Re: Memory leak with ThreadSpecificData

[Josh Clark <id...@gmail.com>]

Has anything else happened with this issue?



On 10/20/05, Curt Arnold <ca...@apache.org> wrote:
>
>
> On Oct 18, 2005, at 8:10 AM, Steindl Christian wrote:
>
> > Hi,
> >
> > I use the VLD (visual leak detector) with my applications and log4cxx
> > and noticed quite some memory leaks. Well, in my specific
> > application a
> > huge number of threads might be spawn in the course of time. Is
> > there a
> > way to get rid of the ThreadSpecificData again? My application runs on
> > Win32, and so the APRInitializer::tlsDestruct() won't be called.
> > Consequently I've patched log4cxx by introducing a method to
> > explicitly
> > free the ThreadSpecificData, but that's of course not transparent. Is
> > there an intend to provide the same behaviour of the APR for Win32
> > eventually? Any other ideas about that problem?
> >
> > Thanks,
> > Christian
>
>
> I do not know of a portable way to be notified of thread termination,
> so any solution would require user cooperation of notifying log4cxx
> that a particular thread is terminating. I'm packing up and moving
> at the moment and will be offline for the next few days. If there is
> no other discussion, log a bug report at http://issues.apache.org/
> jira and we will add it to the list. If you can share more details
> of your solution, it may be helpful.
>