[redland-dev] Redland for Revolution & questions

[Dave Beckett]

On Thu, 14 Oct 2004 13:54:23 +0200, Suzan Foster <su at islief.nl> wrote:

> > It's usually an issue with runtimes when you get memory exceptions in 
> > dll's. I was having similar problems with raptor code. When compiling 
> > with VC++ check that you're building all libraries with /MD for 
> > release and /MDd for debug, /MT and /MTd for the executable. With your 
> > code it sounds like you're passing file handles over runtime 
> > boundaries.

Could you expand on that a bit more?  What do those options do?
Seems this is worth recording for others building redland on win32.
I'll add it to http://librdf.org/notes/porting.html and maybe promote
it to core docs.

The point about memory across dlls was brought up before.  One fix
was to make all redland & rasqal's memory routines use raptor's so
that they all used the same heap.  If you look in
  http://librdf.org/raptor/libraptor.html
that means replacing librdf_malloc (aka malloc) with raptor_alloc_memory
and so on.

I think I've got patches for that in the latest cvs, after the earlier
work of Chris Pointon.

Dave