Slashdot Mirror
Does Open Source Fail the Acid Test?
Norman Lorrain
writes "This month's Computer
magazine is running an article
by Ted Lewis in which he
predicts Open Source software will hit a brick wall. "
← Back to Stories (view on slashdot.org)
← Back to Stories (view on slashdot.org)
1. Although some Free Software Projects are trying to put everything and the kitchen sink into one integrated package, others aren't. I expect those that do this will die of bloat as Ted says.
2. But many projects won't do this. Instead they'll use other programs in the proud Unix tradition of "make it do one task, but make it do that task well." This decoupling of code bases and reuse is a driving strength of Free Software.
3. Free software that "makes it" is usually well-coded in comparison to commercial software. This makes it easier to debug even if you're not too familliar with it, making the statement "any bug is shallow to someone" correct. 4. Since the main cost of code is maintenance. companies whose main product is not software but hardware or services would like to reduce this cost. Parallel debugging is one way of reducing the cost. We may therefore see a lot more open-source from companies in the near future.
That said, the Linux kernel could be considered to be tending towards 1. A possible way of dealing with Linus-overload is to move a lot of driver-code out of the kernel, and keep it seperate -- like sound used to be, and PCMCIA is. This would have the added benefit of cross OS portability (Hurd, BeOS, Solaris for x86 etc -- the PCMCIA and GGI projects have made some steps in that direction). By spreading driver development over multiple OS's you increase the number of developers and reduce the overall cost in terms of each OS.
GPL is the result of RMS being unable to fix bugs in software he was using.
If it takes off or not doesn't matter. It's not going to go away.
People have made-do thus far without support from companies, and if everyone decided to backout, they'd leave many many lines of useful code in their wake.
- Idcmp
if you're going to nitpick, be wary of nitpicking...
A good counter-argument would be that in practice it works opposite of what he describes. Companies like to sell their products based on how many features they provide. There isn't much to be gained from leaning it down. Whereas open source types tend to make the job easier on themselves by reducing bloat.
Maybe teddy hasn't noticed that...
Open Source has served Linux well in the past because of the tradition of componentization in the traditional Unix userland. As VinodV observed in the Halloween document, Unix-style componentization leads to more robust components (because the component authors don't know what inputs their components will be getting) and enables Unix authors to "party" without stepping on each others' toes.
This form of componentization hasn't applied to certain monolithic bloatware applications (sendmail, emacs, Perl). I personally avoid emacs (using vi or JOVE instead), use qmail instead of sendmail where possible, and am learning awk/sed and Python to avoid the bloat of Perl. Nevertheless, the forces pushing developers and administrators towards this kind of bloatware are subtle and powerful.
GUI applications are a whole 'nother tarpit. Basically, I don't believe Open Source will produce polished GUI applications; instead, you'll get more retrofitted bloatware (GtkEmacs, anyone?) and lots of pointless cloning (57 different IRC clients, anyone?).
Just remember, OSS authors: Source code is a LIABILITY, not an asset. We want you to solve problems using *less* source code, not more.
His "what linux is and isn't" chart seems to go counter to his arguments... but let's just look at the isn'ts:
;-) I know, I know, people are idiots. Vi would never fly in the "real world". That's why we have pico ;-)
1) lacks "Video card support"
I suppose the video signal is getting from my box to the monitor by magic. This smells of FUD. True, there are many cards out there the Linux does not support, but to say it lacks "video card support" implies that Linux has no video capabilities at all.
2) lacks "Wireless LAN support"
I can't comment on whether this is true or not, mainly cause I don't care if it's true of not. If wireless support=mainstream viability, then I must be living on a different planet.
3) a good selection of productivity software
I suppose Applixware, Wordperfect, Star Office, etc. aren't good enough. I think I smell a "no M$ Office=worthless" FUD. It's got perl, it's got vi, what else does it need?
where'd my cookie go?
-adam a
Disclamer: I haven't been able to read the article and I'm fairly new to Linux.
Open Source seems to be lucky. The potenital for planting viruses in something like Linux seems great. If open source spreads like it is meant to, sooner or later the checking process and honest will fail. If this begins to happen and makes it into the mass media, open source would perminitely limited to elites users only.
As long as there are people who refuse to accept substandard software, there will continue to be development of OSS software created in a collaborative effort among them. I cant imagine what life was like before OSS, and I shudder thinking what it would be like after should it fail to exist.
/.
Why compete with MS for users? they have their niche, we have ours. I happen to think MS is good for their market, and OSS is good for the type of people who read
It is a sad day when there is no choice but microsoft software.
I personally believe both sides are good, not to say I love microsoft... but strictly OSS for all users isnt a good idea either as most of em are relatively stupid and couldnt care less what goes into their systems.
I just dont like the idea of trying to annihilate each other. Each serves a purpose, each has its weaknesses.
Flame away,
t0ast.
If the threshold of functionality required for mainstream acceptance of some software exceeds the threshold of functionality required to satisfy the desires of the members of the free software developer community, there is a good chance that software will never have mainstream acceptance, unless there is some supplemental motivation to enhance the software, like money.
Is this a problem?
This smells exactly like the BS that they used to run about the AS/400, except insert "AIX" everywhere you see Open Source. Fuck Microsoft, and fuck columnists who think like this. I work for a telecommunications provider. We have one goal. 99.999(999999999) uptime. I won't ever be able to see that with Windows NT. While I've come close with AIX, and Solaris, and Linux. I'm leaning towards Linux. There's a big difference between having the source, and not. I guess he's making the mistake of thinking the average moron. NEEDS to run Linux. Fuck 'em, let them run OS/2, I need Linux to run my Telco, and screw everyone else. You DO want dial-tone, right? You DO want Voice-over-IP, right? You DO want ADSL, right?
I dunno, forget the article.
Yes, but I'm just some unknown nobody, even in the Linux universe, let alone the "corporate" world. I could go on about the circular logic used to come to the conclutions the origional author made. I could point out the fact that, as we have seen quite strongly, Open Source is increasing the programmer population, not decreasing it, and that creeping-featuritus can't really live in the Open Source model. Good examples of this, from both points of view, are the GNOME and KDE projects. While a real waste of the programming talent (XFCE makes these two things look nearly as bad as an MS product) they are actually bringing more programmers into the projects, keeping the code clean while allowing mega-features in that aren't bloating the programs.
Time will show this to be the case. Just as it showed all those people who, back at SD '95 East, said that there would NEVER be any commercial grade DBMS products ported to linux. I quote a representitive from Computer Associates you, when asked if Ingres would ever be ported to Linux, said "Not only no but HELL NO!!"
Joe
First, the article states that the code in Linux is increasing exponentially and will eventually suffer from "code bloat". While it is true there are new features added to Linux all the time, you certainly don't have to use all of them. That is the beauty of having the source code and compilling your own kernel--you only put in what you want or need and leave everything else out.
Second, he states that there is a scarcity of programmers that can (or are willing to) contribute to Open Source. It seems to me that there are many programmers willing to work on Open Source projects just to see how it is done; i.e. there seem to be quite a few CS students out there just working on this stuff to see how it all works. Many initial developers seem to move on, but so far there have been people to replace them. (The GIMP is an excellent example of this) He also cites that Apache development is now only supported by 20 core programmers, compared to the development team of "thousands". It seems to me that once the main project is done, you don't need thousands of programmers just to maintain a code base and make minor bug fixes and improvements.
I think these two things alone make most of the conclusions of this article way off base.
With commercial software, your speed of development and maintenance are only as good as the number and quality of programmers that you can hire. For each good programmer that you hire, another product looses one. Open source does not suffer from this problem.
One can stay with an open source project while working for a succession of different private projects. The private project have the benefit of your expertise only as long as you are employed there. The open source project may enjoy the benefit of your expertise for as long as you desire to contribute.
The 100 very best programmers on earth could all work on separate projects at work and yet collaborate on a single project in their spare time. Open source has much more potential because it can attract very good talent.
Open source does not have the recruiting problem that private companies have if talented programmers are willing to contribute to existing open source projects rather then try to reinvent wheels.
This is the sort of FUD M$ would like you to believe, but it's simply not true. Free software projects usually have a single or a small group of principle maintainers who guard the 'official' release code very well. Certainly no code makes into release verions without their checking it over - and they have a strong interest in not allowing trojans or other defects into their software since their reputations are relying on it. A lot of these people read code like you read English (or maybe better). Even if a trojan somehow made it through in one release, it wouldn't make it into the second. While it is probably true that only a small percentage of users actually check out the code, if the software is useful and popular enough, that is all that is needed. The defect will be found, people will be alerted, and the code will be corrected.
Don't be daft. The most likely way that viruses
will inundate Linux is if wine gets word running well enough that it can propogate the macro viruses. I'd venture to say that Linux, of all the open source is extremely unlikely to ever have a virus implanted in it considering how well reviewed code submitted for it is.
How is the potential for planting viruses (or virii, depending on who you ask) great if you have the source? It seems to me that there are thousands of people going through the code every day, and someone is sure to notice. However, a binary can contain a virus and easily go undetected with no peer review. I could be wrong, but it seems like it sort of relates to the "security through obscurity" encryption style.
I admit linux still has problems leaving its tech and unix roots and it needs to grow its own way instead of just copying other unix's but if linux ever becomes like windows, I can picture thousands of thousands of angry users fixing the problem.
AN exapmle is the introduction of modules. ALot of linux users were extremely picky about which things they thought wanted to go in the kernel and they wanted the fastest performance on the planet and did not want extra features. The result was modules that you can add and remove. Another problem was ease of use.A couple linux users decided to write installation programs for their distributors for easier installation and another incident when the unix guys were angry about the windows lovers praising about destop intergration and having things work together as one (ms went nuts with this with dll's ole etc) the unix and linux guys wrote desktop enviroments for their widnows managers (kde and gnome).
I for one want to make linux the easist os for all (including mac). I intend to someday write remake of xwindows that has destop intergratoin inside it and I also want to associate fles with icons easier without having to add code in the programs itself. Most of todays linux users dont want this but I want to do this to keep linux growing. Linus himself said that he thinks smp is great but he really wants to see linux come in at the low end. FAce it guys, the high end is already competitive and done being capitalised. its the low end that microsoft has a strong stranglehold and we need lower cost pc's (like the days of the trs 80 and atari). I remember when the apple2 was the rich guys computer (when it cost only 400). Now these same people buy 3,000 pc's. Microsoft is expecting to charge oems 120 or 100 bucks for windows 2000. Its too much for many Europeans. We Americans don't relise that were very rich and average people can't afford 3,000 dollar pc's. If pc's go down in price some more, more of these people will buy pc's and the market will continue to grow. Today, all the anaylists expecct the pc market to standstill in around 2003 and stop growing. We need to make these things cheaper! The oss will free them. If the code gets bloated, people will change it. Unlike other os's if the code gets bloated, its stays bloated and people buy other os's. IF the code was clsoed, then linux could possibly die of this problem.
I don't know what kind of wireless lan this guy is trying to run, but this box seems to be handling my NCR card just fine...
Hi,
l
obviously Linux has/had it's share of bugs, but
all of them have been fixed on short notice. If
you want to have a look at the number of bugs in
a commercial *nix, check out:
--> http://sunsolve.Sun.COM/pub-cgi/us/pubpatchpage.p
All of these have taken weeks, sometimes months
to fix because the source wasn't available. More
often than not, a patch fixes one thing and breaks
three others.
If you think that's bad, keep in mind that Solaris
is probably the best of the current batch of
non-OSS *nixes.
It's impossible to argue when the make the wrong assumptions.
... the availability of source code attracts debuggers"
"The business case for Linux and it's open source cousins is based on two fundamentals: the software is free
This is bunk... As a business users of Linux I feel the more important issues are :
Stability
Standardization
Afordability
Manageability
On the server end Linux handles all of these things much better than any Microsoft product.
They also misquote the Halloween Document (Attributing thing said by Microsoft to things said by Eric Raymond). And flat out misstate a lot of things where do they get that crap about failure rates? If a Linux utility failed on me 9% of the time I would through it out.
Dennis Baker
dennis@cwws.net
The measure of default density, while in many cases useful for gauging the faultiness of a programmer's accuracy, is an outdated way of measuring bugs in a product. To the end user, 1 serious bug in 10,000 lines of code is identical to 1 serious bug in 20,000 lines of code, provided the two programs accomplish the same task. Linux's smaller size in comparison to commercial Unices is an asset, not a liability, and it is ridiculous to suggest that Linux will necessarily bloat to the same size as the others.
As for the liabilities cited by the author:
1. Video card support. (Get a commercial X-server, if you love paying so much)
2. Wireless LAN. (Check out Linux Journal's December issue for the article on "Linux and wireless networking in Africa").
3. Commercial productivity applications. (What Unix version has more productivity apps than Linux?)
The author's good points are lost among his muddle and unreasonable assumptions. The comments about Microsoft seem pretty fair -- unless they get a fine on the order of $15 bil, they've got enough resources that they don't really have to worry about Linux, especially as new markets replace the desktop and back room server as primary revenue streams.
I see no merit, however, in his argument that open source will not influence the way that software is built in the future. Aside from the most obvious examples (Mozilla, sendmail, perl), we could also point out that the upstart (proprietary) BeOS discovered GCC to be dramatically better than commercial compilers. . .
... international economic and political reality.
Why bother with these pathetic technical diversions?
Linux could be much less than the technical
excellence it is and the obviousnous of its
ascent would still hold.
Does anyone think that China is going suddenly start
paying for that 95% of the software that is
pirated in that country?
Someone please prove to me how countries will
soon start exporting major portions of their
GNPes to the US as they computerize?
Windows, Apple, and the lot could be a hell
of a lot better than they are and the obviousness
of the commoditization of OSes would still hold true.
So we have:
(1) World Computerizes and Pays major parts of
earnings to US companies
(2) World Computerizes and continues to pirate
and steal commercial software while ignoring
increasing easy to use and cheap OSS.
(3) World Computerizes and accepts a free,
open source model for OSes as the only ethical
and economic choice.
Which of the above 3 will happen?
Because Windows 2003 might have better PnP,
DirectX, and be more stable is somehow change the
underlying economic and political realities?
This article should be read by the advocates
of the open-source movement because it points
out some issues that the OSS movements
have and thus we can solve the problems by
being aware of them.
The author is quite wrong in his conclusions
though. He does not foresee that OSS movement
will evolve and adapt itself to the new demands
of the future. Market forces and commercial sector
will be fully integrated into the OSS world.
Without this adapatability, OSS would not have
come this far. OSS is sustainable.
OSS (open source software) is not basically about
cost-free software. Its about openness and
freedom. Here the API infrastructure is controlled
by the user-community rather than by a
profit-driven manipulative corporation.
This article makes a number of unsubstantiated claims that on the surface appear to be logical arguments, but which fall flat upon examination.
Take the quote from the article in the summary above: "Success leads to features, and feature creep leads to bloated software." Here, the reader is misled to equate the addition of features with "feature creep". The improvement of an application with necessary, relevant, and reasonable features IS possible without "feature creep". Furthermore, UNIX itself is built along the lines of modularity... which explains such assertions as "Linux has no support for video cards" (well duh, of course it doesn't... X does that job)... and this modularity tends to reduce feature creep to a certain degree.
This is not the only misleading statement made either. There are a number of broad assertions made without any supporting evidence at all, such as: "Already, Apache is losing the performance battle against Microsoft's IIS." Hmm. News to me!
I believe that the educated reader can sit down and rip apart most of this article by identifying both the unsupported statements and the skips in logic, similar to the examples above.
AC wrote:
> GUI applications are a whole 'nother >tarpit. Basically, I don't believe Open
> Source will produce polished GUI >applications; instead, you'll get more
> retrofitted bloatware (GtkEmacs, anyone?) >and lots of pointless cloning (57
> different IRC clients, anyone?).
I disagree with this statement. Problem with
bloated GUI is not in its principle.
Command line applications may be even more
bloated, remember JCL in OS/360?
It is simply one genius come once in the
end of 1960 with concept of shell and redirection,
which made possible splitting command line
apps to independent program.
Now we just need another genius which would
do the same for GUI programming. Tcl/Tk comes
close, but it is not enough.
Corba is not solution either becouse it lacks
something as simple as shell concept
"let other program read output of this program" Lets think about
it and we surely find a solution. There are a lot
of us to think.
Geeez....
About a year ago, we needed to set up a wireless router to a remote building. We decided to slap outdoor antennas on Lucent WaveLAN cards (2Mbit wireless LAN) and roll our own. I shuddered to think about 2 more 24x7 network Windows boxes to maintain. I had never seen or touched Linux, but we heard there were Lucent WaveLAN drivers available. A programmer who had been playing with Linux at home suggested we use a spare 8MB 486SLC2/50 (Imagine a 486/33 with the flu).
You can imagine the rest...that little box also became our ftp server, ran network diagnostics, a name server, company mail, and more. Now we run three Linux boxes (not counting dual boot systems), and half the engineers also run it at home.
I think that the risk of finding trojan in software is directly related to the size or the complexity of the source code and NOT to the fact that it is OSS or not.
The fact in free projects like kde, mozilla or apache is that everybody can modify the source and include any modifications he wants but anybody can review the source and check the source for trojans or backdoors too.
And on the opposite, if there is a backdoor on the NT tcp/ip stack it will be very difficult to find...
Below is what I mailed to the author of the article.
Thanks for an interesting critique on Open Source and Linux. You bring up good points that are certainly worth thinking about. I'd like to point out a couple of mistakes in the article, though, if you don't mind:
Linux 2.0 does not have a multithreaded kernel, although Linux 2.2 does. 2.2 also does include video card support for a number of the more common consumer/hobbyist level cards. If you mean graphics cards instead of video, XFree86, usually used with Linux, has fairly good support for most common chipsets, and of course there's also the possibility of using the (commercial) Metro-X or Accelerated-X from Metrolink and XiGraphics, respectively.
Re failure rates / failure densities: While Linux kernel itself is in the 1.5 MLOC range, the entire tool set, which includes the GNU utilities and much other code under GPL or BSD-liek license, is much bigger. Not having all the source code installed, I can't give you an exact quote on that. That would affect your failure density estimate for the entire operating system installation considerably. Of course, at the reliability levels we're both interested in, hardware failure is a big factor, and although Linux supports RAID storage, it doesn't yet offer any kind of general redundancy clustering. That, not the "lack of a good selection of productivity software" is a feature sorely missing in Linux compared to the commercial UNIX vendors like Sun and HP.
It's clear from the article that you're aware of many of the lead programmers in the various important free software projects (Linux and Apache being two that you mention) having been employed by companies directly or closely related to the open source market. However, you seem to have drawn a strangle conclusion from this fact. It is not at all obvious that their employment in these companies means that they're not available for furthering the development of the projects any longer. To the contrary, in fact. Most of these people are on payroll with the specific task of developing and maintaining the free software project they "came from", and the companies in question readily accept that the code these people produce will immediately be released under the original free license. Why? Either it makes the total product of the company better (as is the case for Whistle Communications, whose file and web server appliances run the Samba file server - Whistle employs Jeremy Allison to develop Samba), or the company's internal use of a free product is large-scale enough to make it cost-effective to have one or more full-time employees developing it (which is how the Apache web server was born).
In other words, commercial distributors of free software don't take away momentum from the development of the software, nor does it mean that the software no longer is free. There is no threat to the free availability of Linux, even if Red Hat or someone else did become a controlling force on the commercial Linux market. The 1000 copyright holders of the Linux kernel ensures that - all of them would have to agree to commercialize the OS under the same commercial entity first. Nor could Microsoft "absorb" Linux, as you put it, for the same reason. They could embrace it, but that would only make it stronger, and all of Microsoft's contributions would also become free.
I'm also curious where your market share figures were based on. You claim that the Apache web server is losing market to Microsoft IIS. The most reliable survey I know of in this market, the Netcraft web server survey, has indicated for several months that Apache is not only the biggest, but also the fastest-growing web server product on the Internet, with a 54% market share this month. IIS market share has actually been fairly flat for the past few months at 23-24%, while Apache is constantly stealing away market from Netscape and other smaller products. Intranet deployments naturally can not be reliably counted, since there are no "sales figures" for Apache or other free servers.
