Virtues of Monoculture, Or Why Microsoft Wins

blackbearnh writes to ask, "Why does Microsoft win the development environment war so often, when we all know it's a lifetime lock-in to Windows? Perhaps it's because the open source community offers too much choice." From the post: "Microsoft offers the certainty of no choices. Choice isn't always good, and the open source community sometimes offers far too many ways to skin the same cat, choices that are born more out of pride, ego, or stubbornness than a genuine need for two different paths. I won't point fingers, everyone knows examples... The reality is that there are good, practical reasons that drive people into the arms of the Redmond tool set, and we need to accept that as a fact and learn from it, rather than shake our fists and curse the darkness."

Comments

FAQ item

[Wonko the Sane]

This really needs to be put in a FAQ somewhere.

Does this author have a valid point? Probably
Is this point, and any relevant discussion, different from the last time this was brought up a few months ago?

Probably not.

Re:FAQ item

[Anonymous Coward]

I don't think the author does have a point. After using Vista, I think it's the money put behind discovering the the sweetspot, and zeroing on that. Is Vista perfect, no. But is that a reflection of my deviation from the exact center of the sweetspot, or microsofts failure to anticipate users wants and needs. No, I think that they cast a wide and tight net, catch 95% of the fish, and the rest serve everyone else. Which is good, a real and public good, which doesn't seem to be recognized by the article. Vista will win me back from KDE, it already has honestly. I even had occasion to use OS X intensively this week, and it is not for me. But that heterogenious nature of the OS spectrum insures no one has to go unserved. If Vista had been a horrible catastrophe in my eyes, KDE would have continued to serve me well. Failing that, 2k. To be sure there is someone with a MacBook out there thinking "I love OS X, but worst case scenerio, but if I had to settle, Gnome is good enough." Now if me and him were to meet there would jungle rythems and one of us would be Kirk, and the other would be a lizard man in a loincloth. But the fact remains that while Microsoft serves most, likely best, all should be served as they wish, even if it's a cloud of rabbid individuals who demand full control, no matter how byzantine, over their user experience. We are all discreet elements in an appearently continuous spectrum.

Re:FAQ item

[Wonko the Sane]

The only definite point I got from that article was "sometimes too much choice is bad". I don't think you can really seriously argue with that statement, but on the other hand it'd not all that helpful either.

Re:FAQ item

[CokoBWare]

/* begin flamebait comment */

Does the open source community do anything to change it's fractured ways since the last time this was mentioned?

Probably not. /* end flamebait end */

I agree with him... I think open-source software is awesome. But there is too much fractured choice in the OSS community, and sometimes businesses are better managed and operate smoother when OSS is not part of the equation.

Just my opinion... I know someone will flame me *toast* *poof*

Re:FAQ item

[mrsteveman1]

No you shouldn't be flamed, you haven't said anything incorrect.

The difference i think is management, which Microsoft has, however flawed, and open source as a general rule, does not. Even within single projects there is useless argument, and forks for ridiculous reasons. In most cases, the required action is for one party to be kicked in the ass, hard. There are RARE cases where the majority of the community sees something going wrong and forks, such is the case with X.org.

Then you have cases like gnome and kde, which each develop totally redundant, sometimes useless ways to do the same thing, sometimes neither one does it well either.

Over and over again i see MAJOR parts of the system literally missing, like a device manager, while other parts, like file managers or office applications (openoffice, gnome office, koffice) are developed 3 or 4 times over in parallel by groups who either refuse to use code from another group simply because it has a G- or a K- in front of its name, or neglect to even look around to see if someone has already coded a similar app that could be used and improved.

In all honesty, gnome and kde have driven me away from linux for everything but core server use, and my next laptop will be a Macbook simply because i'm tired of it all.

Re:FAQ item

[tftp]

groups who either refuse to use code from another group

• How many programmers have you seen who like to use someone's else (NIH) code?
• How many programmers produce code that is worth reusing?
• How many programmers write structured, reusable code to begin with?
• How many of the F/OSS programmers have design specifications finished and approved before the first line of code is written? Compare to the realities of commercial programming. This affects the structure of the code.
• How long will it take you to find the free code instead of writing it from scratch? You need to match: license, language, interface, libraries, and other requirements. You also need a good documentation on the code that you are reusing (guess that excludes many F/OSS projects right here) because if you plan to read through the source you indeed might be better off just writing your own.

Besides, many F/OSS people write code not just because they want to produce something specific, but because they like to write the code. There are many babies in the world, but every woman wants her own, strangely enough.

What commercial coding adds is discipline. Your manager may order you to write this documentation, or to use that library - because he has a reason, good or bad. And he has power to make sure you do it. If the project requires coding an ugly routine in an ugly language a F/OSS coder would rather not do it, and he'd be right - he is not paid to suffer. But a commercial coder will do the job, even if it involves 8052 assembly language instead of Python on Planes :-) Every job has its unpleasant parts, and while a F/OSS coder can skip them a commercial coder can not; if the spec calls for an embedded testing code, for example, or Doxygen comments, you put it in.

Discipline and dictatorial approach affect the result a lot. Basically, every commercial product is designed either by one person, or by very small group of people. This person (or group) has complete control over every aspect of the product; s/he might be wrong but at least the product is consistent, and not designed by a committee as it sometimes happens.

In addition to that, commercial products pass the rigorous testing by the free market, and that testing starts when someone thinks about the very idea of a new product. The project may not go forward until there is a good plan how it will be sold, and to who, and for how much. If these numbers make no sense then the product won't be even made. In F/OSS world, for example, I am free to write - and to release into the world - yet another clone of Vi or Notepad (we have hundreds by now, probably.) These clones haven't been weeded out by the market, and so many of them are not viable - but they are out there, just polluting the set of choices because someone will pick some and will be disappointed. You can't reasonably expect a user to choose one out of so many apps? That is a problem.

And, as someone already mentioned, if you combine AbiWord, KWord and OO's Writer you still don't get MS Word, even though the combined labor that went into all three is probably comparable. Effort dispersed, spent on competing projects is ultimately wasted. But it is so hard to join efforts because compromises and agreements are needed. In a business that would not be a problem.

Re:FAQ item

[chromatic]

How many of the F/OSS programmers have design specifications finished and approved before the first line of code is written? Compare to the realities of commercial programming.

What commercial coding adds is discipline.

Discipline and dictatorial approach affect the result a lot. Basically, every commercial product is designed either by one person, or by very small group of people. This person (or group) has complete control over every aspect of the product; s/he might be wrong but at least the product is consistent, and not designed by a committee as it sometimes happens.

Where in the world have you worked? This is so far from my experience that I'm starting to wonder if I ever worked in software at all.

Re:FAQ item

[tftp]

There are plenty of places that subscribe to CMM and in some markets (government/military, medical/life-related, aerospace, etc.) you can't even get a proposal out if you are not CMM/CMMI all the way through. I work for one of those markets.

Re:FAQ item

[JohnFluxx]

> Does the open source community do anything to change it's fractured ways since the last time this was mentioned?

Take a look at freedesktop.org.

* Sharing of sound system - both Gnome and KDE 4 will work with gstreamer
* Joining of messaging system. It was dcop (kde) and corba (Gnome). Now both will use DBus
* Common themes that make kde and gnome apps look the same.

Plus lots of 'small' points. Both follow the .desktop standard for menu items, actions etc. Both use the freedesktop.org icon naming system, and mimetype system, and so on.

de facto standards

[swordgeek]

What the monopoly says, goes. They define a standard. Because they're MS, they define a standard that's different and incompatible with official standards. You either go with the market, or you swim upstream. This is about as clever as saying, "the reason red is red is because it's not yellow."

Nothing to see here. Market forces and ease of use win over features, stability, or quality.

Re:de facto standards

[hydroxy]

It's all boils down to the fact that very few Open Source people (or so it seems) do it for money. This means that no matter if the product 'sells' to 1000 people or a million, they will keep doing it. MS won't sell something that isn't making money, yet Open Source people will keep devoting their time to something that hardly anyone uses.

Think of it like this... you have 50 car dealerships in a small town. They all get their cars for free from the factory. Everything they sell is for a profit, but because there are so many of them, they barely make any profit. The only ones that shut down, are those who get tired of not making any profit. Some owners are content to keep their dealership open without making a profit, because they get to say they own a dealership, or they have a select few customers who always come in to buy cars, thus making them think that what they are doing is worthwhile.

Re:de facto standards

[Osty]

This means that no matter if the product 'sells' to 1000 people or a million, they will keep doing it.

Really? How many dead apps exist on sourceforge? Freshmeat?

MS won't sell something that isn't making money

That's why they killed Xbox, right? Oh, wait ...

Think of it like this... you have 50 car dealerships in a small town. They all get their cars for free from the factory. Everything they sell is for a profit, but because there are so many of them, they barely make any profit. The only ones that shut down, are those who get tired of not making any profit. Some owners are content to keep their dealership open without making a profit, because they get to say they own a dealership, or they have a select few customers who always come in to buy cars, thus making them think that what they are doing is worthwhile.

Worst. Car analogy. Ever.

Keeping a dealership open when it's selling no cars still costs money even the cars are free to the dealer. Similarly, open source projects that don't make money are done in spare time because the author(s) still have to work to eat. Yes, some authors will keep doing it for love of the problem well past the point where they've scratched their own itch. More will get the app just to point of scratching that itch and then mothball it (see open source apps that haven't been updated in years). Most will never get to the point of scratching their own itch and will die in a planning, pre-alpha, or alpha stage, never to be useful by anyone including the original author.

The open source projects that survive are the ones that can either drum up enough interest to build a large developer community such that the author doesn't have to spread himself so thin, or the ones that are able to scrape up some cash in order to pay a developer or two to work on them. Linux is not solely developed by Linus anymore. Perl is not solely developed by Larry Wall. And so on.

As in

[sugarmotor]

As in - Why not limit the number of websites? Too much choice!

Re:+5 (Obvious)

[TubeSteak]

The very nature of Open Source allows for such diversity that it will obviously be split ito smaller groups of enthusiasts. It's the nature of the beast.

To many choices are demotivating.
People feel overwhelmed and decide not to choose anything.
That is basic psych.

http://www.google.com/search?hl=en&q=too+much+choi ce

Choice Wins

[xzvf]

Interestingly enough, I had this arguement today with a co-worker. Choice and flexibility afforded by open source and more importantly open standards will pay dividends for companies that think long term. The shrink wrapped mono-culture beat can be the less expensive option in the short term (no retraining, prepackaged apps with ready training and documentation, cheap labor). But, open with lots of choices wins in the long run every time because it gives ownership of IT to the companies that use it instead of the companies the produce it. Freedom and choice may be the difficult choice in a short-term return corportate culture, but the companies that embrace open standards will be the long term winners.

Re:Choice Wins

[sheldon]

I'm leaving my company now, and I'm writing up some documentation on the systems I've left behind.

They're all similar. They are web apps which did similar things, so I wrote the applications in similar ways using similar technologies. The build similarly, they install similarly. I would choose new technologies as I went, if they were clearly better but I tried to fit them into my existing assumptions.

Choice is great if you are a rogue cowboy developer. Lot's of stuff all over the place, bits and pieces thrown together. I remember a project we had here, it'd been outsourced to some third party. It came back with just about every piece of free open source software you can imagine. The data entry screens were Java running on Apache, the reporting screens were Python, the admin screens were running Perl scripts. The data entry stuff used Oracle, the reporting used postgres. The whole thing was tied together with some other bits of glue and tape. Thank God the morons who wrote it were horrible architects and the thing couldn't scale, otherwise this piece of unmaintable crap might have ended up in production.

