| mgedmin | apport is hilariously broken in 15.04 | 05:55 |
|---|---|---|
| mgedmin | the stack trace ends up in the Disassembly: field because I think it gets confused by gdb's output | 05:56 |
| Noskcaj | darkxst, Can you please renew my membership to ubuntu-gnome-leaders-board? | 05:58 |
| darkxst | Noskcaj, done | 06:01 |
| Noskcaj | ty | 06:02 |
| darkxst | mgedmin, that would be a pitti Q | 06:02 |
| mgedmin | yeah, maybe I should file a bug report | 06:02 |
| mgedmin | unfortunately "apport is hilariously broken, here, look at this: " is not a good bug report | 06:02 |
| mgedmin | Stacktrace: No symbol "__nih_abort_msg" in current context. | 06:03 |
| mgedmin | SegvAnalysis: Failure: invalid literal for int() with base 16: '=' | 06:03 |
| mgedmin | I can't even | 06:03 |
| mgedmin | it's too early in the morning for me | 06:03 |
| mgedmin | Python Exception <class 'TypeError'> iter() returned non-iterator of type '_iterator': | 06:03 |
| mgedmin | inside the Disassembly field that actually contains the stacktrace for some reason | 06:03 |
| mgedmin | apologies for ranting here | 06:04 |
| darkxst | yeh that is wierd | 06:04 |
| mgedmin | I bet apport is parsing gdb's output and getting confused by something | 06:04 |
| mgedmin | maybe those Python errors | 06:04 |
| mgedmin | (gdb extensions written in Python 2 with gdb linked against Python 3, I guess? because nobody cares?) | 06:05 |
| darkxst | mgedmin, Ubuntu was the first distro to link gdb against python 3 | 06:05 |
| darkxst | most of the upstreams were still on 2, but think fedora may have switched now | 06:05 |
| mgedmin | porting python 2 code to python 3 is something I could probably help with | 06:06 |
| mgedmin | but where does that code live? | 06:06 |
| mgedmin | what package? | 06:06 |
| mgedmin | I filed https://bugs.launchpad.net/ubuntu/+source/gdb/+bug/1449389, got no response (as usual) | 06:06 |
| ubot5 | Ubuntu bug 1449389 in gdb (Ubuntu) "Python Exception <class 'TypeError'> iter() returned non-iterator of type '_iterator'" [Undecided,New] | 06:06 |
| darkxst | usually in dbg or dev for the relevant library | 06:07 |
| mgedmin | ok, ok, I could do my own homework, I suppose | 06:07 |
| mgedmin | can I get gdb to print a list of .py files it has loaded in this debug session? | 06:07 |
| darkxst | was that retrace on ubuntu-gnome or launchpad retracer | 06:07 |
| * mgedmin discovers, again, that apport-retrace is useless | 06:07 | |
| mgedmin | if by "that retrace" you mean the bits I quoted here | 06:08 |
| darkxst | "apport-retrace -g <crash_file>" not entirely useless | 06:08 |
| mgedmin | it's from /var/crash/_usr_bin_gnome-shell.1000.crash on my laptop | 06:08 |
| mgedmin | warning: .dynamic section for "/tmp/apport_sandbox_FQNZiQ/usr/lib/libmutter.so.0" is not at the expected address (wrong library or version mismatch?) | 06:08 |
| mgedmin | warning: Could not load shared library symbols for /usr/lib/x86_64-linux-gnu/libtxc_dxtn.so. | 06:08 |
| darkxst | pacakge form staging or archive? | 06:08 |
| mgedmin | and then the stack trace is nonsensical | 06:08 |
| mgedmin | gnome3-staging, which complicates things I'm sure | 06:08 |
| darkxst | yeh, its a different retracer | 06:09 |
| mgedmin | have I enabled ppa ddebs this time (they got disabled by the upgrade as usual IIRC)? | 06:09 |
| mgedmin | probably not... | 06:09 |
| mgedmin | symbol servers, where are they? (https://randomascii.wordpress.com/2013/03/05/symbols-on-linux-update-fedora-fixes/) | 06:10 |
| darkxst | mgedmin, if you enable debugging for the pretty-printers it will show more useful info | 06:11 |
| mgedmin | how do I do that? | 06:11 |
| mgedmin | info pretty-printer | 06:13 |
| mgedmin | no, that's useless | 06:13 |
| darkxst | I forget haven't worked with them in a while | 06:14 |
| mgedmin | info auto-load python-scripts | 06:15 |
| mgedmin | the gdb pretty-printers come from libglib2.0-0-dbg | 06:18 |
| mgedmin | it's a vivid/main package, no ppas involved | 06:19 |
| mgedmin | source package is glib2.0 | 06:19 |
| darkxst | mgedmin, launchpad retracer does not load them currently | 06:20 |
| darkxst | our retracer does | 06:20 |
| * mgedmin tries UDD again: bzr branch ubuntu:vivid/glib2.0 | 06:20 | |
| mgedmin | what do you mean by "our retracer"? | 06:20 |
| darkxst | atleast i don't think pitti merged my changes for that yet | 06:20 |
| darkxst | crashes on gnome3-staging packages are retraced by a retracer that I run | 06:21 |
| darkxst | normally PPA's dont get retraced | 06:21 |
| mgedmin | ah, for bugs filed against ubuntu-gnome? | 06:22 |
| darkxst | yes, for any package originating from gnome3-team ppa's | 06:23 |
| darkxst | gtg, be back in half an hour | 06:23 |
| mgedmin | thank you for your patience! | 06:23 |
| mgedmin | I know what needs to be fixed | 06:31 |
| mgedmin | bzr branch stiiil running | 06:31 |
| mgedmin | wheee, Vcs-Svn: svn://anonscm.debian.org/pkg-gnome/desktop/unstable/glib2.0/ | 06:32 |
| mgedmin | does Debian link gdb against python 3? | 06:33 |
| mgedmin | no! 2.7 | 06:33 |
| mgedmin | okay, bug still present upstream: https://github.com/GNOME/glib/blob/master/glib/glib.py#L61 | 06:37 |
| mgedmin | woo, last commit was by darkxst \o/ | 06:38 |
| mgedmin | ... bzr branch still running | 06:39 |
| mgedmin | suddenly I don't know if I should file a new bug or if it's ok to add more patches to https://bugzilla.gnome.org/show_bug.cgi?id=720635 | 06:41 |
| ubot5 | Gnome bug 720635 in general "Make gdb pretty-printers compatible with Python3" [Normal,Resolved: fixed] | 06:41 |
| mgedmin | git clone git://git.gnome.org/glib finished in 3 and a half minutes | 06:44 |
| mgedmin | bzr branch ubuntu:glib2.0 still running | 06:44 |
| mgedmin | (downloaded 670 megs so far!) | 06:44 |
| mgedmin | glib/.git is 52 MB | 06:44 |
| mgedmin | imagine what UDD could be if git had been invented a couple of years sooner | 06:45 |
| darkxst | https://git.gnome.org/browse/glib/commit/?id=7991178a752a22274950e54dc4f05b55ae54d756 | 06:46 |
| mgedmin | the same kind of fix needs to be applied to glib.py too | 06:46 |
| darkxst | ok, file a new bug | 06:47 |
| darkxst | I do recall fixing some of those, but maybe was in mozjs or something | 06:47 |
| darkxst | launchpad just grew git support | 06:47 |
| mgedmin | bugzilla makes filing a new bug damn inconvenient :( | 06:48 |
| darkxst | for patches I use git bz | 06:48 |
| darkxst | http://blog.launchpad.net/general/git-code-hosting-beta | 06:48 |
| mgedmin | does git bz file bugs, or just upload attachments? | 06:49 |
| darkxst | can file bugs, attach, apply and push | 06:49 |
| mgedmin | neat! | 06:50 |
| darkxst | don't know if you can file a bug without a patch though | 06:50 |
| mgedmin | I don't suppose anybody packaged git-bz for Ubuntu? | 06:50 |
| mgedmin | of course not | 06:50 |
| darkxst | its only a simple shell script | 06:51 |
| darkxst | git://git.fishsoup.net/git-bz | 06:51 |
| darkxst | drop it somewhere in your path | 06:51 |
| mgedmin | https://bugzilla.gnome.org/show_bug.cgi?id=749092 | 06:55 |
| ubot5 | Gnome bug 749092 in general "gdb pretty-printers fail on Python 3 with a TypeError: iter() returned non-iterator of type '_iterator'" [Normal,New] | 06:55 |
| mgedmin | has a patch | 06:55 |
| mgedmin | that I probably should have tested before uploading | 06:55 |
| mgedmin | yay it works | 06:59 |
| mgedmin | ok, back to the gnome-shell segfault | 07:03 |
| mgedmin | xrandr tells it monitors have changed, etc, etc | 07:03 |
| mgedmin | uh, this looks strange: | 07:03 |
| mgedmin | #14 0x00007fa9fbbd2e4a in g_signal_emit_by_name (instance=0x20, instance@entry=0x7562a0, | 07:03 |
| mgedmin | detailed_signal=0x10 <error: Cannot access memory at address 0x10>, detailed_signal@entry=0x7fa9fd284c77 "monitors-changed") | 07:03 |
| mgedmin | 0x20? 0x10? are those pointers? | 07:03 |
| darkxst | they are meant to be pointers | 07:04 |
| mgedmin | frame 3: g_list_prepend | 07:06 |
| mgedmin | info locals: new_list = 0x729990 = {0x7299a0, 0x66a4650, <error reading variable new_list (Cannot access memory at address 0x4)> | 07:06 |
| mgedmin | ah, it's executing new_list = _g_list_alloc () | 07:09 |
| mgedmin | #0 magazine_chain_pop_head (magazine_chunks=0x7299b0) at /build/buildd/glib2.0-2.44.0/./glib/gslice.c:539 | 07:10 |
| mgedmin | 539 (*magazine_chunks)->data = chunk->next; | 07:10 |
| mgedmin | and chunk is 0xe000000000000000 | 07:10 |
| mgedmin | "Cannot access memory at address 0xe000000000000000" | 07:11 |
| mgedmin | wow, apport in ubuntu has 829 open bugs | 07:17 |
| mgedmin | also, yeah, apport-retrace -g <crash file> is wonderful | 07:25 |
| mgedmin | I was complaining about the slow-and-wrong "let's download a bunch of debug debs and install them and then give you a useless backtrace" feature | 07:26 |
| mgedmin | now I know why apport is broken | 07:39 |
| darkxst | mgedmin, apport-retrace in the archives does not support ppa's | 07:40 |
| mgedmin | I'm talking about local apport | 07:40 |
| mgedmin | https://bugs.launchpad.net/ubuntu/+source/apport/+bug/1453011 | 07:40 |
| ubot5 | Ubuntu bug 1453011 in apport (Ubuntu) "SegvAnalysis: Failure: invalid literal for int() with base 16: '='" [Undecided,New] | 07:40 |
| darkxst | https://code.launchpad.net/~darkxst/apport/per-ppa-config2 | 07:41 |
| * mgedmin kind of wants to package gnome-builder | 11:08 | |
| === aaron__ is now known as ahoneybun | ||
| === adrian is now known as alvesadrian | ||
| lindol | hi all | 14:36 |
| === aaron__ is now known as ahoneybun | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!