[m-dev.] Mutually recursive tailcalls in hlc.

Zoltan Somogyi zoltan.somogyi at runbox.com
Thu Nov 5 17:02:15 AEDT 2015



On Thu, 5 Nov 2015 16:17:33 +1100, Paul Bone <paul at bone.id.au> wrote:
> I setup a small test to see if it was worthwhile investigating further (and
> it is!).  I've compiled mandelbrot in hlc.gc with --no-optimise-tailcalls
> --no-c-optimize.  The C optimisations are disabled because C may be
> optimising tailcalls for me.  Then edited the generated C code to create my
> test.  This example uses direct recursion, that's okay, I'm using it as a
> proxy for mutual recursion because with this small benchmark I can
> confidently modify the hand generated C files to setup my test.

I don't know about others, but I don't understand exactly how
your three versions differ, and exactly how you extrapolate from
direct recursion to mutual recursion. I would need access to the
source code (both .m and .c, if the compiler generated .c is hand
modified), and I would strongly prefer that the exemplar program
itself be mutually recursive.

Zoltan.


More information about the developers mailing list