Follow-up:

Running :

valgrind --leak-check=full --show-reachable=yes ./vwc ../../shakespeare.txt

results in:

==1871== LEAK SUMMARY:
==1871==    definitely lost: 1,159,488 bytes in 72,468 blocks
==1871==    indirectly lost: 594,828 bytes in 72,468 blocks
==1871==      possibly lost: 3,293,008 bytes in 6,709 blocks
==1871==    still reachable: 409,434 bytes in 1,682 blocks
==1871==         suppressed: 0 bytes in 0 blocks
==1871==

So there is definitely something fishy going on here, either:

- the Vala Hashmaps are buggy, or
- the refrence counting of Vala is buggy or
- something eludes me here.

Serge.


On Sat, Jun 11, 2011 at 11:30 AM, Serge Hulne <[email protected]> wrote:
> I mentioned earlier on this mailing  list a problem ("segmentation
> fault": usually the sign of a memory allocation gone wrong) which
> occurs when merely substituting
>
> isalnum()
> for
> issspace ()
>
> in a simplistic scanner, see:
>
> http://mail.gnome.org/archives/vala-list/2011-June/msg00062.html
>
>
> This simple substitution produces a totally unexpected segmentation
> fault when the application (see link above) is run on a text file a
> few hundreds of MB large.
>
> - Is there a way to try to assess what went wrong ?
> - Is there a Valgring-like utility for Vala ?
> - Does anybody have an idea why this is happening ?
> - Can Vala be trusted to develop non-trivial applications (is it ripe,
> debugged, tested, robust enough) ?
>
>
> Serge.
>
_______________________________________________
vala-list mailing list
[email protected]
http://mail.gnome.org/mailman/listinfo/vala-list

Reply via email to