|
|
Log in / Subscribe / Register

X.org, distributors, and proprietary modules

[Posted August 14, 2006 by corbet]

X11R7.1 (also known as X.org 7.1) was released back in May. It contains a number of useful new features, better 3D performance on a number of video adapters, and tons of fixes. It is, in general, the platform that X users probably want to be using. This release is not as widely used as it could be, however, and the associated story illustrates one of the costs of proprietary modules.

One of the developments merged into 7.1 was the AIGLX project, dedicated to the important goal of providing better eye candy for Linux users worldwide. Since this code had gone into the X.org mainline, the Fedora-based AIGLX developers decided that there was no reason to continue to maintain their own version. So the Fedora AIGLX repository stopped seeing updates; Fedora users wanting to use the current AIGLX code could get it straight from X.org 7.1.

The Fedora Core 5 distribution, however, shipped X.org 7.0. So, it was asked: would FC5 be updated to X.org 7.1? A major upgrade of this type might not be something all distributors would contemplate, but Fedora is supposed to move rapidly. As a matter of policy, Fedora tends to fix problems (and security issues in particular) by upgrading to the current release rather than by backporting fixes. So, back at the end of July, it was announced that there would be an X.org 7.1 update for Fedora Core 5.

Just one little problem stood in the way: the binary-only drivers from ATI and NVidia did not work with X.org 7.1 (ATI has since released an update). Perhaps, it was suggested, the X.org update could be postponed until such a time that the proprietary module vendors had released compatible versions? This idea was fairly strongly criticized on the mailing lists; Fedora is supposed to be a 100% free software distribution, and should not have to concern itself with the behavior of proprietary software vendors. Mike Harris, the Fedora X.org maintainer at that time (he has since retired), was quite clear on the subject:

Fedora does not support proprietary drivers at all, and never has, nor has any Red Hat OS that preceded it. Our OS products are not held hostage to the release schedule whims of 3rd party proprietary driver suppliers.

Part of the decision of choosing proprietary software, is making a conscious decision that you are held hostage by the vendor of that software to provide you with support for it. That unfortunate limitation should not expand to encompass all users of open source software. If that happens, everyone loses.

By this reasoning, everybody has lost. The Fedora advisory board met to discuss the issue; the resulting decision was that Fedora Core 5 would not be updated to X.org 7.1. The conclusion was that the interests of Fedora users using proprietary NVidia modules outweigh the interests of other users who would benefit from this update.

Needless to say, this decision has not been met with universal acclaim. One Fedora user asked:

If you were the owner of a company that had just announced plans to open source your drivers, would you feel you had made the right decision if a major linux distribution announced it had changed its mind about releasing the software that enabled your driver to run and delayed its shipment for two months *because* there were still vendors whose proprietary drivers were not updated?

The board has spoken, however, and the decision stands. Fedora users who are not up for the (sometimes hair-raising) experience of running from the development repository will have to wait for Fedora Core 6 to get X.org 7.1.

Lest anybody think that this is a Fedora-specific issue, a visit to this Gentoo forum discussion may be of interest. X.org 7.1 remains masked in Gentoo for the same reason - lack of proprietary vendor support - and over half of the people voting in the attached poll believe that situation should continue. Interestingly, only the x86 and amd64 architectures are being held back. The other Gentoo-supported architectures, for which NVidia and ATI modules are never available anyway, have moved forward to the current X.org release.

In both cases, distributors are acting in what they believe is the best interest of their users. Regardless of what one thinks of the outcome, it is encouraging that quite a bit of thought is clearly being put into the effects of changes on the user base. What is rather less encouraging is that the best interest of (at least) Fedora and Gentoo users is in the hands of proprietary module vendors, and that this dependency is imposing a cost on all users, whether they use the modules in question or not. These vendors should not have veto power over the release plans of free software distributions. One can only look forward to the day when current video hardware from all vendors can be used on 100% free systems.

to post comments

X.org, distributors, and proprietary modules

Posted Aug 14, 2006 19:46 UTC (Mon) by oak (guest, #2786) [Link] (1 responses)

One also hopes that MicroSoft hasn't kindly requested these companies to
abstain from updating their modules until Vista (with similar visual
enhancements to the Desktop) is out. Who knows how long it could then
take... ;-)

Microsoft is likely involved

Posted Aug 15, 2006 19:16 UTC (Tue) by qu1j0t3 (guest, #25786) [Link]

Xbox contracts probably affect open source drivers: http://slashdot.org/comments.pl?sid=191703&cid=15750306

X.org, distributors, and proprietary modules

Posted Aug 14, 2006 19:50 UTC (Mon) by warmcat1 (guest, #31975) [Link] (5 responses)

Much as I agree with the conclusion of the article, in fact there is the consideration that upgrading xorg may well bust some existing and working FC5 installs that are going to nightly yum the update. Of course the update would likely enable many more users to run better, for example this Samsung Q35 laptop only has working non-vesa video because it is running development repo xorg packages.

What is the meaning of FCn? It really has to mean that the set of packages that make it up are largely fixed in stone, it would suit many users that it meant that you will only get security updates. I have friends and family on Fedora boxes that I manage for them remotely and if they sucked down a new xorg it runs the risk of causing me more trouble than it would bring them benefit.

If FCn is "stable" and development is "CVS", might be interesting to offer instead "CVS snapshot" sets of development packages that work well for adventurous FCn users, libc versions and so on allowing. This big debate could then have been lessened by being able to point FC5 folks to the new xorg if they wanted it, even by something like fedoa-adventurous.repo if they want to buy into newer stuff with less risk than mainlining development.

Wrong distro for stability, then

Posted Aug 15, 2006 6:19 UTC (Tue) by eru (subscriber, #2753) [Link] (1 responses)

What is the meaning of FCn? It really has to mean that the set of packages that make it up are largely fixed in stone, it would suit many users that it meant that you will only get security updates. I have friends and family on Fedora boxes that I manage for them remotely and if they sucked down a new xorg it runs the risk of causing me more trouble than it would bring them benefit.

But if you really want this, you (or at least the friends&family computers you help manage) should really be running RHEL, CentOS or other "enterprise" distribution, which has a policy of doing only bug and security fixes after each major release. I'm running CentOS on a couple of computers on which the emphasis is on getting mundane work done with minimal need to worry about the platform "living" too much under my feet, and so far it has worked well. But this obviously is a boring solution :-).

Wrong distro for stability, then

Posted Aug 15, 2006 7:38 UTC (Tue) by warmcat1 (guest, #31975) [Link]

> at least the friends&family computers you
> help manage) should really be running RHEL,
> CentOS

Yes good advice, I have several servers on CentOS for the same reason. But Fedora's having the latest stuff (the RHEL/CentOS are still on 2.6.9) is not just appealing to the need for shiny things: many recent laptops and so on won't work well or at all on 2.6.9 and the level of xorg drivers in the "really stable" branches of The Redhat Way.

Fedora is the right answer I think. The question about issuing xorg 7.1 to FC5 was a difficult one, after all Fedora "doesn't support" binary drivers so it didn't have to not break those folks. But OTOH there is a ceaseless thumping of the Ubuntu drums coming from the jungle day and night and many people who currently need their binary drivers for a working desktop would certainly considered this beyond the limit.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 10:05 UTC (Tue) by liljencrantz (guest, #28458) [Link] (2 responses)

No, this upgrade, had it happened, should _not_ have busted any existing working FC5 installs that use the nVidia binary driver. In a correctly installed system the driver is installed as an RPM with dependencies to a specific X.org version, so X.org wouldn't be updated until the binary driver was available. I don't know if the nVidia drivers provided by the various repos out there have correctly implemented dependencies, but if not, then they should be fixed, and this would not be an issue.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 10:37 UTC (Tue) by warmcat1 (guest, #31975) [Link] (1 responses)

Newsflash - not everyone is using livna RPMs. Even for those that are, if the new kernel got installed and was used next boot, the last I heard the livna scripts send you to vesa. Okay so that is better than text mode, but it still breaks 3D, xv and so on for those users.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 14:50 UTC (Tue) by liljencrantz (guest, #28458) [Link]

If you're not using RPMs like those from Livnia, your system will break _every_time_ you upgrade your kernel, anyway. I'm not saying that one _shouldn't_ use non-packaged drivers, but if you do, you _will_ have to manually check and work around dependencies.

X.org, distributors, and proprietary modules

Posted Aug 14, 2006 20:25 UTC (Mon) by iabervon (subscriber, #722) [Link] (7 responses)

Wasn't one of the points of R7 that the distribution is modularized, to the point where it would make sense to ignore the overall version, and ship 7.1 except with a switch for a 7.0-with-nVidia/ATI server? Gentoo, at least, ought to be able to have proprietary drivers limit the server version, although I don't think portage is actually quite nifty enough yet to figure out what that means. (I.e., it should, but does not, eliminate package versions which lead to conflicts as configured when identifying the most recent version of a package; it could stop people from installing server 1.1 with non-matching drivers, but the user would have to figure out that masking >=1.1 resolves the conflict, rather than the system just doing the right thing automatically.)

X.org, distributors, and proprietary modules

Posted Aug 14, 2006 20:51 UTC (Mon) by drag (guest, #31333) [Link] (3 responses)

I don't think the idea of modularization was ever made to allow increased binary compatability with obsolete drivers. That would make things easier and would be nice, but I don't think it was the point behind modularlization.

The idea is that by moving to a modular build system you acheived several things.. For instance:
moving from X-specific build system to one that is popular (abiet considured seriously flawed by many) means that X developers don't have to designed build tools anymore. Also many more new developers would be familar with the new one.

Dividing up the system into clearly defined parts aided in dependancy tracking.. That way with clearly defined borders on functionality you could have people learn to work with a paticular section of code without having to make them understand everything else. Helps out new developers.

Also having it in modular sections aids greatly in distros like Debian with doing upgrades and such. With this you can upgrade peicemeal without breakage of X or apps that depend on X. Before you had to pretty much upgrade it all all the time.

With drivers and such it wasn't so much a big deal. You could always compile and run drivers seperately from X.

Thing is with binary only drivers your going to have this problem..
Your going to be forced to (on occasion)
1. run obsolete kernels for compatability
2. run obsolete X for compatability
3. run obsolete distros for compatability
3.5. Only be able to run very specific distros for compatability
4. any and all above may have serious serious bugs or security hazards you won't be able to use because they will break your binary compatability.
5. Unable to run a lot of open source software that may cause breakage with binary only drivers if they ran together or that they require certain patches or newer versions that are incompatable with your binary drivers.
6. unable to get effective support from Kernel developers/X developers/Redhat etc etc.

The easiest, quickest, cheapest, and most effective solution to this problem is:
Don't buy hardware that requires binary drivers!
Not to dificult.

the only thing that you can't find nowadays that is open source is is Drivers for nvidia cards or R500 and newer ATI cards (r1xx through r4xx have Free and open source drivers. This is everything from ATI 7200 to ATI x850, the r300 drivers may still require you to compile from CVS for greatest effectiveness)

I suggest a nice 945g intel motherboard and Pentium-D 9x0 cpu. They are cheap, easy to find, have everything you need (sata, pata, video, audio, gigabit network, usb 2.0, and some models have firewire) and they are inexpensive. Otherwise that Core Duo stuff is looking very very nice.

