Re: [ob] Regenerate src/configure with 000227

[Geoffrey Keating <geoffk@apple.com>]

On Tuesday, February 18, 2003, at 01:21  PM, Andrew Cagney wrote:

>>> Just FYI, I've committed the attached as `obvious'.  It regenerates 
>>> src/configure using (hopefully) the correct autoconf.  Without this 
>>> the build barfs with the weird syntax error:
>> I would argue against any autoconf *snapshot* being the "right" one.
>> Aren't we supposed to be using the official fsf release of 2.13?  Your
>> change added the sitefile code, which wasn't there before, so it's not
>> just a bugfix - it's a feature change as well.
>
> To expand on DanielJ's comment.  When fixing a GDB / BINUTILS autoconf 
> botch, the final patch _always_ results in the addition of the 
> sitefile stuff.  To me, having that in the diff _is_ normal.
>
>> Plus, you need to test this "obvious" change in the gcc tree and apply
>> it there also - the trees are out of sync now, but they should be in
>> sync.
>
> Should it instead be re-generated with pure 2.13?  Better first 
> question though is what did GeoffK use?  (I'm guessing that it was 
> Geoff's regen that broke it).
>
Yes, I used pure 2.13 as downloaded from ftp.gnu.org.

I think I've worked out what happened.  I did a 'cvs update', but 
didn't think to do a 'cvs update -d', and my tree was pretty old 
(because I only use it to do this) so config/accross.m4 and 
config/acx.m4 were not included in the regeneration.  If I regenerate 
with the updated tree, I get the same configure as I got for GCC.  This 
new configure differs from the current one by not having a --site-file 
flag, and in no other interesting way.