The plugin for Netscape under Linux is available, I've got it installed at work and at home. And I'm actually very impressed with the Shockwave format:
Compact vector animations which download quickly over a modem. Okay, it can be overused, but then so can DHTML.
Like much in our world, the merchant shipping industry is a haven for jobsworths, tightwads and schemers. Rather than have tanks cleaned properly in port (at a cost) unscrupulous shipowners encourage crews to simply flush them at sea. Stand on any beach near a shipping lane, and you can guarantee there'll be tar-like deposits everywhere from this heinous practise.
There's little governments can do even if they have a fair idea which ships or copmanies are to blame. Do it in international waters, and no-one can get you. And if you get caught... Well, almost every large merchant vessel flies under a flag of convenience. This enables the captain or owners to say ``well, sue me in Liberia''.
A friend who works as a deck oficcer commented to me once that another problem is hopeless crews. OFficers tend to be trained and conscientius(sp?), but crews often don't speak the same language as the officers and have no training. In the British merchant navy, many crews are Philipino simply because the owners can get away with paying them shit wages.
Anyways, enough of the ranting. I just hope no more of these little penguins succumb to the oil.
They've updates the developer site since the last time I looked at it. The excellent Objective C docment that dates from the NeXTSTEP era is included under the Cocoa section, whereas last time I looked I'm sure it was under the 'Legacy' section. That bodes really well for the future. It's a shame that Apple don't either:
1) GPL the OpenSTEP API so more apps are written in a cross platform manner (imagine stripping down MacOSX and installing WindowMaker).
2) Help the GNUstep guys to acheive the same things as outlined in point one.
That's fantastic news. Unfortunately my access to a news server is sporadic, so I can't check whether the C++ rumours were well founded. I suppose I could check Deja News, but I find it painfully slow.
What we need now is for Apple to sponsor an atempt to get an ANSI standard for Objectove C. Ideally Stepstone, the GNU Objective C people and Apple should form the committee. If they ever do, and a good standard results then I might be able to remove -Wno-import from my compiler flags under GCC!
There was some talk on the Objective C newsgroups about Apple dropping ObjC for C++. I also heard that they had reversed this decision. Does anyone out there know how closely Mac OS X is tied to NeXTSTEP and the OpenSTEP API's?
Let me guess, your a real man. You code in emacs or vi. Well under X I'd rather use NEdit anyday. Rather than having to learn the archane syntax and multi keypress commands of emacs I'll stick with an editor that lets me get real work done quickly.
And on the console I'll stick to vim. Loads up fast on the overloaded servers I telnet to. Doesn't require ridiculous amounts of resources and disk space like emacs does.
When I want to learn another Lisp variant (I already know Scheme thank you very much), I may give emacs another chance. Until then...
Something relevant to the Slashdot article? Well what could one say beyond ``nice new features and bugfixes''. Relevant to the Yahoo! FreeBSD advocacy? Well, I think I illuminate the reason why Yahoo! use it.
As for my general comments on FreeBSD, I gained all my experience of FreeBSD (as opposed to NetBSD and pre-Solaris SunOS) while at Yahoo!.
Do you see the relevancy there? The logical flow of commentary? Maybe not - in which case I'll just point out that this is Slashdot, and any reasoned comments are a godsend compared to the Natalie Portman or gritsboy posters.
Or maybe you're just bitter that I can see faults in FreeBSD. Well, rest assured I can see the flaws in all the operating systems I use, so unfortunately for you I can't be labeled a Linux/Solaris/whatever bigot.
But there's the rub, had this been an article on a new Linux distro that I use, I may have been posting critiques of Linux. There's your relevancy.
Chris Wareham
The Queen's English - getting quite offtopic
on
FreeBSD 3.4 released
·
· Score: 2
The common intpretation of `elitism' is as a pejorative. This could be attributed to a fear of ones peers, but at the close of the twentieth century I link it to something else. Elitism - a belief that something is better than something else - has been one of the most destructive forces in this century. Look at Nazism, Social Darwinism or the Khmer Rouge. Elitism has been adopted as a term for paternal or downright oppressive ideas.
