Re: Bison error on snap-shot again.

[Marc Espie <espie at quatramaran dot ens dot fr>]

In article <3185.909733200@hurl.cygnus.com> you write:

>  In message <Pine.LNX.3.96.981028111306.2904g-100000@tc.net>you write:
>  > Unless I miss my guess, the 1026 snapshot (full ss) is calling for bison.
>  > This is the same problem I was catching in 1019, but I've cleaned that ss
>  > out of the systems.  Anyone know what's up?
>  > Rand.
>I've modified the snapshot script to touch parse-scan.c so that it always
>appears up to date in snapshots (it's a generated file).  I've also updated
>the egcs_update script in the contrib directory.

Hum.. I've just tried a new method to build egcs: start from the whole cvs
archive, use lndir to create a copy, then rm what you don't use...

as expected, the make process started to use lots of gnu-tools... which 
I happen to have.

Surprisingly though, some of my tools seem to be badly out of date. Even
gperf-2.7 does not recognize the necessary options to rebuild egcs... and
I've got a similar problem with bison in gcc/java later on.
Maybe I've missed something, but I expected to get the latest gperf from
my fav. gnu mirror.

I think it might be a good idea to provide a shopping list of the utility
versions needed to recreate egcs. If anything, if we use the cvs read-only
access, it is because we want to tinker with the source. By any chance,
if some of these tools are still in development, and egcs relies on features
that still are not supported by a generally available version, this is a
problem... how can we work in an `open' environment which turns out not
to be so open after all ?

And yes, I've read the FAQ, which says that you're supposed to touch the
generated files. But then, if you need to tinker with the source files,
how are you going to regenerate them ?