Re: Compiling guile-gobject experiences

[Andy Wingo]

On Sat, 13 Sep 2003, Marius Vollmer wrote:

> Kevin Ryde <address@hidden> writes:
> 
> > I don't think there's any way to portably make a C call from runtime
> > arguments, in general, but it wouldn't be hard to setup the right
> > thing for each CPU+ABI, with a bit of assembler.  Or maybe gcc could
> > help.
> 
> There are already libraries out there for this.  I think "libffi" and
> "libffcall" would be the search terms.

I think that as Andreas has shown, the real slowness is in the GOOPS
side of things. This sort of assembler trickery should not be the focus
of optimization at this point, imo -- the low-hanging fruit is found in
goops.scm (and really should be plucked and placed into goops.c).

So while we might be led down this path eventually (I hope not, but
that's in the future), we shouldn't really think about it now -- we
already have something that works, it just needs to be faster.

Regards,

wingo.

ps. If we wanted to decrease memory utilization by using one C wrapper
function for calling many wrapped C functions (and that would be the
quantity optimized by that sort of solution), I think that since
argument numbers and types is known at compile-time it would be more
fruitful to generate (in C) the needed 1-to-many wrapper functions. But
optimizing memory use is not as urgent as optimizing for speed now.