On Wed, 19 Apr 2006, Kristian Høgsberg wrote:

Hi,

I've set up a yum repository with AIGLX packages for FC5. It's on download.fedora.redhat.com:

http://download.fedora.redhat.com/pub/fedora/projects/aiglx

and there's a aiglx.repo file there that can be dropped into /etc/yum.repos.d. With that,

$ yum --enablerepo=aiglx update

should pull down the AIGLX package set. The packages are just the rawhide RPMs rebuilt for FC5, so they're a little rough around the edges. That said, if the metacity compositor isn't enabled (see below), things should be pretty stable.

Yum chokes pretty hard on that repository.

aiglx [4/6] //var/cache/yum/aiglx/repomd.xml:16: parser error : Opening and ending tag mismatch: link line 15 and head </head> ^ //var/cache/yum/aiglx/repomd.xml:22: parser error : Opening and ending tag mismatch: img line 22 and a <a href="/"><img src="/images/header-fedora_logo01.png" alt="Fedora Project"></a

^ //var/cache/yum/aiglx/repomd.xml:23: parser error : Opening and ending tag mismatch: a line 22 and div </div> ^ //var/cache/yum/aiglx/repomd.xml:27: parser error : Opening and ending tag mismatch: img line 27 and a <a href="/Download/"><img src="/images/header-download.png" alt=" ">Download</a

^ //var/cache/yum/aiglx/repomd.xml:28: parser error : Opening and ending tag mismatch: img line 28 and a ef="/About/Projects/"><img src="/images/header-projects.png" alt=" ">Projects</a

^ //var/cache/yum/aiglx/repomd.xml:29: parser error : Opening and ending tag mismatch: img line 29 and a <a href="/About/FAQ.html"><img src="/images/header-faq.png" alt=" ">FAQ</a> ^ //var/cache/yum/aiglx/repomd.xml:29: parser error : Opening and ending tag mismatch: a line 29 and span a href="/About/FAQ.html"><img src="/images/header-faq.png" alt=" ">FAQ</a></span

^ //var/cache/yum/aiglx/repomd.xml:30: parser error : Opening and ending tag mismatch: a line 28 and div </div> ^ //var/cache/yum/aiglx/repomd.xml:31: parser error : Opening and ending tag mismatch: a line 27 and div </div> ^ //var/cache/yum/aiglx/repomd.xml:52: parser error : Entity 'nbsp' not defined <div class="fedora-corner-tr">&nbsp;</div> ^ //var/cache/yum/aiglx/repomd.xml:53: parser error : Entity 'nbsp' not defined <div class="fedora-corner-tl">&nbsp;</div> ^ //var/cache/yum/aiglx/repomd.xml:60: parser error : Entity 'nbsp' not defined <div class="fedora-corner-br">&nbsp;</div> ^ //var/cache/yum/aiglx/repomd.xml:61: parser error : Entity 'nbsp' not defined <div class="fedora-corner-bl">&nbsp;</div> ^ //var/cache/yum/aiglx/repomd.xml:93: parser error : Entity 'copy' not defined Copyright &copy; 2003-2005 Red Hat, Inc. All rights reserved. ^ //var/cache/yum/aiglx/repomd.xml:96: parser error : Opening and ending tag mismatch: br line 96 and div <br> </div> ^ //var/cache/yum/aiglx/repomd.xml:111: parser error : Opening and ending tag mismatch: br line 95 and body </body> ^ //var/cache/yum/aiglx/repomd.xml:112: parser error : Opening and ending tag mismatch: br line 94 and html</html> ^ //var/cache/yum/aiglx/repomd.xml:113: parser error : Premature end of data in tag div line 92

^ //var/cache/yum/aiglx/repomd.xml:113: parser error : Premature end of data in tag span line 26

^ //var/cache/yum/aiglx/repomd.xml:113: parser error : Premature end of data in tag div line 25

^ //var/cache/yum/aiglx/repomd.xml:113: parser error : Premature end of data in tag div line 21

