Weekly Edition
Return to the Development page
| |||||||||||||
Vignatti: X on Wayland
Tiago Vignatti describes the mechanism by which X clients will be supported on a Wayland display server. "All X windows created from now on will be redirected to offscreen pixmap and stored on a DRM buffer (via the xwayland video driver); that’s how compositing works on Wayland. The idea is that a X client will behave very likely as a regular Wayland client. Therefore, there’s no protocol calls or any major task involved on xwayland and all happens seamlessly, with the protocol 'conversion' penalty close to nil."
(Log in to post comments)
Vignatti: X on Wayland Posted Jun 13, 2012 16:42 UTC (Wed) by theophrastus (guest, #80847) [Link]
Apologies for starting the comments out with a particularly out-of-the-loop question, but in the phrase: "...and stored on a DRM buffer (via the xwayland video driver)" does 'DRM' stand for Digital Rights Management? and if so, why, as a inveterate open source enthusiast, shouldn't i have a knee-jerk negative response to that? thankee.
Vignatti: X on Wayland Posted Jun 13, 2012 16:45 UTC (Wed) by blecoint (subscriber, #131) [Link]
Actually it stands for Direct Rendering Manager: DRM as in DRI (Direct Rendering Infrastructure) which you should be more familiar with.
Here it the wikipedia link http://en.wikipedia.org/wiki/Direct_Rendering_Manager
Vignatti: X on Wayland Posted Jun 13, 2012 16:58 UTC (Wed) by theophrastus (guest, #80847) [Link]
thank you. raw nerve endings semi-soothed now.
Vignatti: X on Wayland Posted Jun 13, 2012 16:51 UTC (Wed) by ebiederm (subscriber, #35028) [Link]
DRM Direct Rendering Manager
I think the initials confuse a lot of people.
Vignatti: X on Wayland Posted Jun 13, 2012 18:42 UTC (Wed) by scientes (guest, #83068) [Link]
https://en.wikipedia.org/wiki/Three-letter_acronym :
According to the Jargon File, a journalist once asked hacker Paul Boutin what he thought the biggest problem in computing in the 1990s would be. Paul's straight-faced response was: "There are only 17,000 three-letter acronyms."
Vignatti: X on Wayland Posted Jun 13, 2012 18:50 UTC (Wed) by theophrastus (guest, #80847) [Link]
perhaps there should be a parallel TLA to FLA as IPv4 to IPv6?
Vignatti: X on Wayland Posted Jun 13, 2012 19:05 UTC (Wed) by Cyberax (✭ supporter ✭, #52523) [Link]
Sure. There's even an RFC for it: http://www.ietf.org/rfc/rfc5513.txt?number=5513 !
Vignatti: X on Wayland Posted Jun 13, 2012 19:51 UTC (Wed) by JohnMorris (subscriber, #73531) [Link]
You mean an ETLA?
Vignatti: X on Wayland Posted Jun 14, 2012 6:01 UTC (Thu) by csamuel (✭ supporter ✭, #2624) [Link]
Not to mention FETLA (a further ETLA) and an EFETLA (extended FETLA).. :-)
Vignatti: X on Wayland Posted Jun 13, 2012 20:20 UTC (Wed) by carlm (guest, #84710) [Link]
Pondering my next laptop purchase to replace my ageing IBM: Intel CPU (with SSE, MMX, VT-x and AES instruction set), ECC RAM, SLC SSD, Nvidia PCI express SLI GPU (from GTX range) with support for DRM, DRI and the various 3D API's. Decent LCD (IPS, not TFT, and forget about RGB CRT). Throw in an XXL LIB for on the go and support for USB 3.0, DVD (incl. PAL), hot-swappable HDD, VGA, DVI. Throw in TRS/RCA... I'll be playing/recording plenty of PCM, WAV with XMP, CBR MP3, VBR OGG and ABR AAC. Add some funky GaP/SiC/AlN LEDs and let the fun begin. FTW!
All clear?
