[10:39] whee, got my hardy t-shirt [10:39] and other merchandise [10:39] too bad that both coffee cups were broken [10:39] damn UPS :) [10:51] tjaalton: no coffee then :-P [10:52] right, tea is better :) [11:13] bryce: btw, should you drop gutsy from the versions_current.html page, and add intrepid? maybe even put intrepid in the middle to minimize the need to scroll :) [14:12] If someone has some spare time, I would like some assistance in setting up a USB to VGA adepter which I can't get to work. [14:13] sorry, I need to finally fix lrm/nvidia [14:33] tseliot: I'm about to fix nvidia diverting/symlinking libwfb, which should fix at least bugs like "pink shadows with compiz" and "FF crashes on certain sites" :P [14:34] tjaalton: what was causing the problem? [14:35] it should use the libwfb.so provided by the server [14:35] and not divert it and symlink against the nvidia one [14:35] "In practice, I don't think that was communicated clearly enough to the distributions" [14:35] said aaronp [14:36] since my packages are a customised version of the lrm they are affected by this problem. I'll have a look at the packaging scripts [14:37] I changed preinst to remove the divert instead of adding one, and rules no longer symlinks that [14:37] should be enough [14:40] tjaalton: this line, right? dpkg-divert --add --rename --package nvidia-glx@@NV_LEGACY@@-envy --divert /usr/lib/nvidia/libwfb.so.xserver-xorg-core /usr/lib/xorg/modules/libwfb.so > /dev/null [14:40] yes [14:41] I'll put a debdiff somewhere soonish, so you can review it and grab what you need [14:41] there's also bug 118605 which is simple [14:41] to fix [14:41] ah, no ubotu [14:43] yes, I would like to read the debdiff [14:44] tjaalton: in the rules we should install the wfb without making the symlink, right? [14:45] tseliot: well, I changed the symlink libwfb.so -> libnvidia-wfb.so.1, like the installer apparently does [14:46] tjaalton: maybe I should file a bugreport against my packages too: [14:46] https://bugs.launchpad.net/ubuntu/+source/linux-restricted-modules-envy-2.6.24 [14:48] can someone please take a look at this 6 day bug: 224479 [14:49] or just link to them, bugs 212648 and 186382 [14:49] we need ubotu here [14:50] tjaalton: my source code is a bit different [14:51] but you have the same bugs, no? [14:52] only this bug AFAIK [14:53] the atieventsd.sh from fglrx is buggy, so 118605 affects you too [14:53] uh, authatieventsd.sh [14:56] :-/ is this the patch which you included? [14:56] http://launchpadlibrarian.net/12737164/authatieventsd.patch [14:56] yes [14:57] my packages were introduced yesterday and I already have 2 bugs to fix ;) [14:57] there are also a number of upgrade bugs from gutsy [14:58] against xorg [14:58] I mean filed against xorg currently [14:58] how can I help? [15:00] search them and file against the correct package :) [15:00] I'm not sure if there's anything to fix though, unless you want to update the old packages [15:01] ok [15:01] let me know when the debdiff is ready [15:01] please ;) [15:02] sure, I need to test it first [15:17] komputes: I replied to the bugreport [15:18] tseliot: thanks [15:43] tseliot: http://users.tkk.fi/~tjaalton/dpkg/lrm-diff [15:43] not tested yet though [15:44] I'll have a look at it. Thanks :-) === ubottu changed the topic of #ubuntu-x to: 06 May 21:00 UTC: Community Council | 07 May 21:00 UTC: Server Team | 08 May 13:00 UTC: Desktop Team | 09 May 04:00 UTC: MOTU | 14 May 21:00 UTC: Server Team | 15 May 13:00 UTC: Desktop Team [16:16] damn bot :) [16:18] tjaalton: I think it would be wise to remove the reference to the wfb in the postrm.in too even though the script looks for the diversion [16:18] and there's another thing [16:18] why? [16:18] better keep it there for awhile [16:19] if we don't