Rakudo and NQP bigint status (2011-11-14)

status.txt

Status of the bigint branches in rakudo and nqp

nqp:
 * all important functions implemented, including
   pow_I and conversion from and to double

 * I've added mp_get_long and mp_set_long functions to
   3rdparty/libtommath/. Dunno if that's the best approach, or if
   our own extensions to tommath should be kept in a separate place.
   The extensions could also be submitted upstream.

 * nqp::isbig_I is broken on 32bit platforms (seems to
   always return 1, people have reported. I don't have any 32bit machine
   available for testing)

rakudo:
 * all Int functions, ops and methods are implemented

 * radcalc in Perl6::Actions seems to work fine, but the codegen
   for rats is broken. It is special-cased in SymolTable.add_constant,
   and that special case assumes that Int can unbox to int and back again
   without loss.
   All my attempts to fix that have failed, it seems to interact very
   weirdly with the rest of the world (or maybe I just don't understand
   the code well enough)

 * Adding some debug statements to radcalc() showed that it is called twice
   for each Rat literal (maybe also for other literals). This is probably not
   specific to the bigint branch, but maybe worth investigating.

 * since the broken Rat literals make the spectests rather noisy, it's
   hard to figure out if more than Rat literals are broken.