A while ago/. had an interview with one of the Debian project leaders and one of the questions was about KDE & QT licensing issues. He said something to the effect that the QPL for QT 2.x is "good enough" and Debian will distribute KDE 2.0 once it's released, however, the *current* license (for QT 1.x) contridicts GPL somehow so that Debain cannot distribute KDE 1.x. But as I said, there will be no problem with KDE 2.0. Is this not the case anymore?
c) This is not about politics or philosophy. We are not willing to break the law, which seems to be what distributing KDE entails at this point. Unlike commercial distributors, we are an all-volunteer group, and can't afford to risk the chance of a lawsuit.
This is a completely clueless post that certainly does not deserve to be moderated up.
Rambus is faster, albeit more expensive, I think we all know that
Do you own a stock in Rambus or what? You are 2/3 right, though: Rambus is more expensive, I think we all know that.
The specs I've seen show 1 ns latencies, compared to the 8 ns on SDRAM
I don't know what specs you are talking about, but Rambus's latency is much higher than that of SDRAM. And it just so happens that latency matters much more to the real world performance than bandwidth.
Let's see, 100MHz SDRAM has a bandwidth of 800 MB/s. 800MHz Rambus has a bandwidth of 1600 MB/s. (BTW, it's actually 400MHz but it transfers data on both the rising and falling edge of the clock cycle, just like DDR SDRAM, essentially doubling the clock speed). That means a computer with 800MHz Rambus RAM is supposed to be 2x as fast as one with standard 100MHz SDRAM, right? Wrong! Check out some real world benchmarks, not Rambus Inc's marketing propaganda. The reality is that Rambus is just a hair faster (about 5-10%). The 2x bandwidth of Rambus is canceled out by significantly higher latency. Notice that this is 800 MHz Rambus we are talking about. Due to extremely high cost of production and low yields you are much more likely to find 700MHz or 600MHz Rambus in PCs. 700MHz Rambus cannot even claim to be faster than SDRAM in any real world benchmarks, and the 600MHz one is absolutely useless.
The bottom line is that Rambus is not faster. Period. It's just a shitty overpriced technology. Even now you'll be hard pressed to find real world applications where Rambus does better. Even for those few applications, would you pay six times the price for a 5-10% speed increase? And when DDR SDRAM finally becomes available, there will be no place for Rambus. DDR SDRAM has a higher bandwidth than Rambus (2100MB/s compared to 1600MB/s) AND lower latency AND, since it's an open spec and very similar to regular SDRAM, it will have lower prices.
Rambus will never become anywhere near as cheap as SDRAM. Remember that Rambus is a proprietary technology and requires the manufacturers to pay royalties. Also, it is much more expensive to produce than SDRAM and the yields are low. I just hate to see SDRAM prices being artificially inflated. What pisses me off the most about this whole Rambus fiasco is that the company is resorting to threats and lawsuits to push their inferior technology into the market. ___
Switch to cable. I have used Rogers cable for quite a while and I am rather pleased with it. I never used Bell DSL stuff because I heard from lots of people that:
1) The PPPoE is an ugly cludge. This POS protocol just doesn't work. The dialers are available for Windoze and MacOS, but both are extremely buggy, crash all the time and you just get disconnected once every blue moon. It's a pain in the ass to get it to work with Linux and Bell people don't support it (obviously -- hey, they can't even get it to work on windoze!). 2) Your IP address changes all the time, prventing you from running a mail/ftp/web server. I do that with cable though, because the IP address is essentially static. 3) You hear noise in the phone line when you make a call. The noise traps they install are far from perfect. 4) It's much slower than cable (I got the top speed of 310Kb/s on Rogers in Toronto).
I read in several places that/boot is the only exception that absolutely requires ext2. Now, I also heard some random noises that this may not be the case anymore. Either way, I could live with that.
... and more convenient. I *always* make 2 partitions whenever I do a *workstation* install of Linux -- one for / and one for swap. The swap partition is 64 Mb (same as RAM), the / is 2 Gb. That gives you plenty of space and no hassles. And I insist that this is the best way to configure a workstation/desktop machine. It's *much* easier than having multiple partitions and gives you the most flexibility. (You may need to create a small/boot partition as well if you have a huge HD).
