Tell me that a week after the Mac-on-Linux folks get their hands on a prerelease. If it's not made to work with Mac OS X for Intel fairly rapidly, I may die of shock. My bet is on support WELL before the first official Intel-based macs hit the shelves.
I don't expect to be able to run Mac OS X natively on non-Apple hardware. I don't think I'd want to, given the likely driver quality etc. However, running it under MoL sounds *very* attractive, and a Xen port I would positively kill for.
Note that my perspective is mostly that of a developer. More than anything, I just want to be able to conveniently test on Mac OS X. It is *such* a pain ssh-ing into the slow eMac at work and using VNC to test things - especially since the braindead thing only WOLs from sleep, not poweroff. Being able to use my nice fast Athlon for test builds and actually test each configure.in.in change on Mac OS X before checking it in would be bliss.
Sadly, I doubt Apple will ever do anything as useful as release a Mac OS X developer edition that runs on Xen, so I'll most likely be stuck with MoL (hopefully).
As the other poster noted, "the best" is rather ambiguous. I wouldn't touch Portage with a barge pole, not only because of efficiency (my time is worth more than a 1% runtime speed increase) but also because as a developer, I don't trust it. I've seen too many *weird* problems from Gentoo users, and most of them go away when they recompile some other random part of their system. Scary.
In a world of stable APIs and ABIs, Portage might be awesome. In this world of unstable, often subtly changing app interfaces, limited insight into dependencies, and weird interactions, I don't think it's all that good an idea. I'll take RPM & yum/YaST/urpmi or dpkg and apt any day.
Knoppix's detection is pretty darn good, but then again everything "just worked" on my FC4 system too. I think this is coming along nicely on most systems.
I think BSD rc scripts are even worse than System V ones. I remain bewildered as to why, in this day and age, we can't have something less horrifyingly archaic and painful that provides the ability to run rc scripts for backward compatibility. Preferably something not quite as complex and full of XML as SMF (from Solaris 10).
As for installers... IMO, boring. They're all pretty good. I couldn't fault FC4's for example.
Anyway - it *is* possible to combine all these. You'll end up with a distro that has other major shortcomings though, such as a user base of four, irregular security updates, and very little reason to trust that it'll stay with the times. Perhaps more importantly, you won't have much package QA or testing across all that wide a range of hardware, so your distro's quality won't be up to scratch.
Also, you'll find that upstream developers just won't be interested. "Your app doesn't compile out of the box on 2bit linux, 1998 unmaintained edition. Help!" . Riight. Note that for this purpose I *am* an "upstream developer".
Personally, I don't think the answer is to seek perfection in distros. I think the answer is to seek more consistency - not necessarily by forcing everything to be the same, but by providing a level of consistency and certain things that apps can rely on to be the same. The LSB is theoretically a good start on that, but the work done by FreeDesktop.org is in my view much more so by view of seeing rapid and comprehensive adoption.
If you want to help Linux distros improve, help FreeDesktop.org .
I was a geek in high school. I'm sure many here were. Y'know what? Any isolation I felt - and yes, there was quite a bit in the first few years - was largely my own fault. I had people try really hard to be nice, to include me, etc and I just didn't know how to respond. I mostly got confused, or thought they were trying to tease / mock me. In my defense, that's because that used to be a problem, but I think mostly because of my lack of social skills rather than any geekyness. I was one of those people who wakes up one day and says "huh. People. Who would'a thought."
One I learned some utterly minimal social skills, things improved a lot. This frequently meant walking home talking about coding / maths with a couple of friends, though that's not all I did. Anyway... the point is, I never, *EVER* saw any trouble or hostility from the "mainstream" once I reached high school. The worst I got was a bit of teasing every now and then, and even that was pretty tame and only during the first two years. We have a term here - "Year 9s are animals." It's kinda expected. After that, any distance from them was mostly because I just didn't "get" their interests - which at the time mostly seemed to involve getting so drunk they threw up. Yay. Not.
Having come from that, and reading stories like this, is terrifying. I've never heard of anything even remotely like this here (Western Australia). It makes me even more frightened that before about the increasing American media influence and cultural influence* here - because it makes me even more inclined to believe the place, collectively, is insane, though obviously most of it's citizens are just fine.
Pass the cluebat.
* Yes, I know I'm responding to an American article on a website largely full of Americans. I wouldn't call either of these the mainstream American, though.
