Re: Should logior (and friends) operate only on inums?

[Rob Browning <rlb at cs dot utexas dot edu>]

"Greg J. Badros" <gjb@cs.washington.edu> writes:

> It's now back to working on all unsigned longs (I never meant to
> change it).  Bignums were never supported, and I won't be writing
> that patch (and I'm not even convinced that that should be in the
> core, actually).

On a related note, I'd like to see conditionally compiled
gh_scm2longlong and gh_longlong2scm support.  I wrote a preliminary
(and slightly broken) patch that I sent a while back, and I may take
another stab at it soon unless someone else gets motiviated.

My hack was very inefficient as it just called scheme level
arithemetic ops from C to piecewise construct/deconstruct the bignums
from longlongs (whenever a bignum was required).  The "right" way to
do this (as Jim pointed out at the time) would probably be to define,
implement, and use a more general C level API for building/picewise
manipulating scheme "integers" from C, but I don't have time for that
right now, and until that's done, my hack seemed a semi-reasonable
approach.

-- 
Rob Browning <rlb@cs.utexas.edu> PGP=E80E0D04F521A094 532B97F5D64E3930