...but reported as a bug, because it did not match the ISO standard. I wonder who the hell did this, and the maintainer just insisted on not changing it back to just Taiwan. I am not making any political statement here. I think an open source software would be more political neutural.
That's contradictory. What you're saying is: Debian uses a name from the ISO standard, but that's offensive. You want it to be something different from the standard, and you think that's not taking a political stand?
Think about it, suppose the name of 'Canada' was changed to 'Canada, 51st state of the USA', even though that's non-standard. Isn't changing it away from the ISO standard name like that a political statement? Then why isn't changing from the ISO standard 'Taiwan, province of China' a political statement?
Whether or not the standard names are any good is a whole different question, of course. But the one thing that Debian is being here is politically neutral. I can understand why Taiwanese people may not like it, but the reality of world politics is that China has a lot more power, which is why few countries have separate diplomatic relations with Taiwan (China doesn't like them doing that, since you only have diplomatic relations with other countries).
So yes Taiwan's place in the world is sucky. However, it's nothing to do with Debian, which is just using the ISO standard name for countries. That standard reflects the reality of the world today, which is that China is more powerful. All you have to do is change the standard.
Jesus people, just redirect the fucking output to a temporary file or to less(1) if something is going to flood the terminal.
The problem is not that it floods the terminal; that's what you expect to happen from a build. The problem is that gnome-terminal is slow and takes a lot of CPU to draw the output. You can say 'just save it to a file then' but that's just a workaround for the real problem of gnome-terminal being slow compared to xterm/Eterm/etc.
If Epiphany started up too slowly, you wouldn't say 'just browse in w3m then', it's a weakness of the program, not of the way people are using it.
Optimizing code which doesn't need optimization is Bad with a capital 'B'. When optimizing code, there is almost always a tradeoff between efficiency and maintainability.
There is truth in what you say, but I don't think this is a counterpoint to what he said. What he is lamenting is the lost art of being aware and caring about performance at the micro-level, as opposed to the macro-level.
In a way, he acknowledges that knowing assembly itself isn't required. Simply knowing the way the machine works (and having a basic understanding of assembly) provides most of the benefits. Good programmers do not pass 2K structs by value to functions in C, they pass pointers. Good programmers move independent code out of loops. And that is better code in every way - more efficient, because it saves executing it repeatedly, and because if it's independent, then looping it doesn't actually make semantic sense. Later readers of the code might wonder if they're just misreading it, and maybe it isn't independent code... now they have to check for obscure relationships.
The best point to take away from the article is that efficiency choices are not just the "big" ones like which algorithm to use, they are everywhere. Every programming decision has efficiency concerns, from the choice of algorithm down to the program design (how much data must be passed around?) and the size of data structures (big ones use more memory, which could affect cache locality).
That doesn't mean that for every choice, you must pick the one that is the most efficient. It simply means that you keep effiency in mind, and you take it into account when choosing. That way when you pick a simpler (but slower) algorithm, you're at least knowingly making the tradeoff... same if you pick the faster algorithm.
Here in Australia, Telstra (50% government owned) is a retailer of DSL, as well as being supplying DSL wholesale to other ISPs who can resell it.
Recently, Telstra dropped the price to around AU$30/month, which compares to the $20/month dialup costs. Considering that you save money on dialup phone calls, that's relatively close to parity price, for something much faster and lower latency.
What did it take for this price cut? The ISP resellers provide much better plans and better service than Telstra, so it dropped the price savagely (probably around 40-50%) and has seen DSL takeup skyrocket as a result, leaving the other ISPs in the lurch as they were paying more to Telstra for wholesale DSL than it was selling to customers at retail! This was eventually somewhat rectified.
However, the point is, all it takes is one company trying to price cut to increase volumes. Other companies will want to match it, since they are willing to drop margins to maintain numbers and stop customer churn doubling or tripling.
The point isn't that it groups them together, it's that it displays them together. Like, I click on the one subject line and it displays the whole "conversation" (yet, like USENET readers), with options to collapse.
In mutt, Esc-v collapses/uncollapses the current thread, and Esc-V does it for ALL threads.
