[<<][staapl][>>][..]
Wed Jun 25 02:25:19 CEST 2008

questions

The problem is not answers; it's asking the right questions. An
attempt:

 Q: Why are multiple passes for the rewriter so essential? Given a
    satisfactory answer to this, is it better to rewrite first to a
    simple qw,cw language, or are per-target patterns better?

 Q: Is it possible to see non-compilable pseudo assembler results
    somehow as type errors or contract violations, and associate
    blame?

Following John Nowak's advice, let's have a look again at the Joy page
about rewriting, and Backus' Turing Award lecture.

http://www.latrobe.edu.au/philosophy/phimvt/joy/j07rrs.html
http://www.stanford.edu/class/cs242/readings/backus.pdf



[Reply][About]
[<<][staapl][>>][..]