Re:Other distros *do* produce betas/Mandrake first
on
RedHat 6.2 - RSN
·
· Score: 1
I hadn't realized that Bero had moved over to working with RedHat (previously, he had merged BeroLinux with Linux-Mandrake, to produce a Pentium compiled Mandrake). I haven't kept up with RedHat, have they also been moving towards providing Pentium compiled releases?
The PentiumGCC period was a dark time for Mandrake:-), but fortunately, gcc/g++ 2.95.2 is about a trillion times less buggy than any release since 2.72.2 (for which the bugs were at least mostly known), and I'm happily using it. Bero has done a lot to make sure packages (including recent kernels), are stable under gcc 2.95.2, and frankly, seeing him working for RedHat is a good sign for them.
Perhaps a Napster-like variation could perform such a service (or even Napster itself, you tell me:) Clients offer space, and in turn, they get their data backed up. Encryption could be built in, etc.
B.S. If the authors were distributing the list of "protected" sites, then maybe they would be commiting a crime of copyright infringement. But figuring something out, and describing it to people, should never be considered a crime (short of martial law). Trade secret, shmade secret. Like someone said earlier, the way to protect their URL list is to only distribute it in HASHED form!
BAD laws are being enacted to protect the lazy; namely companies that think everything they do is sacrosanct. That companies can buy the legislation to create these laws, does not make it right. If CyberPatrol had any brains, they would be *hiring* the people who figured out their lame encryption scheme (or other persons knowledgable in the field), and consulting them as to how best to distribute their list while retaining its secrecy.
The mantra of cryptography is that "security by obscurity" is NOT security, and CyberPatrol should NOT have relied on their encryption being a "trade secret" (as should now be obvious to them). Who is to say their competitors hadn't already discovered their blocking list? That is the only conceivable reason to keep their list private, and they've done a poor job of protecting against it.
You are confused. Red Hat does allow other people to "make money" from their code (and many people are attempting this; Mandrake Linux is an example of a branched-from-RedHat distribution). What they cannot do, is keep from releasing their modifications to RedHat's source code, if they distribute modified binaries. That is what the GPL requires.
On the other hand, the BSD license (apparently) allows someone to make modifications and distribute the binaries, without releasing the source to their changes (as long as the original authors get attribution.)
Hopefully that clears it up for you (assuming you weren't trolling).
Absentee voting should help these people out. They do not HAVE to vote on election day.
That said, I like waiting for the last minute, since I may always learn something about all the wacky propositions we have in California to want to change my mind. Absentee voting by mail sucks because you have to vote a week or more in advance (and apply several weeks ahead of time). A good, easy, secure internet solution is nice because you would be able to vote at the last minute (ideally without having to apply for such a voting option too far ahead of time).
In the future, I think the majority of voting will be done on computer, and it will help increase turnout. However, it will probably not do much to increase turnout for the poor (where the largest percentage of untapped voting power resides).
The comment in the story pays to use Acrobat Reader, but Ghostview will render most PDFs (and will do it faster and nicer than Adobe's Acroread (IMO). It won't do the fancier interactive of some PDFs (they are very rare).
This is undoubtably the most corrupt administration we have ever had.
That is a bold statement... Just look at the last few administrations. I can still not believe Bush (Sr.) could pardon his co-conspirators and not have received a public lashing. (At Christmas time... BRILLIANT!)
As for Gore... Well, I weep for the future. We'll see how well he does without help from the Federal reserve.
And Bush jr.? Why not just eliminate the middleman and vote for Texaco as president?
But if the game is easy to port to a PC (DirectX, Windows, etc..), it will probably be cheaper on a PC (since there is no $8 surcharge), and thus, it undercuts the market for this box. They would probably require certain restrictions on pricing of the PC version to get the "watermark" (to avoid undercutting).
But then, as a game manufacturer, I'd be pissed. Artificially high prices will encourage piracy, and there are a LOT more potential PC owning customers than X box customers...
