We have linux, freebsd, openbsd, netbsd, solaris, tru64, sunOS, AIX, HP-UX, etc. Hmm... how many of those are produced by the same vendor? This isn't an issue of someone boosting NT and MSOffice. In fact, NT was only mentioned in passing in the article. The problem for Compaq is that they're faced with marketing and R&D costs for half a dozen different OSes. That means a large amount of wasted effort and confusion. Why develop Tru64 for IA64 if it will be a multi-million dollar effort that only brings in peanuts? Worse, what if its only effect is to replace systems that would've run Tru64 on Alpha, which would have given Compaq profit on both the chip and the OS? In the long run, I can't see Tru64 as a contender. It has to face down Sun, MS, Intel, Linux, and HP, all of which have superior resources and mindshare. But in the short run at least, not porting to IA64 is probably the wiser choice. --JRZ
I absolutely love KDevelop; it made me appreciate that it's actually easier to program for KDE/Qt than for MFC/Win32 (which I also do), if you have the right tools. More importantly, I wanted to mention that an article on KDE/KDevelop programming (by yours truly) should be showing up in "Visual Developer" magazine in the November issue, God willing. Anybody interested in something that goes a little bit beyond the included tutorials should check it out. It's very cool to see a traditionally Windows-oriented magazine branching out to cover serious Linux programming topics. When is there going to be a "Linux Programmers' Journal"? It's really strange to me that all these magazines have "How to hack the Linux kernel" articles sitting right next to "How to install X for the first time". --JRZ
I have to share one more bio story, and I swear it's true. One of my friends finished his first year as a biology undergraduate at a state school. Got a reasonably interesting summer job as a lab assistant making about $8 an hour. One of his partners, a POSTDOCTORATE biologist, made THE SAME AMOUNT OF MONEY! Absolutely the truth. Maybe he was an idiot, I don't know. But there's a difference between being "money grubbing" and simply wanting a reasonably comfortable lifestyle (after one year of undergrad CS, I was making more than double what the postdoc made, hmm....). Supply and demand, baby. --JRZ
I live in Northern Virginia (home of UUNet, AOL, etc.) and every other day I see another news story lamenting the incredible lack of talented computer/engineering professionals. All these studies also say that we're simply not graduating people fast enough from credible colleges, and many of the jobs out there can't be filled by people with technical institute degrees (though many also can). Yet here comes the National Science Teachers' federation, or whoever, telling us that we need FEWER computer specialists? Maybe they've been working with those chemicals a little too long, but I know a decent number of people in physics, and they are FRIGHTENED by the prospects out there for a PhD physicist. Research in some pure sciences has become so esoteric and so expensive, that only a few top institutions can do really cool stuff. Of course, it ain't easy to get a job at one of those spots. So should we wonder why people enjoy CS, where you can tackle the coolest, most cutting-edge problems from the comfort of your own workstation anywhere in the world? I especially loved the comment that "CS simply doesn't teach the kind of critical thinking that you learn from physics." What a blatant lie! Computer Science students will generally be exposed to deep mathematical theory (abstract reasoning), algorithms (abstract problem solving), hardware (hands-on learning), research techniques and methodology, and systems/applications programming (applied problem solving). What critical thinking was not covered there? I find even the most mundane of these topics (say, debugging, which we all hate) to be a lot more stimulating than titrations! (no offense to the chemists) Academia is like that, as is any large bureaucracy. Everyone needs to preserve his own little niche by convincing the world that comparative Middle Eastern philosophical history, or whatever, is the most important subject in the world. --JRZ
1) I don't see massive speed improvements coming anytime soon. Sun wants to push the server version. The press hype stresses the fact that so needs a powerful computer to run. So basically, they let the desktop version continure to have issues, so the thin-client version looks very attractive in comparision. The Sun Community Source License gives Sun final control of all patches. Which means you can expect simple bugfixes (it crashed when I did this, so I fixed it) to be used, but anything bigger is very doubtful. 2) Ouch. What does this mean for poor Corel? Yes, I know that WordPerfect is vastly superior in many ways, but its recent mild resurgence in the Windows world has had a lot to do with its positioning as a low-end office suite. Manufacturers looking for a cheap way to add value need look no further than StarOffice. Not a big deal in the short run, but it could prove really dangerous eventually to those poor Canadians. 3) Hopefully they just won't bloat it any further. Do what MS has always hoped to do: wait for hardware to catch up. With 128 megs of RAM and a 600+ mHz PC, SO would be quite bearable. 4) Anybody used Applix Anywhere, Applix's free-ish Java office suite? Man, that's slow as a dog. Wonder how it's doing commercially. --JRZ
The one actual change to the program mentioned by Sun so far (aside from generally more interest in server distribution) has been their intention to switch to XML for better compatibility. This is fantastic overall, even though XML makes fairly bloated files in terms of size. --JRZ
