newlib build machinery

[Ralf Wildenhues]

* Jeff Johnston wrote on Tue, Aug 18, 2009 at 11:38:31PM CEST:
> Ralf Wildenhues wrote:
> >Seriously though, override.m4 is currently the lever to
> >
> >- force exactly one Autoconf version for all configure scripts that
> >  include this file,
> >- fix all Autoconf bugs known to the version used,
> >- and used throughout GCC and binutils, and parts of gdb already.
> >
> >This also means that, while there are likely more bugs exposed by
> >updating to new tools now, it's also easy to fix them throughout the
> >tree (and there are more likely people able to fix them for you).  In
> >that way, it's at least debatable which version to use will be better
> >for you; but you decide.

> It would be simpler to copy override.m4 over so I can avoid an
> additional directory specifier for each aclocal, but in my initial
> experiment I specified -I ../config and the aclocal.m4 referenced:
> depstand.m4, lead-dot.m4 and override.m4 from the config directory.
> If it is highly recommended to use the config directory, then I
> guess I'll do so.  In either case, do I need to remove confsubdir.m4
> if we are using override.m4?

If you want to copy override.m4 to newlib/, that is fine as well.
In that case, you can drop confsubdir.m4 (and anyway confsubdir.m4 is a
no-op for Autoconf versions != 2.59).  However, any way you choose to
go, while you can copy over override.m4 already, I'd like to ask you to
wait with updating the Autoconf version until toplevel has moved to
2.64 (and maybe just a little longer).  Rationale is that there is at
least one bug in 2.64 that we still need to work around in 2.64; thus
you might have to keep track for updates for toplevel config/override.m4.

Thanks,
Ralf