[<<][staapl][>>][..]Wed Feb 21 23:55:34 GMT 2007
yes, time to get it done.. overall, i'm quite happy with the
result. it's a lot better than the previous two. i can't really see
much further from here, other than elaborating towards higher
abstractions (different language), and fixing some jump related simple
the bad guy is quoted method code, which has a strange conflict of
concepts. more on that later.
another thing i miss is inline cat code, i.e. for generating
tables. i think i better do this in a different file, and only in
scheme: no more intermediate cat-only files. 1 1.1 16 table-geom
then the lack of proper run-time semantics is kind of weird. the
partial evaluator replaces this, but in an implicit manner: not
everything is accessible, and the bit depth is different.
about literal opti: still not completely happy, since the patterns
should do the literal preprocessing automaticly.
looking at pic18.ss gives me a warm fuzzy feeling :) most of the
knowledge is encoded in 2 patterns: assembly substitution patterns and
recursive macros. language support is encoded in 2 more: some asm
state monad and writer monad.
the thing which would help a bit is reducing the redundancy for the
rewriter macro specification. the way it is right now is very
readable, but maybe a bit too much clutter. on the other hand, it
might be a bit overengineering.