Well, as the author of the first "Open Business Plan" article, I have to say that it's great to see a more open debate about the future of Red Hat. I enjoyed this new piece very much, but I think it's only fair to point out the areas where I disagree with it. Of course, if you look at the first comment on Freshmeat (by Benjamin Scherrey), he seems to have already posted most of these comments too. Here's my overall issue: Red Hat has a business model. They want to provide support and services for Linux. She the fundamental question is: how to they increase their Linux services business? A focus on gcc (as per point 1), for instance, would put Red Hat in conflict with Cygnus, which has made a very successful business for years by controlling the market for gcc support and expansions. Cygnus knows that its future is directly dependent upon these compilers, so they have a very aggressive plan of improvement for the GNU compiler system already. Additionally, what CEO says, "Gee, I think that operating system sounds great for our business, but I'm not happy with their compiler's adherence to ANSI standards, so let's go with a different choice instead"? I think that points 2 and 4 of this article are examples of the way in which many users prefer to look at Red Hat as a software distributor, rather than a services company. RHAT has never claimed that their market would be the desktop. Why? Because servers need expensive support from the vendor, while desktops don't. That's why the company is willing to see others provide the productivity software, while they stick close to the more lucrative server/OS strategy. The issue of a subscription is more of something that SuSE would do (actually, I believe they already do it). Why? Because SuSE appeals to hobbyists (I'm a SuSEr myself). They bundle all the software on Earth, much of which is irrelevant to businesses, and they don't worry too much about integrating the various packages or providing graphical administration. But business unsers want polish, ease of use, and itegration, not just the latest thing to come out. They also see upgrading as a difficult process to be undertaken only when compelling. Remember, millions of copies of DOS shipped last year! Why? Because many folks couldn't find a reason to upgrade from even this incredibly outdated technology. These are not customers who feel the need to go from kernel 2.2.11 to 2.2.12 as soon as the new one comes out. Also, retailers HATE that Linux distros go from 6.1 to 6.2 before even the 6.0 version sells out. It makes their inventory worthless. If you look through any major computer store, you'll probably still find copies of OpenLinux 1.3 on the shelves (possibly being resold under a different name). I've seen CompUSA put OpenLinux 1.3 next to OpenLinux 2.2 and charge MORE for the old version! I have to disagree most strongly with point 6 for many of the same reasons that Mr. Scherrey pointed out. Borland's strength is still in Windows software development tools, a business about which Red Hat knows nothing at all. Worse, they're a software developer, and RHAT does NOT want to be in the software business, they want to be in support. Even more importantly, both Cygnus and Borland already recognize the importance of their Linux strategies. Cygnus is NOT going to drop Linux support in gcc or Code Fusion any time soon, because that's the vast majority of their market base! Similarly, Inprise is already committing to a future as a cross-platform tools provider. As an independent company, they will contribute at LEAST as much to Linux as they could if they had been acquired. Red Hat needs to acquire targets who, if left alone, will NOT help the Linux OS as much as they could. Any acquisition would also have to address Red Hat's desperate need to gear up its consulting arm. I have mixed feelings on the LSB. It's somewhat disappointing that they have not yet been able even to articulate a vision for how they see the Linux standards process playing out in the future. If they could create a base that moves quickly and allows vendors to add their own value, it could be interesting. However, Red Hat holds such a dominant position in the market that they can afford to make their own standards and have others follow them. Look at "Code Warrior for Red Hat Linux"! Red Hat never asked them to support their distro alone; Metrowerks just didn't want to commit right away to supporting more than one system. The bigger challenge will be Corel's Linux, which has big corporate backing, but doesn't even use the RPM format. If Corel and Red Hat could agree on standards, however, that would basically force the LSB members to follow along. Well, like I said, it's great to see an increasingly open discussion of these issues. I hope these discussions set a precedent for more, similar debates in the future. --JRZ
Ok, so I'm an undergraduate, but I thought I'd add my comments anyways from what I've seen/heard and know about the faculty. If you're into theory/algorithms, it's a pretty hard core place. Sedgewick, Tarjan, and their buddies can compete with anybody out there in the field (IMHO, Donald Knuth would be an exception to that statement). A lot of the folks on this side of the department say that the best CS grad student was a math undergrad student. There's also a new program in applied and computational computing which looks really promising. It's all about integrating CS and other fields that require intensive computer modeling, and they're putting some very cool folks in the program. Also a pretty strong program in computer/network security, but I don't find that stuff so interesting, so I couldn't tell you much (except that Ed Felten, the government's tech guy in the MS case is involved in that program). So, basically, I think I agree with the overwhelming sentiment of the others here today: the school you choose should depend on what you want to do.
