In a press conference held today at LinuxWorld 2004 in New York, members of the old X consortium, some members of the disbanded XFree86 core developer group, and Havoc Pennington of freedesktop.org announced that X.org and XFree86 have essentially merged, and that the reformed group is working together to bring “not just more eye candy but new functionality” to the X server for Linux and Unix. Our Take: We wonder what this means for Keith Packard’s X implementation hosted at fd.o. Update: Apparently the above is not accurate, only a few (one?) developers joined X.org.
Is this reminiscent of the proverbial adage, “You don’t know what you have until you’ve lost it” ?
I find this movement very interesting. X.org was a no-player the last few years, XFree86 was losing its steam, and Keith Packard was winning the masses with his project. Clever of X.org and XFree86 to get together, I hope this will put them back on track.
Hopefully the nice words about the “new, more open style” are, in fact, backed up by future actions and communications. There certainly were a lot of complaints about the old XFree86 core team in their old structure.
That’s the good thing about standards, everyone can make their own implementation of it. I’m sure fd.o’s version will be backwards-compatible with the legacy, traditional XFree86, so there should be no problems. People who want the traditional “experience” don’t have to do anything, and those who have servers that need double-buffered translucent window compositing can get Xouvert from fd.o or whatever it is called these days.
“go you good thing!” is what I say to this development. As Raven said “Hopefully the nice words about the “new, more open style” are, in fact, backed up by future actions and communications.”, completely agree and do you know what is even better? this is UNIX centric and not Linux centric.
If in the future X development really picks up steam, not only will Linux benefit but *BSD and any other *NIX out there that uses XFree86.
Also, I am looking forward to seeing some of the good extensions being merged into the tree such as Cairo Graphics and XDamage. The one thing that will be nice when the full embrace of these extensions by GTK/Qt.
GTK/Qt need to work closer with the X community, work with the extension creators to improve integration between X/GTK and Qt so that issues such as cut and paste, drag and drop and numerous other GUI interaction techniques.
If the toolkit and X developers work closer, the eventual result will be a X with a VERY GOOD HIG resulting toolkit variety on UNIX without any of the downsides we see today. If the HIG is bolted straight to the X Windowing system, toolkit developers and developers in general are then forced to conform to a set of standards because utimately, the people who lose out when there are 100s of toolkits with each doing their HIG different results in users confused about things.
Some people mention a “grand unified toolkit”, that isn’t the solution, the solution is to harmise the external differences (differences that users seen with the applications that they use) whilst still ensuring that each toolkit can concentrate on the benefits they have over their competition.
If Borland and Microsoft were able to do it 10 years ago, I am sure that the OSS community should be able to achieve this small feat.
Now that the dead weight is gone, we can finally get somewhere. Hopefully they will clean up X a lot, even before they start adding stuff, and it should hopefully drop the compile time down.
We do need some good integration with the toolkits and X, but still allowing freedom to choose which and whatever you want so everyone stays happy. Everything working together with each other.
will it still take a weekend to get direct rendering to work?
When was the last time we saw so much news/interest in the X Server? Hopefully this new dedication will bring out some young faces that may have been turned off by the former XFree86 attitude.
does this mean that keith p is back on the XFree hacking team?
does it mean all those nice pdf based rendering models we’ve heared about are going to be implemented any time soon?
or does it, as i suspect, mean that two groups of old hacks are friends again and this announcement isn’t going to make the slightest difference to ‘joe user’?
Your results may vary, but direct rendering works pretty much out of box on most distros. Installing the NVIDIA drivers have never taken more than a few minutes on any distro I’ve tried, and I’ve done it on: Fedora, SuSE, Libranet, Debian, Gentoo, and Mandrake.
I don’t know how easy things are with the other drivers, though.
And if you would only get direct rendering in FreeBSD (-STABLE or -CURRENT, I run both) to work OOTB with my Radeon 9500-RivaTuner-modded-to-9700, I would be most grateful, my friend.