I'm impressed that you still associate it with it's original dictionary meaning, but that doesn't stop others interpreting it differently. The English language evolves, and todays slang will be part of tomorrows codified language. I'm particularily aware of this as a learnt Finnish at university. The codified language is overseen by a government body, and all loan words (foreign words adopted into Finnish) are closely scrutinised. This has lead to a dialect that reflects a certain section of Finnish at best. In fact, no one speaks the language as it is in the dictionary.
The same is true of English. In the UK we have societies that campaign for the `Queen's English', an illusory dialect that scoffs at slang or regional pronunciation. This is `elitist' as in the popular definition of the word, and tries to stop the spoken language evolving.
Some of this is evident in the way ``old school'' Unix people look at Linux and projects like GNOME/KDE. It simply masks their fear of the new, just as Nazism often masked the German middle classes fear of the future.
This is why many people suspect FreeBSD bigots are clinging onto an antiquated set of tools, API's and configuration methods. This attitude is clearly not evident in the FreeBSD developers - but then what incentive is there in maintaining the statis quo?
I suppose it would be elitist to point out that inconsistencies like "with my prefference of OS, merely with my preferrence of people" are very frightening to see in a programmer?
But these are attributes [remotely administer... depend on not to crash] you want to have in virtually any computer system!
Having briefly worked in a Windows NT only environment, I couldn't agree more. However for reasons beyond many people's control they are often lumbered with such systems that they know aren't stable or easy to administer. An irreplacable system may depend on an OS feature, or only be available for a buggy version of that OS.
In an ideal world, I would choose the most suitable system for a given purpose. In the real world I'm often given an existing system, or company politics demands that I use brand X for a new one.
I'm currently working on a system that cries out to be rewritten, (it's a Perl hack thats become unmaintainable, but don't tell Tom). I'm not able to rewrite it though, because I would have to move it from Solaris to NT. Company policy stipulates that all new development is ported to Microsoft operating systems and tools.
As a consequence I'm having to rewrite bits of the system, adhering to the old design, and running it on an out of date version of Solaris.
For a server OS I require a system I can remotely administer and depend on not to crash. FreeBSD suits this purpose very well. I have historically worked for companies with remote (think on the other side of the world, or next continent) server farms. Developing on those machines was either painfull, or out of the question because they were production machines. This meant I needed a desktop system that I could use for development. As I'm often the person evaluating new software (entire packages or simply upgrades), root access to a machine that no one else depended on was also essential.
For the desktop I require something that I can rapidly upgrade and has a wealth of niceties. The typical Linux distro comes with a massive amount of software, and something like RPM make upgrading a doddle.
I found FreeBSD performed poorly on low end hardware, something that Linux excels at. The libraries on FreeBSD 2.2.8 were odd - they deprecated BSD system calls that I had been used to on SunOS. I had much more success coding on the more SVR4 like Linux, and then compiling on FreeBSD for production purposes.
Arguably, you could consider Windows as a good desktop OS. It has lots of simple toys, and you can telnet to your development box. Run an X server and it's like you're on a Unix workstation. The flaw with this logic is if you're developing for a Unix OS, why not run one? This is why I always replace NT with Linux when I arrive at a new contract.
To conclude, I'm not knocking FreeBSD. It is a fine OS, and if you're from a BSD background (as opposed to a System V one) it makes the logical choice. My *personal* choice is Linux for the desktop, and my posts state my *personal* issues with FreeBSD in that environment. And with version 4.0 it looks like most of these will disappear.
Yes, but the FreeBSD attitude I encountered was along the lines of ``Why make it easy'' or ``If it is easy it must be crap''. As I noted, I can configure my networking with route and friends. I can configure my kernel by editing a commented text file. But this shouldn't preclude the inclusion of simple to use tools, or make me any less of a `power user' if I use them. Your whole comment just reeks of elitism, and that isn't anything to be proud of. Most people are elitist because they're insecure, not particlularily good in their field. That's why I refrained from lumping FreeBSD's maintainers with the users.