I think IBM is actually the major patent holder on arithmetic coding (www.patents.ibm.com backs me up on that one).
Also, although it can certainly vary, I wouldn't say that Arithemtic coding is faster than Huffman. It has traditionally been considered slower (although hardware advances may have negated this).
Does anyone know of a device that will take serial line input, and generate keyboard output (meaning physically output on a PS/2 type connector). We have an important hardware card w/ crappy proprietary MSDOS software, that expects keyboard input only (seems to read the keyboard buffer directly).
We would like to automate many things, but there is no easy way to do it (that we've found). TSR's that copy serial to keyboard don't seem to work well (they don't allow ALT-Key combos, which we need to control the app). Running under DOSEMU doesn't work because the hardware uses DMA. Bar code readers also don't provide Meta keys (at least, not the ones we need).
A simple box that plugs into the keyboard port, and parses serial input into key strokes, would be ideal.
palette of 4096 colors (16 per line, I think, using Amiga "Copper"-like tricks.)
Blitter w/ logical ops and scaling features (more powerful than Amiga)
Previously mentioned sprite hardware (much more powerful than Amiga)
Sound specs suspiciously similar to Amiga
Basically, the idea and implementation of the helper chips (including a special hardware multiply, and graphics and sound processors) evolved directly from much of Jay Miner's work, although Needle & Mical certainly innovated on their own. Their Sprite hardware engine (as I said before) was quite a powerful feature. STUN Runner showed this off quite well.
The next step was moving away from integer ops and sprites, into true 3D. Hence the 3D0 (the next RJ Mical, Dave Needle (and David Morse(?)) collaboration).
The Lynx borrowed many ideas from the Amiga (with many innovative touches of its own), as would be expected from two guys who worked so closely with Jay Miner. The Lynx sprite engine was ultra-powerful (you gave it a linked list of sprites to draw, and it drew them; no preset limits).
The Lynx's main Achilles heel (IMO) was the relatively short battery life, which meant mainly action games, and few adventure games (ala Mario-whatever). Still, I have a couple of "Handys", and enjoy playing multi-player carts. It still impresses those who have never heard of it ("Wow! Is this new?")
If he didnt patent it, B&N or another competeter might have and used it against him.
Fine, so they buy into the patent bogosity to protect themselves. Why can't they then license the patent to B&N (and others) for a fair return, or until the patent licensing fees are paid for, then sit on it? IBM has many patents that they hold as protection, but which they do not litigate with. And many of those patents are far more legit than the Amazon patents.
I agree with this. I'm actually GLAD to have been forced away from Amazon (I was a loyal customer, thanks to their excellent service) I've found many other retailer's with better prices, better service, or both (I recently ordered a boatload of technical books from Books-A-Million, and I found World Party Music just to name a couple.) There are now MANY good competitors to Amazon, and even if they gave up their patent shenanigans, I'd be hard pressed to return. If anything, by forcing a small but active group of people to look elsewhere, they are increasing awareness of their competitors.
Look at www.noamazon.com and find other merchants, and you'll do well as a consumer.
We still are using mission critical apps that use NetBEUI as part of a data gathering and distribution system for scientific experiments. A DOS box collects the data and distributes it to NT clients (using NetBEUI) for further processing. The apps are closed and proprietary, so we don't have much choice about the networking.:(
However, if we can get this supported by VMWare, we can allow the currently dual booting analysis stations to remain in Linux mode all the time (which is what we prefer).
Time to check into the source, and fire off a request to VMware...
WHY is the universe accelerating? Where is the Universe getting the force to constantly accelerate itself??!!
I'm no expert, but I don't think the expansion of the universe is accelerating. Rather, it may not be decelerating fast enough for gravity to ever pull it all back in (ie. it may expand forever, since gravity gets weaker between more distant objects), which is what is called an "open" universe. "Closed" is if gravity eventually will be able to reverse it all. Either way, the rate of expansion should be slowing due to gravity...