Actually, I was going to write something about this in the original article. I love Qt and would really like, at the very least, for them to release a low-cost/free version for commercial development on Unix/Linux, but keep charging for the cross-platform version (of course, LGPL all around would be even better for the consumers, but they're still a business). It seemed, however, that one of the other Linux distributors would probably be better suited to make this move. It would be pretty awkward for RedHat to abruptly drop GNOME/GTK right now, but if they didn't move to Qt as their primary platform, it wouldn't be a very good investment. Still, something has to give with the current Qt licensing system. It's ridiculous to charge $1500 per developer for a widget set, even if it is fantastic. How much is MS Visual Studio Pro now? $750? And companies with an absolute focus on cross platform capability can always settle for Java for free, despite its faults. There you get a large, skilled developer base and your choice of many IDEs. If Troll-Tech (or their future owner) doesn't wise up in the next year, they'll rapidly be made irrelevant in the commercial markets when Swing and fast JVM's really become common. --JRZ
I totally agree with you about the patchwork license and free software issue. Anyone who feels strongly about using GPL-only software can not really use KDE. However, I'd take issue with your comments on the core technology and the object model. Bonobo and ORBit have a lot of good features, but they inherently suffer from GNOME's focus on C-programming and ORBit doesn't even have C++ bindings! While MICO is a very large ORB, KDE 2.0 will actually use tinyMICO, a scaled-down version that cuts out the unnecessary crap. MICO is a very serious CORBA 2.2 implementation, with many more features than ORBit. ORBIT, however, is clearly the faster ORB, much as MySQL is faster than, say, Oracle (ooh, nice tie-in to a recent "Ask/."), because it's stripped down. More importantly, KOM/OP is a fantastic object model with a very easy learning curve, and the KDE2.0 daemon implements an extremely easy-to-use naming system. I've been nothing but impressed by KOM/OP and the thought that went into it, especially as I use KOffice, which is already quite advanced. I don't mean to start a flame war. But it is important to look at the details of these implementations. Ideally, we'll get a level of object interoperability in the near future. --JRZ
QT/KDE themes are a bit confusing, because there are different types. The current KDE themes in 1.1.2 are pretty typical pixmap stuff, not super fast, but I don't have any problems on my PII/350 with a crappy 4 MB on-board ATI. Qt 2.0 themes, however, can be quite fast. They don't rely on pixmaps, but just override a virtual drawing function. Very cool stuff. My one flamebait on the GTK/Qt debate: if you use C++, Qt is a dream: pure, object oriented libraries. Not a wrapper like GTK-- (which isn't bad, though) or MFC in windows. If you use C, stick to GTK. --JRZ
Well, Instant Messenger as a tradmark isn't really ridiculous. For AT&T to come out with a similar product and give it an identical name is really shady. What if Mitsubishi came out with a car called the "Plymouth" and argued that, hey, Plymouth is a popular name. Bullsh*@#$%&t. We don't complain about IBM which has a TM on "International Business Machines", and I don't see how you can get much more generic than that. It's just fashionable to bust on MS and AOL these days. --JRZ
1) Hopefully by now we all know that no such tax has been proposed in any real way (especially since the USPS is entirely self-funding these days). But the topic is still relevant, because there WILL be internet taxes of some sort. 2) Paying (in some way) for network services based on usage (rather than "unlimited use") is a MUCH more economically efficient. Think about all the spammers who wouldn't be in business if they had to pay $.01 for each email. What if a basic net connection were free, but you had to pay 1 cent per email and 5 cents for each meg of download? Hey, it would screw me too, but it might make us think about the fact that network resources ARE limited, and some clowns suck it up without thinking about that. I'm not saying that this should be a tax, though. 3) Sales tax will eventually have to be applied to internet purchases. As more commerce moves online, it's inevitable, and it's no worse than paying the sales taxes we do now. I'd say, half of the tax goes to the destination state of the sale, half to the originator. Then you don't get much of a break for locating offshore to avoid the source tax. --JRZ
Check out the whitepapers on developer.gnome.org and developer.kde.org. Yup, obviously a whitepaper is quite different from thorough design description, but a few of these docs, especially those related to KOM (KDE Object Model) lay out the costs/benefits and architecture of the chosen solution. As far as the topic of whether engineers can write good design documents, I'd say that many can't, but a few are actually VERY good at laying out all the ideas ahead of time. In those cases, the documents can be quite helpful to others. The most important part, however, is that the document should be a useful roadmap to the programmer who writes it. If you think all the project's aspects through before committing a line of code, sure, some things will have to be changed as the project progresses. But at least you won't end up with a poor hack that needs to be redone from scratch. The final barrier that can exist is that many engineers I've worked with don't have English as their first language. Worse still, some are C programmers and a few are even stuck on Hungarian notation. Ah, I see, "struct _qXrtsh" has a field called "lpszTbsdx", yes, that makes great intuitive sense to me. --JZ
Diminishing Marginal Return
on
Athlon Reviews
·
· Score: 1
Any time you can get an increase in CPU speed directly proportional to the increase in cost, you should jump on it (as in this case). Typically, high end processors demonstrate hard core diminishing marginal rates of return. Thus, the ten percent increase in performance you get by stepping up to the latest Xeon might increase your outlay by 50+%. Those $13,000 PA-RISC workstations, to take a different example, are NOT ten times as fast as a solid $1300 system. They'd be lucky if they could clock in at twice the speed. There's simply no way to say: "Ok, you can buy 1 point of SPECInt for $20." --JZ
