[Python-projects] Crash detected in ASTNG 0.13.1- Proposed patch included.

[Sylvain Thénault]

On Wednesday 02 May à 16:38, Pierre_Rouleau at ImpathNetworks.com wrote:
> Bonjour Sylvain,

Bonjour Pierre 
 
> I ran into a ASTNG problem when pylinting a module that imports a SWIGged
> module.
> 
> I am using:
>      pylintw.py 0.13.1,
>      astng 0.17.0, common 0.21.2
>      Python 2.4.3 (#69, Mar 29 2006, 17:35:34) [MSC v.1310 32 bit (Intel)]
> 
> The traceback is:
> 
> Traceback (most recent call last):
[snip]
>   File "c:\python24\lib\site-packages\logilab\astng\manager.py", line 197, in
> astng_from_something
>     modname = klass.__module__
> NameError: Unknown C global variable
> imcapserver.py:1: [F0002] logilab.astng._exceptions.ASTNGBuildingException: 1-
> Unable to load module imcapserver (Unknown C global variable)

duh!
 
> I used the following patch to fix the problem:
> 
> --- manager.py  2007-05-02 16:24:28.146250000 -0400
> +++ manager.1.py        2007-05-02 16:23:26.505625000 -0400
> @@ -197,7 +197,11 @@
>                  modname = klass.__module__
>              except AttributeError:
>                  raise ASTNGBuildingException(
> -                    'Unable to get module for object %r' % obj)
> +                    'Unable to get module for object %r' % obj)
> +            except NameError:
> +                raise ASTNGBuildingException(
> +                    'Unable to get module for object ??')
> +
>          try:
>              name = klass.__name__
>          except AttributeError:
> @@ -366,3 +370,4 @@
>      def __repr__(self):
>          return '<Project %r at %s (%s modules)>' % (self.name, id(self),
>                                                      len(self.modules))
> +
> ----------------------
> 
> I did not really investigate why repr(obj) fails, but the patch appears to fix
> the problem

hum, I think we could even catch bare Exception to avoid such weirdness
when accessing external code. I'll check this in the repository.

> What I did, though is to print obj, and that caused Python itself to crash,
> something I don't see very often!

indeed... 
Thanks for the report and the patch, cheers
-- 
Sylvain Thénault                               LOGILAB, Paris (France)
Formations Python, Zope, Plone, Debian:  http://www.logilab.fr/formations
Développement logiciel sur mesure:       http://www.logilab.fr/services
Python et calcul scientifique:           http://www.logilab.fr/science