Now, when installing Linux on a *server*, I certainly split the HD into several partitions:/, swap,/var,/tmp, and/home. If the machine runs a mail spool, then/var/mail gets a separate partition as well. That way the program binaries reside on one partition (/), user data on another (/home), temporary stuff on another (/tmp), and finally logs, spool, and other stuff on another (/var).
In either case, I see absolutely no point in putting/usr on a separate partition./usr contains program binaries, libs, and other stuff -- just like/bin,/sbin, and/lib. Why would you want to separate the two if they contain exactly the same kind of data? That does not make any sence.
There is a historical reason why/usr gets a separarate partition. To fsck a partition it must be either unmounted or mounted read-only. In the old days of Unix, once a partition was mounted read only, it could not be remounted read-write. I don't quite remember what the fsck scheme was. But anyway, this is not the case these days. If you look at the boot messages in Linux, you'll notice that the / partition first gets mounted read-only, then fsck checks all partitions (including/), once the check is successful, the / partition is remounted read-write.
Reality check: Cached data can be transfered quicker. Many HDD have around 2Mb of cache.
uhhm yeah. That's gonna do much good. It's only beneficial for occasional reads of small files, nothing else. Especially not sustained reads/writes.
There can be two HDD on this bus, don't just think about one HDD.
uhhm, so what??? only *one* drive can work at a time (the other one waits). IDE is *NOT* concurrent, contrary to what your post would seem to imply.
Transfer rates are going up all the time as HDD technology improves. Best to get a faster rate now than be bottle-necked in a few months.
uh-huh. sure. We still haven't crossed 33Mb/s "bottleneck", *3 years* after the introduction of UDMA33.
ATA/100 is serial you can have longer thin cables. The ATA/33-66 used those horrible short ribbon cables, its worth having this standard just for this reason.
I don't think so. If you really want longer ribbon length, you might as well go SCSI. It will also give you concurrency and up to 15 devices per SCSI host adapter. They are beating a dead horse with this ATA stuff. Then again, it's a pretty good marketing gimmick. For an average Joe ATA/66 = twice as fast as ATA/33. ___
Yeah, just like ATA/66, ATA/100 is just a marketing gimmick. HDs cannot even go up to 33MB/s, so 66 and 100 is pretty much useless. The only time it will help is when the data had been prefetched into the built-in HD cache. So you'll see a little improvement when reading small files once in a while, not sustained read though. And even then, that's what the OS file system cache is for.
They'd do much better if they'd finally make the IDE concurrent, so that more than one device per channel can work at a time; and increase the number of devices per channel. Oh wait, that would require a whole new interface. Never mind. On the second thought, we already have this interface. It's called SCSI. ___
While I agree with what RMS says, I want to ask: who cares? Motif is dying anyway. I have not seen a single *new* product that is being developed using Motif. It is still being used, but only in old apps that are still being maintained. All *new* apps use either GTK or QT. So, in a few years Motif will die its natural death anyway. Who cares that they changed the license? Nobody's gonna use it anyway, so I don't see what the fuss is all about.
Yeah, just like the other poster said, there is no way to *convert* an ext2 partition to a ReiserFS partition and vice versa. You need to back up the entire partition, reformat it as Reiserfs, and then restore the data. Just use cpio, gzip the output and burn it on CD or something.
I had a Pioneer 24x cdrom and I had major problems with it. Not only did any distribution not read the CD during the installation, it wouldn't even boot off of the CD! (btw, win98 wouldn't boot off of it either). I also had lots of problems with it just reading stuff from it. It often made the system unstable and caused all kinds of other intresting problems.
Later on I found out that this is typical of *all* Pioneer cdroms. It is well documented in the Linux kernel documentation. Apparently Pioneer cdroms violate the ATAPI protocol somehow which makes them very unreliable. They still work with windoze because it does not try to multitask them. I got rid of this piece of crap a long time ago, bought a 32x Creative and never had problems since. I don't know if their DVD-roms or SCSI cdroms have similar problems, but I am definitely never buying any more Pioneer crap -- one was enough for me.
Other people have already pointed out good reasons for very limited overall performace gains. There is one more thing though. In most of these tests (except for very low resolutions) the performace was limited by the *video card*, and not the CPU. Check the graphs. At 640x480x16 you can notice some performance difference, while at 1024x768x32 most CPUs perform pretty much the same -- because they work faster than the video card.
