Getting closer…
OK, I’ve made good progress today; nearly there but the problems are getting hairier.
The packaging is sorted, and we’re using libxml2 instead of expat. All the symbols now appear in the right places, and we can run almost to the end of example.py.
But there appear to be a couple of issues with either the C code or defines causing something important to compile out. I need to resolve these before releasing the Windows binaries.
The filename is getting mangled; the sample “file://../data/dc.rdf” is being resolved to “./data/dc.rdf”; I don’t know how easy it is for me to fix this since C is not my strong suite (the rules are similar but different for C++, which I am a bit more comfortable with). There seems to be some Windows-specific code in the uri-to-file routines but it’s a bit of mess and probably the source of some bugginess. I might comment it out and see if I can isolate the problem. For the moment I just put the file where it thinks it might find it.
The second problem is a crash after printing out the first (only?) query result. It looks like it’s a call to free( ) since I’m ending up in the debug heap, so I need to look at the query cleanup routines. I’d rather be leaking than crashing at this stage. I can ignore it and continue, so it may be a double-free attempt.
The third problem is a crash when attempting to serialize the model to file; the file is created but nothing is written; looks like an uninitialized pointer somewhere.
I’ll be comfortable releasing once these specific issues have been addressed. Hopefully they will not be too hard to fix in the end.