Re: [Tinycc-devel] TinyCC's future

[Peter Fröhlich]

Hi all,

On Apr 11, 2005, at 04:30, address@hidden wrote:

> I just have wrotten a IA32 backend, something like lightning but only 
> for IA32
> (RtAsm aka Runtime assembler); I use it to generate runtime code. It 
> would be
> easy built a bytecode intermediate language and use it as middle 
> representation
> between fron and back ends. TCC then only move C code into RtAsm and 
> this one
> into IA32, RISC or PPC etc.

Just my $0.02 again: If you split the compiler into frontend and 
backend, it is worth considering using a higher-level structure like an 
AST instead of linear bytecode, at least if the size of the backends is 
a concern at all. It's easier to process an AST into decent code in one 
pass, harder to do on lower-level representations.

I am assuming that nobody is actually suggesting turning tcc into a 
highly optimizing compiler? Then a linear representation (based on SSA 
probably) would be a good choice, but I don't think that would be in 
the "spirit" of the tcc project.

Peter
--
Peter H. Froehlich <><><><><><> http://www.cs.ucr.edu/~phf/
OpenPGP: ABC2 9BCC 1445 86E9 4D59  F532 A8B2 BFAE 342B E9D9