On Mon, 03 May 2010 10:50:37 -0400, Adam Jackson a...@nwnk.net wrote:
To be clear: blocks here are devPrivates hanging off some object and
bytes are number of bytes in those blocks ?
yes. With devPrivates automatically allocated now, you always get one
even if no-one ever stores something in a
On Tue, 4 May 2010 20:07:02 +0300, Tiago Vignatti tiago.vigna...@nokia.com
wrote:
Hi,
Let me know how this idea looks to you. It's a very messy code though and hard
to review.
I haven't tested with several cursors, neither hotplug. But I don't see any
apparent reasons why not work.
On Wed, 5 May 2010 17:54:07 +0200, Peter Korsgaard peter.korsga...@barco.com
wrote:
+
+/* 3x3 coordinate transformation matrix for abs devs in row major order
*/
+float transform[9];
Please use a struct pixman_f_transform for this, along with the
associated pixman matrix
On Wed, 5 May 2010 17:14:41 +0300, Tiago Vignatti tiago.vigna...@nokia.com
wrote:
Keith, no one took this one. Care to pull it?
This seems like a reasonable change, but it seems like it requires a
custom build of libXfont, right? In which case, we should be fixing that
library and having the
On Wed, 05 May 2010 22:23:43 +0200, Peter Korsgaard jac...@sunsite.dk wrote:
Any other comments before I resend?
No, the code looks fine.
--
keith.pack...@intel.com
pgpLJ0eV9XvvE.pgp
Description: PGP signature
___
xorg-devel@lists.x.org: X.Org
On Thu, 06 May 2010 05:28:22 +0700, Mikhail Gusarov dotted...@dottedmag.net
wrote:
x11perf -compwinwin10 gives 2% increase (278000-285000).
x11perf -noop gives 3% increase (2150-2220)
Hard to argue with that. Seems like something we want.
--
keith.pack...@intel.com
On Wed, 5 May 2010 16:25:28 -0400, Adam Jackson a...@redhat.com wrote:
+
+/* if we don't need to backfill, we're done */
+if (pWin-backgroundState != BackgroundPixmap)
+return pPixmap;
+if (pWin-background.pixmap != None)
+return pPixmap;
+
Presumably this
On Wed, 5 May 2010 23:57:00 +0200 (CEST), Mark Kettenis
mark.kette...@xs4all.nl wrote:
You can probabaly use a version smaller than 9 as there have been a
few ABI bumps since xf86EnableAccess() became a no-op. But you'll
have to bump the ABI anyway when you remove the interface from the
size dynamically.
This looks good to me, and is certainly something that has been
requested by many people.
Reviewed-by: Keith Packard kei...@keithp.com
--
keith.pack...@intel.com
pgpVW3g41Cpsv.pgp
Description: PGP signature
___
xorg-devel@lists.x.org
On Thu, 06 May 2010 16:58:22 +0200, Éric Piel e.a.b.p...@tudelft.nl wrote:
It would only require to find the best way to apply the matrix on the
values received from relative devices, right?
It seems like the only sensible plan for a projective transform would be
to apply the transform after
On Thu, 06 May 2010 10:40:34 -0400, Adam Jackson a...@nwnk.net wrote:
It keeps the logic out of the drivers, which I strongly approve of.
Where does this leave you after startup? Framebuffer with no CRTCs
connected seems logical enough, but it might not be something the
drivers are prepared
On Thu, 6 May 2010 10:47:29 -0700, Aaron Plattner aplatt...@nvidia.com wrote:
I can't help but think that this would all be simpler if Composite could
wrap PaintWindow. ;)
The old pixmap will be gone by the time PaintWindow is called. And, you
can't exactly keep it around as PaintWindow may
On Thu, 6 May 2010 13:12:03 -0700, Jamey Sharp ja...@minilop.net wrote:
-/* Various of the DIX function interfaces were not designed to allow
- * the client-errorValue to be set on BadValue and other errors.