I agree that a trend towards not teaching complex computer subjects is dangerous. I was taught assembler, but program in C or Objective C. My point is that knowing assembler makes my code in C slightly more optimal, but I'm not going to code in assembler if I can avoid it. Perl is often touted as a language that non computer-scientists can benefit from - it gets the job done without necessarily requiring knowledge of what's going on in the guts of the machine. Are you saying that all those people who aren't programmers by trade, yet use Perl for the odd hack should be sneered at? It certainly looks that way judging by your post.
Having most likely burnt my bridges I'll leave it at that.
The recent discussion on what constitutes a Mozilla alpha seems a bit academic now. I've also been trying the recent nightly builds, and stability has been improving rapidly.
My definition of an alpha is when the product is close to feature complete. Mozilla is almost that, and once the major bug fixes are done it's beta time - possibly with M13? It certainly looks that way.
I'm looking forward to replacing Communicator with Mozilla once and for all. Then I'll be down to just two applications that rely on Motif - one of which I'm rewriting to use GTK+ (the moxfm file manager), and another that will prbably have to remain a Motif app (NEdit).
Chris Wareham
A couple of comments on Yahoo! usage of FreeBSD
on
FreeBSD 3.4 released
·
· Score: 3
Having worked for Yahoo! for several months, I have the following points to make about their usage of FreeBSD:
David Filo tried an early version of Linux (a Slackware distro I beleive), but didn't have much success in getting it installed. The legend at Yahoo! is that a copy of FreeBSD was passed onto him, and he only gave it a try because it installed first time. A slightly dubious reason for choosing one OS over another, but at the time FreeBSD was certainly more mature than Linux.
Yahoo! use large numbers of uniprocessor machines with each HTTP request marshalled out on a round-robin basis. When the load gets too great, it's simply a case of adding more machines and tweaking the boxes that assign requests. This approach was taken because multiprocessor setups are that much less reliable.
My experience of FreeBSD was that it makes a fantastic server OS, but version 2.2.* didn't cut it as a desktop platform. I have a copy of 3.3 sitting in my desk drawer and I'm keen to see what's changed from a workstation user's perspective, so don't flame me about using an old version.
My major gripe with FreeBSD is that upgrading can be a pain. The ports and packages collection is not the panacea that many FreeBSD users seem to believe. My hard drive soon became full of half-working applications, with little or no control over what was being installed and where.
This may have been an issue with running FreeBSD 2.2.8 when version 3.1 was already available, but the ports and packages I tried came from the 2.2.* branch. Now that I've left Yahoo! I'm back to using Linux on the desktop, and NetBSD on my Sun servers.
There is undeniably an elitism amongst FreeBSD users that I neither liked nor could I understand. One of the regular complaints I heard was that Linux installation and configuration tools were uneccessary bloat. Well, I can hand edit a kernel config file on both Linux and FreeBSD as well as the next guru, but I'd much rather use 'make menuconfig'. This kind of carping was sheer elitism, and underlined that a sea change in FreeBSD users attitudes was required (if not in that of the developers). Otherwise, the OS would become nothing more than a *tiny* niche platform. With the alternative installation programs touted for version 4.0, it looks like this change is coming.
The problem with the TurobSparc goes beyond the bootloader. The mmu is slightly different to all other 32bit SPARC processors, and none of the SparcLinux developers have acces to one.
If Sun or anyone else could donate a SS5 170Mhz - or if RedHat could buy David Miller one - then there might be some movement on theis issue. I for one would like to upgrade my SS5 110Mhz to a TurboSparc chip, but I haven't even looked into the possibility of doing this thanks to the inability to use Linux on it.
I do recall that NetBSD runs on the Turbo, so perhaps you want to look into installing it, and then adding RPM to give it that RedHat feel.
I've been programming Motif for a couple of years, ever since Sun dropped OpenLook and my then employer switched to Solaris from SunOS.
Motif has an intersting history, as it was partially funded by none other than Microsoft. They saw it as a useful way of testing UI design prior to the big overhaul to Windows 3.1 that resulted in Windows 95.
As I've always been a little bit partisan about Motif (XView was far easier to program with), I've often wondered if Motif is best seen as an Windows 95 GUI prototype. Motif certainly needs an overhaul in places - the file dialogs for instance.
