Re: [Orbit-python-list] Marshalling a boolean in O-P 0.3.1 causes a segfault

[Roland M. <99....@ai...>]

Roland Mas (2001-12-09 16:46:09 +0100) :

> Christian Robottom Reis (2001-12-08 09:42:51 -0200) :
>
>> On Wed, 21 Nov 2001, Roland Mas wrote:
>>
>>>   It seems version 0.3.1 generates a segfault when marshalling
>>> booleans that 0.3.0 does not have.  I attach a brief test-case.
>>
>> Ok, I'll try and look into this. I have to point out that I didn't cover
>> booleans in the testsuite up to now, which is how this regrettable
>> regression was even possible :-)
>
> Well, I applied a patch I received from the Debian user, and it seems
> to fix the bug.  The user confirmed he can't reproduce it, and my
> test-case doesn't reproduce it either.
>
>   That means you just have to add the stuff into the test-suite :-)

Um.  It seems we still have a segfault lurking somewhere (with CVS as
of approx. 24 hours ago).  Unfortunately, the test-suite doesn't tell
us where...  Maybe you'd care to increase the verbosity, because what
I currently get is:
,----
| $ ./runtest.sh 
| ......................./runtest.sh: line 39: 12327 Segmentation fault      python Test.py $*
| $
`----

  Yeah, I know, at some point I'll have to learn how these things
work, but currently I'm busy on other things.  Namely, I'm replacing
PyArg_Parse and Py_BuildValue calls with other functions, since it
seems that these two functions have been sort of deprecated (and it's
hard to find doc on them...),

Roland.
-- 
Roland Mas

Latitude 43.60039° N / Longitude 6.99555° E