divert the wfb why should we try to remove it? Anyway this won't break anything [16:19] there's another potential problem [16:20] in the preinst.in === tjaalton changed the topic of #ubuntu-x to: Ubuntu 8.04 released! | https://wiki.ubuntu.com/X [16:22] let me put a few things on pastebin === ubottu changed the topic of #ubuntu-x to: 06 May 21:00 UTC: Community Council | 07 May 21:00 UTC: Server Team | 08 May 13:00 UTC: Desktop Team | 09 May 04:00 UTC: MOTU | 14 May 21:00 UTC: Server Team | 15 May 13:00 UTC: Desktop Team [16:24] tjaalton: http://pastebin.com/mee4a53f [16:24] for example, have a look at line 63 [16:25] I check and remove diversions created by both me packages and the standard lrm [16:25] those are legacy === tjaalton changed the topic of #ubuntu-x to: Ubuntu 8.04 released! | https://wiki.ubuntu.com/X [16:27] have a look at line 121 etc. [16:27] yes? [16:27] my packages remove those diversions when you uninstall them [16:28] maybe you could do the same with the original lrm [16:28] just in case my diversions are not removed [16:28] sorry, I don't understand [16:28] those are removed by lrm too [16:28] ok, an example: [16:29] if nvidia-glx-envy creates a diversion [16:29] and for some weird reason this is not removed [16:29] your preinst will look for diversions made by nvidia-glx with grep [16:30] and will catch the diversions made by nvidia-glx-envy too [16:30] right? [16:30] I bet you don't divert /usr/X11R6/lib/libGL.so.1? [16:30] but will try to remove only the diversions made by nvidia-glx [16:31] dpkg-divert --add --rename --package nvidia-glx@@NV_LEGACY@@-envy --divert /usr/lib/nvidia/libGL.so.1.xlibmesa /usr/lib/libGL.so.1 > /dev/null [16:32] and the grep doesn't match that [16:33] doesn't match what? [16:33] line 63 [16:34] doesn't match that file [16:34] er, diversion [16:35] I still fail to see the problem :) [16:35] yes, I know. It was just an example [16:36] I just want to make sure that the lrm and lrm-envy can coexist [16:36] which they do, at least here [16:36] maybe the lrm versions should be dropped [16:37] no point in duplicating all this hackery [16:37] this will be solved in Intrepid [16:37] I need the lrm-envy to work with DKMS [16:38] which is something I can't do with lrm on a stable release [16:38] right? [16:38] right [16:38] when the users upgrade to Intrepid [16:39] if they use the lrm-envy, those packages will be replaced by the lrm [16:39] which (hopefully) we'll improve at the UDS [16:40] I'm not a big fan of duplication of efforts, really [16:41] good [16:41] :) [16:44] tjaalton: BTW this is the preinst.in which I use for my packages: http://www.albertomilone.com/nvidia-glx-envy.preinst.in [16:46] pretty much identical to the one in lrm [16:46] yes, exactly, I can keep it that way [16:49] tjaalton: your patch looks good. I'll test it. [16:52] tjaalton: I submitted a couple of small PS3 related patches for dexconf in xorg to the ubuntu-x mailing list over the week end. Are they ok for inclusion? [16:54] munckfish: the updated fb check looks fine, but I'm not sure about the other one. there should be a fix for the server somewhere [16:54] although it's simple [16:56] tjaalton: sure fine. Lets leave the second one for now then. [17:26] tjaalton: a small fix to the patch for fglrx [17:27] the first 2 lines of that patch should be: [17:27] --- common/etc/ati/authatieventsd.sh 2008-03-19 10:56:01.568196236 +0100 [17:27] +++ common/etc/ati/authatieventsd.sh 2008-03-19 11:04:31.272102426 +0100 [17:27] otherwise it won't find the file [17:27] fixed in the debdiff [17:28] ok, perfect [17:28] I'm building the packages right now [17:30] mine are almost done [17:33] mine are done. I'll test them on my ATI card on my testing box [17:33] you have KDM? [17:33] no, but I can install