What's the current recommendation for speeding up bits of guile code.

[Rob Browning]

Right now in gnucash, we code in both Guile and C, and there are a
number of us who'd like to do as much as possible of the work (if not
all of it :>) in Scheme.

However, our current approach when things are too slow in Guile, has
been to move them over to C (if we can't just fix them by writing
better Scheme code).  However, we'd love it if there were another good
alternative that would just speed up some of the Guile code.

We've though about some of the options we've heard of, but we didn't
know what Guile's plan for the future, if any, was on this front.  If
there is a plan, we would probably be able to contribute help, if
wanted, to move things along and maintain them.  The possibilities
we've either thought of or heard about are:

  1) hobbit - we've never used it, and don't know what the current
     status or long-term plans with respect to it are.

  2) Keisuke Nishida's VM - same.

  3) Possible frankenstein marriage of stalin with guile somehow.
     Probably not feasible in the general case, though for side-effect
     free functions where the data in/out is negligible wrt
     computational effort, this might be useful, and I can see how it
     might be implemented.

So is there a plan, and have we overlooked other options?

Thanks.

-- 
Rob Browning <address@hidden> PGP=E80E0D04F521A094 532B97F5D64E3930