You also suggested that the Linux market share is 3.6%, which even if it were true (which is impossible to verify with no concrete data on actual Linux installations) is quite misleading, when a majority of Linux installations are low- to mid-range server systems. In this market segment, UNIX is not quite as weak as your article would lead to believe. Dataquest's server market survey in October last year showed UNIX market share growing, dominating over NT over two-fold. It's difficult to find reliable data on the market share of Linux or the various free BSD derivatives, but it would appear to be somewhere around 14% of the server market, concentrated into web and file servers. According to Datapro, Gartner Group and IDC, Linux is the fastest growing non-Microsoft OS, in fact according to some reports it may be the ONLY non-Microsoft OS growing market share.
I can't help but think that your initial position might have been somewhat biased. Admittedly, so is mine - I like working on this platform. I guess it remains to be seen whether your acid test proposition holds or not.
This article is FUD, pure and simple, probably paid for by a "very large software house".
It has already been shown that GNU software has a significantly lower failure rate than equivalent commercial software.
I'm currently enrolled at UNLV and I'm having a tough time finding :/
anyone else that uses Linux... *sheesh* All the programmers have this idea they're going to work
for microsoft and make lots of money? *sheesh*
Anyway... I've got to start a linux group on campus
as I know there are others there... just not where they're hiding...
The argument in the article to disregard defect rates and instead use defect density is a completely bogus argument.
.signature file.
Consider the following two programs:
Program A contains 10 bugs in 100 lines of code.
Defect rate = 10
Defect density = 10%
Program B contains 10 bugs in 1000 lines of code.
Defect rate = 10
Defect density = 1%
Each program contains the same number of bugs, but the bloat of program B covers up this fact under Mr. Lewis' metric.
Secondly, neither of the measures mentioned in the article concern themselves with either the likelihood of tripping over the defect, nor the results.
If my program contains 2 defects, both are weighted equally. If one of those defects is that it spontaneously destroys your hard drive, that is a much deadlier defect than if the defect simply doesn't save my
Further, no weight is given to the possible relevance of the bug. Consider a bug that makes it impossible to even launch an application (one that EVERY user would trip over) and a bug that is only tripped when you select 12 different items, click three buttons, pause for 10 seconds and then hit cancel (which is likely to only ever be seen by the developer).
Without having a decent metric upon which to determine code quality, Mr. Lewis' arguments about the reliability of Open Source are incredibly weak.
I won't even get started on his attempt to extrapolate userland defects into kernel space...
It's useless responding directly to this kind of FUD - but I can respond to Senegan's eagerness to publish such stories. His lack of editorial wisdom seems incurable, unless one considers possible motivations for such apparent cluessness.
If this continues soon there will be little but stuffed penguin adds and such articles planted to generate controversy (translate "hits" for $lashdot). It seems that if the author of this story is correct about anything he's correct when applying his analysis to $lashdot, which seems lacking in quality control these days.
Well, one of the good things about the OSS community is when a website (like a system or an application) is no longer serving its purpose somebody can spin off another which does. We aren't beholden to $lashdot or anybody else.
Although I intended not to honor this story by responding to it, I will say that all the author's criticisms apply to commercial software, not OSS. $lashdot is becoming like commercial software, driven by "hit count", advertising, and the budding journalism careers of its editors.
It may be time to move on. $lashdot is having problems keeping its database uncorrupted these days, anyway. The next step will be selling all our names to junk mailers.
Oh, but the faithful will say. These people are making such great sacrifices for "us", putting in long hours for no pay to bring us the latest and greatest. And I'd like to sell you an NT system to manage your nuclear power plant.
Trash the Slash
If everybody is working on the GUI, why does everyone still work on the command prompt.
Is Apache, BIND, Sendmail, Qmail, the Linux Kernel, the GNU toolkit, the egcs compiler, Perl, Python accessible without GUI.
The "dirty work" that you talk about is the "fun" part for a true system programmer
I think Linux is starting to cross the chasm btw.
Quite a few of Lewis comments seem to be straight out of a marketing book with this title. The key point is that "crossing the chasm" is hard to do and the way to do it is to focus on one market exclusively and refuse to be drawn outside of that market until you have conqured it completely. Then find another realistic market to tqke over. Wath MS they are very good at it and they started with DOS, it was their strongest card.
I would suggest that this market for Linux is Apache/Linux Web Server. Apache holds 54% of sites, that is not 54% of pages served however, and that is important to remember. If the Linux/Apache combo can take this market over then expansion into more areas will follow. I think the oss community would be well advised to focus carefully on dominating one market segment. I would suggest that Apache might look to a OSS sql database for serving web pages from and extend into that market next. php or similar maybe.
Interesting thought that western Government is built on democracy and openness, the same could be true of software. Human nature doesn't thrive in a secretive power hungry structure, it usually destroys itself through its own arrogance.
However I don't expect that to happen because free people are contrary, they don't like the focus of hierarchical structures and so Linux strength is its diversity,as shown by the range of platforms it runs on.
However I actually believe that Linux will be successful in low cost appliances because a US$40-50 componet in your $400 component price item is a big deal, and if you can use Linux for that, then that is compelling when you roll 1M units. So embedded units systems may pick up o Linux. Netwinder for example, but netwinder must be able to be a client for Citrix, I am not sure if it can?
Other niches will be successful to, the thing is though Linux is written for programmers by programmers.
Pay attention to his facts though, not very accurate, looked at the mail and dns systems of the internet recently?
another AC wrote:
>Have you ever seen the little doom clone that the Microsoft programmer put into excel
No, but I'd love to.
Any instructions for how to get at it ?
Classic marketing ploy by someone with a conclusion in search of a collection of words to justify it.
"No Wireless LAN support", slow Linux release cycle, Apache starved for innovation...
This article is the worst one I have seen in a long time.
Anthony "forgotten password" David
Linux will succeed as a tool to tie together networks of Dos, Windows, and Mac computers. It shines in terms of serving printers, modems, group email, and files, and it's convenient to use, once installed. The file system, if it crashes, is fairly easy to recover, much easier than Windows, but not as easy as Dos.
It probably won't work on the end user desktop. It's seems to be to difficult to write apps for it. Doesn't seem like there will ever be any programmable databases like Dbase, Nutshell, Q&A. Without these you can just forget it.
For languages, businesses have their choice of C++ and C++, both curses upon productive society. The rest - Perl, Tcl/Tk, Python - all have their own serious omissions in critical areas that make them unusable for serious work like customer lists.
You L1n3ks h4krz need to pick up some of those old Dos apps and see firsthand how to build a full featured application. Paradox 4.5 DOS (not Windows) will lead you in the right direction. It'll teach you more about ergonomics and design of the feature set than ten years of listening to some dopey college professor.
>It's useless responding directly to this kind of FUD - but I can respond to Senegan's eagerness to publish such stories. His lack of editorial wisdom seems incurable, unless one considers possible motivations for such apparent cluessness.
/. more then leave. Now.
/. will mention it.
/. sold YOUR email. They don't have mine.
Maybe its because us readers want stories like this? Why don't you go to zdnet and shut up.
>If this continues soon there will be little but stuffed penguin adds and such articles planted to generate controversy (translate "hits" for $lashdot).
You're just pissed because you don't get as many hits. If you don't want to give
> Well, one of the good things about the OSS community is when a website (like a system or an application) is no longer serving its purpose somebody can spin off another which does. We aren't beholden to $lashdot or anybody else.
Well go set one up then and stop complaining. If its really any good I'm sure even
> Although I intended not to honor this story by responding to it, I will say that all the author's criticisms apply to commercial software, not OSS.
$lashdot is becoming like commercial software, driven by "hit count", advertising, and the budding journalism careers of its editors.
Perhaps, but so is zdnet and all the others. Why don't you just sell your computer and not bother.
> It may be time to move on. $lashdot is having problems keeping its database uncorrupted these days, anyway. The next step will be selling all our names to junk mailers.
Idiot. Slashdot is a web site, not a person. Since when has a bunch of HTML been able to manage a database? I'm sure it'll be fixed shortly. This is just a retrofit of fact to suit your arguement. Badly.
Personally I couldn't care less if
> Oh, but the faithful will say. These people are making such great sacrifices for "us", putting in long hours for no pay to bring us the latest and greatest.
Nope, they just have lots of stories I and others like to read. Vote with your feet, you bore us all.
AC
Steps:
1. Take code below and copy to file 'bug.sh'
2. Do 'chmod 0777 bug.sh'
3. Do './bug.sh'
4. Notice how bugs/lines of code ratio has changed
-----[cut here]----------------
#!/bin/sh
#
# Adds 30,000,002 lines of bug free(tm) code to Linux
# for that 'extra low bug density'(tm) feeling
#
file='/usr/src/linux/include/bug_density.h'
echo "/* include/bug_density.h */" > $file
echo "/* Bug Density Patch 11/02/1999 */" >> $file
echo "/* Anonymous Slashdot Reader */" >> $file
echo >> $file
echo "void bugdensity(void)" >> $file
echo "{" >> $file
echo " if(0){" >> $file
for i in 1 2 3
do for j in 1 2 3 4 5 6 7 8 9 0
do for k in 1 2 3 4 5 6 7 8 9 0
do for l in 1 2 3 4 5 6 7 8 9 0
do for m in 1 2 3 4 5 6 7 8 9 0
do for n in 1 2 3 4 5 6 7 8 9 0
do for o in 1 2 3 4 5 6 7 8 9 0
do for p in 1 2 3 4 5 6 7 8 9 0
do echo ' fprintf(stderr,"And another line of code");' >> $file
done; done; done; done; done; done; done; done
echo "};">>$file
echo "}">>$file
echo "/* end of bug_density.h*/">>$file
------[end]-------------
Wow this guy has read 'crossing the chasm' book
WOWWWWWWWWWWWWW!@
That book is good but meant only for COMMERICAL
software moron!, get a life. So has the Porsche lost it?
I mean its not mainstream at $50,000 is it.
Simplistic views breed narrow minds!
Who gives a ratts ass if linux is mainstream or not, because it works well and all techoss love it and MS earns less $$$
ANyone pro NT, anit-linux must own MS shares, so dont listen to them
I feel that the author of this wondrous article who hides behind a secured server misses the point of the article he quotes " .. The Bazaar .." where they talk about the differences in culture between a potlatch ( give-away culture ) and the more "corporate" ( give me culture ).
... "Even if Linux proves to be the better client, Microsoft can always use tying - the creation of MANDATORY DEPENDANCY among applications - to maintain it's monopoly."
... thinking of Diamond cards in particular. My poor Fire GL Pro 1000 is just gathering dust, nice card doesn't play well with others ... )
Languaging like "absorb and extend" reeks of Borg mentality and homogenius culture.
I think what attracts me to all the flavours of un*x is the personalization of the operation environment. It is putting the choice of interface design back into the hands of users.
Look at all the various window managers for XFree86. Sure some of them may be buggy, I think the wonder and attraction is that as a user, one can communicate with the developers to make feature requests, offer bug fixes, portability clues to other OS's...
The author seems to have a love affair with M$. True, I do enough of my own bashing of M$. If it wasn't for them, I wouldn't be making so much $$ in supporting WindozeXX. Thanks Billy!
I also think the author is flawed in his calculations about the code size blah blah blah. Here he is assuming a traditional economic model of continued growth. How many times have apps been rebuilt from the ground-up as major design paradigms in the developers efforts changed the way they thought out the API's? ( And some apps keeping traditional API structures to boot !).
Here's a clue
sic "..And the ploys of a WEAKER entry level player trying to upsurp the STRONGER players". Count how many times the author uses weak in this diatribe of ivory tower mind tweak.
Another response points to the authors predominant focus through his collection of books "How to make money off your personal computer".
I tend to apply a permaculture model to computing. It is like any other tool. There is an old Peruvian saying, "a man is known by the tools he uses".
Does it really matter was OS is best as long as there some benefit from the practice of use? What do we use computers for anyways other than to strucuture reality into coherent forms, make connections, build alliances, create art, express the mundane, the commonality, and the bizarre bazzaar?
I use NT right now to post this. I do my development on Solaris. I like Linux because I can see the next trend in the Gnome/Gtk apps. I support Macs because the redhead who needs help is a cutie. It's all good.
Then again, I tend to favour the give-away side of the model, to use tcom products to create better designs.
If the designers @ Microsoft want to figure out how to get rich by directing where I want to go today,... yeah right,... not there!
And as far as usability and productivity, using any flavour of un*x seems to be more productive regardless of flavour ( RedHat, Suse, OpenBsd, Slackware, Debian, Solaris ). I just downloaded Corel WordPerfect for my RedHat installation. What else to people use computers for other than writing or making digital art?
Oh yeah, no mention of Gimp? It rocks!
Oh poor Microsoft, we bash on you, oh boo-hoo-hoo. A good example of a give-me culture where M Gates has more money ( a resource ) that all the poverty nations in the world put together. I think there is something wrong with that. Oh yeah, he just donated 100 million to path.org, about 2.17e-6 % of his actual wealth.
It comes to haves and have nots. However, the have-nots can breath life into those old 386 machines and have world-class operation systems ( speaking of linux ).
Oh wow, I am out in galatic central! Worf! Plot a course back to Earth! Engage!
So when I think about what goes on in the world, we still have problems with over crowding, smog, diminishing natural resources, the lure of better faster machines with better faster more intuitive interfaces. How many times have I tried to walk my Dad though the control panel on WinXX? Or explain that "this program has performed an illegal function" was NOT because AOL was breaking the law.
So Microsoft this and Microsoft that. Hey, it it works for you great! If you have half a clue about making the most out of your computing experience, then maybe you want to check out some flavor of linux.
Productivity, hmm the author mentions sendmail, bind, ftp, and GNU gcc ( ecgs ? ). I think he is tripping when he says that there is poor video support ( maybe read that HCL, it is the same for WinXX and any other system , or get support from the european markets
Back to my original thought of design paradigms. One thing I like is all the true innovation of the oss developers creating tools such as gcc, python, perl, tcl, all the libs, mysql, postgresql, apache. Roll your own app! Donate some help to the alpha code warriors/warrioresses!
Waiting for the fifth generation programming languages. Voice interface, somewhere in the 24th century...
OK, must sleep, end of rant...
Closing clouds, open source tweaks the corporate model, kant we all just get along? Whatever works best, maximum leverage of available resources to mutual benefit. Plant a tree or two..
Anonymous Coward
"I am a alien from outerspace,
Come to save the human race,
Living in my breifcase,
Living in my suitcase,
I am an alien I say,
I am the word professor,
I am a word processor"
Lee Scratch Perry
My thoughts are the result of extreme exposure to multiprocessing systems
running multithreaded enviroments
There is no way to do serious data entry on a Linux workstation.
Don't give me a song and dance about XForms or Netscape forms. There's no way to validate input as it's being done.
Set up a web database, and enter your phone bill 100 times. Can you do it in less than 45 minutes? I didn't think so. That's because the client has to send it over the network to the server, the server checks it, rejects it, and sends it back over the network to you. Can't get anything done that way.
The way it's done is with FIELDS, with a real lookup window. The data is checked as it's being typed in.
DOS has apps that do this, Windows tries and often succeeds, but nowhere in Linux can this be found.
And Linux won't succeed until someone like me has an app like this.
Bullshit... anybody who knows the source code and the operating system well enough to be able to stick a virus in there just WOULDN'T DO IT. Have you ever heard anybody who's gotten past the initial "gee, this is kinda hard to use" aspect of Linux and actually used it for a significant period of time complain about it, other than say what features they want in it? Personally, I've never heard of anybody running Linux for a significant amount of time and realizing that Windows was better and switching back (well... except maybe a long time ago because of hardware support.) Anybody who uses Linux for any amount of time gains a respect for the operating system and wouldn't code a virus into it. And they also wouldn't want a virus in code that they run themselves...
Matt Spong
spong@wam.umd.edu
It's understandable that OSS products are challenged on the possibility of virus-vunerability, bugginess, and code bloat. Understandable, because these are real, legitimate concerns that we face all the time when we use commercial software.
It's also understandable that OSS users respond to this challenge with "huh?" Simply because these issues are never notable in open source software - they represent the strengths of the code that open source developers put so much pride in.
Mr. Lewis claims that
"Unix has more than 10 million lines of code, while linux has only 1.5 million".
This is inaccurate.
While the 2.0.36 kernel has (by my count) 918835 lines of code, the 2.2 kernel has 1603773, which is nearly twice as much, and it is not the only vital part of the system. If you include X, the compiler, the C library, as well, you're up to about 7 million.
And then there's BIND, the shell, all the shell utilities used by the startup scripts... In fact, if you go the whole hog and include everything on the distribution CD, it is much more. Counting all the lines of C, C++ and shell script code on a the Red Hat 5.2 source CD-ROM, I get a total of 18804260 lines (in 432 packages). That's over 18 million lines of code. Naturally, not all of this is vital for exerybody.
I think Mr. Lewis has been reading Darrell Huff.
- James Youngman
(who can't remember his Slashdot login).
How many people are involved developing the Linux "core"? Don't compare it to how many people are developing GUIs... there is much choice in GUI and that is an advantage of Linux. Compare it to how many people are working on the "core" of other OSes. My guess is that Linux has far more!
They probably didn't publish them because they were frightened!!!
Where has this guy been?
First, you can't compare a 25 year old unix OS
to a 5 year old Linux OS. It's like saying
my dad is wiser than me. Well, no shit, Sherlock!
But let's see Linux will be 25 years from now.
Second, it HAS passed the damn test. He obviously
doesn't realize that he's been using this stuff
for YEARS!!!!
Scarcity of talented programmers? Was that supposed
to be an original thought or is he just saying what
everyone else is saying? Come on! The world is
in general lacking people who sit down and like to
think and read.
Bah! This guy just writes flame-bait.
He bases everything on a flawed assumption that all (open-source) developers as stupid and can't organise themselves into a team as well as commercial developer groups can.
/dev/null
cp * >
Perhaps this may be somewhat of a valid point for open source, but lets look at another acount of something that is not a "fad."
I read an article (sorry I'm gonna take the Scott Adams approach to quoting) about a year ago, I don't remember who it was by or where it was from, but it was hanging on one of my professor's doors. It pretty much said that computing in general has taken a turn towards feature creep. Do you *really* think that you need 16 M of RAM to run a WORD PROCESSOR?? I used to be able to run a word processor on 4K of RAM on a TRS80 (yes I had one). Outside of a few select group (and maybe even them), who really uses all the features of a word processor.
Outside of that tangent, look at support. Has anyone ever tried to get good, speedy support from someone like Micro$oft? They need a credit card number before they will even listen to your call. Presently I'm trying to get a bugfix from --company omitted-- that I should've been able to download over the webpage that is costing me $10 for them to burn a cd because they can't figure out how to make a patch for it.
Perhaps this problem isn't limited to OSS, and perhaps the people who say that most computing systems that pass this "Acid Test" are a little narrow minded in their judgement. I'm not against progress, I'm just against someone singling something out because its presently working as well as anything else around.
I know it asked for non-flame comments, but right now I am too tired to express a well-thought reply to this article (maybe later).
Just want to point out that this seems like a flagrant case of FUD, and it is an EXCELLENT candidate for a flame.
I haven't looked into it, but XForms should be simple to modify to add this capability.
Unless you mean to run a full verification on each and every thing as you type it in. If that's the case it's not that hard. It's just a bit more work. This is really the job of the client program. It should be doing this before sending it to the server. The server should just store things. It's faster this way.
If it's so important to you, why not write it? If you don't program, how about going to the Free Software Bazzar and putting up some money for it?
I think this is the problem of the programmer of the apps you use. It wouldn't hurt to have a lib that could be used by those programmers to make it easy.
It's the same algorithms and languages whether you're on Windows or Unix. The details of the API are relatively insignificant. A person should be able to transfer CS knowledge back and forth. So the question is, which environment would CS students prefer to work in, given the choice, and given that it will not decide what job you can get. Since Windows source licenses are not normally available (and Windows generally sucks in so many ways), Unix is an obvious choice for university people. Free Unices are that much more better because restrictions on source are less than commercial Unices.
Your football analogy is just backwards. Windows is seen by many incoming students as the only platform choice. It's Unix that broadens horizons. I don't think any American CS student in the last few years has discovered the magic of Windows after laboring under the misperception that Unix is all there is.
