Post a Comment
An open source IRIX would be great. There are still a lot of SGI fans who really love their machines and the whole IRIX system. And there are still a lot of high-end applications available for it. Even if software is obsolete compared to other current systems, it should never go to waste, because you never know who might benefit and do something interesting with it.
Well, they did give dontate the very cool XFS filesystem of theirs to Linux as open source already.
http://oss.sgi.com/projects/xfs/
I am not sure there is anything else all that special about Irix compared to any other *nix.
They have taken http://www.osnews.com/subthread.php?news_id=15741&comment_id=159336... bunch of their "know-how" in other areas that SGI was known-for like NUMA, OpenGL etc and put it into Linux too.
See all their open source efforts here, its actually a pretty impressive list:
http://oss.sgi.com/projects/
> Solaris is based on AT&T SVR4, which SCO now own.
> Having to rewrite Solaris to remove SVR4 code is why open-sourcing Solaris took so long.
Do you know this as a fact? The story I collected is different.
"We paid a big, big bag of money a decade ago to get IP rights to do what we wanted to do with Solaris", said Scott McNealy. [ http://news.com.com/Sun,+HP+SCO+probably+wont+touch+us/2100-1016_3-... ]
"We have seen what Sun plans to do with OpenSolaris and we have no problem with it," McBride said. "What they're doing protects our Unix intellectual property rights." [ http://www.eweek.com/article2/0,1759,1785664,00.asp ]
Based on those quotes, it's unreasonable to think Sun had to remove SVR4 code, which SCO owns according to you.
1. SCO own SVR4. That's a well-known fact, and is the basis of their IBM/Linux and Novell lawsuits. That Solaris is based on SVR4 is also a well-known fact - many in the UNIX industry were rather pissed off at Sun and AT&T at the time and formed a (now-moribund) organisation to produce a rival UNIX, OSF/1. When other companies then aligned themselves with AT&T and Sun, this once again divided the UNIX market into two camps. These facts can be verified at Wikipedia, among other places.
2. SVR4 is based on SVR3 (AT&T UNIX) with features from BSD.
3. It might be true that Sun paid money to whomever was the UNIX owner at the time to do whatever they wanted with Solaris. If so, that begs the question why they had to release it piecemeal to make sure they weren't contravening licence agreements.
4. It's well known that SCO's beef is with the GPL and the fact that Linux on x86 has wiped out its UNIX profits. (The old SCO used to be the big UNIX vendor in terms of numbers, because its UNIX ran on x86.) It's also well-known that SCO for a long time, and possibly even now, distributes code under the "unconstitutional" GPL licence.
3. It might be true that Sun paid money to whomever was the UNIX owner at the time to do whatever they wanted with Solaris. If so, that begs the question why they had to release it piecemeal to make sure they weren't contravening licence agreements.
Probably because many codes are not part of SVR4 that purchase covers.
It is not a well known fact that SCO owns SVR4. 95% of the royalties still go to Novell, Novell can over-rule any changes SCO wishes to make to licensing terms and the APA explicitly excluded all copyrights and patents from the transfer. Hell, SCO still needs Novell's permission to even sign up new licensees, which is the basis of Novell's counterclaims against SCO. According to the text of the APA, Novell can direct SCO to make any changes to license agreements that Novell wants to, without getting any permission from SCO, and take these actions on their own if SCO doesn't listen.
Judge Kimball said it is unclear if SCO got ownership of any "Intellectual Property" whatsoever in the APA. The fact that it is unclear to a judge means that it is not a "well known fact" that SCO owns SysV. At best, it means it is uncertain. At worst, it means SCO are liars.
Edited 2006-09-06 00:06
3. It might be true that Sun paid money to whomever was the UNIX owner at the time to do whatever they wanted with Solaris. If so, that begs the question why they had to release it piecemeal to make sure they weren't contravening licence agreements.
At that time, AT&T still owned the license to System V. Sun and AT&T crosslicensed, and SVR4/Solaris were developed concurrently. Sun saw it, among other things, as a chance to 'rationalize' the gap between BSD-ish unix and AT&T-ish unix. The big-bag-of-money that Scott is speaking of is the licensing fee, and at that time AT&T did want a very big bag for the license.
it is much more involved than that...
http://www.levenez.com/unix/history.html#07
Solaris 1 is based on SunOS 4.1
SunOS 4.0 has influences from 4.3BSD (Tahoe)
SunOS 3.2 has influences from UNIX System VR3
therefore Solaris 1 has roots on 4.3BSD and is not based on UNIX System VR4, however as you can see Solaris 2.0 got influenced by UNIX System VR4.
it is not a trivial story...
