| superm1 | bryce, ping. i've been suddenly seeing a few bugs getting reported with mythfrontend shooting out BadMatch and RenderBadPicture on AMD & NV open source drivers. Intel & closed drivers appear to be fine. i was hoping you might have some pointers to how to really debug these? https://bugs.edge.launchpad.net/ubuntu/+source/mythtv/+bug/341898/comments/4 | 03:38 |
|---|---|---|
| ubottu | Ubuntu bug 341898 in mythtv "Mythtv frontend does not display any fonts" [Undecided,New] | 03:38 |
| bryce | superm1: that's a client/server protocol communication error, that indicates a bad parameter was passed (Picture), so you could use xtrace (or xscope or xmond) to analyze the traffic between client and server | 04:21 |
| bryce | the error could either be that the API has changed (unlikely), or that something in those drivers or the xserver is generating invalid (NULL?) values for that parameter for some reason | 04:21 |
| superm1 | bryce, has there been any major changes in the protocols that a QT application such as myth should be changing to reflect? | 04:21 |
| superm1 | since posting that ping someone reported that it failed in VirtualBox too, so i'm guessing vesa is also affected | 04:22 |
| superm1 | i'll see about using some of those tools to look at the traffic between the client and server then, thanks | 04:22 |
| bryce | well, -nv has not changed significantly, and -vesa certainly hasn't | 04:22 |
| bryce | so if it is being seen as a regression with those two drivers, I'd tend to suspect xserver or one of its libs instead | 04:23 |
| superm1 | yeah i had thought that too, perhaps the closed drivers were diverting one of the libs that changed or something so they weren't affected, but Intel working confused me; | 04:23 |
| superm1 | yeah it's definitely outside of mythtv code. it's coming in calls for QT to show() it's widgets. i'd suspect this is going to crop up in other apps that use QT3 too then? | 06:45 |
| RAOF | tjaalton: Oh. The reason libdrm-dev only depends on libdrm-nouveau1 on x86 is a thinko; that qualifier should be dropped. Urgh. | 07:31 |
| dholbach | hiya | 08:02 |
| dholbach | jcristau: I was just chatting with geser about the x11proto-xext-dev / libxi-dev thing - do you know if upstream is aware of it and thinking of a better solution to it? | 08:03 |
| dholbach | (debian bug 499858) | 08:03 |
| ubottu | Debian bug 499858 in x11proto-xext-dev "x11proto-xext-dev: Missing libxi-dev dependency" [Important,Open] http://bugs.debian.org/499858 | 08:03 |
| === crdlb_ is now known as crdlb | ||
| jcristau | dholbach: dunno. how many packages are affected? | 09:18 |
| jcristau | dholbach: actually... apps that want to include XTest.h probably use libxtst, so libxtst-dev Depends: libxi-dev would seem to make sense. | 10:48 |
| dholbach | jcristau: I'm not an X expert - far from it - but libxtst-dev does not include XTest.h | 10:50 |
| jcristau | dholbach: XTest.h provides the prototypes for libXtst | 10:52 |
| dholbach | jcristau: just trying to find out how many packages there are - libfakekey is another candidate | 11:07 |
| jcristau | also build-deps on libxtst-dev | 11:08 |
| dholbach | ok | 11:09 |
| dholbach | jcristau: libxtst would still need to workaround it though, but I'll try to collect some more data | 11:09 |
| dholbach | jcristau: this is what I found using a very unscientific method: http://paste.ubuntu.com/135963/ | 11:38 |
| dholbach | my gut feeling says there's lots more :-/ | 11:38 |
| dholbach | making libxtst-dev Depends on libxi-dev would fix most of them, I guess | 11:39 |
| jcristau | ratpoison has a libxtst-dev b-dep | 11:54 |
| jcristau | so i think you should go for that | 11:54 |
| dholbach | jcristau: you're right - I overlooked it | 11:55 |
| dholbach | thanks | 11:55 |
| dholbach | jcristau: I'll talk to bryce about it - thanks a bunch! | 11:58 |
| jcristau | np | 11:58 |
| jcristau | sorry i didn't think of that solution earlier.. | 12:01 |
| dholbach | jcristau: don't worry - you guys do fantastic work anyway :-) | 12:01 |
| dholbach | jcristau: are you going to do it in Debian too? | 12:17 |
| dholbach | (libxtst build-depends on libxi-dev, lixtst-dev depends on libxi-dev) | 12:18 |
| dholbach | in that case we should be able to sync the package :-) | 12:18 |
| jcristau | XInput.h is in x11proto-input-dev in debian still | 12:19 |
| dholbach | ahhh! | 12:20 |
| jcristau | i suppose we could do that change anyway, but.. | 12:20 |
| dholbach | ok, in that case, I'll just upload it and let bryce fix whatever needs fixing after the release :) | 12:20 |
| dholbach | jcristau: I just stumbled across it when I sponsoring a few bugfixes :) | 12:20 |
| jcristau | yeah i'm surprised this didn't come up earlier | 12:21 |
| dholbach | same here | 12:21 |
| tjaalton | it was moved in intrepid because of the input-properties changes, but then upstream moved it back for XI 1.5 | 12:21 |
| dholbach | we should have more regular rebuild tests, if you ask me :) | 12:21 |
| tjaalton | so we're just waiting for XI2 and things should be in sync again :) | 12:21 |
| dholbach | tjaalton: I'll upload the workaround in a bit, just FYI | 12:23 |
| tjaalton | dholbach: ok | 12:24 |
| dholbach | ok my friends - that was it from me - take care :-) | 12:26 |
| === seb128_ is now known as seb128 | ||
| === maco_ is now known as maco | ||
| bryce | http://news.softpedia.com/news/Goodbye-Ubuntu-7-10-107519.shtml | 23:50 |
| bryce | I'm going to go ahead and unsub us from lrm-2.6.22 at this point, even though it's a few more weeks until 7.10 is officially unsupported. | 23:51 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!