I find it quite interesting when the author talks about the incredible impending code bloat, as if Linux the same thing as an open-source windows. Microsoft's mega-kludge functions as a monolithic entity, and encompasses a huge amount of code for common functions. Linux is a completely different beast, with different maintainers for each component of its system. The kernel, libc routines, and the daemons and commands form the base system, each seperate and distinct. Even pruned down to a minimal form, it is still "GNU/Linux".
The system becomes more attractive and useable as things like X and window managers are added, but because of Linux's component-driven technique, just because one portion of the system has a problem doesn't necessarily mean the entire system is affected.
This is especially true in light of the variety of choices for certain modules, such as window managers, shells, and programming languages.
Overall, I think the "core" of Linux is very careful to avoid code-bloat so it can be as small as necessary. I haven't seen a lot of new features added to ls or fdisk lately... The programs that grow the most will have to deal with the added complexity in a way that works for them, but just because say GNOME or KDE goes under in a swath of bloat and bugs doesn't affect users of other products.
Of course the kernel is probably the best example of the need to avoid gratuitous features and most of all, bugs. This may serve as the metric or "trial by fire" of how future large projects may need to function (or avoid functioning) to avoid de-evolution.
Finally, as others have pointed out, as the popularity of Linux grows, so does the base of programmers and maintainers of software. Somehow the author of the article assumes that there are a finite number of people to work on OSS projects. His assumptions can be summed up as:
OSS as a model will never work. -This is because the total code size will eventually defeat efforts to maintain it.
OSS is maintained by a fixed, non-increasing number of programmers.
OSS's growth rate is large only in respect to its current size.
OSS can be absorbed by Microsoft, as Microsoft is "strong" and OSS is "weak".
Writing OSS has no reward and must become commercial software to survive.
Whether or not OSS succeeds in the dominating a large portion of the world, or simply colonizing a smaller segment of it, it will always stand on its own simply because it is its own reward. It is the only way individuals can express themselves through coding and collaborate with others around the globe in such a organized manner. The ability to take code and extend it for yourself and pass it along for others is OSS's unique characteristic. As the number of people who discover this grows as well as the users of this code, OSS will continue as a success.
Larry B.
I don't know why people bash emacs so much. Its available on just about any platform (i use it here @ work on windoze at my PC and on our AIX box, Xemacs and emacs at home(Linux, NT) and school (Solaris,HPUX) It is wonderful being able to use the same editor on all those platforms writing C,C++,Java and whatever else my heart desires.
Yeah, i could use VI or whatever, but emacs RULES!
all this world needs is a convenient way to verify the authenticity of the package you're installing. problem is, for most things this already exists but nobody uses it. because it's a PITA and i'd rather trust people. (kind of like driving without a seatbelt... you never think you're going to be in an accident, so you don't wear them, then pow!)
One of the authors biggest misconceptions is
that linux doesn't have the ability to consume and
extend, that is linux's greatest ability, it can
absorb the features of other products that are
valuable and forget about the features that aren't
important or needed. Athough i can see where
project complexity will become an issue, i can also
see the comunity learning an effective means of
handling it.
Shads
I don't know where Ted Lewis got his figures from, but the Microsoft ones mentioned earlier in this thread were not done by Microsoft and actually include a URL to their source:
/ bench.html
http://www.zdnet.com/pcmag/features/webserver98
Apache tops the list on one of the benchmarks, but of course not the one MS cite.
Aside: has anyone else checked out Ted Lewis' site and the people who link to it? Let's just say his PhD supervisor is an interesting person.
mml
Simply put the article is full of assertions with little or no substantiated proof. I don't know the person who wrote the article but he can't possibly have studied OpenSource,Unix or Linux to any extent. Otherwise he would know better.
Just one example of this FUD:
"As the "Linux 2.0: What It Is and Isn't" sidebar shows, Linux has yet to incorporate many of the features and applications that will doom it to a complex future."
Now what are these "many" features? Exactly 3, 1 of which is "no Video support". Linux doesn't have it because it doesn't BELONG in Linux, it belongs in X. A second missing "feature" is "no applications". How is this a "missing feature"? How could the growth of the number of Linux applications possibly affect the size of Linux itself? You could have WordPerfect,Smartsuite and Office all ported to Linux and not one line of kernel code will change. So the only "missing feature" that is relevant from his list is "no Wireless Lan support". While I'm sure this is important to someone, adding this support is unlikely to cause major hardship to kernel developers.
The rest of the article is likewise unworthy of having been published in the first place. If I was the paranoid type I would suggest that this article was funded by the unnamed gorilla from Redmond.
Doesn't Emacs stand for Extended Memory And Constantly Swapping?
>But Emacs and Perl? Bloated? I think not. I've
>never had emacs behave in a slow, buggy or crashy
>manner
What does "slow, buggy, or crashy(is that even a word?)" have to do with bloat? What about Emacs 24M size?
>Yes there are a lot of features I don't use, but
>how on earth does that make emacs bloated?
Um...That IS the definition of bloat. Word 97 is not bloated because it is slow, it is bloated because it has 1000000 features no one (ok, maybe 1 person) uses. Look up bloat in medical (vet) book. Should be something like "Swelled with excess gas. Potentially deadly." Not slow, and "Buggy". I hope to God you never enter the software industry. It would be like Dilbert; "We can add as many useless features as we want, and as long as it is fast, it is good!".
>Hell, it doesn't even *LOAD* alot of the features >I'm not using. Just because something is
>extensible, it's bloated?
Yes, if the bloat features are installed as "default". Netscape allows plugins, You do NOT get every plugin when you download Metscape. Big difference.
>Finally, if you're calling Perl bloated. You
>obviously have never used it for more than half
>an hour. Perl has got to be the best programming
>for rapid prototyping. Things that would take
>hours to program in any other language, takes
>maybe half the time using Perl.
(The best programming ______? You mean language?)
Yes, Perl IS bloated. If they simply distributed the core Perl, and left the other crap on CPAN where it belongs it would probably not be. You have a goatee don't you? I thought so. Just because you like something does not mean it is not bloated. Have you ever used M$ VB? Powerbuilder? These are also nice RAD programs (My apologies to the anti M$ folks). But they are ALSO bloated.
>Frankly, you've got a lot to learn before you go
>criticizing other peoples software.
Frankly, you've a lot to learn.
--Peter
--Appropriate
I used to be an IEEE affiliate, and get Computer, but when I changed jobs, and my new co. wouldn't pay for memberships, I had to decide between Dr. Dobbs' and Computer - you can guess which I chose. One reason was that not only was Computer overly academic, but it was *also* full of ivory-tower bs.... The ultimate example of this was the front cover of the last Jan. issue (1994?) I got, which was presenting OO as the Silver Bullet. Sorry, guys, but I've been in the real world *way* too long, and I'll go along with Dijkstra (do they still have y'all read his seminal '68 article, "No Silver Bullet"?).
mark "been on m/fs, *&* pcs, *&* w/s..."
Considering that this is an article on a pay site that couldn't even figure out how to convert it from the proprietary Acrobat format to HTML, I feel fairly secure in assuming they don't "get" the Net or the current state of computing at all. Its flawed premise and numerous inaccuracies are further indications, whether or not one believes OSS will return to its original dominant position in microcomputing. If posted to Usenet, this would be a very effective troll - as we can see right now on Slashdot (and I'm guilty as the rest of you.)
sorry if this was covered already (we need an autosummarize feature when there are this many comments already -- or would that be code bloat?)
what kind of assumptions did this guy make where he assumed he could do some curve fitting to extrapolate an exponential function to project the number of lines of code in the kernel? let's see, in 1991, it was barely a twinkle linus's eye, then about 1994 things started seriously coming together, then version 2 came out, and that was pretty feature complete, and now what's left is more hardware support and things like better plug-n-play , USB, et al. if anything the graph should be fairly linear from 1996 on. for most purposes, the kernel is feature complete, something it hasn't been. don't believe me? ask linus or alan what's going to be in version 3. they don't even know if there is going to be a version 3. it wouldn't surprise me to go from 1.5 million to 2-3 million for 2.4 in 2000/2001, but that's pretty linear when you think of where the project is going.
of course when you're trying to fit a curve, it sucks to have the parameters change half way through; apparently he felt it was better to just ignore that because he could make his case that in a couple of years with exponential growth, linux will be as bloated as windows is/has been for some time.
Either that, or Delphi. Unless he's talking about NeXT/Rhapsody, in which case he might mean Objective C, though I doubt it.
Outside of RAD tools, C and C++ are the primary systems and applications programming languages for every major operating system in existence.
BTW the phrase "serious omissions in critical areas that make them unusable for serious work like customer lists" sounds completely nonsensical to me
Unless-- hmm, customer lists-- sounds like he comes from an IS/financials or management background, and considers SQL or 4GLs the same type of language as C/C++? ("Well, they're all computer languages, right?") If so, of course, his complaint boils down to the standard complaint about a lack of applications, since from a systems programmer's point of view, a 4GL authoring environment is just another appplication. One that, perhaps, has not been written yet, but there's no good reason to believe that one never will be.
~k.lee
What's XForms got to do with the viability of Open Source? As you pointed out it isn't really, really open so it doesn't qualify.
If, however, you're commenting on the fact that XFCE/XFWM is based on XForms and that makes it something unusable... I'd suggest you try it before you make judgments. If you did you'd see how wrong preconcieved opinions can be.
Joe
And in this particular case we talk about an editor that is so user unfriendly that one cannot even open a document ( C-x C-f ) save it ( C-x C-s ) and exit (C-x C-c ) without doing A LOT of RTFM.
Well, I've got an emacs session open right now -- open a file? Try the "Files" pull-down menu and choose "open file". Save the file? Once again -- do the "Files" pull-down menu and choose "Save buffer" or "Save buffer as". Exit? Choose "Exit emacs" from the "Files" pull-down menu.
Now, that wasn't so hard, was it?
It seems to me we want to distinguish OS and its core tools and other softwares more clearly than how it's done in this article. OS and its core tools are the base on which many other softwares can flourish because it's free. I'm going to call "OS and its core tools" simply OS below.
If a blueprint of an automobile design were an OS, current situation is like all automakers having to purchase a blueprint from one company because the blueprint imitated successfully the automatic transmission invented by others and the mass hasn't discovered (and perhaps is not interested in discovering) the fun of driving a stick car.
But the design the blueprint offers is not that great. Moreover the parts makers are often left at the mercy of this single company.
Given all the importance of a blueprint it's natural to see a movement to making some blueprint that's accessible by everyone. Linux is such a phenomenon. There are too many people to gain for this not to happen. Note that for an OS to be such a common blueprint, it must be free.
So I don't think Linux popularity or the open source movement behind it is just a fad nor a "strategy for the weak". The premise is based on real needs and real desire held by many.
Hidehiko Ichimura
University of Pittsburgh
The whole thing is a piece of bad work. First he explains that exponential growth in market share is something that cannot be sustained, then he makes some plot suggesting exponential growth in lines of code could. The figures are not really accurate, as they omit the existing GNU work when Linux started. He can't be bothered to check his facts even enough to spell Linus Torvalds' name correctly.
He does not understand the metrics of error between utilities tested in proprietary Unices and the GNU tools: in both cases, exactly the same utilities were tested, and a lesser percentage of them found to be defective.
He does not understand that Linux *is* already commercially employed, developed and sold, and he does not understand that it cannot become a proprietary system.
He does not understand the the code size of Linux mostly is split among different, separately maintained and manageable projects. (we don't need no stinking graphics server in the kernel!).
While certainly an argument sceptical for the future of Linux development might be construed when carefully examining the movement and conditions it works under, he utterly fails to do so.
I would certainly hope for better done analysis of the dangers and pitfalls Linux will have to face.
This article is simply an opinion piece based on little more than imagination and hand-waving.
>What do you think?
>Will china copy RedHat linux distribution or
>copy one cd with windows 2000?
Actually China will use TurboLinux. 1 million
copies of it were sold in the far east last
year. 1 MILLION! It outsold NT at surveyed
POS locales.
It will be 2 with 3 gradually occurring over the
few years.
Obviously we need more apps, commercial apps too,
but given an increasing easy to use Linux which
at this point with KDE almost equal to NT and
which will soon equal or exceed 98 / NT.
The choices were not multially exculsive. But
the undeniability of 3 is obvious.
"You mean like Emacs? Is there anything that program is not trying to be? Editor, email, newsreader..."
.el and info files are really necessary for emacs to run.
IMHO this is just a distribution issue that gets frequently misclassified as an intrinsic weakness of Emacs. It's a little like saying Linux is bloated because Apache takes up so much space.
As many have noted, Emacs is really an engine for running text-editing applications, not a text editor. The Other Side's equivalent is Visual Studio, which provides syntax-directed editing support for multiple authoring languages. The difference, of course, is that to extend or shrink Emacs you can just load or unload an elisp file. Extending VS requires a bit more effort (ha).
If somebody would just prepare & distribute a truly minimal Emacs distribution for home platforms, I think complaints about bloat would decrease. Only a handful of
(OTOH, if you are running a large site, e.g. at a university, you probably want to install a large number of elisp packages because chances are at least some of your users will need most of them. I personally use at least a half-dozen programming modes regularly.)
~k.lee
what i was pointing out was even if you grab the source from a "reliable source", you still should use the checksum, which is a PITA...
login: cypherpunks
password: cypherpunks
let's take a look
This is true, but changing. More applications, particularly web-based applications are being written in platform-independent programming enviornments. Right now, environments like Java and Tcl/Tk are useable only for small applications, but if the next generation of platform independent enviornments is up to the task, I bet you will see mainstream applications starting to use them
1. dy/dx = y(1-y). That's the logistic differential equation in its simplest form. It's a much better model for biological growth after a certain point because exponential growth becomes unwieldy after a while. At least he recognizes that W2K (the _other y2k bug :-) is going to become totally unwieldy even faster. And Microsoft is not going to have the wisdom to stop feature creep and bloat (Office 97? HA!) while Linus may. I don't see the logistic equation as quite applying to software, but the exponential growth may become more linear.
2. Someone else already mentioned this, but his use of percent reliability rates is totally fucked up. If the ratio of LOC to bugs (= defect density) stays the same, the percent stays the same. See #1 above here - why should Linux get that much bigger? It's also more modular than Win, which helps in getting specific people to fix specific bugs (and modules are easier to rewrite if something gets too big). Some things (like many of the fileutils) will probably not be updated once in a blue moon except for bugfixes - they are _stable_!
3. Complete ignorance of the GPL. You can't embrace and extend Linux because you can't "steal" the open source to use it in a proprietary project. The most billyboy can do is emulate the Linux API, which would give more credit to Linux than he's willing to grudge. Funny, I see GNU in that article but not GPL.
4. Apache vs IIS - This depends so much on benchmarking conditions that I'm not even going to get into it. I think there's an article in Dr. Dobb's Journal in the past year or so on benchmarking idiocies. Read it and others like it.
2. Someone else already mentioned this, but his use of percent reliability rates is totally fucked up. If the ratio of LOC to bugs (= defect density) stays the same, the percent stays the same. See #1 above here - why should Linux get that much bigger? It's also more modular than Win, which helps in getting specific people to fix specific bugs (and modules are easier to rewrite if something gets too big). Some things (like many of the fileutils) will probably not be updated once in a blue moon except for bugfixes - they are _stable_!
I WILL ELABORATE ON THIS.
Assuming that the frequency of encountering a bug increases with size of code given constant defect density is not 100% incorrect but is still a fallacy. What the article assumes is that each line of code will be hit for any configuration no matter the size of the codebase. As the codebase grows, it's often feature creep, or for an OPERATING SYSTEM, IT IS OFTEN ADDITION OF NEW DRIVERS, only a few of which will be in use at any given time. The reliability rate therefore WILL NOT fall as fast as the codebase increases given constant defect density, and it MAY and SHOULD increase (ooh, now i'm starting to write like an RFC) as the defect density falls due to bugfixes.
Er... I didn't pay to read it. The convention here is to use cypherpunks/cypherpunks, so far it's worked for me every time.
Corel is actively contributing to the WINE project. Not every company is as open-minded, but if this effort on Corel's part succeedes (for both
Corel and WINE), then others are bound to take notice. A forward looking company only needs to realize that by contributing to an open source project they are helping generate a product or tool they need at much less of a cost then doing it from scratch themselves.
Recall the basic Model-View-Controller pattern.
This allows the creation of non-monolithic, non-bloatware applications by separating out these functions. Heck, the existing GUI front-ends to a number of Unix command-line programs shows this.
Unfortunately there's a lot of code and apps out there that falls into the GUI bloatware trap, because it's easier to grow such apps from a small start without a lot of upfront planning, which is probably the model most new OSS development follows.
It doesn't have to be that way, and the apps that survive long term will use a more compartmentalized model. The advantage to allowing different interfaces to plug into a given apps funtionality are obvious.
- Al (who is currently working on a generic framework for just such an app model.)
This guy seems to be spouting unsubstantiated personal opinions right and left as if they were facts. Not very convincing, especially coming from someone who states that Linux has no video card support! Let's look at some of his points:
1) "Limited supply of competent developers." This assumes that open source does not create new competent developers. The best way to learn how to code is by reading code produced by other, more competent developers! Thus open source itself creates more competency in the industry. The other fallacy in this statement is that it assumes Linux's competition does not suffer from the same problem! Once the majority of competent developers are working on OSS, it becomes impossible for any competing software to improve at a greater rate!
2) "Linux will become more less reliable as it expands exponentially." This is partially true, systems do tend towards greater complexity. However, Linux is derived from the Unix philosophy of "simple is good!" and "modular is good!", thus it can NEVER become as complex or bloated as most comercial software, especially NT5, the epitomy of a monolithic architecture. What will probably happen is that many enhancements to Linux will be available as optional "plugins" (much like the model Mozilla is moving towards.) Dynamically loadable kernel drivers, anyone?
3) "Caldera, Red Hat, or some other company will comercialize Linux." This has already happened. The implicit assumption that an OSS/commercial hybrid is non-viable is pure FUD. GPL-derived commercial software must by it's very nature be more responsive to the customer; open standards give the customer control, not the marketing department. Customers can always switch to another Linux distro with minimal impact on their operations -- much less impact, then say, switching from Win95 to WinNT. (And soo far, you're lucky if half of you're WinNT4 applications work under WinNT5!!!) CHOICE IS GOOD! (Even in commercial software.)
4) "Some magical commercial OS will appear out of nowhere to replace both Linux and Windows" This statement can only be the result of massive crack intake. Any developer can tell you it is an order of magnitude easier to change existing code to do what you want it to do then to create something from scratch. GNU/Linux thus has a tremendous head start on any new competitor. (Or at least any competitor that isn't derived from GPLed code, which wouldn't be competition at all.)
5) "Greed is everybody's greatest motivator" While this may be true of the author (How much were you paid to write this article?) it is the pinnacle of arrogance for him to assume he knows what motivates everybody else. Obviously, monetary greed has NOT been the motivation for the creation of GNU/Linux in the first place! How could the motivation for these people suddenly change?
6) "Less productivity software available." Currently this is still true, but becoming less true every day. Yes, emulator technology like WINE must be available and robust before Linux makes big inroads into the desktop market. But I thought we were talking about SERVERs, for which productivity software is irrelevant!
I could spend all day shooting holes in these poorly conceived arguments. At best, they point out some pitfalls we should be aware of and avoid. At worst, it is complete FUD, possibly even bought and paid for by those boyz from Redmond...
I can't seem to find the URL for YOUR news site.
From the tone of your comments, your site must be
way better. If this is the case, then you are
doing all of us a disservice by not sharing.
cypherpunks
Ted Lewis's article is inherently flawed, and is an obviously constructed Astroturf Public Relations attack on Open Source software that was patently sponsered by Microsoft. His article has no basis in reality. His incessant references to "the weak" in describing Open Source adherants is simply an attempt to utilize psychological warfare techniques to plant the seeds of contempt for Open Source in the weak minds of his target audience - this is transparent. Refrences to M$ as "the strong" is simply the counterpoint to the previous refrences to Open Source software - this is also transparent.
