> Generating the code is the easy part. Parsing and grammar is
> a bit more
> complex, but error detection and recovery is where things get dicey.
> IME, YMMV.

I disagree. Lex and parse are trivial compared to (efficient!) code
generation. This of course depends on what you actually call code
generation: a simple first-fit tree coverage using statement templates
is also trivial, but not very optimal.

Specifically for a PIC the irregularities of the data (file registers)
and code addressing are major issues.

Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: www.voti.nl
consultancy, development, PICmicro products

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.