And as a sidenote, I recall a former Open Group programmer saying that they toyed with the idea of making Motif open source a few years back. Damn shame they didn't, as I'm sure both the Lesstif and Open Group teams could come up with something far leaner and meaner if they pooled their experience.
Having programmed with Qt and GTK+ (along with Motif, XView and the Athena widgets), I can confidently say that while GTK+ looks ugly to the uninitiated, it is far easier to use than Qt.
The casts that most people complain about in GTK+ are actually an elegant solution to the problem of an Object Oriented toolkit in a essentially non-OOP language. This allows some very clever error handling. The odd thing is that every criticism of GTK+ I have seen is of the "the code looks ugly" variety. These people should try actually *coding* in it sometime.
As for Qt, I will reiterate my point that it is a fine toolkit. I feel that ultimately a novice will have more grasp of C than C++ and will therefore prefer GTK+ for dabbling in X programming.
As I suggested in my first post, many Windows programmers are transitioning to Linux, so C++ may not be an obstacle to using Qt. In that case both toolkits make excellent options for X applications.
As for Qt coverage in the Beginning Linux Programming book, I assume the authors fondness for the GNU license is the main reason for Qt's ommision.
Having programmed with Qt and GTK+ (along with Motif, XView and the Athena widgets), I can confidently say that while GTK+ looks ugly to the uninitiated, it is far easier to use than Qt. The casts that most people complain about in GTK+ are actually an elegant solution to the problem of an Object Oriented toolkit in a essentially non-OOP language. This allows some very clever error handling. The odd thing is that every criticism of GTK+ I have seen is of the "the code looks ugly" variety. These people should try actually *coding* in it sometime. As for Qt, I will reiterate my point that it is a fine toolkit. I feel that ultimately a novice will have more grasp of C than C++ and will therefore prefer GTK+ for dabbling in X programming. As I suggested in my first post, many Windows programmers are transitioning to Linux, so C++ may not be an obstacle to using Qt. In that case both toolkits make excellent options for X applications. As for Qt coverage in the Beginning Linux Programming book, I assume the authors fondness for the GNU license is the main reason for Qt's ommision. Chris Wareham
1) The authors a big fans of GPL software. 2) The book focuses on C not C++.
As the book is geared towards those transitioning from Windows, it may be an oversight to not include any C++ stuff. However, for those just starting out on Linux, GTK+ is a far simpler introduction to X programming than Qt.
And before someone flames me, let me add that my own personal preference is GTK+, but Qt is also a damn file GUI toolit - especially for cross platform (Win32 and Unix) apps.
I compiled it without the debugging code, as I don't expect to be working on it any time soon - just wanted a quick look at the improvements. I think you'll find that most of the Bugzilla comments regarding performance come from people testing the precompiled versions. These have the debugging code and extra symbols specifically to try and get some valid feedback. However, the vast majority of people are only having a quick peek at Mozilla and are either incapable or unwilling to file a useful bug report.
Personally I've got a soft spot for Sun pizza style cases, and improbably huge 24bit framebuffer cards. Must be a nostalgia thing I suppose. Although someone claimed to be able to get hold of recent Ultra workstations in Pizza boxes instead of those boring mini-towers.
"SGI will try to give Linux a boost by releasing some of its OpenGL graphics software into the open-source community, Stedman said."
Like GPL-ing it? While MesaGL is a worthy alternative, it would be in SGI's best interests to get OpenGL into wider usage. I suppose there may be licensing implications if they were to try and GPL the whole shebang, as many companies have already *purchased* licenses. So perhaps we'll see an open source 'official' OpenGL when those licenses expire...
Runs blisteringly quick on a PII 233Mhz laptop. Also appears quite stable.
But then I'm running Linux. I assume from your ``IE4 on the same hardware'' comment that you're running under Windows. But then everything runs like a dog under Windows...
Slashdot Mirror
The plugin for Netscape under Linux is available,
I've got it installed at work and at home. And I'm actually very impressed with the Shockwave format:
Compact vector animations which download quickly over a modem. Okay, it can be overused, but then so can DHTML.