- * Rather than changing interfaces and breaking untold code we introduce
- * a new
Here's a patch to randrproto.txt that specifies the protocol level
semantics for the proposed changes:
From 1265b555ebc49e7e5bc23de7441e7686748e97e1 Mon Sep 17 00:00:00 2001
From: Keith Packard kei...@keithp.com
Date: Thu, 6 May 2010 15:11:52 -0700
Subject: [PATCH] Proposed RandR changes for per
On Thu, 6 May 2010 17:58:29 -0400, Adam Jackson a...@redhat.com wrote:
This does make wide ellipses slower, by a factor of twoish:
Signed-off-by: Adam Jackson a...@redhat.com
Reviewed-by: Keith Packard kei...@keithp.com
--
keith.pack...@intel.com
pgpKjrKM66hVc.pgp
Description: PGP
On Fri, 07 May 2010 10:30:16 -0400, Adam Jackson a...@nwnk.net wrote:
Which way around do rotations apply? If I want my 1920x1200 screen
rotated right, rotations presumably includes Right, but is my pixmap
1920 wide or 1200 wide?
Some clarification appears necessary. The size is the area
On Fri, 07 May 2010 20:41:53 +0200, Simon Thum simon.t...@gmx.de wrote:
While that's true, a sensible subset of transforms could be applied
directly onto relative axes. One would need to beef up the sub-pixel
motion buffers a bit, but it's doable.
Sure, what wording would you suggest to
On Fri, 7 May 2010 18:24:19 +0300, Tiago Vignatti tiago.vigna...@nokia.com
wrote:
It's enough to display animated cursors only if IsAnimCur returns
true.
This is in the screen block handler for displaying the sprite on the
current screen; won't this end up changing the sprite multiple times
On Fri, 7 May 2010 14:58:05 -0400, Kristian Høgsberg k...@bitplanet.net
wrote:
resources = clientTable[client-index].resources;
+eltptr = clientTable[client-index].elements;
for (i = 0; i clientTable[client-index].buckets; i++) {
for (this = resources[i]; this; this
On Fri, 07 May 2010 15:21:57 -0400, Adam Jackson a...@nwnk.net wrote:
I suspect that'll be cleaner. I do like the idea of SetWindowPixmap,
it's just subtle, and seems out of scope for CRTC pixmaps.
I feared the GL issues more, but perhaps it won't be nearly as
bad.
What I was getting at was
On Fri, 7 May 2010 20:23:56 -0700, Jamey Sharp ja...@minilop.net wrote:
Signed-off-by: Jamey Sharp ja...@minilop.net
Reviewed-by: Keith Packard kei...@keithp.com
I liked the random mixture of FALSE/0 for the final DoChangeGC argument,
interesting that there was only a single call
On Fri, 7 May 2010 20:23:56 -0700, Jamey Sharp ja...@minilop.net wrote:
Signed-off-by: Jamey Sharp ja...@minilop.net
Reviewed-by: Keith Packard kei...@keithp.com
--
keith.pack...@intel.com
pgpgba2pLMCxw.pgp
Description: PGP signature
___
xorg
On Fri, 7 May 2010 20:23:58 -0700, Jamey Sharp ja...@minilop.net wrote:
Signed-off-by: Jamey Sharp ja...@minilop.net
Reviewed-by: Keith Packard kei...@keithp.com
--
keith.pack...@intel.com
pgpLxIUtForkn.pgp
Description: PGP signature
___
xorg
On Fri, 7 May 2010 20:23:59 -0700, Jamey Sharp ja...@minilop.net wrote:
In order to execute a wire-level ChangeGC request, we need to look up
the resources named by any XIDs in the value-list. Various places in the
server already have pointers to the resources they want to set into the
GC,
On Thu, 6 May 2010 11:00:37 -0700, Jamey Sharp ja...@minilop.net wrote:
Signed-off-by: Jamey Sharp ja...@minilop.net
Reviewed-by: Keith Packard kei...@keithp.com
--
keith.pack...@intel.com
pgpCatZy3FxOK.pgp
Description: PGP signature
___
xorg
size dynamically.