Describing M$'s criminal behavior, in attempting to crust Netscape, as a skillful application of the [steal]-and-extend strategy is unconsionable. This was criminal behavior, the extent of which is only now coming to light. Creating an organized criminal conspiracy to crush your competitiors is not evidence of competent management, or a strong company, it is evidence of a management team that is grossly incompetent, and a company with the complete inability to produce superior products - M$ is the poster company for this category. In 1998, Linux obtained a 5% market share, this will only increase. Remember, we're at the base of a geometrically progressing adoption curve. Without an illegal criminal conspiracy preventing rapid Linux deployment as an OEM operating system, M$ would lose a further 20% of the market in 1999. Even by deploying their most criminal and underhanded tactics, M$ will still lose at least 10% of their market to Linux in 1999.
In a strange marketing twist on reality, Mr. Lewis describes Linux as being on "the verge of being squashed by [M$] and will probably not survive in its present form," - this is the reverse of the true situation. Microsoft will not survive 1999 in its present form, partially due to its inability to produce viable products without the ability to illegally leverage its operating systems (as Linux becomes ascendent), and partially due to the division of M$ by the DOJ and the proposed criminal prosecution of the M$ management, under RICO statutes.
Another oddity is Mr. Lewis's statement that Linux defect densities will get worse, when in fact what we see is just the opposite occurring in Linux. But we do see an increase in the defect densities of Microsoft's own products - in fact the rate of increase of defects in M$ products seem, from one release to another, appears to be progressing at a geometricly increasing rate.
Mr. Lewis goes on to laud the M$ core development team of 400 'professional' programmers. So M$ has 400 marginally 'professional' programmers working on Windows (in its entirity). This hardly compares to the 200+ Linux kernel hackers, the 600+ of so GUI hackers working on KDE and Gnome, plus the XFree86 people, the FSF/GNU team, and the tens of thousands of others working on other ancillary functions of Linux. Then there are the full time testers, they number in the millions. In Linux, beta tests go out to millions of programmers and users, as opposed to M$ using a paltry few thousand. Which product would you rather use? The "woefully under supported" Linux, or the comparatively unsupported Windows family of essentially untested products?
Then there are the security issues, which Lewis conveniently ignores. You can't trust M$, they have made this clear by their employment of illegal and predatory business practices. With closed source software, your company's future is completely in the hands of another firm - one that is, at its best, indifferent to your needs. At worst, your firm will be targeted for assimilation, or destruction, as your one time supplier expands into your industry in a desperate bid to maintain an unsustainable level of growth. Allowing another company to have complete control of the mission critical software upon which your firm depends is the height of irresponsibility. Since you are at their mercy, they are in a position to leverage their monopolistic control to extort whatever they desire from your firm. This situation will only become more unbearable in the future.
From a security standpoint, it is unconscionable to utilize software that you are not able to fully analyze. Without access to the source code, this is impossible. The continuous stream of security holes being reported in M$ products should drive this point home. Whether these security problems are simply bugs, or intentionally emplaced espionage enabling features is irrelevant. If you place sensitive information on a computer or network without the ability to insure and confirm that the system is secure, you are negligent in your duties. Legally, this could easily be argued to comprise a failure, by you, to exercise due care in protecting company secrets, and thus expose you to an unacceptable level of personal liability. The security ramifications are even more ominous if your software provider becomes a competitor, one with first hand technical knowledge of your systems - and the technical proficiency to surreptitiously breech these security measures with ease. Since M$ solely uses the steal-and-extend strategy, placing your intellectual assets on a M$ server is simply a form of corporate suicide.
Microsoft does not produce commercially viable products, Linux does. The choice is clear.
YES!
Clearly, Mr. Lewis' experience is in marketing, not software development. Most of his criticism of OSS is based on the fundamental assumption that drawing parallels between Microsoft software development and OSS software development is valid... this certainly hasn't been true to date. Mr. Lewis suspiciously seems to know a LOT about M$ software development, and applies what appears to be valid criticisms of M$ development to Linux, without appreciation of the fundamental differences between the two models. His fundamental assumptions are flawed and his reasoning is flawed. In fact, any schoolkid should be able to shoot holes in his logic, so don't feel so proud that you recognize the obvious inconsistencies. The biggest question that came to my mind upon reading this was this: Was this article paid for by Microsoft or not? (By the way, pointing out a flaw in your own product to gain the reader's trust is one of the tactics described in The 22 Immutable Laws of Marketing, so the fact that this article mentions in passing that NT will collapse under it's own rate does NOT rule out the possibility of this being M$ sponsored FUD.)
Well, I don't know about any of you, but I've gotten junk mail from the IEEE Computer Society a few times and considered joining it. However, after reading that steaming pile of dinosaur dung I simply have no respect for that organization.
I wrote membership@computer.org and informed them of why I would never be joining them, and that their organization has definitely been diminished in my eyes.
> due to the "winner take all" nature of
...
> standards, there can be room for only one
> OS. i'd rather that OS be linux than windows,
> wouldn't you?
Winner take all standards like TCP/IP, SMTP, FTP,
SNMP, NFS, HTTP, XML, DOM, ECMA-262, CORBA, PCI,
ISA, PPP, DHCP, SQL, LDAP,
There will be room for more than one OS just as
there always has been.
share a copy of the online stuff, since I am not a IEEE member and can't get online...
discretion guaranteed
blaise_toad@yahoo.com
Emacs implementations are not text editors. They are lisp environments that boot up an editor by default. I like that kind of flexibility - it's the right idea. If you want to build a pure text editor, you get VI, which I'll never touch other than to type :q .
Here's my take on emacs and bloat. Look at XEmacs. It's open source based on more open source. One of the things slated for the next major release is a rewrite for a modular design. XEmacs will take less space on your drive if decide not to install the optional emailers, newsreaders, web browers, whatever.
That aside, long live bloat!
It seems that nowdays, if your magazine is not well read or is losing readership write an article about Linux. This guy is a nobody.....a troll...enough said
Another problem is that Linux LOC is composed from kernel code and driver code. The logarithmic growth of the Linux kernel size has been primarily due to the explosion in driver support.
On the other hand WinNT drivers are developed independently. How many bugs do you think would be in the combined total of WinNT LOC and all supported driver LOC? I'm willing to bet a lot!
The author didn't take this difference between Linux and WinNT into account. It makes his figures absolutely useless.
Jeld, you're an ass.
Go home.
Blanket statements aside ... It does hold water.
... no developers => no popular apps => no more platform.
... why ?
:-).
I'm a REAL live CS major, and yes I have written code (for money) on all the platforms mentioned. I have used and abused every type of computer I can get my hands on.
Given a choice I will always pick a Unix over say a macintosh or windows box.
If the developers leave (don't actively develop for) your platform doesn't mean anything in the commercal world. Think about this. "Popularity" of a platform depends on what applications are available
I graduate in may, by that time i'll have two years of real world experience to boot.
If I was to start a commercial project in may with my money it would not be targeted at windoze or the mac
Because I'd rather not start something that will just be embraced and extended (read: hijaked) by microsoft, as VERY large companies in that dared to enter microsofts world have had happen to them, I'd be a small company with a snowball's chance in hell of not splattering myself on microsofts window.
Btw: the mac is basically a word proccesor and yes liberal arts types tend to use word processors a lot more than c comilers
Uhm, looks like you have been beaten with the "cypherpunks" clue stick. Now go out into the world of "registration required" sites w/o losing your privacy.
Happy surfing.
You've lost me tiger
'LiNoOx ViRuZ AlErT!! KoDe ReD!!'
Dynamically loadable kernel modules ;)
If you don't want your kernel bloated..
it doesn't have to be
power to insmod and rmmod
ya nigga
eye haq 4 u
The run Klyx
Besides the "letters to the editor" email address (computer@computer.org) here are addresses for various editors including Editor-in-Chief, managing editor, staff editor, and assistant editor-in-cheif at Computer Magazine:
:)
aburgess@computer.org,
j.aylor@computer.org,
d.carver@computer.org,
kkroeker@computer.org,
t.lewis@computer.org
And of course I threw in ted lewis for good measure
What the hooey are you talking about?
This was an excellent reference to an abysmal
article which was important for many of us to
read and might have been missed otherwise.
Please take your trolls elsewhere.
I have mixed feelings about, for example, emacs. I actually really like Xemacs-- but all of what I like should be available in a very lightweight editor. Problem is, I already know and love "( C-x C-f ) save it ( C-x C-s ) and exit ( C-x C-c )" and every chance I give myself to learn vi goes wrong. I just don't like vi. At the same time, everyday when I first start xemacs, I get angry-- it takes soooo long to load and it takes sooo much memory. Maybe a gtkemacs with a nice, small gecko renderer would be a blessing. Plus, the emacs team I was on lost the "editor wars" at the Linux expo in Durham to the vi team. We have to do something in the vi-direction.
And I agree wholeheatedly with the original poster's remark
> GUI applications are a whole 'nother tarpit. Basically, I don't believe Open Source will produce polished GUI applications; instead, you'll get more retrofitted bloatware (GtkEmacs, anyone?) and lots of pointless cloning (57 different IRC clients, anyone?).
This is what's happening and I frankly think we are wasting a lot of time on crap like ksolitare. It doesn't have to be this way, but it's an easy way to code, I suppose. Rewriting old crap with new widgets.
It might be useful, in this regard, rather than sneering at everything that Apple has ever done, to take a close look at the idea behind AppleEvents. While there are certainly some problems with AppleEvents (for example the fact that so much of the burden aof parsing is placed in the app) the basic idea is very powerful and is exactly what people are asking for here.
I'm no fan of emacs and I can't comment on sendmail but I have to say, calling Perl bloatware stuns me. I may be wrong but I can hardly imagine anyone who has experience with Perl could call it bloated. What's bloated?
- the space on disk? -- a few MBs
- the space in memory? -- generally pretty efficient
- the speed? -- puhlease...
and you call it monolithic?!? Take a look at CPAN...
Chris Kuhi
(not a coward, just lazy)
ok there are many, many different window managers i'll give you that one but how many of 'em are actually the same?(fvwm variants like 95 don't count) the reason there are so many different ones is because there are so many different things that ppl need in a wm for example... i'm not big into customizing my wm like in Enlightenment or scwm... nor do I need plugins like in FVWM... nor do I need a wharf like in AS/WM... I like WindowMaker alot and I think its great along with Enlightenment and AS and many others(I personally dislike fv just my personal opinion) but they often have features i don't need or don't want... the one that I like best is Blackbox... blackbox fits my needs perfectly I don't need pretty icons or anything like that... all I need is something that gives me virtual desktops and a rootmenu and thats what blackbox gives me... other ppl may want complete control over how their desktop looks and others may want a wm that looks like a macintosh or even win95... its all up to them... that is why there are so many window managers... there are so many different needs and tastes...
I don't understand why Open Source software is so different that normal software that would make it fail the acid test. So what if the source code is free? Consumers will still buy the product. There's more to a software product than the intellectual property, there's packaging, manuals, support, and name recognition. The GPL even says that your software doesn't have to be gratis. If I develop software under the GPL, sell the binaries on CD with manual for $500 with a 90-day support plan, and then sell the source for $1000 with extensive developer support, it's still following the licence. Of course, whoever buys the binaries and source can redistribute it. But who are companies going to trust--a noname redistributor or the actual developer of the product? It's all about name recognition.
Sendmail I can see. But you haven't lived until you try to configure sendmail, and I'm sure we all know that. But Emacs and Perl? Bloated? I think not. I've never had emacs behave in a slow, buggy or crashy manner which would suggest bloatedness. Yes there are a lot of features I don't use, but how on earth does that make emacs bloated? Hell, it doesn't even *LOAD* alot of the features I'm not using. Just because something is extensible, it's bloated?
Finally, if you're calling Perl bloated. You obviously have never used it for more than half an hour. Perl has got to be the best programming for rapid prototyping. Things that would take hours to program in any other language, takes maybe half the time using Perl.
Frankly, you've got a lot to learn before you go criticizing other peoples software.
as well as the currently snowballing way of things, has already proven him sadly mistaken.
End of story.
Zoloft
The trick is to preview your post, then go back to the original edit screen and clean up any errors. You can't do 'edit' => 'preview' => 'submit'.
What part of "gestalt" don't you understand?
You want a good Windows Text editor, try Vim. :)
:)
It's also a good Unix/Be/etc... editor that everyone should use. Of course this is all IMHO.
I read the internet for the articles.
I'm currently enrolled at UNLV and I'm having a tough time finding anyone else that uses Linux... *sheesh*
So convert your friends :) Since I came here as a freshman last semester, I have seen something like nine new Linux installations just in my dorm building, and my roommate, who works for university software support, says that they are gearing up to formally support Linux along with Windows and MacOS :)
I'm at the University of Massachusetts at Amherst, by the way.
I use Emacs for all my programming and don't even know the simple key commands you mentioned here. When running in X, it's much (much, much) friendlier.
I'd hardly call the text editor "archaic". Rather, I find it (with appropriate modes loaded and font-lock enabled) far better than the MSVC++ interface I used to work with. Please be specific in your criticisms of the editor proper. I can't defend the mail/news, calander and the like as I don't use them.
As for Lisp... . All I've needed to do so far is just read others' code and make very minor changes (setting variables and the like). I don't need to _know_ lisp to do these things (though I will be needing it for an AI class next year).
...aren't that hard to use, particularly with RPMs. If you're checking RedHat-compiled packages, you need only one key.
Posted by The Orge Captain:
As I understand the industry at this time I find the commnets made in this article to be fairly uneducated. I bieleve there is a fad going on but it is not the open source movement I bieleve that this will be a trend in the future as software gets closer to the user being more custom for its intended purpose rather then one size fits all of today. The fad I see is people who place linux on there desktop machines and with some of the members of my age group who install Linux thinking that they will become some sort of hacker by installing it. Linux is already to complicated for the average user to run because it was never designed to be run by the average user the people who were constructing it work constructing it for there own use and so there was little attempt to make things not necessarly easy, but straight forward there. There is a tremendous amount of upkeep involved in keeping any unix system runing paticularyly in securing it (Unlike Microsoft the unix community actually fixes vulnerbilitys in a timely fashion). I see linux as NTs nemesis, educated users running the operating system on higher preformace workstations and departmental servers. (What I bileve is needed in the corporate enviroment is a less complicated operating system that runs only what corporate destop users need a browser, word processing etc, enterprise ware. This would reduce the amout of problems that are created by windows in a corporate enviroment (horendous up keep and abuse) Linux is and excellent operating system and Im glad that I have it.
I am not a professional programmer, so I cannot tell if this guys arguments against perl are valid or not, although it seems to me that OO elements in perl 5.x are completely unneeded by a scripting language and perl is not useful for anything but scripting and small to medium CGIs.
There's debate over the usefulness about OO in perl - some people love it, some don't - but Perl doesn't force any computing paradigm in particular on you. You don't have to use the OO stuff if you don't want to.
The notion that perl is only useful for small-to-medium CGIs is ludicrous, however. I use Perl to manage my life. I design and maintain a website with hundreds (and soon to be thousands) of pages, all built in three different versions (text-only, html 3.2, html 4.0/css) that would be impossible to do without the Perl preprocessing compiler I use (and wrote). I also use ePerl heavily on the server side to interface with the mySQL databases.
I've been programming professionally for 15 years (and hacking for 20), and I was raised in the trenches of assembler and C. Perl is unlike anything I've even used. It's incredibly powerful, much faster than any high-level language has any right to be, and it can do just about anything. You might not want to write Photoshop in Perl (or you might), but its creators call it a "swiss-army chainsaw" for very good reasons. The hundreds of libraries of free, tested code add immense functionality without bloating the language. "Swiss army tactical nuke" might be a better moniker.
I can't imagine doing the work I do without Perl.
My two bits.
If you actually analyze carefully Ted Lewis' article -- an exercise I don't recommend -- you'll see that none of his statistical arguments hold water -- even in terms of the mathematics, let alone their factual premises.
That is, where it isn't simply incoherent. Consider, for example, the following:
What does this sidebar actually say [p. 125]? That Linux has:- Multithreaded kernel
- SMP
- Multiple FS support
- Disk striping and mirroring
- Multiprotocol networking
- X11 GUI
- GNU tools
- all the internet tools
Linux lacks:- Video card support
- Wireless LAN support
- "A good selection" of productivity tools.
OK, in other words everything needed for a flexible and high-powered server OS is right there. Notice how lame the "lack" list is, and that the same lacks also apply to the other Unixes -- even granting that there are widely-used video cards that XFree doesn't support (questionable, and how important is this in a server, anyway?), that X.25 doesn't count as a 'wireless LAN' (all the IT managers out there who depend on wireless LANs please raise your hands), and that "a good selection" requires more than two complete suites (StarOffice and Applix).Thus adding additional video card support to XFree, additional kernel drivers for wireless LANs, and two or three more office suites will obviously bloat the Linux kernel, while SMP, multiple file systems and net protocols, and RAID support did not. Sure. Right.
This article would be more at home in a collection of postmodernist academic essays (or perhaps press releases from government bureaucrats) than in a technical journal; it's a wonderful example of content-free writing.
Craig
Look at all the contributors to the kernel mailing list on such topics as memory management and file systems (are these "CORE mechanics" enough for you?).
Everything is fascinating to somebody; Alan Cox got hooked originally because he became fascinated by, of all things, ethernet cards.
And believe me, if there's a fundamental bug in the memory management or your ethernet card driver, whether these things interest you or not you'll notice it, and probably report it in a whiny and offended tone!
Craig
Yep, damn right I'm greedy. I want great software ASAP. And I know that Microsoft is never going to get me there.
Ooh, a sarcasm detector. Oh, that's a real useful invention.
TedC
If you don't like Emacs, use vi (I like pico).
:)
If you don't like sendmail, use qmail.
It isn't like you don't have any choices, unless you install windows. (let me know when windows comes with multiple, independent editors and e-mail servers, and everything else. Then let me know when you find a good *text* editor for it, and a good e-mail server, etc, etc. Then let me know if you had to port UNIX to do it.
As for the article, it was completely wrong. Testing programs for errors has nothing to do with how long their source code is. Seeing as how Linux is currently *the* *dominant* *unix* in the market, I don't see how the other players could do any better. Linux may not scale quite as well on other people's specific enterprise hardware configurations, but it can't do worse than NT. Therefore, Microsoft also fails the acid test.
What was the test again?
pb Reply or e-mail; don't vaguely moderate.
As for the nuts and bolts, many of the best open source programmers work on those aspects. Alan Cox maintains vast swathes of the low levels of the Linux kernel, and owns responsibility for 2.0 maintenance (yes, 2.0.36 is quite recent, and 2.0.37 pre-patches are coming out periodically). Stephen Tweedie and Ingo Molnar do a lot of work on the VM system. Whatever their fame or lack thereof in the outside world, their peers are well aware of their accomplishments.
And the macro viruses still wouldn't work on the linux side of things because they are dependant on WORD which isn't avaible for Linux, so a macro viruses based on MS WORD wouldn't have the slighest idea of how to acess the Linux filesystem. Also since WINE would actually be handling the actual disk I/O and not a MS program under it, it shouldn't be to difficult to have WINE look for and direct this sort of activity to /dev/null instead of to a hd if it does in fact proves to be a problem.
I suspect you and your friends who work for the MS PR department would like very much for people *NOT* to know about these kinds of articles that appear in mags like this one was published in so they can go uncontested. Unfortunely for you, Slashdot exposes this kind of garbage to the light of day so it can be subjected to the ridicule that it deserves. Don't for one second think you are fooling anyone with your snarky comments concerning Senegan's motives for alerting people to this article. The tricks you and others at Microsoft pulled with OS/2 and other computers like the Amiga and Atari ST isn't going to work with OSS and Linux....
This guy is obviously twisting truth here in pursuit of an agenda. Last I saw someone do testing of Apache vs. IIS (ZDnet, no less), Apache beat it quite handily in the speed department. And, Apache also serves to answer another of his arguments, namely code bloat. Apache handles this nicely through the use of modules (as does perl..) Don't want that piece of bloat? Ok! It doesn't get loaded at runtime. The gimp is a nice illustration of that, also
Re: Solaris vs. Linux code size/defect ratio, last time I installed Debian, it had *lots* more to it. Of course, in typical Free Software fashion, it's all modularized, so you only get what you ask for. And, I've seen solaris boxes get wedged in X somewhat frequently. kfm even brought a Solaris x86 box here to its knees. It was causing things to go in extreme slow motion, wouldn't die with kill -9, and even survived a switch to single user mode... And nfs was *not* involved. I've never seen that type of thing happen with Linux.
