public inbox for gcc@gcc.gnu.org
 help / color / mirror / Atom feed
From: Paul Brook <paul@nowt.org>
To: Pablo De Napoli <pdenapo@yahoo.com>, gcc@gcc.gnu.org
Subject: Re: A suggestion to make it easier to develop gcc front ends
Date: Sat, 05 Apr 2003 21:30:00 -0000	[thread overview]
Message-ID: <200304052151.12871.paul@nowt.org> (raw)
In-Reply-To: <20030405150258.49354.qmail@web41511.mail.yahoo.com>

On Saturday 05 April 2003 4:02 pm, Pablo De Napoli wrote:
> Dear developers of gcc:
>
> I want to make a suggestion of an improvement to gcc. The goal of this
> improvement would be making easier to develop new front-ends to gcc for
> different languages.
>
> My idea is that the backend and the front-ends of the gcc should be
> clearly separated. The backend should be implemented as a library. Then
> if someone wants to develop a new language, he would include the
> apropiated headers (say #include <gccbackend.h>) and link with the
> library.
>
> The backend library would implement the convertion from RTL code to
> assembler code, generating debugging code and (language-independent)
> optimizations.
>
> The gcc backend library should have a well-documented and (hopefully)
> stable API. So that when the gcc backends improves, we don't need to
> change every time the gcc front-ends source.

It still isn't entirely done, but a large proportion of the things you 
suggest have already been implemented in the tree-ssa branch. Specifically 
there is the GENERIC/GIMPLE intermediate tree representation. Generating 
these trees (the frontend) and optimizing/generating code from them (the 
backend) are two distinct processes, with the exception of a bit of setup 
code.

I'm not sure I'd describe the interface as well documented or stable as it's 
still fairly new, but It certainly could (and IMHO should) be.

I know of at least one person who routinely does work on g95 by developing 
the fronted in a seperate tree and linking with libbackend.a.

I haven't looked at the mainline/release branches recently, so couldn't say 
how much this is true for those.

Paul Brook

      reply	other threads:[~2003-04-05 20:51 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2003-04-05 20:51 Pablo De Napoli
2003-04-05 21:30 ` Paul Brook [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=200304052151.12871.paul@nowt.org \
    --to=paul@nowt.org \
    --cc=gcc@gcc.gnu.org \
    --cc=pdenapo@yahoo.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).