Re: About a new gcc feature. I hope you read this e-mail

[Fergus Henderson <fjh at cs dot mu dot OZ dot AU>]

On 17-Jan-2003, lucas75it@tiscali.it <lucas75it@tiscali.it> wrote:
> Now, can we introduce a new substep that give us the ability of visit 
> the AST ?

One problem with this is that it would require having a *stable*
interface to the AST.

> We can also modify the previous sequence of passes by introducing some 
> others walkers. For example:
> 
> 1 - parsing
>    1.1  walk around AST
> 2 - tree optmization
>    2.1  walk around Opt AST
> 3 - RTL generation
>    3.1  walk around RTL
> 4 - Sibiling call optimization
> 5 - ................
> 6 - ................

This would require a stable interface to RTL, etc.

The result would be that either

	(1) programs which use these walkers break with new versions of
	  GCC

or that

	(2) GCC's internal data structures remain fixed,
	  and hence GCC does not add any new features (optimizations,
	  languages, etc.) that would require changing its internal
	  data structures

If (1), then the feature is not worth bothering with, IMHO.
But (2) would impede progress.  And I don't see any other easy alternative.

-- 
Fergus Henderson <fjh@cs.mu.oz.au>  |  "I have always known that the pursuit
The University of Melbourne         |  of excellence is a lethal habit"
WWW: <http://www.cs.mu.oz.au/~fjh>  |     -- the last words of T. S. Garp.