New, Modularized X Window Release Now Available for Download

By Leon Shiman, X.org -- X11R7.0 is the first release of the complete modularized and autotooled source code base for the X Window System. It is the first major version release of the X Window System in more than a decade. X11R6.9, its companion release, contains identical features and uses the exact same source code as X11R7.0, but uses the traditional imake build system. (Read the rest of the announcement below)

These changes in source code management, which give openness and transparency to the source code base and employ current technology, invite a new generation of developers to contribute, building on the long tradition of the X Window System. The new modular format offers focused development and rapid, independent updates and distribution of tested modular components as they are ready, freed from the biennial maintenance release timetable.

X11R6.9 is comprised of many distinct components bonded in a single tree, based on imake. X11R7.0 splits that set of components into logically distinct modules, separately developed, built, and maintained by the community of X.Org developers. This simultaneous release gives a transition point for developers, builders, and vendors to adapt their practices to the new X.Org modular process.

X11R7.0 supports Linux and Solaris at this time, with other support pending. X11R7.1, the first modular roll-up release, is scheduled mid-2006. While the monolithic tree will continue to be fully supported and released, new feature development is expected to concentrate on the modular code base.

The X11R7.0 and X11R6.9 releases are the work of more than fifty volunteer contributors worldwide, working under the release management team of Kevin Martin (Head), Alan Coopersmith, and Adam Jackson, with the support of Red Hat, Sun Microsystems, and the unsupported, generous contribution of effort by Adam Jackson.

All X Window System Releases are available from ftp.X.Org and mirror sites worldwide (see http://wiki.x.org/Mirrors). They are distributed under the MIT ("X") License by the X.Org Foundation LLC. Information concerning organization, activities, and mailing lists can be found at www.X.Org. Membership is free and open to contributors. Sponsorship is encouraged to support the global activities of the X.Org Foundation. Current X.Org Sponsors include Sun Microsystems, HP, IBM, StarNet Communications, AttachmateWRQ, Hummingbird, and Integrated Computer Solutions Incorporated [ICS].

In continuous use for over 20 years, the X Window System provides the only standard platform-independent networked graphical window system bridging the heterogeneous platforms in today's enterprise: from network servers to desktops, thin clients, laptops, and hand-helds, independent of operating system and hardware.

* LINUX is a registered trademark of Linus Torvalds. "Solaris" is a trademark of Sun Microsystems. All company names are trademarks of their registered owners.

-------------------

Have an important announcement or article to share with Slashdot readers? Send the complete article (or a proposal) to roblimo (at) slashdot (dot) org.

In other news

[LittleLebowskiUrbanA]

Xfree86 continues their self-imposed slide into obscurity.

Great...

[KangKong]

more autotool hell, woohoo.

Good

[revividus]

I'd guess that 99% of Linux users (myself included) do not hack away at the X source code.

On the other hand, I'd guess that for the 1% who do hack X, this will make thier lives easier. Heck, it might even mean more people decide to work on X, which OSS dogma tells us is a Good Thing(TM), and it probably is.

Re:New developers

[hackwrench]

The autotools are hard enough to learn

Sounds to me that there should be something better than imake and autotools. Something that can be easily applied to any digital project, not just codebases. Something that makes it easy for a person to have their own personal fork that keeps track of what files in the original tree the changed files are based off of and can notify the person of changes to the original project's files, so that improvements can quickly be assimilated across all forks. Anybody else have their laundry list?

Re:In other news

[IntergalacticWalrus]

Not to mention their homepage now talks more about donating money to them than anything else. As if there were any reason to give them money anymore. That's just sad.

Oh wait, they just released 4.5 and they say "it's just terrific"! Wow! I can't wait to try it in all those obscure Linux distributions that still use it (because they still haven't noticed X.org yet)!

I usually don't complain...

[dalutong]

And I am a huge proponent of Free software. But I sure would like to know when X will support today's new technologies and trends. rotating your screen is very difficult. and you can't have accelleration when you do. even resolution changes are difficult (xrandr helps, but you still can only move between the resolutions provided at the X server start, which doesn't help if you've plugged in a different monitor.) Switching between dual displays is hard.

can't think of anything else at the moment.

Re:What this means [OT]

[Shimmer]

Spelling tip: Grammar isn't spelled the way you think it is.

Re:Good

[Fnord]

No, what this means is that of the people who want to hack on the X source code, 99% were unable to get into it because of the interdependant mess that the code was, and the inabillity for most people to commit back.

Now 99% of the people who want to hack on X will be able to find a small isolated module to start on. And now those modules may be able to evolve without breaking the whole. I've wanted to hack at X for a long time, now I very well might.

platform-independent?

[msbsod]

In continuous use for over 20 years, the X Window System provides the only standard platform-independent networked graphical window system...

