Further, if Red Hat wanted to make things awkward for CentOS, they could do whilst still complying with the letter of the GPL (e.g. by not releasing src.rpms - just the pristine sources and any patches they apply, and maybe the %prep and %build parts of the specfiles).
The fact that they don't do this indicates to me, at least, that Red Hat tacitly approve of CentOS and friends.
unless you're good enough (and want) to work for Intel, AMD, nVidia or some other major designer, architecture (as typified by novel designs) seems dead.
I don't understand how you can say this. Chip architecture is alive and well in not just the major companies but many startups and smaller firms as well. x86 architecture doesn't scale well still (although this is attempting to be remedied).
I won't deny that x86 is a pretty sucky 'architecture'.
There are a number of companies who are creating chips that solve memory and execution scale problems for requied devices (e.g. edge network devices like load balancers, content processors, IDS/IPS, firewalls, etc). these markets may not be the desktop PC market in size, but they are still many millions of dollars of business.
Fair point; yes there is some interesting work out there, implementing ASICs out there for pattern matching (similarly DSP). That said, IME, such ASICs tend to be uncompetitive (i.e. considering only performance/price ratio, rather than elegance) compared with commodity x86 hardware by the time they get to market. That doesn't strike me as a particularly safe career path.
Perhaps I was focussing a bit too tightly on general purpose CPUs, though. The last interesting CPU architecture was Transmeta's Crusoe, which appears to be going nowhere fast. Where are the novel CPU architectures of this decade?
I studied CS (Systems Architecture) from 1992-1995. At the time I entered, the x86 PC was popular, but not dominant - there were M68K Amigas, Macs, STs and Suns. There were SGI and DEC MIPS machines. Sun, Fujitsu and Solborne were pushing SPARC. The DEC Alpha was the hotrod processor. And the British-designed Transputer CPU looked like an interesting idea for massively parallel systems.
By the time I came out, the writing was pretty much on the wall, and these days, you just throw x86 boxes at the problem (as long as heat or power aren't a concern, anyway).
Don't get me wrong; knowing how computers work from the metal up is very handy and quite fulfilling (in the same ways that Physics is), but unless you're good enough (and want) to work for Intel, AMD, nVidia or some other major designer, architecture (as typified by novel designs) seems dead.
On the upside, embedded still seems OK, and should only improve - especially in the low-power portable segment. Also, electronics guys seem to have real problems getting their heads around software at times, so that might be another avenue to explore.
The reason that you can't get a board that meets your needs is because your needs are unmeetable. IIRC the Athlon 64 FX can't be paired in a dual processor configuration.
That said, there were a number of 486-based SMP machines around pre-1995. The 486 wasn't SMP-capable, but the custom chipset used made such usage possible.
To the original poster, the cheapest way to have your own custom motherboard is probably to buyout Giga-Byte, or Asus or ECS, or something.:-)
That sounds like a lazy excuse for not implimenting a decent architecture for loading outside drivers. I'll use DLL's as an example. I can replace a DLL to my program, without recompiling it and it will still work properly. You can make your own DLL and replace the one I have, and the program will still function properly.
...provided that the API supplied by the new DLL matches that provided by the old.
The Linux kernel needs this for drivers, and it's probably similar to the method Windows uses.
It already has it, has had it for about nine years, and it's called modules. I've many times compiled driver modules external to the kernel and used them as replacements for kernel-supplied modules. It works. Now, if you're saying that the Linux kernel should better support closed, binary-only driver modules (e.g. nVidia), that's a whole other argument.
If you don't care about Sun's support, you can buy the same chassis from the designers, Newisys. Sun didn't design these Opteron systems themselves! A UK-based reseller of Newisys hardware is Rainbow-IT. Never used 'em, just found 'em when I was googling a while back.
dd_rescue skips read errors, so maybe it skips write errors too (though maybe not, given its intended use). If it doesn't, it should at least provide a good start for a tool that does.