The keybindings are, of course, changeable, so don't complain that they're cryptic, just set them to something simpler if you'd use them a lot.
I like that there is only one JVM that cannot run my Java.
Well, I certainly don't do much Java code, so I'm not trying to pretend to be an expert here.. but have you tried your code under Kaffe or Classpath? The last time I looked at them (a couple of years ago) they were quite a way away from finishing off Java 1.1, let alone 1.4. And I understood that people still end up testing under multiple JVMs anyway.
Sure, GPLing Java has the risk of leading to a number of forks. But I think in practice it will basically invigorate Java. Netscape opened up Mozilla and there have been no forks of it - Netscape and now mozilla.org are seen as the natural caretakers, with a strong moral authority.
Making it GPL kills most of the motivation for people forking (commercial gain from an improved proprietary version). Individuals can still do so, but don't tend to have the weight of companies, so they tend to want to submit their work back to the main (Sun) Java, or they just end up being ignored.
Comparing Java to things like Perl, Python and Ruby - all of these are languages with reasonably large userbases, and none of them has ever had a serious chance of forking. Sun has the bonus of even having a trademark they can use as a combined carrot/stick for the forks.
Companies out there are still going to choose to use Sun Java by the hundreds, with the bonus that more people can work on fixing bugs, or keeping platform ports more up to date. It seems clear to me that ISVs will still almost universally use a version of Java that has the certification logo.
The vast majority of people who use Linux do so only because it is free, not because of any technical/useability advantage.
Millions of pirated Windows copies prove you wrong. If people can pirate a free Windows or they can install a free Linux.. and they choose Linux, then obviously price isn't the distinguishing factor.
Most of the application developers I know who used to develop for Linux have agreed with me that OS X is light years ahead of Linux in every aspect: performance, design, security, everything. Apple putting their framework on Linux would be akin to putting a Ferrari body around a Lada engine.
I don't really understand what you mean at all. Most of the stuff you're praising is just framework-level stuff anyway, not OS level.
In case this is disguised bitching about X and the toolkits (GTK, Qt etc.).. if they wanted, Apple could ditch X and use their own display code, no problems.
The issue is how to make Java free to redistribute. Under the current license, no JVM can be included in a totally free distribution. Is there some license that can protect the trademark and portability that allows free distribution? Can such a license be written that is acceptable to Sun?
First of all, the trademark is a separate thing to the source. You can leave the trademark under a separate licensing deal, so even if people fork the code (that's OK), they can't use the trademark unless Sun agrees. That lets Sun insist on you passing the Java test suite before letting you use the logo.
Now all the IT executives have to check for is the Java logo - which they should already be doing, since that's what it's there for, to let them choose a JVM with confidence.
Making Java GPL or something very similar would ensure that no-one could fork it and hide the results in a way that prevented merging (either by them, or by third parties such as Sun themselves). At the same time, even if somebody came up with some funky MiniJava designed for phones, and all the phone vendors flocked to it, Sun could take their changes and merge them into their Java tree.
Even if Sun opens Java in a GPL-way, they will still be the custodian of Java, with enormous moral authority on the code. It's amusing that everyone fears forking, when in reality it lets everyone - including the Linux distros - unite behind the one codebase.
If Redhat, SuSE, Debian all shipped Sun's JRE - so it was there, on the CD, ready to go, that would do nice things for Java usage. Of course there will be forks, but that doesn't matter because it lets people experiment safely, and their changes can be merged. People don't want 1000 versions of Java, they want one strong one that everyone can distribute and thus unite behind.
There will always be pressure to merge forks into the main Java tree.
I wish Debian developers would pay more attention to these developments (especially to 'no-frills' and 'bleeding-edge' part.)
They do. And that's why they have the unstable distro, just for you. In fact, even cooler, you don't have to wait for a release to run all this cool new stuff, you can upgrade and run it now!
And in fact, I understand you can do the same with Fedora too, just upgrade as you go. If you have the bandwidth to download Fedora, you certainly have the bandwidth to run Debian unstable.
The first is Jeff Waugh's proposal that Epiphany replace Firefox as a the de-facto Linux browser.