it [17:34] shall I install KDM and use it [17:34] ? [17:34] that's the way to trigger the bug [17:34] and test that the fix works [17:34] but it should be tested [17:34] ok, I'll do it [17:34] yes, of course [17:35] this is why we have -proposed [17:35] ok, the libwfb.so symlink needs to be removed in preinst [17:35] dpkg-divert: rename involves overwriting `/usr/lib/xorg/modules/libwfb.so' with different file `/usr/lib/nvidia/libwfb.so.xserver-xorg-core', not allowed [17:36] ouch [17:37] shall we test its existence with [ -h /usr/lib/xorg/modules/libwfb.so ] ? [17:37] and remove it? [17:38] or is it always a symlink? [17:38] probably [17:38] no [17:38] ok then we can check that it's a symlink and then remove it [17:39] yeah, shadows work [17:41] something like this? [17:41] if [ -h /usr/lib/xorg/modules/libwfb.so ]; then [17:41] rm -f /usr/lib/xorg/modules/libwfb.so [17:41] fi [17:41] but better indented :-P [17:42] I know, the -f is useless [17:42] that should work [17:43] I'll put it before the if [ "$(dpkg-divert --list /usr/lib/xorg/modules/libwfb.so | awk '{ print $7 }')" = "nvidia-glx@@NV_LEGACY@@" ] [17:44] since I will have to remove also the diversion created by my current lrm-envy [17:47] I'm building the packages again [17:51] I'm wondering if the error message means something else.. [17:52] which error? [17:52] dpkg-divet [17:52] rt [17:52] do you still get that error despite my suggestion? [17:53] I haven't tried, it takes a while to build [17:57] maybe the error is just misleading, I'll build the new package now [18:00] I'll test it here too [18:16] works! [18:16] phew [18:19] er, no [18:20] can't find libwfb.so :) [18:21] dpkg-divert --remove didn't move the library back in place [18:21] probably since the symlink was removed, hrmh [18:23] -> [18:23] mmm... [18:29] tjaalton: shall we add this line: cp -f /usr/lib/nvidia/libwfb.so.xserver-xorg-core /usr/lib/xorg/modules/libwfb.so [18:29] and remove this one? [18:29] rm -f /usr/lib/xorg/modules/libwfb.so [18:30] so that both files are available (and are identical but with different names) [18:30] and that the diversion will be removed [18:31] but of course we should do something like if [ -f /usr/lib/nvidia/libwfb.so.xserver-xorg-core ]; then [18:52] tjaalton: damn, the fglrx driver locked up when I logged out with KDM (despite the patch) [18:53] I'll have a look at the log [19:08] tjaalton: nothing interesting in the log. Sigh. I have asked ATI's mailing list for news on this bug [19:08] that patch doesn't work for me :-( [19:35] I think the better solution would be to remove the diversion in postinst, then the old libwfb.so link should be out of the way anyway [19:35] ie. removed with the old package [19:35] I'll test that tomorrow [19:42] tjaalton: ok, I'll let you know if ATI's staff knows the solution to the other problem [19:53] tjaalton: ok I'll fix up versions_current and status_current today-ish [20:17] bryce: ok cool [20:23] tjaalton: I also went through and unsubbed us from all the obsolete packages (xserver-xorg-driver-*, etc.) so the page will be more concise [20:30] bryce: ooh, nice [20:31] http://people.ubuntu.com/~bryce/Xorg/versions_current.html has intrepid on it now (but blank) [20:32] http://people.ubuntu.com/~bryce/Xorg/status_current.html is intrepified [20:36] tjaalton: btw, I'm probably going to go through and WONTFIX a ton of displayconfig-gtk bugs within the next week or so. I've prepared a page explaining the situation: https://wiki.ubuntu.com/X/DisplayConfigGtk [20:37] also, I've been thinking about the older lrm packages and their bugs. I'm thinking there's a vanishingly small chance we'd actually fix any pre-hardy lrm issues, so am thinking we should go through and wontfix them or move them to 2.6.24 if they still look