By talking about the Earth, I didn't mean to imply that it existed 13 billion years ago. But the region of space we are in now did exist 13 billion years ago. Let's just assume that the Earth formed mostly out of local matter (say within a billion light years:).
Well, I know that the more distant the object, the faster it is moving away from us. Let's say that the quasar was moving at 10% light speed (from our point of view). The light was emitted 13 billion years ago, so the object is now another 1.3 billion light years away than we see it. Okay fine. But wouldn't that mean that it should have taken 130 billion years for the light to have gotten 13 billion light years away?
An earlier post helped explain with a balloon example, and I know that such an example is often used to try to demonstrate the "inflation" of the universe; and perhaps that explains this all. It just is a bit tricky for me to grasp. I get confused when astronomers talk about such distant objects as being "young", and that the farther away we see, the closer to the big bang we are looking (ie. perhaps just a couple billion years after it happened). The "inflation" is supposed to explain how things can get to be 13 billion light years away, while only being a couple billion years old. If something is 13 billion light years distant, why is the universe not at LEAST 26 billion years old (since if it were receeding at the speed of light, that is the minimum round trip time.)
Drawing black dots on a balloon shows how, as the balloon inflates, the dots move farther apart, and yet the dots themselves aren't moving across the surface of the balloon. Okay fine... Does that mean what we see as receeding matter is still an artifact of an expanding space, while in reality, we are not really moving (much) on the surface of our balloon (ie. space-time)? Is our local movement, due mainly to the coalescing effects gravity, just a minor perturbation of a continually expanding space? Rather than thinking of everything being blasted apart from a single point, is everything really just staying more or less in the same place, but shrinking at such a fast rate that the density of space it continually decreasing, and distorting our scale to make it appear that everything is actually moving away from us?
If something is 13 billion light years away, and everything started in the same place as we did (ie. Big Bang), then wasn't it a lot closer to us 13 billion years ago, and so how come it took 13 billion years for the light to finally reach us.:) Is this some subtle Relativity effect I don't understand?
Alternately, if the light *started* towards us 13 billion years ago, and the emitter is almost as old as the universe, then isn't that object now about 26 billion light years away? And if so, how can the universe only be 15 billion years old?
I FINALLY had to start browsing at 1, after just too much wasted time. I had hoped it wouldn't happen.
I'd like to have the option of marking the topicality of my own post. There are times when I want to post a Troll, or Flamebait, etc. just for fun. It would be nice, as a net citizen, to be able to mark my post as such for those who want to auto-avoid. This basic premise could be made more robust. Maybe there could be a default "First Post" thread for EVERY article, where posts can't be moderated. Then give people the option to avoid ever seeing this thread, if they want. Kind of like the old "crash" program on timeshare systems, or a "grafitti wall" in cities; it deters (somewhat) the malicious forms of these activities.
When I travelled to Indonesia (among other places) in mid-1997, I found small shops that had tons of bit copied, or independently mastered, bootleg DVDs. I was shocked because when I left a few months earlier, I hadn't even seen DVD players on the market (I thought they were a year away, or so). There were bootleg DVDs of movies that had only just been released in the U.S that summer!
So, actual bootlegging worldwide is already a problem, but it is apparently done without the need for DeCSS or equivalent. And stopping DeCSS will not solve this particular problem (although being able to rigorously enforce the regional coding might help slow the spread...)