When the original Athlon was first released, Tom also ran a 3D Studio rendering benchmark, which is purely FPU-intensive. It did show Athlon to be about 40% faster. I wonder why he didn't do it this time...
This is off topic, but when will potato be released? It's been frozen since the middle of january, isn't it time to release it already? Would any Debian developers care to comment? ___
There's been a lot of noise about Athlon not being able to do SMP. But the truth is that P3 cannot do it either. The last chipset that was able to do SMP was BX. But Intel has phased it out in favour of *ghm* i820 and i840. These ones don't do SMP. And now they are under recall. Not to mention that they totally suck because of the Rambus RAM interface. Great job, Intel!
The only P3 chipset right now (aside from BX) that is actually worth anything is VIA's. But it doesn't do SMP either.
Sure, you could get Xeon if you need SMP, but it costs an arm and a leg, and lags in CPU clock compared to P3/Athlon (is it still stuck at 700MHz or did they release a faster one?).
So Intel is just as bad in the SMP department as AMD. In fact, I would argue that AMD is better. At least they have not screwed up their existing chipsets trying to push some overpriced proprietary memory. AMD says they'll have SMP by the end of the year. When that happens, AMD will wipe the floor with Intel.
It's about time you switched. The old set up used round robin DNS to load balance the servers! Uhhm, hello???? I am actually surprised it worked at all. I sure hope this time it's a more sensible setup. (hint hint: www.linux-vs.org)
go to www.m-w.com and look up the word "irony" If you still don't get it... oh well. There are 2 types of people: the ones with a sence of humour and the ones without.
Supermount liberated me from having to use the mount command to read a floppy. I mean, I absolutly love linux, but I really feel that you would have to be a serious geek with the intention of making life hard if you enjoy using the mount command to access a CD-Rom, Zip, floppy, etc
I don't think so. I mean how hard can it be to type "mount/mnt/cdrom"? Also, you can set up KDE and GNOME to automatically mount cdrom & floppy when you click on the desktop icon. You then have to right-click it and click unmount to unmount it. But hey -- I guess it can be a nuissance.
though mandrake is my distribution of choice, I am still interested in trying Debian
Mandrake is great for a desktop (well, looks like you already noticed that;-), but you simply cannot beat Debian for a server. I can describe it with one word: unfuckenbelievable! First, Debian undergoes a very long freezing process (like 4 month...). Potato (2.2) has been frozen since the middle of January. It is already rock-solid but Debian developers still don't feel it's ready to be labeled "stable". When they label it as stable they actually mean it. *That* is much more important to a server than the latest GUI gizmos. Besides that, Debian package system is much more advanced than rpm. You can upgrade your entire distribution with one command: "apt-get dist-upgrade". You can install any package with "apt-get install foo" -- it will auto-magically figure out all the dependencies and download & install all the required packages. (with rpm you have to do it manually, so upgrading the system is a pain in the ass). However, I do not feel Debian is "nice" enough (as in user-friendly) to be used on a desktop, especially not by a newbie.
We cannot please EVERYONE with just one product - therefore we can create several variants to appeal to different preferences and needs.
Exactly! That's why I run Mandrake on my workstation and Debian on my server.
Just wondering if X 4.0 is stable enough for every day use. I heard the original release was not quite there yet, but it got better since then. So, can anybody comment on stablity?
Not quite. GPL does not allow you to link a non-Free program with one that's GPL. So, for example, if a library is released under GPL, you cannot use that library in a binary-only software.
However, binary-only modules for the Linux kernel are allowed because Linus specifically said so. (I think this is called a GPL exception).
If somebody has more knowledge about the issue, please corret me.
Because ATI sucks. Next time you spend big bucks on a piece of hardware, you'd better check out reviews first. ATI Rage Fury MAXX costs almost as much as GeForce DDR, yet GeForce beats the crap out of it. (Who the hell came up with the name "Rage Fury" anyway???). The only thing ATI has that NVidia doesn't (AFAIK) is hardware DVD. But who cares? Even Celeron 300 is fast enough to play software DVD. OTOH, GeForce has hardware transform & lightning -- *that* is a very useful feature.
The chip also boosts the best support for DirectX and D3D.