Other then video cards there is no reason you should be running Binary only drivers. You can easily find open source supported hardware. Nowadays even Wifi devices are almost universally supported by open source drivers. There realy isn't a reason to run ndiswrapper unless you very unlucky. (of course wifi needs a bit more maturing to make it trouble free in Linux as anybody can tell you)

X.org, distributors, and proprietary modules

Posted Aug 14, 2006 21:16 UTC (Mon) by iabervon (subscriber, #722) [Link] (2 responses)

Well, as you mention, modularization lets you upgrade the server separately from everything else (there can't be dependancies, because everything else could be talking to a server over the network, anyway). So being stuck with the server from 7.0 doesn't mean you can't use the latest versions of everything else or vice versa. I remember one of the specific reasons for modularization being that it took too long for all of the client-side stuff to be simultaneously stable, and so new server versions capable of supporting new drivers were unacceptably delayed; obviously, this works in the opposite direction as well, so a desireable effect is increased local client compatibility with obsolete servers (and therefore obsolete drivers).

(Personally, I've never used a proprietary driver, or even a separately-distributed free driver, even with the (old) nVidia card I have in one of my machines.)

X.org, distributors, and proprietary modules

Posted Aug 14, 2006 22:28 UTC (Mon) by drag (guest, #31333) [Link] (1 responses)

Ya I don't know for sure. Maybe a X developer would be able to clarify.

To me the point has always been so that you are able to run newer drivers then your current X version has, not so much you can remain binary-compatable with old drivers.

The idea is that as driver mature you should be able to install new versions for bug fixes and such.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 9:53 UTC (Tue) by daniels (subscriber, #16193) [Link]

You can, it's just that the ABI was broken in the move to 7.1. I think the specific change was adding privates to glyphs to allow glyph caching, which was actually an Xgl change ...

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 2:38 UTC (Tue) by dberkholz (guest, #23346) [Link]

> Wasn't one of the points of R7 that the distribution is modularized, to the
> point where it would make sense to ignore the overall version, and ship 7.1
> except with a switch for a 7.0-with-nVidia/ATI server? Gentoo, at least,
> ought to be able to have proprietary drivers limit the server version,
> although I don't think portage is actually quite nifty enough yet to figure
> out what that means.

In fact, Gentoo does exactly that, although it doesn't generally come up in discussions. All of the headers, libraries, etc are the same, and when we talk about 7.0 vs 7.1 being stable, all we really mean is server+drivers.

You're correct about the binary driver requiring a certain server version -- it doesn't yet work as seamlessly as one would wish. One proposal suggests that the binary driver package install a file masking the newer server and drivers, to make this work a bit better.

X.org, distributors, and proprietary modules

Posted Aug 24, 2006 13:39 UTC (Thu) by mikachu (guest, #5333) [Link] (1 responses)

I just compiled xorg-server 1.1.1 on gentoo with the abi breaking patch reverted, and am happily running nvidia-kernel 8762 on it. Maybe a special xorg-server-mutilated-1.1.1 could be introduced? :)

X.org, distributors, and proprietary modules

Posted Aug 24, 2006 21:14 UTC (Thu) by mikachu (guest, #5333) [Link]

scratch that, 8774 was just released with 7.1 support

X.org, distributors, and proprietary modules

Posted Aug 14, 2006 20:29 UTC (Mon) by tjc (guest, #137) [Link] (11 responses)

I suspose that as the Linux user base increases, distributions will divide along lines such as these.

I'm not sure that Ubuntu has upgrade yet either. I thought that they had, but I still see xorg 7.0 listed on the package list page.

X.org, distributors, and proprietary modules

Posted Aug 14, 2006 21:01 UTC (Mon) by drag (guest, #31333) [Link] (7 responses)

That's to bad.

I bought hardwares specificly because of Free software support and I want to try out AIGLX, but I don't have the time or the willingness to upgrade to 7.1 on my own from source.

It's a good thing I am using Debian unstable, then eh? At least I know they are a bit more millitant about Free software and all that. Maybe Debian will start to beat out Ubuntu or FC now?

Hoes what for politics?

It's all a bit irritating.

X.org, distributors, and proprietary modules

Posted Aug 14, 2006 21:14 UTC (Mon) by warmcat1 (guest, #31975) [Link]

> I bought hardwares specificly because of
> Free software support and I want to try
> out AIGLX, but I don't have the time or
> the willingness to upgrade to 7.1 on my
> own from source.

You should try the relevant packages from development repo. Just temporarily enable fedora-development.repo and start with, eg

yum install xorg-x11-*

see what the requires situation wants you to do. You can always go back to the FC5 ones by hand in a console if it doesn't work out.

X.org, distributors, and proprietary modules

Posted Aug 14, 2006 22:35 UTC (Mon) by palapa (guest, #612) [Link] (2 responses)

debian unstable is still using xorg 7.0.22 (same as testing)

X.org in Debian

Posted Aug 17, 2006 8:35 UTC (Thu) by Wummel (guest, #7591) [Link]

X.org 7.1 packages will first (and partially have already) enter the experimental branch. Debian releases for a roughly a dozen different architectures, and porting the X.org beast to these is not easy. If you are interested in the progress of this effort, you can join the debian-x mailing list.

X.org, distributors, and proprietary modules

Posted Aug 17, 2006 9:50 UTC (Thu) by micka (subscriber, #38720) [Link]

GIven that Debian (hopes to) get freezed soon, I see some reason not to change X version.
Anyway, I don't it as a "we won't break proprietary modules", as fglrx is already severely broken on 2.6.17 linux kernel (don't know for 2.6.16 which is currently in etch)

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 14:22 UTC (Tue) by tjc (guest, #137) [Link] (2 responses)

I bought hardwares specificly because of Free software support [snip]
So did I. But I'm not too concerned.

Historically, if a large distribution does the "wrong" thing, someone will fork it. Once upon a time, Red Hat refused to include KDE, so Mandrake was created. I wouldn't be suprised if someone does something similar with Fedora, creating a "we don't care about your proprietary drivers" version.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 20:51 UTC (Tue) by skvidal (guest, #3094) [Link] (1 responses)

Just wait! In 2 months FC6 will be out and it will have those drivers.

Then you'll have all your happiness.

Just a few months.

Or install FC6Test2 and work from there.

-sv

X.org, distributors, and proprietary modules

Posted Aug 17, 2006 14:26 UTC (Thu) by tjc (guest, #137) [Link]

Just wait! ... Just a few months.
You mean just this one time? I think this will become a regular occurance, if we are all going to hang about and wait for companies to update their proprietary drivers.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 2:54 UTC (Tue) by elanthis (guest, #6227) [Link] (2 responses)

Dapper is using 7.0, but Edgy is already upgraded to 7.1.

X.org, distributors, and proprietary modules

Posted Aug 17, 2006 9:05 UTC (Thu) by jospoortvliet (guest, #33164) [Link] (1 responses)

well, not here, then. I run edgy, but xorg has version 7.0.22ubuntu7 and
nothing newer. I have to run with the nv driver anyway, atm, as the nvidia
proprietary driver doesnt compile with the new 2.6.18-rc4 kernel, and i'm
starting to get more and more annoyed with these proprietary drivers. they
really DO hold you back, now and then. sucks that the nv driver is so
irritatingly slow...

X.org, distributors, and proprietary modules

Posted Aug 26, 2006 1:00 UTC (Sat) by crimsun (guest, #13750) [Link]

You've mistaken the packaging version of the 'xorg' metapackage for the actual version of the X Window System server (in the 'xserver-xorg-core' package, version 1:1.1.1-0ubuntu8).

Edgy has had bits and pieces of X.Org 7.1x since early July. As of right now it has something resembling 7.1+ :

$ head -5 /var/log/Xorg.0.log

X Window System Version 7.1.1
Release Date: 12 May 2006
X Protocol Version 11, Revision 0, Release 7.1.1
Build Operating System: Linux 2.6.15.7 i686

X.org, distributors, and proprietary modules

Posted Aug 14, 2006 22:07 UTC (Mon) by elama (guest, #262) [Link] (22 responses)

Isn't it funny somehow?

The whole thing started with one stubborn GNU not willing to work with buggy printer drivers. And now that one stubborn GNU has become a mighty herd the whole stampede is stumbling over proprietary video drivers and their lack of consequence....

FWIW, I'm stumbling with the herd, but I wish I could be as stubborn and consequent as the first GNU.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 2:26 UTC (Tue) by drag (guest, #31333) [Link] (21 responses)

hehe..

Got "RMS envy"?

You can 'be strong and not stumble' if you want. You just have to do things that people from the 'best tool for the job' crowd thinks is completely irrational; which is to be willing to sacrifice for your ideals.

As far as hardware goes there are four groups of manufacturers of hardware in reference to open source software.
1. Vendors that are willing to support with documentation and/or drivers.
2. Vendors that simply want to make sure their stuff works in Linux.
3. Vendors that want to make their stuff work as long as it's binary only drivers.
4. Vendors that don't care.

So on Linux and Free software support it realy depends on what your goals are. Are the goals of Linux distros to be the most popular usefull Operating system possible? Or is it the goal of Linux distros to be the most popular and usefull Open Source/Free software operating system there is?

Remember that right now there is no obvious legal, technical, or financial reason why drivers need to be open source. Among the most supported and popular hardware out there it is supported by mostly closed source drivers. Atheros for Wifi. ATI or Nvidia drivers for Linux. Nvidia drivers for their motherboards.

These drivers are often considured the most featurefull and best performing drivers out there by many people. Anytime somebody asks about wifi they are told by at least a couple people to buy atheros cards. Anytime somebody asks about video cards they are told to buy Nvidia.

So vendors in group #2 are the pivitol group. They will more likely choose to go closed source if they can, but they will choose to go open specs or open source if they can.

If you deny binary only drivers then you will get open source support for vendors in group 1 and 2, but will loose the support of vendors in group 3.

As a "for instance" Texas Intruments was interested in support Linux. They wanted to make sure their wifi was supported. So obviously they had a desire to see this happen, and if closed source was unnacceptable then they probably could of been pressured into support a open source driver. But they don't have to have a open source driver for linux because people don't object to binary only drivers. So they went and worked with Conextent and made sure that their wifi driver was compatable and released a press release saying that they supported Linux.

However if you want the support of group 3 you will get binary only drivers from group 2. So if you support group #3 you gain teh support of groups 1, 2, AND 3, but most drivers will end up being closed source.

So for distros and developers to think about is:
A. do they want the most support of vendors possible, then in that case they support binary only drivers.
or
B. do they want the most open source drivers support possible, then in that case they deny all support and use legal restrictions to avoid having group #3's support.

Of course it's not all black and white like that. But it is true that many vendors will support binary-only drivers if they can get away with it, but will release documentation and driver support if they have no other choice.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 3:10 UTC (Tue) by elanthis (guest, #6227) [Link] (19 responses)

"1. Vendors that are willing to support with documentation and/or drivers."

The problem is that this group largely doesn't exist in the video driver realm. As has been noted, Intel's recent code release still using binary blobs for certain functionality.

You may think that "best tool for the job" people are sacrificing their ideals, but when it comes down to it, I'd rather that the scientists and doctors who are using high-end hardware requiring proprietary drivers be able to do their work than allow people to die in order to hold up their ideals over something and completely irrelevant to peoples' lives as software.

Sure, they would be better off if they could use that hardware with Free drivers, as they wouldn't have to worry about vendor lock-in or an inability to debug problems they run into. But the whole value of that Freedom is to allow them to do their job better and with fewer restrictions; making choices that eliminate their ability to do that job in order to improve their ability to do that job just doesn't make any sense. It's like killing for peace.