No, shred uses specially-chosen patterns in order to maximise the chances of making data unrecoverable from media recorded using older techniques (e.g. MFM and RLL), even using electron microscopy (i.e. techniques one might expect a hostile foreign intelligence agency to use). Shred was originally really intended for use when wishing to securely erase a single file. However, with modern journalling filesystems, this is no longer guaranteeable, so shred is deprecated for this purpose as it gives a false sense of security.
Shred also does a total of nine random passes, and, to be fair, it actually has its own RNG rather than just relying on/dev/[u]random. For this alone, it might be worth using (though Gutmann's paper suggests only "a few" random passes are necessary for modern discs; I interpret "a few" as two or three).
Your point about larger block sizes is well made, but I'm not sure what dd will do if it encounters a write error on a 512 byte block if its block size is greater than 512 bytes. In fact, IIRC, dd will terminate on a write error, so you probably need something that'll just skip over write errors.
...works with more than just Smartlink modems. I use it with the Agere softmodem built into my Toshiba laptop. I do, however, have to use AT commands to limit it to a 33.6 connection, as a V.90 connection actually ends up slower with all the retransmits.
The real solution is, of course, to use external RS232 modems.
Using shred isn't necessary for any modern hard disc. Something like 'dd if=/dev/urandom of=/dev/hda obs=512' is sufficient, as good as you're going to get if you're working from behind the drive's on-board electronics and is faster than shred. This is paraphrased from the epilogue of Peter Gutmann's classic paper http://www.cs.auckland.ac.nz/~pgut001/pubs/secure_ del.html. PRML has been widely used since at least 1998 (a 6GB IBM drive I bought then was PRML).
Actually, the RAID cards you refer to as 'firmware RAID' perform most of the RAID operations in the driver of the OS (though some, at least, provide enough BIOS support in order to get the OS booted). Promise and Highpoint's cheaper cards work like this, and Linux's ide-raid drivers used to support some of these chipsets (along with the manufacturer's own drivers).
IMHO, the only reasons to use these as anything other than bog-standard ATA controllers are a) if you have a pre-existing RAID setup that you wish to continue to use b) to migrate from or c) because your OS doesn't provide its own software RAID (XP Home, Windows 9x). As bog-standard ATA controllers, they're as good as any other (I use a Promise MBUltra 133 with two 200G Seagate discs and Linux md myself).
Everything the parent has written tallies with discussions I've had with a handful of Russian and Ukrainian friends I know, especially the 'be smart, or be a relative of a party official if you want a decent lifestyle' bit.
Linux is a miniscule fraction of the consumer market. There is absolutely no reason for any manufacturer, be it a Taiwanese OEM, or Intel / NVidia to open their source. And they've demonstrated they don't have to - either they simply ignore Linux or they ship binary drivers anyway.
That's the present situation, to some extent, but that's exactly the attitude I'd like to see change. You're right that the Linux userbase isn't massive, but it's probably somewhere around the size of the Mac userbase, and it's about time we started being picky about the scraps that the hardware vendors drop from their table.
Name me a modern graphics card that has really good Linux drivers. nVidia's binary drivers are marginally better behaving than ATI's binary drivers, but neither really works well.
[snip]
The biggest problem I see is that there's nothing to show the manufacturs that you're voting with your dollars based on linux compatibility. If I buy an nVidia card instead of an ATI card, nothing tells nVidia that I did it because their drivers are marginally better. They may think it's their flashy commercials that got me, or the bundles, or heck, even the box art.
True, but ATI in the past have given sufficient documentation (and Freely-licensed code!) to the XFree project such that there are reliable, Free, 3D-capable drivers for the Radeon range upto the 9250. That might be old-hat for a gamer, but for many users (and some less-hardcore) gamers, that's plenty fast enough.
Given how long it took XFree to incorporate ATI's code into a shipping release of XFree, it's not really surprising that they went the nVidia route (why bother writing drivers for an OS if they're only available to most users around the time you're trying to clear that chipset out at firesale prices ready for the next model?). But ATI have in the past shown willing, and I'm committed to buying their products as a small way of showing that that was the Right Thing to do. I've also emailed them to let them know that this is the case. I know I'm only one person, but if more people did this...
