This is the mail archive of the
guile@sourceware.cygnus.com
mailing list for the Guile project.
Re: Proposal for a Guile binary file format
- To: guile at sourceware dot cygnus dot com
- Subject: Re: Proposal for a Guile binary file format
- From: Lynn Winebarger <owinebar at free-expression dot org>
- Date: Mon, 22 Nov 1999 05:44:12 -0500 (EST)
On Mon, 22 Nov 1999, Klaus Schilling wrote:
> Marius Vollmer writes:
> > > > I would pay the price of slower loading of uncompiled Scheme source
> > > > when we would instead have a good compiler that produces binary code
> > > > that can be loaded fast and executes fast.
>
> Is it actually possible to compile procedures involving some form of
> eval efficiently? Stalin refuses to do that. And 'naos' needs eval or
> local-eval a lot, at least until K.Wright shows me that it is not so.
>
A while back there was a little discussion about Stalin on
comp.lang.functional (probably been more since) - sounded like it obtained
impressive optimization results by requiring the code to be statically
analyzable.
As for compiling procedures involving eval, I'd imagine it would have a
lot to do with how convolved the construction of the expression you're
eval'ing is, if it's available at compile time (i.e. not interactively
input), and how much computation you're willing to spend on abstract
interpretation to gather up the potential eval'ed code.
Lynn