Yup, the server is having a bit of trouble today. Too much good publicity;) Try http://fara3.cs.uni-potsdam.de/~smeier/kdevelop_ho me/ for a mirror. Please, please, report bugs so that KDev 1.0 non-beta will be as clean as possible! --JZ
Re:Memory Usage ?(or Memory more important than CP
on
Some KDE news
·
· Score: 1
Ok, I just have to second the advice that you switch to KDE1.1.1 or better to improve speed, but I'd also recommend trying WindowMaker + KFM (the KDE file manager). With WindowMaker you already have the dock, background manager, etc, so you don't need a panel or kbgndwm or anything like that. KFM with the new KDE 1.1.2 hi-color icons looks damn cool and it fills in that empty space on the left side of the screen (I use dock to the right, clip at the bottom) that always seemed so asymetrical. Plus, KFM by itself only needs a few megs of memory. --JZ
Why is this still my typical party conversation? Me/her: Her: So, what're you majoring in? Me: Well, I don't know... maybe CompSci. Her: Oh. . Do you do that sort of thing a lot, I mean computer stuff? Me: Oh, you know, I to keep it diverse, but there's so much happening in computers these days, it's really interesting stuff. Her: Well, did you do that sort of thing a lot in high school? Me: No, I actually wanted to me a comp lit major, or maybe econ, but the CS department here is just really great, and it's a fun place to be. Her: Well, I gotta go... Me: No, wait! Really, I just want to start a net company and get rich! I swear! Her: Oh. That's cool. Give me a call sometime. Preferrably right before your company goes public. --JZ
The optimism isn't wrong, that is, but the angle is way off. Which OS will take better advantage of this new EasyPC spec: a) The OS whose company was deeply involved in writing the spec and which is coming out with a specific version just to support it (Millenium) or b) The OS that gets little, belated info about new hardware initiatives, can't run WinModems (which are surely a central part of EasyPC), and only has experimental PnP and USB support now?
The real future for Linux comes when more hardware vendors are willing to write drivers for the LinModems of the future. Oracle's $150 Linux NC will be a nice start. And if Corel gets good penetration, it will certainly help. --JZ
I know it seems scary, but it's a perfectly valid, and even necessary, legal step. If people can freely spout libel all across the internet and get away with it just because they're "Anonymous Cowards," it will become a MAJOR problem. Think about it this way: if I own a software company, I could post fake, libellous leaks about my competitors on stock sites across the country, claiming to be an employee who wishes to remain anonymous. Should a smart investor take advice from an anonymous leak? Of course not. Are there a lot of stupid investors out there? I don't think I have to answer that one. I think it was Salon that had an article about major studios creating fake fan sites for their movies (duh, that's a shocker), but, much worse, also creating fake anti-fan sites to badmouth their competitors. When this kind of blatant lying rises to the level of true libel, we need to have some recourse. Remember, you can NOT be sued for simple, constitutionally protected speech or plain old stupid ideas. Just libel. And that's the way it should be. --JZ
Ok, we all know we're talking about a low-end E-commerce site here, so here's what's missing from the package: -Easy to use database -Graphical database tools -Simple, graphical server/e-store configuration tools -Web authoring tools They can't sell a package that centers on MySQL, because that would incur their weird licensing costs (Although they can include it as an option), but maybe a PostgreSQL/MySQL option would be allright. KMySQL is a great, free database frontend that has plugins to work with other DBs (no Postgres yet, though) and could be great if RH would invest in it a little. Templates are easy and/or freely available. Hmm... what does that leave undone? Easy tools for building e-commerce apps. Something easier than, say, Zope or another real app server. People just want a "Yahoo Store" or the like. --JZ
Well, I would say: no biggie, tons of other top app servers out there will/already do support Linux (which is certainly true), but every mention of app servers just reminds me how cool JSP and EJB are. And how slow Blackdown is. Linux needs a top-notch JVM before its server-side Java can take off. Check out WebTechniques benchmarks of GNUJSP on apache on Linux. The poor author adores the OS and webserver, but, even with tweaking, it still took SECONDS to server a simple JSP. No joke. What we need from Sun is not their app server, but their support for a real Java2 on Linux. Maybe IBM can do it. . . --JZ
I use KDevelop almost exclusively now. There are maybe 2 or three crashing bugs that I've seen, but most of those are rare and/or fixed in CVS versions. It is by far the best way to learn about KDE/Qt programming, because it includes great documentation and an AppWizard to get you started. Needless to say, I'm a big fan. --JZ
Unfortunately, it's not quite there yet. Even more unfortunately, I haven't seen any change in their web site in months. It doesn't have a class parser/class browser, which is really the key thing that I need in an IDE. The stub code is there for it to be added in, but the project seems a little stalled. Check out www.netbeans.com. They have a pure Java IDE that's actually quite good. Also free for non-commercial use and/or trial. --JZ
Slashdot Mirror
We have linux, freebsd, openbsd, netbsd, solaris, tru64, sunOS, AIX, HP-UX, etc. Hmm... how many of those are produced by the same vendor? This isn't an issue of someone boosting NT and MSOffice. In fact, NT was only mentioned in passing in the article. The problem for Compaq is that they're faced with marketing and R&D costs for half a dozen different OSes. That means a large amount of wasted effort and confusion. Why develop Tru64 for IA64 if it will be a multi-million dollar effort that only brings in peanuts? Worse, what if its only effect is to replace systems that would've run Tru64 on Alpha, which would have given Compaq profit on both the chip and the OS? In the long run, I can't see Tru64 as a contender. It has to face down Sun, MS, Intel, Linux, and HP, all of which have superior resources and mindshare. But in the short run at least, not porting to IA64 is probably the wiser choice. --JRZ
I absolutely love KDevelop; it made me appreciate that it's actually easier to program for KDE/Qt than for MFC/Win32 (which I also do), if you have the right tools.
