[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: [Tinycc-devel] windows port
From: |
Erik Poupaert |
Subject: |
RE: [Tinycc-devel] windows port |
Date: |
Sun, 16 Mar 2003 13:12:20 +0100 |
I was thinking of tinycc as a backend for a front end compiler we've been
toying for months now, to build. It would be a java-like type of compiler,
with lexer & parser done with SableCc. The incredible speed of tinycc would
be a boon for such project. The availability of libtinycc favours this kind
of use too.
It looks like quite a few people are looking at tinycc for this kind of use.
It is a missing link for compiler builders. Today everybody has to build
their own backend, since the dominant toolchain (gcc) has a complex backend
and is not really fit to re-use.
A first issue is that the backend should be able to generate code for at
least windows (PE) and linux (ELF).
Another issue would be that the backend should have a "lenient" mode. The C
language requires you to declare prototypes in one way or the other (e.g.
header files) prior to using a particular function. For manually written
code, it is definitely a necessary validation. For compiler backends,
however, it introduces the burden of producing header files, while the
necessary validations have already been done by the front end.
Would it be a lot of work to add some kind of "lenient" mode, in which
prototypes are considered to be correct as used?
I think tinycc of all compiler back-end technologies comes closest to a
re-usable component for compiler projects. It could save everybody a lot of
time.