Slashdot Mirror
A while ago /. had an interview with one of the Debian project leaders and one of the questions was about KDE & QT licensing issues. He said something to the effect that the QPL for QT 2.x is "good enough" and Debian will distribute KDE 2.0 once it's released, however, the *current* license (for QT 1.x) contridicts GPL somehow so that Debain cannot distribute KDE 1.x. But as I said, there will be no problem with KDE 2.0. Is this not the case anymore?
c) This is not about politics or philosophy. We are not willing to break the law, which seems to be what distributing KDE entails at this point. Unlike commercial distributors, we are an all-volunteer group, and can't afford to risk the chance of a lawsuit.
Uh-huh. Who's gonna sue you? FSF?
___
Rambus is faster, albeit more expensive, I think we all know that
Do you own a stock in Rambus or what? You are 2/3 right, though: Rambus is more expensive, I think we all know that.
The specs I've seen show 1 ns latencies, compared to the 8 ns on SDRAM
I don't know what specs you are talking about, but Rambus's latency is much higher than that of SDRAM. And it just so happens that latency matters much more to the real world performance than bandwidth.
Let's see, 100MHz SDRAM has a bandwidth of 800 MB/s. 800MHz Rambus has a bandwidth of 1600 MB/s. (BTW, it's actually 400MHz but it transfers data on both the rising and falling edge of the clock cycle, just like DDR SDRAM, essentially doubling the clock speed). That means a computer with 800MHz Rambus RAM is supposed to be 2x as fast as one with standard 100MHz SDRAM, right? Wrong! Check out some real world benchmarks, not Rambus Inc's marketing propaganda. The reality is that Rambus is just a hair faster (about 5-10%). The 2x bandwidth of Rambus is canceled out by significantly higher latency. Notice that this is 800 MHz Rambus we are talking about. Due to extremely high cost of production and low yields you are much more likely to find 700MHz or 600MHz Rambus in PCs. 700MHz Rambus cannot even claim to be faster than SDRAM in any real world benchmarks, and the 600MHz one is absolutely useless.
The bottom line is that Rambus is not faster. Period. It's just a shitty overpriced technology. Even now you'll be hard pressed to find real world applications where Rambus does better. Even for those few applications, would you pay six times the price for a 5-10% speed increase? And when DDR SDRAM finally becomes available, there will be no place for Rambus. DDR SDRAM has a higher bandwidth than Rambus (2100MB/s compared to 1600MB/s) AND lower latency AND, since it's an open spec and very similar to regular SDRAM, it will have lower prices.
Rambus will never become anywhere near as cheap as SDRAM. Remember that Rambus is a proprietary technology and requires the manufacturers to pay royalties. Also, it is much more expensive to produce than SDRAM and the yields are low. I just hate to see SDRAM prices being artificially inflated. What pisses me off the most about this whole Rambus fiasco is that the company is resorting to threats and lawsuits to push their inferior technology into the market.
___
Switch to cable. I have used Rogers cable for quite a while and I am rather pleased with it. I never used Bell DSL stuff because I heard from lots of people that:
1) The PPPoE is an ugly cludge. This POS protocol just doesn't work. The dialers are available for Windoze and MacOS, but both are extremely buggy, crash all the time and you just get disconnected once every blue moon. It's a pain in the ass to get it to work with Linux and Bell people don't support it (obviously -- hey, they can't even get it to work on windoze!).
2) Your IP address changes all the time, prventing you from running a mail/ftp/web server. I do that with cable though, because the IP address is essentially static.
3) You hear noise in the phone line when you make a call. The noise traps they install are far from perfect.
4) It's much slower than cable (I got the top speed of 310Kb/s on Rogers in Toronto).
___
I read in several places that /boot is the only exception that absolutely requires ext2. Now, I also heard some random noises that this may not be the case anymore. Either way, I could live with that.
___
... and more convenient. /boot partition as well if you have a huge HD).
/, swap, /var, /tmp, and /home. If the machine runs a mail spool, then /var/mail gets a separate partition as well. That way the program binaries reside on one partition (/), user data on another (/home), temporary stuff on another (/tmp), and finally logs, spool, and other stuff on another (/var).
/usr on a separate partition. /usr contains program binaries, libs, and other stuff -- just like /bin, /sbin, and /lib. Why would you want to separate the two if they contain exactly the same kind of data? That does not make any sence.
