desktop-exec crashes when using gir bindings to launch a non-existant app

[Thomi Richards]

When I use the gir bindings for libUAL like so:

from gi.repository import UpstartAppLaunch

UpstartAppLaunch.start_application('sadfsdf')

The 'start_application' method returns True, and the crash attached to this bug occurs. Ideally, start_application (and start_applicaiton_test) should return False if the application cannot be started.

ProblemType: Crash
DistroRelease: Ubuntu 14.04
Package: upstart-app-launch 0.3+14.04.20140220-0ubuntu3
ProcVersionSignature: Ubuntu 3.13.0-12.32-generic 3.13.4
Uname: Linux 3.13.0-12-generic x86_64
NonfreeKernelModules: nvidia
ApportVersion: 2.13.2-0ubuntu5
Architecture: amd64
CurrentDesktop: Unity
Date: Thu Feb 27 08:31:37 2014
ExecutablePath: /usr/lib/x86_64-linux-gnu/upstart-app-launch/desktop-exec
InstallationDate: Installed on 2014-01-06 (51 days ago)
InstallationMedia: Ubuntu 14.04 LTS "Trusty Tahr" - Alpha amd64 (20140105)
ProcCmdline: /usr/lib/x86_64-linux-gnu/upstart-app-launch/desktop-exec
ProcEnviron:
 LANGUAGE=en_NZ:en
 PATH=(custom, user)
 XDG_RUNTIME_DIR=<set>
 LANG=en_NZ.UTF-8
 SHELL=/bin/bash
Signal: 5
SourcePackage: upstart-app-launch
StacktraceTop: main ()
Title: desktop-exec crashed with signal 5 in main()
UpgradeStatus: No upgrade log present (probably fresh install)
UserGroups: adm autopilot cdrom dip lpadmin plugdev sambashare sudo

[Apport retracing service (apport)]

StacktraceTop: main (argc=<optimized out>, argv=<optimized out>) at /build/buildd/upstart-app-launch-0.3+14.04.20140220/desktop-exec.c:68

[Ted Gould (ted)]

So there are two parts to this bug:

What libUAL is doing is saying "was I able to ask upstart to start this application?" and returning whether it was able to do that. If the application was started or failed to start is signaled in the started or failed signals. I think this is useful as it allows the distinction of what happened in a relatively low level library. Perhaps higher level functions are needed, I'm not sure.

Aborting on invalid app ids. I think that it's important that we get bug reports on invalid appids being sent to UAL, it shouldn't happen and is a sign of a bug in higher levels of the stack. Perhaps it should be a recoverable error instead of a crash, but really they're both just ways of getting the info into whoopsie. More than happy to discuss ideas on how to better report this, but I do think it needs to be reported so that we start to see those higher level bugs.