Bug#412069: patch for beryl support

2007-03-28 Thread Julien Cristau
On Wed, Mar 28, 2007 at 10:41:32 +0200, Robert Millan [ackstorm] wrote:

> 
> Ok.  Can we add hooks to permit that unofficial beryl packages for etch add
> the missing bits in xorg.conf without breaking the debconf-based generation?
> 
> See attached patch.
> 
Not going to happen for etch, sorry.

Cheers,
Julien


signature.asc
Description: Digital signature


Bug#412069: patch for beryl support

2007-03-28 Thread Robert Millan [ackstorm]
On Wed, Mar 28, 2007 at 09:23:20AM +0200, Brice Goglin wrote:
> tags 412069 +wontfix
> thank you
> 
> 
> 
> Trying to summarize:
> 
> [...]
> 
> So I would personally not enable any of these options by
> default (hence tagging as wontfix). It is already very late
> in the release cycle, changing the default config now is
> too dangerous according to all the above.

Ok.  Can we add hooks to permit that unofficial beryl packages for etch add
the missing bits in xorg.conf without breaking the debconf-based generation?

See attached patch.

-- 
Robert Millan

ACK STORM, S.L.  -  http://www.ackstorm.es/
diff -ur xorg-7.1.0.old/debian/local/dexconf xorg-7.1.0/debian/local/dexconf
--- xorg-7.1.0.old/debian/local/dexconf	2007-02-13 11:02:09.0 +0100
+++ xorg-7.1.0/debian/local/dexconf	2007-03-28 10:40:00.0 +0200
@@ -351,6 +351,7 @@
 if [ "$DEVICE_USE_FBDEV" = "true" ]; then
   printf "\tOption\t\t\"UseFBDev\"\t\t\"$DEVICE_USE_FBDEV\"\n" >&4
 fi
+(shopt -s nullglob ; for i in /etc/X11/xserver.debconf.d/device/* ; do $i >&4 ; done)
 printf "EndSection\n" >&4
 
 ### MONITOR
@@ -428,6 +429,12 @@
 EndSection
 SECTION
 
+### Extensions
+exec 4>"$DEXCONFTMPDIR/Extensions"
+echo "Section \"Extensions\"" >&4
+(shopt -s nullglob ; for i in /etc/X11/xserver.debconf.d/extensions/* ; do $i >&4 ; done)
+echo "EndSection" >&4
+
 # Close file descriptor 4 before we delete temporary files
 exec 4<&-
 
@@ -443,7 +450,7 @@
 
 SPACER=
 for SECTION in Header Files Module InputDeviceKeyboard InputDeviceMouse \
-   Device Monitor Screen ServerLayout DRI; do
+   Device Monitor Screen ServerLayout DRI Extensions; do
   if [ -e "$DEXCONFTMPDIR/$SECTION" ]; then
 eval $SPACER
 cat "$DEXCONFTMPDIR/$SECTION" >>"$OUTFILE"


Processed: Re: Bug#412069: patch for beryl support

2007-03-27 Thread Debian Bug Tracking System
Processing commands for [EMAIL PROTECTED]:

> tags 412069 +wontfix
Bug#412069: patch for beryl support
Tags were: patch
Tags added: wontfix

> thank you
Stopping processing here.

Please contact me if you need assistance.

Debian bug tracking system administrator
(administrator, Debian Bugs database)


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Bug#412069: patch for beryl support

2007-03-27 Thread Brice Goglin
tags 412069 +wontfix
thank you



Trying to summarize:

About Composite, upstream just got a commit saying
"Enable Composite by default now that it disables itself in
 the known bad cases."
(and it has been said that
 "fglrx disables the DRI when Composite is enabled.")

About XAANoOffScreenPixmap, Michel said:
"The option is only needed to work around bugs in XAA. Because
 basically everything is a pixmap when windows are redirected
 (i.e. when a compositing manager is running), it effectively
 disables most 2D acceleration in that case and still a large
 portion of it without a compositing manager. So this option
 can't be enabled by default, because it hurts performance i
 all cases *except* when running a GLX compositing manager."

About AddARGBGLXVisuals, there was no actual answer. It seems
nVidia specific. Since xorg.conf would have to be edited to
enable the nVidia driver, this option could be added at the
same time.