Remember that the nyquist rate is greater than 2 times the highest reproduceable signal (not greater than or equal). There are actually an infinite number of 100 Hz sine waveforms that can be reconstructed from a 200Hz sampling frequency (by varying the amplitude and phase) That is just a mathematical pedantry though. Anything less than 100Hz is unique (mathematically; this assumes a perfect filter is used)
The problem w/ consumer digital sound is simply that the sampling rate and SN ratio, while fine on paper, require that the post-D/A filtering be very good. Unrealistically good (Heck, it was very ambitious for the time in which it was designed). But it is very hard to design zero-phase filters with a steep cutoff and low ringing, and I think this is what the audiophiles typically complain about. If the sampling rate had been 100,000Hz and 24-30 bits of quantization, then the filters could have been much more gentle in the high end, the D/A could have been very linear where it counted, and audiophiles who claim to like the "warmth" of analog systems, would be very pleased. Methods like oversampling, etc. are very helpful in this respect, but have their own implementation issues.
I personally think that "warmth" that analog audiophiles often talk about is simply a learned response to the type of gentle noise that good analog systems inherently have. A good digital system is all about reducing or removing noise, since the type of noise it introduces (aliasing typically) is quite unpleasant. But the noise of analog systems has a different and distinctive feel, which many people simply call "warmth". It is still an artifact of playback, though, and not an example of better reproduction of the signal. In other words, gaussian noise is something that people are just trained to accept, and when you mess with it, some people get upset.
Anyway, CDs were a good compromise at the time, for producing decent consumer audio that didn't degrade with plays, and shouldn't really be held as an example of what digital audio is truly capable of. Perhaps, in the future, we will have DVDs with much higher grade audio that WILL show everyone that digital can be both "warm" and accurate. Audiophiles will still want to play them on tube-amps, however.:)
Well, as of Mandrake 7.0 (and I think even 6.1), the default compiler is Gcc 2.95 (7.0 uses the more reliable 2.95.2). They do not appear to be using PGCC anymore, although they do ship egcs 1.1.2 (probably for backwards compatibility and kernel building)
I do know that recent kernels have incorporated numerous patches, submitted by Bero (of Mandrake), in order to allow it to be compiled by modern Gcc versions (ie. 2.95+). A quick check seems to indicate that kernel 2.2.14 is the minumum for those wanting to safely build with Gcc 2.95.
In a nutshell, use Gcc 2.95.2 if you can, and don't use PGcc if reliability is a consideration (from personal experience).
Slashdot Mirror
I hadn't realized that Bero had moved over to working with RedHat (previously, he had merged BeroLinux with Linux-Mandrake, to produce a Pentium compiled Mandrake). I haven't kept up with RedHat, have they also been moving towards providing Pentium compiled releases?
:-), but fortunately, gcc/g++ 2.95.2 is about a trillion times less buggy than any release since 2.72.2 (for which the bugs were at least mostly known), and I'm happily using it. Bero has done a lot to make sure packages (including recent kernels), are stable under gcc 2.95.2, and frankly, seeing him working for RedHat is a good sign for them.
The PentiumGCC period was a dark time for Mandrake
Perhaps a Napster-like variation could perform such a service (or even Napster itself, you tell me :) Clients offer space, and in turn, they get their data backed up. Encryption could be built in, etc.
PS. Get back to studying for CS255, David!
B.S. If the authors were distributing the list of "protected" sites, then maybe they would be commiting a crime of copyright infringement. But figuring something out, and describing it to people, should never be considered a crime (short of martial law). Trade secret, shmade secret. Like someone said earlier, the way to protect their URL list is to only distribute it in HASHED form!
BAD laws are being enacted to protect the lazy; namely companies that think everything they do is sacrosanct. That companies can buy the legislation to create these laws, does not make it right. If CyberPatrol had any brains, they would be *hiring* the people who figured out their lame encryption scheme (or other persons knowledgable in the field), and consulting them as to how best to distribute their list while retaining its secrecy.
The mantra of cryptography is that "security by obscurity" is NOT security, and CyberPatrol should NOT have relied on their encryption being a "trade secret" (as should now be obvious to them). Who is to say their competitors hadn't already discovered their blocking list? That is the only conceivable reason to keep their list private, and they've done a poor job of protecting against it.