Average users aren't in much better of a position in many cases, either. The vast majority of my friends don't use any software other than games (and the OS the games run on). Proprietary, closed-source games. For these people, the whole reason for owning a computer is to play these games. They don't need or want the source. They'd love to be able to "share with their neighbor" only because they'd prefer to get their software gratis. To them, software Freedom is entirely useless as anything other than a way to spend less money. Running Linux or another Free OS does nothing for them. Running Free drivers does nothing for them. Even if they had a fully functional Free OS, the message is lost on them anyways because they WANT to play non-Free games, and taking that option away from them doesn't place them among group of true Free Software users, it simply removes them from the group of software users.

I am definitely a "best tool for the job" sort of guy, and I do firmly believe that Open Source is pretty much always the best tool for the job from a technical and social level. I also know that many people don't care about the best tool for the job, and instead care about the best tool for the profit, and that these people create a lot of useful tools that Free Software hasn't yet been able to mimic, and until then, most people are going to be stuck using those monopolistic proprietary tools.

I believe the games anecdoce is possibly the most important. Games are what drive a majority of constumers to buy PCs, are what drives a majority of consumer-level hardware improvements, and are the gateway for sales of a lot of other hardware and software.

Until the Free Software movement can manage to start producing games on par with the proprietary offerings of the day, Free Software and its ideals has little hope of really reaching out to a majority of users or directing most companies.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 4:14 UTC (Tue) by bojan (subscriber, #14302) [Link]

> They don't need or want the source.

Which is quite understandable. Not every user is a developer. But, they still do benefit from the fact that the source is distributable and modifiable by others, as the cost of development and distribution is greatly reduced. This is especially true long term.

> Running Linux or another Free OS does nothing for them.

Of course it does. It lowers the cost of the development and distribution of the OS, therefore bringing down the cost of their machine. Something, I'm sure, they will have an interest in.

It is easy to focus attention on immediate and short term effects of free software end pronounce that "most users don't care". They may not be entirely sure as to why they *should* care (mostly due to lack of infromation), but the economic effects of free software are surely going to make them care, as these determine how much money users have to part with in order to do what they want to do.

PS. This is exactly why Microsoft isn't willing to let OEMs go and demands they sign a contract with Microsoft to ship every single PC preloaded with Windows. The "software cost" then gets counted as "hardware cost" and users don't have any other choice but to accept it as something that "simply is the way it is". Bill may be many things, but he sure isn't stupid (as the balance on his bank account clearly shows :-).

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 4:20 UTC (Tue) by drag (guest, #31333) [Link] (17 responses)

""You may think that "best tool for the job" people are sacrificing their ideals, but when it comes down to it, I'd rather that the scientists and doctors who are using high-end hardware requiring proprietary drivers be able to do their work than allow people to die in order to hold up their ideals over something and completely irrelevant to peoples' lives as software.""

How many Linux users using Nvidia binary drivers are doing it because they are doctors and require high end 3d graphics?

How many people need these graphics? How many people are going to loose their jobs or loved ones or anything like that becuase they suddenly uninstall flglx or they sudden go a buy a ralink card and get rid of their atheros one?

Probably the answer is: 'not very many at all'.

Think about it. How many of us NEED nvidia.ko on our machines? I know I am better off once I got rid of that crap.

But going around and saying "oh I wish I don't need binary drivers" is something else entirely. The answer is _You_probably_don't_.

"Average users aren't in much better of a position in many cases, either. The vast majority of my friends don't use any software other than games (and the OS the games run on). Proprietary, closed-source games. For these people, the whole reason for owning a computer is to play these games. They don't need or want the source."

Then why on earth would they ever want to run Linux?!

Windows provides better performance. Better compatability and probably better stability. It's simply a much much better tool for playing propriatory video games. There is no contest.

It's never going to happen that propriatory game companies will ever support Linux properly. Becuase if Free software becomes popular it will probably put them out of business.

I am not saying that it's morally superior to be pro-free software or anti-free software. I am saying there is a pratical choice.

What we have now is this sort of bizzare middle ground that doesn't realy work. A bog with unsteady footing.

Every distro supports closed source drivers. Even Debian. I can install module-assistant and go m-a a-i nvidia && apt-get install nvidia-glx and it will probably work. But they don't 'officially' support it. So it leaves users in Limbo. They get broken kernels with broken drivers and nothing ever works out of the box and nothing ever is effortless. Hear about a security update? You can't use it because now you can't play Doom3. If those drivers are officially supported they you would be able to get the latest kernel update and still be able to play Doom3.

If the drivers are denied completely by Ubuntu/Debian/FC/et al then people who want Free software drivers will be able to easily find out what is supported and will get the best quality free software drivers they can get.

Right now we have a huge number of people with half-working hardware bitching about FLGX drivers and NDISWrapper BS and how the kernel needs a stable ABI and this and that when perfectly good drivers for their hardware exist for their R200/R300/R400 ATI video cards and their Broadcom wifi and their TI wifi and they don't know it because the Ubuntu forums are full of howtos on getting craptastic ndis stuff running and how ATI sucks.

Anyways. For games there are lots of nice Free ones.

If your friends like going out and spending 400 dollars on a PC + 200 dollars on a video card + 100 dollars on Windows + 100 dollars on assorted windows BS in order to get it usable + 70 dollars for a video game worth about 4 hours of gameplay only to find out that previously installed game broke their dvd drivers with it's copy protection then that's their problem.

They'll learn that fancy != good, eventually. :-)

And I sincerly doubt that FSF is going to be any good at game making. Other people are working on that.

For instance Tremulous. It's a great game, but it doesn't have any players because everybody is busy with struggling to get Cedega or Wine to barely work.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 4:24 UTC (Tue) by drag (guest, #31333) [Link] (14 responses)

http://debianlinux.net/games.html

There is a fairly up to date list of fancier open source games.

How many people know gamers and they had long conversations about what they'd like to change aobut this or that game or how they'd like to make a model or their own character and stuff like that?

With open source games they can actually DO that. Making video games or contributing with testing and such would be a fun hobby for a lot of people, but I bet that didn't know most of the stuff in the above link existed.

mmorpg, racing games, flight simulations, lots of FPS, stragety games, etc etc.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 9:37 UTC (Tue) by elanthis (guest, #6227) [Link] (7 responses)

"With open source games they can actually DO that."

No, no they cannot. Entire teams of artists, musicians, and veteran game coders take many months or years to push out even mediocre games these days.

Even given that I know how to code, which I do, I still completely lack the time or resources to make my ideal game, and there certainly isn't an OSS game I could springboard off of. I'd have to do everything from the ground up.

The only games that OSS is even close to catching up on is FPS games and "classic" games. Modern action games, modern platformers, modern adventure games, and modern RPGs are all far beyond anything any OSS offering has even come close to producing.

X.org, distributors, and proprietary modules

Posted Aug 16, 2006 5:57 UTC (Wed) by drag (guest, #31333) [Link] (6 responses)

"Even given that I know how to code, which I do, I still completely lack the time or resources to make my ideal game, and there certainly isn't an OSS game I could springboard off of. I'd have to do everything from the ground up."

What do you mean there is no OSS game for you to use?

GPL'd Quake3 not good enough for you?

Ogre3d gaming engine not good enough for you? Crystal Space not up to scratch?

This is the shit I am talking about. You're so convinced that only closed source companies are capable of making good video games that you don't even pay attention to what is happenning around you.

And you not alone. Most people don't even know this stuff exists. All the focus is on getting stupid Wine to work.

Have you ever seen the game mod community?

People take gaming engines based on Quake3, Enemy Territory, Quake2, UT2004, Halflife2.

COUNTERSTRIKE was one of the most popular game of all times and it was a MOD that was made at no cost and freely downloaded!

They don't charge money for any of it. But they all come from a Windows culture were closed source is natural and you should restrict access just because it's how it's done.

Linux needs to cultivate a culture of making games and making 3d stuff easy and 'just work'. Wings3d and Blender should be made aviable right beside Gimp. GTKRadient should be a standard feature.

How many hundreds of games have hobbiests produced? I remember back when Quake2 was popular it was easy to make models for it. People made litterally THOUSANDS of models. Out of those there were at least nearly a hundred that were as good or better then the ones that Quake2 actually came with.

A lot of these things come realy damn close to the quality that top commercial games offer.

Ogr3d and CrystalSpace as well as a couple other game engines are Linux centric and completely open source with support for a veriaty of scripting languages. These things can be used to produce graphics and such that are as high quality as anything else out there. Exceptionally fast, flexible, and Free software. Most of the end users are Windows folk though.

All you need is something fun you want to do. Lots of people do it all time. Lots of programmers would love to make decent games. In fact the majority of programmers seemed to want to become programmers because of games they played.

You need to make it easy for artists and sound people who want to play around to make stuff and they will. It happens all the time.

The stuff is starting to pick up more and more. It's slow, but open source stuff is gaining momentium.

Even it's just as the basis for indie game makers, I can live with that. Python for instance is probably one if not the most popular scripting languages in games, commercial or not. LibSDL is a very popular API used by many indie gamers. Runs on directx in windows, opengl in Linux. Much easier to use then either and games are much easier to port.

X.org, distributors, and proprietary modules

Posted Aug 16, 2006 20:34 UTC (Wed) by piman (guest, #8957) [Link] (5 responses)

> GPL'd Quake3 not good enough for you?

No, it's not. Quake 3 is seven years old. The proprietary gaming industry has surpassed almost every bit of technology in it, from AI to graphics to networking.

> I remember back when Quake2 was popular it was easy to make models for it.

A model is not a game.

> Ogr3d and CrystalSpace as well as a couple other game engines..

A 3D engine is not a game.

> Lots of programmers would love to make decent games.

But they can't. From the software end alone: We have reasonably good 3D engines, and now we're getting okay modelling tools (but most game development studios have customized tools). Where's the generic AI libraries, or network prediction? Things like MMOs, with gigabytes of data and beefy server requirements, are totally beyond us right now. Things like Xbox Live are years away (if anyone's even working on them).

Writing games is hard. It requires a unified vision between everyone involved, it requires varied skill sets in writing, programming, and art, and then also game design is a field on its own. Free games are lagging behind proprietary games by years. And as long as we're basing new games on the Doom (1!) engine, or even the Quake 3 engine, they're going to remain laughably behind. Look at the press around Wesnoth; it's a best-of-breed free game, and it really is amazing what they've achieved. It's also not going to impress gamers unless they've been asleep for a decade.

> You need to make it easy for artists and sound people who want to play around to make stuff and they will. It happens all the time.

This might be the biggest misconception around free gaming. If you have enough programmers pounding out code for what they need, yes, you'll get lots of useful things. But if you want something cohesive, like a game, you can't rely on that. You need programmers talking to artists talking to designers, and you need to keep this up for the whole project. You won't get an excellent game by throwing random art, music, and code into a blender and looking at what comes out.

Besides that, unlike a word processor, where you're trying to accomplish a task and can use different software to get there, a game is a world unto itself. If your friends are playing Doom 3, and you want to play with them, it doesn't matter how good your free game is.

Quality free games aren't impossible. But they're very, very hard, and they need to develop their own communities.

(Or maybe you're speaking from experience, and I'm just ignorant of some segment of free software gaming. What have you worked on?)

Where have we heard this before?

Posted Aug 19, 2006 5:44 UTC (Sat) by GreyWizard (guest, #1026) [Link] (4 responses)

We're lucky compilers and kernels don't need a unified vision and various skill sets. Such things are easy to make compared to proprietary games that never crash, scale from wrist watches to supercomputers and protect sensitive data from determined attackers. Word processors and web browsers are never at disadvantage when most peole are using a different product since proprietary formats and incomplete support for standards is impossible. Programmers working in their spare time could never be motivated to make games, which most of us find boring. Plus games need artwork and music which no one would create without compensation. We all know how easy it is for talented artists and musicians to find work.

