From: Joshua N Pritikin <joshua.pritikin@db.com> >Hm. You seem confused about what I am proposing. Let me emphasize that >any perl5 implementation of perl6 will be a (C?) code generator. This >generated C code will then compile into a perl6 binary. In this way, >perl6 could be distributed as perl5 code, C code, or as a binary. This is a very good idea, particularly since it would allow people to also optimize their code for readability vs. speed (whether or not the perl5 generated code inlines certain things, for instance - for a segmented architecture or a debug build, the tokenizer could split into separate functions, but for a PC or server build, the various functions could actually all be inlined) and help trim down the forest of #ifdefs and macros as well. (For instance, we wouldn't need the empty random MUTEX_LOCK() calls; they would never appear in the generated C for a build that doesn't need them.) -- BKS ________________________________________________________________________ Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com