Suprised at how “overlooked” this area is, considering its centrality to everything…..in one sense you could say its “under-developed” relative to the heavy development in other areas – hope some of Keith’s ideas come to fruition and are firmly at the centre of things as we move on…..
We wonder what this means for Keith Packard’s X implementation hosted at fd.o.
Well, if it dosnt integrate back into xorg (the new name for Xfree86) then it can only mean one thing…choice.
This might be the best thing since syntax highlighting came to xedit(1).
I am glad that all sorts of new things are happening with X, but I mostly want my old desktop back, only better. I remember the old days when button-3-copy and button-2-paste worked everywhere, and I really miss it. I remember the thrill of sneakily replacing OLVWM and FVWM on my Slackware box with Bowman. I was happy to have X, and thought it performed amazingly well with 8Mb of RAM and an ok VGA card.
I was thrilled, having used NeWS at school and wanting very badly to have that at home instead of Geoworks Ensemble.
I really hadn’t seen any of this slowness everyone is talking about. Maybe I was just lucky with video cards, motherboards, and processors. I just started using a 2ghz Celeron at work lately though, a Systemax hack job from Office Depot, and did find X to be very unresponsive with my particular combination of hardware, and if it is that bad for a lot of people, I am all for whatever needs to be done.
It should pretty much work out of the box, iirc agp is in the generic kernel, all you will need to do is use kldload to load the radeon module.
This is certainly a good news.
For a moment I thought that Xfree members called it quits for good but as always, I am wrong.
If I say so my self, this would make a huge difference in oss desktops. I hope they support 3DWM there too.
well, I’m happy as a clam.
inspite of slashdot crowd diatribes ‘less ditch X in favor of <fill the blanks>’, there is simply no substitute for X as far as I know. yes, there are several projects announced a few years back. their intention was to replace X. at the time, as I remember their greatest achievement was the ability to draw a few polygons. It appears they still can draw a few polygons. (remember the joke: Polynomial. Polygon). X is alive and kicking and at this point it is certainly rejuvenated.
I used the new fd.o X server exclusively last week, and with the exception of Wolfenstein: Enemy Territory not running and Epiphany crashing on certain web pages, the server was looking pretty good. If you read the docs, you should be able to get it going in less than an hour. (slack9.1)
We’re well on our way to a next generation open-source desktop. Whether fd.o or this new Xorg server become the defacto graphics system, I think it’s safe to expect rapid growth in the next year or two. By the time Longhorn gets released, it may already look dated.
I should note that Keith’s current work looks awesome, but does nothing in the way of HIG or consistancy. All it does is shadows, great transparency, smoother transform, and smoother dragging windows. It’s up to Gnome and KDE to fill in the blanks, I personally think Gnome is taking more of an initiative, but that’s my opinion.
when we we expect to begin the fruits of this labour? 6 months? 18? call me skeptocal but i think the realtest is what they’ll do with this new group. best-of-luck!
Your comment convinced me to try out the new FD.O server. I was extremely impressed. Installation was a matter of just running the script supplied on the site. No configuration was required, since the server auto-detects your input devices. The only snag was I had to edit one of the source files to set the proper PCI-ID for my GeForce4Go. After that, the accelerated Xnvidia server worked just great. For something that is such an early state of development, its extremely impressive.
Performance is good for such an early release. Even though the drop-shadow is unaccelerated, it runs quite fast on my hardware. Expose lag is non-existant, but then again, it was pretty rare in XFree86 either. Resize performance is dismal, but that’s to be expected given that the memory allocator is apparently due for a rewrite (a resize in a backing-stored system like this involves numerous allocations/frees). Perhaps they will do something like Apple does, which is allocate a full-size buffer once at the beginning of a resize operation, then trim it to the correct size at the end of the operation. This thing has a *voracious* appetite for RAM. I’ve got a 1600×1200 display, which means at 16bpp (the highest Xnvidia supports at the moment) each full-screen window takes up 4MB. Of course, in the final version people will be running at 32bpp, so that’s 8MB per window. If you’re like me and have lots of windows open simulaniously, you’re looking at about 200MB+ just for window buffers. Of course, RAM is cheap, and if you look at it another way, its only costing you about $30 worth of RAM
When the GL-acceleration is added to the server, and the code is tuned for production use, it is definitely going to *spectacular*.