You are confused. Red Hat does allow other people to "make money" from their code (and many people are attempting this; Mandrake Linux is an example of a branched-from-RedHat distribution). What they cannot do, is keep from releasing their modifications to RedHat's source code, if they distribute modified binaries. That is what the GPL requires.
On the other hand, the BSD license (apparently) allows someone to make modifications and distribute the binaries, without releasing the source to their changes (as long as the original authors get attribution.)
Hopefully that clears it up for you (assuming you weren't trolling).
That said, I like waiting for the last minute, since I may always learn something about all the wacky propositions we have in California to want to change my mind. Absentee voting by mail sucks because you have to vote a week or more in advance (and apply several weeks ahead of time). A good, easy, secure internet solution is nice because you would be able to vote at the last minute (ideally without having to apply for such a voting option too far ahead of time).
In the future, I think the majority of voting will be done on computer, and it will help increase turnout. However, it will probably not do much to increase turnout for the poor (where the largest percentage of untapped voting power resides).
The comment in the story pays to use Acrobat Reader, but Ghostview will render most PDFs (and will do it faster and nicer than Adobe's Acroread (IMO). It won't do the fancier interactive of some PDFs (they are very rare).
That is a bold statement... Just look at the last few administrations. I can still not believe Bush (Sr.) could pardon his co-conspirators and not have received a public lashing. (At Christmas time... BRILLIANT!)
As for Gore... Well, I weep for the future. We'll see how well he does without help from the Federal reserve.
And Bush jr.? Why not just eliminate the middleman and vote for Texaco as president?
But if the game is easy to port to a PC (DirectX, Windows, etc..), it will probably be cheaper on a PC (since there is no $8 surcharge), and thus, it undercuts the market for this box. They would probably require certain restrictions on pricing of the PC version to get the "watermark" (to avoid undercutting).
But then, as a game manufacturer, I'd be pissed. Artificially high prices will encourage piracy, and there are a LOT more potential PC owning customers than X box customers...
We'll just have to wait for more details.
I think IBM is actually the major patent holder on arithmetic coding (www.patents.ibm.com backs me up on that one).
Also, although it can certainly vary, I wouldn't say that Arithemtic coding is faster than Huffman. It has traditionally been considered slower (although hardware advances may have negated this).
Does anyone know of a device that will take serial line input, and generate keyboard output (meaning physically output on a PS/2 type connector). We have an important hardware card w/ crappy proprietary MSDOS software, that expects keyboard input only (seems to read the keyboard buffer directly).
We would like to automate many things, but there is no easy way to do it (that we've found). TSR's that copy serial to keyboard don't seem to work well (they don't allow ALT-Key combos, which we need to control the app). Running under DOSEMU doesn't work because the hardware uses DMA. Bar code readers also don't provide Meta keys (at least, not the ones we need).
A simple box that plugs into the keyboard port, and parses serial input into key strokes, would be ideal.
Thanks in advance.
http://sls.mcs.usu.edu/~kurto/lynx/faq.html
Other notable Amiga-derived features include:
- palette of 4096 colors (16 per line, I think, using Amiga "Copper"-like tricks.)
- Blitter w/ logical ops and scaling features (more powerful than Amiga)
- Previously mentioned sprite hardware (much more powerful than Amiga)
- Sound specs suspiciously similar to Amiga
Basically, the idea and implementation of the helper chips (including a special hardware multiply, and graphics and sound processors) evolved directly from much of Jay Miner's work, although Needle & Mical certainly innovated on their own. Their Sprite hardware engine (as I said before) was quite a powerful feature. STUN Runner showed this off quite well.The next step was moving away from integer ops and sprites, into true 3D. Hence the 3D0 (the next RJ Mical, Dave Needle (and David Morse(?)) collaboration).
The Lynx borrowed many ideas from the Amiga (with many innovative touches of its own), as would be expected from two guys who worked so closely with Jay Miner. The Lynx sprite engine was ultra-powerful (you gave it a linked list of sprites to draw, and it drew them; no preset limits).