Unfortunately, it's taken me having problems with nVidia's drivers, the geForce2go in my laptop and the 2.6.10 kernel to bring me to this point; FWIW, nVidia's drivers used to work fine for me on the same hardware with a 2.4 kernel, but now they're quite broken and require me to linger on an older version AND patch the nVidia kernel driver AND patch the kernel. At some point, I expect nVidia's drivers to stop working entirely on this hardware, since they're completely unresponsive to the problem in the forums. And, in their defence, why should they be when they've already got my money (via Toshiba)?:(
The dist makers need to sit around the table with the driver writers and work out a dist-neutral binary API
No. The hardware vendors, if they wish to profit from the expanded market that offering Linux compatibility wins them, need to work with the community on our terms. Otherwise, screw 'em; there are other, community-friendl<-y|-ier> vendors to choose from.
The combination of X/fontconfig/freetype already provides better sub-pixel font rendering than Windows' ClearType, and has had it for some time (IMHO, Mac OS X's font rendering is still king of the hill, last time I compared). More details here. RH80 and up make the settings accessible through gnome-font-properties (as does any other distro with a recent version of GNOME, probably).
Compost... I mean composite is the worst option for video, no matter who makes the cable.
Well, that depends. I have a TV which has a rubbish S-Video signal path (colours are over-saturated and smeared, even testing with different source equipment and cabling), but which has a reasonable composite path (the picture was much better than the S-Video, even when I tested using an old, freebie audio RCA cable.
No, I also said that a Powerbook with equivalent performance and functionality to my Toshiba cost about the same. Apple hardware was once significantly more expensive, and may be in the future, but I don't believe it has been recently, or is at the moment. And that's without taking integration and depreciation into account.
I don't even own any Apple hardware (oh, I tell a lie - I have a 3rd-hand 68040 Mac Centris that I got for free!); all my bought-as-new hardware has been x86 running Linux. But if I was buying notebook hardware today, I would make sure to consider a Mac.
Same here, I'd love to go mac if the following two problems could be corrected:
1) the overwhelmingly higher cost of Apple hardware
2) the complete and utter lack of new game releases for Macs.
I can't help with the second, but as to the first...
Myself and a friend bought laptops at about the same time just over three years ago; I went for a Toshiba Satellite 3000-214 (PIII-M 933MHz, 256MB RAM, 20G disc, CD-R/DVD-ROM drive, nVidia GeForce2go graphics, 14" TFT). He went for a Powerbook. Not only did the Powerbook cost about the same amount (about 1500GBP), the hardware is better-integrated with MacOS than the Toshiba is (either with Windows XP or Linux), the battery lasts about twice as long at 5-6 hours (and still does, unlike the Toshiba battery which now only lasts ~45 minutes) AND the Powerbook is worth about twice what my Toshiba is now worth (source: eBay) at about 700GBP.
I don't think the legal issues can be that serious because pretty much every other distribution out there supports MP3 with their audio packages, and quite a few support NTFS. I think its more of RH not wanting to admit their wrong.
Say you're someone who feels "Linux" infringes your "Intellectual Property" Rights. Who would you sue? If you don't pick Red Hat (or Sun or IBM, if you can find a way of getting them on the hook), you need to step down from the board.
Red Hat know they're #1 (or #3, at least) target, and so respond by being extra-cautious.
Slashdot Mirror
The fact that they don't do this indicates to me, at least, that Red Hat tacitly approve of CentOS and friends.
This is exactly why the FSF asks authors to sign ownership of their works over to the FSF; so the FSF can take on any GPL violations that may occur.
I don't understand how you can say this. Chip architecture is alive and well in not just the major companies but many startups and smaller firms as well. x86 architecture doesn't scale well still (although this is attempting to be remedied).
I won't deny that x86 is a pretty sucky 'architecture'.
There are a number of companies who are creating chips that solve memory and execution scale problems for requied devices (e.g. edge network devices like load balancers, content processors, IDS/IPS, firewalls, etc). these markets may not be the desktop PC market in size, but they are still many millions of dollars of business.