More importantly, I wanted to mention that an article on KDE/KDevelop programming (by yours truly) should be showing up in "Visual Developer" magazine in the November issue, God willing. Anybody interested in something that goes a little bit beyond the included tutorials should check it out. It's very cool to see a traditionally Windows-oriented magazine branching out to cover serious Linux programming topics.
When is there going to be a "Linux Programmers' Journal"? It's really strange to me that all these magazines have "How to hack the Linux kernel" articles sitting right next to "How to install X for the first time".
--JRZ
I have to share one more bio story, and I swear it's true. One of my friends finished his first year as a biology undergraduate at a state school. Got a reasonably interesting summer job as a lab assistant making about $8 an hour. One of his partners, a POSTDOCTORATE biologist, made THE SAME AMOUNT OF MONEY! Absolutely the truth. Maybe he was an idiot, I don't know. But there's a difference between being "money grubbing" and simply wanting a reasonably comfortable lifestyle (after one year of undergrad CS, I was making more than double what the postdoc made, hmm....). Supply and demand, baby. --JRZ
I live in Northern Virginia (home of UUNet, AOL, etc.) and every other day I see another news story lamenting the incredible lack of talented computer/engineering professionals. All these studies also say that we're simply not graduating people fast enough from credible colleges, and many of the jobs out there can't be filled by people with technical institute degrees (though many also can).
Yet here comes the National Science Teachers' federation, or whoever, telling us that we need FEWER computer specialists? Maybe they've been working with those chemicals a little too long, but I know a decent number of people in physics, and they are FRIGHTENED by the prospects out there for a PhD physicist. Research in some pure sciences has become so esoteric and so expensive, that only a few top institutions can do really cool stuff. Of course, it ain't easy to get a job at one of those spots. So should we wonder why people enjoy CS, where you can tackle the coolest, most cutting-edge problems from the comfort of your own workstation anywhere in the world?
I especially loved the comment that "CS simply doesn't teach the kind of critical thinking that you learn from physics." What a blatant lie! Computer Science students will generally be exposed to deep mathematical theory (abstract reasoning), algorithms (abstract problem solving), hardware (hands-on learning), research techniques and methodology, and systems/applications programming (applied problem solving). What critical thinking was not covered there? I find even the most mundane of these topics (say, debugging, which we all hate) to be a lot more stimulating than titrations! (no offense to the chemists)
Academia is like that, as is any large bureaucracy. Everyone needs to preserve his own little niche by convincing the world that comparative Middle Eastern philosophical history, or whatever, is the most important subject in the world.
--JRZ
1) I don't see massive speed improvements coming anytime soon. Sun wants to push the server version. The press hype stresses the fact that so needs a powerful computer to run. So basically, they let the desktop version continure to have issues, so the thin-client version looks very attractive in comparision. The Sun Community Source License gives Sun final control of all patches. Which means you can expect simple bugfixes (it crashed when I did this, so I fixed it) to be used, but anything bigger is very doubtful.
2) Ouch. What does this mean for poor Corel? Yes, I know that WordPerfect is vastly superior in many ways, but its recent mild resurgence in the Windows world has had a lot to do with its positioning as a low-end office suite. Manufacturers looking for a cheap way to add value need look no further than StarOffice. Not a big deal in the short run, but it could prove really dangerous eventually to those poor Canadians.
3) Hopefully they just won't bloat it any further. Do what MS has always hoped to do: wait for hardware to catch up. With 128 megs of RAM and a 600+ mHz PC, SO would be quite bearable.
4) Anybody used Applix Anywhere, Applix's free-ish Java office suite? Man, that's slow as a dog. Wonder how it's doing commercially.
--JRZ
The one actual change to the program mentioned by Sun so far (aside from generally more interest in server distribution) has been their intention to switch to XML for better compatibility. This is fantastic overall, even though XML makes fairly bloated files in terms of size.
--JRZ
Well, as the author of the first "Open Business Plan" article, I have to say that it's great to see a more open debate about the future of Red Hat. I enjoyed this new piece very much, but I think it's only fair to point out the areas where I disagree with it. Of course, if you look at the first comment on Freshmeat (by Benjamin Scherrey), he seems to have already posted most of these comments too.
