This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: AutoGen as a GNU package
- To: bkorb at allegronetworks dot com
- Subject: Re: AutoGen as a GNU package
- From: Geoff Keating <geoffk at geoffk dot org>
- Date: Mon, 16 Apr 2001 12:23:18 -0700
- CC: rms at gnu dot org, bkorb at gnu dot org, craig at jcb-sc dot com, jfc at mit dot edu, rearnsha at arm dot com, rth at redhat dot com, kenner at nyu dot edu, law at redhat dot com, jason at redhat dot com, meissner at redhat dot com, davem at redhat dot com, mark at codesourcery dot com, bernds at redhat dot com, wilson at redhat dot com, gcc at gcc dot gnu dot org
- References: <B4DFCB7CDE2DD4118F690008C78694160100B62E@tahoe.allegronetworks.com>
- Reply-to: Geoff Keating <geoffk at redhat dot com>
> From: Bruce Korb <bkorb@allegronetworks.com>
> Date: Mon, 16 Apr 2001 10:13:46 -0700
> > Based on the evaluators' report, I've decided I would like to make
> > AutoGen a GNU package. But there is one point they told me which
> > seems like a problem that ought to be solved:
> >
> > Version 5.x is the current, Guile-enabled version, but
> > version 4.x is the series required to build the include
> > fixer for GCC. (Can't change -- no Guile for CygWin )-:
> >
> > It seems like a very bad thing to need to keep an old version around
> > indefinitely. Can you explain the reasons to me in more detail?
> > Let's look for a solution.
>
> If we can assume GCC maintainers all have access to "Standard GNU Tools"
> and Guile is included in "Standard GNU Tools", then all maintainers
> would have access to the prerequisites to running AutoGen 5.x.
I think that this is the wrong way to look at it.
The GNU project is based on giving the users of our software the
freedom to read and change it. If we provide files that require a
tool to build that is not available on some platform(s), then an
important freedom is lost, because the user cannot readily change that
part of the code.
So, it is not just "maintainers", it is all users who must be able to
rebuild the program from source.
In this particular case, I think it would be OK to require Guile to
create fixincludes if we stopped using fixincludes on the platforms
that can't support Guile, because then all the users who would need to
run it could run it. I don't know if this is feasible.
I have added the main GCC mailing list to the CC, as this change seems
like it would affect everyone using GCC not just those people with
blanket write privileges.
> What I do not know is if these assumptions are correct. Therefore,
> I am CC-ing the opinion of the "Blanket Write Privs" GCC folks.
> To make it clear to them, the question is basically this:
>
> Can we change the AutoGen version required to rebuild fixincludes
> from 4.x to 5.x? Doing so will also require that the Guile library
> be installed on a GCC-maintainer's system (in addition to AutoGen 5.x).
> Leastwise, any maintainer that touches the fix definitions.
>
> It would probably take me a couple hours to convert the template.
> It would render it incomprehensible to AutoGen 4.x.
--
- Geoffrey Keating <geoffk@geoffk.org>