This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: PCH merge bootstrap failure on systems without flex
- From: "Kaveh R. Ghazi" <ghazi at caip dot rutgers dot edu>
- To: mark at codesourcery dot com, pfeifer at dbai dot tuwien dot ac dot at
- Cc: dje at watson dot ibm dot com, gcc-bugs at gcc dot gnu dot org, gcc at gcc dot gnu dot org, geoffk at redhat dot com
- Date: Mon, 10 Jun 2002 10:32:03 -0400 (EDT)
- Subject: Re: PCH merge bootstrap failure on systems without flex
- References: <Pine.BSF.4.44.0206101202120.46487-100000@naos.dbai.tuwien.ac.at>
> From: Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at>
>
> On Sun, 9 Jun 2002, Mark Mitchell wrote:
> > We want generated files in tar files so that you can build GCC from
> > a release tarball with a minimum of installed tools.
> > [...]
> > We've had this debate before; let's not to it again.
>
> Well, that we've had this debate before doesn't mean we came to a commonly
> accepted conclusion.
>
> Requiring flex where we already require bison is one thing, but requiring
> a full development chain including the auto-tools is something we should
> carefully evaluate.
Agreed.
> > We don't want them in CVS because we assume people using CVS are
> > developers and we assume developers can get the tools.
>
> This assumption neglects testers (of which we have too few, especially on
> not-so-common platforms, and many of which use guest accounts with limited
> resources to perform their testing).
> Gerald
Yeah, I'm in that boat.
While flex by itself isn't such a big deal, I think its important to
evaluate the cost when we increase these kinds of dependencies.
Certainly, I'm against deleting auto-tool generated files from CVS.
--Kaveh
--
Kaveh R. Ghazi Director of Systems Architecture
ghazi@caip.rutgers.edu Qwest Solutions