You've really got to cut down on the use of NEVER.
Case in point: I did a phone proxy debian installation with a guy who was completely computer-illiterate. Even though I had to supply a special driver for the network card (it wasn't supported in the standard kernels at the time), emailing it to him to be saved on a floppy and transfered to the Linux box wasn't hard at all. And, it worked the first time, and expectedly.
Remember - the world and it situationas are not static. Just because Linux is for computer literates now doesn't mean that it won't be accessible to most everyone in a few years. It's already managed to make itself much more accessible than Unix has typically been.
Another package which is better/farther along
Real life concerns (usually revolving around real job or school workloads)
A project leader who doesn't know how to attract people to his banner, or know how to set reasonable goals
A Boring Project which doesn't capture the interest of the parties involved for the long haul, or ever attract enough people to have completed
In the case of the first, that's just simple software darwinism... either the coder helps them out or simply looks for something else to do. Saves us from wasted effort in duplication.
The second, well, not much can be done about that... ya gotta eat, but I've noticed that a lot of people will go and do that stuff and come back, or at the least hand control over to a competant fellow developer.
The third can be summed up in pretty much one word: Freedows. ;)
The last is the only one that worries me, and has been identified by others as a 'trouble spot' for OSS software development under the so-named bazaar model... people often shy away from the uncool/boring projects for obvious reasons. I personally see this as an excellent opportunity for profit-motivated coders to fill the gaps in with commercial offerings... YAPMFL (Yet Another Profit Margin From Linux).
I question this 'untested update' thing you mention though... if you're looking for code stability, chances are good you've already latched onto a decent distro (rh, deb, suse et al) where a centralized body compiles, tests and packages changes prior to release. I've personally found redhat pretty good about supplying stuff which just works out of the .rpm, and seems quick on the draw to fix buggy things (although it took me 2 days to finally DL the wu.ftpd exploit-proofed RPM... they REALLY need to work on that site mirroring from updates.redhat.com).
--
rickf@transpect.SPAM-B-GONE.net (remove the SPAM-B-GONE bit)
"People will pay big bucks for the luxury of ignorance."
The graph given in the report showing the exponential growth in the kernel code is at odds with the data from: http://ps.cus.umist.ac.uk/~rhw/kernel.versions.htm l . Using the tar.gz files listed therein (granted, it is compressed) I made a graph of the kernels over time. Here's the results in ASCII (my apologies for the size and formatting): Size in MB- --+----++- --+----++
12 ++----+-----------+----------+----------+--------
+ + + + + + +
| A |
10 ++ AA ++
| AAA |
| AAA |
| AAA |
8 ++ AA ++
| AA |
| AAA |
6 ++ AAA ++
| AA AA A |
| AAAAAA AAA |
| AAA |
4 ++ A ++
| AAA |
| AAA |
2 ++ AAA ++
| AAAAAAA |
| AAAAAAA |
+ + AAAAAAAAA + + + +
0 0 ++-AAAAAAAA-------+----------+----------+--------
1992 1993 1994 1995 1996 1997 1998 1999
There is no exponential growth in the archive file sizes. I've got graphics graphs, but no site to host them. Anyone want to contribute web space for my extraction source code and graphics?
Linus didn't type that number, someone else was transcribing his phone conversation to IRC. That latter person misstyped. The source to the Linux kernel is about 1.5MLOC.
However, the article is still wrong because the error test wasn't comparing kernels, but the utilities, and the GNU utilities are not dramatically different in size than the commerical utilities, in terms of KLOC.
SPF support for most open source mail servers can be found at libspf2.
you rock. i think that may be enough. You've got the point. that'll work.
Reality Warriors are hard to find. Anyone with the hunger must suffice. No surrender.
Hang on. We'll help.
Brak: What's THAT?
Thundercleese: A light switch.. of TOTAL DEVASTATION!
Secondly, the shortage in programmers. A distributed development model means that you almost -can't- have the same kind of shortage a company could have.
Thirdly, the modular design of Linux solves complexity issues, design issues and stability issues.
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
..I suppose that proprietary software (for example, Win2000 or Netscape 4.x) is less bloated?
Daniel
Hurry up and jump on the individualist bandwagon!
IIRC, Emacs is just the core LISP environment and editor..all those other things are separate programs that happen to run inside Emacs.
Daniel
Hurry up and jump on the individualist bandwagon!
I sent a response to Computer the day the
issue arrived in the mail. I've already received word that it will be published, if they have space.
I have a copy of my letter on my home machine; maybe I can dig it out and post it.
Well, CORBA works on the level of objects and methods, not programs. It does have the concept of methods returning values.
But the point of the UNIX shell was that programs read and wrote streams of data, which isn't really the case in graphical applications. The closest analogue would be to have one program's window contained in another's, which X of course supports (usually called "swallowing" a window in another). CORBA can be used to negotiate this. This is how BABOON and OpenParts work, AFAIK.
For one, the message you were responding to never said that emacs was good for everyone and everything. It just said that it wasn't bloated. Perhaps it's hard to use, and perhaps you don't like its extension language, but that doesn't have anything to do with it being bloated.
Two, not being a programmer, I'm not surprised that you don't like emacs. I am a professional programmer and it makes my life a hell of a lot easier. I won't get into the details, but it really saves a lot of effort. To be fair, a lot of programmers don't like it either, which is fine for them.
Three, I agree that having to know lisp to configure it is a pain, but this is no longer the case. There's now a pseudo-graphical configuration mode which requires little to no lisp. I still wouldn't consider it a walk in the park, but it is easier than it used to be.
Finally, have you ever used Gnus (the news reader)? It's far from bad. I wouldn't call VM bad either, but it's not outstanding.
That all said, I agree that emacs is a bit archaic. And it would be better if it were modularized somewhat. As much as I love Gnus, it could do to have a real GUI and some machine-native code to speed it up.
And I'm not sure that all US students do learn LISP. I did do work with AI (in college) so I did, but I can't speak for everyone else. Perhaps they should teach LISP or some kind of functional programming language to all students, though, just because it's good to be able to think in that paradigm.
Just in case anyone is still reading this thread... (sorry, my home computer is broken)
Sort of. They could be said read streams of events, and they write drawing requests. But this level of interaction is basically useless for most scripting purposes (though there are a few where it's useful). Just to be clear, the events they read are like "this keycode was pressed; the mouse button was pushed down at coordinate x,y; the cursor entered your window". The drawing requests are like "draw a line from x1,y1 to x2,y2". Also, the event streams are targeted at windows, not programs (where windows include both top-level windows and inidividual widgets). Piping these together would be basically useless except for things like UI testing.
If you want to talk about high-level events, that's where CORBA comes in.
What I find interesting is that there really have been no revolutionary improvements in complexity management in all this time. There have been incremental improvements, such as software tools like more advanced languages to automate some of the most menial work of programming or revision control systems that provide improved concurrency control and communication about software changes, and somewhat improved design methodologies like object-oriented design, but despite some grandiose claims nothing has really solved the problem that designing and building complex software is hard.
Open Source software is not immune to this problem. What source code availability and open development processes have done is introduce other incremental improvements, mainly increasing the available labor for implementation and debugging. Successful open-source projects tend to remain focused on fairly specific goals (usually development of specific applications) and have efficient centralized management. These usually translate into advantages over commercial software development projects working on projects of the same scale because open development tends to distribute work over more people and more time than commercial developers can afford.
Lewis doesn't seem to acknowledge that commercial software is equally susceptible to the complexity problem, or that the smaller number of participants in most commercial projects put them at a disadvantage for doing the grunt work of implementation and debugging. His assertion that open source software is somehow going to be unable to obtain wide market share is, I think, based more on cultural and psychological factors such as the pervasive belief that you're getting more if you pay for something than if you get it for free rather than on any inherent weaknesses of open source software relative to commercial software.
However, I do think it's important for Open Source software advocates to realize that it is just as susceptible to the complexity problem, and that distributed development is only an incremental rather than a revolutionary improvement in programming technology -- a brute-force solution that uses greater amounts of conventional effort to attack the problem, rather than a fundamental change in approach that would allow similarly complex projects to be done with less effort.
There are a lot of other flaws in Lewis's article, such as the assumption that Open Source software is about market share rather than people solving problems, that large commercial enterprises are "strong" and open development is "weak", or the various factual errors in his characterization of Linux and other open source software, but the biggest flaw to me is his apparent assumption that commercial development is capable of solving problems that open development is not even though both use essentially the same software engineering methods.
Bravo. Spoken like a true prodigy
Who would like to volenteer supporting a 200 user office moving from M$ to Linux ?
Bain
Sanity is a majority vote.
For one thing this article only looks at the 2.0 Linux kernel, and another it does not do any comparisons to number of programmers vs. defect density or try to figure out the defect density of the various microsoft operating systems.
Also if a high-end *nix can have 60% defects and still remain on par with linuxs defect density then what does that say about bloat? Considering linux has virtually all of the functionality of the high-end *nix at the core level.
As for the arguments on defect density and bloat increasing with linux that *could* be true if everyone who decided to code opensource apps or work on them did not have the unix mindset of making one thing do what it's supposed to and do it well, and not integrate everything into it.
As long as there is a will to do something, something will be done.
Here we have a magazine with such a clue, they distribute their articles in PDF format (making it harder to read and limiting their ability to make money from banner ads) and require people to log in to read their stuff ("go away, new customers! we don't want you around!").
And we have their "think tank" consultant from the "Technology Assessment Group" (translation: we can't do it, so we slam everyone who can).
Even as a columnist, he seems to be lacking on the competence curve. With all the bad spelling ("Torvold"??), erroneous facts (best video support outside Win9x, and a preferred platform for new advances in wireless LAN technology), distortions (the "defect density" BS), bad math (Linux has three times the developers per line of code that NT has, even given his questionable stats, and Linux is undersupported?), and outright lies (Apache "losing the performance battle to IIS"), it's hard to imagine that he even did any research, much less deliver any thoughtful conclusions. The majority of the article is a dissertation on his own theories of how the software market works (all delivered without a single cite or bibliographic reference); thus, when Linux violates some of his "basic principles", it's no wonder that it comes up short in his "analysis".
If it weren't for the fact that this article isn't likely to be thrown on my desk as Exhibit A why someone shouldn't use Linux, I wouldn't even waste my time on it.
Anyone have a copy in HTML or something... I hate PDF.
This guy is rediculous. He talks about Linux's reliability being suspect? I don't care what his reasons or arguments are. The proof is in the pudding. Linux development/quality is not slowing down, and it's not going to. OTOH, I can think of one megacorp who's software is gonna be way late :)
--
-- Knowledge shared is power lost. -- Aleister Crowley
I really hate to use language like this but it's all BS.
This is the most monumental pile of fud I have read in years. So let me relax and tear his claims apart.
1 : Linux is compared to Unix when Reliability is tested. Not Windows NT. This is significant because the only selling point of NT in Big offices is that it's cheaper to deploy and therefore lower quality is acceptable for none critical costs. As long as Linux is Better than NT it's good enough to crush it. Solaris and Aix can stay where they are for now.
2 : All his Embrace and Extend scenarios assume that the GPL doesn't exist. Simply put anyone who sells Linux must keep it open. End of story.
3 : He compared Linux size to that of other unixs and NT based on the number of Lines of code. Sure Linux is small by that measure. Now put it against SCO based on features included in the Kernel. Suddenly Linux is the lumbering giant.
4 : Apache's once large dev team has shrunk to less than 20 because the hype has died down, or maybe it's because the developers have all the features they want so only a dozen good men are needed to fix security breaches and tweak performance.
5 : IIS matches apache in performance. This is the biggest BS of the lot. It's also why I like the idea that Apps based on QT won't be ported to Windows. Apache may be slower than IIS under light loads when running on NT with a large computer that's bearly stressed. Compare Apache on Linux to IIS on NT when both are on identical hardware. Huge difference.
6 : Linux is compared to Unix on market share based only on the $$$ spent. Linux is free to cheap, therefore The fact that it's equal to all the other unixs combined in volume means nothing to this "reporter".
7 : He talks about management difficulties and how Linux development has slowed down since the 2.0 release. In the same section he states how the management problems are being effectively dealt with. I.e. Delegate.
8 : He compares Linux dev team to NT Dev team based on numbers of programers and testers. Linux has 200 full time while NT has 400 full time. Linux has 1000 Par time, NT has 0 part time. NT has 60,000 Beta testers max and 250 internal testers who can peak at the source code. Linux hase over 100,000 testers who peak at the source code. Including me.
Too bad that didn't make it in.
This is the most beautiful fud fest I have sean in a long time and yes. It's BS but it's beautiful BS. We are in the 3rd stage of Ms Ghandi's 4 stage struggle. They are fighting us now. Next year we win.
--= Isn't it surprising how badly I spell ?
The vast majority of the OSS/Linux talking heads seem to neglect the INTENSE stratification that exists in the computer business. I think Microsoft downplayed this very factor when they decided to feed NT to the common user, and are now paying very heavily for it . . . what's this there's going to be another consumer-grade OS based on DOS . . why? because NT is not compatible with cutting edge gaming at its core and will never be. I sure hope that the moron that came up with that strategy has been ripped a new one by his boss.
I know very little about the requirements of the server market, but I suspect that if reliability and stability is a strong factor then this great rise in NT usage might start to fall away with the advent of W2k, that is unless W2k is initially released in a very stable form heh heh heh. The next viable solution is a Unix-type platform, and it seems to me that this is where Linux will start to make some serious inroads against NT & the bix Unixes.
Then there's the consumer/business market. I initially clump them together because both of these major markets tend to use similar hardware, whereas the server market uses distinctly more advanced hardware. On the business end, Windows is pretty strong because it's fast. What I mean is that you can order a hundred or more PCs with Windows pre-installed and with a minimum of effort, all these PCs can be hooked up to a central network, and employeeds can be plunking numbers into spreadsheets and emailing jokes within a very short time. While Linux might offer advantages in maintance, configurability, and performance, it will be hard to convince the business folks to go for it because it doesn't come preinstalled (although this might be changing), and when something doesn't work right there isn't a centralized someone to call and bitch at. Forget that Microsoft charges for support, if this is a business that isn't a real concern. Add to this the additional cost of training virtually all the employees to use this OS that most have never seen and some have never heard of.
As far as the consumer market goes, there exist many sublevels. At the top (money-wise) there are the gamers. These are the people who plunk down many thousands of dollars on a predictably regular basis just so they can prance around as cyber-warriors (I have nothing against gamers, but sometimes I wonder if the tons of money might not be better spent). I think this is where Microsoft fumbled badly. The gamers are constantly looking for ways to get their machines to run faster, and with the talk of QuakeIII being multithreaded so it can use multiple processors, and the growing realization that for 32-bit programs that consume a lot of memory, NT offers a large performance advantage over 95/98, it should be obvious that Microsoft saw a way to feed an OS to a segment of people that are willing to spend a lot of money on a very regular basis. Unfortunately the very things that make NT run better on high-end machines also make it a nightmare for games that require direct-hardware access. That Microsoft thought (and possibly still think) that they could have their cake and eat it too still amazes me. But I have digressed. This is another market that Linux will not have an influence in. The emmergence of cutting-edge hardware along with their accompanying cutting-edge drivers (for 95/98) happens at such a rapid pace that there's no way Linux could keep up. Couple this with an already severe lack of game support (there are a lot of mainstream games other than Quake and Heretic), and the stage is set for 95/98 to rule this market segment.
Now for the consumers that are knowledgeable about their machines, and are eager to learn more. Generally these tend to be students at all levels, aspiring young programmers that are trying to learn as much as possible as quick as possible and don't have to worry about things like a 9-to-5 job. Here is where Linux already lives, and is probably growing rapidly. Personally, this is where I fall. I've only been messing with computers for a couple of years now, but I've already worn out Windows 95 and just the other day I hosed my NT installation so that I might be "all Linux". I maintain a small 98 installation just for games. I think this segment of the consumer market is approximately the same size as the gamer segment, and possibly just a little larger. The only problem is that within this segment are also the people who use computers for what they are intended, and hence they don't need the latest new hardware so their dollars tend to stay close to home instead of being scattered into the market. While this is a good personal philosophy, it really sucks for the market segment, as it lessens the importance of the segment. Linux also amplifies this sentiment. I have two machines at my house, one is a dual PPro that is my primary machine (and even though it is considered antiquated by many, it is still blazingly fast for everything that I ever need to do), and the other is a 486/66. When I was given the 486 it was slogging away with win95, and I was very impressed at the performance increase that was gained just by changing the OS to Linux. So the people who never felt like they needed a huge machine to begin with now will have yet another reason not to spend their money.
Then there is the last and largest subgroup of the consumer market. I call this group the email-drones. The people who belong to this group are the vast multitudes that are buying PCs because Betty down the street has one and if Jenny gets one then she can send email to Betty who can then reply to her email, all without picking up the telephone. And then there's the chat and the ICQ and before you know it the beauty salon of the 50s has moved into the den and onto the computer screen. This market group is single reason for the existance of both the iMac and the $500 PC. Everyone was shocked when the iMac didn't come with a floppy drive . . well, when I was mentioning this in passing conversation to my mother, she said right away "What do you need one of those for?". I rest my case. I'm also not trying to be in the least sexist by seemingly populating this large market segment with stereotypical images of the "gossiping housewife". I use this image as a behavior descriptor only, as it applies to an equal number of males as to females. Anyway, back to the point, here is yet another market where Linux will never make any serious inroads. My parents are still using an antiquated P75 machine, and everytime my father grumbles about things taking a long time to load and/or programs crashing I think about setting up a Linux installation on their machine, as I know Linux could easily fulfill every one of their computing needs and it would do it faster and better. But I already get too many calls from them whenever something happens on their machine that they don't understand and I shudder to think about trying to have them add a directory to their PATH or some such over the phone. Once again, Windows wins because it is easy, not because it is the best.
So what does this have to do with the article? Well, I think that Linux will continue to grow in both market size and stature, and that it will become a real competitor to Microsoft in many areas, especially in the high and low end. But unfortunately the real meat is in the upper and lower middle of the market, and I suspect that the ability for Linux to penetrate this area is slim to none, but not for any of the reasons described in the article, rather for the simple reason that the demands of those market segments do not mesh with the attributes and philosophy of the Linux OS. Personally I think this is a good thing. There's a reason I have a 98 installation, it plays Age of Empires REAL well, and it gives me a place to go to "relax" at the computer. I like having the "play" separated from the "work" in this manner, and while I remain a strong admirer and advocate of the Linux OS, I think the first question that must always be asked is "What do you plan on doing with your PC?". And I'm glad that Linux is not the solution to the reply "Umm . . I don't know . . games and stuff . . "
Sean
RFC2119
Personally, I've always found the core mechanics the most fascinating, (sometimes to the exclusion of all else). The nuts and bolts are what makes Linux powerful and elegant.
As far as fame goes, Linus receives the most by dealing with core kernel code, the core of the core as it were. Alan Cox, working on core-kernel and driver code (mainly), currently receives second billing. RMS's contributions to make, glibc, (long list) gets far less attention than deserved. And for the developers working on the "fancy" projects such as X11, XFree86, Gnome, KDE, StarOffice and Gimp, great as their contribution is, there is far less name-recognition.
/* MAGIC THEATRE
ENTRANCE NOT FOR EVERYBODY
MADMEN ONLY */
The author states that code bloat and featurism will become a problem in the long term. Well, free softeare has been arounf for a long time and we've seen no real evidence of it yet. Most free software projects go through an early phase of rapid development and then settle down to a quiet middle age. Remember how often GCC and Emacs were released in the early 90's? Now minor releases are made at about 1 year intervals.
What really happens when a project matures is that the original programmers strike off into new areas (like GUI desktops) and leave their previous work to those interested in maintaining it.
I'd be more convinced if he had given some (ANY!) examples of popular open-source software floundering under its own success. Until he can do that, his whole rant is mere speculation.
BTW, what did he mean by "Linux going commercial?" at the end of the article? It's GPL, ace.