I think the minutes simply say he suggested the possibility 'Epiphany become the official Linux port of Firefox'. That's not the de-facto Linux browser by any means, except in any sense that someone could say that the default GNOME web browser is the de-facto Linux browser (an arguable issue, given the popularity of KDE on many distros).
That said, I don't think the suggestion makes any more sense. Firefox having basically the same UI on multiple platforms is an important strength that would be totally compromised by letting Epiphany take on its Linux role.
This is heading towards proprietary extensions territory, a la Netscape/IE.
Not really. Would you call GNOME itself a proprietary extension, given that it doesn't come from a standard?
Parts of XUL (like XBL) have been proposed for standardisation, but the best argument for standardisation is having a working implementation. It helps when people can try out the idea and see how it works in practice.
In that sense, it wouldn't be bad for GNOME/Mozilla to do some integration work and let people build GNOME apps in XUL. Then they can propose it as a standard, hopefully before MS does with their XAML stuff. We really don't want any more web momentum tilting Microsoft's ways - it's just too bad to have a single company doing it.
Yes, if Linux is going to be the OS for newbies. Yes, if Linux is going to be the OS for the desktop. The users won't care *why* it doesn't work, just *that* it doesn't work.
Because, after all, Microsoft writes all the Windows drivers for hardware out there, right?;) Seriously though, you cannot have universal hardware support without the manufacturers getting off their butts and supporting the OS for you. No-one, MS included, has the resources to write decent drivers for all those gazillions of sound cards, printers, cameras etc.
Once again, I think people miss the point. Linux's sound card support is actually very good, given that very few of those drivers are being written by the vendors. When you think of it that way - the commmunity has supported YOUR hardware - it's pretty good.
I wonder if they are running a Debian based or Debian itself, and Debian has another hole in it.
Funny. Too bad that was just a regular kernel hole, not one special to Debian's kernel. Any other distros can simply count themselves lucky the attackers didn't choose them.
...and when the car breaks down or they want to upgrade it, they go to a mechanic. The few that try to do it themselves either know what they're doing or break something pretty quickly.
Exactly. Most of them aren't changing the oil or installing new radios etc. They just use it, they don't try to change it. Really both Linux and Windows are much better in that regard - "oh you want to install a better radio? that means you need better cables.. I'll put that in for you!"
When cars do that, I'll consider them as good as computers in this respect.
Seems to work just fine for Knoppix; no, they're not as multi-platform as the base Debian dist is, but there's no reason the hardware detection scripts couldn't be ported. If you have a problem with your hardware detection, hack the script.
Different architectures have different methods of booting; some can only use one bootloader, some can use more than one. The point is, it's not just a matter of "hardware detection" so you can find the right drivers, the way you fundamentally boot and display things does not work the same over different architectures.
That doesn't mean what you're suggesting is impossible, but I don't think it's as easy as you were implying.
the people who care the least about a new version of Gentoo are the Gentoo users. It's a beautiful model... they way Debian was supposed to work.
You mean like the way the people who care the least about a new version of Debian are existig Debian users, because most of them just dist-upgrade anyway? In fact because many of them run unstable and keep following it, ignoring the fact a frozen release happened at all?:)
GTK is the GIMP Tool Kit, you know, that graphics program? It was originally written by the authors of GIMP to make their own program easier to write. I'm afraid the politics had nothing to do with it. GTK was originally written for GNOME no more than Qt was originally written for KDE.
Now why don't we try to mimic that interface into Linux instead of everything Windows has?
We are. See Keith Packard's work on the freedesktop.org X server. Otherwise.. Apple sent cease and desist letters to all the places you could get MacOS X themes. MacOS X is attractive but I could see myself getting bored with that look.
A flawless implementation of a crap algorithm is still crap.
No.. a flawless implementation of a crap algorithm just doesn't scale well. Of course bug rate is not the only criteria used when evaluating software, but people spend hundreds of man-hours fixing bugs.
It demonstrates that the quality of open source code is not automatically worse than professional proprietary code (which some people believe is the case). The important thing is that it's at least an attempt at formal study (and not simply personal collating of anecdotal reports).
