[gpc] Re: GCC integration?

Steven Bosscher stevenb@suse.de
Fri Dec 17 09:46:00 GMT 2004


On Friday 17 December 2004 08:14, Thorsten Glaser wrote:

> Since gcc 4.0 will be out of the door not in less
> than two months from now, and given that gcc 4.1,
> which ought to solve at least the basic problems
> with tree-ssa etc. is not even in sight yet, I
> propose that there will be a gcc-3.5 release with
> support for Pascal, based upon gcc-3.4, current
> gpc and Waldek's patches.

What basic problems with tree-ssa do you see?  As far as I can tell it
pretty much works for everyone now.  You have to be more specific than
this.  Same for Waldek's patches.  If you mean the patches against the
backends, I think you'll find it hard to get them accepted.


> While it might also be worthwhile to integrate
> Hiroaki Etoh's ProPolice SSP, and probably other
> fixes (Apple stuff?), this would mean gcc 3.5
> must be an "official" release.

The problem with ProPolice is not on the GCC side, the maintainer of
that patch is simply not willing to address the comments and concerns
that have been expressed over and again.  And what "Apple stuff" are
you talking about.  And who, do you think, will want to spend effort
on backporting whatever it is you have in mind?


> For others, developers, especially OS vendors, this
> would also be great, because history has shown that
> gcc 3.x branches tend to end after 3.x.3 or 3.x.4,

For GCC 3.3 at least this is not true.  GCC 3.3 is really the first of
the GCC3 series that received wide acceptance.  For the other ones,
apparently nobody was interested in continuing the release branch.

I don't know why you think your idea is great for OS vendors.  As far
as I can tell they're pretty happy with GCC 3.3 (only Fedora Core 3 is
based on GCC 3.4, as far as I can tell), everyone appears to be aiming
for GCC4 for the next major release.


> and for the aforementioned reasons 4.0/4.1 will not
> gain wide acceptance outside of Gentoo (j/k) soon;

What is this assumption based on?  I don't see any "aforementioned"
reasons.


> if one were to incorporate generic 3.x bugfixes and
> maybe backport 4.x bugfixes into a 3.5 series,

GCC 3.x and GCC 4.0 are internally so very different already that you
will find very few bugfixes for GCC 4.0 that you can backport to any
of the GCC 3.x releases.


> this
> could not only get much adoption, but maybe support
> from the large (e.g.) GNU/Linux vendors (Novell?).

If we could fly with the speed of light, and got the support of NASA,
we could visit our friends on Centauri Alpha.  But that's just as
unlikely and hypothetical as your claim here.  ;-)

If any of the large vendors would have been interested in a new GCC3
based release, they would have asked by now.


> This support would lead to a faster distribution of
> gpc to "the masses" than releasing a standalone gpc
> version now and then integrating it into 4.1/4.2
> ever could.

Except that you'd be integrating with an end-of-life infrastructure.
So, one, two years from now when the GCC4 series will be mainstream
compilers, you have to start over with your integration work to keep
gpc available to the masses.


> Besides, support for 3.4 is "almost" done 
> for the more common platforms; I doubt gpc will make
> it into mainline before it runs on almost all platforms,

You could make it so much easier for yourselves if you'd just skip
GCC 3.4 and work against mainline instead.  When you do that, you can
generate target independent code (namely functions as trees, not RTL),
and *poof* gpc would work on all targets, just like that.  The only
concern would be the runtime library, but if it's portable to POSIX-
like targets, that's not such a big concern either.

Of course this also causes other challenges for gpc.  Things like the
SET_TYPE are gone, you'll have to produce code for that yourself in
the front end.  There probably are other things like that which need
rethinking.  On the other hand, the Ada crew (GNAT) managed to do it,
so why not gpc.


> and having it in gcc cvs will surely help to both get
> more eyes on the code (Nathanael said something like
> this IIRC) and get these who do changes to the backend
> more aware of gpc.

IMHO it's those hacking gpc that need to be more aware of GCC ;-)

I'm also a bit puzzled why you'd want a GCC 3.5 based on GCC 3.4 now,
just to get gpc integrated, when for years gpc has hardly tried to be
integrated at all.  Having a GCC 3.5 as you suggests seems to me like
a waste of effort.  Why not just work against GCC 3.4 if you really
want to stay GCC3-based? 

Gr.
Steven



More information about the Gcc mailing list