This article was so full of factual errors and misconceptions that I think the author didn't do any research at all, and is completely ignorant of the topic he writes about.
The business case for Linux and its open source software cousins is based on two fundamentals: the software is free or nearly so, and the availability of the source code attracts a large number of debuggers - again for free or nearly so.
No, the business case for Linux and OSS is based on these two fundamentals: it performs better than its proprietary counterparts, and it fails less often. The author goes to great lengths to describe how total cost of ownership is a key factor in buying decisions, yet fails to realize that TCO is influenced most by unit failure rate, and studies have shown that unit failure rate for Linux is far less then for Windows. Thus, TCO should be far less for Linux simply because you don't need support as often.
This is a simple math bug. The failure rate for Linux is 9 percent. If the size of Linux were ten times its current size, the failure rate would probably still be 9 percent. That's what percents are - ratios. His "defect density" sounds to me like exactly the statistics he is refuting (thus, the defect rate of Linux is 9%, compared to 15-43% for other Unix), since density is a ratio. 60% of 10 million sounds like a lot more than 9% of 1.5 million.
"Linux, once again, has had over 1,000 people submit patches, bug fixes, etc. and has had over 200 individuals directly contribute code to the kernel. ... Microsoft's core development team consists of 400 full-time program-mers and 250 testers... When compared to the size of the Windows NT effort, Linux is woefully undersupported.
Microsoft has only two times the core programmers, but they have 6 times as many lines of code to work on, and Linux is the one that's undersupported? Sounds like Linux has three times as many programmers per line of code than Microsoft.
Already, Apache is losing the performance battle against Microsoft's IIS.
Didn't he see the recent article by ZDNet showing that Apache outperformed IIS by a wide margin?
Even the opensource community admits to this weak-ness: "The biggest roadblock for open source software projects is dealing with exponential growth and management costs as a project is scaled up in terms of innovation and size" (http://www.opensource.org/halloween1.html).
He quotes a Microsoft source saying Linux is in trouble, then attributes it to the Linux community. Of course Microsoft is going to put Linux down. When taken in the correct context with the correct attributions, this statement loses his purpose.
So we can rule out any scenario in which Microsoft takes over Linux.
The GPL would prevent this anyway. Even if Microsoft tried to "absorb and extend", they'd never be able to do it without violating copyright laws, because any extending they did would have to be released as soon as they shipped it to any customer. Even if they try to tie applications to "their" Linux, any work they'd done to make it "their" Linux would have to be released under the GPL, and could be replicated in other Linux variants.
A third scenario is most likely: Linux will turn commercial
Linux is commercial. What do you think Cygnus, Red Hat, Caldera, SuSE, and Pacific HiTech are doing with it? They certainly aren't charities - they're out for profit.
To make a lasting impression, software developers still must cope with absorb-and-extend and other techniques of the strong. To do so, they will have to retain a certain amount of proprietary code in their products and charge for their intellectual property, one way or the other.
First off, as stated before, absorb and extend just won't work with GPL'd code. It can't. Any work Microsoft puts into enhancing Linux (sans applications, which are effectively independent of Linux itself) becomes part of Linux, which only improves it for everyone else too. Any support Microsoft puts into Linux would only help the Linux community. Second, Linux's "intellectual property" isn't the code, it's the people behind it. Companies like Cygnus, Red Hat, and Caldera make lots of money off OSS by having the right people and services, not by having proprietary code. Microsoft can't duplicate that without a significant investment in people and time.
Grrrr... I don't know why a hatchet job by some clueless economist should make me angry, but it does. My only consolation is, tomorrow I'll have forgotten his article, but he'll still be afraid of Open Source.
fish and pipes
who? guys who enjoy such a work
hany
Or, more accurately, perhaps hasn't yet passed this acid test. It can only be said to have failed that test if, after some "reasonable" period of time, it hasn't passed it; otherwise, simply by discussing a product early enough in its life cycle, one could dismiss almost any new product as "failing this acid test".
Perhaps it will fail the test, and perhaps it won't. His article should be treated as a prediction to be tested against future reality, not as a firm description of what will happen. (A phrase that's more and more in my mind these days is "Just because somebody says something, that doesn't necessarily mean it's true.")
"... The concept of free software is a well-known and frequently practised strategy of the weak"
He then draws a parallel between AT&T's licensing of SystemV with Linux's openness. AT&T's strategy was a market-driven, commercial venture, Linux's openness isn't.
"When HP and Sun acknowledge Linux as a viable challenger, we will see its rapid deconstruction as these competitors first embrace and the extend its advantages. Linux is on the verge of being squashed be the strong and will probably not survive in its present form."
Here's the question of code forking again and I must admit that it's worried me before. Thankfully, the open licenses under which the Open Projects operated do not permit it, as Linus said in the MSNBC chat.
About the failure rate of Unix utilities: "... it should not be confused with defect density, a more reliable metric... Unix has more than 10 million lines of code, while Linux has only 1.5 million." His conclusion is that Linux's defect density (lines_of_code/failures) is much higher.
Wrong! The correct ratio is functionality/failures. If a product does what it should do in less lines of code, that's actually a benefit because that's the mark of better, cleaner code that's much easier to debug.
"Thus, Linux's reliability is suspect. In fact, we can expect Linux defect densities to get worse..."
The point is that I've run Linux at home for the last 7 years at home on four different computers and at work for the last 5 years on many, many different computers and in that time have had four unscheduled crashes. Yup, can count them on ONE hand. I have also used Windows for the last two months and since have had more than 20 crashes and have had to re-install 3 times.
Linux gives me more functionality than Windows, for a better price and keeps my personal data (the most important part of my computer) safe.
He later goes on to compare the number of full-time programmers at work on Linux and Microsoft, and the number of beta-testers pointing out that Microsoft has so much more.
Well that doesn't reflect very well on them, now, does it? Considering that they're challenged and therefore but on the same footing as the Open people.
"... Support diminishes still further when the hype wears off an open source application... Mozilla's mailing list declined by 58 percent... Apache..was developed by a cast of thousands is now supported by fewer than 20 core members. Already, Apache is losing the performance battle against Microsoft's IIS."
Mozilla's mailing list: Expected. That's called stabilization. Apache's development team: That's all that's needed, baby! Apache's performance: One wonders where he gets his numbers.
Page 2.
"... Unfortunately, these exceptionally talented programmers (Open source) are in limited supply and, as any open source program becomes widely distributed, this talent will become increasingly scarce."
Limited supply? As far as I known the human race hasn't become infertile, all of a sudden. Anybody can become an Open Source Programmer, in fact, zillions of kids now in school are getting in on the act. That's your supply.
And this is where I quit. The next bits made me angry enough that the only thing I could reply was "F**k off and die, you idiot!" I guess that qualifies as flaming. Oh well.
DannyC.
cypherpunks cypherpunks cypherpunks cypherpunks cypherpunks cypherpunks cypherpunks cypherpunks cypherpunks cypherpunks cypherpunks cypherpunks cypherpunks cypherpunks cypherpunks cypherpunks cypherpunks cypherpunks cypherpunks cypherpunks cypherpunks cypherpunks cypherpunks cypherpunks cypherpunks
(you get the idea)
I don't think Slashdot posts any articles that haven't been 'cypherpunks' enabled.
**>>BELCH
Measures of defect density are meaningless, for the most part. I should know, because I have worked as a Software Metrics dude as a full time job. Now, I must admit that I haven't read the article, but if he says that Linux has a higher defect density than other Unices, then this can be accounted for quite easily.
Consider one difference between Linux and other unices, namely the definition of a "standard" distribution. There isn't really any such thing in Linux. As a result, we pretty much have an infinite number of systems, since any two programs that interoperate within the system can be classed as a sub-system. Ergo, we have a lot more problems with interoperability. If you want to bump up the defect density for propaganda reasons, you just count as many individual incompatibilities as you want, but treat the "area" as being fixed. So you can basically prove anything in terms of your numbers.
There's another difference here between Linux and other unices. Namely, users are expected to have a bit more common sense when it comes to ironing out the kinks. If there's a problem with one bit of software, they can often leave it aside and work out how to fix it later. They can *still* have a system that works well, and still has a lot more features than the equivalent "other" unices.
Also, compare this with Microsoft's method of producing software. They don't give a damn about defect density. They realise too that it doesn't tell you anything. Instead, what they do is classify bugs according to their impact. Then they trade off testing/bug-fixing so that they only fix the major, high-impact stuff. Then they release what are effectively beta versions and let the customers find out the niggling errors that aren't too serious.
This seems to be a new form of FUD tactic from a pro-microsoft head. Since linux is the enemy, simply pit linux heads against unix heads. It doesn't matter if the issue is irrelevant. It diverts attention from the real issues.
Beware of statisticians: numbers are an easy source of divisiveness.
I saw a great deal of legitimate gripes with this article. One thing that was not mentioned yet was the growth of the internet's partial responsibility for the OSS revolution. I have to wonder if that was a stratigic omition. The net definatly makes it much more feasible for developers to spontainiously coordinate with little to no overhead. That flexibility of the internet, renders figures that do not weigh the recent popularity of ipv4 in fair proportion, slanted and antique in nature.
... yes. Yes, yes, a bit, a bit. She has got a wart.
With the solid statistics just mentioned backing it up, I knew I knew the material in this article had originated from somewhere else. After a little searching I turned up a group who had done similar research and turned up similar results.
That group's transcript...
Did you dress her up like this?
No, no... no
She Turned me into a newt.
A Newt?
I got better
what do you do with witches?
Burn! Burn, burn them up!
And what do you burn apart from witches?
More witches! Wood! So, why do witches burn? [pause]
B--... 'cause they're made of wood...?
Good!
We have found a witch, may we burn her?
What also floats in water?
Bread!
Apples!
Uh, very small rocks!
I'm not a witch, I'm not a witch!!
They dressed me up like this.
No we didn't!
No! No!
And this isn't my nose, it's a false one!
So, how do we tell whether she is made of wood?
Build a bridge out of her!!
A duck! Exactly! So, logically... If she weighs the same as a duck, she's made of wood? And therefore?
A witch! A witch! A witch! A witch!
Right, remove the supports!
[whop] [creak]
A witch! A witch!
It's a fair cop.
Burn her! Burn her!
Novel theory: Modern Man evolved from psychopath
Heh.. tru running Emacs on a 386SX-16 with 8 megs of RAM. Then you'll understand the true nature of Emacs: Eighty Megs And Constantly Swapping.
I think there is a world market for maybe five personal web logs.
I found Mr. Lewis to be full of himself, and this finding was substantiated when I went to the website http://www.friction-free-economy.com that promotes his book. In the section called "about the book", he goes on and on about how science and technology joined economy in the industrial revolution, and how the current powers that be/were didn't understand what was going on. Let me point to a couple of phrases that I found particularly amusing, given his stance on OS's role in the current "revolution":
- "Drucker's Law still applies: the people in the midst of the revolution don't know what is hitting them. And they won't know until after the IPO1 is over. Like passengers in a speeding boat, spectators in the software age know the river's current is swift, but they don't know where the raging falls lie."
Too true. But I you find it rather pathetic to rant about how people failed to understand the industrial revolution, and then commit an article which thoroughly demonstrate that he hasn't properly understood the networking revolution that he claims to be preaching? Mr. Lewis has no idea what's hitting him, and thus serves as a poor guide for others.
- "It may be too fast for royalty [drawing a parallell to the powermongers of the Industrial revolution, here], but the software economy is on its way. It may be a mystery to the establishment, but it is well understood by the Netheads2 in Wired World3 . It may violate the doctrine handed down by classical economists, but it does follow a set of laws. It may be just in time."
A mystery indeed, Lewis, and indeed one that you don't grasp the way you claim. Oh, and by the way, we apologize for violating your doctrine. The final sentence in the previous paragraph points to the next:
- "The late 20th century marks the beginning of the end. Within the next 20 years we will discover the new laws of the software economy. We have early warning signs - Netscape Communications Corp. parleyed 16 months of software development work into an IPO valuation of $3 billion. Companies like Microsoft and Adobe Systems which were unknown a decade ago are now the darlings of the stock market, and the nouveau riche telecommunications industries like 3COM, Cisco Systems, and Bay Networks have turned from small-cap industries into powerhouses of the new century."
Okay, people: This guy is a friggin Ph.D., he holds degrees in Mathematics and Computer Sciences, so let's not write him off as an idiot (even though he claims Microsoft was unknown a decade ago). His angle is just skewed. He has published numerous books, and the one that probably sheds the most light on his skewed vision on computing, is the one from 1976 called "How To Profit From Your Personal Computer".
From this, we may postulate that Mr. Lewis is probably hooked on the income side of computing. Don't get me wrong, though, I have no problem with anybody making money by facilitating my work. However, this places Lewis among the current day's power-mongers (I may be inflating his ego here, but the guy is Chairman of Computer Science at the Naval Postgraduate School in Monterey, Califorina, which may or may not be a big deal - kind of hard for me to know from where I sit), and if we were to allow a parable from his own account of the industrial revolution, he should thus be one of the last to know what hits him in this revolution. According to his recent writings, he appears dead on schedule...
Far from being discouraged by Mr. Lewis futile attempts to discredit the OS movement (some of his points are valid, though, had they not drowned in the rest of his rubbish), we should see this as a confirmation that OS is on the right path. This path will need some adjustment from time to time, and some of the necessary adjustments may stem from people with no better understanding of the cause than that of Mr. Lewis.
Those who don't know their history, are doomed to repeat it...
I don't think Mr. Lewis has a firm grasp on the whole reason Free Software (FS) exists.
:)
:)
First off, Linux and FS in general was not started to squish MS. I still don't think it exists to squish MS. That's not the point. This is a community center that all are welcome in. These community volunteers are not here to build a non-profit organization to put the Embassy Suites out of business. Just a place for people who want a good, down to earth, quality place to go. One that they, too, are free to contribute to. A place that can exist after the founders leave, for whatever reason.
I don't know about the majority here, but I started using linux for a few core reasons.
1) I wanted UNIX. Period. UNIX has power that NT just can't achieve. It's flexible, stable, logical, and versitile.
2) I wanted to learn to code. Having source code to everything is a big plus there. I can peruse the kernel, KDE, Afterstep, sendmail, whatever I want. That makes for some great learning material. Plus, I can reuse this code freely.
3) I am never held captive to a bug again. Without even claiming to be a programmer (I am not) I can say that I have fixed a bug for myself. A NIC driver. Not some glamorous GUI office suite. Just a NIC driver for a card I needed to get working right at that very time. Stores closed, no money, choice between 2 NICs, one broken, and one with a driver that would not compile. Read the error messages from the compiler, fixed what logically seemed to be the problem, and Shazam... a working network.
None of this has anything to do with MS other than you can't do it with MS
The biggest reason I try not to use Windows (of any version) is that I have lost work. Alot of work. Windows freezes, and occasionally takes a partition with it. Not often, but how often does it take?
I do disagree with MS in their business practices, but that still is not the reason I use linux and GNU software. I use it because it's just better. I can get things done. And I can learn, which is very important to me.
Do I care if corporate (insert country here) adopts linux? Only to the extent that I may get the one application I really need that isn't on linux yet, and that hardware developers will start contributing open sourced drivers. I think I'll get the drivers and apps with or without this, so in the end, it really doesn't matter.
As to the remarks about bloat and featurism. I see a checks and balances here. The programmers ARE the users. If RedHat decided to start playing "embrace and extend" tomorrow, what would happen? Well, I think the first thing that would happen is the keepers of the code (us... the users and programmers, by grant of the GPL) would just revert the code. And stop using RedHat. RedHat would sink like a rock, and they must know that. Aside from being able to fix and use software on your own terms, this is the most important aspect of Free Software. Without us, they don't exist. And since anybody can start a company to take their place, using the very package they put out, they hold on by honor and reputation alone. Anything they add that isn't liked by the rest of the FS programming world will be an orphaned bastard child left on the side of the road. Plenty of other distro's to take their current place at the top (they are top in the US, anyway, for the time being.)
Those of us who really want to use linux/*BSD/HURD/etc... are in no danger of losing it. We will still wake up tomorrow and have all the code. Nothing lost, nothing to worry about. And there will always be someone that wants to make it do something THEY need. And we will all benefit. New day, same way.
This is linux. You will NOT be assimilated, but you are always welcome to join in. We're having fun over here
> The Unix error defect could have been as high as
> 60% and still parrelled that of Linux.
That's absurd. Think about. In pratical use, it's how often it fails for you. 60% is unusuable. Linux's bugs are far from making it unusable. While his comparison based on lines of code may make sense to him, it doesn't make any sense to me.
Call it OpenNews if you want or OpenJournalism..
I hate clicking on article to find out i have to pay to read it.. Just say no!
StickBoy
--- "The problem is not that the world is full of fools, it's that lightning isn't being distributed correctly." -- Mar
I fear he may be right about feature creep leading to bugs. Look no further than Microsoft for a perfect example.
I see even classic Slashdot is now pretty much unusable on dial up anymore.
you can look at it either way. The manufacturers need to feel a greater demand for it than they do now. Wouldn't it be easier for the people who know exactly how the hardware works to write a driver for an OS for which they can get the source than the other way around?
I see even classic Slashdot is now pretty much unusable on dial up anymore.
His argument seems to hinge on the fact that there are only so many programmers with the will and talent to produce quality open source software. The problem with this logic is that open source software itself actually CREATES more talented programmers because would be coders and programming students can learn by reading actual CODE! While I don't believe that ALL code should be open source, I do believe the operating system and core components should be, otherwise you end up with microserfs using the advantages that only they as the OS provider have to squash everyone else. An OS owned by the people is the only way to prevent that.
It is clearly true that an expensive system with bloat, high defect density, and a remarkable snobishness towards customer feedback has indeed become quite mainstream, particularly in the operating system market, so this argument is fairly weak from the beginning.
Now, it is not really clear whether Linux will or will not suffer from the kinds of bloat and creeping featuritis that he describes. Speaking of the kernel itself, its creation still lies with one person who reads and re-reads every single patch which is applied to be sure everything is done correctly. This is a distinguishing characteristic of Linux kernel which almost certainly contributes greatly to the tightness of its code, particularly given its large featureset.
As for maintainance, the problem is clearly alot easier with open source software. One has the same recourses as before (complain to the author/vendor) with the added benefit that they can complain to any arbitrary third party and get them to fix the problem, then submit a patch to have it fixed generally if they so choose.
Basically, the trouble with his argument is that it is entirely speculation, without any evidence to support it. It is logical to think that free software could not possibly work conceptually, particularly with proprietary software being so much more profitable, but it is also demonstrable that this is not the case.
In general, it is free software's more efficient nature (compare RedHat's employee count and market penetration to Sun's employee count and market penetration) that will make it work, most likely.
I think the MS page is Slashdotted. :-)
Or maybe their scripting needs debugging...
Microsoft VBScript runtime error '800a01f4'
Variable is undefined: 'URL2Code'
/NTServer/inc/global.inc, line 30
While many of his arguments were flawed as has been covered by other contributors, the fact that he butchered Linus' last name severely discredits the entire article IMO.
;)
Let's face it folks, nobody *knows* what the future will bring. It's all mental masturbation. All I know is it's fuckin' great to be alive in a world where I have a better-than-Ultrix 4.2 (what I was brought up on) Unix workstation on my desk *at home*. I only dreamed of that 5 years ago.
As far as the future of Linux goes, let the people decide. Linux is what you make of it. Nobody will force bloat down your throat (hehe). The modularization of the kernel is such a wonderful thing. I'm not a kernel programmer, but I can't understand why the central kernel code itself should need to rise exponentially... I'd guess more along the lines of *linearly* if it's being done 'right'. And I imagine the module code will increase with whatever the market brings. But that's just mental masturbation
Orp
A squid eating dough in a polyethylene bag is fast and bulbous, got me?
Compare Netscape Communicator to Gecko.
'nuff said. Lewis's claim of bloat killing
OSS software is bunk, because the link between feature creep and bloat isn't there.
Feature creep in OSS means somebody offering new command utilities, maybe a GUI here and there. If I don't put it in my hard drive, it isn't bloat.
So according to this guy, if I write a killer app, compile it for Linux, and sell it binaries-only, that will somehow make Linus wither like the wicked witch of the west once watered?
