Re: Implementing loadlibrary, calllib, etc.

[John Swensen]

On Mar 24, 2010, at 3:15 PM, Ryan Rusaw wrote:

> Another option would be to use the libraries from the clang compiler
> project, http://clang.llvm.org/index.html, to parse and extract the
> necessary information from the header files.
> 
>> 
> 

I took a look at swig and it looks like extracting the necessary bits from swig 
might be work work than it is worth.  On the other hand, maybe we should just 
use swig altogether and then wrap all the specifics of loadlibrary, callib, 
etc. around swig.  If I remember correctly, it seems that someone had a fairly 
complete SwigOctave implementation.  clang looks like a much better solution 
that gccxml.  It looks very easy to extract the same information as provided by 
gccxml, without having to make a system() call to execute gccxml and depend on 
an XML library for parsing the output.  There is a simple example given at 
http://amnoid.de/tmp/clangtut/tut.html about how to extract global variables 
from a source file.  It seems that parsing structs, functions, etc. should be 
equally as simple.

JWE, I will definitely take baby steps and post progress to the mailing list.  
I was originally planning to make this and OctaveForge package, but your 
comment made me wonder if you thought it should go in the core of Octave?


John Swense