When Microsoft release updates, you get them straight away. When Linus, etc. release updates, Debian might get around to putting it into the stable branch in 3-4 years.
That's ignorance or trolling on your part. Take a look at the frontpage of the Debian website - see the security alerts there? Each one of them is accompanied by backported fixes to stable. That's the whole point of the Debian security team, to make sure stable gets all critical security fixes.
One of the stupidist things linux/unix evolved into is a complete and utter lack of updating/rearchitecting the programming api's.
Actually things like the c10k connection problem (see http://www.kegel.com/c10k.html) show the various new APIs that people are using for specific problems.
For problems like buffer overflow, see snprintf():widely implemented on various Unixes, now part of C99 and SUSv2 standards. Also asprintf() which does away with the problem in a different way, by being a memory allocating printf(), whose return value you must free.
The reason is that most of the fundamental programming APIs are working well, and various libraries compete to provide other interfaces (eg. db3 vs sqlite for binary database storage, GTK vs Qt for widget library, etc etc).
Microsoft Terminal Services require orders of magnitude less bandwidth because it works at the toolkit level (as in, "draw a button and a drop-down menu"
I don't think this is what it does. I've worked a little on RDP (Remote Desktop Protocol), used by MS Terminal Servers, and from what I saw, it does indeed send pixmaps (and text strings, which the client renders). What the protocol does have is 2-level caching - there's a memory cache of bitmaps (so the server can say "draw cached bitmap 14 at this position"), and a persistent disk cache (so that next time you connect, the server might not have to send those bitmaps the first time it draws them).
If you're interested, check our the source code to rdesktop.
I suspect he thinks that they will not be a realistic alternative, and I also suspect he's right.
There is an enormous installed base of X applications out there, apps which would have to be rewritten. Sure, they could run over an X compatibility layer, but that tends to nullify most of the rewrite advantages in the first place.
If things like Xcb provide lower latency, and things like XDamage provide flexible management of windows/buffers for advanced effects, then the time invested in a rewrite probably aren't worth it.
Slashdot Mirror
...but reported as a bug, because it did not match the ISO standard. I wonder who the hell did this, and the maintainer just insisted on not changing it back to just Taiwan. I am not making any political statement here. I think an open source software would be more political neutural.
That's contradictory. What you're saying is: Debian uses a name from the ISO standard, but that's offensive. You want it to be something different from the standard, and you think that's not taking a political stand?
Think about it, suppose the name of 'Canada' was changed to 'Canada, 51st state of the USA', even though that's non-standard. Isn't changing it away from the ISO standard name like that a political statement? Then why isn't changing from the ISO standard 'Taiwan, province of China' a political statement?
Whether or not the standard names are any good is a whole different question, of course. But the one thing that Debian is being here is politically neutral. I can understand why Taiwanese people may not like it, but the reality of world politics is that China has a lot more power, which is why few countries have separate diplomatic relations with Taiwan (China doesn't like them doing that, since you only have diplomatic relations with other countries).
So yes Taiwan's place in the world is sucky. However, it's nothing to do with Debian, which is just using the ISO standard name for countries. That standard reflects the reality of the world today, which is that China is more powerful. All you have to do is change the standard.
Jesus people, just redirect the fucking output to a temporary file or to less(1) if something is going to flood the terminal.
The problem is not that it floods the terminal; that's what you expect to happen from a build. The problem is that gnome-terminal is slow and takes a lot of CPU to draw the output. You can say 'just save it to a file then' but that's just a workaround for the real problem of gnome-terminal being slow compared to xterm/Eterm/etc.
If Epiphany started up too slowly, you wouldn't say 'just browse in w3m then', it's a weakness of the program, not of the way people are using it.
Optimizing code which doesn't need optimization is Bad with a capital 'B'. When optimizing code, there is almost always a tradeoff between efficiency and maintainability.
There is truth in what you say, but I don't think this is a counterpoint to what he said. What he is lamenting is the lost art of being aware and caring about performance at the micro-level, as opposed to the macro-level.