Signed-off-by: Dave Airlie airl...@redhat.com
Reviewed-by: Keith Packard kei...@keithp.com
--
keith.pack...@intel.com
pgpBZKkic9VPK.pgp
Description: PGP signature
___
xorg-devel@lists.x.org: X.Org development
Archives: http
On Fri, 7 May 2010 06:18:45 +1000, Dave Airlie airl...@gmail.com wrote:
I only do it for canGrow drivers, others keep the old behaviour, I
assume if you canGrow then you can do something interesting when a
display shows up later like resize the framebuffer within any
constraints.
Ajax: I
On Fri, 7 May 2010 15:43:23 -0400, Kristian Høgsberg k...@bitplanet.net wrote:
Eeek, you're right. And this is different from the other cases
because here we don't actually delete the resource which caused
another resource to be added so we'll end up doing it again. I don't
have a good
On Sat, 10 Apr 2010 19:16:40 +1000, telf...@lnx-bsp.net wrote:
+ if (! pDstPix) /* Usually implies display driver rejected
width, height, depth, etc... */
+ {
+ err = BadValue;
+ goto bail;
+ }
+
The only error you can
When resetting the server, pScrn-EnterVT must be unwrapped or the
next server generation will end up wrapping the wrapper and causing an
infinite recursion on EnterVT.
Signed-off-by: Keith Packard kei...@keithp.com
---
hw/xfree86/modes/xf86Crtc.c|2 ++
hw/xfree86/modes/xf86RandR12.c
These are some cleanup patches currently sitting on my
'fix-private-usage' branch but which have nothing to do with the
privates change. I'd like to get them merged to master so that I can
get the API-change for privates reviewed and integrated soonish.
Signed-off-by: Keith Packard kei...@keithp.com
---
config/udev.c |6 +++---
hw/dmx/dmxgc.c|2 ++
hw/dmx/glxProxy/glxcmds.c |2 +-
3 files changed, 6 insertions(+), 4 deletions(-)
diff --git a/config/udev.c b/config/udev.c
index 452fb5a..a56fcd6 100644
Signed-off-by: Keith Packard kei...@keithp.com
---
hw/dmx/dmxextension.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/hw/dmx/dmxextension.c b/hw/dmx/dmxextension.c
index 4e55533..c8e8dae 100644
--- a/hw/dmx/dmxextension.c
+++ b/hw/dmx/dmxextension.c
@@ -1155,7 +1155,7
The CursorScreenKey array is supposed to be used by the DDX for cursor
private data, but midispcur was abusing it to hold cursor bits private
information. Create a separate set of privates for the dispcur cursor
bits information.
Signed-off-by: Keith Packard kei...@keithp.com
---
mi/midispcur.c
If the driver fails to free the screen pixmap in its CloseScreen
function, fb is about the only place available. This probably
shouldn't be done though; it won't clean up the pixmap correctly.
Signed-off-by: Keith Packard kei...@keithp.com
---
fb/fbscreen.c |5 -
1 files changed, 4
On Sat, 8 May 2010 11:54:30 +0200 (CEST), Mark Kettenis
mark.kette...@xs4all.nl wrote:
+#if 0
case CT_PIXMAP:
case CT_UNSORTED:
case CT_YSORTED:
@@ -390,6 +391,7 @@ void dmxChangeClip(GCPtr pGC, int type, pointer pvalue,
int nrects)
/* These clip types are
On Sat, 8 May 2010 12:57:37 +0200, Julien Cristau jcris...@debian.org wrote:
Can we make the stuff in InputAttributes const instead?
Yeah, that appears to mostly work. I can't figure out how to get this
line to compile:
attrs.tags = xstrtokenize(udev_device_get_property_value(udev_device,
On Sat, 8 May 2010 16:39:15 -0700, Jamey Sharp ja...@minilop.net wrote:
Wow, the rabbit hole just keeps getting deeper. Here are five new
cleanup patches, two new bug fixes, and an improved version of my
previous ChangeGC rework patch series.
There is an infinite amount of cleanup work that
On Sat, 8 May 2010 16:39:16 -0700, Jamey Sharp ja...@minilop.net wrote:
Signed-off-by: Jamey Sharp ja...@minilop.net
Yeah, seems sensible enough to me.