Fair point; yes there is some interesting work out there, implementing ASICs out there for pattern matching (similarly DSP). That said, IME, such ASICs tend to be uncompetitive (i.e. considering only performance/price ratio, rather than elegance) compared with commodity x86 hardware by the time they get to market. That doesn't strike me as a particularly safe career path.
Perhaps I was focussing a bit too tightly on general purpose CPUs, though. The last interesting CPU architecture was Transmeta's Crusoe, which appears to be going nowhere fast. Where are the novel CPU architectures of this decade?
By the time I came out, the writing was pretty much on the wall, and these days, you just throw x86 boxes at the problem (as long as heat or power aren't a concern, anyway).
Don't get me wrong; knowing how computers work from the metal up is very handy and quite fulfilling (in the same ways that Physics is), but unless you're good enough (and want) to work for Intel, AMD, nVidia or some other major designer, architecture (as typified by novel designs) seems dead.
On the upside, embedded still seems OK, and should only improve - especially in the low-power portable segment. Also, electronics guys seem to have real problems getting their heads around software at times, so that might be another avenue to explore.
That said, there were a number of 486-based SMP machines around pre-1995. The 486 wasn't SMP-capable, but the custom chipset used made such usage possible.
To the original poster, the cheapest way to have your own custom motherboard is probably to buyout Giga-Byte, or Asus or ECS, or something. :-)
The Linux kernel needs this for drivers, and it's probably similar to the method Windows uses.
It already has it, has had it for about nine years, and it's called modules. I've many times compiled driver modules external to the kernel and used them as replacements for kernel-supplied modules. It works. Now, if you're saying that the Linux kernel should better support closed, binary-only driver modules (e.g. nVidia), that's a whole other argument.
If you don't care about Sun's support, you can buy the same chassis from the designers, Newisys. Sun didn't design these Opteron systems themselves! A UK-based reseller of Newisys hardware is Rainbow-IT. Never used 'em, just found 'em when I was googling a while back.
dd_rescue skips read errors, so maybe it skips write errors too (though maybe not, given its intended use). If it doesn't, it should at least provide a good start for a tool that does.
Shred also does a total of nine random passes, and, to be fair, it actually has its own RNG rather than just relying on /dev/[u]random. For this alone, it might be worth using (though Gutmann's paper suggests only "a few" random passes are necessary for modern discs; I interpret "a few" as two or three).
Your point about larger block sizes is well made, but I'm not sure what dd will do if it encounters a write error on a 512 byte block if its block size is greater than 512 bytes. In fact, IIRC, dd will terminate on a write error, so you probably need something that'll just skip over write errors.
The real solution is, of course, to use external RS232 modems.
Using shred isn't necessary for any modern hard disc. Something like 'dd if=/dev/urandom of=/dev/hda obs=512' is sufficient, as good as you're going to get if you're working from behind the drive's on-board electronics and is faster than shred. This is paraphrased from the epilogue of Peter Gutmann's classic paper http://www.cs.auckland.ac.nz/~pgut001/pubs/secure_ del.html. PRML has been widely used since at least 1998 (a 6GB IBM drive I bought then was PRML).
dmraid, apparently. Once you've migrated the data off your array, do yourself a favour, and rebuild it using Linux md RAID. :-)
Check http://linux.yyz.us/sata/faq-sata-raid.html and http://linux.yyz.us/sata/sata-status.html. There's a good chance you'll be able to use your existing SATA "RAID" card with open drivers included in the standard kernel, and without any performance penalty.
IMHO, the only reasons to use these as anything other than bog-standard ATA controllers are a) if you have a pre-existing RAID setup that you wish to continue to use b) to migrate from or c) because your OS doesn't provide its own software RAID (XP Home, Windows 9x). As bog-standard ATA controllers, they're as good as any other (I use a Promise MBUltra 133 with two 200G Seagate discs and Linux md myself).
Everything the parent has written tallies with discussions I've had with a handful of Russian and Ukrainian friends I know, especially the 'be smart, or be a relative of a party official if you want a decent lifestyle' bit.