/usr gets a separarate partition. To fsck a partition it must be either unmounted or mounted read-only. In the old days of Unix, once a partition was mounted read only, it could not be remounted read-write. I don't quite remember what the fsck scheme was. But anyway, this is not the case these days. If you look at the boot messages in Linux, you'll notice that the / partition first gets mounted read-only, then fsck checks all partitions (including /), once the check is successful, the / partition is remounted read-write.
I *always* make 2 partitions whenever I do a *workstation* install of Linux -- one for / and one for swap. The swap partition is 64 Mb (same as RAM), the / is 2 Gb. That gives you plenty of space and no hassles. And I insist that this is the best way to configure a workstation/desktop machine. It's *much* easier than having multiple partitions and gives you the most flexibility. (You may need to create a small
Now, when installing Linux on a *server*, I certainly split the HD into several partitions:
In either case, I see absolutely no point in putting
There is a historical reason why
___
see the subject
___
That's how all caches work, you know. The OS is smart enough to read ahead and write behind (well, at least the OS I'm using ;-)
___
uhhm yeah. That's gonna do much good. It's only beneficial for occasional reads of small files, nothing else. Especially not sustained reads/writes.
There can be two HDD on this bus, don't just think about one HDD.
uhhm, so what??? only *one* drive can work at a time (the other one waits). IDE is *NOT* concurrent, contrary to what your post would seem to imply.
Transfer rates are going up all the time as HDD technology improves. Best to get a faster rate now than be bottle-necked in a few months.
uh-huh. sure. We still haven't crossed 33Mb/s "bottleneck", *3 years* after the introduction of UDMA33.
ATA/100 is serial you can have longer thin cables. The ATA/33-66 used those horrible short ribbon cables, its worth having this standard just for this reason.
I don't think so. If you really want longer ribbon length, you might as well go SCSI. It will also give you concurrency and up to 15 devices per SCSI host adapter. They are beating a dead horse with this ATA stuff. Then again, it's a pretty good marketing gimmick. For an average Joe ATA/66 = twice as fast as ATA/33.
___
Yeah, just like ATA/66, ATA/100 is just a marketing gimmick. HDs cannot even go up to 33MB/s, so 66 and 100 is pretty much useless. The only time it will help is when the data had been prefetched into the built-in HD cache. So you'll see a little improvement when reading small files once in a while, not sustained read though. And even then, that's what the OS file system cache is for.
They'd do much better if they'd finally make the IDE concurrent, so that more than one device per channel can work at a time; and increase the number of devices per channel. Oh wait, that would require a whole new interface. Never mind. On the second thought, we already have this interface. It's called SCSI.
___
While I agree with what RMS says, I want to ask: who cares? Motif is dying anyway. I have not seen a single *new* product that is being developed using Motif. It is still being used, but only in old apps that are still being maintained. All *new* apps use either GTK or QT. So, in a few years Motif will die its natural death anyway. Who cares that they changed the license? Nobody's gonna use it anyway, so I don't see what the fuss is all about.
___
Yeah, just like the other poster said, there is no way to *convert* an ext2 partition to a ReiserFS partition and vice versa. You need to back up the entire partition, reformat it as Reiserfs, and then restore the data. Just use cpio, gzip the output and burn it on CD or something.
___
I had a Pioneer 24x cdrom and I had major problems with it. Not only did any distribution not read the CD during the installation, it wouldn't even boot off of the CD! (btw, win98 wouldn't boot off of it either). I also had lots of problems with it just reading stuff from it. It often made the system unstable and caused all kinds of other intresting problems.
Later on I found out that this is typical of *all* Pioneer cdroms. It is well documented in the Linux kernel documentation. Apparently Pioneer cdroms violate the ATAPI protocol somehow which makes them very unreliable. They still work with windoze because it does not try to multitask them. I got rid of this piece of crap a long time ago, bought a 32x Creative and never had problems since. I don't know if their DVD-roms or SCSI cdroms have similar problems, but I am definitely never buying any more Pioneer crap -- one was enough for me.
___
Other people have already pointed out good reasons for very limited overall performace gains. There is one more thing though. In most of these tests (except for very low resolutions) the performace was limited by the *video card*, and not the CPU. Check the graphs. At 640x480x16 you can notice some performance difference, while at 1024x768x32 most CPUs perform pretty much the same -- because they work faster than the video card.