In a way, he acknowledges that knowing assembly itself isn't required. Simply knowing the way the machine works (and having a basic understanding of assembly) provides most of the benefits. Good programmers do not pass 2K structs by value to functions in C, they pass pointers. Good programmers move independent code out of loops. And that is better code in every way - more efficient, because it saves executing it repeatedly, and because if it's independent, then looping it doesn't actually make semantic sense. Later readers of the code might wonder if they're just misreading it, and maybe it isn't independent code... now they have to check for obscure relationships.
The best point to take away from the article is that efficiency choices are not just the "big" ones like which algorithm to use, they are everywhere. Every programming decision has efficiency concerns, from the choice of algorithm down to the program design (how much data must be passed around?) and the size of data structures (big ones use more memory, which could affect cache locality).
That doesn't mean that for every choice, you must pick the one that is the most efficient. It simply means that you keep effiency in mind, and you take it into account when choosing. That way when you pick a simpler (but slower) algorithm, you're at least knowingly making the tradeoff... same if you pick the faster algorithm.
Here in Australia, Telstra (50% government owned) is a retailer of DSL, as well as being supplying DSL wholesale to other ISPs who can resell it.
Recently, Telstra dropped the price to around AU$30/month, which compares to the $20/month dialup costs. Considering that you save money on dialup phone calls, that's relatively close to parity price, for something much faster and lower latency.
What did it take for this price cut? The ISP resellers provide much better plans and better service than Telstra, so it dropped the price savagely (probably around 40-50%) and has seen DSL takeup skyrocket as a result, leaving the other ISPs in the lurch as they were paying more to Telstra for wholesale DSL than it was selling to customers at retail! This was eventually somewhat rectified.
However, the point is, all it takes is one company trying to price cut to increase volumes. Other companies will want to match it, since they are willing to drop margins to maintain numbers and stop customer churn doubling or tripling.
The point isn't that it groups them together, it's that it displays them together. Like, I click on the one subject line and it displays the whole "conversation" (yet, like USENET readers), with options to collapse.
In mutt, Esc-v collapses/uncollapses the current thread, and Esc-V does it for ALL threads.
The keybindings are, of course, changeable, so don't complain that they're cryptic, just set them to something simpler if you'd use them a lot.
I like that there is only one JVM that cannot run my Java.
Well, I certainly don't do much Java code, so I'm not trying to pretend to be an expert here.. but have you tried your code under Kaffe or Classpath? The last time I looked at them (a couple of years ago) they were quite a way away from finishing off Java 1.1, let alone 1.4. And I understood that people still end up testing under multiple JVMs anyway.
Sure, GPLing Java has the risk of leading to a number of forks. But I think in practice it will basically invigorate Java. Netscape opened up Mozilla and there have been no forks of it - Netscape and now mozilla.org are seen as the natural caretakers, with a strong moral authority.
Making it GPL kills most of the motivation for people forking (commercial gain from an improved proprietary version). Individuals can still do so, but don't tend to have the weight of companies, so they tend to want to submit their work back to the main (Sun) Java, or they just end up being ignored.
Comparing Java to things like Perl, Python and Ruby - all of these are languages with reasonably large userbases, and none of them has ever had a serious chance of forking. Sun has the bonus of even having a trademark they can use as a combined carrot/stick for the forks.
Companies out there are still going to choose to use Sun Java by the hundreds, with the bonus that more people can work on fixing bugs, or keeping platform ports more up to date. It seems clear to me that ISVs will still almost universally use a version of Java that has the certification logo.
The vast majority of people who use Linux do so only because it is free, not because of any technical/useability advantage.
Millions of pirated Windows copies prove you wrong. If people can pirate a free Windows or they can install a free Linux.. and they choose Linux, then obviously price isn't the distinguishing factor.
Most of the application developers I know who used to develop for Linux have agreed with me that OS X is light years ahead of Linux in every aspect: performance, design, security, everything. Apple putting their framework on Linux would be akin to putting a Ferrari body around a Lada engine.
I don't really understand what you mean at all. Most of the stuff you're praising is just framework-level stuff anyway, not OS level.
