2007-04-25 13:21:06 |
Martijn Faassen |
bug |
|
|
added bug |
2007-05-24 09:28:26 |
Martin Pitt |
apport: status |
Unconfirmed |
Confirmed |
|
2007-05-24 09:28:26 |
Martin Pitt |
apport: statusexplanation |
|
Indeed the current hook already does not create a report for scripts in your home directory:
if not apport.fileutils.likely_packaged(binary):
return
which returns true if the executable path starts with the usual system directories (/usr/bin, /lib, etc.).
So, if I understood you correctly, 'import apport.report' is too heavy here? The code can certainly be reshuffled to postpone that import until after above check, which would reduce the amount of parsed code for an exception in non-packaged Python programs. |
|
2007-05-24 09:28:35 |
Martin Pitt |
apport: status |
Confirmed |
In Progress |
|
2007-05-24 09:28:35 |
Martin Pitt |
apport: statusexplanation |
Indeed the current hook already does not create a report for scripts in your home directory:
if not apport.fileutils.likely_packaged(binary):
return
which returns true if the executable path starts with the usual system directories (/usr/bin, /lib, etc.).
So, if I understood you correctly, 'import apport.report' is too heavy here? The code can certainly be reshuffled to postpone that import until after above check, which would reduce the amount of parsed code for an exception in non-packaged Python programs. |
|
|
2007-05-24 09:39:10 |
Martin Pitt |
apport: status |
In Progress |
Fix Committed |
|
2007-05-24 09:39:10 |
Martin Pitt |
apport: assignee |
|
pitti |
|
2007-05-24 10:16:29 |
Martin Pitt |
apport: status |
Fix Committed |
Fix Released |
|
2007-05-24 10:16:29 |
Martin Pitt |
apport: statusexplanation |
|
apport (0.81) gutsy; urgency=low
.
* apport/report.py: Remove '[apport]' default bug title prefix. (LP: #94819)
* apport/crashdb_impl/launchpad.py: Tag new bugs with
'apport-<problemtype>'. This replaces the former '[apport]' prefixing.
* debian/local/setup-apport-retracer: Specify a path in '.' command and
use sh again. Yay for me needing three attempts before actually RTFMing
how '.' works (which is really nasty and strange IMHO).
* bin/apport-chroot: Fix symlinks before repackaging the chroot tarball in
'install' and 'installdeb' modes.
* debian/local/setup-apport-retracer: Install python-libxml2 and python-apt.
* bin/launchpad-crash-digger: Supply --auth instead of the deprecated
--cookie to apport-chroot.
* bin/apport-chroot: Fix identifier name in command_retrace().
* debian/local/setup-apport-retracer: Set APPORT_CRASHDB_CONF to the local
crashdb.conf.
* bin/apport-chroot: Unset APPORT_CRASHDB_CONF for login and retrace.
* bin/launchpad-crash-digger: Check the release of a bug and whether we have
a chroot for it before untagging it. This avoids loosing tags for bugs we
do not yet have a working retracer chroot for.
* bin/apport-retrace: Do not abort with an exception if package installation
fails. Give a proper error message instead and point to -u. (LP: #115681)
* apport/crashdb_impl/launchpad.py, update(): Create a temporary directory
and use proper file names for the new attachments. With TemporaryFile(),
attachment file names ended up as '<fdopen>'. (LP: #115347)
* apport/report.py, add_os_info(): Add field 'NonfreeKernelModules' which
lists loaded kernel modules which do not have a FOSS license. This is
particularly helpful for quickly checking for restricted graphics drivers.
(LP: #103239)
* apport_python_hook.py: Move the apport.* imports into the try: block and
move the likely_packaged() test to the top, to avoid importing
apport.report and creating a Report object for non-packaged scripts. This
makes the entire code more efficient and robust against errors in the
apport modules. (LP: #109955)
* apport/report.py, add_gdb_info(): Intercept OSError from gdb invocation
(which might be segfaulting itself) and just do not put any gdb output
into the report. The automatic retracers can try their luck again.
(LP: #112501)
* bin/apport-retrace: Fix handling of packages which are still known to
/var/lib/dpkg/status, but do not have an apt record any more; treat them
like virtual packages and just issue a warning instead of falling over.
(LP: #107474)
|
|
2009-07-08 22:40:16 |
Launchpad Janitor |
branch linked |
|
lp:ubuntu/karmic/apport |
|
2012-07-06 05:21:08 |
Launchpad Janitor |
branch linked |
|
lp:debian/experimental/apport |
|