[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: g++ compiler
From: |
Paul Pluzhnikov |
Subject: |
Re: g++ compiler |
Date: |
Sun, 09 Oct 2005 12:12:36 -0700 |
User-agent: |
Gnus/5.1006 (Gnus v5.10.6) XEmacs/21.4 (Jumbo Shrimp, linux) |
ganeshb@gmail.com writes:
> The functionality I am expecting is that,
>
> 1. apply my annotations (directives) and modify the source code.
> 2. apply g++ optimizations (esp. inlining) over the modified code.
2a. reconstruct the optimized IL tree into valid/compilable C++
> 3. Repeat 1-2 until convergence (some condition set by me).
>
> I would like to know if this is an achievable task using g++.
Everything is achievable in theory.
In practice, step 2a *alone* will probably take you a lot longer
than a month (I've worked on such a reconstructor, and would estimate
2 (mythical :) man-years of effort to write one from scratch).
OTOH, it you don't need "all possible" trees to be correctly
reconstructed (i.e. you only care about a handful of hand-coded
inputs), then this might be doable in a week or two.
Also note, that the code after optimization may not be easily
convertible back into C++, only to "lowered" C (I never looked at
optimized IL, so don't know for sure).
Cheers,
--
In order to understand recursion you must first understand recursion.
Remove /-nsp/ for email.