I believe you’re supposed to use Xvesa regardless of what graphics card you have, Xnvidia only has “optimizations” for one specific card use in testing. I’m glad you got to see how quickly things are progressing though :-).
hey!
what about Xouvert and XServer?
those two projects really rulz
Hi, Vulpes, how are you?
From Eric Anholt’s ‘DRI on BSD’ web site:
“FreeBSD:If kldstat shows your card module is loaded but it nothing shows up in dmesg | grep drm, you may not have a supported card. Notable unsupported cards that people are often confused about:
-Rage Pro (being developed in mach64-0-0-6-branch)
-Rage Mobility (Not referring to the Mobility M4 and higher or the Radeon Mobility chips, this one is a Rage Pro variant).
-Radeon 9200/9500/9700
As you can see from the above, I have one of the “notable unsupported cards” (two of them, perhaps, if you take the Riva Tuner mod into account). ATI is being slow about releasing driver particulars for cards that, at least in the case of the 9700, are still selling for $200 and up.
From Xfree86.org’s website:
“XFree86 has not Merged with X.Org
[23 January 2004]
There are several news items claiming that X.Org and The XFree86 Project have merged. This is a blatant lie. The XFree86 Project remains an independent organisation. There have been no discussions between The XFree86 Project and X.Org about a merge, let alone any agreement.”
Installation was a matter of just running the script supplied on the site. No configuration was required, since the server auto-detects your input devices.
Sounds like it is dumbed down already. I think they should not waste their time making it easy to install, and instead focus on adding useful features so that it will be adopted more quickly. Don’t follow in M$’s footsteps!!#!
Don’t know where you pulled that info from, but it’s not on http://www.xfree86.org
Eh? Seems to work just great over here! Like I said, I had the change the PCI-ID in the code to match my hardware. It appears to at least have a bit-blit optimization that the stock Xvesa server does not.
They changed the message. Maybe it was deleted while they were editing it. It says now:
There are several news items claiming that X.Org and The XFree86 Project have merged. This is untrue. XFree86’s relationship with X.Org has not changed since 1999 when we became an honourary member of X.Org. The XFree86 Project remains an independent organisation both legally and operationally. It will continue to as an independent organisation as mandated by its mission statement and bylaws. There has been no discussion with X.Org about any such merge, let alone any agreement to a merge.
X.Org is a vendor-sponsored organisation, formed by vendors to best suit the interests of those vendors; XFree86 is an independent volunteer organisation, with a focus on the individual. Therein lies the rub.
The “blatant lie” part has been removed, and of course, in typical cowardly fashion, there is no mention that it was updated.
Damage will only really start to get done once the the distributors decide to support Keiths implementation. When/if that occurs, then you’ll see the final exodus of coders from the XFree86 camp. The sad part is, the XFree86 core still refuse to accept that they’re the ones who created their own demise.
XFree86 Has Not Merged With X.Org
[23 January 2004]
There are several news items claiming that X.Org and The XFree86 Project have merged. This is untrue. XFree86’s relationship with X.Org has not changed since 1999 when we became an honourary member of X.Org. The XFree86 Project remains an independent organisation both legally and operationally. It will continue to as an independent organisation as mandated by its mission statement and bylaws. There has been no discussion with X.Org about any such merge, let alone any agreement to a merge.
X.Org is a vendor-sponsored organisation, formed by vendors to best suit the interests of those vendors; XFree86 is an independent volunteer organisation, with a focus on the individual. Therein lies the rub.