Re: Potential reference miscounts and segfaults in plpython.c

On Thu, Feb 23, 2012 at 8:35 PM, Daniele Varrazzo
<daniele(dot)varrazzo(at)gmail(dot)com> wrote:
> On Mon, Feb 20, 2012 at 12:09 AM, Jan Urbański <wulczer(at)wulczer(dot)org> wrote:
>
>> BTW, that tool is quite handy, I'll have to try running it over psycopg2.
>
> Indeed. I'm having a play with it. It is reporting several issues to
> clean up (mostly on failure at module import). It's also tracebacking
> here and there: I'll send the author some feedback/patches.
>
> I'm patching psycopg in the gcc-python-plugin branch in my dev repos
> (https://github.com/dvarrazzo/psycopg).

The just released psycopg 2.4.5 has had its codebase cleaned up using
the gcc static checker.

I haven't managed to run it without false positives, however it's been
very useful to find missing return value checks and other nuances. No
live memory leak has been found: there were a few missing decrefs but
only at module init, not in the live code.

Full report about the changes in this message:
<https://fedorahosted.org/pipermail/gcc-python-plugin/2012-March/000229.html>.

Cheers,

-- Daniele