Worst case RMS won't invite me to his Superbowl party.
Then there's his statement "but Linux is still a Unix, and Unix is still losing market share". That's not FUD. It's fudging.
Hmm. Let's see.
Sendmail, BIND, Apache, INN, are probably some of the MOST used software on the internet. The versions bundled with most "commercial" *nixes are derived from the OSS versions, and most installations immediately replace vendor-supplied version with the latest OSS editions.
Nope, that's not mainstream. Clearly, we hit the brick wall and just kept going.
The Slashdot effect is great if it helps free projects, but if you quote from pay or registration sources, you just dignify their behavior and bring new customers to them.
I find it really annoying when I click on a link on Slashdot just to find that I have to register or pay some company. No, thanks.
And I find it preposterous that some people here actually comment on a short quote from a larger article that they could not read in total because you have to pay for it. Read it completely and THEN you can comment on it.
Bloatware is a result of one thing: a company's iterative processes through version numbers to get more revenue. Open-source projects are immune to this trend, because open-source projects aren't driven by corporations, but by necessity.
Programs in the open source model exist to solve a problem, not to get people to buy it.
As far as 'not main-stream acceptance', what about Apache? sendmail? bind? XFree86?
Granted, not everyone is using those programs, but if he's trying to predict what Linux (any *nix really) will be as far as a desktop operating system in several years... well... he's nuts.
Also, he fails to grasp that new talented programmers are created daily. Linux is quite popular on campuses. I can't cross campus without hearing about it somewhere. Linux can easily move in to capture future talented programmers in a way that Apple and Microsoft's astroturf campaigns cannot possibly attain.
Real computer science majors won't touch Microsoft or Apple products. That's why the business building has the shrine to Microsoft, and not any of the many computer buildings. (Macs are for the liberal arts people).
But these are *new* Linux users, not old time "Unix is everything" people. I hear people asking their friends what distro they should install and I see people with Running Linux books on their desks in my Computer Science classes. It was NOT this popular last year or the other three years that I've been attending college.
I go to the University of Texas at Austin. There's LOTS of Linux people around!
Consequently, a lot of the pro-Linux folks from Dell and IBM get hired from here. Many of my professors are also working at IBM and they all talk about how you can't walk through their buildings without seeing penguin stickers on the doors.
It's encouraging that there will be a future beyond working with the broken Win32 SDK; and with the trial looking so well, a future beyond working for Microsoft.
(Not that I ever would)
...because, while the author questions other people's statistics, he doesn't provide any inkling as to where he got most of his. So when he says that Linux has a higher 'defect density', he doesn't quantify this -- most likely because he cannot.
But this article is a great argument if one assumes that the industry will stay as it is, which has always been risky. IBM and others thought that mainframes would always rule the earth, Microsoft was once unknown too. Hell, I remember not very long ago when everybody did the BBS thing and life was good. But things change.
This article also seems to assume that CIOs will continue to have huge supplies of cash to throw at technologies and at blaming others (i.e. look at the support industry at present for what it really boils down to), and just as the computer has whet the corporate appetite for more profits, there has to be another target after all is eeked out from these machines. And guess who'll be in line? That's right, the ones who keep buying things for them.
But the largest problem, and the problem that the Linux community has to deal with right now, is that we're in a position where we just don't have to care about being commercial, or being big, or so on. This article treats Linux as a business, which it never intended to become. But more and more now I see people who are treating it as if it should be profit-motivated, and it's a shame that the chase after money has caught so many of us. Yes, Linus can be considered a bottleneck, but only if one feels that a certain schedule of releases must be kept up, which just isn't the case. Too often, we seem to be sacrificing our ideals to win approval of business, and by-and-by most of what business cares about here is making money, not keeping freedom.
If a woodchuck could, would it be too lazy to?
He may or may not have some valid comments there about why OSS software will succeed or fail.
Who cares?
I don't give a rat's fart whether OSS succeeds or fails in the commercial market. I use it because I like it. Not because everyone else says I should use it.
If businesses decide not to go with OSS software it's their choice, not mine. I'll still stick with Linux.
Gentlemen, start your penguins
Maybe some of the OSS projects will go down, probably others will arise....
Stability is a question of moving, not unlimited growth. Take a look at evolution...
Regards
kampi
-- a blessed +42 regexp of confusion (weapon in hand) You hit. The format string crumbles and turns to dust
I do not have any conclusive refutation for the conclusions (that Linux will die of instability, be embraced-and-extended by MS, or problems increasing market share), but I can pick a hell of a lot of holes in Todd Lewis' reasoning.
Firstly it is a more-or-less established fact that the OS market is not friction-free. Any OS provider experiences increasing returns as support from hardware OEMs and ISVs comes on line. This is roughly were Linux is now.
Secondly he seems to be comparing apples with oranges in trying to calculate the defect density. He says Linux has only 1.2million lines of code. This is the figure for the kernel. But he takes the defect rate for the *entire*system* including the bundled utilties, and compares it to the (probably correctly calculated) defect density for some (unnamed) commercial Unix. He goes on to say that all one-off figures are suspect, but supposes that this somehow casts doubt on Linux' reliability.
His list of features missing from Linux is truly pathetic. Linux (or rather XFree86) has video card support second only to Windows 98, and anyway this is an irrelevance on a server platform. Same goes for productivity software. Show me a wireless LAN worth using and I will believe the last point. Hw goes on to say that the addition of these feature will destabilise the system. How, exactly ? These are pifflingly small additions compared with the work already done.
He goes on to make some dubious comparisons between the Linux and Apache core team sizes and those of the competing teams in Microsoft. He does not seem to know that there is no general correlation between team size and productivity or stability of the resulting software, and willfully ignores the fact that the core teams are only a small part of the development effort and an even smaller part of the debugging effort.
He quotes the suspect statistics from the halloween document showing a decline in mail list traffic on the Mozilla groups (proves nothing), and the equally suspect statistic that IIS is faster than Apache (yes, but only on Win32, beyond 12 users or so any Linux platfrom running apache thrashes IIS on NT into the ground).
He believes that Linux is 'just Unix' and Unix is losing market share to NT. I think this misses the salient point. People want to use Intel hardware because it is cheap - they very much do not want to use MS operating systems. Linux runs better on Intel hardware than any other OS - therefore it is primarily competing with NT as a low cost OS, not with Unix as a Unix.
Finally he seems to beleive that Linux is vulnerable to embrace and extend, but he does not say how. Any attempt by Microsoft or anyone else to do so would face onrunning competition from the open source community and an obligation to release any modifications to the core system. That would make Linux a harder target to beat than, say, Netscape. Phew, that was long ...
Hi all,
There is something I think we in the Linux community aren't being completely honest about. Sure, the number of new Linux software applications in development are astonishing, and yes these applications are more than a match for the offerings of conventional software houses. But I think we are perhaps in too much of a hurry to show off these new developments, because they are for the most part in development - a great number in very early stages of development - and people out there in the world, including the media, are mistakenly beginning to think that these applications are what is on offer from the Linux community.
The result is that tens of thousands of people who are new to the Linux community - including the media - download the latest development releases of things like Gnome, Enlightenment, etc., and then get frustrated with stability issues, poorly implemented features, or just installing the thing!
Part of the problem is that so much has been happening over the past year that most Linux applications are in development. There are so many unfinished applications on offer that it is difficult for people to even find stable applications that they can use.
We really need to emphasise to new users and the media that Linux _is_ super stable and a promising alternative to other OS's, but _only_ if really stable, and usually fairly mundane, applications are used.
People used to the Windows world are accustomed to downloading every update they can find. This is not a safe policy in the Linux world because most of the time these 'updates' are untested.
We and the media has created a situation where new Linux users are hungry for every new and glossy app we can develop. The hype, in my opinion, is getting a little out of hand, and people are being mislead.
Why don't we encourage RedHat or one of the other Linux distributors to counterbalance their RawHide distribution with a RockSolid distribution. Hence we coud always point the media, business users, and new Linux users to the RockSolid distribution, and keep them enthused about Linux by keeping them informed of what applications will soon be declared 'rock solid'.
I'm not a professional programmer myself (whatever you define that to
be), but the tone of your comment suggests a lack of familiarity with
perl or emacs.
Perl is in fact a good scripting language, but the scope of
scripting is much larger than you think, which could explain your
disdain for its OO features. Scripting languages are often used to
write rather large (100,000+ line) applications. OO provides features
for complexity management that become very useful when applications
reach that size. Also consider that other popular scripting languages
(including Python and Tcl (through the [incr tcl] extension) provide
OO structure. Scripting does not just refer to 20 line bourne shell
scripts that pipe run a couple of useful binary apps together.
As for emacs, I'll first note that I am an avid user of emacs and I
disagree with your perspective of it. Emacs was not written to be
"user-friendly" text editor. It is more an extremely extensible text
editing environment. Flexibility appears to be much more a design
criterion than short learning curve. I agree that emacs has a high
learning curve to comprehend all the features, but the effort spent
familiarizing yourself with operating it reaps many benefits. I found
that once I got over the hump of learning a large portion of the
commands in emacs that I could work very productively. Human-Computer
Interface issues experts have shown (sorry, I have no references) that
human motor memory works much better using a keyboard than using a
mouse -- try hitting a menubutton without looking at your
screen.Memorization of key commands makes one much more productive
because you rely less on screen feedback to execute commands. Perhaps
the difference could be compared to that between having to stare at
the keyboard and type with one finger and being a touch typist.
Learning how to touch type takes some time, but the benefits are
grand.
As for useability, I'm not sure what version of emacs you have used,
but the versions I use (both Xemacs and emacs) have pull-down menus.
I've always preferred a program with menus that allow me to learn to
use the application. Over time I memorize the key commands and
forsake use of menus any longer. I would also note that, at least on
unix, that emacs motion and editing key commands are rather pervasive;
they work in applications such as Netscape, FrameMaker, and any tcl/tk
based gui application. I'm not sure you could call them standard,
but they can be found many places.
In a quick look, I loved this one the most, even more than the growth rate assumption above it. Now this is a scientifically proof clause to use as the basis for an explanation of how the OSS movement will "implode".
Laugh of the week, if it only didn't have a load of less critical readers. I hate the power of clueless media.
I think, therefore thoughts exist. Ego is just an impression.
I think that the development of these 'RockSolid' distros is inevitable, and a good thing. It only makes good business sense...
We'll probably see this when some distro (probably Red Hat) feels they have enough momentum, and a rock solid app base to make a push for the home desktop environment. I also believe that this may happen in the relatively near future (a year or two.)
Frankly, we're almost there anyway...I'm a computer guru by no ones definition, but i have very few problems with the applications that i use crashing (I'm using SuSE 5.3, btw.) I'm not saying that I haven't had some difficulties in getting my system up and running, but these have been almost completely hardware related.... I'm well aware that I differ from most people in that I'm willing to try and dig to solve a problem, even if I've no idea of what I'm doing. However, I'm reasonably convinced that the amount of digging that I've actually had to do is pretty small.
Two other things I've not seen mentioned yet.
1) As time goes on, I think that the portability of Linux will be what makes it viable...particularly since _someone_ will be motivated to get it working on the bleeding edge (at least until something better comes along.)
2) If Linux becomes too bloated, or can't do the job anymore, then something else will come along that will. This isn't a bad thing...it's healthy.
"The things we wizards have to put up with."--Jethro Bodine
Zontar
(somewhere in tenn.)
Il n'y a pas de Planet B.
>Frankly, a great deal of viruses have gotten into
>closed source program than one can count...
>usually these programs are benign and called by
>the name "easter eggs"! Have you ever seen the
>little doom clone that the Microsoft programmer
>put into excel?
``Easter eggs" started to appear because corporations did not want their programmers to attach their names to the software. (The corporations were afraid that head-hunters could then lure the best people from said companies to other jobs.) But since it is a human trait to want to attach one's identity to one's work, these best people figured out how to leave some sign that a real person did have something to do with the software.
In the Open Source/Free Software model, this need to attach one's identity to something is met by a line or two in the comments:
# Written by Joe Blow
# (c) by Joe Blow, under the terms of GPL
> How hard do you think it would
>be for one of those programmer to put a few more
>as a trap door because they are pissed of at the
>company?
Disgruntled employees are endemic at every high tech corporation. Treat the employees better, make the code available for peer review, & this problem will go away.
Geoff
I think I see a trend here. Maybe for them it really would be easier to muzzle the entire internet than to produce p
He says that Linux hass 1.5 million lines whereas unix has 10 million lines. What a load of garbage.
What a load of garbage! He is comparing kernel
to kernel+applications. Solaris is supposed to have 8 million lines of code and that refers to kernel+standard utilities that come bundled with Solaris. What is this crap of Linux having less defects because it has less code ( and thereby subtly implying it does less).
Thanks for a beautifully written article.. It's always good to see something like that on Slashdot..
There are a few things that maybe we'll have to agree to disagree on though... You say that Linux won't make inroads on the market of the 'average home user' (read gamer/email drone).. I've found that quite a few people who fit into this category who've actually installed Linux, and been happy with it.. A small learning curve (maybe the same as going from win 3.x to win 9.x) and they're there..
I've also had a _lot_ of Win users wandering past my workstation at work (where I set up a Linux box to handle the department's webserver, fileserver etc.), and look in awe at my basic Windowmaker screen with the clip.
Everyone wants one, and now, most of the people on the floor who own PCs are running Linux at home, to check out this OS that seems to do so much more than Windows.
I'd agree with your points, if you stated that there's a long way to go before Linux makes inroads into those markets...
I believe it will.. Not to the dominance of the market that MS have.. But I believe they'll maintain a reasonable presence.
Never is a long long time.
I still have a nice long list of lots of names back in '94-'95 who were telling me that Linux would never make it mainstream, that it'd never be a commercial viability and you'd never have non-guru users.
Some of them I still phone up to laugh at before I ask them out for a beer.
They're now the ones asking for my advice on how Linux can be used to increase the reliability of their company's information systems.
All that said, it's merely my experience of the situation, so what I write is coloured by my own experience and bias.
It'll still be interesting to see what, despite all the FUD that's spread, happens.
These are, indeed, interesting times..
Malk.
... is what people gain by slagging on OSS. Assume, then, that he is not part of an elaborate plot to FUD the OSS world to death (hatched by MS, of course). What does anyone gain by making the world safe for corporate america? Do middle managers sit in their offices thinking, Damn, I have to adopt a policy that makes more criminal, white-collar, lowlife a$$h*le$ rich; lets get our publication company to slag on that product that isn't technically owned by anyone. I see why ZD is a FUD machine: no one there is smart enough to type ls and then interpret the results. But the rest of the world? It just confuses me. You're picking on the little guy.
ZOMG I WOULD LOVE TO KNOW ABOUT YOUR FEELINGS ON MACINTOSH VERSUS WINDOWS, VI VERSUS EMACS, AND HOW YOU'RE NOT A DORK
"To qualify as a world-class success and not just a fad, each new product or method must pass the acid test of 'crossing the chasm' that separates early adoption from mainstream acceptance. Linux, and open source in general, fails this acid test."
I strongly disagree with this. The key question here is, what is the mainstream market? It is interesting to note that the OSS that has been around for a while has been very widely adopted among its intended market: software developers. Would anyone say that Emacs, vi, grep, sed, gcc, gzip, tar, to name just a very few, do not have mainstream acceptance amongst developers? And what about the server world: BIND, Sendmail, Apache, not to mention the Berkley TCP/IP stack?
Now, it is true that there is not wide spread adoption of GNU/Linux on the desktop. No one is arguing that, and that is not the immediate goal. GNU/Linux is currently being (re)positioned as a server OS. Its original 'market' was simply hobbyist OS hackers. A perfect example of its new market are ISPs. What proportion of hobbyist OS hackers and ISPs run GNU/Linux or *BSD systems? More than enough, I am sure, to qualify as 'mainstream acceptance'.
Pronouncements that OSS is just a 'fad', and that it 'fails this acid test' are certainly premature. IMHO, the evidence points to the exact opposite conclusion. In the markets it was designed for GNU/Linux and OSS already have mainstream acceptance. The question is whether they will be able to gain such acceptance in other markets, such as the workstation and desktop markets. This has yet to be seen, and to state that OSS has failed the acid test is like stating that Microsoft has failed the acid test because the majority of toasters do not run WinCE.
"The more the open source paradigm succeeds, the more untenable it becomes because maintenance and support grow more complex, and costs increase due to a scarcity of talented programmers. Success leads to features, and feature creep leads to bloated software."
Here the author seems to get confused between individual OSS projects, and the 'OSS paradigm' itself. Ignoring such distinctions the argument seems to go:
1) Maintenance and support for OSS grows more complex as that software succeeds.
2) [Implied] There is a limited pool of talented programmers willing to work on OSS projects.
Therefore:
3) Costs increase because this pool is exhausted due to 1.
Firstly, the statement that success breeds features is not necessarily sound. Success does not have to equate to feature bloat, and it does not have to mean increased complexity. 'ls' is a very successful utility. Is it bloated? I think not. One key design principle of UNIX, and therefore of GNU/Linux, is that of modularity. This directly counters feature creep. For non-utility apps this principle is not as strong, however it is still important. X is a classic example: it works with all the different WMs, which (should) work with all the different X apps. (Please! No GUI flames! Maybe I shouldn't even mention this as an example, oh well...)
Also, if a project is succeeding that implies that it is becoming more reliable, getting better documentation, and gaining a larger user base. Now, it is true that as the user base grows the testing becomes more exacting, more bugs will be found, and more features/enhancements requested, but:
One key aspect of OSS, notably espoused by ESR, is that as the user base grows, so does the support base. This counters the third point. As an app becomes more widely used, more people will want to hack at it, fix bugs, write docs, etc. It could be argued that the proportion of developers in the user base is too small to alleviate the problem. This is simplistic, however. Programmers are not the only ones who contribute to a project. Writing documentation, sending in bug reports, sharing ideas, providing feedback: these are all very important contributions that anyone can make.
Furthermore, it is perfectly possible for companies to pay people to work on OSS. If the cost of paying for the maintenance of existing OSS software is less than the cost of buying or developing and maintaining alternate versions of the software, then the argument is irrelevant. Note that in some cases it is impossible to maintain proprietary software at any price (e.g. when the owner goes out of business).
If I have the time I might try to address the rest of the points in the article, but for now these will do.
- "I never could learn to drink that blood and call it wine" - Bob Dylan (Tight Connection to my Heart)
I had my classes in Russia where they taught Pascal and C
FORTRAN and BASIC anyone?
What's LISP?
<^>_<(ô ô)>_<^>
It is possible for opensource software to be supported by commercial entities w/o becoming proprietary.
e.g. in http://www.opensource.org/for-suits.html several models are introduced. I think that with IBM, Dell, and compaq now (or soon) offering hardware with Linux installed one bears repeating:
If IBM and Compaq switched just one tenth of their AIX and Digital Unix teams to Linux programming, they could contribte immensly....
The slashdot post now links to a password protected article. I assume that the magazine disliked the slashdot effect. Anyone know where to find another copy?
The author is correct to a certain extent in believing that the romantic ideal of volunteer development of very large open-source applications will become increasingly strained in the future.
But the development of Linux is not simply going to collapse. Because it has become crucial to the businesses of so many companies, it is in their interests to spend a lot of money maintaining and improving it. It's simply a matter of re-adjusting the idea of 'an infinite number of hackers' to 'an infinite number of software / hardware companies etc'.
As a member of the scientific community I can appreciate the problems of scale up. What works on a lab bench doesn't necessarily work on the factory floor. The points brought to light in this artcile are worthwhile if they inspire the Linux community to work on solutions for scale up. The community (in order to attain the lions share of the market) needs to continuously recognize emerging challenges and develop from its chaos a means of meeting the needs of the community much the same way as a laboratiry does. The goal of scale up is to remove the austerity with which a procedure is completed. Take something seemingly complex and fine tune it such that anyone can see the steps involved. This doesn't make the procedure less complex or the one performing it any smarter it just carefully defines the parameters needed to complete the job. The so called shortage of talented people is overcome by a surplus of technicians. Before Win98 was released I watched an infomercial for it. The two Win98 programmers that appeared on the show were not boy geniuses they barely had a better working knowledge than I do (and that is pretty pathetic)they were technicians. The framework is already in place for hoardes of young semi-experienced programmers to help out; check out freshmeat. Centralized control and distribution of patches, fixes, and new ideas are abundant.