But when you're trying to write documentation to hand stuff off to the next person, it is so much easier if what you have left behind is all similar to other stuff. It's just so much more maintainable, and easier to train the new guy in.

That's what wins in the long term. It's not raw freedom and choice, it's making intelligent choices and then sticking with them.

Re:Choice Wins

[KingMotley]

That assumes that companies actually want ownership. Most don't. As a developer, I don't want ownership of the IDE. I don't want to have to learn how to recompile it, or learn the inner workings of it. I don't get paid to change the font it uses from the default, nor do I really want to try and help a fellow developer, only to realize his program won't compile because of some optimization he did to his compiler so that it ran .01% faster.

I don't demand that sears give me the complete schematics to the hammer I bought, along with detailed instructions how to cut down a tree to form the handle, and smelt the metal for the head. I just want to hit something with it, and it to work. That's something that I think many in the OSS fan clubs fail to realize that most people who really want to use something DON'T want to tinker with the internals of it.

Re:Choice Wins

[marcosdumay]

"That's what wins in the long term. It's not raw freedom and choice, it's making intelligent choices and then sticking with them."

And how do you plan on makng inteligent choices without 'raw freedom and choice'? The only point that I can take from your post is that lack of choice is (badly) a replacement for good management. Well, that is a nice point, but it doesn't seem you where trying to make it.

Re:Choice Wins

[littlewink]

It came back with just about every piece of free open source software you can imagine. The data entry screens were Java running on Apache, the reporting screens were Python, the admin screens were running Perl scripts. The data entry stuff used Oracle, the reporting used postgres. The whole thing was tied together with some other bits of glue and tape. Thank God the morons who wrote it were horrible architects and the thing couldn't scale, otherwise this piece of unmaintable crap might have ended up in production.

Does "piece of unmaintainable crap" refer to Java, Perl, Apache, PostGreSQL, Oracle, Python or all of them? Because each of those has a reputation far better than yours.

Or is it supposedly a "piece of unmaintainable crap" because of the architects' design?

Before you post again, try to determine exactly _where_ the problem lies. Here you attempt to besmirch, to no effect, the very pillars of Open Source (Java, Perl, Apache, PostGreSQL, Oracle, Python) yet, when distilled down, your only possible criticisms are that

1. Open Source has multiple languages and
   
2. the architects built something that did not scale.
   

Point 1 is obvious and silly. Point 2 is questionable at best, because you're making the claim by yourself with no supporting evidence.

Maybe you weren't willing to examine the system to see what was wrong and fix it? So you left. Should we trust the word of a man who quits a company and then criticises it's new systems? I think not.

Hmm

[pembo13]

This article assumes that open source developers are aiming at becoming Microsoft like. Maybe they're just in it to make good software: not a profit, not make money for shareholders, or anything that that Microsoft is obviously aiming for. And the article is also using a very narrow definition of "win", one which I'm not sure is possible for OSS to attain.

Re:Hmm

[tftp]

Maybe they're just in it to make good software

Then you'd better define what does the word "good" mean here. How do you measure goodness? One common and sensible metric is popularity, since it measures the willingness of people to use the product. Is a software that is fast and secure good, even though nobody uses it? (because the interface is command line, in Mayan for example.)

MS surely measures goodness by sales numbers and by market penetration; that flows from the fact that MS Office, [= 2003] for example, just works for most of the people. You install it, and you are in business. That's good. On the other hand, take LaTeX - it can do things that you'd have to struggle with in MS Word, but you also must be a rocket scientist (or a mathematician, which is the same) to use LaTeX. That is seen as seriously ungood.

It's not the choice

[AKAImBatman]

The problem isn't the choice, it's the follow-through. Open Source software maintains its momentum as long as there is an itch to scratch. As soon as that itch is satisfied, the work stops. Even if the code is unsuitable for your average joe. Technically, this is where the commercial distributions are supposed to pick up the slack and do the rest of the work. You know, offer an integrated Linux environment. Something to make all that money we're throwing at them worth something. But they don't. And I have no idea why.

Perhaps the most telling event was when I got a copy of Sun's Java Desktop System. It was a complete SUSE-based distro with Sun's unified desktop on top of it. I forget what the exact problem was, but in order to change a *BASIC* system setting, the instructions required that I directly edit a system file.

Excuse me?

This little gaffe was repeated by Mandrake with its command-line audio setup. RedHat with its inability to automatically handle its own damn package format. So on and so forth. I forget how many times integrated tools should have existed, and... well... didn't. I won't even get into the "broken by design" GUI choices of GNOME.

Now Ubuntu has been slowly trying to push this back; to make Linux a bit more user-friendly. But it's just one distro among many. There needs to be a concerted effort from all companies that SELL Linux. They need to give as good of an experience as they can possibly give. Simply repackaging the same software with a new GUI theme isn't going to cut it. They need to actually spend some money on covering the gaps that the unpaid community isn't going to cover. (I mean, let's be reasonable. They're not getting paid to develop a boring dialog and test flipping the switch 300 different ways.)

The development tools themselves are fine. In fact, Java is pretty well covered by Eclipse and Netbeans. If Linux distros make more of an effort to integrate the (now OPEN SOURCE!) Java into the system, they can make developer's lives even easier. Mono is also an acceptable choice, but the key thing is to get it integrated. Make your commercial Operating Systems FEEL like commercial Operating Systems. Not hobby OSes that have a nice coat of paint on them. In other words, maybe you commercial guys could pull your weight a little? Maybe?

* Ok, you can start flaming me now. I'm sure I've said something that offends distro X fanboys. Bring it on so I can ignore it. :-/

Re:It's not the choice

[Anonymous Coward]

Sigh...

It is idiotic posts like that the make us all dread these types of stories on Slashdot.

Sigh...

It is idiotic posts like this that make us all dread these types of stories on Slashdot.

Re:It's not the choice

[Tablizer]

The point I'm trying to make is that I like the choice. I like that I could use FreeBSD for my Postfix server but Slackware with a custom kernel for my routing and firewall needs. I liked that I could create the tools I needed rather than having to work with the very fixed monoculture that is Windows and its networking tools. I like that I could build streamlined servers that didn't possess a whole bunch of shit that has absolutely nothing to do with what I am doing.

You assume that everyone else likes what you do. My favorite tools fell out of style and don't really exist anymore. Thus, it doesn't matter what I like. It is what the majority likes that matter. MS might be mediocre, but at least it is consistent mediocre. When you step into a different MS shop, you don't see a custom-made OS built by some Lisp or Perl guru that you have to reverse engineer and learn. That takes too long. OSS needs more de-facto standards. Sometimes mediocre standards are better than great-but-uncommon tools. SQL has survived because it is a common good-enough standard, not because it is the best possible relational language.

Re:It's not the choice

[Anonymous Coward]

While I agree with "The problem isn't the choice, it's the follow-through." I'm going to slighly change the argument.

I think there are two types of developers. Those that make "legos" and those that make "sculptures."

The first type make rock solid building blocks. They want to solve problem x, or possibly problem set X. But avoid assembling the pieces (they are more powerful divided, think about how in Unix you are given a handful of command line tools and can build really cool shell scripts easily).

A worse crime is assembling the pieces but doing a piss poor job and creating a tool that is neither powerful or simple (Microsoft's use of "Wizards" comes to mind, but there are plenty of open source tools that meet this criteria).

Then the sculpters come along. They see these great tools, and assemble them really nicely, so that everyone can enjoy them. The problem is that there are very few talented sculpters out there, worse still, even if you know what a tool should look like, you may not be able to assemble it (well enough to be user-friendly anyway).

The example of sculpters that is going through my head is Apple. For years they've been looking at other people's work and assembling it very nicely. Yet, in terms of creating legos how "innovative" is Apple? In my opinion, not very (relative to all the work they've done). But they are very innovative at putting the legos together.

A good cathedral (monoculture) should look at the bazaar, and take the best stuff. While the bazaar should be left to explore new tools and ideas. Both have their place (yin and yang kinda thing).

I will also join you in the praising of ubuntu on this front. I believe they are exactly what FLOSS needs.

Re:It's not the choice

[Burz]

Its enough to make someone go back to basics on their C64 ;-)

Seriously, I appreciate and agree with your comments. But more than that, the fact is that FOSS vendors do not promote a default modern development environment that budding application developers would appreciate. Its all geared toward system-level developers (which probably explains why our applications are managed in centralized repositories as if they were just another part of the OS).

Its all terribly backward when applied to the personal computing domain, because influential PC (incl. Wintel and Mac) users expect a clearly-defined platform that encourages the installation and development of third-party products.

Windows has Visual Studio + MSDN as a default choice for developers.

OS X has Xcode + ADC as a default choice as well.

"Linux" (or LSB or whatever distro like RedHat) has...... ?

Right now, FOSS mostly gets major applications that were developed commercially for Windows then ported to Unix for niche appeal before being pushed out of the proprietary market. The home-grown Linux apps tend to have many shortcomings that, IMO, stem from a system programmers trying to scratch the application itch while not really knowing what they're doing... many are trying to fill a void that application programmers won't go near because the latter can't identify a clearly defined and promoted development platform. Also, the former collectively look down their noses at the latter quite intently.

So outside of the systems development sphere, the other important players in the "Desktop Linux" game - end users and application developers - are not seeing a consistent and easily identifiable platform.

Hits the nail on the head...

[Sancho]

We celebrate the diversity of choices available to solve a problem and call it freedom. IT managers and CIOs look at it and call it chaos, confusion and uncertainty. That's a pretty solid difference between the two, and honestly, I think it's probably a good point. Linux desktops compete with other Linux desktops. Gnome competes with KDE. It's still choice, and I still like choice, but fragmentation and a hundred different ways of doing things makes it hard to find the information you're looking for online, makes it hard to support (Helpdesk workers complain about having to support more than 3 versions of Windows!), and makes it hard for the user to choose.

Re:Hits the nail on the head...

[Coryoth]

fragmentation and a hundred different ways of doing things makes it hard to find the information you're looking for online, makes it hard to support (Helpdesk workers complain about having to support more than 3 versions of Windows!), and makes it hard for the user to choose. Right now the choice in Linux seems to be vast and confusing because Linux in general is a niche system, so all the different choices are all equally niche options. As Linux slowly gains popularity (and it will -- if nothing else it is imporving faster than Windows: compare Windows 98 with Redhat 5.2, then compare Windows Vista with Ubuntu 7.04) you won't find the vast array of options all gaining equally from the influx of users. Ultimately, despite the vast array of distributions, only a very small handful (probably 2 or 3 at most) will gain any significant share. For most people Linux will simply be whatever the most popular of those 2 or 3 distros turns out to be. There won't be a plethora of choice, there'll be one distro that everyone you know uses, and then a whole bunch of other niche Linux stuff that only the geeks care about. The same will happen with desktop environments and development libraries: at most 2 options will be supportable as mainstream choices and the rest will be firmly relegated to small niche options -- they'll still be at least as popular as they are now, they will just be completely eclipsed by the increased popularity of the 1 or 2 winners.

Once all of that happens you'll find that, for almost any average user, there is 1 or at most 2 ways of doign things, and all the information online that is easy to find is all relevant to you. Support will be for the 1 or 2 most popular distributions, which will have very standardised configurations that everyone uses. Sure, the hundred different ways of doing things will still exist, and anyone geeky enough to mess with them will find that things are little different than they are now -- its just that most people who aren't interested will never even be aware of all of that. In the same way that most average Windows users are only dimly aware of Linux as a small niche player, eventually we'll reach a point where most average Linux users are only dimly aware of other distros as small niche players. Nothing changes for us geeks, and nothing changes for average users (except the OS they consider "standard").