When the original Athlon was first released, Tom also ran a 3D Studio rendering benchmark, which is purely FPU-intensive. It did show Athlon to be about 40% faster. I wonder why he didn't do it this time...
___
You do need to be root to run Q2, Q1, as well as all other older games that rely on SVGAlib
___
This is off topic, but when will potato be released? It's been frozen since the middle of january, isn't it time to release it already? Would any Debian developers care to comment?
___
There's been a lot of noise about Athlon not being able to do SMP. But the truth is that P3 cannot do it either. The last chipset that was able to do SMP was BX. But Intel has phased it out in favour of *ghm* i820 and i840. These ones don't do SMP. And now they are under recall. Not to mention that they totally suck because of the Rambus RAM interface. Great job, Intel!
The only P3 chipset right now (aside from BX) that is actually worth anything is VIA's. But it doesn't do SMP either.
Sure, you could get Xeon if you need SMP, but it costs an arm and a leg, and lags in CPU clock compared to P3/Athlon (is it still stuck at 700MHz or did they release a faster one?).
So Intel is just as bad in the SMP department as AMD. In fact, I would argue that AMD is better. At least they have not screwed up their existing chipsets trying to push some overpriced proprietary memory. AMD says they'll have SMP by the end of the year. When that happens, AMD will wipe the floor with Intel.
___
You just can't possibly come up with a more moronic name!
___
It's about time you switched. The old set up used round robin DNS to load balance the servers! Uhhm, hello???? I am actually surprised it worked at all. I sure hope this time it's a more sensible setup. (hint hint: www.linux-vs.org)
___
Don't we already have drive-through church? Try to beat that!
___
go to www.m-w.com and look up the word "irony"
If you still don't get it... oh well. There are 2 types of people: the ones with a sence of humour and the ones without.
___
I don't think so. I mean how hard can it be to type "mount /mnt/cdrom"? Also, you can set up KDE and GNOME to automatically mount cdrom & floppy when you click on the desktop icon. You then have to right-click it and click unmount to unmount it.
But hey -- I guess it can be a nuissance.
though mandrake is my distribution of choice, I am still interested in trying Debian
Mandrake is great for a desktop (well, looks like you already noticed that ;-), but you simply cannot beat Debian for a server. I can describe it with one word: unfuckenbelievable! First, Debian undergoes a very long freezing process (like 4 month...). Potato (2.2) has been frozen since the middle of January. It is already rock-solid but Debian developers still don't feel it's ready to be labeled "stable". When they label it as stable they actually mean it. *That* is much more important to a server than the latest GUI gizmos. Besides that, Debian package system is much more advanced than rpm. You can upgrade your entire distribution with one command: "apt-get dist-upgrade". You can install any package with "apt-get install foo" -- it will auto-magically figure out all the dependencies and download & install all the required packages. (with rpm you have to do it manually, so upgrading the system is a pain in the ass).
However, I do not feel Debian is "nice" enough (as in user-friendly) to be used on a desktop, especially not by a newbie.
We cannot please EVERYONE with just one product - therefore we can create several variants to appeal to different preferences and needs.
Exactly! That's why I run Mandrake on my workstation and Debian on my server.
___
Just wondering if X 4.0 is stable enough for every day use. I heard the original release was not quite there yet, but it got better since then. So, can anybody comment on stablity?
___
Not quite. GPL does not allow you to link a non-Free program with one that's GPL. So, for example, if a library is released under GPL, you cannot use that library in a binary-only software.
However, binary-only modules for the Linux kernel are allowed because Linus specifically said so. (I think this is called a GPL exception).
If somebody has more knowledge about the issue, please corret me.
___
wow! me wipes drule off the chin...
___
Because ATI sucks. Next time you spend big bucks on a piece of hardware, you'd better check out reviews first. ATI Rage Fury MAXX costs almost as much as GeForce DDR, yet GeForce beats the crap out of it. (Who the hell came up with the name "Rage Fury" anyway???). The only thing ATI has that NVidia doesn't (AFAIK) is hardware DVD. But who cares? Even Celeron 300 is fast enough to play software DVD. OTOH, GeForce has hardware transform & lightning -- *that* is a very useful feature.
The chip also boosts the best support for DirectX and D3D.
What is that supposed to mean?
___