In case this is disguised bitching about X and the toolkits (GTK, Qt etc.).. if they wanted, Apple could ditch X and use their own display code, no problems.
The issue is how to make Java free to redistribute. Under the current license, no JVM can be included in a totally free distribution. Is there some license that can protect the trademark and portability that allows free distribution? Can such a license be written that is acceptable to Sun?
First of all, the trademark is a separate thing to the source. You can leave the trademark under a separate licensing deal, so even if people fork the code (that's OK), they can't use the trademark unless Sun agrees. That lets Sun insist on you passing the Java test suite before letting you use the logo.
Now all the IT executives have to check for is the Java logo - which they should already be doing, since that's what it's there for, to let them choose a JVM with confidence.
Making Java GPL or something very similar would ensure that no-one could fork it and hide the results in a way that prevented merging (either by them, or by third parties such as Sun themselves). At the same time, even if somebody came up with some funky MiniJava designed for phones, and all the phone vendors flocked to it, Sun could take their changes and merge them into their Java tree.
Even if Sun opens Java in a GPL-way, they will still be the custodian of Java, with enormous moral authority on the code. It's amusing that everyone fears forking, when in reality it lets everyone - including the Linux distros - unite behind the one codebase.
If Redhat, SuSE, Debian all shipped Sun's JRE - so it was there, on the CD, ready to go, that would do nice things for Java usage. Of course there will be forks, but that doesn't matter because it lets people experiment safely, and their changes can be merged. People don't want 1000 versions of Java, they want one strong one that everyone can distribute and thus unite behind.
There will always be pressure to merge forks into the main Java tree.
I wish Debian developers would pay more attention to these developments (especially to 'no-frills' and 'bleeding-edge' part.)
They do. And that's why they have the unstable distro, just for you. In fact, even cooler, you don't have to wait for a release to run all this cool new stuff, you can upgrade and run it now!
And in fact, I understand you can do the same with Fedora too, just upgrade as you go. If you have the bandwidth to download Fedora, you certainly have the bandwidth to run Debian unstable.
Even running sid feels like you're "behind the curve" (in terms of what my Gentoo friends are emerging)
Really? What stuff are you thinking of?
The first is Jeff Waugh's proposal that Epiphany replace Firefox as a the de-facto Linux browser.
I think the minutes simply say he suggested the possibility 'Epiphany become the official Linux port of Firefox'. That's not the de-facto Linux browser by any means, except in any sense that someone could say that the default GNOME web browser is the de-facto Linux browser (an arguable issue, given the popularity of KDE on many distros).
That said, I don't think the suggestion makes any more sense. Firefox having basically the same UI on multiple platforms is an important strength that would be totally compromised by letting Epiphany take on its Linux role.
This is heading towards proprietary extensions territory, a la Netscape/IE.
Not really. Would you call GNOME itself a proprietary extension, given that it doesn't come from a standard?
Parts of XUL (like XBL) have been proposed for standardisation, but the best argument for standardisation is having a working implementation. It helps when people can try out the idea and see how it works in practice.
In that sense, it wouldn't be bad for GNOME/Mozilla to do some integration work and let people build GNOME apps in XUL. Then they can propose it as a standard, hopefully before MS does with their XAML stuff. We really don't want any more web momentum tilting Microsoft's ways - it's just too bad to have a single company doing it.
Because, after all, Microsoft writes all the Windows drivers for hardware out there, right?
Once again, I think people miss the point. Linux's sound card support is actually very good, given that very few of those drivers are being written by the vendors. When you think of it that way - the commmunity has supported YOUR hardware - it's pretty good.
must.. resist.. temptation to moderate...
I wonder if they are running a Debian based or Debian itself, and Debian has another hole in it.
Funny. Too bad that was just a regular kernel hole, not one special to Debian's kernel. Any other distros can simply count themselves lucky the attackers didn't choose them.
...and when the car breaks down or they want to upgrade it, they go to a mechanic. The few that try to do it themselves either know what they're doing or break something pretty quickly.
Exactly. Most of them aren't changing the oil or installing new radios etc. They just use it, they don't try to change it. Really both Linux and Windows are much better in that regard - "oh you want to install a better radio? that means you need better cables.. I'll put that in for you!"