Meh, Novell can just over-rule what SCO says and give them the right to do it anyways. Novell still has the final say on what licensees are allowed to do with Unix code.
And what they licensed from "SCO" (SCO is nothing more than a rogue glorified licensing agent on behalf of Novell) probably makes up very little of Irix.
http://www.novell.com/licensing/indemnity/pdf/10_7_03_n-sco_sgi.pdf
http://www.novell.com/licensing/indemnity/pdf/10_10_03_n-scoandsgi....
Did they license most of Irix from SCO as a Unix derivative or is it like Solaris and a BSD derivative?
Dude! License IRIX from SCO? Drop back a bit and think this through a couple times and see what other impressions you get.
SCO doesn't own IRIX. No matter whether you mean [the] Santa Cruz Operation, now owned by SUN after changing it's name to Tarantella and switching focus from operating systems to interoperability software a la Citrix, nor by The SCO Group, formerly Caldera, a failed Linux vendor that currently gets a 5% commission from Novell on Unix SysV license sales. IRIX is a trademark of SGI and is their product. They may still license some SysV code. It's hard not to do since things like init scripts or printer script "filters" are all unpublished property of $vendor. Or you could replace the init system scripts with a binary a la Solaris.
The only reason they still exist is because about 7 months before bankruptcy they were able to attract investors by suing IBM for $3 beeeeelyon!
They showed Jay Schulist's clean room version of Berkeley Packet Filter code as one example of proof of their infringement claims.
These guys are clowns who have parlayed hundreds of millions of investor dollars into a failure so abject that it attempts to delay the inevitable with any available option.
> Well, they did give dontate the very cool XFS filesystem of theirs to Linux as open source already.
I am aware of the SGI's contribution to Linux.
Not surprising since SGI is one of the earliest Unix vendors to pretty much give up their own arch and OS to Linux on Intel (Itanium).
However you cannot compare the amount of work for open-sourcing a filesystem (which runs on top of VFS that's commonly used by nix*) and SMP/NUMA code to the amount of work for open-sourcing the whole OS.
Of course, I also hope SGI can open-source all the cool technologies in IRIX, if not the whole OS.
Sun had the goals of keeping control of the systems, more or less, and generating a vibrant and Sun-oriented commmunity. I think IRIX fans would be happy just to get a code dump. Sure, it isn't a way of getting a vibrant development community going, but it certainly would be sufficient to enable driver updates to keep legacy applications working on newer systems, and more.
"What kept them?"
Seriously, as several people have said, this was obviously on the cards. IRIX hasn't had an update in *ages*.
For some nostalgia, you can make KDE look like an IRIX system if you go into the "Style," "Window Decorations" and "Colours" dialogue boxes and choose the right options.
Just goes to show, if Linux is gaining any ground, it is mostly at the expense of Unix. Like the Greek gods, the son replaces the father. Perhaps ReactOS will will fare better as a Windows replacement.
Not to say no one switches from Windows to Linux (I did for one) but it's just taken more seriously as a server OS right now, with which its price, sets it up to slowly replace its inspiration. I've mixed feelings there, I wouldn't have tried out a Unix-like environment without Linux and I never had a stake in Unix, emotionally or otherwise, but it's somehow a shame each time a branch thereof fades out.
I guess there's no point in keeping on a workforce to maintain IRIX when they can use Linux for free.
Didn't IBM massively cut its AIX workforce when they moved to Linux so as to take advantage of the free labor that produces Linux? A corp can save money by using free Linux rather than paying programmers to maintain some proprietary Unix.
Didn't IBM massively cut its AIX workforce when they moved to Linux so as to take advantage of the free labor that produces Linux?
I remember being told in 2001 that IBM will replace AIX with Linux in 2004 (roughly), didn't happen.
Today, majority of SMP POWER-system users run AIX, although Redhat and SuSE enterprise versions are 'certified' on POWER, and IBM contributes code to make sure Linux kernel runs well on POWER.