That's the present situation, to some extent, but that's exactly the attitude I'd like to see change. You're right that the Linux userbase isn't massive, but it's probably somewhere around the size of the Mac userbase, and it's about time we started being picky about the scraps that the hardware vendors drop from their table.
[snip]
The biggest problem I see is that there's nothing to show the manufacturs that you're voting with your dollars based on linux compatibility. If I buy an nVidia card instead of an ATI card, nothing tells nVidia that I did it because their drivers are marginally better. They may think it's their flashy commercials that got me, or the bundles, or heck, even the box art.
True, but ATI in the past have given sufficient documentation (and Freely-licensed code!) to the XFree project such that there are reliable, Free, 3D-capable drivers for the Radeon range upto the 9250. That might be old-hat for a gamer, but for many users (and some less-hardcore) gamers, that's plenty fast enough.
Given how long it took XFree to incorporate ATI's code into a shipping release of XFree, it's not really surprising that they went the nVidia route (why bother writing drivers for an OS if they're only available to most users around the time you're trying to clear that chipset out at firesale prices ready for the next model?). But ATI have in the past shown willing, and I'm committed to buying their products as a small way of showing that that was the Right Thing to do. I've also emailed them to let them know that this is the case. I know I'm only one person, but if more people did this...
Unfortunately, it's taken me having problems with nVidia's drivers, the geForce2go in my laptop and the 2.6.10 kernel to bring me to this point; FWIW, nVidia's drivers used to work fine for me on the same hardware with a 2.4 kernel, but now they're quite broken and require me to linger on an older version AND patch the nVidia kernel driver AND patch the kernel. At some point, I expect nVidia's drivers to stop working entirely on this hardware, since they're completely unresponsive to the problem in the forums. And, in their defence, why should they be when they've already got my money (via Toshiba)? :(
No. The hardware vendors, if they wish to profit from the expanded market that offering Linux compatibility wins them, need to work with the community on our terms. Otherwise, screw 'em; there are other, community-friendl<-y|-ier> vendors to choose from.
Looks like Bitstream Vera Sans to me.
The combination of X/fontconfig/freetype already provides better sub-pixel font rendering than Windows' ClearType, and has had it for some time (IMHO, Mac OS X's font rendering is still king of the hill, last time I compared). More details here. RH80 and up make the settings accessible through gnome-font-properties (as does any other distro with a recent version of GNOME, probably).
Well, that depends. I have a TV which has a rubbish S-Video signal path (colours are over-saturated and smeared, even testing with different source equipment and cabling), but which has a reasonable composite path (the picture was much better than the S-Video, even when I tested using an old, freebie audio RCA cable.
I don't even own any Apple hardware (oh, I tell a lie - I have a 3rd-hand 68040 Mac Centris that I got for free!); all my bought-as-new hardware has been x86 running Linux. But if I was buying notebook hardware today, I would make sure to consider a Mac.
1) the overwhelmingly higher cost of Apple hardware 2) the complete and utter lack of new game releases for Macs.
I can't help with the second, but as to the first...
Myself and a friend bought laptops at about the same time just over three years ago; I went for a Toshiba Satellite 3000-214 (PIII-M 933MHz, 256MB RAM, 20G disc, CD-R/DVD-ROM drive, nVidia GeForce2go graphics, 14" TFT). He went for a Powerbook. Not only did the Powerbook cost about the same amount (about 1500GBP), the hardware is better-integrated with MacOS than the Toshiba is (either with Windows XP or Linux), the battery lasts about twice as long at 5-6 hours (and still does, unlike the Toshiba battery which now only lasts ~45 minutes) AND the Powerbook is worth about twice what my Toshiba is now worth (source: eBay) at about 700GBP.
Say you're someone who feels "Linux" infringes your "Intellectual Property" Rights. Who would you sue? If you don't pick Red Hat (or Sun or IBM, if you can find a way of getting them on the hook), you need to step down from the board.
Red Hat know they're #1 (or #3, at least) target, and so respond by being extra-cautious.
Of course, IBM's solution only works in conjunction with IBM's software (Windows-only, AFAIK).