apport (0.44) feisty; urgency=low
.
Some more 'Need for Speed' optimizations:
.
* backends/packaging-dpkg.py, _check_files_md5(): Also accept a md5sum
string in addition to a md5sum file.
* backends/packaging-dpkg.py, get_modified_files(): Compare package file's
ctime and mtime against the package list file's mtime and only md5sum the
files that are newer. This drastically reduces the amount of md5suming
(usually to zero) and thus speeds up the information collection.
* backends/packaging-dpkg.py: Use a single hackish 'dpkg-query --show *'
as a portable variant of 'cat /var/lib/dpkg/status' to pre-fill the status
cache with all packages instead of calling dpkg -s on every single package
we query. This changes the time for figuring out dependencies and their
versions from 'unbearable for many packages' to 'barely noticeable'.
.
New feature: per-package apport hooks to collect additional information:
.
* apport/report.py: Add method add_hooks_info() which executes a function
add_info(report) from /usr/share/apport/<package>.py. Also add
appropriate test cases. This provides per-package hooks for apport.
* apport/ui.py: Call add_hooks_info() in the information collection thread.
.
Bug fixes:
.
* apport/report.py: Add some more test cases for _check_interpreted() for
Python scripts.
* apport/python_hook.py: Check for a correct ExecutablePath in
test_general().
* apport/python_hook.py: Use fileutils.likely_packaged() instead of
checking for /tmp and home, so that we ignore stuff in /usr/local, too.
Closes: LP#81244
* apport/python_hook.py: If we figure out an ExecutablePath which is not
actually an executable, do not create a report. This particularly affects
interactive python sessions where sys.argv[0] is empty and thus
ExecutablePath ends up being the current directory. Add test cases.
Closes: LP#81237
apport (0.44) feisty; urgency=low packaging- dpkg.py, _check_files_md5(): Also accept a md5sum packaging- dpkg.py, get_modified_ files() : Compare package file's packaging- dpkg.py: Use a single hackish 'dpkg-query --show *' dpkg/status' to pre-fill the status info(report) from /usr/share/ apport/ <package> .py. Also add interpreted( ) for python_ hook.py: Check for a correct ExecutablePath in general( ). python_ hook.py: Use fileutils. likely_ packaged( ) instead of python_ hook.py: If we figure out an ExecutablePath which is not
.
Some more 'Need for Speed' optimizations:
.
* backends/
string in addition to a md5sum file.
* backends/
ctime and mtime against the package list file's mtime and only md5sum the
files that are newer. This drastically reduces the amount of md5suming
(usually to zero) and thus speeds up the information collection.
* backends/
as a portable variant of 'cat /var/lib/
cache with all packages instead of calling dpkg -s on every single package
we query. This changes the time for figuring out dependencies and their
versions from 'unbearable for many packages' to 'barely noticeable'.
.
New feature: per-package apport hooks to collect additional information:
.
* apport/report.py: Add method add_hooks_info() which executes a function
add_
appropriate test cases. This provides per-package hooks for apport.
* apport/ui.py: Call add_hooks_info() in the information collection thread.
.
Bug fixes:
.
* apport/report.py: Add some more test cases for _check_
Python scripts.
* apport/
test_
* apport/
checking for /tmp and home, so that we ignore stuff in /usr/local, too.
Closes: LP#81244
* apport/
actually an executable, do not create a report. This particularly affects
interactive python sessions where sys.argv[0] is empty and thus
ExecutablePath ends up being the current directory. Add test cases.
Closes: LP#81237