The Lynx's main Achilles heel (IMO) was the relatively short battery life, which meant mainly action games, and few adventure games (ala Mario-whatever). Still, I have a couple of "Handys", and enjoy playing multi-player carts. It still impresses those who have never heard of it ("Wow! Is this new?")
Fine, so they buy into the patent bogosity to protect themselves. Why can't they then license the patent to B&N (and others) for a fair return, or until the patent licensing fees are paid for, then sit on it? IBM has many patents that they hold as protection, but which they do not litigate with. And many of those patents are far more legit than the Amazon patents.
Man, what a total bastardizaton of that song...
r ics.html
PS. Can I say bastardization on TV?
For the song lyrics, see the following:
http://www.beef-cake.com/interactive/lyrics/bluly
Look at www.noamazon.com and find other merchants, and you'll do well as a consumer.
Thanks Amazon! :-)
PS. Can't help with the neural nets, sorry...
We still are using mission critical apps that use NetBEUI as part of a data gathering and distribution system for scientific experiments. A DOS box collects the data and distributes it to NT clients (using NetBEUI) for further processing. The apps are closed and proprietary, so we don't have much choice about the networking. :(
However, if we can get this supported by VMWare, we can allow the currently dual booting analysis stations to remain in Linux mode all the time (which is what we prefer).
Time to check into the source, and fire off a request to VMware...
So, to us, this is potentially far from useless.
I'm no expert, but I don't think the expansion of the universe is accelerating. Rather, it may not be decelerating fast enough for gravity to ever pull it all back in (ie. it may expand forever, since gravity gets weaker between more distant objects), which is what is called an "open" universe. "Closed" is if gravity eventually will be able to reverse it all. Either way, the rate of expansion should be slowing due to gravity...
But maybe someone else will correct me. :)
By talking about the Earth, I didn't mean to imply that it existed 13 billion years ago. But the region of space we are in now did exist 13 billion years ago. Let's just assume that the Earth formed mostly out of local matter (say within a billion light years :).
Well, I know that the more distant the object, the faster it is moving away from us. Let's say that the quasar was moving at 10% light speed (from our point of view). The light was emitted 13 billion years ago, so the object is now another 1.3 billion light years away than we see it. Okay fine. But wouldn't that mean that it should have taken 130 billion years for the light to have gotten 13 billion light years away?
An earlier post helped explain with a balloon example, and I know that such an example is often used to try to demonstrate the "inflation" of the universe; and perhaps that explains this all. It just is a bit tricky for me to grasp. I get confused when astronomers talk about such distant objects as being "young", and that the farther away we see, the closer to the big bang we are looking (ie. perhaps just a couple billion years after it happened). The "inflation" is supposed to explain how things can get to be 13 billion light years away, while only being a couple billion years old. If something is 13 billion light years distant, why is the universe not at LEAST 26 billion years old (since if it were receeding at the speed of light, that is the minimum round trip time.)
Drawing black dots on a balloon shows how, as the balloon inflates, the dots move farther apart, and yet the dots themselves aren't moving across the surface of the balloon. Okay fine... Does that mean what we see as receeding matter is still an artifact of an expanding space, while in reality, we are not really moving (much) on the surface of our balloon (ie. space-time)? Is our local movement, due mainly to the coalescing effects gravity, just a minor perturbation of a continually expanding space? Rather than thinking of everything being blasted apart from a single point, is everything really just staying more or less in the same place, but shrinking at such a fast rate that the density of space it continually decreasing, and distorting our scale to make it appear that everything is actually moving away from us?
Ummm, sorry for so many questions...
If something is 13 billion light years away, and everything started in the same place as we did (ie. Big Bang), then wasn't it a lot closer to us 13 billion years ago, and so how come it took 13 billion years for the light to finally reach us. :) Is this some subtle Relativity effect I don't understand?