As if choice is inexistant on MS platforms

[xenocide2]

There's plenty of choice on Windows. The only difference is that these choices involve paying money for things whose worth you can't evaluate until you've used them for longer than a month. Branding helps tremendously in such a situation, as does bundling, both of which MS has in spades.

Some examples of choices developers have on windows platforms:
* IDEs - visual studio, eclipse, netbeans, dev-c++, codewarrior, just to name a few I've used
* The various .NET languages
* Databases
* Webservers, IIS, apache, or something else?
* antivirus, Vista tried pretty hard to end all of these though.

If you're just moaning about how Microsoft has a large vertically integrated set of tools, well, there's Java. Nobody does this, because its stupid and they have the choice not to.

Re:As if choice is inexistant on MS platforms

[casings]

While I agree with your point, these points are more or less choices made during the design of a program or platform and are made either depending on the project or the company. I believe the article is referring to not only these standard design choices, but also the fact that there are literally infinite numbers of configurations using open source software which go beyond simple design choices, and which most firms want some kind of standard over.

AFAIK, managers don't want to have to spend the time making sure the environment is properly setup when that time could be used to make important design decisions.

Re:As if choice is inexistant on MS platforms

[Osty]

There's plenty of choice on Windows. The only difference is that these choices involve paying money for things whose worth you can't evaluate until you've used them for longer than a month.

If you're going to claim that you have to pay for everything on Windows, you probably ought to choose better examples.

IDEs - visual studio, eclipse, netbeans, dev-c++, codewarrior, just to name a few I've used

Visual Studio (Express versions), Eclipse, NetBeans, Dev-C++ -- all free

The various .NET languages

Again, free. You don't need anything but a text editor and the .NET SDK (free) to build .NET applications. Also, other languages like F# (variant of ML) and IronPython (uh ... Python on .NET) are free as well. In fact, IronPython is even open source.

Databases

SQL Server Express, PostgreSQL, MySQL, SQLite, all free. Sure, SQL Server (non-Express), Oracle, and db2 are not free, but Oracle and db2 are not free on Linux either.

Webservers, IIS, apache, or something else?

IIS is "free" (comes with the OS you paid for), and Apache is obviously free as well.

If you're just moaning about how Microsoft has a large vertically integrated set of tools, well, there's Java. Nobody does this, because its stupid and they have the choice not to.

They have the choice not to moan about Microsoft's vertically integrated toolset? Or they have the choice not to use Java?

For independent or small developers, an integrated set of tools isn't really all that important (though nice). For a medium to large business, it's critical if you want to get anything done in a reasonable amount of time without reinventing the wheel over and over again. That toolset doesn't have to be Microsoft, but they do provide a compelling set of developer utilities (Visual Studio is easily one of the best IDEs available for any price, for example). That's what the article was getting at -- when it comes to developers, Microsoft Gets It(tm) (queue Ballmer's infamous "Developers developers developers" video here).

Re:As if choice is inexistant on MS platforms

[Osty]

On an unrelated note, I wasn't aware the .NET SDK was free. Interesting, but I can't see the justification in using F# over Ocaml. Of course, if you're using .NET you might as well get an MSDN kit and visual studio, so I can see where they're going with it.

.NET SDK 2.0. Also, as I mentioned, Visual Studio Express products are available for free for many different uses (app development, web development, database development, game development, etc). If you're developing software professionally, you probably would want to spring for a full Visual Studio product, but for individuals and small projects the Express SKUs work great. XNA's Game Studio Express is built on top of Visual C# Express, for example, and allows you to build games for Windows and/or Xbox 360.

As for why you would use F# instead of OCAML, IronPython instead of Python, Ruby.NET instead of Ruby, etc, is for interoperability. Because the languages are implemented to compile down to IL (.NET's Intermediate Language), they can be used with any other .NET language. For most people it's just a novelty, but it's still cool in a geeky way. BTW, this is no different than building compilers to target languages to the Java VM, like Jython. Why would you use Jython rather than Python? Because it can more easily interact with Java components and other languages compiled to the Java VM. Not a big deal if you're only doing Python work, but invaluable if parts of your product are written in Java and it makes sense to build other portions with Python.

de facto arguments

[Anonymous Coward]

"What the monopoly says, goes. They define a standard. Because they're MS, they define a standard that's different and incompatible with official standards."

Would that be the .NET standard, or the Mono standard?

"Nothing to see here. Market forces and ease of use win over features, stability, or quality."

Well considering the story is about development environments (I'm not certain how desktop environments got dragged into this). Are you saying that Visual Studio is incapable of producing quality code that's stable? Or is two and three a function of the programmer wielding the tool, and not the tool itself?

The reason why everyone uses Microsoft....

[linguae]

...is not because open source offers too many choices. After all, Microsoft's products are still a choice among many other solutions. The reason why much of us still use Microsoft software (to be more specific, Windows, Office, and programming environments related to those two products) is because of a few reasons:

1. Microsoft Windows is simply where the user base is; hence, software is prioritized to be written for Windows. For example, I recently had to buy Windows XP to install on my MacBook because my statistics class required the use of Minitab for its assignments, where no OS X or Linux version exist. AutoCad, a popular engineering tool, is also Windows only.
2. People are comfortable with Microsoft products, and don't feel like switching to alternatives, even if the alternative is technologically superior or can improve their productivity. Why don't they switch? Well, many people figured that they've invested a lot of time learning a product, and they don't want to spend that same amount of time learning an alternative. As much as we geeks wish otherwise, not everybody is very interested in the tools that they use.

Microsoft alternatives do have their merits. To use operating systems as an example, OS X is a great general purpose OS, and I love the customizability of open-source OSes such as Linux and BSD. However, most software is written for Windows (you're guaranteed to find a Windows software package for almost anything, whereas you'll have to search harder when using an alternate platform), and if you have a Windows problem, somebody will know how to fix it.

I admit, I'm not the biggest fan of Microsoft. However, most of us can't avoid their software, like it or not. After nearly a year of not having a Windows machine, I installed a Windows partition to do class assignments. I don't like Windows, but I need to do what is necessary to complete the assignment. For some people, replace assignment with job and add "to pay the bills."

I don't think MS's monopoly will last forever. But, for now, expect to be still using Windows and other Microsoft solutions. When you are in a lion's mouth, wiggle until you wiggle yourself out.

Consumers hate choice

[Synn]

This is why there is only 1 car manufacturer. 1 brand of soda, shoes, toothe paste and so on.

Consumers would get confused if they had to choose from 15 different versions of laundry detergent, so we only have Tide.

Re:Consumers hate choice

[heinousjay]

It doesn't require much investment of time to learn the differences involved in dispensing laundry detergent. They are effectively drop in replacements replacements for each other, from a user's perspective.

As are all your examples, actually. The differences lie in taste, not capability.

You've basically demonstrated that you don't get the point in the article at all. I guess that is why this keeps getting posted, over and over again. It's the sledgehammer technique.

Re:Consumers hate choice

[plasmacutter]

As are all your examples, actually. The differences lie in taste, not capability.

You've basically demonstrated that you don't get the point in the article at all.

I think he got the point perfectly.

Os is also a difference in taste.

I switched to linux from windows because it suited my taste, and to mac from linux because it suited my taste better.

I don't see the difference.. an os is an os is an os.. theyre there to provide basic computer operation, filebrowsing, and a foundation from which to run specialized apps. all 3 of these choices (and many more) provide that functionality.

They use different applications, but then again cars use different parts, different gas.

Now if you want to argue it stems from the american mentality of "technology is for geeks so i dont want to do any of the legwork", that's another story, but his comparison was valid.

i can take any x86 pc and "drop in" linux or bsd as a replacement (or osX if it's a mac).

Lifetime Lock-in?

[MrSteveSD]

Why does Microsoft win the development environment war so often, when we all know it's a lifetime lock-in

It's not a lifetime lock-in when they discontinue your entire development environment and language. Yep, by discontinuing the VB6 language they saved us from that terrible lock-in. Now we are free to re-type those millions of lines of code (and years of effort) in another language on any platform we like. How thoughtful of them.

Spooky

[Rorian]

I was thinking about all of this just yesterday - At work I develop code almost entirely in Visual Studio (98, 2k3 and 2k5) with a little netbeans on the side when I have to deal with Java. I installed Ubuntu 7.04 on the weekend (haven't used Linux in a while, thought I'd have another play). Now I have KDevelop, Eclipse, GCC+Vi, GDB, DDD, KDB, and about 15 other tools that all provide portions of what MS VS wraps up in one neat package, and none of them do it with half the quality. Eclipse is not bad, but 1. It's written in Java, and fairly slow, 2. Debugger integration was average at best and 3. The GUI is overly verbose, borders are too big etc. There are other issues but I guess most of them dissapear with prolonged usage (I'm still not 100% happy with Visual Studio, and I've used that for years).

My point is, Microsoft has made it MUCH easier for developers than Linux, at least for in-house software development. I must admit that there are some benefits to Open Source development tools for distributed development, but not all that many - Svn/Cvs are equally as usable under Windows (if not easier, with tortoisesvn/cvs), Cygwin covers a lot of gaps for GNU-Win32 development, etc.

I don't know if it's the amount of choice or what, and I must admit I haven't used KDevelop in a long time, it may be really awesome by now, but I really don't look forward to the day when M$ explodes and I am forced onto Linux/Mac OS X (I hear the Mac OS X IDE of choice is pretty nasty).

Re:Spooky

[Rorian]

uhuh.. what if I want to provide a full blown application suite? You are right, Linux can do all sorts of wonderful things like that (but then, so can Cygwin under win32.. if html2pdf doesn't exist in Cygwin yet, porting cannot be all that hard) so that point is slightly moot. I've also found that Python is fantastic tool for thousands of such odd-jobs like that, and Python is just as happy under Win32 as Linux.

The fact is that for large application development, Visual Studio IS a fantastic tool and I would use it before any other development environment I have experienced to date, for such a task. As for Windows being a closed platform, care to actually elaborate on how it is closed, and how this closed nature has a negative impact on daily productivity?

I appreciate Linux advocacy, I used to be a Linux advocate myself, and ran Linux exclusively for a long time, but the fact is that I am not going to be Pro-Linux just because it's cool. Windows does everything I want (which generally consists of watching the occasional movie, bit of music, web browsing and the odd bit of World of Warcraft). Linux does all of this as well, but just not quite as nicely, so why bother?

Perhaps I'm with the wrong crowd here, being Slashdot and all, but I'm starting to get bored with the whole "Support Linux because it's not Microsoft" rant - How about supporting the best tool for the job, no matter what that tool is? It's just a little bit more mature.

monoculture bad

[bugi]

Keep in mind that monoculture has disadvantages too.

Variety, for example, is necessary for adaptation, creativity and resistance to disease.

Thanks, we know

[The Bungi]

From TFA:

Guess what? Microsoft has a pretty good development suite on their hands. To be honest, C# is largely what I'd do if I could rewrite Java from scratch with no concerns for backward compatibility. It has a couple of really cool features, like the virtual, override and new keywords that let you specify what should happen when you cast a class to it's base class and then call a method on it that's defined in both.

Guess what? We all know that. We've tried Eclipse and KDevelop and Glade all your other tools. You have pretty cool languages but you insist on keeping your barriers artificially high by forcing a primitive toolset on everyone so that only the anointed few can develop software for your platform.