Where have we heard this before?

Back in the early 1990's the conventional wisdom said the Linux kernel was a pale, puny toy compared to giants like AIX, HPUX and Solaris. Maybe it was a fun hobby, but it could never catch up to legions of professional developers since no one could ever get paid to work on free software. Several giant corpses later, the last proprietary Unix standing has itself turned to free software to compete for developers. Usable desktop environments were supposed to be beyond us too, until we made some. For a while pundits couldn't stop talking about how we could never catch up to proprietary desktops, but they've stopped now. We have working demonstrations of impressive three dimensional features proprietary systems can't yet match.

And so on and so forth. Be patient. We'll get around to remaking the game industry in due time.

Where have we heard this before?

Posted Aug 19, 2006 6:15 UTC (Sat) by piman (guest, #8957) [Link] (3 responses)

> We're lucky compilers and kernels don't need a unified vision and various skill sets.

Compared to games, they do not. A mediocre C compiler is suitable for a year-long project in a university CS program, for one or two students. A mediocre game is a year-long project for one or two programmers, one or two artists, a composer, and a game designer (who may or may not be one of the programmers). A professional game is the product of 20-50 people working full-time for 1-2 years.

The rest of your first paragraph is a strawman. It's not hard to find half a dozen artists who want to work on a game. It is hard to find half a dozen artists who want to work on the same game, and to enforce a consistent style between them when it's all volunteer work. (Try enforcing a consistent coding style, then consider if you had a team of programmers whose entire task was to work on coding style.) You'll see this problem across most free games.

> We have working demonstrations of impressive three dimensional features proprietary systems can't yet match.

To be really blunt, no. We have working demonstrations of impresive 3D features that are *on par* with proprietary systems. And we only get them on most systems by running proprietary drivers.

> Be patient. We'll get around to remaking the game industry in due time.

I didn't say it was impossible. I said you're not going to get there using a seven year old engine, or by ignoring the most popular kinds of games.

Moping Doesn't Lead to Games

Posted Aug 19, 2006 17:12 UTC (Sat) by GreyWizard (guest, #1026) [Link] (2 responses)

Comparing a C compiler developed by university undergraduates in a year to a mediocre professional game is like claiming game development is easy because one person can write a graphical chess game in six months while making a C compiler that can translate five programming languages, target a three platforms and optimize half as well as GCC takes more time. Efforts to build games and compilers come in a variety of shapes and sizes, so considering what it takes to be competitive is the only way to do an apples to apples comparison.

A competitive game might require two years of work from fifty people, but a competitive compiler or kernel is a ten year effort requiring hundreds of experts in dozens of specialized domains. No game ever created has enjoyed as much effort from such diverse contributors as Linux, GCC and other large free software projects. Do you suppose it's easy to reach consensus among even half a dozen brilliant egomaniacs working on a web browser, or to attract talent and enforce a consistent visual style in a desktop environment? Think again.

Proprietary games achieve what they do because an industry has formed around effective economic and social models. We in the free software community have found similar models compatible with our values for kernels, web servers, databases, browsers, desktop environments and more. We'll do the same for games. That's why engaging free software games that not as good as the best proprietary games are nothing to worry about. Like C compilers created in university courses, these projects are opportunities to learn the craft as well as find better ways to organize effort.

> We have working demonstrations of impresive 3D
> features that are *on par* with proprietary systems.

Which proprietary systems are on par? I suspect we'll have to agree to disagree on that, but five years ago you would have been moaning about how far behind free desktop systems are. I'm glad the developers who actually caught up ignored you.

> And we only get them on most systems by running proprietary drivers.

So? Graphics chip sets from Intel (which is committed to free software drivers) are readily available, as are R200 ATI cards for which free drivers work well. Reverse engineering efforts for R300 and NVIDIA cards are underway and future generations of CPUs for which specifications have always been available seem likely to have superiour graphics processing built in. Proprietary drivers are a serious problem but this is one battle among many that the free software community is winning.

Are you going to pretend that reverse engineering graphics card interfaces is easier than making games too?

> I said you're not going to get there using a seven year old engine, or
> by ignoring the most popular kinds of games.

Accusing me of attacking straw man arguments works better when you don't indulge yourself. I don't see anyone arguing that we should ignore proprietary games or that at the moment the best of them are not more technically advanced and polished than the best free software games. But we have to start somewhere. Moping and exaggerating the difficulty of making games won't get us there either.

Moping Doesn't Lead to Games

Posted Aug 20, 2006 7:00 UTC (Sun) by piman (guest, #8957) [Link] (1 responses)

> five years ago you would have been moaning about how far behind free desktop systems are.

I've been running exclusively free desktops for the past five years (and I mean exclusively -- only recently did I install non-free firmware, still no Flash, Java, Real, etc). Guess what? I still moan. They're usable; that doesn't mean they're not behind, and certainly they were five years ago.

> Graphics chip sets from Intel (which is committed to free software drivers) are readily available

And useless for modern 3D games.

> But we have to start somewhere. Moping and exaggerating the difficulty of making games won't get us there either.

If you think I'm just moping, or I'm exaggerating, I'd suggest you take a look at http://pydance.org and http://www.sacredchao.net/~piman/angrydd. Now put your money where your mouth is: What experience do you have that justifies your opinion that free games are going to just magically get written without serious changes to how we approach the problem?

If you notice, I keep stating: Professional-quality free games are not impossible. However, they do not exist yet, and no one has any idea how to make them. The games industry has a decade head start on free gaming, and unlike Microsoft in 1995, they're not going to sit still for another decade.

Moping + Boasting != Argument

Posted Aug 25, 2006 4:39 UTC (Fri) by GreyWizard (guest, #1026) [Link]

> [Free desktops are] usable; that doesn't mean they're not behind, and
> certainly they were five years ago.

Be specific. What can proprietary desktops do that free software desktops can't? Don't point to a lack of proprietary content or drivers. Those are social and poltical problems, not technical ones.

> [Graphics chip sets from Intel are] useless for modern 3D games.

I mentioned this in the context of advanced three dimensional desktop features, for which such chip sets are effective. You have also failed to address the availability of free software drivers for R200 cards which will support many but not all modern games, the promise of more competitive products of this kind from Intel in the near future and the likely convergence of CPU and GPU hardware.

> If you think I'm just moping, or I'm exaggerating, I'd suggest you take a
> look at http://pydance.org and http://www.sacredchao.net/~piman/angrydd.

Proving that game development is hard takes more than pointing to the inadequacy of your own efforts. Since we are discussing the relative difficulty of kernel, compiler and game development you would at least need to point to both games and kernel or compiler features you've implemented to make this relevant. Even if you could do so, that would not excuse dismissing the points I've made without an argument, as you have done.

Argue on the merits and spare me the boasting.

> What experience do you have that justifies your opinion that free games
> are going to just magically get written without serious changes to how we
> approach the problem?

Free software games do get written and some are quite good. I assume you intended to refer to free software games that can compete with the best proprietary games on their own terms. In that case, I never claimed that will happen without changes to how we approach the problem. Read the post you responded to and you'll discover that I said exactly the opposite. Approaching problems in a variety of ways is something the free software community is especially good at.

As for experience that justifies my opinion, I've done elementary graphic design, written OpenGL applications and worked on Linux kernel network card drivers. The first requires pointing, clicking and patience. The second requires a reference book and a reasonable grasp of simple math. The third requires mastery of concurrency, data structures, detailed hardware architecture and then some. A game is more than the first two but a kernel is much more than the third.

> The games industry has a decade head start on free gaming, and unlike
> Microsoft in 1995, they're not going to sit still for another decade.

They're not? That seems to be at odds with prevailing wisdom around the game industry these days. All the serious gamers I talk to say things like: "I think the industry is stuck in a loop. The same old crap keeps coming out." There are exceptions and interesting arguments against this as well, but it's far from clear that the proprietary games industry will be difficult to catch.

We don't disagree about whether current free software games can compete with the best proprietary games on their own terms today: they can't. Our difference seems to be in how we interpret that. Where you see overwhelming obstacles I see opportunity.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 14:18 UTC (Tue) by wilck (guest, #29844) [Link] (5 responses)

mmorpg, racing games, flight simulations, lots of FPS, stragety games, etc etc.

Show that list to a 16-year-old gamer from your neighborhood. I am sure she'll be excited about it... for about 10 minutes.

People want to play the games that are hyped in the press, they want first-class graphics, sound, and gameplay. Kids want the same games that their friends play. TuxRacer et al., nice as they may be, don't qualify.

TransGaming is bringing part of that universe to to the Linux world. But of course, you get the necessary FPS only with NVidia or ATI hardware. Sigh.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 20:17 UTC (Tue) by drag (guest, #31333) [Link] (2 responses)

Like I said before,

If your looking for commercial games support to drive Linux adoption rates your going to be sadly dissapointed.

It WILL NEVER HAPPEN.

You know why? Becuase Linux is Free software.
Who are the big game developers nowadays?
EA? Microsoft? Valve?

I mean seriously. These guys hate linux. They are so pro-closed source, pro-propriatory that's it's not even funny. They will never volentarially support Linux. Not even if Linux had 10 percent of the game market. Not even if there was a demand for it.

If Linux succeeds they loose a lot. They like the status quo, they like the control they have over the market, over the platform, over the hardware. They don't want it to change. They are making money hand over fist and they don't want to risk it.

Would you?

If even partially the goal of people like FSF or GNU are relalised they they would probably face huge financial losses. You'd have better luck trying to get SCO to open source OpenUnix.

Beleive it or not the #1 or #2 people own computers at home is to play video games.

Sure young people with free time are targetted by big game makers, but only becuase young people are easy to manipulate. Easy to sell to.

However the vast majority of people who own a computer don't look at computer gaming magazines. They don't go out and by a fancy gaming video card. They just want to have fun after work for a hour or so, that's all.

Games never support Linux?

Posted Aug 16, 2006 1:54 UTC (Wed) by nicku (subscriber, #777) [Link] (1 responses)

It WILL NEVER HAPPEN.

You know why? Becuase Linux is Free software. Who are the big game developers nowadays? EA? Microsoft? Valve?

I mean seriously. These guys hate linux. They are so pro-closed source, pro-propriatory that's it's not even funny. They will never volentarially support Linux. Not even if Linux had 10 percent of the game market. Not even if there was a demand for it.

These companies are less driven by ideology than by the prospect of making money. Ideology affects some of their managers, but the managers who can demonstrate profit will tend to win out.

Games never support Linux?

Posted Aug 16, 2006 5:32 UTC (Wed) by drag (guest, #31333) [Link]

Their 'idealogy' is "I make a game, I restrict access, I sell access, I make lots of money'. That's pretty much oppisite of the entire Linux ethos.

How are you going to restrict access to a game? How are you going to impliment DRM and cdrom checks when the owner of the computer controls all of it down to the kernel?

You think the kernel developers are going to help StarForce (http://www.star-force.com/)impliment it's "Advanced anti-piracy solutions for software distributed on CD/DVD-ROM and CD-R, as well as via the Internet; license management and DRM technologies."

THAT'S their Idealogy. They are pro-propriatory, drm restrictions, right managements, anti-piracy. That sort of thing isn't going to realy work out well on Linux. However if we begin to support binary only drivers and support DRM for Linux then they will probably start to seriously considure support Linux.