Here's my overall issue: Red Hat has a business model. They want to provide support and services for Linux. She the fundamental question is: how to they increase their Linux services business?
A focus on gcc (as per point 1), for instance, would put Red Hat in conflict with Cygnus, which has made a very successful business for years by controlling the market for gcc support and expansions. Cygnus knows that its future is directly dependent upon these compilers, so they have a very aggressive plan of improvement for the GNU compiler system already. Additionally, what CEO says, "Gee, I think that operating system sounds great for our business, but I'm not happy with their compiler's adherence to ANSI standards, so let's go with a different choice instead"?
I think that points 2 and 4 of this article are examples of the way in which many users prefer to look at Red Hat as a software distributor, rather than a services company. RHAT has never claimed that their market would be the desktop. Why? Because servers need expensive support from the vendor, while desktops don't. That's why the company is willing to see others provide the productivity software, while they stick close to the more lucrative server/OS strategy. The issue of a subscription is more of something that SuSE would do (actually, I believe they already do it). Why? Because SuSE appeals to hobbyists (I'm a SuSEr myself). They bundle all the software on Earth, much of which is irrelevant to businesses, and they don't worry too much about integrating the various packages or providing graphical administration. But business unsers want polish, ease of use, and itegration, not just the latest thing to come out. They also see upgrading as a difficult process to be undertaken only when compelling. Remember, millions of copies of DOS shipped last year! Why? Because many folks couldn't find a reason to upgrade from even this incredibly outdated technology. These are not customers who feel the need to go from kernel 2.2.11 to 2.2.12 as soon as the new one comes out. Also, retailers HATE that Linux distros go from 6.1 to 6.2 before even the 6.0 version sells out. It makes their inventory worthless. If you look through any major computer store, you'll probably still find copies of OpenLinux 1.3 on the shelves (possibly being resold under a different name). I've seen CompUSA put OpenLinux 1.3 next to OpenLinux 2.2 and charge MORE for the old version!
I have to disagree most strongly with point 6 for many of the same reasons that Mr. Scherrey pointed out. Borland's strength is still in Windows software development tools, a business about which Red Hat knows nothing at all. Worse, they're a software developer, and RHAT does NOT want to be in the software business, they want to be in support. Even more importantly, both Cygnus and Borland already recognize the importance of their Linux strategies. Cygnus is NOT going to drop Linux support in gcc or Code Fusion any time soon, because that's the vast majority of their market base! Similarly, Inprise is already committing to a future as a cross-platform tools provider. As an independent company, they will contribute at LEAST as much to Linux as they could if they had been acquired. Red Hat needs to acquire targets who, if left alone, will NOT help the Linux OS as much as they could. Any acquisition would also have to address Red Hat's desperate need to gear up its consulting arm.
I have mixed feelings on the LSB. It's somewhat disappointing that they have not yet been able even to articulate a vision for how they see the Linux standards process playing out in the future. If they could create a base that moves quickly and allows vendors to add their own value, it could be interesting. However, Red Hat holds such a dominant position in the market that they can afford to make their own standards and have others follow them. Look at "Code Warrior for Red Hat Linux"! Red Hat never asked them to support their distro alone; Metrowerks just didn't want to commit right away to supporting more than one system. The bigger challenge will be Corel's Linux, which has big corporate backing, but doesn't even use the RPM format. If Corel and Red Hat could agree on standards, however, that would basically force the LSB members to follow along.
Well, like I said, it's great to see an increasingly open discussion of these issues. I hope these discussions set a precedent for more, similar debates in the future.
--JRZ
Ok, so I'm an undergraduate, but I thought I'd add my comments anyways from what I've seen/heard and know about the faculty.
If you're into theory/algorithms, it's a pretty hard core place. Sedgewick, Tarjan, and their buddies can compete with anybody out there in the field (IMHO, Donald Knuth would be an exception to that statement). A lot of the folks on this side of the department say that the best CS grad student was a math undergrad student.
There's also a new program in applied and computational computing which looks really promising. It's all about integrating CS and other fields that require intensive computer modeling, and they're putting some very cool folks in the program.
Also a pretty strong program in computer/network security, but I don't find that stuff so interesting, so I couldn't tell you much (except that Ed Felten, the government's tech guy in the MS case is involved in that program).
So, basically, I think I agree with the overwhelming sentiment of the others here today: the school you choose should depend on what you want to do.
Actually, I was going to write something about this in the original article. I love Qt and would really like, at the very least, for them to release a low-cost/free version for commercial development on Unix/Linux, but keep charging for the cross-platform version (of course, LGPL all around would be even better for the consumers, but they're still a business).
It seemed, however, that one of the other Linux distributors would probably be better suited to make this move. It would be pretty awkward for RedHat to abruptly drop GNOME/GTK right now, but if they didn't move to Qt as their primary platform, it wouldn't be a very good investment.