Reviewed-by: Keith Packard kei...@keithp.com
--
keith.pack...@intel.com
pgphOLz8HImDf.pgp
Description: PGP signature
On Sat, 8 May 2010 16:39:17 -0700, Jamey Sharp ja...@minilop.net wrote:
Also make it stop being a macro.
Signed-off-by: Jamey Sharp ja...@minilop.net
Also looks fine. I like the missing semi-colons in the original.
Reviewed-by: Keith Packard kei...@keithp.com
--
keith.pack...@intel.com
On Sat, 8 May 2010 16:39:28 -0700, Jamey Sharp ja...@minilop.net wrote:
@@ -500,7 +500,7 @@ compNewPixmap (WindowPtr pWin, int x, int y, int w, int h)
XID val = IncludeInferiors;
ValidateGC(pPixmap-drawable, pGC);
- dixChangeGC (serverClient, pGC,
On Sat, 8 May 2010 22:53:05 -0700, Jamey Sharp ja...@minilop.net wrote:
Hi Keith! All your comments are, of course, already addressed in the
next and final patch I sent in the series--including the glxChangeGC
refactoring. Do you just want them squashed, or reordered somehow, or
what?
Hrm. I
On Mon, 10 May 2010 07:33:21 -0400, Kristian Høgsberg k...@bitplanet.net
wrote:
I can certainly do that. Something like this look ok?
Yeah, this looks fine.
Reviewed-by: Keith Packard kei...@keithp.com
--
keith.pack...@intel.com
pgp6DzSnIcaVm.pgp
Description: PGP signature
On Mon, 10 May 2010 20:05:56 +0300, Tiago Vignatti tiago.vigna...@nokia.com
wrote:
Why increase the number of privates for mandatory components on Xorg? I guess
the idea, which should be stated very loud, is to extensions and drivers only
use private mechanism.
The mi software cursor code
On Mon, 10 May 2010 15:34:53 -0400, Adam Jackson a...@redhat.com wrote:
Don't allow setting the alpha map of a picture to itself, or to a
picture that already has an alpha map, to prevent cycles.
Signed-off-by: Adam Jackson a...@redhat.com
I'd say that setting an alpha-map to a picture with
On Mon, 10 May 2010 09:21:53 -0700, Keith Packard kei...@keithp.com wrote:
On Mon, 10 May 2010 14:30:12 +0700, Mikhail Gusarov dotted...@dottedmag.net
wrote:
Mikhail Gusarov (6):
Clean {X,XNF}{alloc,calloc,realloc,free,strdup} from pre-C89 baggage
Document allocation
On Tue, 11 May 2010 09:09:56 -0700, Jeremy Huddleston jerem...@apple.com
wrote:
Fixes regression introduced in 9de0e31746d5f0d9d39d11c94ec3cbc04a9935fc
Appears to follow the API changes to me.
Reviewed-by: Keith Packard kei...@keithp.com
--
keith.pack...@intel.com
pgpQUtgbx16tb.pgp
On Tue, 11 May 2010 14:12:52 -0400, Adam Jackson a...@nwnk.net wrote:
I just want it to not trivially crash my server. I could care less
about correctness for this.
Then I think the only needed fix is to check for the trivial loop:
index 48693b8..78e8c9b 100644
--- a/render/picture.c
+++
On Thu, 13 May 2010 00:25:30 +0700, Mikhail Gusarov dotted...@dottedmag.net
wrote:
JFYI: I have rebased this patchset to the current master. Such global
shallow restructurings rot like ripe bananas.
Thanks, I've merged this. Can I get you to go find all of the warnings
that this adds and
I'm currently tracking four patch sets that have been out for review
for a while:
git://people.freedesktop.org/~keithp/xsever fix-private-usage
git://people.freedesktop.org/~jamey/xserver cleanups
git://people.freedesktop.org/~vignatti/xserver bus-cleanup
Signed-off-by: Keith Packard kei...@keithp.com
---
config/udev.c |6 +++---
hw/dmx/dmxgc.c|2 ++
hw/dmx/glxProxy/glxcmds.c |2 +-
3 files changed, 6 insertions(+), 4 deletions(-)
diff --git a/config/udev.c b/config/udev.c
index 5e8d8da..d706a77 100644
Here's a sequence of six short patches that clean up various private
usages in the server to make the devPrivates API change
smaller. They're all fairly separable, so feel free to review just
those that interest you.