Of course one of the core problems is someone like this, who from the article seems like the quintessential "Microsoft sux" type-A personality suddenly realizes that Microsoft (and Borland and others) have been writing far superior development tools for the past ten years that actually increase developer productivity and having great success at it. What an idea! Having to learn 14 different tools to get something done might be good for bragging and leetness, but they kill productivity. In the real world, that kills the deal.

Imagine what kind of killer product you would have if you paired Ruby with a good IDE and a good graphical debugger. Or Python. A good front-end to MySQL that's actually easy to use. Or an admin tool for Apache that makes sense. But "ease of use" is not "leet", so no dice. "We don't want VB in Linux". That's a great attitude, and it will continue to perpetuate the idea that Windows is the only "easy" platform to write software in, with Microsoft tools. There's no reality distortion field here - that's just the truth.

So much potential wasted because of a culture that idolizes unecessary complexity as if it were a badge of honor.

Re:Thanks, we know

[mattgreen]

Well put. You see blowhards on here who insist that Linux should be hard to use, because, damn it, it was hard for them to learn it. Or some tripe about how computers should require a license to use. And it amuses me when I inquire why OSS alternatives have a hard time parsing C++ for autocomplete: "it's hard to do." Cry me a river, of course it is. I thought the many eyes of open source could solve *any* problem? In reality, nobody wants to, so it stays in a state below half-assed. Visual Studio limps on by and at least gets right enough to be useful.

Visual Studio is far from perfect, but for a tool that I used day in and day out, I don't have many complaints overall. How many programs can you really say that about?

Re:Thanks, we know

[Llywelyn]

Imagine what kind of killer product you would have if you paired Ruby with a good IDE and a good graphical debugger. Or Python.

Eclipse w/ PyDev is fairly killer and not difficult to set up. Objective-C w/ XCode and Interface Builder is also very nice to work with.

Re:Thanks, we know

[MechaBlue]

C#.NET is wonderful, far better than Java.
- Exception handling is much better. Rather than having the compiler complain when I forget to handle an exception, it encourages me to wade through many pages of docmentation and manually trace it.
- Generics have a great naming convention. Instead of being able to remember 1 name and interface for a class, I need to know 2. It even encourages me to learn both by forcing me to modify sizeable portions of the code when moving from non-generics to generics.
- The documentation is so much better than Java, too. Rather than have 1 page per class and making me use the page-up and page-down keys, it lets me use the much more convenient type-to-search-and-click-through-6-pages approach.
- It's great that the solution file changes every time that I pick my nose. It makes using Subversion and CVS so much easier, especially with a large team.
- The kept all the good stuff from Java, such as: ArrayList arrayList = new ArrayList();
- Also, I don't have to worry about another creating a non-standard implementation of the CLR and trying to lock-out clones. Microsoft can do this all by itself.
- I don't need to worry about trying to get .NET applications working on non-Windows platforms because MS has only released a reference implementation for Windows.
- I don't need to worry about adding in code generation for repetitive tasks because there doesn't seem to be any facility for it.

I'm sure that there are workarounds for some of these issues but, really, I don't care. I learned Java a decade ago and, if I want to program in something like Java, I'd rather do it in Java than learn C#.NET. Especially when I can use vim or Eclipse to crank out a Ruby or Bash script to tie together a few simple programs. C#.NET is Java that tried to not be Java, but only succeeded in the bad ways.

Admittedly, Visual Studio 2k3 and 2k5 have advantages over Eclipse. The comments in C# are somewhat nice (// comments are better than /* */ comments when commenting out large blocks of code, though the XML thing is really verbose). The word-completion is totally sexy. I could do without the curious glitches, though...

Re:Thanks, we know

[QuantumG]

You aint just whistlin' dixie brother. I'm an experienced Open Source developer (and professional software engineer) and I just refuse to touch autoconf, etc. As for rpm/deb packaging? Someone else, do it for me, please.

Maybe

[Timesprout]

Just maybe its that until quite recently the Microsoft development environment was vastly superior to anything being offered by the Open Source community. There were quality development and debugging environments from MS and Borland. More dev editors than you could shake a stick at. There was easy integration with multiple databases and it was easy to develop slick front ends to this data. There was tooling availabe for easy project management and application testing.

Maybe Microsoft actually copped on to the fact that businesses wanted tools to build the apps they needed while the Open Source community were patting themselves on the back about how cool and fantastically leet they were for having text editors and shell scripts.

Whats interesting is that the two current leading Java Open Source IDEs (Eclipse and Netbeans) are both tools which started out life intended as commercial offerings but were donated to the community by IBM and SUN.

Self-defeating argument

[Gregory Cox]

The author of the article shoots himself in the foot here:

Should I use iBatis or Hibernate? XFire or AXIS? Perl, PHP or Ruby? Debian, Fedora, Ubuntu or Suse? Make the wrong decision, and you can waste a ton of time, as we found out on a recent project when we wasted a week try to make AXIS2 work for a web service project, only to find out that XFire was the right choice.

So by his own admission, XFire and AXIS don't compete with each other directly - they're different tools for different jobs.

The solution to his problem is not to get rid of one of them, but to put something in the documentation for AXIS saying "If you're working on a web service project, you may be better off using XFire." The fact that an open source project is able to recommend another solution like this is a strength. In contrast, a company (in this case Microsoft) with a vested interest in promoting its own monoculture is unlikely to tell potential customers to go elsewhere.

Conversely, if you have a choice of near-identical tools, any of which would be acceptable, why not just pick the one which is most popular at the time? This leaves you at the mercy of programming fads, but if you want a monoculture, you should expect that anyway.

Re:I hears yah

[Wonko the Sane]

"This one time, I tried to use a 1/2" socket wrench and wasted a bunch of time until I realized that I actually needed a 9/16" box end wrench"

"Having this many different tools is too confusing. Instead of socket wrenches, box end wrenches, open-end wrenches and hammers, we should just use crescent wrenches for everything"

This is dumb

[Tim_UWA]

So what if I have a choice between openoffice.org, abiword or LaTeX to produce my documents? As long as I save it as a PDF, anyone can read it. If I use Microsoft Word, I either install a 3rd party program to save it as a PDF, or require that the people I send the document to have the same version of Word as me, running on the same platform. Disclaimer: I didnt' read the article.

You're looking at the wrong market

[Rix]

The people paying money for commercial Linux distributions aren't Uncle Ed and Aunt Martha. They're corporate or government IT departments, and they don't need every setting available to the GUI.

What you want requires a commercial distribution target for pre install on low end consumer grade equipment. Even ignoring the Microsoft tax, such a proposition is a bit shakey, financially.

Here's another reason...

[davmoo]

A major problem, in my opinion, that a lot of open source products have is the lack of decent documentation. Everyone wants to code, no one wants to document. I'd be here all day if I tried to list all the open source products I've looked at and tried that I gave up on simply because there was no documentation, or only poor documentation, and I had to move to something commercial that was at least properly documented.

And before someone drags out the dead horse named "why don't you document some of these projects", the answer is I can't document something I can't figure out how to use. And as others have already pointed out, a lot of open source software is not intuitive.

Apache vs IIS

[RalphTheWonderLlama]

I prefer Apache since it's free but....

This goes for virtually every non-default configuration of Apache and IIS but here are a couple examples.

Allowing only certain IP addresses to access a website:
Apache -
1. Research on the web how this is done using Google.
2. Find something called "mod_authz_host" and an example of its use here http://httpd.apache.org/docs/2.2/mod/mod_authz_hos t.html .
3. Get confused by all the different examples.
4. Attempt to insert "code" into httpd.conf to limit access to certain IP addresses.
5. Test to see if it worked.
6. Research, edit, test more as needed.

IIS -
1. Click a few buttons
2. Enter IP addresses allowed.
3. Test (it works first try as expected since editing was intuitive).

Use SSL:
Apache -
1. Research on the web how this is done using Google (lots of research).
2. Install something called OpenSSL
3. Copy a few files to a windows directory
4. Find an openssl.cnf file that doesn't exist with the OpenSSL install for some reason.
5. Create a SSL certificate using command line.
6. Due to legal/political constraints, download a different copy of Apache with SSL from a strange 3rd party website and replace current copy of Apache that you had installed.
7. Make several changes to httpd.conf file.
8. Install this new Apache as a service using command line if needed.
9. Make several more changes to httpd.conf file (uncommenting LoadModule line, including ssl.conf in an IfModule thing).
10. Copy the certificate files made earlier to an Apache directory.
11. Edit ssl.conf file on several lines to identify server name, document root directory, then also include the certificate path.
12. Restart Apache, pray it works.

IIS -
1. Go to website properties using GUI
2. Click Directory Security tab
3. Click Server Certificate
4. Follow Web Server Certificate Wizard to create certificate.

For extra credit, require SSL connection - In Directory Security tab, Secure Communication area, click Edit, and check the Require secure channel SSL checkbox. I gave up on that for Apache and figured out some way to just forward requests to https (a bit of a hack it seems).

Things just seem more intuitive when using IIS rather than editing conf files and hoping things work in Apache. There is a lot less frustration. It's a shame. Yes I did look for 3rd party Apache config GUIs and couldn't find anything that looked good.

Re:Apache vs IIS

[dodobh]

There is Webmin as a GUI for Apache.

On the other hand, once you have things working:

Commit Apache config to SVN. Put common config in one file, per virtual host configs in their own files.
Checkout config(s) on new box, with automatic merging of base and per virtual host configs.
Reliably roll out the same configuration multiple times, across host upgrades, reinstalls and datacentre moves.

Do the same thing for every service you run. One host? Ten? A hundred? A thousand? ...

The end result of the difference between the Windows philosophy and the Unix philosophy is that Unix adds a bit of a startup cost, but keeps operating costs low.

Re:Apache vs IIS

[hey]

Lets say you have a virtual host on your Apache box and you want to clone it with minor mods.
Easy... cut'n'paste in the config file. But on IIS... I wouldn't know where to begin.

Re:Apache vs IIS

[AlexTurner]

I have run both IIS and Apache. 100% correct,IIS is easier to set up. However, it has proven very much harder indeed to make it 'bomb proof'. There is a critical difference between open source and close source that we seem to be missing here. If you write open source, people can see your source code. If the code is for a mission critical component (like a web server) then people are _really_ going to care that it is well written!

I have a motto about the differences in systems running on Linux (or FreeBSD etc) and Windows. On the xnix platforms, it takes days to set up and it runs for years without crashing. On Windows it takes minutes to set up and runs for hours without crashing.

Oh - and one other little point. It is very interesting to see the word 'monoculture'. One of the major challenges for farmers when they grown monocultures is preventing the spread of disease (like viruses) because lack of differentiation of species makes it so easy for disease agents to spread. Sound familiar?

Up to Vista, I found Windows best for home and business functions. Linux/FreeBSD was best for 24/7 servers. Now I thin ubuntu has taken the lead for the home as well. But the barrier for the business systems is much much higher, the alternatives will really have to pull their socks up if they are interested in entering the market in a large way.

AJ [www.nerds-central.com|nerds-central.blogspot.com| twitter/AlexTurner]

Re:Apache vs IIS

[DaleGlass]

I dread working with MS GUIs. Everything has 20 tabs, sometimes pops up extra windows and requires half an hour of clicking. It's not possible to figure out what the full current config is without clicking everywhere. Then there's my favourite annoyance: Fixed size windows with list views that sometimes require scrolling to see all the information in them.

Take for instance setting up a DHCP server.
Task: Switch 100 hosts from static IP address to DHCP, configured to give them the same address they currently have, by MAC.

