View Full Version : Feature Request: Rotate screen like XRotate
oimel
January 25th, 2007, 03:50 PM
Hi,
I'm using compiz for my daily work but miss one feature. I already searched for it in this forum, but it seems nobody else wants it ;) :
I want to rotate my screen by 90° to work with portrait documents. Without Xgl this feature is possible with the Xrotate extension in Xorg without restarting the X server.
Would it be possible to implement this in compiz so that a special button or shortcut rotates my screen?
RYX
January 25th, 2007, 04:29 PM
Hi, oimel!
I think (i.e. I hope) features like this will be available once the input-redirection is implemented into the core (or the x-server). Technically, the rotation is not very difficult to implement - but you wouldn't be able to use the window(s) while rotated (because the input wouldn't rotate with the screen/window) ...
But patience will sell out - this feature will be much better in compiz than it is everywhere else (looking-glass, metisse, ...) - because as far as I can say David implements it once it's finished, not once it's wanted ... (though I'd like to have it, too)
:)
stjepan
January 25th, 2007, 04:47 PM
With metisse you can rotate windows: http://www.youtube.com/watch?v=vaQBG_stv3g
http://insitu.lri.fr/metisse/
oimel
January 25th, 2007, 05:29 PM
Oh, sure, I didn't take this input-redirection-thing into account; I just thought it should be really easy to rotate the screen with OpenGL. But of course I also want my mouse to behave accordingly :)
stjepan
January 25th, 2007, 05:56 PM
No input redirections.
RYX
January 25th, 2007, 06:10 PM
Concerning the input-redirection, I share David's opinion to wait for an "official" feature in the X-server rather than creating some custom workaround (like metisse or beryl). I am sure it will not take very long time ...
lagorgy
January 25th, 2007, 06:23 PM
Concerning the input-redirection, I share David's opinion to wait for an "official" feature in the X-server rather than creating some custom workaround (like metisse or beryl). I am sure it will not take very long time ...
is there an announced day for input redirection?
RYX
January 25th, 2007, 06:29 PM
I don't know of any date. It is no planned feature for xorg 7.3 ... But I guess now that Mandriva officially uses metisse as its desktop and metisse supports input-redirection there will be a bigger interest in implementing input-redirection into xorg. Ubuntu, Suse and all other distros that use xorg surely don't want to fall behind ...
(Though it's only a personal feeling and not based on official information)
:)
stalynx
January 26th, 2007, 09:10 PM
As much as I agree with DavidR on waiting until input redirection arrives in the xorg tree, it might not actually happen for months or a year. Hell, search for "input redirection" on the xorg mailing list and you get three results. That speaks to the actual priority of it getting implemented.
Maybe DavidR should consider adding workarounds that are modular in nature so when input redirection finally comes about it will be easier to remove/replace the code.
mikedee
January 27th, 2007, 04:41 PM
Hell, search for "input redirection" on the xorg mailing list and you get three results.
I think they call it event redirection
http://freedesktop.org/wiki/Software/XEvIE
This might have something to do with it too.... At the end of the day though it will all be ready when it is ready.
Any hacks done at the compiz or plugin level would work, but they would never work RIGHT (ie there would be lots of unsolvable bugs and edge case problems), so really it could be done but would people like the result?
You then have to look at developer time, you can say developers will work on what they like, which is true. In my experience developers normally have more than 1 interest and they do sacrifice one project to work on another. If that mystery developer spent all their time on hacking something like this and continuously fixing bugs, then their time is wasted. They would also be upset when the real solution arrives and all their hard work is dumped. Even adding a workaround solution is very hard.
ATM I feel that there is too much else that could be done to try to beat the server to some sort of input redirection, we just have to realise that it is a very hard problem which they are all really keen to solve. Once its ready it will be the best thing you could hope for, but we just have to wait unfortunately.
Having said that, someone may decide to create a workaround, but I do not think its very likely.
Just my opinion anyway ;)
stalynx
January 27th, 2007, 08:41 PM
XEvIE is related but distinct from input redirection. XEvIE is more about getting events from the xorg server that correlate to hardware cursor/keyboard events. Input redirection is about moving these events around to different windows.
Anyway we are in full agreement how hackish any attempts would be to achieve behavior equivalence to input redirection without input redirection in the xorg server. However the basic problem is the disorganized state of xorg. Also it should be mentioned that the primary hackers for xorg are paid developers. Not to say that unpaid people don't do important things but the majority of "new feature" work is being done by people who get paychecks. Also they all work for different companies whose priorities are all in different places.
If RedHat thought input redirection was important they would have pushed the code into the tree already. The same goes for Novell, Intel, etc. OpenGL compositing was in a similar state until DavidR got the ball rolling with XGL. Which interestingly received similar criticism as being "hackish".
So here we are waiting for input redirection and who knows who is working on it and what is its state. So even though making hacks to get similar effects is bad engineering it is entirely a pragmatic decision considering input redirection's state in xorg.
RYX
January 27th, 2007, 09:17 PM
I agree - if nobody tries it, it'll take even longer to be done (if that's what you mean) ... I guess that David is not entirely against those "hackish" implementations, I think he just doesn't want to put work into it because he has enough to do ...
In theory, input redirection could be done by compiz and it shouldn't be much more than applying 2d/3d window transformation-matrices to all input ... maybe someone (who has really nothing better to do) can look at the way metisse does its input-redirection.
Vasek
February 12th, 2007, 01:38 PM
Hello!
Have you seen this? Looks like David Reveman took the input redirection in his own hands. IMHO it is the best way it could happen, isn't it?
http://hoegsberg.blogspot.com/2007/02/pushing-envelope.html
vBulletin® v3.7.3, Copyright ©2000-2008, Jelsoft Enterprises Ltd.