Re: backport blockers for in-tree build

[Basile Starynkevitch <basile at starynkevitch dot net>]

On Sun, 26 Sep 2010 16:51:56 +0200
Ralf Wildenhues <Ralf.Wildenhues@gmx.de> wrote:

> * Gerald Pfeifer wrote on Sun, Sep 26, 2010 at 04:44:05PM CEST:
> > On Sun, 26 Sep 2010, Ralf Wildenhues wrote:
> > > builds.  Either users have gotten more diligent at checking prior bugs
> > > in the bugzilla (yeah, right), or we *have* fixed most of those in the
> > > trunk. 
> > 
> > :-)
> > 
> > In install.texi we currently have the following:
> > 
> >   First, we @strong{highly} recommend that GCC be built into a
> >   separate directory from the sources which does @strong{not} reside
> >   within the source tree.  This is how we generally build GCC; building
> >   where @var{srcdir} == @var{objdir} should still work, but doesn't
> >   get extensive testing; building where @var{objdir} is a subdirectory
> >   of @var{srcdir} is unsupported.
> > 
> > Do you see anything we may want to adjust there?  Presumably not yet,
> > but as part of closing PR 37210 ?
> 

And in another message:
 
> * Basile Starynkevitch wrote on Sun, Sep 26, 2010 at 05:27:56PM CEST:
> > Do you suggest to build GCC in its source tree today?
> 
> No, I wouldn't.
> 
> > Would you suggest that to new GCC developers?
> 
> No, I wouldn't.


Maybe the phrasing should be even more strong. E.g.

  First, we @strong{highly} recommend that GCC be built into a
  separate build directory from the sources which does @strong{not}
  reside within the source tree. Do otherwise (e.g. building with source
  directory same or parent of build directory) at your own risk.

Or even stronger:

  We require that GCC be built into a separate build directory from the
  sources which does @strong{not} reside within the source tree.
  
And yes, I would like configure to warn about that.  Half of the
partners which I asked to build GCC from source made that mistake!
(and I don't know well enough non-Linux systems to propose a
configure.ac patch which would work eveywhere)

Another configure wish would be that the topmost configure checks that
all the arguments are valid for the gcc/configure. But that is complex
to implement, and ours configure.ac files are too complex already!

Cheers.
-- 
Basile STARYNKEVITCH         http://starynkevitch.net/Basile/
email: basile<at>starynkevitch<dot>net mobile: +33 6 8501 2359
8, rue de la Faiencerie, 92340 Bourg La Reine, France
*** opinions {are only mine, sont seulement les miennes} ***