Somehow I question the claim that the X Window System is still platform-independent. To me it looks like a unix-centric development. There are other operating systems, like VMS, and they come with older versions of the X Window System, too. But the "autothis-and-that" tools all are written for Unix features, like the file specification, syntax of options, compilation tools etc.. None of the differences among various operating systems are addressed in the new scheme and somehow I doubt they will be in the future. Of course, one could adapt other operating system to Unix, but people who chose not to use Unix certainly did that because they do not want to their software to be Unix-alike. Not that I want to judge here which operating system is the best (after all this is /. :-), but I like to suggest that either the people who are developing the X Window System work on this part of their software or drop the claim that they produce platform-independent software.

In any case I appreciate the X Window System very much, thanks.

Re:Good

[karlto]

As a user, my life is made easier by the result of the developers' work. If it is easier for them to do this work, I'm sure that every user will see a benefit.

Re:nVidia

[jZnat]

At least nVidia actively supports Linux. Take a look at ATi and tell me they're doing anything helpful with their own abysmal Linux support.

Effect on end user

[jd]

• Modular code is easier to maintain, so expect fewer bugs and a rapid explosion in the number of features. It is also (generally) easier for binary-only extensions, so expect more hardware vendors to support it.
  
• Modular code means that the compiler cannot take advantage of any knowledge of other files when optimizing the code, but this doesn't matter much as the original tree didn't do that either. Commercially optimized versions of X might be fractionally larger and/or slower, though.
  
• Gentoo users are in for an looooong run-up to Christmas. Especially if there is a bug in the e-build.
  
• Fedora Core users will suffer greatly, unless the RPM specs correctly instruct RPM to deinstall legacy components from the old structure. Fedora users will also need to be careful about any RPM files that refer specifically to the old X11 RPMs. The same is true for other package-based distributions - package dependencies may not be tracked correctly, leading to outdated dependencies. At best, updates might fail unexpectedly.
  

Re:nVidia

[Sparks23]

I would think that there'd be an initial delay, but that in the long run it will actually be rather faster this way. Since if a video card driver wants to not share their info, they can now in theory write a modular driver for X11 and release a little binary video driver module, instead of having to release binaries of the entire X11 system.

Granted, the reality may be different than the ideal, but we can hope, right?

Looking foreward to modular X

[demon_2k]

I'm for one looking foreward to modular X.
I know that the changes don't mean much at the moment, not to the end user anyway. I'm curious how will this affect the developement process, if more developers will jump on the X.org wagon as the article suggests. Will we see releases more often? I'm also curious how will this affect video card menufactores, and ultimetly their curtomers. I don't know what about the rest of you. I see that there's a bit of mixed feelings about all this but, I'm excited about this. I can't wait to see what kind of an affect it fill have say... 2 year from now

means less painful updating your system

[aleator]

the xorg as it is now is about 110MB (binary for i686) in size. it comes out about 2 times a year. means that you have to download every year around 230MB of data to keep your X up-to-date.

BUT (!) actually, you are only 2 weeks of the whole time really up to date, because most of the libraries and drivers are outdated, just a week after the release came out. this means, that you download 230MB and are waiting the whole time for new releases hating the whole system it is organised.

new, the modularised organisation gives the developers and package maintainers the ability to update just one library at a time - to release it immediately it is known to work fine with the rest and the user has the binary of this small library (e.g. 2MB) ready for download in about a week after its release. this means you still download over the year about 200MB of updates, but you are not waiting for relases to fix your problem, because every week or month, a new release of the PARTS of xorg come out and fix problems and add features. this way, the user profits faster from the whole lot of features that come out and fixes that solve problems. (of course, in the old system, you were always able to get the whole sources (hundreds of MB) and compile them yourself (hours to days of compiling, can fail if you use wrong compiler or wrong checkout-time when getting sources))

in the modular organisaiton, also a newbie can then recompile only one part of X, because of the less time it takes and a more transperent process

==> end user gets updates more frequently, has to wait less and has much less pain updating only parts of X

Auto-configuration

[EdMcMan]

Hopefully this will mean that soon X will be able to probe more and use the config file less.

Anyway, it is great that X.org is finally bringing some more work on X. XFree was content to sit around and twiddle their thumbs for the most part.

Re:Fully Modular

[temojen]

Hopefully it'll also mean we can install X client software on a server without also needing to install an X server and fonts too (kind of useless waste of space when you don't have a monitor, keyboard or mouse on the machine)

Re:you're confused

[penguin-collective]

I mean, in terms of being able to remote-log-on to a computer and get the GUI? Windows Remote Desktop runs quite well.

X11's network transparency is a lot more than just remote desktop display; neither Windows nor OS X have anything like it.

But that's only a small part of it. Window management, damage, rendering, automation, and a lot of other facilities are far better designed and more powerful in X11 than on other platforms.

Re:I usually don't complain...

[Burz]

Xorg is also a bear to configure, either by hand or automaticaly. You are only supplied with a CLI tool (with no options) that does a terrible job most of the time. In order to change the default resolution, depth or refresh rate most users must rely on distro-specific tools which are unable to handle the xorg.conf file with confidence.

