[Daca-general] scan-build and metrics gsoc proposals and DACA

[Raphael Geissert]

Hi Michael, everyone,

[Adding paultag and holger to CC, as they apparently talked about
jenkins.d.n and build.me the other day]

On 18 March 2013 02:49, Michael Tautschnig <mt at debian.org> wrote:
> [Keeping Sylvestre and Zack in CC, not sure you are subscribed]

Ditto

> For a lack of time, I haven't yet looked into the GSoC proposals. Yet I am also
> facing the very same issues in my current research activities, where I'm trying
> to run our software verification tools on all the packages in Debian.  There is
> still quite a bit of work ahead of me until I eventually get there, but at least
> automated builds using our own research compiler infrastructure are happening
> (resulting in some >100 bug reports..., usertagged goto-cc).

I just took a look at some of them and they seem very useful. Are
those bug reports the result of manual verification? or more
precisely, has there been a need of manual verification to avoid
filing false positives or are all the issues detected known (or even
proved) to be true positives?
And is it possible that such tool will eventually be publicly available?

> I really see data processing as the biggest challenge, and I don't yet have a
> solution. But there are a few things I could contribute:
>
> - One of my students (MSc project, not GSoC) is going to look into this over the
>   next few months. I will be happy to share the results.

That'd be great, thanks!

> - I'd like to investigate whether jenkins (with auto-generated jobs) could be
>   used to take care of all the scheduling and triggering bits, plus
>   notifications, etc. This would immediately come with the potential for scaling
>   to build farms, as jenkins natively supports a master/slave setup.
> - I am currently working towards acquiring hardware via the university. Once
>   I've got some (more), I'll play with a jenkins setup. If this succeeds, I'll
>   happily share it.

Would your student be focusing on jenkins only, or would she/he
explore alternative solutions first (at least as "related work") and
then focus on jenkins?
Yesterday I mentioned in #-irc that I don't see jenkins as a generic
solution for all the problems at hand. Especially when focusing on
more than just one tool; but that's just my impression, I haven't
actually tried to do it.

Holger, Paul: perhaps you are interested in the above.

Cheers,
-- 
Raphael Geissert - Debian Developer
www.debian.org - get.debian.net