Sorry about the moron-gram you got. I don't have mod points to squash him appropriately right now.
Your comments are interesting. Do you know of any reliable resources where I can find out more about such things? I'm no audio nut by far, but I *am* interested in finding out what I can do to improve my audio gear without being hopelessly ripped off.
That makes me sound like I don't use Java because it's "not enough better."
I don't use Java because I can't use the libraries I need without (AFAIK) doing lots more work than I have to under C++, and because I really don't like Swing. At all. Anyway, I'm mostly working on existing projects where the language and toolkit choice isn't exactly mine to begin with.
Aaah, Qt iterators. *sobs*. Yes, Qt definitely has some problems.
When it comes to the macros, moc, and signals/slots though... I'm really not convinced it's a big deal. While it is conceptually ugly, it works around the miserable limitations of C++'s reflection/introspection and its ultra-basic RTTI scheme. And, in practice, it works well and without fuss. I value things that work well and stay out of my way to let me get the real work done.
If C++ had decent reflection and introspection (and a couple of other things) I'd agree with you that Qt should stick to being just a toolkit - not a general programming library and set of language extensions. Unfortunately, it doesn't, and Qt for me provides a way to use C++ while retaining my sanity. Just. If I didn't use Qt, I'd probably end up using similar ugly hacks with a meta-compiler, macros, or an IDL compiler to get the facilities I need, and *that* would be ugly.
As for memory management - yep, I agree it's mind numbingly stupid and frustrating. Why the C++ committe didn't mandate a basic garbage collection implementation as part of the std lib, and some decent introspection/reflection capabilities, is beyond me. They threw in the kitchen sink and most of the rest of the pluming system when they put in templates (you know you can write functional programs with templates that are executed by compiling the file, right?), so surely, say, a property system wouldn't have gone astray?
I've looked at GTKmm - that's what Inkscape are using. Looks interesting, but as you say it's poorly documented and the API isn't what I'd call attractive. If I hadn't joined an existing project built on Qt, I'd probably have tried it out.
What gets me about Java is that it's not enough better, for all the fuss it gets. What pisses me off the most is that it lacks a built-in property system so you STILL have to do the archaic old "lets write simple accessors for every attribute, just in case we later need to make them do more" crap. A good property system negates that need completely - you can simply create a property that looks like a value member, but runs the accessors when you get/set it. Still, at least it does provide some reflection and introspection facilities and some (marginally) better support for functor objects etc so you don't have to use hacks like Qt's moc.
There's also the matter of dependencies. I couldn't use Java without dependencies with any of the programs I currently work on. Not without re-implementing a number of the libraries I use. What's worse, I'd have to create and maintain Java bindings for them. LittleCMS is but one example. It's not Java's fault, but it still sucks:S
I did read Guido's comments on typing. I can see his concerns (especially when it comes to programmers from C/C++/Java backgrounds who're moving to Python), but I'm not sure it's a good reason to completely exclude the facility.
I know you can do a lot of this with decorators now - but I also know they're sufficiently convoluted to make any sane developer want to cry and wish to go back to their nice, simple C++ templates;-)
OK, small issue: Python is strongly typed too (rather more so than, say, the fairly weakly-typed language C). It's just not/statically/ typed.
Anyway... I personally share your desire for better static checking. I'd rather the compiler *didn't* do it though, and got on with the job of compiling. In my view static checking is the job of an additional tool (though it might use parts of the compiler to help it along, and the compiler might have a convenience "run a static check too" option) dedicated to the job.
I wish Python had more facilites for/optional/ static typing where appropriate, and better support for things like before/after contracts. Assertions just don't cut it.
That said... I'm not too sure about Eiffel. It sounds like it goes a bit far in the other direction.
Right now, my dream language would be Python with a *good* standard GUI toolkit (Tkinter doesn't count), selective static typing - perhaps by interface/protocol not just class - and the ability to transparently run Haskell procedures. That would be awesome.
Well... the very newest, top-end cards may be out of reach of all but those who consider "quad opteron" a serious option for their next gaming box. Don't laugh, there are certain to be a few out there.
Thing is, it doesn't matter. Doing so:
means they can still claim to have "the fastest card on the market" even if they can't afford to sell many (remember, at small yields it can cost MEGABUCKS to make these things);
makes the other cards in their range look more reasonably priced by comparison
makes the "mid range" look higher - again, probably helps push up what people will consider buying;
helps make sure they have reviews of the production part published, games listing support for them, etc by the time production volumes rise and prices on the card are dropped to saner levels; and
Probably doesn't cost them much anyway; they may as well try to get megabucks for a few cards while ramping up volumes.