Still, something has to give with the current Qt licensing system. It's ridiculous to charge $1500 per developer for a widget set, even if it is fantastic. How much is MS Visual Studio Pro now? $750? And companies with an absolute focus on cross platform capability can always settle for Java for free, despite its faults. There you get a large, skilled developer base and your choice of many IDEs. If Troll-Tech (or their future owner) doesn't wise up in the next year, they'll rapidly be made irrelevant in the commercial markets when Swing and fast JVM's really become common.
--JRZ
I totally agree with you about the patchwork license and free software issue. Anyone who feels strongly about using GPL-only software can not really use KDE. /."), because it's stripped down. More importantly, KOM/OP is a fantastic object model with a very easy learning curve, and the KDE2.0 daemon implements an extremely easy-to-use naming system. I've been nothing but impressed by KOM/OP and the thought that went into it, especially as I use KOffice, which is already quite advanced.
However, I'd take issue with your comments on the core technology and the object model. Bonobo and ORBit have a lot of good features, but they inherently suffer from GNOME's focus on C-programming and ORBit doesn't even have C++ bindings! While MICO is a very large ORB, KDE 2.0 will actually use tinyMICO, a scaled-down version that cuts out the unnecessary crap. MICO is a very serious CORBA 2.2 implementation, with many more features than ORBit. ORBIT, however, is clearly the faster ORB, much as MySQL is faster than, say, Oracle (ooh, nice tie-in to a recent "Ask
I don't mean to start a flame war. But it is important to look at the details of these implementations. Ideally, we'll get a level of object interoperability in the near future.
--JRZ
QT/KDE themes are a bit confusing, because there are different types. The current KDE themes in 1.1.2 are pretty typical pixmap stuff, not super fast, but I don't have any problems on my PII/350 with a crappy 4 MB on-board ATI.
Qt 2.0 themes, however, can be quite fast. They don't rely on pixmaps, but just override a virtual drawing function. Very cool stuff.
My one flamebait on the GTK/Qt debate: if you use C++, Qt is a dream: pure, object oriented libraries. Not a wrapper like GTK-- (which isn't bad, though) or MFC in windows. If you use C, stick to GTK.
--JRZ
Well, Instant Messenger as a tradmark isn't really ridiculous. For AT&T to come out with a similar product and give it an identical name is really shady. What if Mitsubishi came out with a car called the "Plymouth" and argued that, hey, Plymouth is a popular name. Bullsh*@#$%&t. We don't complain about IBM which has a TM on "International Business Machines", and I don't see how you can get much more generic than that.
It's just fashionable to bust on MS and AOL these days.
--JRZ
1) Hopefully by now we all know that no such tax has been proposed in any real way (especially since the USPS is entirely self-funding these days). But the topic is still relevant, because there WILL be internet taxes of some sort.
2) Paying (in some way) for network services based on usage (rather than "unlimited use") is a MUCH more economically efficient. Think about all the spammers who wouldn't be in business if they had to pay $.01 for each email. What if a basic net connection were free, but you had to pay 1 cent per email and 5 cents for each meg of download? Hey, it would screw me too, but it might make us think about the fact that network resources ARE limited, and some clowns suck it up without thinking about that. I'm not saying that this should be a tax, though.
3) Sales tax will eventually have to be applied to internet purchases. As more commerce moves online, it's inevitable, and it's no worse than paying the sales taxes we do now. I'd say, half of the tax goes to the destination state of the sale, half to the originator. Then you don't get much of a break for locating offshore to avoid the source tax.
--JRZ
Check out the whitepapers on developer.gnome.org and developer.kde.org. Yup, obviously a whitepaper is quite different from thorough design description, but a few of these docs, especially those related to KOM (KDE Object Model) lay out the costs/benefits and architecture of the chosen solution. As far as the topic of whether engineers can write good design documents, I'd say that many can't, but a few are actually VERY good at laying out all the ideas ahead of time. In those cases, the documents can be quite helpful to others. The most important part, however, is that the document should be a useful roadmap to the programmer who writes it. If you think all the project's aspects through before committing a line of code, sure, some things will have to be changed as the project progresses. But at least you won't end up with a poor hack that needs to be redone from scratch. The final barrier that can exist is that many engineers I've worked with don't have English as their first language. Worse still, some are C programmers and a few are even stuck on Hungarian notation. Ah, I see, "struct _qXrtsh" has a field called "lpszTbsdx", yes, that makes great intuitive sense to me. --JZ
Any time you can get an increase in CPU speed directly proportional to the increase in cost, you should jump on it (as in this case). Typically, high end processors demonstrate hard core diminishing marginal rates of return. Thus, the ten percent increase in performance you get by stepping up to the latest Xeon might increase your outlay by 50+%. Those $13,000 PA-RISC workstations, to take a different example, are NOT ten times as fast as a solid $1300 system. They'd be lucky if they could clock in at twice the speed. There's simply no way to say: "Ok, you can buy 1 point of SPECInt for $20."