So I would personally not enable any of these options by
default (hence tagging as wontfix). It is already very late
in the release cycle, changing the default config now is
too dangerous according to all the above.

Regarding after Etch, the X server is already much newer in
experimental, Composite will be enabled safely by upstream,
the XAA problems might be fixed, EXA might work much better,
... no reason to enable any of these options now then.

Brice



-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Bug#412069: patch for beryl support

2007-02-28 Thread David Nusinow
On Wed, Feb 28, 2007 at 10:39:13AM +0100, Robert Millan [ackstorm] wrote:
> On Tue, Feb 27, 2007 at 06:25:51PM -0500, David Nusinow wrote:
> > On Tue, Feb 27, 2007 at 10:26:09AM +0100, Robert Millan [ackstorm] wrote:
> > > On Mon, Feb 26, 2007 at 11:02:02PM -0500, David Nusinow wrote:
> > > > 
> > > > I haven't decided if enabling composite by default is a wise move yet. 
> > > > No
> > > > one in Debian has done a real analysis as to what the downsides of such 
> > > > a
> > > > decision are.
> > > > 
> > > > While compiz and beryl may be very sexy, I don't want to break or 
> > > > degrade
> > > > performance on systems not running them simply for the convenience of 
> > > > not
> > > > having to modify xorg.conf. It's not out of the question, but I don't 
> > > > want
> > > > to blindly enable options in our default settings for the bling of it.
> > > 
> > > Than can we conditionalise it?  We could add a question with priority
> > > 'normal' that most users won't see.  Then in the post-etch era, the Beryl
> > > community can add a "reconfigure xserver-xorg and enable beryl settings"
> > > step in their how-tos.
> > > 
> > > If we were to do that, I suppose we would prefer to have 
> > > XAANoOffscreenPixmaps
> > > and AddARGBGLXVisuals as well.
> > 
> > I'd be fine with that. You'll have to ask the translation team for
> > permission on this one though. I've cc'ed Christian in case he's not
> > reading this report.
> 
> Can I assume that, since you agreed to Composite by default in the other
> sub-thread, we're only discussing XAANoOffscreenPixmaps and AddARGBGLXVisuals
> here ?

Exactly. I don't think composite needs a debconf question.

 - David Nusinow


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Re: Bug#412069: patch for beryl support

2007-02-28 Thread Michel Dänzer
On Wed, 2007-02-28 at 11:45 +0200, Timo Aaltonen wrote:
> On Wed, 28 Feb 2007, Robert Millan [ackstorm] wrote:
> 
> > That sounds more like an upstream solution.  IMHO, in Debian we should do it
> > in the conffile because:
> >
> >  - It's more transparent.  User can tell this is enabled without going 
> > through
> >the patchset.
> >
> >  - It's easier to disable if it turns out to be a problem (performance?).

I agree.

> See the patch below.
> 
> Messing with xorg.conf is nasty, please do it like Ubuntu (and Fedora) has 
> done since August and enable it in the server code.

As Robert said, this really needs to be done upstream. It's getting
tiresome that some distros ship patches such as this without pushing
them upstream.


-- 
Earthling Michel Dänzer   |  http://tungstengraphics.com
Libre software enthusiast |  Debian, X and DRI developer



Bug#412069: patch for beryl support

2007-02-28 Thread Timo Aaltonen

On Wed, 28 Feb 2007, Robert Millan [ackstorm] wrote:


On Tue, Feb 27, 2007 at 06:26:45PM -0500, David Nusinow wrote:

On Tue, Feb 27, 2007 at 11:33:12AM +0100, Robert Millan [ackstorm] wrote:

On Tue, Feb 27, 2007 at 10:55:55AM +0100, Michel Dänzer wrote:

My point is that if we don't figure that out satisfactorily in time, we could
just enable the Composite extension, which sounds fairly safe, and seems to be
enough for Intel cards.  And Intel happens to be the card brand we have the
best driver support for (compared to a RE'd driver for ATI and nothing for
nVidia), so focusing on it makes sense to me.


Yes, that should be mostly safe at this point. The only exception that
comes to mind offhand is that fglrx disables the DRI when Composite is
enabled.