valid issues in hardy [20:38] then once the older lrm packages are cleared of X bugs, we could unsub x-swat from those packages, and not have all the non-x bugs in them counting against us [20:38] but what do you think? [21:03] tjaalton: why do we have to remove the diversion in the preinst when such diversion is already removed in the postrm? [21:04] when, say, nvidia-glx is removed that diversion is removed too. Therefore we only have to make sure that the new nvidia-glx doesn't create that diversion any longer. [21:06] * tseliot knows that he won't sleep because of this problem :-P [21:12] tseliot: :-) (been there) [21:19] Okay, so X crashed on me twice. But, now I can't recreate it. [21:19] Is it worth filing a bug, or is that just going to be a waste of everyone's time? [21:20] if you have a backtrace of the crash, it can be investigated [21:20] I have the backtrace from the Xorg.0.log, does that count? [21:22] yep, that's good [21:23] often those backtraces aren't detailed enough, but sometimes they're enough to identify if the bug is a dupe of another bug or something [21:23] ted1: you should enable apport if you are still running hardy [21:24] Does it get disabled? [21:24] I didn't think it got ever turned off, but I was figuring that X would be too low level for it. [21:29] ted1: it's disable in stable to not annoy users [21:30] ted1: you need to edit /etc/defaults/apport to enable it [21:32] tseliot: because we want that diversion to disappear on upgrade [21:33] and I said postinst, preinst apparently is not the place to do it [21:33] seb128: I didn't know that, thanks. It is on now. [21:34] bryce: yes, I've been thinking of doing the same (closing pre 2.6.24 lrm bugs) [21:35] then when fglrx/nvidia is ripped from lrm we would have much more chance of actually knowing what are the important issues, now they are only buried in the noise [21:38] fedora also includes debugging symbols during development phase (or at least they used to do that). don't know how feasible that would be.. [21:39] (re: apport, stable release) [21:42] * bryce nods [21:44] tjaalton: you mean they don't strip unstable builds? [21:44] seb128: I'm not sure how it's done, I'll just ask :) [21:45] that would not be possible in ubuntu [21:45] it would mean rebuilding the whole archive before release, right? [21:45] having ubuntu fitting on one cd is already a fight most of the time [21:45] oh right [21:45] and that too [21:46] apport and the retracers usually work alright [21:48] the tough cases will always be painful to debug [21:48] I guess no tools help there [21:48] right, tools don't do everything for you [21:52] tjaalton: do you know if there's a way we could configure X to print out full backtraces in Xorg.0.log instead of the reduced backtraces? [21:57] bryce: I wonder what --enable-debug does [21:58] tjaalton: any ideas on this apt-get update error - http://pastebin.ubuntu.com/10611/ ? [21:58] "Hash Sum mismatch" [21:59] ah [21:59] bryce: -logverbose 5 perhaps? [21:59] bryce: a busy mirror I guess [21:59] our mirror got hosed because it used to mirror a.u.c [21:59] bryce: I'm getting crummy access to the US mirror right now. It keeps jumping IPs. [21:59] ah, hmm, so would switching to a different mirror do it? [22:00] I switched apt-mirror to use se.a.u.c, works nicely [22:00] and current too [22:00] ok [22:01] ahh much better [22:05] yay - intrepified: http://people.ubuntu.com/~bryce/Xorg/versions_current.html [22:05] lunch. bbiab [22:08] sweet, a dynamic page :) [22:35] great, mesa-7.1 coming soon [22:35] maybe by UDS [22:35] sweet [23:03] * bryce fusses with css [23:24] ok, this looks fairly good... http://people.ubuntu.com/~bryce/Xorg/versions_current.html [23:25] cool, you dropped all the apps that are in bundles [23:25] yup