For instance, my distro did not initially setup DPMS (power-saving) feature, so I added the option to xorg.conf myself with a text editor. Now I cannot change the default settings using the disstro-supplied GUI tool-- Thee only way it will let me is to first restore the original backup configuration (the version it knows how to handle).

Linux is plagued by perhaps a dozen half-assed Dsiplay setting tools.

Why does Xorg leave it to the dsitros and the end-user?? THEY are the experts on the conf file format, so they should provide a way to change individual settings without disturbing the others. It reads the bloody xorg.conf file every time it starts, so why can't it handle writing it back to disk?? Is it too hard to add one more API funtion saveConf() so that a user-facing configuration tool can make changes without screwing up?

Alternately, why not swicth to using an XML conf file? That would result in 3rd-party tools being able to change the configuration much for sure-footedly.

Re:New developers

[Q2Serpent]

Any modern distributed version control system?

Re:Fully Modular

[wasabii]

On Debian X has been built as a large tree, but torn apart and packaged into modules, for a long time. I believe this is true about RH too.

xfonts-base, xfonts-75dpi, xserver-xfree86, few more.

What they've done now is break apart the upstream source.

Re:Longer building time

[msuarezalvarez]

And, pray, build time is a metric for exactly what?

Re:That's a shame.

[msuarezalvarez]

Build time is only an issue for developers (and people with lots of time and a passion for watching gcc command lines pass by...), most importantly, usually no one needs to rebuild the whole thing: 99% of the time, you rebuild only the little part you are working on (and, notice, that little part migth have become littler with modularisation). Build time is essentially a non-issue.

It is quite boring to watch this "autotools suck" meme live on. Sure, they can be a pain, but that is usually solved by RTFM; sure they sometimes feel like you need to perform demonic invocations in order to do something, but they sure work, and do so well enough that the very people who maintain a huge beast like X.org are willing to cope with it. Come up with an alternative, good enough that people are willing to commit themselves to using it in real life projects, and then I'll listen with pleasure to your ramblings about how much autotools suck.

YES!!! I have waited so long for this!

[poofyhairguy82]

I don't understand the people saying that nothing big for end users comes with Xorg 7. For me Xorg 7 is my best Christmas present! Am I a Xorg hacker? No, I'm an eye-candy nut!

With Xorg 7 comes the chance for the first stable composite extension! So Xcompmgr will stop crashing (as much)! Also, by using my own guide I can get an accerated desktop with a ATI 9250 card that uses EXA (which is more stable than Nvidia's renderaccel)! [ubuntuforums.org] So maybe...just maybe...I can get a Windows 98 level stable accerated desktop before 2005 ends, thereby beating Vista out the gate by a year. And since the KDE compositor is near stable, I can enjoy menu transparancies now when I log into Kubuntu without fear of crashing!

Also the new driver interface will bring improvements to the closed Nvidia driver once they get their head around it, and my 6600 GT will hopefully give me decent performance with Skippy-xd by the time Dapper comes.

Of course, this won't help most users because composite won't be turned on by a major distro for at least a year or two but for those of us on the Linux Eye Candy edge there is a whole new world open today.

By far Xorg is the most primitave part of the Linux desktop compared to the alternatives (especially with Openoffice.org2 out there) and this release is the first step towards the wonderful desktop that OSX people have now and Vista people will have next year. I can't wait soon enough for drop shadows, real transparancies, and minimize effects that do not suck!

Re:nVidia

[NeoChaosX]

ATi released the specs for their Radeon 9250 chips and older models, so open-source hackers can make our own drivers (it's also the reason why the EXA feature in 7.0 supports ATi cards right now, and not nVidia cards). IMO allowing programmers to make their own open-source drivers with the official specs would be considered a lot more "helpful to Linux" than putting out working but closed-source drivers.

Re:Fully Modular

[MattBurke]

There are more operating systems out there running X11 than your favourite RPM-based overly-dependant brand of Linux. Most of them have the sense to distinguish between the need for X client libraries and an X server. I refer to the Free/Net/OpenBSD, Solaris, and AIX servers I have not 10 yards from me now which all happily run X11 apps without having an X server installed. Also I doubt Linux distributions which do things 'properly' like Slackware and Debian would dare install an X server without explicit confirmation although I may be corrected on this latter point.

Re:Why is this imporant?

[octopus72]

99%? Are you sure? Try using generic vesa driver and see how "fast" it is. It isn't all about drivers. It's also about general architecture improvement, meaning with new driver core (currently only experimental) you will be able to use fast hardware accelerated composite extension. There are also other numerous "under th hood" improvements (brand new MESA for example)and most importantly modularisation of the whole project. Personally I feel that parallel relase of monolithic/modular version wasn't necessary. Much work implemented into something that won't be relevant in 6 months with modular-only 7.1.