Chris Wareham
Like much in our world, the merchant shipping industry is a haven for jobsworths, tightwads and schemers. Rather than have tanks cleaned properly in port (at a cost) unscrupulous shipowners encourage crews to simply flush them at sea. Stand on any beach near a shipping lane, and you can guarantee there'll be tar-like deposits everywhere from this heinous practise.
... Well, almost every large merchant vessel flies under a flag of convenience. This enables the captain or owners to say ``well, sue me in Liberia''.
There's little governments can do even if they have a fair idea which ships or copmanies are to blame. Do it in international waters, and no-one can get you. And if you get caught
A friend who works as a deck oficcer commented to me once that another problem is hopeless crews. OFficers tend to be trained and conscientius(sp?),
but crews often don't speak the same language as the officers and have no training. In the British merchant navy, many crews are Philipino simply because the owners can get away with paying them shit wages.
Anyways, enough of the ranting. I just hope no more of these little penguins succumb to the oil.
Chris Wareham
They've updates the developer site since the last time I looked at it. The excellent Objective C docment that dates from the NeXTSTEP era is included under the Cocoa section, whereas last time I looked I'm sure it was under the 'Legacy' section. That bodes really well for the future. It's a shame that Apple don't either:
1) GPL the OpenSTEP API so more apps are written in a cross platform manner (imagine stripping down MacOSX and installing WindowMaker).
2) Help the GNUstep guys to acheive the same things as outlined in point one.
Chris Wareham
Objective-C is here to stay
That's fantastic news. Unfortunately my access to a news server is sporadic, so I can't check whether the C++ rumours were well founded. I suppose I could check Deja News, but I find it painfully slow.
What we need now is for Apple to sponsor an atempt to get an ANSI standard for Objectove C. Ideally Stepstone, the GNU Objective C people and Apple should form the committee. If they ever do, and a good standard results then I might be able to remove -Wno-import from my compiler flags under GCC!
Chris Wareham
There was some talk on the Objective C newsgroups about Apple dropping ObjC for C++. I also heard that they had reversed this decision. Does anyone out there know how closely Mac OS X is tied to NeXTSTEP and the OpenSTEP API's?
Chris Wareham
Let me guess, your a real man. You code in emacs or vi. Well under X I'd rather use NEdit anyday. Rather than having to learn the archane syntax and multi keypress commands of emacs I'll stick with an editor that lets me get real work done quickly.
...
And on the console I'll stick to vim. Loads up fast on the overloaded servers I telnet to. Doesn't require ridiculous amounts of resources and disk space like emacs does.
When I want to learn another Lisp variant (I already know Scheme thank you very much), I may give emacs another chance. Until then
Chris Wareham
Something relevant to the Slashdot article? Well what could one say beyond ``nice new features and bugfixes''. Relevant to the Yahoo! FreeBSD advocacy? Well, I think I illuminate the reason why Yahoo! use it.
As for my general comments on FreeBSD, I gained all my experience of FreeBSD (as opposed to NetBSD and pre-Solaris SunOS) while at Yahoo!.
Do you see the relevancy there? The logical flow of commentary? Maybe not - in which case I'll just point out that this is Slashdot, and any reasoned comments are a godsend compared to the Natalie Portman or gritsboy posters.
Or maybe you're just bitter that I can see faults in FreeBSD. Well, rest assured I can see the flaws in all the operating systems I use, so unfortunately for you I can't be labeled a Linux/Solaris/whatever bigot.
But there's the rub, had this been an article on a new Linux distro that I use, I may have been posting critiques of Linux. There's your relevancy.
Chris Wareham
The common intpretation of `elitism' is as a pejorative. This could be attributed to a fear of ones peers, but at the close of the twentieth century I link it to something else. Elitism - a belief that something is better than something else - has been one of the most destructive forces in this century. Look at Nazism, Social Darwinism or the Khmer Rouge. Elitism has been adopted as a term for paternal or downright oppressive ideas.
I'm impressed that you still associate it with it's original dictionary meaning, but that doesn't stop others interpreting it differently. The English language evolves, and todays slang will be part of tomorrows codified language. I'm particularily aware of this as a learnt Finnish at university. The codified language is overseen by a government body, and all loan words (foreign words adopted into Finnish) are closely scrutinised. This has lead to a dialect that reflects a certain section of Finnish at best. In fact, no one speaks the language as it is in the dictionary.