If we have to choose between optimizing for free drivers that work sanely
(either intel or radeon) and non-free drivers that are partly broken, I
think it's clear what is better.

Ok with enabling Composite by default then?


Yeah, I'm fine with that, although I think the better method is to enable
it directly in the server by default.


That sounds more like an upstream solution.  IMHO, in Debian we should do it
in the conffile because:

 - It's more transparent.  User can tell this is enabled without going through
   the patchset.

 - It's easier to disable if it turns out to be a problem (performance?).


See the patch below.

Messing with xorg.conf is nasty, please do it like Ubuntu (and Fedora) has 
done since August and enable it in the server code.


Users still can disable it with their config.

--- xorg-server-1.2.0/os/utils.c.orig   2007-02-26 11:08:08.0 +0100
+++ xorg-server-1.2.0/os/utils.c2007-02-26 11:08:21.0 +0100
@@ -138,7 +138,7 @@
 #ifdef COMPOSITE
  /* COMPOSITE is disabled by default for now until the
   * interface is stable */
- #define COMPOSITE_DEFAULT FALSE
+ #define COMPOSITE_DEFAULT TRUE
 _X_EXPORT Bool noCompositeExtension = !COMPOSITE_DEFAULT;
 #endif




Bug#412069: patch for beryl support

2007-02-28 Thread Robert Millan [ackstorm]
On Tue, Feb 27, 2007 at 06:25:51PM -0500, David Nusinow wrote:
> On Tue, Feb 27, 2007 at 10:26:09AM +0100, Robert Millan [ackstorm] wrote:
> > On Mon, Feb 26, 2007 at 11:02:02PM -0500, David Nusinow wrote:
> > > 
> > > I haven't decided if enabling composite by default is a wise move yet. No
> > > one in Debian has done a real analysis as to what the downsides of such a
> > > decision are.
> > > 
> > > While compiz and beryl may be very sexy, I don't want to break or degrade
> > > performance on systems not running them simply for the convenience of not
> > > having to modify xorg.conf. It's not out of the question, but I don't want
> > > to blindly enable options in our default settings for the bling of it.
> > 
> > Than can we conditionalise it?  We could add a question with priority
> > 'normal' that most users won't see.  Then in the post-etch era, the Beryl
> > community can add a "reconfigure xserver-xorg and enable beryl settings"
> > step in their how-tos.
> > 
> > If we were to do that, I suppose we would prefer to have 
> > XAANoOffscreenPixmaps
> > and AddARGBGLXVisuals as well.
> 
> I'd be fine with that. You'll have to ask the translation team for
> permission on this one though. I've cc'ed Christian in case he's not
> reading this report.

Can I assume that, since you agreed to Composite by default in the other
sub-thread, we're only discussing XAANoOffscreenPixmaps and AddARGBGLXVisuals
here ?

-- 
Robert Millan

ACK STORM, S.L.  -  http://www.ackstorm.es/


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Bug#412069: patch for beryl support

2007-02-28 Thread Robert Millan [ackstorm]
On Tue, Feb 27, 2007 at 06:26:45PM -0500, David Nusinow wrote:
> On Tue, Feb 27, 2007 at 11:33:12AM +0100, Robert Millan [ackstorm] wrote:
> > On Tue, Feb 27, 2007 at 10:55:55AM +0100, Michel Dänzer wrote:
> > > > My point is that if we don't figure that out satisfactorily in time, we 
> > > > could
> > > > just enable the Composite extension, which sounds fairly safe, and 
> > > > seems to be
> > > > enough for Intel cards.  And Intel happens to be the card brand we have 
> > > > the
> > > > best driver support for (compared to a RE'd driver for ATI and nothing 
> > > > for
> > > > nVidia), so focusing on it makes sense to me.
> > > 
> > > Yes, that should be mostly safe at this point. The only exception that
> > > comes to mind offhand is that fglrx disables the DRI when Composite is
> > > enabled.
> > 
> > If we have to choose between optimizing for free drivers that work sanely
> > (either intel or radeon) and non-free drivers that are partly broken, I
> > think it's clear what is better.
> > 
> > Ok with enabling Composite by default then?
> 
> Yeah, I'm fine with that, although I think the better method is to enable
> it directly in the server by default.

