Re: expansion, memoization, and evaluation...

[Rob Browning]

Mikael Djurfeldt <address@hidden> writes:

> Oops...  This reminds me of another consideration I had when opting to
> work on Scheme source: While methods are normally optimized at
> generic application time, goops source can be compiled offline.
>
> If the optimizer does source --> source transformation it's reasonably
> easy to use it together with an offline compiler.  It's more difficult
> to explain the memoized code to the compiler...

OK, I'm confused (and I'm pretty sure most of the difficulty is on my
end :>).  I'm not completely familiar with how things work now, so
could you explain a bit if you have time?

In the above, am I right in presuming that by "work on Scheme source",
you're referring to the way your goops code uses the combination of
the scheme source and an envt representation during the process (that
I don't yet know a lot about) of optimizing an invocation?

Also in the above, when you say "optimizer does source -> source
transformation", which optimizer are you referring to, and more
generally, how would the offline compilation process go in your
thinking?

  scm-sexp -> expanded-sexp -> goops-optimized-sexp -> .o file?

or does the goops optimizer have to work in the dynamic envt at
runtime?  If so, is there a way we can build a goops optimizer that's
more efficient than just falling back on eval?

-- 
Rob Browning
rlb @defaultvalue.org, @linuxdevel.com, and @debian.org
Previously @cs.utexas.edu
GPG starting 2002-11-03 = 14DD 432F AE39 534D B592  F9A0 25C8 D377 8C7E 73A4