___
xorg-devel@lists.x.org: X.Org
If the driver fails to free the screen pixmap in its CloseScreen
function, fb is about the only place available. This probably
shouldn't be done though; it won't clean up the pixmap correctly.
Signed-off-by: Keith Packard kei...@keithp.com
---
fb/fbscreen.c |4
1 files changed, 4
---
render/render.c |7 +++
1 files changed, 7 insertions(+), 0 deletions(-)
diff --git a/render/render.c b/render/render.c
index c9d3e8e..a3e96c0 100644
--- a/render/render.c
+++ b/render/render.c
@@ -1145,6 +1145,12 @@ ProcRenderAddGlyphs (ClientPtr client)
This mostly ensures that the cursor is created after the server is up
and running with all of the DevPrivate keys initialized. However, it
will also save a tiny amount of memory on systems where the invisible
cursor is never used.
Signed-off-by: Keith Packard kei...@keithp.com
---
xfixes
These aren't used anywhere.
Signed-off-by: Keith Packard kei...@keithp.com
---
dbe/dbe.c |4 +---
dbe/dbestruct.h |4
2 files changed, 1 insertions(+), 7 deletions(-)
diff --git a/dbe/dbe.c b/dbe/dbe.c
index 1cda9f5..dc4bada 100644
--- a/dbe/dbe.c
+++ b/dbe/dbe.c
@@ -1456,7
The CursorScreenKey array is supposed to be used by the DDX for cursor
private data, but midispcur was abusing it to hold cursor bits private
information. Create a separate set of privates for the dispcur cursor
bits information.
Signed-off-by: Keith Packard kei...@keithp.com
---
mi/midispcur.c
On 12 May 2010 20:27:11 +0200, Soeren Sandmann sandm...@daimi.au.dk wrote:
The alphaMap structure gets mirrored in pixman images, and it does
recurse through its alpha_map pointer:
Can you make it stop?
in which I suggest to simply BadMatch out if you try to set a picture
as its own alpha
On Wed, 12 May 2010 16:41:55 -0400, Matt Turner matts...@gmail.com wrote:
On Wed, May 12, 2010 at 4:33 PM, Mikhail Gusarov
dotted...@dottedmag.net wrote:
This patchset contains cleanups of various warnings, both introduced and not
introduced by recent X*alloc purge.
Newly introduced
Here's a script I wrote to switch allocator calls around:
#!/bin/sh
sed -i \
-e 's/\[Xx]alloc\ *(/malloc(/' \
-e 's/\[Xx]free\ *(/free(/' \
-e 's/\[Xx]realloc\ *(/realloc(/' \
-e 's/\[Xx]calloc\ *(/calloc(/' \
$@
I think that's all that's required. Suggestions on which repo
...@dottedmag.net
For the series:
Reviewed-by: Keith Packard kei...@keithp.com
--
keith.pack...@intel.com
pgpmSE9YEryfr.pgp
Description: PGP signature
___
xorg-devel@lists.x.org: X.Org development
Archives: http://lists.x.org/archives/xorg-devel
Info
Walsh ewa...@tycho.nsa.gov
Reviewed-by: Keith Packard kei...@keithp.com
--
keith.pack...@intel.com
pgpa8zPfqE6oe.pgp
Description: PGP signature
___
xorg-devel@lists.x.org: X.Org development
Archives: http://lists.x.org/archives/xorg-devel
Info: http
On 13 May 2010 03:03:47 +0200, Soeren Sandmann sandm...@daimi.au.dk wrote:
It's not really well defined if the alpha origin is different from
(0, 0), though.
Good point.