The same is true of English. In the UK we have societies that campaign for the `Queen's English', an illusory dialect that scoffs at slang or regional pronunciation. This is `elitist' as in the popular definition of the word, and tries to stop the spoken language evolving.
Some of this is evident in the way ``old school'' Unix people look at Linux and projects like GNOME/KDE. It simply masks their fear of the new, just as Nazism often masked the German middle classes fear of the future.
This is why many people suspect FreeBSD bigots are clinging onto an antiquated set of tools, API's and configuration methods. This attitude is clearly not evident in the FreeBSD developers - but then what incentive is there in maintaining the statis quo?
Chris Wareham
I suppose it would be elitist to point out that inconsistencies like "with my prefference of OS, merely with my preferrence of people" are very frightening to see in a programmer?
Nope. Just pedantic ;-)
Chris Wareham
But these are attributes [remotely administer ... depend on not to crash] you want to have in virtually any computer system!
Having briefly worked in a Windows NT only environment, I couldn't agree more. However for reasons beyond many people's control they are often lumbered with such systems that they know aren't stable or easy to administer. An irreplacable system may depend on an OS feature, or only be available for a buggy version of that OS.
In an ideal world, I would choose the most suitable system for a given purpose. In the real world I'm often given an existing system, or company politics demands that I use brand X for a new one.
I'm currently working on a system that cries out to be rewritten, (it's a Perl hack thats become unmaintainable, but don't tell Tom). I'm not able to rewrite it though, because I would have to move it from Solaris to NT. Company policy stipulates that all new development is ported to Microsoft operating systems and tools.
As a consequence I'm having to rewrite bits of the system, adhering to the old design, and running it on an out of date version of Solaris.
Chris Wareham
For a server OS I require a system I can remotely administer and depend on not to crash. FreeBSD suits this purpose very well. I have historically worked for companies with remote (think on the other side of the world, or next continent) server farms. Developing on those machines was either painfull, or out of the question because they were production machines. This meant I needed a desktop system that I could use for development. As I'm often the person evaluating new software (entire packages or simply upgrades), root access to a machine that no one else depended on was also essential.
...
For the desktop I require something that I can rapidly upgrade and has a wealth of niceties. The typical Linux distro comes with a massive amount of software, and something like RPM make upgrading a doddle.
I found FreeBSD performed poorly on low end hardware, something that Linux excels at. The libraries on FreeBSD 2.2.8 were odd - they deprecated BSD system calls that I had been used to on SunOS. I had much more success coding on the more SVR4 like Linux, and then compiling on FreeBSD for production purposes.
Arguably, you could consider Windows as a good desktop OS. It has lots of simple toys, and you can telnet to your development box. Run an X server and it's like you're on a Unix workstation. The flaw with this logic is if you're developing for a Unix OS, why not run one? This is why I always replace NT with Linux when I arrive at a new contract.
To conclude, I'm not knocking FreeBSD. It is a fine OS, and if you're from a BSD background (as opposed to a System V one) it makes the logical choice. My *personal* choice is Linux for the desktop, and my posts state my *personal* issues with FreeBSD in that environment. And with version 4.0 it looks like most of these will disappear.
As simple as that
Chris Wareham
Dare we ask for a bit of professional knowledge?
Yes, but the FreeBSD attitude I encountered was along the lines of ``Why make it easy'' or ``If it is easy it must be crap''. As I noted, I can configure my networking with route and friends. I can configure my kernel by editing a commented text file. But this shouldn't preclude the inclusion of simple to use tools, or make me any less of a `power user' if I use them. Your whole comment just reeks of elitism, and that isn't anything to be proud of. Most people are elitist because they're insecure, not particlularily good in their field. That's why I refrained from lumping FreeBSD's maintainers with the users.
I agree that a trend towards not teaching complex computer subjects is dangerous. I was taught assembler, but program in C or Objective C. My point is that knowing assembler makes my code in C slightly more optimal, but I'm not going to code in assembler if I can avoid it. Perl is often touted as a language that non computer-scientists can benefit from - it gets the job done without necessarily requiring knowledge of what's going on in the guts of the machine. Are you saying that all those people who aren't programmers by trade, yet use Perl for the odd hack should be sneered at? It certainly looks that way judging by your post.