OpenOffice.org was ported quickly to Mac OS X. Mac users went "eew, X11" and turned up their noses - and rightly so, the X11 integration on Mac OS X is shit. I should know. Scribus, an app I contribute to, has been working on Mac OS X for ages - it's a C++ app, yet all it took was a recompile and a few endianness fixes. That, however, doesn't cut it - you get nowhere unless it's been ported to all Apple's special platform specific APIs for doing everything, none of which can be neatly integrated into an X11 app - you have to move across to Aqua first. Qt made that fairly easy, but there's still all this f**ing integration work left.
<rant> Apple could've made this much easier by making their X11 integration NOT SUCK and giving X11 developers some easy to use libraries for simple access to the Mac menu bar, dock, VFS, etc. They evidently felt tying developers to Mac OS X's APIs was more important. </rant>
Anyway... OpenOffice's platform API abstraction didn't fit well with the "Mac way" of doing things, so they had to do some fairly major work to make it all work right. The best way ended up being the Java cocoa bindings. Note that the Java here is not portable, it relies on the Cocoa bindings for Java that only work on Mac OS X.
Blame Mac OS X. Java deserves only some of the credit here, and I suspect many of the issues would've remained if OO.o was written in Java.
Sure, Swing is slow as molasses, but the Java language isn't. Using SWT, or the Java cocoa bindings, it'd be much less crap. You lose your GUI portability, but then you'll have to customise your GUI heavily for each platform anyway.
I'll still stick with C++ and Qt, especially with Qt4 GPL for Windows.
It seems to be a choice between dirty swamp water (Java), somewhat cleaner but might be radioactive swamp water (C#), and swamp water with sewerage in it (C++).
I can't wait until PyQt4 comes out - GPL for Windows, Mac OS X, and UNIX/Linux. Python has its own faults, and PyQt more, but... it sucks much less than the others for many tasks.
The real issue isn't Java - it's javax.swing, the Java Swing toolkit, and the java.awt, the Abstract Window Toolkit . AWT looks awful, has a clumsy event model, and is low level and clunky. Swing provides higher level widgets, but an over-complex API, and it's still slow as hell.
Ever tried SWT from Eclipse? It's the Java widget toolkit that doesn't suck! (as much).
That said, I'm very happy with C++ and Qt. Well, except the C++ bit, but I find Java just as gag-worthy in different ways (Java 1.5 goes some way to rectifying the issues with generics and the collections framework, though. It's still WAY too verbose and static for a truly nice language, though).
Yep... and fair enough, too. As I said, the fellow can be rather abrasive - and ships crippleware from what you say (I haven't tried it).
I'm don't disagree about that. I just don't think that justifies the attitude of entitlement that some folks seem to have here, and I'm not convinced that alone is a reason to write off his other work.
That said, I'm not particularly inclined to rely on this for anything important in the long term, given his licensing track record.
The bigotry being displayed here is astonishing. Between whining about cdrecord, making uninformed snipes about how Linux is better, and writing off Solaris because of the CDDL, it's a pretty poor show. I know slashdot can do better than this:S
note: I have concerns about the CDDL too, but it ONLY MATTERS if you want to contribute your code into the core codebase, use Solaris code in your own, or redistribute modified Solaris code. The contributor agreement only matters if you want to have your code merged into Solaris - you can simply maintain an outside patch/dist if you have a problem with it. I'm 99% sure none of the loud complainers here will be doing any of the above anyway.
I also tried Solaris 10 - and got rid of it. It's not much of a desktop yet - old software, and it needs a comprehensive package collection of libs and GNU tools REALLY badly. It does, however, serve some people's needs fantastically, especially in the server space. Let's not write something off entirely because "sun are bad, mmkay" or because it doesn't have the latest GNOME.
As for cdrecord... come on. The fellow can be abrasive but I don't see how that's important here, and he can do what he wants with his code. He did license it under the GPL in the first place, which I for one appreciate, so we can use it and the extended DVD-supporting derivatives of it available in Linux distros. I don't see why him deciding *not* to give away *more* of his work draws such incredible indignation here. Sure, it'd be nice (FSF zealous would argue "morally required"), but really it's his work and his code.