Vignatti: X on Wayland Posted Jun 13, 2012 20:26 UTC (Wed) by hummassa (subscriber, #307) [Link]
Once you find it, please post make, model and price here? ;-)
Vignatti: X on Wayland Posted Jun 13, 2012 21:07 UTC (Wed) by nix (subscriber, #2304) [Link]
Once you've got all of those you can replace it with a single giant acronym to rule them all.
Vignatti: X on Wayland Posted Jun 13, 2012 21:08 UTC (Wed) by nix (subscriber, #2304) [Link]
And that *whoosh* sound was the joke flying right over my head until just a bit too late. Why is there no delete-dammit button that works for five seconds after you post or something? :)
Vignatti: X on Wayland Posted Jun 13, 2012 21:14 UTC (Wed) by hummassa (subscriber, #307) [Link]
I'm with you (see my sibling post). In those cases, I just hang my head and try to cry silently enough not to call any attention to myself.
Vignatti: X on Wayland Posted Jun 14, 2012 1:03 UTC (Thu) by drag (subscriber, #31333) [Link]
You mean 'PC'?
Vignatti: X on Wayland Posted Jun 14, 2012 1:55 UTC (Thu) by salimma (subscriber, #34460) [Link]
Nah, PS/2 ...
Vignatti: X on Wayland Posted Jun 13, 2012 21:08 UTC (Wed) by cwillu (subscriber, #67268) [Link]
A POS without an SSD, DYT?
Vignatti: X on Wayland Posted Jun 22, 2012 15:51 UTC (Fri) by BenHutchings (subscriber, #37955) [Link]
Ogg isn't an acronym or abbreviation. :-P
DЯM Posted Jun 14, 2012 2:21 UTC (Thu) by gmatht (guest, #58961) [Link]
That was before unicode... it is like the IPv6 of TLAs ;)
DЯM Posted Jun 15, 2012 8:24 UTC (Fri) by fb (subscriber, #53265) [Link]
> That was before unicode... it is like the IPv6 of TLAs ;)
Priceless! Best comment of the month!
Vignatti: X on Wayland Posted Jun 14, 2012 8:39 UTC (Thu) by rvfh (subscriber, #31018) [Link]
Sorry for posting a serious comment ;-)
This looks like we could have plain X on Wayland, so the question is: can we have Wayland on our computers by the end of the year? Most of our distribution could use Wayland directly and xwayland would cope with the older stuff, right?
Can't wait... :-P
Vignatti: X on Wayland Posted Jun 14, 2012 12:12 UTC (Thu) by sorpigal (subscriber, #36106) [Link]
> Remark that the X protocol doesn't define policy but the WMs, and this is a challenge for Wayland, that does define.
When was it decided that policy in the windowing system was a good idea after all?
Vignatti: X on Wayland Posted Jun 14, 2012 12:14 UTC (Thu) by rahulsundaram (subscriber, #21946) [Link]
That's been the general trend in Linux for a number of years now.
Vignatti: X on Wayland Posted Jun 14, 2012 12:19 UTC (Thu) by cmm (guest, #81305) [Link]
after people have used X11 for a minute or two.
Vignatti: X on Wayland Posted Jun 15, 2012 11:26 UTC (Fri) by sorpigal (subscriber, #36106) [Link]
A lack of policy anywhere is a problem, but I don't see that putting it in the window system is right. No policy in X gave us a mess to deal with, but it also means that we weren't stuck with twm.
Vignatti: X on Wayland Posted Jun 15, 2012 13:35 UTC (Fri) by cmm (guest, #81305) [Link]
perhaps if people felt stuck with twm, they would've been improving twm instead of going off and writing a bazillion other variously-broken things instead.
Vignatti: X on Wayland Posted Jun 15, 2012 15:31 UTC (Fri) by nix (subscriber, #2304) [Link]
Because the world would have been a *better* place without fvwm, gwm, KWin, compiz, XMonad, Awesome and dwm! Yeah, we can come up with just one perfect window manager which has all the properties of all those projects no matter how diametrically opposed, and twm would clearly have been the right place to start! The first window manager anyone thought of should just have been tied into X at the start, and then everyone could extend it as long as they didn't mind hacking their X server!