Having most likely burnt my bridges I'll leave it at that.
Chris Wareham
The recent discussion on what constitutes a Mozilla alpha seems a bit academic now. I've also been trying the recent nightly builds, and stability has been improving rapidly.
My definition of an alpha is when the product is close to feature complete. Mozilla is almost that, and once the major bug fixes are done it's beta time - possibly with M13? It certainly looks that way.
I'm looking forward to replacing Communicator with Mozilla once and for all. Then I'll be down to just two applications that rely on Motif - one of which I'm rewriting to use GTK+ (the moxfm file manager), and another that will prbably have to remain a Motif app (NEdit).
Chris Wareham
Having worked for Yahoo! for several months, I have the following points to make about their usage of FreeBSD:
David Filo tried an early version of Linux (a Slackware distro I beleive), but didn't have much success in getting it installed. The legend at Yahoo! is that a copy of FreeBSD was passed onto him, and he only gave it a try because it installed first time. A slightly dubious reason for choosing one OS over another, but at the time FreeBSD was certainly more mature than Linux.
Yahoo! use large numbers of uniprocessor machines with each HTTP request marshalled out on a round-robin basis. When the load gets too great, it's simply a case of adding more machines and tweaking the boxes that assign requests. This approach was taken because multiprocessor setups are that much less reliable.
My experience of FreeBSD was that it makes a fantastic server OS, but version 2.2.* didn't cut it as a desktop platform. I have a copy of 3.3 sitting in my desk drawer and I'm keen to see what's changed from a workstation user's perspective, so don't flame me about using an old version.
My major gripe with FreeBSD is that upgrading can be a pain. The ports and packages collection is not the panacea that many FreeBSD users seem to believe. My hard drive soon became full of half-working applications, with little or no control over what was being installed and where.
This may have been an issue with running FreeBSD 2.2.8 when version 3.1 was already available, but the ports and packages I tried came from the 2.2.* branch. Now that I've left Yahoo! I'm back to using Linux on the desktop, and NetBSD on my Sun servers.
There is undeniably an elitism amongst FreeBSD users that I neither liked nor could I understand. One of the regular complaints I heard was that Linux installation and configuration tools were uneccessary bloat. Well, I can hand edit a kernel config file on both Linux and FreeBSD as well as the next guru, but I'd much rather use 'make menuconfig'. This kind of carping was sheer elitism, and underlined that a sea change in FreeBSD users attitudes was required (if not in that of the developers). Otherwise, the OS would become nothing more than a *tiny* niche platform. With the alternative installation programs touted for version 4.0, it looks like this change is coming.
Chris Wareham
The problem with the TurobSparc goes beyond the bootloader. The mmu is slightly different to all other 32bit SPARC processors, and none of the SparcLinux developers have acces to one.
If Sun or anyone else could donate a SS5 170Mhz - or if RedHat could buy David Miller one - then there might be some movement on theis issue. I for one would like to upgrade my SS5 110Mhz to a TurboSparc chip, but I haven't even looked into the possibility of doing this thanks to the inability to use Linux on it.
I do recall that NetBSD runs on the Turbo, so perhaps you want to look into installing it, and then adding RPM to give it that RedHat feel.
Chris Wareham
Admittedly it isn't entirely faultless, but I can think of far worse places to live. Like Liverpool. Or Bradford.
Disclaimer: I grew up near Guildford
Chris Wareham
Open up the documentation for Motif, and who is listed amongst the sponsors? Yup, that's right Microsoft.
Chris Wareham
I've been programming Motif for a couple of years, ever since Sun dropped OpenLook and my then employer switched to Solaris from SunOS.
Motif has an intersting history, as it was partially funded by none other than Microsoft. They saw it as a useful way of testing UI design prior to the big overhaul to Windows 3.1 that resulted in Windows 95.
As I've always been a little bit partisan about Motif (XView was far easier to program with), I've often wondered if Motif is best seen as an Windows 95 GUI prototype. Motif certainly needs an overhaul in places - the file dialogs for instance.