^ //var/cache/yum/aiglx/repomd.xml:113: parser error : Premature end of data in tag div line 20

^ //var/cache/yum/aiglx/repomd.xml:113: parser error : Premature end of data in tag body line 18

^ //var/cache/yum/aiglx/repomd.xml:113: parser error : Premature end of data in tag link line 14

^ //var/cache/yum/aiglx/repomd.xml:113: parser error : Premature end of data in tag meta line 13

^ //var/cache/yum/aiglx/repomd.xml:113: parser error : Premature end of data in tag link line 7

^ //var/cache/yum/aiglx/repomd.xml:113: parser error : Premature end of data in tag meta line 6

^ //var/cache/yum/aiglx/repomd.xml:113: parser error : Premature end of data in tag head line 4

^ //var/cache/yum/aiglx/repomd.xml:113: parser error : Premature end of data in tag html line 3

^ aiglx 100% |=========================| 4.1 kB 00:00 http://download.fedora.redhat.com/pub/fedora/projects/aiglx/x86_64/repodata/...: [Errno -1] Error importing repomd.xml for aiglx: Error: could not parse file //var/cache/yum/aiglx/repomd.xml Trying other mirror. Cannot open/read repomd.xml file for repository: aiglx failure: repodata/repomd.xml from aiglx: [Errno 256] No more mirrors to try. Error: failure: repodata/repomd.xml from aiglx: [Errno 256] No more mirrors to try.

Kristian

I tried this just for fun on my ATI r300 and was surprised that it almost worked. I get what looks like an alpha channel (blue screen with shadows on it). True to the notes I can turn compositing off and everything is back to normal.

Is there a focus for this work (mail list, wiki page, irc meeting)? Do you want feedback to bugzilla?

-Cam

On Wed, 2006-04-19 at 16:03 -0700, alan wrote:

On Wed, 19 Apr 2006, Kristian Høgsberg wrote:

...

Hi,

I've set up a yum repository with AIGLX packages for FC5. It's on download.fedora.redhat.com:

http://download.fedora.redhat.com/pub/fedora/projects/aiglx

and there's a aiglx.repo file there that can be dropped into /etc/yum.repos.d. With that,

$ yum --enablerepo=aiglx update

should pull down the AIGLX package set. The packages are just the rawhide RPMs rebuilt for FC5, so they're a little rough around the edges. That said, if the metacity compositor isn't enabled (see below), things should be pretty stable.

Yum chokes pretty hard on that repository.

no, actually, yum doesn't choke at all:

https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=189443

-sv

Is it possible (as in: would it work) to use metacity with a KDE desktop, or is this fun only for a gnome desktop at this point?

Joel Rittvo

Does it work with the latest nvidia driver ?

I think it was not mentioned the old aiglx repo should be removed from our /etc/yum.repos.d.

On 4/20/06, Ignacio Vazquez-Abrams ivazquez@ivazquez.net wrote:

On Thu, 2006-04-20 at 00:33 -0400, Joel Rittvo wrote:

...

Is it possible (as in: would it work) to use metacity with a KDE desktop, or is this fun only for a gnome desktop at this point?

metacity is just a window manager. There's almost no reason you can't use it in place of kwin.

-- Ignacio Vazquez-Abrams ivazquez@ivazquez.net http://fedora.ivazquez.net/

gpg --keyserver hkp://subkeys.pgp.net --recv-key 38028b72

-----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2.2 (GNU/Linux)

iD8DBQBERxG9oK1Hsnseh8QRAgIMAJ4uuvWERoY87ewDOR9NfPN/rdLbZQCeL87y QiU95cT8j0Vu5pprhQR7wfc= =WchA -----END PGP SIGNATURE-----

-- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list

-- http://clunixchit.blogspot.com

On Wed, Apr 19, 2006 at 07:30:51PM -0500, Josh Boyer wrote:

On Thu, 2006-04-20 at 00:45 +0100, Cam wrote:

...

Kristian

I tried this just for fun on my ATI r300 and was surprised that it almost worked. I get what looks like an alpha channel (blue screen with shadows on it). True to the notes I can turn compositing off and everything is back to normal.