(This sort of diametric opposition, lack of agreement, and lack of clear right answers is entirely *why* X went in for 'mechanism, not policy' in the first place. I'm astonished that this has apparently been forgotten by the Wayland devs, given that some of them are the same people who came up with that policy for X. What has changed to make it a bad idea? Nothing, as far as I can tell, except for a desire for consistency whether or not people like the consistent result more than what they could have had otherwise.)
Vignatti: X on Wayland Posted Jun 15, 2012 16:27 UTC (Fri) by Cyberax (✭ supporter ✭, #52523) [Link]
If in return for this we got Linux on at least 50% of desktops? Yes, the world would be a better place.
Oh, and Wayland is most certainly the epitome of "mechanism, not policy" principle. The whole Wayland is just about 10k of clear C code with minimum of policy.
All the policy is encapsulated in the compositor. Right now the reference Weston compositor is not very modular, but nothing stops you from writing XMonad compositor for Wayland. In fact, it'll work even better than XMonad on X.
Vignatti: X on Wayland Posted Jun 15, 2012 17:09 UTC (Fri) by nix (subscriber, #2304) [Link] If in return for this we got Linux on at least 50% of desktops? Yes, the world would be a better place.I don't care how many desktops Linux is on: it seems clear that desktops are not where the future lies for most random consumers anyway. I care that my working environment is not driven into lowest-common-denominatordom by people who think that whatever policy suits *them* necessarily suits me as well. X has done a very good job of avoiding that. I'd be glad if Wayland did too (*without* having to rewrite huge chunks of Weston, thanks).
Vignatti: X on Wayland Posted Jun 15, 2012 17:43 UTC (Fri) by drag (subscriber, #31333) [Link]
> I care that my working environment is not driven into lowest-common-denominatordom by people who think that whatever policy suits *them* necessarily suits me as well. X has done a very good job of avoiding that.
I want copy and paste that is not broken.
This is something that X and X applications has had 28 years to fix and there shows no sign in sight.
I will take 'lowest common denominator' over 'no common denominator' or whatever X has in regards to copy+paste/drag+drop.
Vignatti: X on Wayland Posted Jun 15, 2012 18:45 UTC (Fri) by hummassa (subscriber, #307) [Link]
Would you care to elaborate on that? I use Linux exclusively on the desktop for the las 15 years and all the time I can jus select something in me app, and middle-click to paste it on other app, or use ctrl-c ctrl-v just like I do in the other OSs.
Vignatti: X on Wayland Posted Jun 15, 2012 19:22 UTC (Fri) by rahulsundaram (subscriber, #21946) [Link]
You weren't aware of copy/paste issues in 15 years? That seems unbelievable. Copy something, close the application where you copied it from and then try pasting it or copy something more complex than plain old text and try. If you aren't running something like klipper, you will understand.
Vignatti: X on Wayland Posted Jun 15, 2012 19:30 UTC (Fri) by hummassa (subscriber, #307) [Link]
> If you aren't running something like klipper, you will understand.
So, NOW I know why I haven't seen any problems. I use kde (and klipper) since before the the turn of the century. And anyway, I usually cut and paste between open apps. I didn't even *know* it was supposed to work otherwise... ;-)
Vignatti: X on Wayland Posted Jun 15, 2012 20:32 UTC (Fri) by nybble41 (subscriber, #55106) [Link]
It's not even like Windows works any differently; it just has something like Klipper as a standard background process. If Klipper didn't have a tray icon (necessary to provide features not available in Windows, like clipboard history) you wouldn't even be able to tell the difference.