And as a sidenote, I recall a former Open Group programmer saying that they toyed with the idea of making Motif open source a few years back. Damn shame they didn't, as I'm sure both the Lesstif and Open Group teams could come up with something far leaner and meaner if they pooled their experience.
Chris Wareham
Having programmed with Qt and GTK+ (along with Motif, XView and the Athena widgets), I can confidently say that while GTK+ looks ugly to the uninitiated, it is far easier to use than Qt.
The casts that most people complain about in GTK+ are actually an elegant solution to the problem of an Object Oriented toolkit in a essentially non-OOP language. This allows some very clever error handling. The odd thing is that every criticism of GTK+ I have seen is of the "the code looks ugly" variety. These people should try actually *coding* in it sometime.
As for Qt, I will reiterate my point that it is a fine toolkit. I feel that ultimately a novice will have more grasp of C than C++ and will therefore prefer GTK+ for dabbling in X programming.
As I suggested in my first post, many Windows programmers are transitioning to Linux, so C++ may not be an obstacle to using Qt. In that case both toolkits make excellent options for X applications.
As for Qt coverage in the Beginning Linux Programming book, I assume the authors fondness for the GNU license is the main reason for Qt's ommision.
Chris Wareham
Having programmed with Qt and GTK+ (along with Motif, XView and the Athena widgets), I can confidently say that while GTK+ looks ugly to the uninitiated, it is far easier to use than Qt. The casts that most people complain about in GTK+ are actually an elegant solution to the problem of an Object Oriented toolkit in a essentially non-OOP language. This allows some very clever error handling. The odd thing is that every criticism of GTK+ I have seen is of the "the code looks ugly" variety. These people should try actually *coding* in it sometime. As for Qt, I will reiterate my point that it is a fine toolkit. I feel that ultimately a novice will have more grasp of C than C++ and will therefore prefer GTK+ for dabbling in X programming. As I suggested in my first post, many Windows programmers are transitioning to Linux, so C++ may not be an obstacle to using Qt. In that case both toolkits make excellent options for X applications. As for Qt coverage in the Beginning Linux Programming book, I assume the authors fondness for the GNU license is the main reason for Qt's ommision.
Chris Wareham
Qt was probably not covered for two reasons:
1) The authors a big fans of GPL software.
2) The book focuses on C not C++.
As the book is geared towards those transitioning from Windows, it may be an oversight to not include any C++ stuff. However, for those just starting out on Linux, GTK+ is a far simpler introduction to X programming than Qt.
And before someone flames me, let me add that my own personal preference is GTK+, but Qt is also a damn file GUI toolit - especially for cross platform (Win32 and Unix) apps.
Chris Wareham
I compiled it without the debugging code, as I don't expect to be working on it any time soon - just wanted a quick look at the improvements. I think you'll find that most of the Bugzilla comments regarding performance come from people testing the precompiled versions. These have the debugging code and extra symbols specifically to try and get some valid feedback. However, the vast majority of people are only having a quick peek at Mozilla and are either incapable or unwilling to file a useful bug report.
Chris Wareham
"Sexiest hardware in existance ..."
Personally I've got a soft spot for Sun pizza style cases, and improbably huge 24bit framebuffer cards. Must be a nostalgia thing I suppose. Although someone claimed to be able to get hold of recent Ultra workstations in Pizza boxes instead of those boring mini-towers.
Chris Wareham
"SGI will try to give Linux a boost by releasing some of its OpenGL graphics software into the open-source community, Stedman said."
...
Like GPL-ing it? While MesaGL is a worthy alternative, it would be in SGI's best interests to get OpenGL into wider usage. I suppose there may be licensing implications if they were to try and GPL the whole shebang, as many companies have already *purchased* licenses. So perhaps we'll see an open source 'official' OpenGL when those licenses expire
Chris Wareham
Runs blisteringly quick on a PII 233Mhz laptop.
...
Also appears quite stable.
But then I'm running Linux. I assume from your ``IE4 on the same hardware'' comment that you're running under Windows. But then everything runs like a dog under Windows
Chris Wareham