Otherwise for somebody like EA or Microsoft to start distributing software and games and such for Linux it would require somebody to start a gaming company that leverages the Linux platform so successfully that it starts to threaten their business model. Then we will start to see some action.

Just like in everything else Linux has been successfull with.

Get back to me in about 20 years or so.

X.org, distributors, and proprietary modules

Posted Aug 17, 2006 18:12 UTC (Thu) by grahammm (guest, #773) [Link] (1 responses)

Yet these games have much more 'lasting' appeal than the modern 3d-graphic, surround sound games. How many of the "current" games will still be available and being played in 30 years time? Most of the text-only games I encountered at university 30 years ago are still available and being played now.

FPS Games in the Future

Posted Aug 17, 2006 22:13 UTC (Thu) by zlynx (guest, #2285) [Link]

People still play Doom and Quake. when those games were released, they were the ultimate in 3D FPS graphics. It doesn't hurt them that id Software released them to open source.

It was just a year or two ago that some friends and I set up a Quakeworld LAN game. That was barrels of fun. With monkeys.

I have no doubt that today's FPS games will still be played by someone in the future.

Game incompatibility myth

Posted Aug 15, 2006 14:47 UTC (Tue) by broom (guest, #2914) [Link]

I bought a number of Loki games several years ago, when I had a much earlier incarnation of my home PC. I can't even remember exactly what the PC was back then (32 bit, x86 of one kind or another).

Since then, I've changed motherboards, CPU (to Athlon 64 in 64-bit mode), memory, graphics card, linux distribution, hard drives, cdrom drives, and even the case. It does, however, still have the original floppy drive, bought in 1995!

The Loki games I have all still work at least as well as they did back then!

Linux is a fine platform for proprietary games. To a large extent, this was due to Loki creating and releasing a lot of open source support, while keeping only the games themselves proprietary.

Proprietary software & vital work

Posted Aug 15, 2006 18:36 UTC (Tue) by Max.Hyre (subscriber, #1054) [Link]

[...] I'd rather that the scientists and doctors who are using high-end hardware requiring proprietary drivers be able to do their work than allow people to die [....]
A couple of lessons on the hazards of proprietary software [LWN subscriber-only content] See paragraph 5.

X.org, distributors, and proprietary modules

Posted Aug 17, 2006 19:02 UTC (Thu) by lysse (guest, #3190) [Link]

> You just have to do things that people from the 'best tool for the job' crowd thinks is completely irrational; which is to be willing to sacrifice for your ideals.

Presumably being willing to die for one's ideals is also completely irrational, yet there have been plenty of people who have effected lasting change in the world by doing just that. Given that, I think people could perhaps bear to live without the latest shiniest video card.

X.org, distributors, and proprietary modules

Posted Aug 14, 2006 22:20 UTC (Mon) by grouch (guest, #27289) [Link] (8 responses)

When will these distributors provide proper names for the distributions, such as nVidia's Fedora Core and ATI's Gentoo? This would at least alert users as to who controls the distributions. Users shouldn't be left with the mistaken impression that these distributions are independent of those video card makers.

OK if it's in FC6

Posted Aug 14, 2006 22:48 UTC (Mon) by JoeBuck (guest, #2330) [Link] (1 responses)

I think it could be argued that a major change to the X version shouldn't happen without a distro upgrade in any case, and FC6 will be out soon. Choosing not to upgrade X in FC5 is reasonable even in the absence of the proprietary-graphics issue.

That will give the proprietary module people a bit more time to get their act together and port their modules.

As long as the Fedora folks proceed with getting the new Xorg into FC6 I think it's OK.

OK if it's in FC6

Posted Aug 14, 2006 23:16 UTC (Mon) by spot (guest, #15640) [Link]

Indeed, Xorg 7.1 is in development and on track for FC-6.

X.org, distributors, and proprietary modules

Posted Aug 14, 2006 23:02 UTC (Mon) by tetromino (guest, #33846) [Link]

I suppose I can understand why nvidia hasn't updated their drivers. They are about to release a new generation of hardware (geforce 8) and probably want to put all their driver updates, both for new video cards and for xorg 7.1, into a single release. But it sucks for the users when a nvidia's timetable and the free desktop development timetable are so out of sync...

Right now, nvidia's proprietary driver is essentially the only choice for the (very significant number of) people who use an nvidia card with linux. The free nv driver is broken and useless. Hopefully the nouveau project will produce a working driver soon, so that desktop distros wouldn't be so dependent on a proprietary driver.

X.org, distributors, and proprietary modules

Posted Aug 14, 2006 23:03 UTC (Mon) by drag (guest, #31333) [Link] (3 responses)

That's the trouble.

It's not the distros are the ones being controlled. It's that the distros are desigined to be used by end users and the end users are the ones being controlled by propriatory stuff.

What maybe distros should do is to provide certification and recommended hardware for their system.

Linux is at the point now were people WANT to run Linux and they will modify their purchasing decisions based on Linux compatability. But most people aren't going to understand binary vs open source drivers and don't know which hardware vendor does this or does that.

Or to put it another way.. Even though Distros are mainly Free software this Free software is more valuable to people then the computer it runs on.

But people have limited resources and money and therefore if their computer can't run a distro then they simply won't be able to run that distro.

So what people like FC and Ubuntu should do is have recommended hardware and certified hardware system.

If I am fairly clueless user and want to run FC I would like to be able to go to FC and find out:
what proccessor I should buy
what wifi card I should buy (this is a BIG one)
what printer I should buy
what scanner should I buy
what motherboard and/or chipset should I buy
what video card should I buy
and maybe were I should buy it from.

This would make my purchasing decisions easier.

This is different from a 'compatability' list.. this is recommended hardware list. This is the 'ideal' hardware for best compatability and best support with your distro of choice.

A couple of examples are:
the KRP, which is favored hardware for building a HDTV Linux PVR. Using this certification system I should be able to go out and buy a computer, plug Knoppmyth into it and have a working HD-ready PVR in under 20 minutes. Full installation end-to-end.
http://www.mysettopbox.tv/KRP.html

Gamix is designed to create a standardized platform you could build or buy for playing games. The idea is that game developers could build gaming DVDs for this platform and it should work universally well on any other PC somebody builds with certified componates. The game developer has full control with the entire software stack. Obviously targetted for Linux support, but not nessicarially.
http://www.gamix.com/

Try to use the cheapness, flexibility, and performance of PC commodity hardware without the hassle of the end user dealing with driver issues or Windows.

This is different from 'compatability lists' because recommended hardware lists are much more focused, much more attainable.

With Windows you have the 'recommends windows xp' stuff along with their windows driver criteria stuff. With OS X you have Apple controlling the hardware completely and thus everything 'just works'. Linux can 'just work' for people also, but you need to have good hardware support.

If the recommended hardware for Ubuntu or FC or whatever works out then you can start to define a open common hardware criteria system. Much like how the LSB is for software compatability you can determine a LHB for a Linux hardware compatability.

X.org, distributors, and proprietary modules

Posted Aug 14, 2006 23:18 UTC (Mon) by drag (guest, #31333) [Link] (1 responses)

Oh and a couple others that I just remembed..

Redhat's HCL list
https://hardware.redhat.com/hwcert/index.cgi

D. J. Berstein's 20060107 Unix workstation hardware and build guide
http://cr.yp.to/hardware/build-20060107.html

Linuxprinting.com suggested printers
http://www.linuxprinting.org/suggested.html

Arstechnica June 2006 hardware buyers guide (windows oriented)
http://arstechnica.com/guides/buyer/system-guide-200606.ars

So on and so forth.

Most people just want stuff they don't have to mess with. A guide or recommended/suggested hardware can do this for them.

X.org, distributors, and proprietary modules

Posted Aug 17, 2006 20:18 UTC (Thu) by bronson (guest, #4806) [Link]

Too bad djb went with nvidia.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 2:34 UTC (Tue) by dberkholz (guest, #23346) [Link]

> It's not the distros are the ones being controlled. It's that the distros
> are desigined to be used by end users and the end users are the ones being
> controlled by propriatory stuff.

Exactly. Distributions exist to make the end user's task of maintaining a full system easier. That's a core part of the Gentoo philosophy, at least, and it doesn't allow us to take stances like "We're breaking everyone who uses binary drivers."

X.org, distributors, and proprietary modules

Posted Aug 14, 2006 23:26 UTC (Mon) by bojan (subscriber, #14302) [Link]

> nVidia's Fedora Core

Come on, it's far from that. Actually, if you wish, you can run X11 7.1 with AIGLX enabled on FC5 now, without compiling anything:

http://download.fedora.redhat.com/pub/fedora/projects/aiglx/
http://fedoraproject.org/wiki/RenderingProject/aiglx
http://fedoraproject.org/wiki/RenderingProject/AiglxOnFedora

And, all this will be available by default in FC6, which is probably going to be released mid-October:

http://fedoraproject.org/wiki/Core/Schedule

nVidia engineers submitted the change to X.org 7.1

Posted Aug 14, 2006 23:49 UTC (Mon) by dlang (guest, #313) [Link] (1 responses)

it's interesting that the nvidia engineers were the ones who submitted the changes to X.org 7.1 that break their drivers and they are the last ones who haven't released a compatable driver.

I'm sure teh conspiracy folks could have a field day with this one.

nVidia engineers submitted the change to X.org 7.1

Posted Aug 15, 2006 3:12 UTC (Tue) by elanthis (guest, #6227) [Link]

The submitted one of the changes, but not all of them. There are several different things which changed ABI in X.org 7.1. NVIDIA was responsible for the Xv changes (which were necessary in order to do hardware Xv acceleration with compositing), but not the other ABI changes (like fonts).

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 2:16 UTC (Tue) by mrshiny (guest, #4266) [Link] (18 responses)

Well, as much as it pains me to admit it, I am one of the nVidia users who would be in a very bad spot if Fedora Core 5 upgraded X and there wasn't a nVidia driver available. The sad reality is that there is no open source driver that meets my needs. If there were, I'd use it, but it JUST DOESN'T EXIST. And considering the situation with FC5 when it was released (no binary module support), it'd be silly for Fedora to break binary graphics drivers now.

As it stands, I am going to have to wait until a NVidia driver is available for my linux distro. The free NV driver is unacceptable slow (if it even works on my current card... haven't tried it in years). If FC6 ships with X.org 7.1 (which I think it should) and nVidia doesn't have a driver, then I will have to wait until that driver arrives before installing FC6. If nVidia is late releasing the driver, I will be mad at them and consider switching cards, as I did with ATI when I had driver issues with them. But if Fedora breaks my working system, it won't be nVidia that I'll blame; it'll be Fedora.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 6:13 UTC (Tue) by arjan (subscriber, #36785) [Link] (5 responses)

This entire saga is a balance between what you, as non-free linux user (you are not using an open source OS, see Dave Airlied's OLS talk) want and what the people who bought an open source supported card need. The later is of course all Intel graphics (40% market share) and a chunk of ATI users (R300), both of which have drivers that made a huge leap forward in X 7.1, adding lots of hardware support.

It's not a clear case that your position is worth more than the other users position...

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 7:50 UTC (Tue) by warmcat1 (guest, #31975) [Link] (4 responses)

> It's not a clear case that your position
> is worth more than the other users position...

It's not clear, but it is pretty clear a chunk of the Fedora userbase cliff face would crack off and fall into the sea over having their working desktops broken, tainted or not. It's legitimate to shrug one's shoulders over that and go on anyway, but obviously after a few such decisions Fedora as a project will be looking at an eroded and calcified userbase with reduced new user blood coming in because of the word of mouth. Fedora can even shrug its shoulders at that situation, since it has the bubbling wellspring of Redhat money underpinning it so doesn't have to care if all around is a desert.