That sounds more like an upstream solution.  IMHO, in Debian we should do it
in the conffile because:

  - It's more transparent.  User can tell this is enabled without going through
the patchset.

  - It's easier to disable if it turns out to be a problem (performance?).

-- 
Robert Millan

ACK STORM, S.L.  -  http://www.ackstorm.es/


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Bug#412069: patch for beryl support

2007-02-27 Thread David Nusinow
On Tue, Feb 27, 2007 at 11:33:12AM +0100, Robert Millan [ackstorm] wrote:
> On Tue, Feb 27, 2007 at 10:55:55AM +0100, Michel Dänzer wrote:
> > > My point is that if we don't figure that out satisfactorily in time, we 
> > > could
> > > just enable the Composite extension, which sounds fairly safe, and seems 
> > > to be
> > > enough for Intel cards.  And Intel happens to be the card brand we have 
> > > the
> > > best driver support for (compared to a RE'd driver for ATI and nothing for
> > > nVidia), so focusing on it makes sense to me.
> > 
> > Yes, that should be mostly safe at this point. The only exception that
> > comes to mind offhand is that fglrx disables the DRI when Composite is
> > enabled.
> 
> If we have to choose between optimizing for free drivers that work sanely
> (either intel or radeon) and non-free drivers that are partly broken, I
> think it's clear what is better.
> 
> Ok with enabling Composite by default then?

Yeah, I'm fine with that, although I think the better method is to enable
it directly in the server by default.

 - David Nusinow


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Bug#412069: patch for beryl support

2007-02-27 Thread David Nusinow
On Tue, Feb 27, 2007 at 10:26:09AM +0100, Robert Millan [ackstorm] wrote:
> On Mon, Feb 26, 2007 at 11:02:02PM -0500, David Nusinow wrote:
> > 
> > I haven't decided if enabling composite by default is a wise move yet. No
> > one in Debian has done a real analysis as to what the downsides of such a
> > decision are.
> > 
> > While compiz and beryl may be very sexy, I don't want to break or degrade
> > performance on systems not running them simply for the convenience of not
> > having to modify xorg.conf. It's not out of the question, but I don't want
> > to blindly enable options in our default settings for the bling of it.
> 
> Than can we conditionalise it?  We could add a question with priority
> 'normal' that most users won't see.  Then in the post-etch era, the Beryl
> community can add a "reconfigure xserver-xorg and enable beryl settings"
> step in their how-tos.
> 
> If we were to do that, I suppose we would prefer to have XAANoOffscreenPixmaps
> and AddARGBGLXVisuals as well.

I'd be fine with that. You'll have to ask the translation team for
permission on this one though. I've cc'ed Christian in case he's not
reading this report.

 - David Nusinow


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Bug#412069: patch for beryl support

2007-02-27 Thread Robert Millan [ackstorm]
On Tue, Feb 27, 2007 at 10:55:55AM +0100, Michel Dänzer wrote:
> > My point is that if we don't figure that out satisfactorily in time, we 
> > could
> > just enable the Composite extension, which sounds fairly safe, and seems to 
> > be
> > enough for Intel cards.  And Intel happens to be the card brand we have the
> > best driver support for (compared to a RE'd driver for ATI and nothing for
> > nVidia), so focusing on it makes sense to me.
> 
> Yes, that should be mostly safe at this point. The only exception that
> comes to mind offhand is that fglrx disables the DRI when Composite is
> enabled.

If we have to choose between optimizing for free drivers that work sanely
(either intel or radeon) and non-free drivers that are partly broken, I
think it's clear what is better.

Ok with enabling Composite by default then?

-- 
Robert Millan

ACK STORM, S.L.  -  http://www.ackstorm.es/


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Bug#412069: patch for beryl support

2007-02-27 Thread Michel Dänzer
On Mon, 2007-02-26 at 13:12 +0100, Robert Millan [ackstorm] wrote:
> On Mon, Feb 26, 2007 at 12:39:45PM +0100, Xavier Bestel wrote:
> > On Mon, 2007-02-26 at 11:44 +0100, Robert Millan [ackstorm] wrote:
> > > My intel card can do without either XAANoOffscreenPixmaps or 
> > > AddARGBGLXVisuals.
> > > No difference can be found on first sight (and I tested most basic stuff: 
> > > cube,
> > > skydome, etc).
> > > 
> > > nVidia doesn't need AddARGBGLXVisuals, but it needs XAANoOffscreenPixmaps.

