This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: PATCH: top level makefile cleanup (prep for autoconfiscation)
- From: Nathanael Nerode <neroden at doctormoo dot dyndns dot org>
- To: Alexandre Oliva <aoliva at redhat dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Mon, 28 Jan 2002 17:04:10 -0500 (EST)
- Subject: Re: PATCH: top level makefile cleanup (prep for autoconfiscation)
On 28 Jan 2002, Alexandre Oliva wrote:
> On Jan 28, 2002, Nathanael Nerode <neroden@doctormoo.dyndns.org> wrote:
>
> > The dependencies are all right, but they're far from optimal. There's a
> > fair amount of missing dependency information in the Makefile, as well as
> > some which I know to be bogus. As for the configure.in file, it doesn't
> > have proper dependencies, and relies on a sorting order, which is nasty.
>
> Ok, so here's an idea, since you seem to be enjoying this project :-)
>
> One of my quirks with the single-tree build is the long time it takes
> to configure everything, the fact that it can't be done in
> parallel, and the fact that one tool can't depend on the other having
> already been built in order to do configure tests with it.
>
> In my dream implementation of the top-level configure/Makefile, I'd
> like the top-level configure to create just the top-level Makefile.
> Everything else should be configured and built according to Makefile
> rules. For example, if the compiler wants an assembler to test, it
> configure-gcc should depend on all-gas, but it doesn't need to depend
> on say bison for configuration, only for building, so all-gcc would
> depend on all-bison.
>
> Is this in your plan? If not, perhaps you would consider putting it
> in?
This is in many ways the best way to do things. Unless I hear arguments
to the contrary, I intend to do as much as I can this way.
This involves significant restructuring of the Makefile, but I was going
to have to do some of that anyway, and some of it should be done for speed
anyway.
--Nathanael