[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Apport for Debian



Hi,

Now that Wheezy is released, I'd like to get an opinion from fellow
project members if it is okay to add apport to the Debian unstable queue
so that we can see it in time for Jessie.

Apport [1] is an automated crash reporting tool. It could also be used
as a bug reporting tool, but there are certain features missing (a text
input to add user description), not making it a candidate for reporting
bugs. For crashes, apport does a great job. It can intercept almost all
crashes on the box, including Python exceptions.


The implementation for Debian is very simple. There is an Apport CrashDB
engine for Debian, which drafts the reports, discards the binary crash
data (that can be huge), and sends it as an email to the Debian BTS. It
lacks the feature to search for duplicate bug reports (like reportbug
does, if online) on the BTS though.

Apport for Debian currently resides in experimental with a whopping
popcon stat of 11000+ installs. In the past, I have blogged [2] about
apport's state in Debian, where I received some constructive feedback.

1) Duplicate bug reports: There are high possibilities that we could see
a sudden increase in the number of bug reports, many duplicates. This is
something I'm not sure how we want to evaluate. We could give apport a
try, and leave it to the users to be more conscious when hitting submit.

2) Incomplete / Useless backtrace: This issue has been fixed. If apport
senses that the backtrace is incomplete, it will not allow for the bug
to be reported.

3) Incomplete / Useless backtrace (Missing debug symbols): Just like
point #2, if apport detects that the debug symbol packages are missing
for the package, it will deny filing the bug report.

4) Blacklisting: If a developer wants to opt-out for apport reports,
they can ship hooks into /etc/apport/blacklist.d/ .
There was request to completely blacklist a package on the basis of
package names. This is doable. I hacked a patch in the past but it
wasn't optimal. If there's a need we can work something out in a later
release. I'd like to see this feature in the core apport framework though.


I propose to see apport as part of the Debian archive. Having apport
will help us catch many unnoticed and unreported crashes.

Until some consensus is built, apport will reside in experimental only.



[1] https://launchpad.net/apport
[2] http://www.researchut.com/taxonomy/term/170



-- 
Given the large number of mailing lists I follow, I request you to CC me
in replies for quicker response

Attachment: signature.asc
Description: OpenPGP digital signature


Reply to: