Re: Tree stability going into Version 3.0

["N. D. Culver" <ndc at alum dot mit dot edu>]

[ndc]
> Recently, I've been thinking about writing a 'private tree' to 'gcc
> tree' program which incorporates the GCC back end.

[Mike Stump]
>>We'd much rather have you write a program front end for gcc, then a
>>slurp up the output of your proprietary front end module.

The front end is rather complex; it is a multi-pass globally optimizing/interpreting
gizmo that stores all information in a permanent database and runs in a cross platform
framework. This format doesn't fit well (at all?) within the current GCC architecture.
Besides which, most people probably wouldn't choose to program in my language which
is a homebrew concoction.

My internal AST representation is fairly universal and the parser table generator and
parser permit a lot of what is conventionally done in a front end to be expressed in
the grammar. So one of these days I'll release the grammar wackers which produce
my pretty tree. If I can build a 'my tree' to 'gcc tree' back end program/subroutine
then all a language guy would have to do is write a program which calls the parser,
munch on the tree and then send the tree along to the back end for machine code generation.

Norm