The latter is weird, as TTBOMK the nvidia driver doesn't use XAA at all.

> > > I don't know about ATI.  But ATI seems to be so badly screwed wrt Beryl 
> > > that
> > > we shouldn't care much.
> > 
> > Beryl works quite well on my ATI r300 using XAANoOffscreenPixmaps -
> > well, excepted I can't resize windows and sometimes the whole thing just
> > freezes.
> 
> Yes, but as Brice said before we shouldn't enable these options unless we
> know what the drawbacks are.  Maybe you can investigate about that?

The option is only needed to work around bugs in XAA. Because basically
everything is a pixmap when windows are redirected (i.e. when a
compositing manager is running), it effectively disables most 2D
acceleration in that case and still a large portion of it without a
compositing manager. So this option can't be enabled by default, because
it hurts performance in all cases *except* when running a GLX
compositing manager.

> My point is that if we don't figure that out satisfactorily in time, we could
> just enable the Composite extension, which sounds fairly safe, and seems to be
> enough for Intel cards.  And Intel happens to be the card brand we have the
> best driver support for (compared to a RE'd driver for ATI and nothing for
> nVidia), so focusing on it makes sense to me.

Yes, that should be mostly safe at this point. The only exception that
comes to mind offhand is that fglrx disables the DRI when Composite is
enabled.


-- 
Earthling Michel Dänzer   |  http://tungstengraphics.com
Libre software enthusiast |  Debian, X and DRI developer



Bug#412069: patch for beryl support

2007-02-27 Thread Robert Millan [ackstorm]
On Mon, Feb 26, 2007 at 11:02:02PM -0500, David Nusinow wrote:
> 
> I haven't decided if enabling composite by default is a wise move yet. No
> one in Debian has done a real analysis as to what the downsides of such a
> decision are.
> 
> While compiz and beryl may be very sexy, I don't want to break or degrade
> performance on systems not running them simply for the convenience of not
> having to modify xorg.conf. It's not out of the question, but I don't want
> to blindly enable options in our default settings for the bling of it.

Than can we conditionalise it?  We could add a question with priority
'normal' that most users won't see.  Then in the post-etch era, the Beryl
community can add a "reconfigure xserver-xorg and enable beryl settings"
step in their how-tos.

If we were to do that, I suppose we would prefer to have XAANoOffscreenPixmaps
and AddARGBGLXVisuals as well.

-- 
Robert Millan

ACK STORM, S.L.  -  http://www.ackstorm.es/


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Bug#412069: patch for beryl support

2007-02-26 Thread David Nusinow
On Fri, Feb 23, 2007 at 12:18:06PM +0100, Robert Millan [ackstorm] wrote:
> Package: xserver-xorg
> Severity: wishlist
> Tags: patch
> 
> I know it's too late for Beryl to make it into etch, but can we at least ship
> an xorg.conf that is frendly to Beryl ?  With this patch, only installing
> the beryl packages will be enough to get it working with no further setup
> (provided that OpenGL is working fine).
> 
> Also note that having these extra lines in your xorg.conf is harmless for
> non-Beryl users (at least it didn't produce any new warnings or errors in
> /var/log/Xorg.log for me), so I propose to add them unconditionaly.
> 
> Please, can this make it into etch?  This kind of eye-candy is _awesome_
> to attract new users.
> 
> -- 
> Robert Millan
> 
> ACK STORM, S.L.  -  http://www.ackstorm.es/