I saw the same thing on my laptop with an ATI Radeon Mobility something or other.

Are there any differences between the FC5 and rawhide packages? I was working with Bill Nottingham last night, who has an identical graphics card, and his setup was working with AIGLX whereas mine wasn't. The only difference that we could see was that Bill was running rawhide, and I was running FC5.

The chipset in question is an ATI r250.

josh

Josh Boyer wrote:

Are there any differences between the FC5 and rawhide packages? I was working with Bill Nottingham last night, who has an identical graphics card, and his setup was working with AIGLX whereas mine wasn't. The only difference that we could see was that Bill was running rawhide, and I was running FC5.

There shouldn't be, should just be a backport to FC5. I'll verify this today.

- ajax

Matthias Saou wrote:

Adam Jackson wrote :

...

...

I tried this just for fun on my ATI r300 and was surprised that it almost worked. I get what looks like an alpha channel (blue screen with shadows on it). True to the notes I can turn compositing off and everything is back to normal.

I see this too on an X800, but I think it was only on amd64 kit. I'm looking into it. The weird thing is pretty much everything else about accelerated indirect works (q3 over the network, etc), just not the texture_from_pixmap extension, so it should be pretty straightforward to track down.

I'd love to test this too, but it would be with an RV350 (Mobility Radeon 9600 M10)... what are my chances to have accelerated 3D working if I switch to rawhide? Just curious to know if it's even worth thinking about trying or not.

Rawhide should have DRI enabled for all R300 and R400 series cards, with the caveat that PCIE cards are busted until hopefully the end of the day (need to backport some stuff from Mesa head).

Of course, that's just whether it's enabled, not whether it works. But the M10 stands a pretty good chance.

- ajax

Adam Jackson wrote :

...

I tried this just for fun on my ATI r300 and was surprised that it almost worked. I get what looks like an alpha channel (blue screen with shadows on it). True to the notes I can turn compositing off and everything is back to normal.

I see this too on an X800, but I think it was only on amd64 kit. I'm looking into it. The weird thing is pretty much everything else about accelerated indirect works (q3 over the network, etc), just not the texture_from_pixmap extension, so it should be pretty straightforward to track down.

I'd love to test this too, but it would be with an RV350 (Mobility Radeon 9600 M10)... what are my chances to have accelerated 3D working if I switch to rawhide? Just curious to know if it's even worth thinking about trying or not.

Matthias

On Thu, Apr 20, 2006 at 10:05:09AM -0400, Adam Jackson wrote:

Josh Boyer wrote:

...

Are there any differences between the FC5 and rawhide packages? I was working with Bill Nottingham last night, who has an identical graphics card, and his setup was working with AIGLX whereas mine wasn't. The only difference that we could see was that Bill was running rawhide, and I was running FC5.

There shouldn't be, should just be a backport to FC5. I'll verify this today.

That would be great.

If there's no major differences in the aiglx packages, could the difference in kernel versions between FC5 and rawhide have something to do with it? That was asked on IRC, and graphics stuff is a black box to me so I have no idea.

josh

Jeremy Katz wrote:

Wouldn't it make sense to start enabling this by default at this point?

No. Otherwise I'd have just done it upstream.

What's the downside of doing so?

You break Xinerama. And GLX and Xv redirection still don't work yet, which is sort of a bad thing.

- ajax

On Thu, Apr 20, 2006 at 11:17:16AM -0400, Sam Folk-Williams wrote:

Hi, this is what I get when I execute either of the above commands. This is on a t42 with a ATI mobility 7500, with the livna kmod package installed. Any hints or tips much appreciated:

[root@localhost ~]# /usr/share/themes/Clearlooks-gOrangeous/gtk-2.0/gtkrc:3: Unable to find include file: "icons/iconrc"

depth: 24

type: _NET_WM_WINDOW_TYPE_DESKTOP

type: _NET_WM_WINDOW_TYPE_DESKTOP

type: _NET_WM_WINDOW_TYPE_DOCK

type: _NET_WM_WINDOW_TYPE_DOCK

type: _NET_WM_WINDOW_TYPE_DOCK

type: _NET_WM_WINDOW_TYPE_DOCK

metacity: symbol lookup error: metacity: undefined symbol: glXBindTexImageEXT

[1]+ Exit 127 USE_WOBBLY=1 metacity --replace

Turns out this was due to the kmod-fgrlx package -- thanks Jeremy. Now I am getting the blue screen others have mentioned. Sam

...

-- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list

dragoran wrote:

Kristian Høgsberg wrote:

...

Hi,

I've set up a yum repository with AIGLX packages for FC5. It's on download.fedora.redhat.com:

...

where are the x86_64 packages? I thought the bug was fixed.

They should be up soon, but the X server was crashing consistently for me on x86_64 when I tested it, so I didn't want to push that just yet.

Kristian

On Thu, 2006-04-20 at 13:30 -0400, Sam Folk-Williams wrote:

Turns out this was due to the kmod-fgrlx package -- thanks Jeremy. Now I am getting the blue screen others have mentioned.

BSOD: Blue screen of death on Linux... that's made my day :-)

Richard.

Adam Jackson wrote :

The log _always_ claims to "initialize" the Composite extension, which is broken. If you just see the message:

(II) Initializing built-in extension COMPOSITE

then that just means your server was built with Composite support. If you also see the message:

(**) Extension "Composite" is enabled

then you really do have Composite enabled. If you don't, you need to turn it on in your xorg.conf:

Section "Extensions" Option "Composite" EndSection

This is what I was missing! Now I get a nice blue background, a blue panel, a blue gkrellm... and some blue shades to go with them. Seems like I got to the same point as all the others now :-)

Matthias

Jeremy Katz wrote:

On Thu, 2006-04-20 at 12:41 -0400, Adam Jackson wrote:

...

Matthias Saou wrote: then you really do have Composite enabled. If you don't, you need to turn it on in your xorg.conf:

Wouldn't it make sense to start enabling this by default at this point? What's the downside of doing so?

Jeremy

as long as the composite manager is disabled it should'nt make any differnce...

On Thu, 20 Apr 2006, Josh Boyer wrote:

On Thu, Apr 20, 2006 at 10:05:09AM -0400, Adam Jackson wrote:

...

Josh Boyer wrote:

...

Are there any differences between the FC5 and rawhide packages? I was working with Bill Nottingham last night, who has an identical graphics card, and his setup was working with AIGLX whereas mine wasn't. The only difference that we could see was that Bill was running rawhide, and I was running FC5.

There shouldn't be, should just be a backport to FC5. I'll verify this today.

That would be great.

If there's no major differences in the aiglx packages, could the difference in kernel versions between FC5 and rawhide have something to do with it? That was asked on IRC, and graphics stuff is a black box to me so I have no idea.

Would it be also possible (assuming that it works) of getting x86_64 packages for AIGLX? I don't want to have to move everything on my laptop to rawhide quite yet.

Is it working with XVideo yet?

If not, any workaround for Xine and MythTV?

Ignacio Vazquez-Abrams wrote:

On Thu, 2006-04-20 at 00:33 -0400, Joel Rittvo wrote:

...

Is it possible (as in: would it work) to use metacity with a KDE desktop, or is this fun only for a gnome desktop at this point?

metacity is just a window manager. There's almost no reason you can't use it in place of kwin.

I renamed /usr/bin/kwin to /usr/bin/kwin-real, and then made a symlink between metacity and kwin. Everything seems to be running fine calling metacity in this roundabout way.

Could someone tell me the correct way to tell KDE to use metacity instead of kwin so I can undo this kludge method, though. Thanks!

Joel Rittvo

On Fri, 2006-04-21 at 12:23 +0200, Erwin Rol wrote:

I also see the blue background with shadows only, when using aiglx, with a radeon laptop. What might be interesting is that it works a lot better when i switch from 24bit to 16bit mode (I can't test 32bit since that doesn't seem to be supported with the radeon driver).