The problem with the other case is that you can *later on* change
the picture used as an alpha map to add an alpha map
On Wed, 12 May 2010 20:41:39 -0700, Jeremy Huddleston jerem...@apple.com
wrote:
I just noticed that misc.h now has X_DEPRECATED and X_NORETURN ... why
are we duplicating these? Why not just use the ones from xproto?
An excellent question -- the Xfuncproto.h one completely escaped my
mind.
On Wed, 12 May 2010 20:42:08 -0700, Jeremy Huddleston jerem...@apple.com
wrote:
Use the values from xproto rather than duplicating the effort
Signed-off-by: Jeremy Huddleston jerem...@apple.com
Reviewed-by: Keith Packard kei...@keithp.com
--
keith.pack...@intel.com
pgpB7Sp2J16Pw.pgp
On Thu, 13 May 2010 16:07:44 +1000, Peter Hutterer peter.hutte...@who-t.net
wrote:
+ if (sscanf(s, %ld%1[%], p-value.num, tmp) != 2) {
Could you use
char tmp;
sscanf(s, %ld%c, p-value.num, tmp)
and then check tmp to see if it was '%'?
--
keith.pack...@intel.com
On Thu, 13 May 2010 11:45:02 +0300, Oliver McFadden oliver.mcfad...@nokia.com
wrote:
-PointerNonInterestBoxProcPtr PointerNonInterestBox;
+PointerNonInterestBoxProcPtr PointerNonInterestBox; /** unused */
I'd just go ahead and delete this line. Otherwise,
Reviewed-by: Keith Packard
On Thu, 13 May 2010 07:23:55 -0400, Kristian Høgsberg k...@bitplanet.net
wrote:
Keith, are you waiting for more reviews on this one?
No, just a simple agreement with Adam that his analysis is correct; he
raised a question that I hadn't seen answered yet. If you're in
agreement, then I think
On Thu, 13 May 2010 17:24:06 +0300, Oliver McFadden oliver.mcfad...@nokia.com
wrote:
Wouldn't this break the ScreenRec ABI? I noticed this structure also has
a few other variables marked as unused. I assume for this exact
reason?
Yes, this changes the ABI. You're welcome to suggest
On Fri, 14 May 2010 10:14:29 -0700, Dan Nicholson dbn.li...@gmail.com wrote:
It seems like there could be more than one person looking at a pull
request, checking the review status of the patches, judging if there
isn't anything inherently wrong with the patchset, and seeing if they
merge
On Fri, 14 May 2010 10:36:13 -0700, Jamey Sharp ja...@minilop.net wrote:
XInputExtensionInit calls MakeAtom, which doesn't work without the atoms
table initialized.
Signed-off-by: Jamey Sharp ja...@minilop.net
Cc: Peter Hutterer peter.hutte...@who-t.net
Tested-by: Keith Packard kei
On Fri, 14 May 2010 12:47:06 -0700, Jamey Sharp ja...@minilop.net wrote:
On Fri, May 14, 2010 at 12:40 PM, Jeremy Huddleston jerem...@apple.com
wrote:
Thanks Jamey!
Reviewed-by: Jeremy Huddleston jerem...@apple.com
No problem! Out of curiousity, did you just review the patch, or also
On Wed, 12 May 2010 21:38:05 +0200, Julien Cristau jcris...@debian.org wrote:
On Wed, May 12, 2010 at 10:53:18 -0700, Dan Nicholson wrote:
The attached seems to build without warnings on master.
Are you guys happy with this version now? It appears to be all nicely
reviewed and ready to merge.
On Mon, 17 May 2010 12:52:47 +1000, Peter Hutterer peter.hutte...@who-t.net
wrote:
hw/xfree86/common/xf86Opt.h|4 +++
hw/xfree86/common/xf86Option.c | 47
Just looking at how other OPTV_ values are used, I don't see a patch to
On Mon, 17 May 2010 07:50:25 +0300, Oliver McFadden oliver.mcfad...@nokia.com
wrote:
Previously the cursor code would be called even with a hidden cursor.
This was because the X server used an invisible 1x1 cursor sprite,
rather than a NULL cursor.
This will help performance when
XACE is
defined, so this change seems reasonable to me.