Binary drivers are bad news, no doubt, and keeping them at a distance and in the cold is good, but right now they bridge the gap between what some users must have and what Fedora can provide.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 10:27 UTC (Tue) by k8to (guest, #15413) [Link] (1 responses)

I don't disagree that a distribution must consider usability as well as
freedom, and that there is room for different choices. However, Fedora
has made this choice. Fedora _does not_ support binary drivers. This
bridge was already crossed.

It seems the discussion found some other issues which were important as
well, and so have decided to keep with 7.0 on the balance.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 10:44 UTC (Tue) by warmcat1 (guest, #31975) [Link]

Yes Fedora has always had that stance, I think it is a good stance.

However many Fedora users for whatever reason ARE using binary drivers of one flavour or another, it's a fact. Therefore there are ramifications to 4KSTACKS or xorg 7.1 or whatever the next problem will be that shows the fissure between the open codebase that can move with it en bloc and the closed codebase that is fragile.

It is a political decision how to balance those needs, the damage to binary driver users got plenty of airing on fedora-devel and I am sure it was part of the considerations in not pushing all updating FC5 users on to xorg 7.1. And that's probably a good thing, some realpolitik, because this running sore will be with us for a long time AFAICT.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 10:58 UTC (Tue) by arjan (subscriber, #36785) [Link] (1 responses)

there is another angle as well; and that is the chicken-and-egg one. NVidia won't release a 7.1 driver until a major distro uses 7.1. And if the major distros won't go to 7.1 until nvidia supports it... nothing ever changes and progress is harmed.

Progress is harmed for the majority of users; last market share numbers that were on theregister.net were something like "intel 40%, nvidia 20% ati 20%". Both Intel and ATI users gain from 7.1 with better suppoprt for their hardware, much better support in fact.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 11:14 UTC (Tue) by warmcat1 (guest, #31975) [Link]

Yes, fair enough. But in this particular situation it is not the stasis of "nothing ever changes": nobody argues against xorg 7.1 in FC6. FC6 is very close, many (most?) FC5 people will be updating in the next month or two. Any delay in exposing xorg 7.1 to mass usage on Fedora is therefore very limited.

Stepping between FCn versions is in the hands of the user and he knows he can expect strong differences (eg, FC1 -> FC2 == 2.4 -> 2.6 ). Such a user might receive the result of nVidia binary breakage with a shrug and use nv or vesa until the new binary release, because he knows he can expect 'excitement' of the good or the bad kind from updating to a new release. It seems a good place to bring in a large core component update regardless of the binary question. The fact that an FC5 user can actually use the development repo xorg packages to get the new stuff if they really needed it (as I do) makes it even less objectionable to the FC5 user.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 9:57 UTC (Tue) by russell (guest, #10458) [Link] (9 responses)

You have the option to stop upgrading, downgrade to FC4, or buy a new video card. It was your decision to buy Nvidia. You knew the score before you bought it.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 11:51 UTC (Tue) by smitty_one_each (subscriber, #28989) [Link]

Is downgrading really an option, when it could open up security holes?

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 13:39 UTC (Tue) by mrshiny (guest, #4266) [Link] (7 responses)

Come on, downgrade, or stop upgrading? Please. First of all, this opens me up to security vulnerabilities. Second of all, this means I have to know when an update is going to break my working system. Anyway, I can turn this around: anyone who wants xorg 7.1 right now can just update it manually.

The fact is: Fedora Core 5 has a certain set of features. Compatibility with the nVidia binary driver is one of those features, and it is an important enough feature that the first FC5 kernel update contained a fix that compatibility. An update to a distro (esp. one that is nearing the end of its life-cycle, since FC6 is nearly ready) should NOT remove features without a damn good reason. Breaking your users' systems is BAD.

You can argue that some users actually need the enhancements in 7.1. That's fine, but by NOT delivering 7.1 you're not breaking their working systems, you're simply not supporting their potentially working systems. And it's not as if you can't provide, via something like an optional yum repository, support for 7.1 for people who need it. But to force 7.1 down the throats of users who don't want it (yet)? That's a good way to alienate your userbase.

It's a pretty clear-cut case. Installing xorg 7.1 breaks some users. Not installing it doesn't break any users. FC6 will be out soon and will include xorg 7.1. Therefore, FC5 stays at 7.0.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 14:43 UTC (Tue) by vblum (guest, #1151) [Link] (6 responses)

The simple fact is: RH let themselves be forced to operate based on the whims of a closed-source vendor. This is an extremely bad situation.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 15:16 UTC (Tue) by mrshiny (guest, #4266) [Link]

Let's not get over-dramatic. Fedora decided not to break users' systems. They didn't decide to cancel upgrading to 7.1. They didn't decide that xorg 7.1 will never be in Fedora Core 5, ever. They just aren't shipping it now, but I bet that when the binary drivers are available it will be pushed out as an update. Users who want it now can get it. Users who don't want it have working systems. This is not a question of Fedora abandonning its principles. They aren't going to start shipping the binary drivers, or the mp3 codecs, or any of those other products that users arguably want but violate, in some way, the politics of Fedora/RedHat.

Consider this: pushing xorg 7.1 into yum's updates will force people to upgrade, since it's a cumbersome task to manage this update process and somehow exclude xorg. Furthermore it means that Fedora will stop supporting xorg 7.0 and this may leave non-upgrading users vulnerable to security problems. But by NOT pushing xorg everyone who has a working Fedora system gets to keep using it, and anyone who wants xorg 7.1 can get it somewhere else (like, FC6 or fedora development).

Frankly I'm glad that Fedora made the decision that causes the least harm to their users. And I think it's clear that this stance is the proper stance, no matter what the cause of the delay is, whether it's lack of a popular binary driver, or some other incompatibility that is outside of Fedora's control. This is what a responsible distro does: it packages quality software together, and coordinates and manages the releases to maximize the user's productivity and convenience. So I say Thank You to the Fedora developers.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 15:36 UTC (Tue) by wilck (guest, #29844) [Link] (3 responses)

...on the whims of a closed-source vendor

What a weird interpretation. Did NVidia command the Fedora board not to upgrade? You seem to think that NVidia is somehow acting against the community, or trying to force everyone to comply with their will. In fact they are trying to support Linux - in their own way, which is admittedly suboptimal.

I don't like the hostility in your statement (and others on this page). There are quite a few companies freedom-loving people can justifiably detest. But I don't think ATI and NVidia are among them.

The availability of a closed source driver harms free software development by attenuating the need for free ones - conceded. But that doesn't mean that NVidia (or ATI, for that matter) are enemies of Linux or Open Source. No more than any other company (say Google, say Sun, ...) which doesn't open up all their source code.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 17:27 UTC (Tue) by vblum (guest, #1151) [Link] (2 responses)

My intentions were not so hostile to NVidia. This is mere business logic. The question is who can apply pressure to whom.

RH decided not to upgrade a piece of critical infrastructure because a single(!) company failed to put out their closed source driver in time. From a consumer's perspective, this is not good.

I started out long ago with a similar sentiment like yours, that is, let's give NVidia a break and let's just accept that they're doing their Linux support in a different way. When Arjan van der Ven's (I think) piece on closed source drivers in the kernel came out, I thought it overdramatic at first.

However, it is now clear that any slack on the part of NVidia _is_ able to affect a distribution release adversely after all. Only a single vendor, mind you - but it can threaten RH with the anger of NVidia's users, anger that should be directed at NVidia.

What will happen if this starts affecting more than one driver, as A van der Ven theorized? Already with two perhaps conflicting vendor dependencies, we'll end up in a much more difficult situation ... do you chose X.org 7.5 which supports the latest card by Martian grafix, Inc - or do we ship X.org 7.3 instead, because the widely popular Obsolox Zillienium adapter is supported there, but not later?

Then think three conflicting vendors.

It may be possible to hold up everyone else's releases and follow only a single vendor's schedule, NVidia's. But if we allow one such case, there will be more. What then?

My take is: If vendors like NVidia insist on closed source, fine. But if they are then not fast enough to upgrade their drivers, this MUST be their problem, and that of their users. As it is now, suddenly the pressure is applied to everyone else to NOT upgrade, but not applied to NVidia to upgrade. How very wrong.

With this choice, we are moving towards the unmaintainability of the free software part that Arjan projected as his worst case scenario. That's a long way off - but he may have had a better point than I first thought.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 17:36 UTC (Tue) by mrshiny (guest, #4266) [Link] (1 responses)

I agree with most of what you're saying, except that the following distinction should be made: Fedora Core 5 is already shipped, installed, and stable. There is no compelling reason to break user functionality. I don't think anyone is saying that Fedora should delay including xorg 7.1 in FC6 if the nVidia driver isn't ready. At that point it will be nVidia's fault that their driver doesn't support a popular distribution. But at this point, with FC5, it is Fedora that would be breaking my working system, or, at best, forcing me to A) find out that a particular update will break my system, and B) figure out how to back out that update, and/or prevent it from applying, thus potentially leaving me open to other dependencies breaking and security vulnerabilities.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 19:54 UTC (Tue) by vblum (guest, #1151) [Link]

Point taken; although gentoo's default choice continues to puzzle me a bit. Anyway, may NVidia one day see the light from all this debate ...

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 20:33 UTC (Tue) by drag (guest, #31333) [Link]

"The simple fact is: RH let themselves be forced to operate based on the whims of a closed-source vendor. This is an extremely bad situation."

My understanding is that at least with Nvidia drivers Redhat will refuse to support driver issues with that thing running. It's been a source of quite a few nasty bugs in the past.

For other stuff.. like apps that run on Redhat. Ya certification of propriatory apps is one of their major selling points so they would have to conform to closed source vendors they work with.

X.org free driver support for recent ATI cards

Posted Aug 16, 2006 2:28 UTC (Wed) by nicku (subscriber, #777) [Link] (1 responses)

The sad reality is that there is no open source driver that meets my needs. If there were, I'd use it, but it JUST DOESN'T EXIST.

I use ATI Radeon 9250 cards on all our machines for which free 3D support is excellent, however these cards cannot plug into current motherboards.

At a recent SLUG presentation, an Xorg developer spoke of better support for some later ATI cards in Xorg. Is there free support for later ATI video cards (those that support “PCI Express” motherboards) in Xorg 7.1?

X.org free driver support for recent ATI cards

Posted Aug 16, 2006 21:42 UTC (Wed) by droberge (guest, #10852) [Link]

I recently purchased a PCI Express motherboard and a Radeon X800 GTO card. The 'radeon' driver in X.org 7.1, along with the DRI kernel drivers in 2.6.17 and the latest Mesa support this out of the box, although I haven't had it long enough to see if the 3D is truly stable. AFAICT the 2D portion of the card is the same as earlier cards so the 2D portion of the driver works just fine. The driver supports the higher 9000-series Radeons as well as the X-series (but not the X1000 series).

It was a little difficult to find information on the status of the driver, but the documentation in the DRI Wiki (http://dri.freedesktop.org/wiki/) is getting better every day.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 4:59 UTC (Tue) by kwink81 (guest, #33926) [Link] (7 responses)

I think we are looking at this from the wrong angle.

Begging Nvidia/ATI to give us open specs/documentation will work just about as well as begging Microsoft to release Word under the GPL. Ain't gonna happen.

In the graphics areana, we need to treat hardware like we treat software. just like we can replace closed software with open software, we need to replace closed hardware with open hardware.