Windows: Hand falls off due to clicking for an hour, and mindlessly copy/pasting. Error prone. MS DHCP is completely braindead by the way, and seems to confuse the description and MAC fields (did anybody even test that?)

Linux: Maybe 5 minutes to write a script that: pings subnet, parses output of 'arp -a', and automatically generates a dhcpd config file. If the hosts were Linux ones I could also automatically replace the network settings on all of them. If it goes wrong, fix script then run again. Not all hosts are on? No problem, I can tweak the script to run periodically, detecting new computers and adding the config for them.

Yes, GUIs are easier for tiny tasks, but once you need something uncommon they're a huge pain in the ass. And it's not like there can't be a GUI that generates dhcpd.conf either.

Re:Apache vs IIS

[Mista2]

Horses for courses. One day I was off sick but there was a work project on that required the setup of a webserver to host a php based application. The project had a tight deadline, so I worked from my sick bed 8) Get laptop and connect to internet: working over crappy DSL line, 2Mb top speed 8( run ssh to get to a suse server I had installed earlier in the week for the project, (the server was also running some other services so could not be restarted during the working day) Use Yast to install apache and php requiremements from up-to-date suse internet repository. Use vi to edit the apache config to create virtual server. Use rsync to copy the php application to the server (for windows users, rsync is a little like FTP, but over an ssh connection, and can easily replicate a full folder structure). Test, debug, then secure access to only http, https and ssh. Approx 4 hrs for the whole job. To do the same thing in Windows - Take Previously installed Windows 2K3 server, enable remote desktop access, oh hang on, that wasn't done initially so will have to call someone at the office to do it for me, Oh, and have to install whole citrix gateway infrastructure to support secure encrypted remote access. Add software - IIS, hang on, no network repository for i386 directory, get someone at the office to stick in a CD, then restart. Add php services to IIS - umm, haven't done that before so I don't know how hard it is to do. Run Windows Update to patch - reboot (current installs I have done from Win2K3-R2 CD usually require at least 2 restarts to get fully up-to-date) No easy way to copy the full application to the server (as no way I am opening up MS FTP to the internet without a proxy 8)) so tar or zip it up, mail to the client, get someone to copy tar file to the server, then login and extract it. I think the point is here that the job would have been nearly impossible for me to do that day had the host been a Windows host with IIS. It would have requried the server to be restarted to patch and install software, and none of the secure remote access is available out of the box. Apart from a kernel patch late last year, the web server has never been restarted. Apache has been restarted a few times, but it hands over connections to the new httpd as it shuts down so users don't really see an outage if it's done right. Oh, yeah, I also have a dev server with the same apache config. When I make a change I test it in dev, then copy over the conf file before restarting. Anyone know how to copy IIS settings from one server to another? Please let me know.

Won't be long now.

[Bluesman]

The writing is on the wall for Microsoft. Sure, they've had the developers in the past, but I think it's only a matter of time before they jump ship.

Lots of people aren't too happy with Vista, and OS/X, Linux, and BSD are gaining ground.

The keys to the kingdom are backward compatibility, always an MS strong point. But the open-source community has this nailed -- the POSIX API's are over thirty years old, and won't change. It's only a matter of time before Microsoft changes just enough so that developers can't rely on them to be rock-solid anymore, and there will be a mass exodus.

It will start with specialized apps like video and audio editing software (already happened to some extent) and gradually will work its way through the entire business suite of tools.

Some years ago, I offered to write a Postgresql database app for a small business. They refused, saying they wanted to use SQL Server, even if it was more expensive, since "Microsoft isn't going anywhere anytime soon."

Today, I doubt there would be that same certainty. Ten years from now, I expect the tables will have completely turned, with Linux based apps seen as the "old reliables" that never go away.

Oh, god, what bullshit!

[Russ Nelson]

Oh, god, what bullshit is this! Choice is not a problem! If choice was really a problem, then somebody would create a Linux distro with no choices. "Sit down, shut up, and run the software we choose". Except, nobody does that because nobody wants that.

The whole "choice is bad" meme is complete and utter nonsense. http://angry-economist.russnelson.com/barry-schwar tz-master-chooser.html

The truth about Kdevelop

[QueePWNzor]

I'll admit, the difference between Kdevelop (KDE's IDE) and Anjuta (arguably the GNOME alternative) are trivial but too often annoying in reality. But there frankly is a reason why I, among others use Kdevelop: It simply looks better than everything else in KDE. Now, of course, because it is linked in deeply with KDE as a whole, unless you upgrade every time a new KDE release comes out, the development libraries&junk may fall behind. I upgrade about every 6-8 months, but I see how, if somebody didn't, that could be a huge pain. But Anjuta, though it uses GNOME/GTK+ libraries (GTK+ is used for almost all programs, but the GNOME variants can cause trouble), it isn't so deeply linked in. So this deepens the debate, especially since the two environments go forward at often random paces. When I started using GNU/Linux, I got frustrated at this, and it caused me to use my Windows IDE, Microsoft Visual C++ more often. Because it has its own formats for organization, I started becoming more dependent on it as I added more headers and crap. That is why Kdevelop gives Microsoft market-share. I think, but am not sure that KDE is the most common desktop environment. Ubuntu may have altered the balance, though.

Things to learn from Windows and OSX.

[fyngyrz]

Oh, it's helpful, all right. For instance, the only reason we've not released a port to linux - a free version, of course, we'd like to give back to the community - is because there is no standard GUI layer. It's a hodgepodge of these widgets and those widgets, this license and that license (really meaning, these liabilities and those liabilities.) Windows provides all that. Free. Built in. Plus a large market. So we developed for them. When Windows became intolerable because of activation DRM, we moved to OSX. Nice GUI layer, free, built-in. development proceeds apace, while linux runs servers. Others may have other reasons, but those are ours. The day the linux core gets BUILT-IN windowing and graphics, and I do NOT mean just xwindows or xwindows plus yet another sometimes-there and restrictively licensed widget set, is the day we make a port that we will release to the community. The community can then, of course, use our stuff or not as they see fit - but as is, it's not a choice. That's been the unanimous decision of the linux community: no coherence.

I want to say one more thing. The existence of a standard GUI layer in NO way means that you can't still have everything you have now. You'd just have one more thing, something people could write to as a default, even just as a fall-back.

That's my 2 cents.

Re:Things to learn from Windows and OSX.

[Kristoph]

You seem to misunderstand that Linux is not an operating system, it is a kernel. The various 'distributions' are operating systems built upon the Linux kernel. In the vast majority of cases the distributions which target client solutions offer stable UI layers for any given version.

So you could, for example, create a solution for Ubuntu - currently the most popular client distribution - which includes a windowing system, widget libraries, desktop environment, documentation environment and everything else you may need.

]{

Re:Things to learn from Windows and OSX.

[Simon80]

If you think that it's incoherent for X windows to be in a different module than the kernel, there's something that you don't get. I understand that there's a confusing amount of choice for someone who wants to develop Linux software, but there never will be some kind of "BUILT-IN" Linux(tm) windowing system, the concept just doesn't make sense. However, it is true that a unified API that eases porting between different GUI kits would be nice.

Re:Things to learn from Windows and OSX.

[BlueTrin]

You would then target a fraction of a smaller customer base.

Re:Things to learn from Windows and OSX.

[Osty]

If you think that it's incoherent for X windows to be in a different module than the kernel, there's something that you don't get.

"Built-in" does not have to mean built into the kernel. In the case of Linux, "built-in" might mean "available by default in every distribution without having to do anything special". Maybe that's KDE, maybe it's GNOME, maybe it's something else, but it has to always be there (alternatively, you have to really work at making it not be there, in which case you're probably not the target audience).

However, it is true that a unified API that eases porting between different GUI kits would be nice.

I don't think a unified API solves the problem, so long as projects are allowed to implement however much or little of it as they want. What is needed is a single, de-facto GUI widget kit that is ubiquitous across all distributions of Linux. Or at least provide a single easy way for an application to force the installation of that toolkit if it's not there already (and no, debs, rpms, pkgs, and whatever else don't satisfy that criteria because there's too many of them -- one installation method, one dependency resolution method, one toolkit. Not 10 different installation methods, each with their own dependency resolution models, for 20 different toolkits).

Re:Things to learn from Windows and OSX.

[EsbenMoseHansen]

is because there is no standard GUI layer.

I don't know whether to laugh or cry. You can use any GUI layer you'd like, and it will work everywhere. Want to use QT? No problem. GTK? No problem. Athena? No problem. It all works. In fact, on any given day, I use applications that uses different toolkits (e.g, firefox, SWT, konqueror) and it never bothered me. Yeah, the open file dialog problem still exists, but there is hope in that. And someday, the look will be unified too, have patience. Neither of those are really important, though.

So if that was your reason for not porting, you didn't port for the wrong reasons.

Re:Things to learn from Windows and OSX.

[salimma]

What is needed is a single, de-facto GUI widget kit that is ubiquitous across all distributions of Linux.

Hmm. As far as I know, most Linux distribution comes with GTK+ and Qt installed by default, regardless of whether they use GNOME or KDE as their default desktops. And freedesktop.org is working at standardizing more parts of the desktop..

As far as licensing goes, with GTK+/GNOME you can develop proprietary applications; with Qt, if you want to do so you have to buy a commercial license, and with KDE.. you cannot. Simple as that.

Re:Things to learn from Windows and OSX.

[pwizard2]

The day the linux core gets BUILT-IN windowing and graphics, and I do NOT mean just xwindows or xwindows plus yet another sometimes-there and restrictively licensed widget set, is the day we make a port that we will release to the community.

I think that merging the kernel and the windowing system is a b> terrible idea, and here's why:

1. It needlessly complicates the kernel, meaning that there are going to be more bugs/exploits.

2. Isolating the kernel from the windowing system is good because if X breaks, I can still use the kernel via a bash shell, which would probably give me the functionality that I need to repair/reinstall X. If the kernel and the windowing system were merged, I probably wouldn't have a fully functional kernel if something happened to the windowing system. The only option then would be a reformat.

Re:Things to learn from Windows and OSX.

[eihab]

Then users demanding "Linux" applications are demanding something that will never happen since you can not write an application to a kernel!

Sarcasm aside, I am (and have always been) shocked at how various Linux distributions differ. I'm by no means a Linux expert, but come on, if you can't agree on where to put your init0.d-init6.d (or whatever it is, yes I'm looking at you Gentoo) then how do you expect an outsider to write something that works for "Linux"?

I mean we hear all this crap about Microsoft and how ME breaks 98, and 2000 breaks ME and Vista breaks XP, etc.
But do the math, how many operating systems are we talking about here? now compare that to Linux distributions available today.

I'm not trolling, I'm no shill and I definitely do not work for Microsoft.. but I just really believe that "too" much choice sucks.

I just (literally) bought a new house (closed escrow today) and I've been in the carpet choice dilemma for the past two weeks! (mind you it took me a while to settle on carpet instead of hard wood floors, pergo, etc.).

Come to think of it, I think this carpet analogy (that I'm about to write) is right on money here. OS is like carpet! It's the foundation that you build on. Think of how carpet pretty much dictates what kind of furniture (style/color etc.) you'll be placing on it, OS will do the same with your applications.

I would have a much easier time choosing between 2-3 carpet types and 10-20 colors than the amazing number of styles/options out there (visit the carpet section next time you're at home depot and you'll know what I'm talking about).

Things like carpet and operating systems (I can't believe I'm lumping them up in the same sentence) are not mundane. They are a *huge* deal because of the consequences of the choice (much like posting this without hitting the "Post Anonymously" checkbox). They're something that you will invest a good amount in, and that will be with you for a long time.