Reviewed-by: Keith Packard kei...@keithp.com
--
keith.pack...@intel.com
pgpq5GmvMWVKE.pgp
Description: PGP signature
___
xorg-devel@lists.x.org: X.Org development
Archives: http://lists.x.org
On Tue, 18 May 2010 08:44:27 +0300, Oliver McFadden oliver.mcfad...@nokia.com
wrote:
I suppose for HW cursor it doesn't really matter since we wouldn't take
damage on that, right? We just draw everything normally, then tell the
HW to draw a sprite at X,Y? (Again, haven't looked at HW cursor
On Tue, 18 May 2010 12:54:32 -0400, Adam Jackson a...@nwnk.net wrote:
I'm not a huge fan of more MAXSCREENS-sized arrays, but I guess you
don't have any option here, the bits representations could be different
on each screen.
Reviewed-by: Adam Jackson a...@redhat.com
Thanks. You'll find in
On Wed, 19 May 2010 11:42:10 -0400, Adam Jackson a...@nwnk.net wrote:
Just a bit in the long bitfield list in WindowRec? No reason to burn a
whole int on it. Although we're remarkably close to running over an int
there too.
We can make more :-)
I bet we can macro it away in the autoconf
On Wed, 19 May 2010 11:40:24 -0400, Kristian Høgsberg k...@bitplanet.net
wrote:
How can we be sure another client hasn't connected in the meantime
getting the id of the client who schedules the swap?
I thought the whole point of creating a resource was that it would be
deleted when the client
is probably not much
used as the drivers all need to pull pixmaps out of each window these
days.
But, this is way better than what we've got.
This also marks a flags field for future expansion without needing to
break ABI.
Signed-off-by: Adam Jackson a...@redhat.com
Reviewed-by: Keith
On Wed, 19 May 2010 12:47:27 -0700, Jamey Sharp ja...@minilop.net wrote:
Keith, lest you grumble about the scope of these patches: I believe they
don't conflict with your fix-private-usage branch at *all*. :-) Or at
least, nothing that can't be auto-merged.
No grumbling, just whinging that
On Wed, 19 May 2010 16:04:14 -0400, Adam Jackson a...@nwnk.net wrote:
It's really just an optimization to store it in the ScreenRec, you could
equivalently do:
Other than the whole 'what depth does the root window get' issue...
--
keith.pack...@intel.com
pgp2sxgcoGv9O.pgp
Description: PGP
On Wed, 19 May 2010 17:30:01 -0400, Adam Jackson a...@nwnk.net wrote:
Don't commit this yet, I want to pull some of Jamey's suggestions in.
I'm a' waitin'.
--
keith.pack...@intel.com
pgpM5TgqQlzcY.pgp
Description: PGP signature
___
On Wed, 19 May 2010 17:58:15 -0400, Adam Jackson a...@nwnk.net wrote:
On Wed, 2010-05-19 at 16:04 -0400, Adam Jackson wrote:
I dunno, the whole blackPixel/whitePixel thing is an anachronism. The
idea is that on monochrome framebuffers it's not especially obvious
whether 1 or 0 is black.
On Wed, 19 May 2010 16:09:58 -0700, Jamey Sharp ja...@minilop.net wrote:
Like this? It's more clear that it's correct after the patches I posted
earlier, to make CreateScratchGC static, but I believe it's correct
regardless.
'reusable' seems a bit vague; perhaps something more like
On Wed, 19 May 2010 15:25:43 -0700, Jamey Sharp ja...@minilop.net wrote:
Since this code was using CreateScratchGC and not GetScratchGC,
FreeScratchGC would always call FreeGC, so just call it directly.
I'm having a hard time caring; drivers don't call CreateGC either, so it
seems like we can
Packard kei...@keithp.com
---
PURPLE!
A fine bikeshed color. Of course, perhaps mauve would be nicer? One
minor comment below, which you may either adopt or ignore. Otherwise:
Reviewed-by: Keith Packard kei...@keithp.com
@@ -1149,7 +1153,7 @@ FreeScratchGC(GCPtr pGC
On Wed, 19 May 2010 02:01:18 +0700, Mikhail Gusarov dotted...@dottedmag.net
wrote:
This changes ABI of server as Xalloc/Xfree/Xrealloc/Xstrdup are
no longer exported. OTOH, API is not changed.
