Re: Native code generation and gcc

[Helmut Eller <eller.helmut@gmail.com>]

The following message is a courtesy copy of an article
that has been posted to gmane.lisp.guile.devel as well.

On Sat, Dec 03 2016, Mikael Djurfeldt wrote:

> Are there fundamental reasons for not re-using the gcc backends for
> native code generation? I'm thinking of the (im?)possibility to
> convert the cps to some of the intermediate languages of gcc.

Tail calls come to mind.  GCC is built around the C/C++ ABI and C
calling conventions.  Due to C's varargs, that's almost always some
variant of "caller pops arguments".  I think sibling calls can be
optimized everywhere but not general tail calls.  You could add support
for tail calls to some backends, but that's far from "for free".  The C
calling convention is also not that great for returning a (statically)
unknown number of values.

Sure, there are tricks/kludges to compile Scheme to C.  I think it would
be simpler/less risky to make a C generating backend for Guile before
going directly to GCC.  The main advantage of a direct GCC front-end
would be better control over debugging information.

Helmut