|Storing, accessing type information at run time email@example.com (1999-06-27)|
|Re: Storing, accessing type information at run time firstname.lastname@example.org (Nikola Petrov) (1999-06-29)|
|From:||Nikola Petrov <email@example.com>|
|Date:||29 Jun 1999 02:58:38 -0400|
|Organization:||The Open Group Research Institute|
> Queinnec's Scheme-to-C compiler in _Lisp in Small Pieces_ tags each
> Scheme object (idea #1 above).
I can't help you a lot but I would only comment that, as far as I know
Queinnec, he did most of Scheme compiling with the idea for it to be
1. easy, and 2.distributed. If that is your intention, than the
proposed approach seems to be inevitable. On the contrary, if you wish
to optimize a compiler for a single machine, then probably typing
would better be global. This approach is mostly imperative thinking
and probably another reason for Queinnec to avoid it.
Just the thoughts of one of his students.
Return to the
Search the comp.compilers archives again.