Re: Thoughts on Debian quality, including automated testing
Thomas Hood <firstname.lastname@example.org> wrote:
> Under most of these topics Lars discussed automated testing. Are
> there objections to Lars's concrete proposals (e.g., standardization
> on a way to invoke package specific tests)? Are there other ideas?
> Should Debian do more auditing, for example?
I'm all for automated testing, but I want to support Lars' point that
the burden should be mostly on the people writing the automatisms, not
on individual package maintainers reinventing the wheel.
> Then what other things
> can be done to help individual maintainers fix more bugs and fix them
It would be good if there was a way to find out "problematic" packages,
by extracting information about
- how many bugs does a package have
- how many of them don't have a single response
- how many have not been dealt with for n months (or days/weeks for RC
- how many packages depend on the package
and try to create some rating or ordering. One could then not only
identify packages that could use some work, but also for which of them
it's most useful.
Another good thing would be an effort to go through important packages'
ancient bugs and clear them up. E.g. all those, mostly years-old dpkg
bugs that report a segfault. Either something should be done about
them, or they should be closed. My impression is that dpkg is actively
maintained currently, but had some problems in the past, and the current
maintainers don't have the time to clean up this inherited mess. It
seems to me that this is a problem in other packages, too. teTeX is one
Inst. f. Biochemie der Univ. Zürich