Vignatti: X on Wayland Posted Jun 15, 2012 21:16 UTC (Fri) by cortana (subscriber, #24596) [Link]
And some part of GNOME or GTK+ has done a similar job (for at least text selections) for several years, as I'm perfectly able to paste text once I close the gedit instance from which it was copied. The fact that I don't know where this component may be found is a testament to its reliability, as if I had filed a bug on it I'd have tried to find out. :_)
Vignatti: X on Wayland Posted Jun 16, 2012 9:51 UTC (Sat) by hongli (guest, #75254) [Link]
Well actually, a number of years ago I suddenly came to the realization that copy-paste worked as intended, after suddenly remembering that in the past I had to be careful not closing the application from which I copied. I was confused. Back in the days I ran all kinds of clipboard managers; I even wrote one myself. And now, since Ubuntu >= 8.10 or something, I don't see any indication of a clipboard manager yet everything worked as expected. How can that be? I Googled, but found nothing. I read the source, but couldn't find the location of the clipboard manager within 10 minutes so I gave up and just accepted that things Just Work(tm) now.
Vignatti: X on Wayland Posted Jun 18, 2012 10:15 UTC (Mon) by jku (subscriber, #42379) [Link]
I hope you never actually need the information, but in GNOME (and Ubuntu) clipboard and many other useful little things are handled by g-s-d: http://git.gnome.org/browse/gnome-settings-daemon/tree/pl...
Vignatti: X on Wayland Posted Jun 15, 2012 17:54 UTC (Fri) by Cyberax (✭ supporter ✭, #52523) [Link]
And I don't care about fringe users. Why should the whole architecture of the stack be geared towards them, harming 99% of 'common' users along the way?
I'm not against crazy WMs like RatPoison or XMonad, but only when their existence does not require additional expense from other users.
Vignatti: X on Wayland Posted Jun 15, 2012 19:14 UTC (Fri) by marm (guest, #53705) [Link]
Maybe because those 1% of people are those who bring you your OS kernel, compilers and device drivers? ;-)
Anyway, I don't see the "all policy belongs to the window manager" model in X harming anybody, as opposed to systems which take choices away from the users.
Vignatti: X on Wayland Posted Jun 15, 2012 22:56 UTC (Fri) by Cyberax (✭ supporter ✭, #52523) [Link]
Not really. Most kernel developers use pretty standard distributions these days.
Vignatti: X on Wayland Posted Jun 18, 2012 21:06 UTC (Mon) by gidoca (subscriber, #62438) [Link]
How is rewriting Weston any different from rewriting any X Window Manager?
Vignatti: X on Wayland Posted Jun 15, 2012 17:25 UTC (Fri) by viro (subscriber, #7872) [Link]
Your honour, the defendant stands accused of the heinous crime of mocking a nonsense statement made by the previous poster. As the matter of fact, he has openly used sarcasm, in front of multiple witnesses, without a thought of tender souls who might be forever scarred and driven away by such a reckless display of cruelty (and toxic culture - shan't forget this one, whatever that might mean) to someone who was simply expressing his/her/its opinion. This reckless behaviour cannot be allowed to stand unpunished - we must, for the sake of those tender souls (and the children - will somebody think of sadistic little bastards?) subject the defendant to ostracism or some analog thereof, lest the toxic culture continues unabated. Additionally, we must forbid any reporting of this reprehensible incident, unless accompanied by clear disclaimers stating that reporter does not find it anything other than shameful and not to be approved under any circumstances.
Vignatti: X on Wayland Posted Jun 15, 2012 19:51 UTC (Fri) by nix (subscriber, #2304) [Link]
Bravo!
(you see, this is the sort of elegant mockery which makes the world richer and teaches people something worth learning. If all mockery was of this quality, it would be a good thing. Unfortunately too much of it is a cover for letting off steam on random passers-by.)
Vignatti: X on Wayland Posted Jun 15, 2012 19:44 UTC (Fri) by cmm (guest, #81305) [Link]
> Because the world would have been a *better* place without fvwm, gwm, KWin, compiz, XMonad, Awesome and dwm
as painful as it is for me, as an Awesome user, to claim: yes, it would probably be a much better place.
copy/paste would _work_. there would be some logic and theme in the window-management keyboard shortcuts. applications with long-standing window-management brokenness (try using OpenOffice under Awesome, for instance, or pretty much any Java GUI program) would be unthinkable. I would not have to suffer video tearing on my allegedly well-supported Intel graphics chip, because everyone's environment would be similar enough for this buggy behavior to be universally irritating (providing sufficient motivation for the developers of whatever responsible part to bloody well fix it already). stuff that actually matters would work better.
Vignatti: X on Wayland Posted Jun 15, 2012 19:54 UTC (Fri) by nix (subscriber, #2304) [Link]
Why on earth does everyone say copy/paste doesn't work? It works exactly as well as it does on Windows, which is to say perfectly well iff you're copying something vaguely textual and both apps are still running. Shut the source app down and what you get back might not look quite as good as it did before.
You know how many people I've heard complain about this behaviour, outside this thread, in my entire life? Zero. Nobody other than geeks even *notices* it.
(As for the video tearing, well, that's something else I've never seen, and is not something that wiring policy into the graphics server would do -- or, rather, its existence is merely a consequence of the sort of transition -- to composition -- which is pretty much entirely impossible to implement if you nail lots of policy into the graphics server.)
Vignatti: X on Wayland Posted Jun 19, 2012 12:52 UTC (Tue) by nye (guest, #51576) [Link]
>Why on earth does everyone say copy/paste doesn't work? It works exactly as well as it does on Windows, which is to say perfectly well iff you're copying something vaguely textual and both apps are still running.
Well, there do still seem to be problems with copy/paste in my recent experience.
What I want:
What I have:
Strictly speaking, X is not at fault here. There's something going wrong with Klipper, or KDE, or Qt, or GTK, or some specific combination thereof, but that's sort of the problem - if X specified a policy, there would be much less scope for those things to introduce bugs/incompatibilities/design disagreements.
That's not to say necessarily that the answer truly is to wire a lot of policy into the display server (for one thing, it's probably too late - any attempts to improve the situation by adding more policy risk creating *more* incompatible combinations), but there is a real issue there.
Vignatti: X on Wayland Posted Jun 19, 2012 20:31 UTC (Tue) by hummassa (subscriber, #307) [Link]
I am a heavy KDE user (at least 10h/day, 300 days/year) since 1998. I never ever had a problem within the requirements you describe. I only discovered that you must keep the "from" app running (WRT not losing text formatting) when this thread started... Probably I never had closed the "from" application before. :-D
Vignatti: X on Wayland Posted Jun 20, 2012 12:43 UTC (Wed) by nye (guest, #51576) [Link]
I do have some use cases which are probably rather atypical. I'm going to try to remember to keep a log of when paste does something unexpected. Maybe there are some specific applications triggering it that I've just never noticed.
Vignatti: X on Wayland Posted Jun 21, 2012 20:31 UTC (Thu) by elanthis (guest, #6227) [Link]
When was it decided that having policy in the windowing system meant that policy couldn't be user-configurable or even whole-sale replaceable through plugins and scripting, allowing all of the same feature experimentation and widely divergent feature sets that X11 allows without all the headaches caused by buggy or incomplete reimplementations of the shared bits?
Vignatti: X on Wayland Posted Jun 18, 2012 7:12 UTC (Mon) by pbaum (subscriber, #4514) [Link]
Would this would work also for remote X clients?
Vignatti: X on Wayland Posted Jun 19, 2012 18:08 UTC (Tue) by sdalley (subscriber, #18550) [Link]
See the diagram in Vignatti's blog. From the point of view of the remote X client, the local X server is no different than it's always been. The client sees a remote server IP address plus X-server port number with a socket waiting to accept connections. This socket is initially managed by Xwayland, until the connection event causes it to fork an Xorg server, which takes over the socket and handles X protocol events as it always has done. Further clients are handled by the Xorg server right from the word go.
So the answer to your question is yes.
|
|||||||||||||