> diff -ur xorg-7.1.0.old/debian/local/dexconf xorg-7.1.0/debian/local/dexconf
> --- xorg-7.1.0.old/debian/local/dexconf   2007-02-13 11:02:09.0 
> +0100
> +++ xorg-7.1.0/debian/local/dexconf   2007-02-23 12:07:45.0 +0100
> @@ -351,6 +351,11 @@
>  if [ "$DEVICE_USE_FBDEV" = "true" ]; then
>printf "\tOption\t\t\"UseFBDev\"\t\t\"$DEVICE_USE_FBDEV\"\n" >&4
>  fi
> +cat >&4 < + # needed for beryl
> + Option  "XAANoOffscreenPixmaps" "true"
> + Option  "AddARGBGLXVisuals" "On"

These aren't generally necessary, and I don't believe it's a good idea to
set them as default when they're not needed. The latter may be harmless,
but I don't know enough about the former, since it only appears to be
useful on some ati hardware. There's a reason why upstream hasn't enabled
it by default in the driver.

> +SECTION
>  printf "EndSection\n" >&4
>  
>  ### MONITOR
> @@ -428,6 +433,15 @@
>  EndSection
>  SECTION
>  
> +### Extensions
> +exec 4>"$DEXCONFTMPDIR/Extensions"
> +cat >&4 < +Section "Extensions"
> + # needed for beryl
> + Option  "Composite" "Enable"
> +EndSection
> +SECTION

I haven't decided if enabling composite by default is a wise move yet. No
one in Debian has done a real analysis as to what the downsides of such a
decision are.

While compiz and beryl may be very sexy, I don't want to break or degrade
performance on systems not running them simply for the convenience of not
having to modify xorg.conf. It's not out of the question, but I don't want
to blindly enable options in our default settings for the bling of it.

 - David Nusinow


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Bug#412069: patch for beryl support

2007-02-26 Thread Robert Millan [ackstorm]
On Mon, Feb 26, 2007 at 12:39:45PM +0100, Xavier Bestel wrote:
> On Mon, 2007-02-26 at 11:44 +0100, Robert Millan [ackstorm] wrote:
> > My intel card can do without either XAANoOffscreenPixmaps or 
> > AddARGBGLXVisuals.
> > No difference can be found on first sight (and I tested most basic stuff: 
> > cube,
> > skydome, etc).
> > 
> > nVidia doesn't need AddARGBGLXVisuals, but it needs XAANoOffscreenPixmaps.
> > 
> > I don't know about ATI.  But ATI seems to be so badly screwed wrt Beryl that
> > we shouldn't care much.
> 
> Beryl works quite well on my ATI r300 using XAANoOffscreenPixmaps -
> well, excepted I can't resize windows and sometimes the whole thing just
> freezes.

Yes, but as Brice said before we shouldn't enable these options unless we
know what the drawbacks are.  Maybe you can investigate about that?

My point is that if we don't figure that out satisfactorily in time, we could
just enable the Composite extension, which sounds fairly safe, and seems to be
enough for Intel cards.  And Intel happens to be the card brand we have the
best driver support for (compared to a RE'd driver for ATI and nothing for
nVidia), so focusing on it makes sense to me.

-- 
Robert Millan

ACK STORM, S.L.  -  http://www.ackstorm.es/


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Bug#412069: patch for beryl support

2007-02-26 Thread Xavier Bestel
On Mon, 2007-02-26 at 11:44 +0100, Robert Millan [ackstorm] wrote:
> My intel card can do without either XAANoOffscreenPixmaps or 
> AddARGBGLXVisuals.
> No difference can be found on first sight (and I tested most basic stuff: 
> cube,
> skydome, etc).
> 
> nVidia doesn't need AddARGBGLXVisuals, but it needs XAANoOffscreenPixmaps.
> 
> I don't know about ATI.  But ATI seems to be so badly screwed wrt Beryl that
> we shouldn't care much.

Beryl works quite well on my ATI r300 using XAANoOffscreenPixmaps -
well, excepted I can't resize windows and sometimes the whole thing just
freezes.

Xav




-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Bug#412069: patch for beryl support

2007-02-26 Thread Robert Millan [ackstorm]
On Fri, Feb 23, 2007 at 10:25:56PM +0100, Robert Millan wrote:
> 
> Oh, not really.  Screw nVidia.  Beryl is only for faithful followers of the
> one true Intel card brand.  :-)
> 
> So which lines are we going to add, if any?  I can test which lines can
> Intel do without, but not till monday.