Thanks for fixing libXdmcp as well, but we'll need to get that library
released and the package
On Thu, 20 May 2010 11:01:39 -0700, Jamey Sharp ja...@minilop.net wrote:
Yes/no?
Yeah, CreateScratchGC should be an internal implementation detail of
GetScratchGC. Besides, 'CreateGC' is a shorter name than
'CreateScratchGC', and the GC you get is exactly as described in the
protocol, instead
I tried make distcheck today and had to fix a few things in hw/xwin.
Keith Packard (2):
winpriv.h was moved from hw/xwin to hw/xwin/glx in November
Find windows GL wrapper files in $(srcdir) instead of .
hw/xwin/Makefile.am |1 -
hw/xwin/glx/Makefile.am |5 +++--
2 files changed
On Fri, 21 May 2010 14:43:17 +0300, Tiago Vignatti tiago.vigna...@nokia.com
wrote:
Although API is break, luckily any drivers right now is using such monster.
diff --git a/hw/xfree86/common/xf86str.h b/hw/xfree86/common/xf86str.h
index 485c15a..e5713db 100644
---
On Fri, 21 May 2010 14:43:19 +0300, Tiago Vignatti tiago.vigna...@nokia.com
wrote:
RAC is the champion of remaining trash for sure!
Signed-off-by: Tiago Vignatti tiago.vigna...@nokia.com
Reviewed-by: Keith Packard kei...@keithp.com
--
keith.pack...@intel.com
pgpnmhMUtcGTe.pgp
Description
On Wed, 5 May 2010 17:14:41 +0300, Tiago Vignatti tiago.vigna...@nokia.com
wrote:
Keith, no one took this one. Care to pull it?
I though I had replied suggesting that this should depend on what support was
provided in the system libXfont instead of having the server be
configured separately.
On Wed, 5 May 2010 17:54:07 +0200, Peter Korsgaard peter.korsga...@barco.com
wrote:
This is equivalent to the evdev patch sent earlier:
I'd like to see someone from the input team review this before I include
it in the server. In particular, the last I heard, there was a
suggestion that we
On Thu, 06 May 2010 10:30:25 -0400, Adam Jackson a...@nwnk.net wrote:
On Wed, 2010-05-05 at 15:45 -0700, Keith Packard wrote:
On Wed, 5 May 2010 16:25:28 -0400, Adam Jackson a...@redhat.com wrote:
+
+/* if we don't need to backfill, we're done */
+if (pWin-backgroundState
On Tue, 11 May 2010 14:12:52 -0400, Adam Jackson a...@nwnk.net wrote:
I just want it to not trivially crash my server. I could care less
about correctness for this.
How about this?
From 50be5e8ff00e734506d2bb0c25054d7648f0cf1e Mon Sep 17 00:00:00 2001
From: Keith Packard kei...@keithp.com
On Fri, 21 May 2010 17:10:59 +0100, Chris Wilson ch...@chris-wilson.co.uk
wrote:
The implementation should be fixed though, let's return BadMatch and
stop this silent corruption.
If existing apps are using these formats, it seems reasonable to expect
us to make them work. It seems like the
On Fri, 21 May 2010 19:02:32 +0300, Vignatti Tiago (Nokia-D/Helsinki)
tiago.vigna...@nokia.com wrote:
So makes sense to deprecate the ddx one and use only dix?
Alas, the ScrnInfoRec has a different lifetime than the ScreenRec, so
I'm not sure we could deprecate the DDX one. Would take some
On Fri, 21 May 2010 18:48:14 +0300, Tiago Vignatti tiago.vigna...@nokia.com
wrote:
Send only screen index instead the whole rec for lock and remove the argument
of unlock.
nak -- the arbiter presumably needs to run during server initialization
before the screen is allocated, which means that
601 - 700 of 4633 matches
Mail list logo