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

[Raphael Geissert]

Hi Zack,

On 22 March 2013 14:18, Stefano Zacchiroli <zack at debian.org> wrote:
> OK, thanks for clarifying. It's not clear to me, however, what you
> currently propose as next action for all this. Are you saying that you'd
> rather prefer that no further extensions to DACA current architecture
> (new static analyzers, metrics, etc.) are added, and that you rather
> prefer re-doing the DACA architecture based on Hadoop & co?

I'd rather do the latter. Doing the former would work only for a
little while (a few months, maybe a year), since it is not scalable.
Some quick patching and hacking could be done to make it slightly more
scalable and eventually turn it into something that "can do the job".
It's not impossible.
My main point would be that if somebody can already do the job of
exploring the use of Hadoop and a minimal implementation then it would
be better to spend some time on it instead of working and patching an
existing architecture. But it will always depend on the priorities of
the different projects :)

> For what concern the Debian/Coccinelle work at IRILL, I do not think we
> are going to have enough manpower to invest in redoing DACA architecture
> based on Hadoop. We could either set to contribute the Coccinelle
> integration if someone else is working on the architecture part; or we
> could just develop something specifically for our Coccinelle need, to be
> run on our own hardware, and later on work into integrating it into
> DACA-next-generation; or, last option, we can contribute integration
> with the current DACA architecture of what we're going to work on.
> Either of this options is suboptimal, I know, but that's what we can
> offer, at least in the short term.

Yep, just like I said. What can easily be done to "integrate"
Coccinelle's results would be to generate some html pages and put them
under the qa.d.o/daca/ namespace.
The only blocker for doing that is the fact that daca/ is currently a
symlink to a directory in my home dir. I haven't moved it to
/srv/qa.d.o due to the qa user/group ownership that would be required.

> So, if you think the DACA-next-generation architecture is going to
> happen, we can collaborate on that, otherwise I fear we'll have to
> choose the least bad options of the above.

I'm not sure how much time and when I could work on it, so you choose.
Since I've a bit more experience using gearman I could probably put
something better than the current architecture with it, but data
storage and other things would still need to be addressed.


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