I found the following warning, which might suggest a difference between FC5 and Rawhide.

(WW) RADEON(0): [dri] limiting video memory to one aperture of 65536K (WW) RADEON(0): [dri] detected radeon kernel module version 1.22 but 1.23 or newer is required for full memory mapping. (II) RADEON(0): Detected total video RAM=32768K, accessible=65536K (PCI BAR=131072K)

But they seem to work kind of, glxgears for example works, but the OpenGL part of the window is always ontop. Blender does weird things also, it looks OK but parts of the windows are black when you move the cursor over them (blender does things like highlighting, and the whole GUI is in OpenGL).

Interesting is that in 24bit mode the output of blender is OK. It looks like in 24 bit mode only the OpenGL parts are shown.

- Erwin

Joel Rittvo wrote:

I renamed /usr/bin/kwin to /usr/bin/kwin-real, and then made a symlink between metacity and kwin. Everything seems to be running fine calling metacity in this roundabout way.

Could someone tell me the correct way to tell KDE to use metacity instead of kwin so I can undo this kludge method, though. Thanks!

startkde will honor the $KDEWM environment variable.

- ajax

On Wed, Apr 19, 2006 at 04:00:48PM -0400, Kristian Høgsberg wrote:

$ USE_WOBBLY=1 metacity --replace &

• metacity also has an experimental screen magnifier built in. Like

the wobbly windows this is enabled by setting an environment variable - USE_MAGNIFIER:

$ USE_MAGNIFIER=1 metacity --replace &

As a lot of people are having trouble with these awesome effects - are there other effects to play with? Like, how does one enable drop shadows or transparency -- or is it just these two for now? Also, where are these variables set permanently?

Thanks, Sam

• real translucency in gnome-terminal. Enable this by clicking the

"Transparent background" radio box in the "Effects" tab of the termnal profile editor dialog. gnome-terminal may have to be restarted for this to take effect - pkill gnome-terminal should do the trick.

There are a number of known bugs with these packages, so go easy on bugzilla :). Specifically,

• The damage events doesn't always kick in, so sometimes window

contents doesn't update properly. A workaround for this is to switch desktop back and forth.

• The drop shadows look weird on shaped/argb windows, for example

xeyes, the notify bubbles and well, even the rounded metacity corners. The shadow code is due for an overhaul later, so we're not going to patch over this issue in the short term.

• Perfomance for in-window updates isn't always great. We're

currently doing too much work when updating textures from pixmaps. It's still possible to optimize this further with the current setup, and longer term the memory architecture in the X server and DRI stack will see some changes that will allow us to optimize this further.

Have fun, Kristian

-- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list

On Fri, 2006-04-21 at 14:26 -0400, Sam Folk-Williams wrote:

Sorry -- I figured this out. The drop shadows should just be there if it's working... I'm at the point where with wobbly it's the blue screen, without wobbly the composite works but X eats up a lot of cpu, everything drags, and the colors get sort of weird.... this is the radeon 7000 in a thinkpad t42

Experienced the same thing without wobbly and X eating up lots of CPU. This is a Radeon 9000 Pro.

Hi.

n0dalus n0dalus+redhat@gmail.com wrote:

Isn't 32 bit colour depth just the same as 24 bit with an unused 8 bits?

If you disregard a possible alpha channel, yes. It then comes down to access patterns in the video memory (accessing 32 bit values on 32 bit boundaries may be faster, depending on your system)

On Sat, 2006-04-22 at 10:52 +0200, Erwin Rol wrote:

On Sat, 2006-04-22 at 17:05 +0930, n0dalus wrote:

...

On 4/21/06, Erwin Rol mailinglists@erwinrol.com wrote:

...

when i switch from 24bit to 16bit mode (I can't test 32bit since that doesn't seem to be supported with the radeon driver).

Isn't 32 bit colour depth just the same as 24 bit with an unused 8 bits?

Well the driver must think there is some difference because it bails out with a "not supported" when i select 32bit color depth :-) I dunno if the radeon actually does a 24bit packed (3 byte) or a 24bit unpacked (4byte) store in the frame buffer. Also the Matrox cards support this 10bit per color thing, so they need 30bit per pixel, there 24bit and 32bit really would be different.

For a while now, in XFree86/Xorg, setting depth 24 refers to the actual color depth, not padding. The driver is to set the actual framebuffer depth to 24 or 32 bits as is appropriate for the hardware. AFAIK on most hardware, padding 24bit out to 32 bit performs faster.

No idea how Xorg plans to handle this new fangled HDR thing...

On Mon, 2006-04-24 at 02:17 +0530, Deepan Chakravarthy wrote:

Harvestman is a web crawler written in python. It is now shipped in debian. I would like to maintain harvestman for fedora extras. How do I add harvestman to fedora extras ?? Please guide me . Thanks Harvestman website: http://harvestman.freezope.org/

Go to http://fedoraproject.org/wiki/Extras

/B

Hello:

I had AIGLX working on FC5 using the aiglx.repo for rstrode's directory. I'm now having problems using this new aiglx.repo. I dropped it into my /etc/yum.repos.d today and now my XServer is kaputt. I get the following messages after a failure to load at GDM.

Module Loader Present ... (EE) module ABI minor version (5) is newer than the server's version (4) (EE) Failed to load module "bitmap" (module requirement mismatch, 0)

Fatal Server error: Unable to load required base modules, Exiting...

Does anyone have any idea what happened? I've got the [updates-testing] repo also activated.

I would like to get AIGLX working again. I don't think the experimental screen magnifier was available on the old AIGLX I had running.

Benjy

On 4/19/06, Kristian Høgsberg krh@redhat.com wrote:

Hi,

I've set up a yum repository with AIGLX packages for FC5. It's on download.fedora.redhat.com:

http://download.fedora.redhat.com/pub/fedora/projects/aiglx

and there's a aiglx.repo file there that can be dropped into /etc/yum.repos.d. With that,

    $ yum --enablerepo=aiglx update

should pull down the AIGLX package set. The packages are just the rawhide RPMs rebuilt for FC5, so they're a little rough around the edges. That said, if the metacity compositor isn't enabled (see below), things should be pretty stable.

Here's an overview of the features and tweakable settings in the current versions of the repository packages. Since rawhide has the same packages the instructions below can also be used with a recent rawhide system:

• metacity with compositing manager functionality. Turn it on using

gconf-editor by enabling /apps/metacity/general/compositing_manager or alternatively, use

    $ gconftool-2 -s /apps/metacity/general/compositing_manager --type bool

true

• the metacity build in the repo has wobbly windows as an optional

feature. Set the environment variable USE_WOBBLY to 1 and restart metacity. For example:

    $ USE_WOBBLY=1 metacity --replace &

• metacity also has an experimental screen magnifier built in. Like

the wobbly windows this is enabled by setting an environment variable - USE_MAGNIFIER:

    $ USE_MAGNIFIER=1 metacity --replace &

• real translucency in gnome-terminal. Enable this by clicking the

"Transparent background" radio box in the "Effects" tab of the termnal profile editor dialog. gnome-terminal may have to be restarted for this to take effect - pkill gnome-terminal should do the trick.

There are a number of known bugs with these packages, so go easy on bugzilla :). Specifically,

• The damage events doesn't always kick in, so sometimes window

contents doesn't update properly. A workaround for this is to switch desktop back and forth.

• The drop shadows look weird on shaped/argb windows, for example

xeyes, the notify bubbles and well, even the rounded metacity corners. The shadow code is due for an overhaul later, so we're not going to patch over this issue in the short term.

• Perfomance for in-window updates isn't always great. We're

currently doing too much work when updating textures from pixmaps. It's still possible to optimize this further with the current setup, and longer term the memory architecture in the X server and DRI stack will see some changes that will allow us to optimize this further.

Have fun, Kristian

-- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list