To that end, I suggest anyone who cares take a look at The Open Graphics Project(http://wiki.duskglow.com/tiki-index.php?page=Open-Graphics). It is an attempt to create an open/fully-documented 3D graphics card. If we can get this off the ground, we can ignore Nvidia/ATI instead of being chained by them.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 6:24 UTC (Tue) by k8to (guest, #15413) [Link] (6 responses)

Not to be TOO much of a party pooper on this project, but I have been
eagerly anticipating it for around 2 years now. They are yet to make a
developer-only board available, and the project naming and discussion
seems moderate impenetrable. When I do think I understand it, it seems
they are getting bogged down in unimportant debates, as well as
hair-brained schemes like high end soundcards.

Let me know when significant progress is made, and/or when I can pay
money to acquire hardware that I can use to drive a computer monitor.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 14:58 UTC (Tue) by yodermk (subscriber, #3803) [Link] (1 responses)

I was under the impression that the developer board was Really Soon Now. And it looks cool, if you enjoy hacking that sort of thing.

I don't know how long it will take to make a graphics card after the dev board is out, but I'm guestimating around the beginning of 2008. If it happens by then I'll be happy.

Sucks to wait, but these projects take time. Until then there's Intel.

X.org, distributors, and proprietary modules

Posted Aug 17, 2006 11:23 UTC (Thu) by k8to (guest, #15413) [Link]

Originally I was hoping to buy a dev board, not because I am terribly
capable of significantly assisting, but because I wanted to financially
support the project and track the work in progress.

Unfortunately the original discussions of a possible devboard price were
not realistic, and 200-400 dollars ballooned into the current ~1000$
estimate.

I continue to hope for success.

X.org, distributors, and proprietary modules

Posted Aug 17, 2006 3:24 UTC (Thu) by rloomans (subscriber, #759) [Link] (3 responses)

"I have been eagerly anticipating it for around 2 years now."

It might seem like 2 years, but the project has only been running for about a year.....

X.org, distributors, and proprietary modules

Posted Aug 17, 2006 11:20 UTC (Thu) by k8to (guest, #15413) [Link] (2 responses)

http://marc.theaimsgroup.com/?l=linux-kernel&m=109831...

As you can see, the project is 22 months old. Sorry, I was off by two
months.

X.org, distributors, and proprietary modules

Posted Aug 17, 2006 20:24 UTC (Thu) by bronson (guest, #4806) [Link] (1 responses)

And have they been promising a devboard for that entire time?

X.org, distributors, and proprietary modules

Posted Aug 17, 2006 22:40 UTC (Thu) by k8to (guest, #15413) [Link]

I'm not sure what you mean exactly, but yes they have been trying to
produce hardware for that entire time.

Some of the vendors have already shifted to 7.1

Posted Aug 15, 2006 6:54 UTC (Tue) by lamikr (guest, #2289) [Link] (4 responses)

With binary only drivers it's always a run and catch competition.

Luckily all vendors has not hold back to 7.0 version. Mandriva for example has already moved to 7.1 some months ago in their cooker and will ship that with AIGLX in their next release. (Should be out pretty soon)

And if I have understood correctly, there are now finally available binary drivers at least from ATI to 7.1.

Some of the vendors have already shifted to 7.1

Posted Aug 15, 2006 13:50 UTC (Tue) by mrshiny (guest, #4266) [Link] (3 responses)

Fedora isn't "holding back" to 7.0. FC6 will ship 7.1, which, I think, everyone agrees is a good thing. What some people disagree with (myself included) is whether it's good to ship 7.1 to existing FC5 users. I'd say, no, it's not good, because that breaks some users' systems. That's bad software engineering, breaking a working system for no good reason. The users of FC5 who NEED 7.1 already DON'T have a working system, and so they can't say their situation is any worse. And they can always install 7.1 manually.

Some of the vendors have already shifted to 7.1

Posted Aug 17, 2006 5:15 UTC (Thu) by ronaldcole (guest, #1462) [Link] (2 responses)

It's FC5! Fedora breaks things like Promise FastTrack RAID in the kernel by going to 2.6.17 and doesn't bat an eye!! And since FC5 is the beta for RHEL5, I want 7.1 in FC5!!!

Some of the vendors have already shifted to 7.1

Posted Aug 17, 2006 15:38 UTC (Thu) by mrshiny (guest, #4266) [Link] (1 responses)

If Fedora broke Promise FastTrack RAID because of a kernel update, then I consider that a bug. However if the reason that broke is because the Promise card is a binary driver (I don't know the details) then Fedora is in a tight spot because the kernel devs refuse to provide even a stable API, let alone ABI, and there is no "stable" branch of the kernel that is seriously maintained.

The only "excuse" for breaking the driver is if the new driver is supposed to work but happens to have a bug. This is "normal" when software is updated, even if it isn't ideal. (I can't help but mention that THIS use-case can be avoided if drivers are shipped separately from the kernel).

Some of the vendors have already shifted to 7.1

Posted Aug 21, 2006 2:53 UTC (Mon) by k8to (guest, #15413) [Link]

Well the other reason to break fastrak raid support is that it's harmful. Slowing down disk access
while making youself beholden to proprietary firmware is just not a winning combination.

Backstory: promise makes "fake raid" controllers that do a poorer job than the linux builtin software
raid. They are notable for having many more bugs than linux softraid, being marketed
misleadingly, and damaging performance in most use cases.

Fedora

Posted Aug 15, 2006 8:58 UTC (Tue) by vblum (guest, #1151) [Link] (2 responses)

What a suicidal release policy. If RH ever had the market strength to force a proprietary vendor to upgrade, it has just seriously impaired that ability for the future. This is not even about free software or not; it's a simple case of a very bad business decision.

One can only hope (unrealistically?) that Novell (Opensuse) makes a better choice ...

Fedora

Posted Aug 15, 2006 19:39 UTC (Tue) by aj (subscriber, #39001) [Link] (1 responses)

> One can only hope (unrealistically?) that Novell (Opensuse) makes a better choice ...

openSUSE 10.2 Alpha3 has just switched from X.org 6.9 to 7.1:
http://www.novell.com/coolblogs/?p=451

Nice!

Posted Aug 15, 2006 19:50 UTC (Tue) by vblum (guest, #1151) [Link]

Very encouraging!

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 9:44 UTC (Tue) by russell (guest, #10458) [Link] (6 responses)

When I bought my laptop, I chose an Intel graphics solution because of their support for free drivers. Why didn't everyone else do the same? Looks like closed source drivers aren't the only problem. The people that buy that hardware are also a problem. Sorry if this ruffles your feathers, but why should people who see the value in freedom suffer from the choices of others.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 13:53 UTC (Tue) by smulcahy (guest, #2758) [Link] (4 responses)

I've read a lot in the last week about the triumph that is Intel open sourcing their drivers. This is great news and I'm happy about it. But the sad reality of the matter is that the performance of Intel graphics devices is abysmal for uses such as gaming. For gaming graphics hardware, it's a 2 horse race with Nvidia and ATI (I can dig out reviews and benchmarks if you like).

Ok, as someone has pointed out, if you want to play games, go buy a PC with an Nvidia or ATI card and go play your games under Windows. What about us guys that want to do both? Play some games in the evening on Windows and spend their day doing development in Linux. I know, we're horrible realists who are undermining all the good work of the FSF, but I'm guessing there are a sizable minority of people in this group - does the "free software community" really want to cut us loose?

I'd love to have a decent open source driver for my nvidia and ati cards (I straddle both evil camps!) but in the meantime the binary drivers actually work pretty well with the occasional configuration headache.

I guess what I'm really getting at here is that Intel is not an option for a lot of us, despite their laudable efforts to open source their drivers. I think we need to continue gently encouraging Nvidia and ATI to open source while recognising that there are valid business obstacles to them doing so - the best thing that can happen is that, over time, the number of Linux users running Nvidia and ATI increases to the point where they can no longer ignore us.

This is far less likely to happen if the community as a whole turns us out in the cold. I applaud Fedora and Gentoo for balancing the needs of their users against other priorities, I think they've made a reasonable decision.

/me steps into his asbestos undies

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 15:23 UTC (Tue) by khim (subscriber, #9252) [Link]

I know, we're horrible realists who are undermining all the good work of the FSF, but I'm guessing there are a sizable minority of people in this group - does the "free software community" really want to cut us loose?

Yes: UNIX was (and is) proprietary software, and the GNU project's philosophy said that we should not use proprietary software. But, applying the same reasoning that leads to the conclusion that violence in self defense is justified, I concluded that it was legitimate to use a proprietary package when that was crucial for developing a free replacement that would help others stop using the proprietary package.

Sometimes it's Ok to use proproetary software. If you have to. If people lives depend from your work. Or if you are working on free drivers. But if games is the motivation behind NVidia choice - then you deserve breakage. After all you can start X.org 7.1 with old NVidia drivers if you'll use -ignoreABI option - and if sub-pixel rendering options makes a mess out of your desktop... you knew what you've bought...

X.org, distributors, and proprietary modules

Posted Aug 16, 2006 10:08 UTC (Wed) by russell (guest, #10458) [Link] (2 responses)

If your playing games on windows, then what do you use the 3d aceleration for under Linux? Can you get by with the nv driver?

X.org, distributors, and proprietary modules

Posted Aug 16, 2006 10:14 UTC (Wed) by smulcahy (guest, #2758) [Link] (1 responses)

Hi,

Yes, I don't have any compelling need for 3d under Linux other than for eye candy. The last time I checked the nv driver worked ok apart from 3d. I have an ATI X700 Pro in my desktop at the moment though and that does not seem to work with the free ati or radeon drivers (apologies for my ignorance, I'm no expert on the ati drivers but I do recall trying both before downloading the binary driver - which has consistently suprised me with how well it installs on versions of debian testing).

-stephen

X.org, distributors, and proprietary modules

Posted Aug 18, 2006 3:04 UTC (Fri) by roelofs (guest, #2599) [Link]

> I have an ATI X700 Pro in my desktop at the moment though and that does not
> seem to work with the free ati or radeon drivers...

Unless that's one of the cards with no 2D core at all (and possibly even if it is), you should be able to use the VESA driver, right? Granted, it's not fast, but I do almost nothing but development, and there's virtually nothing (save only 2D scroll rate, perhaps) that would actually be significantly better with acceleration. Of course, that does assume the VESA modes suffice to drive the monitor at its/your preferred resolution. ;-)

Greg

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 15:05 UTC (Tue) by wilck (guest, #29844) [Link]

Your argument appeals to nobody but die-hard Linux fans. Normal users will ask their local PC dealer, who will tell them that NVidia/ATI's the best, and that NVidia/ATI "support Linux". They will find similar information on the Web, too (not on LWN, of course). That's why "everyone else didn't do the same" as you did.

I find it understandable that GPL compliance of drivers is not the first priority criterion for some people when buying hardware. Even for those who know about the problems with proprietary video drivers, their FPS performance may be tempting.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 14:49 UTC (Tue) by rfunk (subscriber, #4054) [Link]

I'm a fan of neither Fedora, NVidia, or binary drivers (I'm a
Debian/Ubuntu fan), but I tend to think that this is actually the right
decision for FC5, as long as FC6 is not held back on 7.0 too. It's
important not to break working systems with a "minor" update.

(Though my opinion may be somewhat colored by my recent experiences
trying (and failing) to get a user onto a current distribution despite
her insistence on continuing to use WordPerfect. At least there I know
that as long as I stick with the old distribution WP will still work.)

