Re: fixinc: PATCH: non-GNU make broken [in] `make bootstrap'

[Geoff Keating <geoffk at geoffk dot org>]

Bruce Korb <bkorb@pacbell.net> writes:

> > I'd like to hear from the fixinc people about this one,
> > as they may wish to change fixinc/Makefile.in instead.
> 
> The "fixinc people" consist of a cadre of one.  :-)
> He can only be relied upon to read email addressed to him
> or containing the six letters, "fixinc" in the subject.
> 
> Anyway, changing fixinc/Makefile.in and fixinc/mkfixinc.sh
> is not the preferred choice:
> 
> > Permission to apply to both paths?
> > 
> > BTW, I know that someone could point out that gcc/fixinc/mkfixinc.sh
> > and gcc/fixinc/Makefile.in should directly use HOST_CC, etc
> > everywhere.  However, that is a far bigger change for a subsystem that
> > only knows little of internal gcc Makefile quirks.
> > 
> > 2001-09-27  Loren J. Rittle  <ljrittle@acm.org>
> > 
> >   * Makefile.in (STAGE2_FLAGS_TO_PASS): Propagate HOST_CC.
> >   (fixinc.sh): Map CC, CFLAGS and LDFLAGS to HOST_* versions for
> >   later recursive make invocation.
> 
> fixinc knows little about GCC's make structure and it has big plans
> for one day incorporating the fix-headers logic and becoming
> installable.  The reason for this is that some people find it
> inconvenient to rebuild all of GCC when all that is really
> needed is the re-editing of a few changed headers.  When this
> happens, it will need to be built with the next stage CC.
> OTOH, we may all be old and grey before this happens.

This would also be good for the very common case where GCC is
distributed in binary form for a non-free OS.  In such a case, the
headers often can't be distributed, so instead fixinc is distributed
and run as part of the install.

-- 
- Geoffrey Keating <geoffk@geoffk.org>