It's his code. He's not obliged to release it, and if his goals in keeping it closed are being met I guess he probably won't.
That said, he did release cdrecord under the GPL, and it has now been extended by others to support device file access and DVD writing. Grab the SRPM of cdrecord / dvdrecord from Fedora Core 4, for example.
Couldn't agree more. Even if InDesign wasn't so much better, I'd jump ship just to be free of 30+ digit mixed-case unhyphenated serial numbers, network broadcasts to detect other Quark installs unless you use an expensive license server (so no disk image based installs for you!), arrogant sales and support, and a program that just hasn't improved much over the last three versions and three thousand dollars.
Even if they wanted to, I think you'll find that Quark can't. Their app is really messy and not very portable - it was apparently a screaming nightmare just to port from MacOS 9 to MacOS X.
My understanding is that they'd have to rewrite quite a bit of it to do a proper Linux port. I guess they could port the win32 version using WINELib, though.
At this point, it doesn't matter much. Until Adobe ports Photoshop, nobody will care. If Adobe ports Photoshop, chances are they'll port Acrobat and InDesign instead. Why should Quark then try to contest a small market with a juggernaut already crammed into it?
There are also other options. I happen to work on one of them, Scribus. It's not that great right now and needs a LOT of work to bring it up to the required levels of features and reliability. It's already extremely useful, however, and it has some ass-kicking features including PDF export reliability that's right up there with InDesign.
I find your comments very interesting. I'ts the first I've heard of really solid areas that Quark does things better, but I can see what you mean. I'll have to investigate those areas with the layout staff at work and see how they feel.
Quark's probably going to get replaced at work soon, because I just can't get it to do simple things right. The single biggest problem is it's handling of EPS and PDF. Save as EPS is buggy, and doesn't embed TrueType fonts even when told to. Placing PDFs on the page is a screaming nightmare - we now use Acrobat to convert them all to EPS instead. For my work, those are MAJOR problems.
I first reported the EPS problem to Quark not long after Quark 4.0 came out, and from what I hear it's still isn't fixed in Quark 6. Why haven't I checked? Because the last Quark demo I downloaded to test HAD SAVE AS EPS *DISABLED*. Yeah... way to let your customers test your software.
Frankly, even if InDesign has some serious issues, I'm inclned to jump ship just so I don't have to deal with Quark anymore. They've been arrogant and unpleasant to deal with, it's hard to buy their software in Australia without paying massive markups to exclusive distributors, and they just don't seem to care what their customers want. At least their prices have been forced down by Adobe, though.
Quark rarely does anything properly when it comes to platform support. Their first major release after OS/X came out just fixed the worst of the hacks so it ran OK under MacOS 9 emulation (Classic). They then charged another large upgrade fee to get one that ran natively... and I'd be shocked if that wasn't still chock full of MacToolbox calls and other elderly crud.
Does Quark still have a future or is the future Adobe and Macromedia?"
Franlkly, Quark lost when InDesign 2.0 came out. Since then, the upgrade path has widely been considred to be Quark 4 -> InDesign 2 -> InDesign CS . Quark 6 ? Yeah.... I heard they released that.
One of the biggest reasons for that is probably that Quark 4 -> Quark 6 and Quark 5 -> Quark 6 upgrades used to cost more than a new copy of InDesign. This, guys, is a really bad plan for keeping marketshare.
Quark's prices have plummeted, but even so all they really have going for them is that most designers are more familiar with Quark. Their technology is embarrassingly inferior in features, reliability, and pretty much everything else.
To top it off, Quark hasn't lost it's customer-hostile attitude to sales and support. Adobe will listen to you, and might even act on what you say. You don't get that from Quark. They pissed off a lot of customers while they had them locked in, and now those customers are jumping ship as fast as they can.
In short... if I was the Quark CEO, I'd be looking for other work too. Unless the company pulls their head out in a hurry, I'd expect them to lose more than just their CEO.
... they won't let you leave from the big house with bars.
You might be well advised to find a better government sooner rather than later should you do so at all. As a citizen of another country with an arguably "better government" (Australia) I'd like to point out that (a) we're trying as hard as we can to be as stupid as America, and (b) Please, please, please put your vote to stopping this stupidity at it's source instead. If all the sane, smart Americans leave we're all f**ed.
Actually, they do. You see "Linux" on some OS detection tools etc because they use the Akamai network - a third party who uses Linux - for content distribution. The underlying servers are still Windows.
To me that's just a sign they're willing to be pragmatic and use what works. It shows them in a good light when compared to the foaming zealots, actually...
Slashdot Mirror
Tell me that a week after the Mac-on-Linux folks get their hands on a prerelease. If it's not made to work with Mac OS X for Intel fairly rapidly, I may die of shock. My bet is on support WELL before the first official Intel-based macs hit the shelves.
I don't expect to be able to run Mac OS X natively on non-Apple hardware. I don't think I'd want to, given the likely driver quality etc. However, running it under MoL sounds *very* attractive, and a Xen port I would positively kill for.
Note that my perspective is mostly that of a developer. More than anything, I just want to be able to conveniently test on Mac OS X. It is *such* a pain ssh-ing into the slow eMac at work and using VNC to test things - especially since the braindead thing only WOLs from sleep, not poweroff. Being able to use my nice fast Athlon for test builds and actually test each configure.in.in change on Mac OS X before checking it in would be bliss.
Sadly, I doubt Apple will ever do anything as useful as release a Mac OS X developer edition that runs on Xen, so I'll most likely be stuck with MoL (hopefully).
As the other poster noted, "the best" is rather ambiguous. I wouldn't touch Portage with a barge pole, not only because of efficiency (my time is worth more than a 1% runtime speed increase) but also because as a developer, I don't trust it. I've seen too many *weird* problems from Gentoo users, and most of them go away when they recompile some other random part of their system. Scary.
... IMO, boring. They're all pretty good. I couldn't fault FC4's for example.
In a world of stable APIs and ABIs, Portage might be awesome. In this world of unstable, often subtly changing app interfaces, limited insight into dependencies, and weird interactions, I don't think it's all that good an idea. I'll take RPM & yum/YaST/urpmi or dpkg and apt any day.
Knoppix's detection is pretty darn good, but then again everything "just worked" on my FC4 system too. I think this is coming along nicely on most systems.
I think BSD rc scripts are even worse than System V ones. I remain bewildered as to why, in this day and age, we can't have something less horrifyingly archaic and painful that provides the ability to run rc scripts for backward compatibility. Preferably something not quite as complex and full of XML as SMF (from Solaris 10).
As for installers
Anyway - it *is* possible to combine all these. You'll end up with a distro that has other major shortcomings though, such as a user base of four, irregular security updates, and very little reason to trust that it'll stay with the times. Perhaps more importantly, you won't have much package QA or testing across all that wide a range of hardware, so your distro's quality won't be up to scratch.
Also, you'll find that upstream developers just won't be interested. "Your app doesn't compile out of the box on 2bit linux, 1998 unmaintained edition. Help!" . Riight. Note that for this purpose I *am* an "upstream developer".
Personally, I don't think the answer is to seek perfection in distros. I think the answer is to seek more consistency - not necessarily by forcing everything to be the same, but by providing a level of consistency and certain things that apps can rely on to be the same. The LSB is theoretically a good start on that, but the work done by FreeDesktop.org is in my view much more so by view of seeing rapid and comprehensive adoption.
If you want to help Linux distros improve, help FreeDesktop.org .
Those stories from those kids are terrifying.
... the point is, I never, *EVER* saw any trouble or hostility from the "mainstream" once I reached high school. The worst I got was a bit of teasing every now and then, and even that was pretty tame and only during the first two years. We have a term here - "Year 9s are animals." It's kinda expected. After that, any distance from them was mostly because I just didn't "get" their interests - which at the time mostly seemed to involve getting so drunk they threw up. Yay. Not.
I was a geek in high school. I'm sure many here were. Y'know what? Any isolation I felt - and yes, there was quite a bit in the first few years - was largely my own fault. I had people try really hard to be nice, to include me, etc and I just didn't know how to respond. I mostly got confused, or thought they were trying to tease / mock me. In my defense, that's because that used to be a problem, but I think mostly because of my lack of social skills rather than any geekyness. I was one of those people who wakes up one day and says "huh. People. Who would'a thought."
One I learned some utterly minimal social skills, things improved a lot. This frequently meant walking home talking about coding / maths with a couple of friends, though that's not all I did. Anyway
Having come from that, and reading stories like this, is terrifying. I've never heard of anything even remotely like this here (Western Australia). It makes me even more frightened that before about the increasing American media influence and cultural influence* here - because it makes me even more inclined to believe the place, collectively, is insane, though obviously most of it's citizens are just fine.
Pass the cluebat.
* Yes, I know I'm responding to an American article on a website largely full of Americans. I wouldn't call either of these the mainstream American, though.
Sorry about the moron-gram you got. I don't have mod points to squash him appropriately right now.
Your comments are interesting. Do you know of any reliable resources where I can find out more about such things? I'm no audio nut by far, but I *am* interested in finding out what I can do to improve my audio gear without being hopelessly ripped off.
That makes me sound like I don't use Java because it's "not enough better."
I don't use Java because I can't use the libraries I need without (AFAIK) doing lots more work than I have to under C++, and because I really don't like Swing. At all. Anyway, I'm mostly working on existing projects where the language and toolkit choice isn't exactly mine to begin with.
Aaah, Qt iterators. *sobs*. Yes, Qt definitely has some problems.
... I'm really not convinced it's a big deal. While it is conceptually ugly, it works around the miserable limitations of C++'s reflection/introspection and its ultra-basic RTTI scheme. And, in practice, it works well and without fuss. I value things that work well and stay out of my way to let me get the real work done.
:S
When it comes to the macros, moc, and signals/slots though
If C++ had decent reflection and introspection (and a couple of other things) I'd agree with you that Qt should stick to being just a toolkit - not a general programming library and set of language extensions. Unfortunately, it doesn't, and Qt for me provides a way to use C++ while retaining my sanity. Just. If I didn't use Qt, I'd probably end up using similar ugly hacks with a meta-compiler, macros, or an IDL compiler to get the facilities I need, and *that* would be ugly.
As for memory management - yep, I agree it's mind numbingly stupid and frustrating. Why the C++ committe didn't mandate a basic garbage collection implementation as part of the std lib, and some decent introspection/reflection capabilities, is beyond me. They threw in the kitchen sink and most of the rest of the pluming system when they put in templates (you know you can write functional programs with templates that are executed by compiling the file, right?), so surely, say, a property system wouldn't have gone astray?
I've looked at GTKmm - that's what Inkscape are using. Looks interesting, but as you say it's poorly documented and the API isn't what I'd call attractive. If I hadn't joined an existing project built on Qt, I'd probably have tried it out.
What gets me about Java is that it's not enough better, for all the fuss it gets. What pisses me off the most is that it lacks a built-in property system so you STILL have to do the archaic old "lets write simple accessors for every attribute, just in case we later need to make them do more" crap. A good property system negates that need completely - you can simply create a property that looks like a value member, but runs the accessors when you get/set it. Still, at least it does provide some reflection and introspection facilities and some (marginally) better support for functor objects etc so you don't have to use hacks like Qt's moc.
There's also the matter of dependencies. I couldn't use Java without dependencies with any of the programs I currently work on. Not without re-implementing a number of the libraries I use. What's worse, I'd have to create and maintain Java bindings for them. LittleCMS is but one example. It's not Java's fault, but it still sucks
Well, I did say "dream language" ;-)
;-)
I did read Guido's comments on typing. I can see his concerns (especially when it comes to programmers from C/C++/Java backgrounds who're moving to Python), but I'm not sure it's a good reason to completely exclude the facility.
I know you can do a lot of this with decorators now - but I also know they're sufficiently convoluted to make any sane developer want to cry and wish to go back to their nice, simple C++ templates
OK, small issue: Python is strongly typed too (rather more so than, say, the fairly weakly-typed language C). It's just not /statically/ typed.
... I personally share your desire for better static checking. I'd rather the compiler *didn't* do it though, and got on with the job of compiling. In my view static checking is the job of an additional tool (though it might use parts of the compiler to help it along, and the compiler might have a convenience "run a static check too" option) dedicated to the job.
/optional/ static typing where appropriate, and better support for things like before/after contracts. Assertions just don't cut it.
... I'm not too sure about Eiffel. It sounds like it goes a bit far in the other direction.
Anyway
I wish Python had more facilites for
That said
Right now, my dream language would be Python with a *good* standard GUI toolkit (Tkinter doesn't count), selective static typing - perhaps by interface/protocol not just class - and the ability to transparently run Haskell procedures. That would be awesome.
Thing is, it doesn't matter. Doing so:
OpenOffice.org was ported quickly to Mac OS X. Mac users went "eew, X11" and turned up their noses - and rightly so, the X11 integration on Mac OS X is shit. I should know. Scribus, an app I contribute to, has been working on Mac OS X for ages - it's a C++ app, yet all it took was a recompile and a few endianness fixes. That, however, doesn't cut it - you get nowhere unless it's been ported to all Apple's special platform specific APIs for doing everything, none of which can be neatly integrated into an X11 app - you have to move across to Aqua first. Qt made that fairly easy, but there's still all this f**ing integration work left.
<rant>
Apple could've made this much easier by making their X11 integration NOT SUCK and giving X11 developers some easy to use libraries for simple access to the Mac menu bar, dock, VFS, etc. They evidently felt tying developers to Mac OS X's APIs was more important.
</rant>
Anyway... OpenOffice's platform API abstraction didn't fit well with the "Mac way" of doing things, so they had to do some fairly major work to make it all work right. The best way ended up being the Java cocoa bindings. Note that the Java here is not portable, it relies on the Cocoa bindings for Java that only work on Mac OS X.
Blame Mac OS X. Java deserves only some of the credit here, and I suspect many of the issues would've remained if OO.o was written in Java.
Sure, Swing is slow as molasses, but the Java language isn't. Using SWT, or the Java cocoa bindings, it'd be much less crap. You lose your GUI portability, but then you'll have to customise your GUI heavily for each platform anyway.
I'll still stick with C++ and Qt, especially with Qt4 GPL for Windows.
It seems to be a choice between dirty swamp water (Java), somewhat cleaner but might be radioactive swamp water (C#), and swamp water with sewerage in it (C++).
... it sucks much less than the others for many tasks.
I can't wait until PyQt4 comes out - GPL for Windows, Mac OS X, and UNIX/Linux. Python has its own faults, and PyQt more, but
The real issue isn't Java - it's javax.swing, the Java Swing toolkit, and the java.awt, the Abstract Window Toolkit . AWT looks awful, has a clumsy event model, and is low level and clunky. Swing provides higher level widgets, but an over-complex API, and it's still slow as hell.
Ever tried SWT from Eclipse? It's the Java widget toolkit that doesn't suck! (as much).
That said, I'm very happy with C++ and Qt. Well, except the C++ bit, but I find Java just as gag-worthy in different ways (Java 1.5 goes some way to rectifying the issues with generics and the collections framework, though. It's still WAY too verbose and static for a truly nice language, though).
Yep... and fair enough, too. As I said, the fellow can be rather abrasive - and ships crippleware from what you say (I haven't tried it).
I'm don't disagree about that. I just don't think that justifies the attitude of entitlement that some folks seem to have here, and I'm not convinced that alone is a reason to write off his other work.
That said, I'm not particularly inclined to rely on this for anything important in the long term, given his licensing track record.
The bigotry being displayed here is astonishing. Between whining about cdrecord, making uninformed snipes about how Linux is better, and writing off Solaris because of the CDDL, it's a pretty poor show. I know slashdot can do better than this :S
... come on. The fellow can be abrasive but I don't see how that's important here, and he can do what he wants with his code. He did license it under the GPL in the first place, which I for one appreciate, so we can use it and the extended DVD-supporting derivatives of it available in Linux distros. I don't see why him deciding *not* to give away *more* of his work draws such incredible indignation here. Sure, it'd be nice (FSF zealous would argue "morally required"), but really it's his work and his code.
note: I have concerns about the CDDL too, but it ONLY MATTERS if you want to contribute your code into the core codebase, use Solaris code in your own, or redistribute modified Solaris code. The contributor agreement only matters if you want to have your code merged into Solaris - you can simply maintain an outside patch/dist if you have a problem with it. I'm 99% sure none of the loud complainers here will be doing any of the above anyway.
I also tried Solaris 10 - and got rid of it. It's not much of a desktop yet - old software, and it needs a comprehensive package collection of libs and GNU tools REALLY badly. It does, however, serve some people's needs fantastically, especially in the server space. Let's not write something off entirely because "sun are bad, mmkay" or because it doesn't have the latest GNOME.
As for cdrecord
It's his code. He's not obliged to release it, and if his goals in keeping it closed are being met I guess he probably won't.
That said, he did release cdrecord under the GPL, and it has now been extended by others to support device file access and DVD writing. Grab the SRPM of cdrecord / dvdrecord from Fedora Core 4, for example.
It "just works"
Couldn't agree more. Even if InDesign wasn't so much better, I'd jump ship just to be free of 30+ digit mixed-case unhyphenated serial numbers, network broadcasts to detect other Quark installs unless you use an expensive license server (so no disk image based installs for you!), arrogant sales and support, and a program that just hasn't improved much over the last three versions and three thousand dollars.
Even if they wanted to, I think you'll find that Quark can't. Their app is really messy and not very portable - it was apparently a screaming nightmare just to port from MacOS 9 to MacOS X.
My understanding is that they'd have to rewrite quite a bit of it to do a proper Linux port. I guess they could port the win32 version using WINELib, though.
At this point, it doesn't matter much. Until Adobe ports Photoshop, nobody will care. If Adobe ports Photoshop, chances are they'll port Acrobat and InDesign instead. Why should Quark then try to contest a small market with a juggernaut already crammed into it?
There are also other options. I happen to work on one of them, Scribus. It's not that great right now and needs a LOT of work to bring it up to the required levels of features and reliability. It's already extremely useful, however, and it has some ass-kicking features including PDF export reliability that's right up there with InDesign.
I find your comments very interesting. I'ts the first I've heard of really solid areas that Quark does things better, but I can see what you mean. I'll have to investigate those areas with the layout staff at work and see how they feel.
Quark's probably going to get replaced at work soon, because I just can't get it to do simple things right. The single biggest problem is it's handling of EPS and PDF. Save as EPS is buggy, and doesn't embed TrueType fonts even when told to. Placing PDFs on the page is a screaming nightmare - we now use Acrobat to convert them all to EPS instead. For my work, those are MAJOR problems.
I first reported the EPS problem to Quark not long after Quark 4.0 came out, and from what I hear it's still isn't fixed in Quark 6. Why haven't I checked? Because the last Quark demo I downloaded to test HAD SAVE AS EPS *DISABLED*. Yeah... way to let your customers test your software.
Frankly, even if InDesign has some serious issues, I'm inclned to jump ship just so I don't have to deal with Quark anymore. They've been arrogant and unpleasant to deal with, it's hard to buy their software in Australia without paying massive markups to exclusive distributors, and they just don't seem to care what their customers want. At least their prices have been forced down by Adobe, though.
Quark rarely does anything properly when it comes to platform support. Their first major release after OS/X came out just fixed the worst of the hacks so it ran OK under MacOS 9 emulation (Classic). They then charged another large upgrade fee to get one that ran natively... and I'd be shocked if that wasn't still chock full of MacToolbox calls and other elderly crud.
Franlkly, Quark lost when InDesign 2.0 came out. Since then, the upgrade path has widely been considred to be Quark 4 -> InDesign 2 -> InDesign CS . Quark 6 ? Yeah .... I heard they released that.
One of the biggest reasons for that is probably that Quark 4 -> Quark 6 and Quark 5 -> Quark 6 upgrades used to cost more than a new copy of InDesign. This, guys, is a really bad plan for keeping marketshare.
Quark's prices have plummeted, but even so all they really have going for them is that most designers are more familiar with Quark. Their technology is embarrassingly inferior in features, reliability, and pretty much everything else.
To top it off, Quark hasn't lost it's customer-hostile attitude to sales and support. Adobe will listen to you, and might even act on what you say. You don't get that from Quark. They pissed off a lot of customers while they had them locked in, and now those customers are jumping ship as fast as they can.
In short ... if I was the Quark CEO, I'd be looking for other work too. Unless the company pulls their head out in a hurry, I'd expect them to lose more than just their CEO.
... they won't let you leave from the big house with bars.
You might be well advised to find a better government sooner rather than later should you do so at all. As a citizen of another country with an arguably "better government" (Australia) I'd like to point out that (a) we're trying as hard as we can to be as stupid as America, and (b) Please, please, please put your vote to stopping this stupidity at it's source instead. If all the sane, smart Americans leave we're all f**ed.
Yay! Another judgement for a totally retarded patent awarded!
Er... no... wait...
they want some of the things Intel /does/ have going for them. Say, Pentium M (fast, low power, and getting cheaper in a hurry) or StrongARM.
Actually, they do. You see "Linux" on some OS detection tools etc because they use the Akamai network - a third party who uses Linux - for content distribution. The underlying servers are still Windows.
To me that's just a sign they're willing to be pragmatic and use what works. It shows them in a good light when compared to the foaming zealots, actually...