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 15:04 UTC (Tue) by mrfredsmoothie (guest, #3100) [Link] (1 responses)

What is rather less encouraging is that the best interest of (at least) ... Gentoo users is in the hands of proprietary module vendors, and that this dependency is imposing a cost on all users, whether they use the modules in question or not.
Pretty small cost, though: 'echo ">=xorg-x11-7.1 ~x86" >> /etc/portage/package.keywords', right?

X.org, distributors, and proprietary modules

Posted Aug 16, 2006 0:41 UTC (Wed) by dlang (guest, #313) [Link]

not quite
echo ">=x11-base/xorg-x11-7.1 ~x86" >> /etc/portage/package.keywords

Xorg 7.1 is in FC6, where it belongs

Posted Aug 15, 2006 15:15 UTC (Tue) by rriggs (guest, #11598) [Link] (5 responses)

Xorg 7.1, with it's ABI changes, has no business going into Fedora Core 5.

FC5, at this point in its lifecycle, is considered by many to be stable. Xorg 7.0 was the most recent version when FC5 started out. Xorg 7.1 does belong in FC6, which is exactly where it will be introduced.

The folks managing Fedora Core are doing a fine job IMO. There is a balancing act to do, and preventing breakage in an existing distribution has to be a primary consideration.

I am a Fedora Core user and I expect breakage when new FC versions are rolled out. I had no problem when 4KSTACKS broke the NVidia drivers I rely on because that happened in a new FC version. But I would have been severely disappointed if they chose to roll out a 4KSTACKS kernel to a stable FC version before the bugs had been ironed out. I am a long-time Red Hat customer and Fedora Core user, but that sort of instability would have me running somewhere else fast.

I can agree that binary drivers do cause issues that a purely open-source solution would not have to face. And I would very much prefer that ATI and NVidia open source their drivers. But to imply that distributors and their users are being held hostage by these drivers is a little over the top. It's merely one consideration in the decision process regarding what packages to include in upgrades.

Put another way: I made the decision to purchase NVidia hardware (again, as recently as last week) because, in the end, it is in my best interest to have excellent 3D hardware support at the expense that proprietary drivers exact. So, if people want to blame someone for this, they can blame me and others like me who make that choice.

Xorg 7.1 is in FC6, where it belongs

Posted Aug 15, 2006 15:47 UTC (Tue) by smulcahy (guest, #2758) [Link] (4 responses)

How else can we show off the goodness of xscreensaver? :)

Xorg 7.1 is in FC6, where it belongs

Posted Aug 16, 2006 2:46 UTC (Wed) by nicku (subscriber, #777) [Link] (3 responses)

How else can we show off the goodness of xscreensaver? :)

Your joke depends on the view that besides games and screensavers, there is no compelling need for 3D drivers. However, if you are interested in astronomy, then Celestia and Stellarium benefit greatly from the excellent 3D support by the free Xorg drivers for my old ATI 9250 AGP graphics cards.

(Try these wonderful software systems even if you weren't previously interested in astronomy :-)

Xorg 7.1 is in FC6, where it belongs

Posted Aug 16, 2006 10:17 UTC (Wed) by smulcahy (guest, #2758) [Link]

Hi,

I actually have come across these and would agree - they are beautiful examples of how 3d can be used other than for eye candy and gaming. Thanks for the pointer though.

-stephen

Xorg 7.1 is in FC6, where it belongs

Posted Aug 16, 2006 10:33 UTC (Wed) by danielpf (guest, #4723) [Link] (1 responses)

Celestia is indeed an excellent example of what can bring could 3D
graphics to non-game applications. There is a huge potential
also in other domains, like medical imaging, or more generally
complex multi-dimensional data rendering.

Xorg 7.1 is in FC6, where it belongs

Posted Aug 17, 2006 11:39 UTC (Thu) by wookey (guest, #5501) [Link]

Indeed. Geodata is another area where 3D can make huge difference.

For example the Survex cave survey visualisation tool benefits enormously from improved OpenGL 3D performance. It's certainly the only reason I ever expect to take any interest whatsoever in 3D graphics performance.

X.org, distributors, and proprietary modules

Posted Aug 15, 2006 22:46 UTC (Tue) by brouhaha (subscriber, #1698) [Link] (2 responses)

Mike Harris is quoted as having written:
Fedora does not support proprietary drivers at all, and never has, nor has any Red Hat OS that preceded it.
Factually incorrect. Various early retail-box Red Hat Linux releases in the 4.x/5.x timeframe actually did include some proprietary software, including a proprietary X server with integrated drivers for hardware that was not supported by XFree86.

X.org, distributors, and proprietary modules

Posted Aug 17, 2006 8:38 UTC (Thu) by tialaramex (subscriber, #21167) [Link]

Red Hat have bundled game demos, time limited versions of backup and RDBMS software and lots of other things in the N+1th CD of their boxed Linux distribution, but they don't /support/ any of that stuff.

Are you sure that the X server you're thinking of isn't in the same category? ie "Here's some stuff which runs on Linux, but is not part of Red Hat Linux"

Of course it might be that Mike means specifically kernel drivers...

X.org, distributors, and proprietary modules

Posted Aug 17, 2006 11:32 UTC (Thu) by k8to (guest, #15413) [Link]

As said by tialrmrmxmrmx, Red Hat never considered their extra CD to be
part of their distribution proper. I have always considered this
something of of a weak argument, myself. However, they have steadfastly
worked at getting rid of the need for nearly every category of such
proprietary software, so I cannot fault their choices too much.

X.org, distributors, and proprietary modules

Posted Aug 17, 2006 19:12 UTC (Thu) by lysse (guest, #3190) [Link] (2 responses)

I wonder if those who do would be so keen to defend the right of vendors to release binary drivers, or insist that the Linux / X11 / Fedora developers should cater to their needs, if they placed a price tag - say $50 - on those drivers? Let's face it, nothing's stopping them; it could happen tomorrow, it wouldn't be at all immoral, and as the binary driver advocates never cease to point out, the Linux user base is far too small for such a move to have an adverse effect on the company concerned.

X.org, distributors, and proprietary modules

Posted Aug 17, 2006 21:29 UTC (Thu) by ncm (guest, #165) [Link] (1 responses)

If they could get the $50 without commercial repercussions, they would be obliged to charge it. Otherwise they would be subject to shareholder lawsuits for failing to extract potential revenue. (Apple was forced to charge for OS upgrades, for that reason.) But I'm not sure what your point is.

X.org, distributors, and proprietary modules

Posted Aug 17, 2006 22:38 UTC (Thu) by k8to (guest, #15413) [Link]

If Nvidia would sell me a license to full specifications for their
hardware for 50 dollars, I would be willing to buy that! If Nvidia was
willing to sell me the source to a Free Software driver codebase for
their hardware for 50 dollars, I would buy that!

X.org, distributors, and proprietary modules

Posted Aug 26, 2006 12:31 UTC (Sat) by zboszor (guest, #40152) [Link]

Let's put the heated debate aside about X.Org 7.1's temporary incompatibility
with proprietary 3D drives.

"One of the developments merged into 7.1 was the AIGLX project,
dedicated to the important goal of providing better eye candy for
Linux users worldwide."

This sentence is not the whole truth. AIGLX is Accelerated Indirect GLX,
and that means a lot more than just to provide eye candy.
Just think about heavyweight 3D applications, not just games, but CAD/CAM
running on a THIN CLIENT with X.Org 7.1.

It is another disruptive technology that only open source
can make possible. An engineering office with many expensive
workstations can be a whole lot cheaper...

Posted Aug 26, 2006 23:16 UTC (Sat) by ajaxx (guest, #35881) [Link] (2 responses)

So, as the current Fedora X maintainer, I'd like to clarify a few points.

The Fedora advisory board met to discuss the issue; the resulting decision was that Fedora Core 5 would not be updated to X.org 7.1.

No, that decision was made before the board's meeting. The X team basically decided that a complete 7.1 backport was a high-risk move with not a lot of end-user benefit, particularly since we already had the important bits of 7.1 rebuilt in the AIGLX repository (which, despite the link given in this article, never went offline to my knowledge, though it did move a bit). Given that, the time requirements for stabilisation for FC6, and the promise of even easier external repository integration in FC6 and following, the 7.1 backport just wasn't a win.

Think about it. If you're just going to backport everything from rawhide to the release update stream, then why bother having releases at all? Putting it in a separate repo makes it opt-in, which we don't really have a better mechanism for yet (sadly).

The board's meeting and subsequent announcement were to make sure that everyone was clear that 7.1 wouldn't be backported wholesale, and to clarify the motivation for doing so particularly with regards to the Fedora mission and policies. From the announcement itself:

[...] it's a major change with only modest benefit, and a better solution is coming soon. That's an argument that resonates with me, and that has nothing to do with proprietary software, and everything to do with stability for users.

The "better solution", of course, being essentially a continuation of the AIGLX repo idea, but for the whole X stack, and for more than one previous release of the OS. And let's be clear here, the changes in 7.1 affect the open drivers too, and bugs have been found that were introduced between 7.0 and 7.1.

Fedora users who are not up for the (sometimes hair-raising) experience of running from the development repository will have to wait for Fedora Core 6 to get X.org 7.1.

This is unfair, the upgrade path from the FC5 AIGLX repository to FC6 is completely supported due to the way we're versioning the packages.

These vendors should not have veto power over the release plans of free software distributions.

And they don't. Full stop. I'm a complete open source bigot and I think the IHVs are doing their customers a disservice by keeping their drivers closed. I've said as much to their face.

I also have a responsibility to keep a working X working. Sometimes there are more important things than running the absolute latest release. This is one of those times. That said, rawhide had 7.1 packaged within days of its upstream release, and before any other distro. So at least we're still head of the class.

Posted Aug 27, 2006 2:53 UTC (Sun) by dberkholz (guest, #23346) [Link] (1 responses)

> That said, rawhide had 7.1 packaged within days of its upstream release,
> and before any other distro. So at least we're still head of the class.

Is that so? Gentoo had it the day after release, so we certainly qualify as "within days" too.

*xorg-server-1.1.0 (23 May 2006)

23 May 2006; Joshua Baergen <joshuabaergen@gentoo.org>
-xorg-server-1.0.99.903.ebuild, +xorg-server-1.1.0.ebuild:
Bump for 7.1 final. Includes various crash fixes, etc.

Anyhow, no point in arguing about who's first. I just want to point out that Rawhide isn't the only place.

Posted Aug 27, 2006 4:23 UTC (Sun) by ajaxx (guest, #35881) [Link]

Yipe! Cheerfully withdrawn.

X.org, distributors, and proprietary modules

Posted Aug 27, 2006 22:33 UTC (Sun) by stolennomenclature (guest, #40166) [Link]

I think it is interesting to speculate as to why Nvidia has not added the texture-from-bitmap call to its drivers yet. Presumably to test and demonstrate the AIGLX technology their must be an open source driver with this functionality. Surely Nvidia can match the achievements of the OS community? Why is it taking so long?

I suspect that one of the comments already posted noting the possibility of Nvidia and ATI waiting for Microsoft Vista (pressure from MKicrosoft either direct or indirect or both) may well be correct.

Intel seems to be making the best long term move - by opening up its graphics drivers whilst already shipping the most GPU's, it may sound the death knell for these other companies. ATI already looks like its being absorbed in the manner of Digital and Compaq.

Hope they do the way of the dodo - good riddance. I'm waiting impatiently to buy the first Intel graphics card in the shops. Anyone want a cheap Nvidia?


Copyright © 2006, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds