apport-retrace strangely exits with 0 return code

[Brian Murray]

apport/sandboxutils.py contains the following code:

    # sanity check: for a packaged binary we require having the executable in
    # the sandbox; TODO: for an unpackage binary we don't currently copy its
    # potential local library dependencies (like those in build trees) into the
    # sandbox, and we call gdb/valgrind on the binary outside the sandbox.
    if 'Package' in report:
        for path in ('InterpreterPath', 'ExecutablePath'):
            if path in report and not os.path.exists(sandbox_dir + report[path]):
                apport.error('%s %s does not exist (report specified package %s)',
                             path, sandbox_dir + report[path], report['Package'])
                sys.exit(0)

I don't think apport should exit with a 0 here, as setting up the sandbox and the retrace was not successful. I discovered this when trying to retrace _usr_lib_checkbox_qt_checkbox-qt-service.2001.crash which is part of apport-test crashes.

[Martin Pitt (pitti)]

Indeed, thanks for spotting! Fixed in http://bazaar.launchpad.net/~apport-hackers/apport/trunk/revision/2968 , this also cleaned up a few other errors to consistently use apport.fatal().

[Launchpad Janitor (janitor)]

This bug was fixed in the package apport - 2.18-0ubuntu2

---------------
apport (2.18-0ubuntu2) wily; urgency=medium

  * Fix PEP-8 error in test/test_backend_apt_dpkg.py.

apport (2.18-0ubuntu1) wily; urgency=medium

  * New upstream release. Changes since our last merge from trunk:
    - apport-gtk: Use GtkWidget::valign property instead of GtkMisc::yalign
      which is deprecated in GTK 3.16. Thanks Iain Lane.
    - sandboxutils, make_sandbox(): Don't exit with 0 (success) if the
      ExecutablePath does not exist. (LP: #1462469)
    - sandboxutils, make_sandbox(): Fix second round of package installs to go
      into permanent root dir as well.
    - apt/dpkg install_packages(): If a requested package version is not
      available from apt in the given archive, try to download it from
      Launchpad. Thanks to Brian Murray!
    - kerneloops: Fix crash when determining the version of a custom kernel.
      Thanks Brian Murray. (LP: #1468154)
    - apt/dpkg install_packages(): Ignore -dbg packages whose descriptions
      contain "transitional". (LP: #1469635)
    - Keep "[origin: ...]" information in Package: and Dependencies: fields
      for native-origins.d/ origins, so that it's possible to retrace them.
      Thanks Brian Murray! (LP: #1470572)
    - Add support for retracing with discovering and enabling foreign
      Launchpad PPA origins, as specified in reports' Package:/Dependencies:
      fields. Thanks Brian Murray!
    - hookutils.attach_wifi(): Shorten value of CRDA if iw isn't available on
      the system. Thanks Brian Murray.
    - Fix wrong assertion in crashdb.test_check_duplicate() which surfaces
      under Python 3.5. (LP: #1474539)
  * test/test_backend_apt_dpkg.py: Disable new test assertion for
    unity-services-dbgsym, which doesn't currently work in the autopkgtest
    (but works fine in trunk and on the retracers). To be investigated.

 -- Martin Pitt <email address hidden> Fri, 17 Jul 2015 11:43:37 +0200