My intel card can do without either XAANoOffscreenPixmaps or AddARGBGLXVisuals.
No difference can be found on first sight (and I tested most basic stuff: cube,
skydome, etc).

nVidia doesn't need AddARGBGLXVisuals, but it needs XAANoOffscreenPixmaps.

I don't know about ATI.  But ATI seems to be so badly screwed wrt Beryl that
we shouldn't care much.

-- 
Robert Millan

ACK STORM, S.L.  -  http://www.ackstorm.es/


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Bug#412069: patch for beryl support

2007-02-26 Thread Robert Millan [ackstorm]
On Mon, Feb 26, 2007 at 11:44:36AM +0100, Robert Millan [ackstorm] wrote:
> On Fri, Feb 23, 2007 at 10:25:56PM +0100, Robert Millan wrote:
> > 
> > Oh, not really.  Screw nVidia.  Beryl is only for faithful followers of the
> > one true Intel card brand.  :-)
> > 
> > So which lines are we going to add, if any?  I can test which lines can
> > Intel do without, but not till monday.
> 
> My intel card can do without either XAANoOffscreenPixmaps or 
> AddARGBGLXVisuals.
> No difference can be found on first sight (and I tested most basic stuff: 
> cube,
> skydome, etc).
> 
> nVidia doesn't need AddARGBGLXVisuals, but it needs XAANoOffscreenPixmaps.
> 
> I don't know about ATI.  But ATI seems to be so badly screwed wrt Beryl that
> we shouldn't care much.

Given all this, can we at least enable the Composite part by default to make
Intel cards work?

Patch attached.

-- 
Robert Millan

ACK STORM, S.L.  -  http://www.ackstorm.es/
diff -ur xorg-7.1.0.old/debian/local/dexconf xorg-7.1.0/debian/local/dexconf
--- xorg-7.1.0.old/debian/local/dexconf	2007-02-13 11:02:09.0 +0100
+++ xorg-7.1.0/debian/local/dexconf	2007-02-23 12:07:45.0 +0100
@@ -428,6 +433,15 @@
 EndSection
 SECTION
 
+### Extensions
+exec 4>"$DEXCONFTMPDIR/Extensions"
+cat >&4 <>"$OUTFILE"


Bug#412069: patch for beryl support

2007-02-23 Thread Robert Millan
On Fri, Feb 23, 2007 at 10:25:56PM +0100, Robert Millan wrote:
> > 
> > About being nVidia specific, I was talking about the other line,
> > "AddARGBGLXVisuals", I have no idea what this option does.
> 
> Neither me, I just got them from the wiki.

I just tried on an nVidia, and it turns out AddARGBGLXVisuals is the only
option for which enabling/disabling doesn't seem to make any difference in
beryl.

As for the other two, Beryl can't do without them (in nVidia, that is).

I'll test Intel tomorrow.  As for ATI, let's say all my ATI cards are
currently 2D junk.

-- 
Robert Millan

My spam trap is [EMAIL PROTECTED]  Note: this address is only intended
for spam harvesters.  Writing to it will get you added to my black list.


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Bug#412069: patch for beryl support

2007-02-23 Thread Robert Millan
> This comment is for "XAANoOffscreenPixmaps". I could say ok to adding
> this option since I remember seeing improvement in Compiz when enabling
> it on my Radeon. But, I don't know what its drawbacks are.
> 
> About being nVidia specific, I was talking about the other line,
> "AddARGBGLXVisuals", I have no idea what this option does.

Neither me, I just got them from the wiki.  If you're concerned about
yet-to-be-found drawbacks, we could make it optional via debconf then?

> > On my Intel chipset, this line didn't break anything.
> 
> So you think we should add nVidia crap to anybody's xorg.conf just
> because there are some users of these evil non-free drivers out there ? :)

Oh, not really.  Screw nVidia.  Beryl is only for faithful followers of the
one true Intel card brand.  :-)

So which lines are we going to add, if any?  I can test which lines can
Intel do without, but not till monday.

> Anyway, as long as nobody explained the drawbacks of these options (and
> there are probably some, or they would be enabled by default), there is
> no way this is going to be enabled by default, sorry. It is important to
> keep the default config working on most boards, including the old ones
> that might not support Compiz/Beryl at all for some reason (for instance
> their video memory could not be large enough to store offscreenpixmaps?).