Choice is a VERY good thing, too much choice may even be perfect for those who know "exactly" what they want.

I live and breathe computers, and I think it's silly that I'm spending all this time researching "the" perfect carpet for me and my family. I also believe that the home decoration specialist that knows what kind/make/model/color/class of carpet they want for the hallway of their second house (and have it written somewhere) would be as frustrated as I am with carpet if they had to choose from the plethora of Linux distributions out there today.

The choice is even worse for software companies that put money or even the whole company on the line when they make a platform choice.

I'm done ranting, gotta hit the sack and hope that I come to a decision about the carpet before we get the keys!

Side [OT] (as if this whole thing wasn't anyway) note:
If you happen to know something about carpet, or where I can get more/good information please post it. No pets, one year old and an average budget. Thanks in advance :)

Re:Things to learn from Windows and OSX.

[Anomolous Cowturd]

Bah. KDE-based skype works just fine in a gnome session. It even puts it's status icon in the appropriate place on the gnome panel.

The only built-in thing in linux is the kernel, and most of that is optional, and that's the way it's going to stay. So "no built-in whatever" will always be true. So what. Require X11 for your app. Pick a common gui toolkit and use it. Compile statically if you're worried. Your reasoning lacks merit. Find a better reason (there are plenty) not to support linux, or shut up.

Re:Things to learn from Windows and OSX.

[ozmanjusri]

how do you expect an outsider to write something that works for "Linux"?

Provide the source code under an open license. It's how we want it to happen.

Re:Things to learn from Windows and OSX.

[cortana]

Why? Are non-Ubuntu users unable to run GTK applications?

Re:Things to learn from Windows and OSX.

[ErroneousBee]

When I did some (about 1 hours worth) research into this, I came to the conclusion that the only 2 obvious choices were wxWidgets and GTK (perhaps SDL if we are talking games).

Seeing as both are available on windows/mac/linux, its really hard to understand why you are saying that there are too many choices. There are two, and they port to all linux distros as well as Win/Mac.

Re:Things to learn from Windows and OSX.

[SQLGuru]

Honestly, though. How many non-/. reading, mainstream users are going to install another widget system on their computer? Few or None. Even if you port your free application to a "Linuxy" version of Windows or OSX, no one will come. If you want my mom to use your program, it better come on CD and run on Windows. If you want my dad to run it, it better come on CD and run on Windows. My sister, well, she'll download it, but it better run on Windows.

Back to the original question, the reason that so many companies choose the Windows development? It just works. I can install MS Visual Studio, connect to MS SQL Server, pull in web services off of IIS. And it just works. I call developing in a Microsoft environment "pointy-clicky-draggy-droppy". You don't really have to think about too many config files or whether something works together. It may be dumbed down, but as a developer, I don't want to really have to think about configuration, I want to think about code.

My primary development environment right now is Eclipse and Java. I hate it. Why? Too many choices and no way to know which is the right choice. I have to deal with this configuration and that configuration and a few .properties files thrown in, too. I spend about six hours each week (sometimes more) fighting the environment I work in because someone made a choice that turns out not to have been the best one (for us - I'm sure for someone else it was the perfect choice).

Layne

Re:Things to learn from Windows and OSX.

[TheRaven64]

Mac users don't complain that FireFox doesn't look like a real OS X app, they complain that it doesn't feel like one. If you want to see what I mean, try double-clicking on the address bar, or using any of the keyboard shortcuts for moving around it that work in every single text box on the system except FireFox. Qt apps tend to be just as bad, which is why I avoid them like the plague.

A lot of the reason OS X is easy to use is motor memory. Applications all use the same shortcuts, so your brain thinks 'perform this high-level action' and then your fingers do it without any more input from your high-level brain functions. There is much less thinking 'what key do I press to do this?' than I've found on any other platform. Last time I was using a *NIX desktop, I had three different applications open with three different ways of skipping to the end of the line in a text box (not counting Vim, because I get enough visual clues to expect terminal applications to behave differently).

the Apple software doesn't really seem to have that much of a unified look anyway You're probably not consciously aware of this, but Apple software has three distinct looks, with three subtly different behaviours. This gives you a subconscious visual clue, which allows you to expect the correct behaviour, often without thinking about it. In spite of this, core behaviour is the same across applications. One of the worst design decisions the major Linux distros made was to theme KDE and GNOME applications so that they looked the same. They don't behave the same, since KDE and GNOME have different UI philosophies, and so you immediately destroy the visual clue that they will behave differently.

Re:Things to learn from Windows and OSX.

[dosquatch]

You seem to misunderstand that Linux is not an operating system, it is a kernel.

I swear to FSM, I'm gonna go postal the next time I see this argument. Come over here and let me beat you with a clue.

The vast majority of computer users are not developers, code warriors, uber-geeks, or anything of the sort. All they want is to check their mail, browse the web, and run a word processor. They don't understand this distinction. They don't want to understand this distinction. So what happens is that they call it whatever they hear it being called.

It's akin to pointing to a parking lot and calling all of the vehicles in it "crankshafts", and then berating some hapless would-be driver for not understanding the distinction that that happens to be just one critical part in the overall makeup, and even though you casually call them all "crankshafts" the vehicles are likely to bear little semblance to one another in form or function.

Yeah, I know. Another car analogy. I'm trying to be feel apologetic...

Re:Things to learn from Windows and OSX.

[vdboor]

because there is no standard GUI layer. Windows provides all that.

No, it does not. Well only sort-of.. The "standard GUI layer" of Windows is limited to the plain widgets we all know from Windows 95. The ones Notepad and WordPad still use. Ugly menu's and big bevel toolbar buttons. If you look closer you'll see Notepad, Windows Explorer, Visual Studio, Office all use different menu's and toolbar handles. They're all custom widgets, not standard.

Most advanced widgets for Windows are part of a commercial widget toolkit you've chosen. This can be MFC, ComCtl, VLC (Borland), Windows Forms (.Net), WPF (.Net3), Qt, and I'm missing others (e.g. remember those big sized OK-buttons a big green check icon inside).

All those different frameworks do have something in common. Windows provides central settings for fonts and color schemes. This makes them all look the same. That's something Linux should really improve.

his license and that license (really meaning, these liabilities and those liabilities.)

You have two good options for Linux:

• GTK+. Free for use in commercial projects (LGPL). It's the base of GNOME.
• Qt. IMHO a enterprise class toolkit (see customer list). Requires a license for commercial work, but I don't see how that's different from a license for Visual Studio. And you'll get commercial support in return too. Qt is the base of KDE.

Re:Things to learn from Windows and OSX.

[demallien2]

May His Great Noodliness banish thee to the realms of meat and three veg for yet another car anolgy. No parmesan for you!

Re:Things to learn from Windows and OSX.

[Man On Pink Corner]

Provide the source code under an open license. It's how we want it to happen.

It's not, however, how USERS want it to happen. Hint: if installing your program requires me to run a compiler, I'm just going to stikc with the closed-source Windows version, and maybe get some actual work done that week.

Re:Things to learn from Windows and OSX.

[Teach]

Choice is a VERY good thing, too much choice may even be perfect for those who know "exactly" what they want.

But as you say, too much choice is crippling. To quote author Madeline L'Engle: "Freedom is a terrible gift, and the theory behind all dictatorships is that 'the people' do not want freedom. They want bread and circuses. They want workman's compensation and fringe benefits and TV. Give up your free will, give up your freedom to make choices, listen to the expert, and you will have three cars in your garage, steak on the table, and you will no longer have to suffer the agony of choice."

Re:Things to learn from Windows and OSX.

[99BottlesOfBeerInMyF]

Then users demanding "Linux" applications are demanding something that will never happen since you can not write an application to a kernel!

The truth of the matter is, you can easily write a "Linux" application that runs on pretty much all major Linux distributions.

I'm not trolling, I'm no shill and I definitely do not work for Microsoft.. but I just really believe that "too" much choice sucks.

I disagree. The real problems we have with many choices are all because we have a monoculture. Because there is one dominant OS and that OS is intentionally incompatible with other things, fragmentation is an issue. If we had four OS's sharing 75% of the market and another 10 OS's sharing the rest we'd have significant investment in all the major ones and in technologies that allow for good cross-platform development and portable applications so it did not matter which OS you are running as much, since it would preclude your using a given application.

I would have a much easier time choosing between 2-3 carpet types and 10-20 colors than the amazing number of styles/options out there (visit the carpet section next time you're at home depot and you'll know what I'm talking about).

I picked out some carpet a month ago. Like any other decision, I had criteria. It needed to go with two different wall colors and needed to have a pattern to hide the stuff tracked in from outside and the garage. If there were only 10-20 colors, I would have had a carpet that was less suited to my needs than what I did get, since I doubt one of these 20 colors would have been a combination of colors complementary to my walls along with a dark and a neutral to help hide dirt and the like. If you don't like having choices, then just pick from a limited selection from one company or collection. I don't see why this is hard. Variety and competition breed innovation and improvement and allow for a better fit to a given task. It is only when we have a monoculture that the problems appear and one vendor can intentionally make sure you have no choices by making sure floors are either compatible with their carpet or are weird and non-standard ones that do not work with anything other than one specialty floor.

The choice is even worse for software companies that put money or even the whole company on the line when they make a platform choice.

So it is better for a company to have only one or two choices, than many choices to find the best fit for their needs? I'm not sure I understand that argument at all.

If you happen to know something about carpet, or where I can get more/good information please post it. No pets, one year old and an average budget. Thanks in advance :)

I don't know a lot about carpet, but I recently did some research and consulted someone who knew a whole lot. I'm happy with the result. Consider what it will look like in two years given the type of traffic you have. Consider changing trends and what will quickly look dated. Spend the extra money for thick, high quality padding underneath the carpet as that makes a whole lot more difference for the price differential than the cost difference between the carpets themselves.

Re:Things to learn from Windows and OSX.

[dosquatch]

So you're saying if my mom starts to call cars "crankshafts", I shouldn't correct her? Or is your analogy exactly backwards from the point you're trying to make?

I'm saying monkey see, monkey do. She's calling them crankshafts because she sees you and a bunch of your buddies calling them crankshafts. I'm saying "if" she's wrong, maybe you should set a different example rather than split hairs and confuse her with information she doesn't need, while continuing to do the same thing for which you're "correcting" her.

Nevermind that I'm not completely convinced she's wrong. Nomenclature changes, and it might be that you don't have a lot of say in that change. Majority rules. We call them "cars" now, but they started out as "horseless carriages". If the majority of folks mean a complete OS when they say "linux", you might just have to suck it up and deal lest you find yourself railing against the wind.

And don't kid yourself, the distinction doesn't really matter. (Blasphemy!) No, but it's true! The kernel, without a platform, is pretty useless. A platform, without a kernel, is much the same. There's an intrinsic linking there that you can't put down as (or with) semantics, and the simple fact is that the demarcation point between the two only matters if you're coding parts of the system. To everyone else, it's just geeks making noise which cues the glazed eyeballs.

Re:Things to learn from Windows and OSX.

[amazon10x]

Yes you do; you just don't know it. You use linux because it works and because it costs no money. Those two aspects hinge upon freedom.

Re:Things to learn from Windows and OSX.

[Phisbut]

As far as licensing goes, with GTK+/GNOME you can develop proprietary applications; with Qt, if you want to do so you have to buy a commercial license, and with KDE.. you cannot. Simple as that.

