* Stefano Lattarini wrote on Sun, Sep 26, 2010 at 11:39:38AM CEST:
> On Sunday 26 September 2010, Ralf Wildenhues wrote:
> >
> > Well the patch surely makes it clear why $curdir was a badly chosen
> > name: it is not clear at which time this directory was "current".
> There's a pending patch (for "tests-init" branch) to rename it ;-).
Ah, good. Thanks.
> > > # aclocal and automake cannot work without configure.ac or
> > > configure.in
> > >
> > > -$ACLOCAL 2>stderr && { cat stderr >&2; Exit 1; }
> > > +"$curdir/aclocal-$APIVERSION" 2>stderr && { cat stderr >&2; Exit
> > > 1; }
> > >
> > > cat stderr >&2
> > >
> > > -grep configure.ac stderr
> > > -grep configure.in stderr
> > > -AUTOMAKE_fails
> > > -grep configure.ac stderr
> > > -grep configure.in stderr
> > > +$FGREP configure.ac stderr
> > > +$FGREP configure.in stderr
> > > +"$curdir/automake-$APIVERSION" 2>stderr && { cat stderr >&2;
> > > Exit 1; } +cat stderr >&2
> > > +$FGREP configure.ac stderr
> > > +$FGREP configure.in stderr
> >
> > This all looks like change for the purpose of changing.
> No, it avoids the use of arguments with $AUTOMAKE and $ACLOCAL, and
> escape literal dots in grep searches.
Well, but the tests as they were were strict, and not wrong. Oh well,
getting into bikeshed issues again, which I was trying to avoid.
> Should I drop this chunk anyway?
However you like.
Patch is OK with modifications.
Thanks,
Ralf