When cars do that, I'll consider them as good as computers in this respect.
Seems to work just fine for Knoppix; no, they're not as multi-platform as the base Debian dist is, but there's no reason the hardware detection scripts couldn't be ported. If you have a problem with your hardware detection, hack the script.
Different architectures have different methods of booting; some can only use one bootloader, some can use more than one. The point is, it's not just a matter of "hardware detection" so you can find the right drivers, the way you fundamentally boot and display things does not work the same over different architectures.
That doesn't mean what you're suggesting is impossible, but I don't think it's as easy as you were implying.
the people who care the least about a new version of Gentoo are the Gentoo users. It's a beautiful model... they way Debian was supposed to work.
:)
You mean like the way the people who care the least about a new version of Debian are existig Debian users, because most of them just dist-upgrade anyway? In fact because many of them run unstable and keep following it, ignoring the fact a frozen release happened at all?
Just for fun I recompiled a static version of ImageMagick using gcc 3.3, with Pentium IV optimizing
:)
But now you don't know whether the main difference is gcc 3.3, or the P4 opts, or the fact that it's statically linked
GTK is the GIMP Tool Kit, you know, that graphics program? It was originally written by the authors of GIMP to make their own program easier to write. I'm afraid the politics had nothing to do with it. GTK was originally written for GNOME no more than Qt was originally written for KDE.
Now why don't we try to mimic that interface into Linux instead of everything Windows has?
We are. See Keith Packard's work on the freedesktop.org X server. Otherwise.. Apple sent cease and desist letters to all the places you could get MacOS X themes. MacOS X is attractive but I could see myself getting bored with that look.
A flawless implementation of a crap algorithm is still crap.
No.. a flawless implementation of a crap algorithm just doesn't scale well. Of course bug rate is not the only criteria used when evaluating software, but people spend hundreds of man-hours fixing bugs.
It demonstrates that the quality of open source code is not automatically worse than professional proprietary code (which some people believe is the case). The important thing is that it's at least an attempt at formal study (and not simply personal collating of anecdotal reports).
When Microsoft release updates, you get them straight away. When Linus, etc. release updates, Debian might get around to putting it into the stable branch in 3-4 years.
That's ignorance or trolling on your part. Take a look at the frontpage of the Debian website - see the security alerts there? Each one of them is accompanied by backported fixes to stable. That's the whole point of the Debian security team, to make sure stable gets all critical security fixes.
One of the stupidist things linux/unix evolved into is a complete and utter lack of updating/rearchitecting the programming api's.
:widely implemented on various Unixes, now part of C99 and SUSv2 standards. Also asprintf() which does away with the problem in a different way, by being a memory allocating printf(), whose return value you must free.
Actually things like the c10k connection problem (see http://www.kegel.com/c10k.html) show the various new APIs that people are using for specific problems.
For problems like buffer overflow, see snprintf()
The reason is that most of the fundamental programming APIs are working well, and various libraries compete to provide other interfaces (eg. db3 vs sqlite for binary database storage, GTK vs Qt for widget library, etc etc).
Microsoft Terminal Services require orders of magnitude less bandwidth because it works at the toolkit level (as in, "draw a button and a drop-down menu"
I don't think this is what it does. I've worked a little on RDP (Remote Desktop Protocol), used by MS Terminal Servers, and from what I saw, it does indeed send pixmaps (and text strings, which the client renders). What the protocol does have is 2-level caching - there's a memory cache of bitmaps (so the server can say "draw cached bitmap 14 at this position"), and a persistent disk cache (so that next time you connect, the server might not have to send those bitmaps the first time it draws them).
If you're interested, check our the source code to rdesktop.
I suspect he thinks that they will not be a realistic alternative, and I also suspect he's right.
There is an enormous installed base of X applications out there, apps which would have to be rewritten. Sure, they could run over an X compatibility layer, but that tends to nullify most of the rewrite advantages in the first place.
If things like Xcb provide lower latency, and things like XDamage provide flexible management of windows/buffers for advanced effects, then the time invested in a rewrite probably aren't worth it.