You got misinformed there. You can develop proprietary applications with KDE. KDE is LGPL'd, and the LGPL allows proprietary applications. The caveat though is that KDE links to Qt, and Qt is either GPL or commercial. So in order to develop a proprietary application with KDE, you need a commercial license for Qt, but it is possible.

Re:Things to learn from Windows and OSX.

[lordtoran]

Who was suggesting freak stuff like a manual compile? Availability of source code allows the distributor to customize the software and provide a package that plays nice with his particular Linux distribution. This is the usual way software for "Linux" is released. Publish the source and leave the detail work to the community.

Re:Things to learn from Windows and OSX.

[Chris Shannon]

.. but I just really believe that "too" much choice sucks.

Very true. This is a very informative Google TechTalk called The Paradox of Choice - Why More Is Less [warning 1 hour] that describes this issue. It's not intuitive, but is logical. It is very hard for people to make a choice, so hard, that often people will avoid making a decision when doing nothing is actually more costly than making a bad choice. This talk is back up by results of some interesting experiments.

Re:I hears yah

[bug1]

I went to bottle shop the other day to buy some beer, to my surprise they had 100 different types, i really enjoyed sampling them all at the time, but today i am sick and bloated.

Just because you can have something doesnt mean you should.

Re:I hears yah

[Helldesk Hound]

> "Having this many different tools is too confusing. Instead of socket
> wrenches, box end wrenches, open-end wrenches and hammers, we should
> just use crescent wrenches for everything"

But sometimes you need a spanner.

Sometimes you also need a screw driver.

And sometimes you need a tool designed specifically to get into that awkward spot that no other tool can get into.

tools are precision instruments designed to manipulate precision parts without damaging those parts.

If you have a one-tool-fits-all approach then you only end up reducing either the variety or the quality of what you can produce.

Because everyone uses windows

[Billly Gates]

Which development environment offers the best integration with MS products? Surprise its visual studio.

If you have a windows shop your stuck with it. Borlands tools do not include things like the VBA for office.

Not that simple, sadly

[Moraelin]

This article assumes that open source developers are aiming at becoming Microsoft like. Maybe they're just in it to make good software: not a profit, not make money for shareholders, or anything that that Microsoft is obviously aiming for. And the article is also using a very narrow definition of "win", one which I'm not sure is possible for OSS to attain.

And that is assuming that OSS developpers are a bunch of nerds in their free time, doing software just for fun. Sad to say, his assumption is closer to reality.

Oh, there are thousands of 1-2 man projects on sourceforge done by enthusiasts in their free time. Chances are you haven't even heard of most of them. They also tend to be small projects.

If you look at what's in your favourite Linux distribution, though, it's a different story. Look at the kernel credits some day. You'll see a lot of people from IBM, Red Hat, etc. Hate to break it to you, but they're doing a paid job there. Others may not be employees at such, but got paid/sponsored by a corporation to develop that stuff. E.g., ReiserFS was pretty much paid for by SuSE.

Other programs there? Mozilla? It even got started because Netscape wanted a browser that can stop MS's onslaught onto their business. Then it got bought by AOL, and nowadays it's Google footing the bill. Open Office? Got started as a proprietary project, then bought by Sun. Nowadays it's Sun doing pretty much the whole work, with people paid to code on OOo. It's costing Sun a lot of money. Etc.

See, the F/OSS that gets taken anywhere _near_ seriously these days is the work of corporations. Pretty much it's just a framework for a bunch of corporations to pool their resources into fighting MS. None of them has the resources to challenge the behemoth single-handedly, and some have already lost against the behemoth when trying to "solo" it. E.g., ask IBM what happened to their OS/2.

Where this long rant is going is: of _course_ those corporations are aiming at becoming the next MS. In fact, some of them were the original (near)monopoly long before MS. IBM used to be _the_ name in computing business, long before MS even existed. (And incidentally was just as underhanded as MS. The term "FUD" was first used to describe IBM's tactics, long before MS even existed.) Sun was _the_ name in professional micro-computers. Etc.

And some of them suffered quite humiliating defeats at the hands of the "beast". IBM created the PC, and everything had to be "IBM PC" compatible. Then MS helped shift that to "Intel x86 compatible". When IBM tried to introduce the micro-channel architecture, it discovered that it no longer is in control of the very architecture it created. The market just ignored IBM and took the PC in the direction other companies wanted. Then even Intel lost control. It became "Windows compatible." It may not have been immediately recognizable as a defeat, but it became blatantly so when Intel had to go ask for MS's permission to implement their own 64 bit extensions... and got told to use AMD's instead. Ouch.

In a sense, MS helped "create" Linux. At the anti-trust trial, MS used Linux as an example in their "we're not a monopoly, other people can still make good OS's" sophistry. It just told everyone what other OS they could use instead, if only it was more up to modern standards. And they just proceeded to help with bringing it there.

At any rate, the short story is: most of the successful F/OSS is the work of corporations, and _of_ _course_ they want to be the next MS. Or at least to take some market share from MS. And _of_ _course_ they'd like to make a profit (indirectly) out of it. That's the whole _point_ of bothering with it.

E.g., Sun isn't developping OOo because it just likes making cool software. It's because at some point people were saying, basically, "Yeah, well, but your workstations don't run MS Word." If the software they pay big bucks for doesn't address that problem, they might as well fire that team and move on.

E

Re:LMAO Non free finally wins.

[Anonymous Coward]

I'm going to give all of that up right now and buy OSX, Vista, Visual Studio and half a dozen software packages that I need to get real work done on these real platforms. Careful. With that attitude you might actually start making money.

I don't get it --why not pick an arbitrary GUI?

[KWTm]

I'm not sure I understand you. You say that because Linux has both GNOME and KDE (and others), there is not one standard GUI on which to develop. But why don't you just pick one? People have access to both, you know.

For example, as a die-hard KDE user, I'll ask: what happens if you just pick GNOME and go with that? If it's a useful program to me, I'll install your GNOME program on my KDE machine. For example, I run GnuCash and not KMyMoney, I run Gnumeric and not KSpread, I run Abiword and not KWord (or OpenOffice.org), and I run Firefox and only occasionally Konqueror. I plan to continue to use KDE for the foreseeable future, and I've never downloaded the default Ubuntu, only Kubuntu.

Unless I misunderstand you, you seem to be saying: "Microsoft has a single door to walk through. But Linux provides double-doors, so I don't know whether to walk through the left one or the right one. So I won't bother, and I'll just stick with the single door because the lack of choice is less confusing."

Re:I don't get it --why not pick an arbitrary GUI?

[Chandon Seldon]

Google doesn't seem to have had a problem with Google Earth. Nero burning rom works fine. None of the various proprietary games for Linux has any problem. Adobe Acrobat Reader works. Realplayer works great.

Sure - the developer who builds the package has to have a basic understanding of the platform. They may even end up having to spend some time researching the problem. But - if you're willing to go to the effort to port to a new platform a couple hours of study shouldn't bother you.

Realplayer is a perfect example of building and packaging a proprietary application for Linux. They built using Gtk for maximum compatibility, and they initially released in .rpm and .tar.gz. They then allowed free redistribution, so all the major distributions repackaged the program and distributed the packages in their non-free repositories at no cost to Real Networks.

Re:I don't get it --why not pick an arbitrary GUI?

[ratboy666]

And this comes up again...

It is perfectly possible for an application author to produce an application and link it statically.

It is also possible to link with -z origin -rpath '$ORIGIN/../lib' (or similar) allowing the bundling of the EXACT shared objects (dlls for you windows folk) locally into a single directory tree.

If this is done, the application can be put into any directory, and will run from there with no further configuration needed. If glibc is included, it will also run on a WIDE range of kernels.

Of course it isn't in the "packaging system" -- but that can also be done (especially to run pre and post scripts).

As to the GUI? Either include the widget set, or use GNOME. What's the problem?

And (ps.) this is what's generally done on Windows. DirectX is included by every application that needs it.

"Build from source" isn't required to cross x86 Linux boundaries -- it exists to cross processor families (Sparc, PowerPC, MIPS, ARM, etc.), and Operating Systems (Solaris, AIX, HP/UX, etc.).

Even there, tools exist (eg. QEMU) that allow running binaries on other archs.

FOSS allows you to CHOOSE your monoculture ..

[cheros]

I'm not convinced the article provides a solid basis for blaming choice as a problem.

I have yet to see people try to find a new toolset every time they build a platform. Usually, an IT shop decided on which tools it will use to do the job (including which hardware, code language and dev framework) and will stick with that choice, simply because that's where their expertise lies. Only when the toolset is not up to the job or there is a simpler/better way to address the task at hand will there be a re-examination and/or switch, and such changes are in both environments (Win/FOSS) also driven by the people doing the job doing the usual looking around for ideas and products - that's simply part of the work (did I just argue that Slashdot reading is essential? Yes! :-)).

After that it's learning how to maximise your use of the toolset and work around the problems with it, and that tends to result in some branching out from the default platform as well. Do MS shows only use 100% MS code? IF SM had their way, sure, but life's not like that. The only difference with an MS shop is that experimenting doesn't immediately cost license fees and instantly creates the risk of a FAST visit being successful, but that too is an issue hat can be managed.

So this 'choice' is a starting issue, not a live ops issue.

The challenge of a monoculture is not that it's mono, it's about who controls the direction. An MS monoculture doesn't really to be driven by user need, witness the heap of crap that is Vista, and the total mess they made of the different versions of .Net. From an IT strategy point of view you're better off with a direction that YOU set, not the vendor. Not only is it cheaper, it's also less driven by a vendor's need to flog new products.

At that point you can start asking questions about true business benefits and TCO.

A question of self-responsibility

[petrus4]

I watched V for Vendetta again last night, and was reading some related material online afterwards. It introduced me to a couple of ideas which although I'd more or less known about instinctively, I possibly hadn't considered from quite that perspective. This is going to appear to be offtopic at first, but bear with me and you'll see the point as it relates to the issue of choice with Linux.

From what I've read, the central element of anarchic thought is apparently the idea of a scenario where people are genuinely self-responsible; where people are able to make decisions and choices about everything they do, and where it can be at least hoped that the need for an external authority is mitigated by said people having an internalised system of morality. In other words, the idea being while they are able to choose to do whatever they like, that people will eventually figure out what they are meant to do on their own.

However we keep seeing (no doubt unfortunately in the minds of some of us) that the above scenario, not only where Linux is concerned but in every other area of their lives, is overwhelmingly not what the vast majority of people truly want. I've found myself reading quotes from both Freud and George Bernard Shaw over the last 24 hours that stated that contrary to the commonly held belief, the majority genuinely do not want freedom, precisely because a prerequisite of freedom is self-responsibility.

This of course is where not only Microsoft in the case of software, but repressive states of all kinds in general life come into the picture. As V said, they offer order, certainty, stability, an absence of chaos, and most importantly, an absence from the need for a person to think for themselves, and all they ask in return is silent, obedient consent. They give people a scenario where decisions are made for them, where no thought whatsoever is necessary, nor responsibility taken for wrong decisions. As the old saying goes, "Nobody ever got fired for buying from IBM."

This is what people overwhelmingly want; what they are trained from the earliest age to want. National governments use the education system these days in order to start negative reinforcement against the exercise of free will within individuals as early as possible, and if such is instilled deeply enough and early enough, the process produces individuals who refrain from exercising choice as much as possible for the rest of their lives thereafter.

If you're wondering why people continue to want Windows over Linux, and continue to complain about the degree of choice inherent in Linux, you might perhaps also want to ask why people are also willing to allow the likes of George W. Bush and Tony Blair to remain in political office. The answer to both questions is the same, for they are in truth both different elements of the same issue; an insistence on avoiding self-responsibility and reasoned, conscious thought within the majority of the population.

How can Linux advocates overcome such, I hear you ask? Instilling independence in those who do not have it already is by necessity an incredibly slow and transitional process. In the case of someone complaining about being overwhelmed by choice, I'd probably start by asking them what it is that they as individuals want to do with a computer, and then direct their attention to a single distribution (or possibly even Windows itself, if appropriate) which will meet their needs. I've tended to notice that people aren't normally wanting a reduction of choice for people other than themselves, when they are asked, but merely want a scenario where they do not need to engage in it. Hence, if they find something which will meet their own requirements, they will very often cease to complain.

Some individuals are inherently lacking initiative and crave situations where they are taken care of by external parties. Sadly, there isn't much any of us can do in the case of such individuals, other than hand them a copy of Ubuntu or Vista, and a smile. Although I fall into the trap myself on here fairly regularly, I also try and tell myself that such people are not worth getting upset over, since they are a reality that we cannot change anyway.

DOH! or, How to write an uninsightful article

[Colin Smith]

What he's really talking about is the Network Effect, but doesn't seem to realise. The Network effect is why we speak national languages instead of reegional ones, why some form of English is ultimately going to replace all others, why TCP/IP is the only protocol our machines talk now. Why all keyboards are querty. Why we use the same currencies, why we all drive on the same side of the road. There is utility in all things being the same.

It also applies to user interfaces, libraries, operating systems etc but to a much weaker level. This simply means that it takes longer for the users of the various interfaces, libraries, development tools to converge on the same solution. The need can't particularly great because if it was, the convergence would be happening far quicker. It'll happen over time in the Linux environment, in the meantime, the market is going through the various Linux softwares and choosing the one which fits their needs best.

Preposterous.

[Inoshiro]

"Every job has its unpleasant parts, and while a F/OSS coder can skip them a commercial coder can not; if the spec calls for an embedded testing code, for example, or Doxygen comments, you put it in."

The percentage of non-FOSS which is documented and the percentage of FOSS which is documented are pretty similar in my experience. Perhaps you are unaware of the incredible 95% of software which is developed commercially, but which is not sold in a shrink-wrapped box in Circuit City. Many companies have internal IT departments which couldn't code their way out of a paper bag in VB, let alone document it. Go read the daily WTF if you think I'm lying.

"s/he might be wrong but at least the product is consistent, and not designed by a committee as it sometimes happens."

Andrew Morton. That's a name I can think of when I think of someone with vision for a particular FOSS project which are willing to say when things (don't) match their vision. Linus Torvalds also fits this bill. There are similar names in other projects, but I'm most familiar with the kernel.

"These clones haven't been weeded out by the market, and so many of them are not viable - but they are out there,"

The weeding doesn't occur at the store level, it occurs at the reputation level. All FOSS stuff is staked on reputation. If you have a high reputation, you are going to be used more and included in more distributions. If you are a crappy app, you'll never see a real user base. Since the programmer is programming for ego (see the somewhat inaccurate women/baby analogy), the programmer should be motivated to produce better work which becomes more popular. The KDE programmers sure seem to have worked to make sure that KDE is useful. The Gnome programmers have also worked towards some mindshare. Given how people used to choose window managers, but now choose desktop environments, I'd say that these programmers have changed the game wrt GUI interfaces on Linux. That sounds much like a market shift, but with eyeballs and hearts instead of money.

"Effort dispersed, spent on competing projects is ultimately wasted."

Oh, I'm sorry, I guess you're just a troll. Or you simply don't understand what FOSS is.

I'd like to know

[epine]

If only I could get a look at the user-agent strings to see how many of the "all praise consistency" crowd have posted their comments under Firefox. Those of you who did should feel a small twinge of moral ambiguity the next time you open a page under tabbed browsing. Microsoft only came out with tabbed browsing when choice put their back to the wall. I was reading the other day that ninety odd percent of the world's food production is confined to twenty odd species of plant and animal. While we're on the subject of restricting choice for the greater good, I hear that arranged marriage offers many practical efficiencies.

Definitely

[NekoXP]

Simply put, he's right in a way. After all, developing software for an environment you choose, means anyone who didn't choose your environment who wants to run your software has to switch, or install more software, and deal with the problems possibly associated with such a thing.

Remember when KDE, GNOME, Xfce and Enlightenment didn't share a desktop API? Look now at how Enlightenment reinvents everything using it's own special libraries? While Enlightenment has some distinct advantages over the way the others are designed, it is a DIFFERENT system. Want to install a GNOME core application on KDE? Well, you have to drag in most of GNOME, still. The same in reverse. Install Enlightenment tools on top? Well you have to drag in the rest of the E17 framework.

Install X on my system, and it still pulls in 5 different sound daemons.. yikes, and yikes again. Xine, MPlayer and GStreamer/Totem too. They all use the same libraries after all, but do I need 3 different ways to play a movie?

I personally prefer GNOME and Xfce if only because they use the same GTK toolkit - however I personally loathe GTK and the GTK API. I don't want to even get started in Enlightenment.

So, when you sit down and use Windows, what do you do? Well, you're pretty much stuck using Windows. And for all intents and purposes, there is a strict set of toolkits and APIs they provide for you (DLL hell wipes that off the map though). There is no "which API do I use to open a window and add a button" if you are using VisualC++ and reading the documentation, it will pretty much railroad you into one choice. But there ARE other choices.. they are just less obvious and less relevant.

I think this is why I like the concept of RAD stuff like Ruby On Rails, however I do hate Ruby, and Python, and I never got into Perl in a big way, and while I'm stuck with PHP, it's because it's closer to C++, which I absolutely love. If I had a choice I'd be coding everything in C++, with a single toolkit, but unfortunately because everyone else makes other choices, I can't.

Does my life deserve to be made this difficult by virtue of the freedom of choice? Probably ;D

I love Linux...but as a software engineer...

[zurtle]

I also love Kubuntu and use it on all my machines at home... However I develop for Windows at work. Developing application in Windows can be as easy as VB. As far as I have seen there is nothing on this (OSS) side of the fence that comes close to it for ease of use, with (reasonably) good debugging. From there it's a skip and a jump into C# and beyond.

As a (bad) example of how far UIs have to go, I use Visual Studio 2005 at work and I use Matlab for my postgrad work. Both professional, closed-source products. The Matlab debugging facilities pale in comparison to Visual Studio's power. Makes life so much easier. I find Matlab a bit archaic, but KOctave seems even worse. I guess Matlab only has to do better than Octave on Linux systems.

Of course my argument is omitting things like Eclipse... but I have only used it a few times and found it too slow (slower than VS.Net!!) and user unfriendly.

Re:+5 (Obvious)

[Aladrin]

"it is unstructured choice that makes it difficult for consumers"

I'll agree with that. Now tell me where the structure is in choosing a distro for Linux.

MS offers choices in their new Vista, and the 5 or so versions that they offer is an almost unbearable choice for consumers. ('Ohhh, do I really need feature X? What if I choose not to get it, and need it later?') At least they have a chart that shows you the features and what you'll be missing if you buy the cheap ones.

Linux offers dozens of distros and I've never yet seen a chart that shows the pros and cons of each one, or even the biggest 5. For instance, check this page. http://www.linuxdevcenter.com/pub/a/linux/2005/10/ 27/what-is-a-linux-distribution.html It says 'Choosing a Linux Distribution' and lists the major ones. Does it tell you what they do? No, it tells you how they were born. WTF good is that?? It does recommend Debian for servers, and Ubuntu for newcomers. But it doesn't say the features at all.

So it's unstructured choice. You could spend months on the net researching distros before you actually found the one that suited you. With the Windows Vista chart, it would take an hour, tops.

BTW, that link was the first result from Googling 'linux distro chart' and none of the other results even seem relevant.

That's the wrong end of the stick

[ChameleonDave]

Although Linux distros have many flaws, they have nothing to do with the reason why most people and businesses use Windows.

Linux and Open Source are new as a mature software solution. Microsoft started dominating the market many years ago now. If everyone bought computers today and gave them to employees that had never used a computer before, then of course we would win. But that is not the situation. Companies have documents in Word format; they have employees who think the big blue e equals the internet; people have paid for licences already. It will take time, but we will win.

If that's the case, shouldn't Apple be in the lead

[jimstapleton]

I thought Apple had fewer software options for most tasks than Windows, at least it has always seemed that way to me.

Likewise, BSD has fewer software options for most tasks than Linux since it has less development, so why does Linux take the OSS business market instead of BSD?

Note: I don't intend this as a flame at all, I like one of the OSes that this says should do good based on the logic, and I don't like the other. I like one of the OSes that looks like it should be bad in this comparison, and I don't like the other. I'm just saying the arguments logic doesn't seem to add up to me.

Linux has lots of monocultures

[massysett]

Linux has lots of monocultures--pieces of software that have become mostly standard:

* the Linux kernel (rather than, say, the Hurd)
* X.org (rather than XFree86, which is now dead)
* bash (rather than ksh, csh, tcsh, or my favorite, fish)
* Apache (I had to look at Wikipedia to see if alternatives even exist)
* MythTV (any other Linux PVRs?)
* GCC, and for that matter, most GNU tools

Perhaps usage standardizes on one piece of software when that benefits people, but usage fragments when there are benefits to choice. Doesn't seem like a problem.

Re:LMAO Non free finally wins.

[analog_line]

apt-get install kdevelop

Here you have two prime examples of "too much choice" in action.

You're assuming that everyone uses an apt-get based package distribution method, and you just can't do that.

You're also assuming KDE is there, and that's an even more flawed assumption.

Dude, what are you talking about?

[Ayanami Rei]

1) Your software still looks like a 16-bit Windows application. I imagine the code isn't very portable to begin with. Hence your difficulty in porting.
2) GTK is your GUI target. If you need a C++-based library, then target wxWindows or QT (added bonus, you get Windows compatibility for free with those two).
There aren't any other choices. It really isn't as complicated as you make it out to be.

Strategic Incompetence

[mkcmkc]

In my experience, nothing makes a monoculturist IT manager happier than being able to reply to some big ugly request for services with a simple

Sorry, but Microsoft doesn't do that.

And in many bureaucratic environments, that's the end of the story.

good strategy

[nightsweat]

Blame the user and tell him he's lazy for not understanding crap that just gets in the way of getting work done.

It's easier and more efficient to ask "paper or plastic" than to follow that up with "paper made from wood pulp or grasses?" "recycled wood pulp or all-new?" "paper from hardwood or from softwood?" "Made from trees from the Northern hemisphere or southern hemisphere?" "Bleached or natural?" "Logo printed on or blank?"

Sure, some people will want all these decisions, but they shouldn't be a requirement of the OS. Solve the Gnome/KDE nonsense and you'll see Linux propagate much more than it already is.

Variety, too spicy?

[CopaceticOpus]

The problem is not the existence of too many choices. The choices should be out there for those who need them, but they should also be transparent for those who don't need them. This is one reason Ubuntu has proven to be so user-friendly. It makes many choices for you by default, which are good choices for most users. It doesn't force users to think about choices that they don't really care about anyway.

On the other hand, I think that open source development often wastes much of its potential by creating too many varieties of products. I have a dozen video players installed on my system, and I'm still searching for a good one. There might be a good one available if the development work hadn't been repeated across so many similar products.

Re:Furiously Spending as we speak

[AJWM]

Unlike Microsoft, which has never engaged in massive layoffs

Sure. When you've hired thousands of "permatemps", you don't lay them off, you just expire their contracts.