Want a patch to make a debconf option for this?

-- 
Robert Millan

My spam trap is [EMAIL PROTECTED]  Note: this address is only intended
for spam harvesters.  Writing to it will get you added to my black list.


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Bug#412069: patch for beryl support

2007-02-23 Thread Brice Goglin
Robert Millan [ackstorm] wrote:
> On Fri, Feb 23, 2007 at 12:54:58PM +0100, Brice Goglin wrote:
>> Robert Millan [ackstorm] wrote:
>>> + Option "AddARGBGLXVisuals" "On"
>>>
>> I thought this option was nVidia specific.
>
> According to the wiki [1], it is "optional if NVIDIA drivers are used".

This comment is for "XAANoOffscreenPixmaps". I could say ok to adding
this option since I remember seeing improvement in Compiz when enabling
it on my Radeon. But, I don't know what its drawbacks are.

About being nVidia specific, I was talking about the other line,
"AddARGBGLXVisuals", I have no idea what this option does.

>
> On my Intel chipset, this line didn't break anything.

So you think we should add nVidia crap to anybody's xorg.conf just
because there are some users of these evil non-free drivers out there ? :)

Anyway, as long as nobody explained the drawbacks of these options (and
there are probably some, or they would be enabled by default), there is
no way this is going to be enabled by default, sorry. It is important to
keep the default config working on most boards, including the old ones
that might not support Compiz/Beryl at all for some reason (for instance
their video memory could not be large enough to store offscreenpixmaps?).

Brice



-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Bug#412069: patch for beryl support

2007-02-23 Thread Robert Millan [ackstorm]
On Fri, Feb 23, 2007 at 12:54:58PM +0100, Brice Goglin wrote:
> Robert Millan [ackstorm] wrote:
> > +   Option  "AddARGBGLXVisuals" "On"
> >   
> 
> I thought this option was nVidia specific.

According to the wiki [1], it is "optional if NVIDIA drivers are used".

On my Intel chipset, this line didn't break anything.

[1] 
http://wiki.beryl-project.org/wiki/Install_Beryl_on_Debian#Configure_xorg.conf

-- 
Robert Millan

ACK STORM, S.L.  -  http://www.ackstorm.es/


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Bug#412069: patch for beryl support

2007-02-23 Thread Brice Goglin
Robert Millan [ackstorm] wrote:
> + Option  "AddARGBGLXVisuals" "On"
>   

I thought this option was nVidia specific.

Brice



-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Bug#412069: patch for beryl support

2007-02-23 Thread Robert Millan [ackstorm]
Package: xserver-xorg
Severity: wishlist
Tags: patch

I know it's too late for Beryl to make it into etch, but can we at least ship
an xorg.conf that is frendly to Beryl ?  With this patch, only installing
the beryl packages will be enough to get it working with no further setup
(provided that OpenGL is working fine).

Also note that having these extra lines in your xorg.conf is harmless for
non-Beryl users (at least it didn't produce any new warnings or errors in
/var/log/Xorg.log for me), so I propose to add them unconditionaly.

Please, can this make it into etch?  This kind of eye-candy is _awesome_
to attract new users.

-- 
Robert Millan

ACK STORM, S.L.  -  http://www.ackstorm.es/
diff -ur xorg-7.1.0.old/debian/local/dexconf xorg-7.1.0/debian/local/dexconf
--- xorg-7.1.0.old/debian/local/dexconf	2007-02-13 11:02:09.0 +0100
+++ xorg-7.1.0/debian/local/dexconf	2007-02-23 12:07:45.0 +0100
@@ -351,6 +351,11 @@
 if [ "$DEVICE_USE_FBDEV" = "true" ]; then
   printf "\tOption\t\t\"UseFBDev\"\t\t\"$DEVICE_USE_FBDEV\"\n" >&4
 fi
+cat >&4 <&4
 
 ### MONITOR
@@ -428,6 +433,15 @@
 EndSection
 SECTION
 
+### Extensions
+exec 4>"$DEXCONFTMPDIR/Extensions"
+cat >&4 <>"$OUTFILE"