I can think of many reasons for IBM to have an in-house OS to keep pace with the innovation on POWER (RAS, virtualization, performance, etc).
I think that now there is nothing special about SGI. No more powerful and *different* workstations. May be the age of the powerful Unix workstation is finished, may be.
SGI should release IRIX as open source, at least the code that can be released without problems. Can be a good starting point to maybe "FreeIRIX" or "OpenIRIX" :the first multimedia-aimed free UNIX clone.
The decadence of SGI seems to be unstoppable. They will die, I think.
Edited 2006-09-05 21:33
>I think that now there is nothing special about SGI. No more powerful >and *different* workstations. May be the age of the powerful Unix >workstation is finished, may be.
That's not true regarding the current SGI offerings of their Itanium workstation. They are powerfull and very different from the rest. Please read the article http://www.osnews.com/story.php?news_id=15582
Anton
All right, may be I not getting the 'whole' picture. Wow, this Linux thing is a true revolution that will be remembered in the computing history books of the future...
However, I think it's a good thing to wipe millon dollars operating systems. But this does not harm any big company, in fact, it's reinforcing it by using free work. Do you think the MS monopoly couldn't in the future to incorporate some free software to Windows (at least at slow rate) ?
This is sad to hear,but not any news. This has been known for long and reported several months ago.
Anyway,Why, Irix is one of the most stable OS:es in the world and works great. Why not open source Irix and port it to x86 instead.
I am pretty fed up with all linux talk all the time. We sell and configure and use linux, solaris, osX and irix and even if i do like Linux, not everything has to be linux, i do prefer Solaris and irix.
Why not open source Irix and port it to x86 instead.
The answer has only been posted several times already: it's based on licensed code, to which they have no redistribution rights.
Besides that, it's not like they got the spare cash to make it 'licensed code-free'.
IRIX xfs produces the lowest I/O Wait time on Linux Distros than any of ext2/3 or Reiserfs or Reiser4.
I've tried it on Mandriva 2006 with all available kind of file systems
with xfs: copying concurrently 4 folders each with 500MB that includes small and large amount of files took 10% of CPU
and thus I was able to start other programs and use them
while on all other linux file systems it made the CPU at 100% and almost froze the system
I have confirmed this behavior on all distros
Then I did a very demanding test to see xfs power:
1. copy CDROM to disk
2. copy 12 instances of the previous folder internally to 12 different locations at the same time
3. transfer 17 GB over gigabit ethernet
4. open openoffice2 and opera9 and mozilla firefox1.5.0.2 at the same time
5. Run glxgears to obtain maximum GPU usage
and CPU was at 50% onlyyyyyyy.
Of course I have tried all kind of kernel updates (20 of them) to execlude the kernel as the source of high I/O wait times on linux with ext3/reiserfs.
I also have tested JFS of IMB AIX 5L to see how good it is, but it didn't hold a candle to xfs though was better than both ext3 and ReiserFS.
So this showed me that what we really need is not just another Unix but just an old linux with proper file system and device drivers will make the distro rock, and maybe then will linux gain serious footsteps towards market share. And thus SGI decision is perfect for them and for us.
I think IRIX died in 1999 when sgi started producing NT Workstations and sold tons of OpenGL patents to nVIDIA and Microsoft (Direct3D). There was a myth floating around about IRIX 7 (a port of IRIX to Itanium), around '99-00, but ultimately sgi decided to go with Linux on Itanium, rather than rewrite IRIX.
A friend of mine has been working on an sgi history and info page. It's still in a very precocious stage, but gives a lot of info on IRIX.
http://bama.ua.edu/~dodso003/irixinfo.html
Edited 2006-09-06 04:57
Itanic is "common" compared to MIPS?
MIPS is embedded all over the place. in millions of end systems. You've been duped by the billions of marketing dollars spent by Intel and HP, to make you think that Itanic is a ubiquitous chip and is deployed everywhere. The awful truth is that they were just trying to appease shareholders who have seen US $10B in Itanic spending go down the drain.
Ironically, Itanic is the very chip that took SGI down when they latched on to it. Now that IRIX is gone, what do they have left? Linux on Itanic? Why in the world would I want that?





