Re: Why does config.guess come up with such wacky stuff?

[Robert Lipe <robertlipe at usa dot net>]

John Hughes wrote:
> On UnixWare 7.1 the config.guess from gcc-2.95.2 sez:
> 
> 	i686-UnixWare7.1.0-sysv5
> 
> That's pretty wierd.  On UnixWare 2.1.3 (which is pretty
> much the same from most points of view) we get:
> 
> 	i386-pc-sysv4.2uw2.1.3
> 
> which is also quite ugly, but at least we don't get the OS
> version shoe-horned into the "manufacturer" field.

I could bore you with the history on this but won't.    It's painful.

The Right Thing To Do is to refresh the GCC config.guess with the master
copy from autoconf.  It will return 'i586-sco-sysv5uw7.1.0'.  Then we
need to sweep through the GCC tree fixing up the case statements that
match on *-sysv5*.

The competing factions have discussed this amongst themselves and with
the autoconf maintainers and agreed this is the right thing to do.  So
far nobody has emitted the keystrokes to fix it in GCC.  

If you submit a patch for any affected files for the UnixWare pieces,
I'll approve and install it.  Most of the regex's match on *-sysv5* so I
don't expect much fallout.  I'd expect a patch resynching config.guess
from autoconf to be non-controversial, though I can't approve/install
anything that isn't SCO-specific in that file.  If a resync from the
autoconf version (highly recommended) is pre-approved, I'll install it
at the same time.

Hint: As you go xargs/grepping through the tree, don't miss the one in
libstdc++/configure.in.

RJL