Makefile/configure patch for manpage generation

[Joseph S. Myers]

On 19 Dec 2000, Gabriel Dos Reis wrote:

> | (b) Do the configure changes in this patch not avoid the problem of
> | attempting to regenerate manpages without sufficiently recent versions
> | installed, or do the Makefile changes in the patch not avoid the problem
> | of relying on #! when running texi2pod.pl?
>
> Not relying on #! when running texi2pod.pl.

That is what

-TEXI2POD = $(srcdir)/../contrib/texi2pod.pl
+TEXI2POD = perl $(srcdir)/../contrib/texi2pod.pl

in this patch is supposed to fix.  Are you saying that things don't work
after this patch?  The only such scenario I can see would be when (a) the
timestamps are wrong; and (b) perl is in the PATH at configure time, and
Pod::Man is sufficiently recent (which should eliminate almost everyone,
since this needs a more recent Pod::Man than is in perl 5.6.0); and (c)
the PATH at make install time is different, and does not include perl or
includes an older version.

Perhaps someone would like to hack GNU diffutils (check with Paul Eggert
first, I guess, to see if there's a more recent development version than
2.7.2 from alpha.gnu.org) to include an option to output diffs for
specified source files before generated files, so that snapshot and
release scripts can do that and gcc_update need only be recommended for
people working from CVS?

-- 
Joseph S. Myers
jsm28@cam.ac.uk