Re: [xml] xmlParserCtxt Leak?

[Daniel Veillard <veillard redhat com>]

On Mon, Jun 16, 2008 at 09:16:20AM -0700, mt2 wrote:
> I implemented the changes you recommended and it works; everything
> seems to be in order now. I've implemented the same situation with
> plain xmlInitParser() and didn't see that "still reachable" data,
> which is the only reason I posted here. Now that I know that is
> "per-thread data", it makes a lot more sense. Valgrind doesn't
> complain anymore either, so my life is a lot happier. Thank you for
> your help and information.

  okay,

> Daniel, thank you, too, for your help. I put the offending code in a
> loop as you suggested, and when compiled against libxml with memory
> debug, all turns out well. Thank you for that hint. If I need help
> again, I will make to sure to do that; this is still a learning
> process for me. The example I posted is exactly the same as my own
> code, but I figured it would be best, though, if you didn't have to
> scroll through pages of comments and such.
> 
> Is there any more information I can read about this "per-thread data"?
> I'd like to learn more.

  Not very glorious, basically it shouldn't exist, but the API
included global variables allowing to change the library behaviour.
This is a big mistake, but well it's here, and to limit the damage
it's better in multithreaded applications to have per thread global
variable values, so there is a module globals.c in the library implementing
that.
  Advice, don't try to use global variables from the library, and
avoid trying to make use of this 'feature'.

Daniel

-- 
Red Hat Virtualization group http://redhat.com/virtualization/
Daniel Veillard      | virtualization library  http://libvirt.org/
veillard redhat com  | libxml GNOME XML XSLT toolkit  http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine  http://rpmfind.net/