Alternately, if the light *started* towards us 13 billion years ago, and the emitter is almost as old as the universe, then isn't that object now about 26 billion light years away? And if so, how can the universe only be 15 billion years old?
I FINALLY had to start browsing at 1, after just too much wasted time. I had hoped it wouldn't happen.
I'd like to have the option of marking the topicality of my own post. There are times when I want to post a Troll, or Flamebait, etc. just for fun. It would be nice, as a net citizen, to be able to mark my post as such for those who want to auto-avoid. This basic premise could be made more robust. Maybe there could be a default "First Post" thread for EVERY article, where posts can't be moderated. Then give people the option to avoid ever seeing this thread, if they want. Kind of like the old "crash" program on timeshare systems, or a "grafitti wall" in cities; it deters (somewhat) the malicious forms of these activities.
When I travelled to Indonesia (among other places) in mid-1997, I found small shops that had tons of bit copied, or independently mastered, bootleg DVDs. I was shocked because when I left a few months earlier, I hadn't even seen DVD players on the market (I thought they were a year away, or so). There were bootleg DVDs of movies that had only just been released in the U.S that summer!
So, actual bootlegging worldwide is already a problem, but it is apparently done without the need for DeCSS or equivalent. And stopping DeCSS will not solve this particular problem (although being able to rigorously enforce the regional coding might help slow the spread...)
Remember that the nyquist rate is greater than 2 times the highest reproduceable signal (not greater than or equal). There are actually an infinite number of 100 Hz sine waveforms that can be reconstructed from a 200Hz sampling frequency (by varying the amplitude and phase) That is just a mathematical pedantry though. Anything less than 100Hz is unique (mathematically; this assumes a perfect filter is used)
:)
The problem w/ consumer digital sound is simply that the sampling rate and SN ratio, while fine on paper, require that the post-D/A filtering be very good. Unrealistically good (Heck, it was very ambitious for the time in which it was designed). But it is very hard to design zero-phase filters with a steep cutoff and low ringing, and I think this is what the audiophiles typically complain about. If the sampling rate had been 100,000Hz and 24-30 bits of quantization, then the filters could have been much more gentle in the high end, the D/A could have been very linear where it counted, and audiophiles who claim to like the "warmth" of analog systems, would be very pleased. Methods like oversampling, etc. are very helpful in this respect, but have their own implementation issues.
I personally think that "warmth" that analog audiophiles often talk about is simply a learned response to the type of gentle noise that good analog systems inherently have. A good digital system is all about reducing or removing noise, since the type of noise it introduces (aliasing typically) is quite unpleasant. But the noise of analog systems has a different and distinctive feel, which many people simply call "warmth". It is still an artifact of playback, though, and not an example of better reproduction of the signal. In other words, gaussian noise is something that people are just trained to accept, and when you mess with it, some people get upset.
Anyway, CDs were a good compromise at the time, for producing decent consumer audio that didn't degrade with plays, and shouldn't really be held as an example of what digital audio is truly capable of. Perhaps, in the future, we will have DVDs with much higher grade audio that WILL show everyone that digital can be both "warm" and accurate. Audiophiles will still want to play them on tube-amps, however.
I am sure that they are working on supporting the Itanium (IA-64) architecture, as more than just a 32 bit OS. Anyone w/ hard facts?
Well, as of Mandrake 7.0 (and I think even 6.1), the default compiler is Gcc 2.95 (7.0 uses the more reliable 2.95.2). They do not appear to be using PGCC anymore, although they do ship egcs 1.1.2 (probably for backwards compatibility and kernel building)
I do know that recent kernels have incorporated numerous patches, submitted by Bero (of Mandrake), in order to allow it to be compiled by modern Gcc versions (ie. 2.95+). A quick check seems to indicate that kernel 2.2.14 is the minumum for those wanting to safely build with Gcc 2.95.
In a nutshell, use Gcc 2.95.2 if you can, and don't use PGcc if reliability is a consideration (from personal experience).