[Python-Dev] Re: python vs. valgrind

[Fredrik Lundh]

Neal D. Becker wrote:

>I just tried running valgrind-2.1.2 on python-2.3.3.

> [many, many more complaints...]

compare and contrast:

$ valgrind python2.1
Python 2.1.3 ...
>>>
==12393== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 17 from 2)
==12393== malloc/free: in use at exit: 78911 bytes in 220 blocks.
==12393== malloc/free: 6367 allocs, 6147 frees, 764773 bytes allocated.

$ valgrind python2.3
Python 2.3.3
==12397== ERROR SUMMARY: 893 errors from 71 contexts (suppressed: 19 from 2)
==12397== malloc/free: in use at exit: 704466 bytes in 227 blocks.
==12397== malloc/free: 2033 allocs, 1806 frees, 1540716 bytes allocated.

(using a slightly older version of valgrind 2.x)

most (all?) errors occur in PyObject_Free and PyObject_Realloc, which indicates
that the problem is either in pymalloc or valgrind itself (and in my experience, while
vladimir and tim almost never mess up, valgrind is *never* wrong...)

    http://www.python.org/doc/2.3.3/whatsnew/section-pymalloc.html

</F>