Re: ustring memory leak

[Chris Vine <chris cvine freeserve co uk>]

On Fri, 2 Apr 2010 14:13:28 +0200
Fabian Jacquet <fabian jacquet gmail com> wrote:
> I already try with "G_SLICE=always-malloc" environment variable and
> here is the  result.
> Leak of 835584/1000 or 835
> Leak of 811008/1000 or 811
> Leak of 724992/1000 or 724
> Leak of 720896/1000 or 720
> Leak of 720896/1000 or 720
> Leak of 724992/1000 or 724
> Leak of 724992/1000 or 724
> 
> So, we still have a memory leak but less.
> Maybe is it interesting for future investigations.

This is very strange, and indeed inexplicable.

So we have it that when you call g_utf8_collate() from more than one
thread, it leaks, but only when the threads are created with the windows
native thread functions and not via the GThread wrappers?  This defies
logic.

I wonder whether the original plan A might not have been the right one.
Possibly it is the way the various libraries link together.  Maybe you
are using a different version of some relevant windows dll than your
glib library was compiled against, or some initialisation that needs
to be carried out hasn't been.

Chris