--JZ
Yup, the server is having a bit of trouble today. Too much good publicity ;) Try http://fara3.cs.uni-potsdam.de/~smeier/kdevelop_ho me/ for a mirror. Please, please, report bugs so that KDev 1.0 non-beta will be as clean as possible! --JZ
Ok, I just have to second the advice that you switch to KDE1.1.1 or better to improve speed, but I'd also recommend trying WindowMaker + KFM (the KDE file manager). With WindowMaker you already have the dock, background manager, etc, so you don't need a panel or kbgndwm or anything like that. KFM with the new KDE 1.1.2 hi-color icons looks damn cool and it fills in that empty space on the left side of the screen (I use dock to the right, clip at the bottom) that always seemed so asymetrical.
Plus, KFM by itself only needs a few megs of memory.
--JZ
Why is this still my typical party conversation? ..
Me/her:
Her: So, what're you majoring in?
Me: Well, I don't know... maybe CompSci.
Her: Oh. . Do you do that sort of thing a lot, I mean computer stuff?
Me: Oh, you know, I to keep it diverse, but there's so much happening in computers these days, it's really interesting stuff.
Her: Well, did you do that sort of thing a lot in high school?
Me: No, I actually wanted to me a comp lit major, or maybe econ, but the CS department here is just really great, and it's a fun place to be.
Her: Well, I gotta go.
Me: No, wait! Really, I just want to start a net company and get rich! I swear!
Her: Oh. That's cool. Give me a call sometime. Preferrably right before your company goes public.
--JZ
The optimism isn't wrong, that is, but the angle is way off. Which OS will take better advantage of this new EasyPC spec:
a) The OS whose company was deeply involved in writing the spec and which is coming out with a specific version just to support it (Millenium)
or
b) The OS that gets little, belated info about new hardware initiatives, can't run WinModems (which are surely a central part of EasyPC), and only has experimental PnP and USB support now?
The real future for Linux comes when more hardware vendors are willing to write drivers for the LinModems of the future. Oracle's $150 Linux NC will be a nice start. And if Corel gets good penetration, it will certainly help.
--JZ
I know it seems scary, but it's a perfectly valid, and even necessary, legal step. If people can freely spout libel all across the internet and get away with it just because they're "Anonymous Cowards," it will become a MAJOR problem.
Think about it this way: if I own a software company, I could post fake, libellous leaks about my competitors on stock sites across the country, claiming to be an employee who wishes to remain anonymous. Should a smart investor take advice from an anonymous leak? Of course not. Are there a lot of stupid investors out there? I don't think I have to answer that one.
I think it was Salon that had an article about major studios creating fake fan sites for their movies (duh, that's a shocker), but, much worse, also creating fake anti-fan sites to badmouth their competitors. When this kind of blatant lying rises to the level of true libel, we need to have some recourse.
Remember, you can NOT be sued for simple, constitutionally protected speech or plain old stupid ideas. Just libel. And that's the way it should be.
--JZ
Ok, we all know we're talking about a low-end E-commerce site here, so here's what's missing from the package:
-Easy to use database
-Graphical database tools
-Simple, graphical server/e-store configuration tools
-Web authoring tools
They can't sell a package that centers on MySQL, because that would incur their weird licensing costs (Although they can include it as an option), but maybe a PostgreSQL/MySQL option would be allright. KMySQL is a great, free database frontend that has plugins to work with other DBs (no Postgres yet, though) and could be great if RH would invest in it a little.
Templates are easy and/or freely available.
Hmm... what does that leave undone? Easy tools for building e-commerce apps. Something easier than, say, Zope or another real app server. People just want a "Yahoo Store" or the like.
--JZ
Well, I would say: no biggie, tons of other top app servers out there will/already do support Linux (which is certainly true), but every mention of app servers just reminds me how cool JSP and EJB are. And how slow Blackdown is. Linux needs a top-notch JVM before its server-side Java can take off. Check out WebTechniques benchmarks of GNUJSP on apache on Linux. The poor author adores the OS and webserver, but, even with tweaking, it still took SECONDS to server a simple JSP. No joke. What we need from Sun is not their app server, but their support for a real Java2 on Linux. Maybe IBM can do it. . .
--JZ
Sorry, you won't see VMWare or WINE working on an RS/6000. Both require an x86 processor.
--JZ
I use KDevelop almost exclusively now. There are maybe 2 or three crashing bugs that I've seen, but most of those are rare and/or fixed in CVS versions. It is by far the best way to learn about KDE/Qt programming, because it includes great documentation and an AppWizard to get you started.
Needless to say, I'm a big fan.
--JZ
Unfortunately, it's not quite there yet. Even more unfortunately, I haven't seen any change in their web site in months. It doesn't have a class parser/class browser, which is really the key thing that I need in an IDE. The stub code is there for it to be added in, but the project seems a little stalled.
Check out www.netbeans.com. They have a pure Java IDE that's actually quite good. Also free for non-commercial use and/or trial.
--JZ