I strongly disagree with the author in that programs need to be bloated with features. Watch the evolution of cars. When a new car comes on the market it is usually complete, functional, and full of features and options. As the line of cars ages and new generations are released they too get heavier and more bloated with 'luxury'items. This continues until the heavy bloated car becomes a dinosaur and is remade.
This guy is a bisuness man, he has no idea what he's talking about. Besides his problems with his logic (which were already noted in some above posts) he seems to be claiming that free software has a very small market share and the market share isn't gonna succeed and its companies are going to make decisions about the future of free software. What a joke. Sure some bisuness have found ways to make a bit of money off free software, and some contribute to it, but in no way do they control it. The majority of free software is simply stuff people do in thier spare time, and they would usually LIKE to see it widely used, but they aren't going to go through any corperate hotshot bs to make it widely used. This is definitly the most ignorant free software FUD i have ever seen. And oh yeah, this IS FUD alright.
I prefer Emacs.
Just because I critize it, doesn't mean that I don't like it. In fact, I would say I am critical because I like it so much.
Emacs is an excellent editor. It does what I need it to do really well. It's just that all the other stuff it is expected to do, doesn't make sense for an editor to do.
vi is nice for fast and dirty editing, but not for long editing jobs (mostly because I'm more familiar with Emacs command, so I'm more productive with it).
Je ne parle pas francais.
Ok, I don't get it. He asserts Linux and Open Source (oddly enmeshed and confused in his article) are doomed to marginality because they are not commercial software.
Ok, let's look at the success of other commercial Unix/Unix utility vendors. Well, guess what, compared to the installed based of Microsoft they are all marginal.
It's not the development model, or commercial vs. non-commercial, or the defect rate in Linux utilities vs HP/UX - its that every other player has to compete with the 2000 lb gorilla of Microsofts installed base.
Beyond that his article is riddle with factual flaws and bad logic, and a general statistical illiteracy. Linux doesn't support video cards? Hmmm... I must be imagining what I am seeing on my screen right now.
Linux will probably never rule the desktop - oh well. It will rise to a position of power in the Server OS market, a position based on its technical merit and performance - which is on par or surpassing that of most commercial Unixes.
Open Source will always be a valid development model, suffering from its inadequacies as does the commercial development model. Some OSS products will see commercial success, some will only continue to be used by the Linux hobbiest. Who cares. The OSS movement has produced some awesome software products - and I like to use many of them - I could care less what this guy recommends his Fortune 500 droids do.
These products will ultimately stand or fall on their own merit, not on the politics of their development model.
-josh
Apparently MSNBC was misquoting Linus some last night. (Last I saw on linux-kernel, Linus was going to check over the transcript. I'm assuming he'll be posting to linux-kernel again after he does that.) The linux kernel has ~1.6 Million lines of code. But the article is not comparing 10 Million lines of kernel code in a commercial Unix to the Linux kernel code. Rather, it is comparing kernel+utilities of a commercial unix to kernel code of linux. It is also interesting that Linux is far more cross platform then any commercial unix. Unicies like HP/UX and Solaris only have to deal with one, or maybe two very controlled hardware environments. Linux deals with somewhere around 10-15 (guesstimate). Overall, linux is in a better position by that benchmark.
Also, Linux is not the only open source project out there. There is still FreeBSD, OpenBSD, and NetBSD as far as other open source unicies go.
I'm thinking that they had 'maxservers = 120' left from the default config :P
Here I am in my trusty lynx, and I'm asked for a password.
That's not the frustrating part. I take a guess at a l/p combo which works, being an old crypto fan. ;-)
But then I found that the article can't be read anyway, because it's in some proprietary format. This is ridiculous. Can't Slashdot consider such formats (as well as passwords) when accepting articles for submission?
Okay..before anyone flames me..let me clarify-I love linux.
Now..I feel that with large pieces of code (esp OSS) when there is no responsibility except moral, there will be instances of bugs coming through. Also people who work for recognition will work on the more "fancier" aspects of things for which they get attention. Who wants to pay attention to the nuts & bolts and the really CORE mechanics which is dull to say the least ?? Let me know!
Hey..you admitted it yourself. I admit the low level is there but then most of the people are working on the GUI. At the end of it all, its what deep within that matters. It exists now..BUT who is working on improving it ????
Your right, but why are there so many desktops (KDE, GNOME, GNUstep, etc.) and so many window managers?
Having 3 different desktops is as stupid as having so many 'kinda BSD' OSs
Funny, I find Xemacs and an Eterm (or five, sometimes) to be the most powerful development environment. Period. has it occurred to you that Emacs (& Xemacs) is meant to be a power tool, not a quick editor?
Try flying an F14 or a Mig. I bet you that they're a hell of a lot harder to fly than a single prop airplane. Guess what, they're also a hell of a lot more powerful. Emacs exists when what you're doing isn't quick and dirty, it's for when what you are doing is big and complex.
Think of it this way: emacs isn't meant to be intuitive to those who haven't used it, it's meant to give maximum power to those who master it. That's a good thing. When you don't want to master it, you don't use it. If you want to master it, you find that you have an extremely powerful tool at your fingertips, even if you may require all ten of them.
Think of it like Kung Fu. It's awfully complex and difficult. (Much more so than emacs, Kung Fu takes decades to master.) The result is awfully powerful and effective.
As for the rest, if you can't learn a language in a week, you're probably not the sort of person (at least yet) who would benefit from emacs anyhow.
Anyhow, nothing aside from food, shelter, water, and air is good for everyone. Different tools for different jobs, to each his own, etc. To quote some really good people during the interface wars (my term), "The only intuitive interface is the nipple. After that it's all learned."
They laughed at Einstein. They laughed at the Wright Brothers. But they also laughed at Bozo the Clown. -- C. Sagan
Could someone put the PDF file up on their web page for me, so I could have a look at it? It sounds really interesting, because I thought right from the start that the open source model had some problems with it.
:)
The only way I figured it could work is if computer communism was established, where tax dollars went to the government to pay the software developers, and everyone got free software.
Of course, that wouldn't work.
There are TWO critical issues for good software - programmer quality and motivation.
Its amazing the good work you can get from fairly average people who are highly motivated. On the other hand, I'm sometimes amazed at the crap I have produced when I couldn't give a damn about the outcome.
I thought one of the key features of these "learned" journals was peer review. Maybe OSS could teach them what that means.
The article in Computing seems to assume a rather restricted definition of success. It used to be a commonly held belief that a computing utopia involved a whole menagerie of operating systems communicating through simple, standardised interfaces. Now it seems that the only acceptable definition of success is the business agenda of market share and world-pervasiveness.
I am a dedicated Linux user. However, I would be disappointed if that operating system came to completely dominate world markets. Whilst it is suited to me, it is not ideally designed for all computer users. For example, I revel in the freedom which X-windows gives me in separating mechanism from policy, for other people this `freedom' may be confusing and disorientating. It's not a patronising attitude, since the option to change systems always remains.
Above promoting one operating system or another, the real achievement of Linux is to be open, to implement existing standards cleanly and transparently, and to do so without a covert business agenda.
There are, believe it or not, people out here who judge the success of an operating system by its fitness to purpose of one particular user base. Linux may, or may not, only be well suited to a small set of geeks and systems administrators. This small group of people could well be a poor prospect from the point of view of marketeers and advertising revenue. But, for me at least, this does not mean that the operating system is not a success. Not everyone is after world domination.
-- dans@chiark.greenend.org.uk
I agree. However, I was taking a worst case scenario. Even _if_ Linux were restricted to that small group of users, then it could still be a success. I don't think it is, but that's the worst case possibility. As I said, I think it has more potential than that, but as a card-carrying geek, I would, woukldn't I?
Sorry, but I do it today. Using Java with fields.
Adults are obsolete children. - Dr. Seuss
There error density in the article is unacceptably high...
/usr/src/linux and *.[ch], Linux 2.2.1 has 1604504 lines of code. Throw .S files in there and the total swells to 1676913 lines of code.
"However, Linux 2.0 lacks the following features:
"* video card support" ???
"* Wireless LAN support" http://www.rage.net/wireless/wireless_howto.html Plus 2.2 most definitely has wireless LAN support.
"* good selection of productivity software" Depends on your definitions here, but that will be a hard position to take in a year.
"UNIX has more than 10 million lines of code, while Linux has only 1.5 million," so Linux has a higher error density (errors per lines of code). Where is he getting this numbers? Is he just counting the kernel? If so, he is probably undercounting severely. No way is a typical UNIX kernel that much larger. If he is counting utilities, then he is way underestimating. Either way, it's a completely bogus comparison. In the text, he is supposedly comparing utilites.
Okay, by my count (using find, wc, and python) on
Now the RedHat 5.2 CD #2 with all the SRPMS is a couple hundred megs. I'd pop it in and find out the exact number, but I don't have it handy. It's got to be at least 300 MB, at a guess. Taking into account roughly the same compression rate as the 2.2 kernel src (about one line per 7 bytes), that's 42 million lines of code for RedHat. It's a guestimate, but at worse I figure +/- 10 million lines, but easily as much as UNIX (whatever UNIX he's talking about where you can count lines of code). SuSE is substantially larger.
He also assumes that the Linux (apparently kernel) code base will continue to grow exponentially in terms of lines of code and will be as bloated as NT (and where will NT be?). A lot of those lines of code go into supporting different platforms and drivers. I suspect maybe 25% of it or less is actually used on any given platform. How many platforms does NT support? (Two, if you count token Alpha support.)
There's more, but basically the bottom line is: This guy is totally smoking crack or something. Oh wait, I get it now! He's dropping acid! That's why Linux doesn't pass the "acid test": He popped a couple microdots and Linux didn't sing a pizza into his nose or something. Bad trip, man!
-- Blame any errors on your own stupidity. All wrongs reserved.
So I downloaded and printed out this article using Acrobat. It printed the first page OK and then printed only the graphics on the following pages.
If he had used Ghostscript no doubt it would have printed it all just fine.
--------
Bill Gates Is My Evil Twin.
I didn't even bother to point out to Ted Lewis that the article doesn't print properly in Acrobat.
Instead, I sent him a rather fierce response, accusing him of doing a hatchet job on Linux and open source. For example, Figure 1 shows the code base of Linux rising by an order of magnitude in the next four years (with no factual basis whatsoever provided for the claim! -- the worst kind of chartjunk), yet the article directly contradicts that by noting the difficulty Linus and the core team are having these days just doing normal tree maintenance.
Frankly, I write this off to pure polemics. I told him if he wanted to write about this subject again he should start from scratch and do it right. But I'm not holding my breath. Lewis is a dinosaur and we can't save him.
--------
Bill Gates Is My Evil Twin.
Ignore the media. Ignore the good press, ignore the bad press, ignore the idiots. Just sit down, relax, and write the code. We are doing just fine, and things will be fine in the future. Focus on the goals, not the press.
-Master Switch, one more element in the machine
Didn't anyone notice the article on ZDNET showing
linux w/ apache faster than NT w/ ISS?
Even the ZDNET guys saw that one!
Fellowship 9/11
Don't Apache and Sendmail count as "wide acceptance"? And I'll laugh if he doesn't think Gecko will be widely accepted. These aren't just niche groups using these products.
I am not a professional programmer, so I cannot tell if this guys arguments against perl are valid or not, although it seems to me that OO elements in perl 5.x are completely unneeded by a scripting language and perl is not useful for anything but scripting and small to medium CGIs. As for emacs, I do not understand why a plain text editor even having a whole bunch of features has to take >20MB of my hard drive? And in this particular case we talk about an editor that is so user unfriendly that one cannot even open a document ( C-x C-f ) save it ( C-x C-s ) and exit ( C-x C-c ) without doing A LOT of RTFM. I inserted those E-macs keys in this note to show to people who might not know emacs capital C means Control key and those are key combos you have to press to get the mentioned results. Now why do I have to know lisp to configure my editor? Because all CS students learn lisp in school? I didn't go to any CS classes in this country, I had my classes in Russia where they taught Pascal and C, and I don't understand at all why do they still have US students learn Lisp even if most of them are not gonna work with AI. I don't wanna learn Lisp because I think it is ugly and I don't wanna learn any programming language to configure a text editor. If you tell me that emacs is not only a text editor, I will even agree. For reasons unknown to mankind emacs is a conglomerate ( default config ) of an archaic text editor, bad mail client, bad news reader, unusable ( meaning that plain bad is too weak for this ) calendar and a few other things of about the same quality.
You can flame me now all you want, I understand that things like emacs are nostalgick, but it doesn't make them good for everybody, imagine you would have to go around in a top hat just because that is what your grand-grand-dad used to wear when he was young.
Everybody Lies. But it doesn't matter since nobody listens.
Lewis' arugment that support will be unmanagable after the user base increases beyond a certain point is dumb. Do you think Microsoft handles all the support for all of it's products? No. Then who do people turn to when they have windows problems? They go to their computer vendor, computer guru friends, and local computer stores. In other words more than just Microsoft people support Microsoft products. The same will happen with Linux. For support people will call VA Research, Red Hat, Caldera, the HOWTOs, the news groups, irc, their next door neighbor. Not every one is going to run to Linus to solve their Linux problems.
On another topic, as available software increases more developers (i.e. windows developers) will switch to Linux. This will mean more hackers will tinker with Linux and produce more patches and kernel code. This will make up for the "supposed" lack of developers for Linux (Lewis said there where about 200 developers for the Linux kernel) compare to MS (400 kernel devs.)
Let's see, we have virtually every major database being ported to Linux. We have games beginning to be written for Linux now(as part of the initial release, not an afterthought or port) as well. There are who knows how many millions of systems running Linux. So, it's gone a bit beyond, "fad" stage when commercial products are showing up for it regularly. As for the problem of bloat, you may have noticed that many distributions have an FTP site where additional packages can be downloaded from. This will allow for the bloat to be removed from the initial release, and to keep things to a reasonable level. So, bloat isn't an issue either, as long as the packages are organized in a way that can be managed by the users. This can be dealt with by offering different "levels" of install for users. I know that Debian GNU/Linux began to implement a "function" question to help make installation easier for beginners with the 2.0 release. I suspect that when 2.1 is released, it will build upon this feature to help make the initial install easier. Redhat and SuSE have used a GUI install to aid users through the install. This shows that it will not take much longer before Linux becomes "accessable" for many endusers to move to.
Vulnerability to viruses is more a consequence of the distribution method than of the software itself. After all, it's not unheard of for even shrink-wrapped software to be infected. What's really at fault is the implicit instruction to "just trust me and run this code".
So what we need is someone to trust. This could be a shrink-wrap packager like Red Hat, or a certificate authority. Automatically checking the signature on new software shouldn't be any more complicated than running a virus scanner - something that many (most?) mainstream users already do.
> To qualify as a world-class success and not just a fad, each new product
> or method must pass the acidtest of "crossing the chasm" that separates
> early adoption from mainstream acceptance.
Besides this not being the only, or even best, criterion of mainstream
acceptance, he fails to define mainstream acceptance. With Linux' user
base in the same ballpark as Mac OS, I'd say it has passed this test
anyway.
> This is because the strategy of the weak no longer works when a product
> becomes a serious threat to competitors. Microsoft demonstrated this in
> its battle with Netscape. Once Netscape seriously challenged Microsoft by
> gaining the dominant share of browser users, Microsoft skillfully applied
> an absorb-and-extend strategy, a proven technique of the strong.
Huh? Netscape was not open source. Netscape didn't "challenge" MS, they
created a market that MS dismissed at the time. MS got market share
through price dumping, product tying and exclusionary contracts, not
"absorb-and-extend".
> Unix has more than 10 million lines of code, while Linux has only 1.5
> million. So the Unix defect rate could have been as high as 60 percent
> and still paralleled that of Linux.
Program A has 100 Klocs and 10 bugs. Program B, with identical
functionality, has 50 Klocs and 6 bugs. Program A is more reliable?!
> Greed is the greatest motivator of all.
Spoken like a true American. He simply doesn't understand the OSS
movement.
> First, Microsoft has many more options than does the open source
> movement. Because it holds the strong position, Microsoft can simply
> absorb and extend Linux.
He also hasn't read the GPL.
> A third scenario is most likely: Linux will turn commercial and Caldera,
> Red Hat, or some other traditional software publisher will push it into
> the mainstream.
He definitely hasn't read the GPL. And he doesn't realize that Caldera
and Red Hat went commercial a couple years ago.
> In their drive to simplify support, they will always choose Windows,
> Solaris, or HP-UX over yet another Unix. After all, total cost of
> ownership is driven by support, not capital costs.
I see this argument a lot and I hate it because it's so stupid. It
implies that no OS will ever replace another. If the cost of adding or
switching an OS was so great, we'd ALL have 3270s on our desks.
> Even if Linux proves
> the better product, Microsoft can always use tying-the creation of a
> mandatory dependency among applications-to maintain its monopoly.
Where can I get the drugs he's taking? The DOJ disagrees with him.
"The cost of freedom is eternal vigilance." -Thomas Jefferson
Lewis seems to think that giving away code (OSS)
is analogous to, and an extension of, the
marketing strategy of giving away a product
(binaries, in this case) in order to increase
market share. This analogy is dubious at best.
The correct analogy would be a company giving
the manufactoring plans and tools needed to create
the object locally.
Half and Half: let me offer a couple of C++ programmer perspectives.
There are numerous projects currently being worked on by OpenSource programmers to port languages and databases such as Clipper, dBase, etc. to the Linux world. These were good DOS based tools which for the most part did not survive the migration into Windows very well, mainly because ODBC, Visual Basic, and Access, etc. made it too damn easy to kludge together small databases, etc. and killed alot of the midrange market for development tools)
That being said, in the Linux world, not counting the Oracles, etc. which are currently porting their products, there are already plenty of fairly powerful databases available such as MySQL, etc. which will also continue to improve.
What we do not have yet is a GUI based development environment in Linux which is as flexible and powerful as the MS, Symantec, Inprise (Borland), and Sybase tool sets.
Stay tuned. It is only a matter of time before the tools develop, whether OpenSourced or by the professional tool sellers such as Symantec.
Once we have effective, easy to use development tools, foundation class libraries, etc., the powerful GUI applications will come, because the market is there.
So don't despair: join the effort.
...Open Source isn't the only answer -- but it's almost always a better value than the alternatives...
I grew up in LV and something you need to realize about UNLV's computer science department.
Historically it has been very conservative, as the major market in LV was the defense industries. As far as a user's group, I can guarantee that if you have an operating Linux system and you post on campus, within a few days you will generate enough interest to either find or start your own.
...Open Source isn't the only answer -- but it's almost always a better value than the alternatives...
>>Basically, I don't believe Open Source will produce polished GUI applications...
I disagree. Although I haven't released yet I and many others have too much pride in our work to leave something "unpolished." Coupled with the fact that the Linux User community is the fastest bug reporting crew I've ever seen, OpenSource Software should be MORE polished than the competition.
> Instead, you'll get more retrofitted bloatware (GtkEmacs, anyone?) and lots of pointless cloning (57 different IRC clients, anyone?).
True, but look at Windows. Their are probably fifty different Word Processors you can buy. But only the top four or five really matter, because their features are polished to the point of survival in a tough market.
Same for Linux if not more so. What I look forward to is a code library that allows me to point my compiler to a target desktop (KDE, GNOME, or TheNextNewGreatestVersion, whatever it is) and be able to release versions of my application without completely rebuilding the components from scratch for each desktop.
Suggestions anyone?
...Open Source isn't the only answer -- but it's almost always a better value than the alternatives...
If you are releasing software with YOUR name on it, not some big bloated company's name, wouldn't you be more interested in keeping the bugs OUT?
But if you release teh code with, then it is easier to get reports on what the bugs are and clean them up, optimize code, and release a beter product bearing only YOUR name, and a list of credits to those who help.
On the whole I find Linux much MUCH less buggy that any M$ product I have used!
I have only had a linux kernel crash ONCE on my system, and that was because I really messed up on the compiling of it. Never tell it that there are no drives that the kernel can support.
Win NT slows seriously and demands a reboot about ever 3 days when I leave it up for that long.
Laugh, it's good for you!