Re: automake 1.7 fails aclocalii test on FreeBSD 4.7-RC

[Matthias Andree]

On Mon, 07 Oct 2002, Alexandre Duret-Lutz wrote:

> These four lines should not happen.  The whole set of `--acdir'
> and `-I' flags passed to `aclocal' are meant to force `aclocal'
> to fetch *anything* from its own directories, and ignore files
> installed system-wise.
> 
> Furthermore /usr/X11R6/share/aclocal/ is a strange location.
> Does it corresponds to anything you typed yourself
> (e.g. --prefix=/usr/X11R6) or does it comes from "elsewhere"?

Hum, FreeBSD likes to install X11 related things into /usr/X11R6 rather
than /usr/local. See X11BASE in the ports (in contrast to LOCALBASE).

> Has this copy of Automake been patched in any way?  (e.g., it
> could be the FreeBSD package of Automake.)

*blush* now that you're asking, indeed the automake 1.5 distribution
patches for FreeBSD have been applied, and they fiddle around with
aclocal.in. That is where the X11 comes from, see the first two patch
hunks. If you feel FreeBSD should not do this, then please contact the
FreeBSD automake port maintainers at address@hidden -- note I did
not run "make check" of the patched automake-1.7.

The patches are: 1. patch-aa: for capitalization to automake.texi (should be
harmless). 2. patch-ab: add --no-split to $(MAKEINFO) (should also be
harmless). 3. aclocal.in patch, attached for reference. That is what
FreeBSD's automake-1.5,1 port used and what I left in, because I did not
look.

> Just to be sure:
> 
> % md5sum automake-1.7/aclocal.in
> 9818714e5743cebf02873d2116a7852d  automake-1.7/aclocal.in

Modified. :-(
MD5 (work/automake-1.7/aclocal.in) = 99615b842b56e932a3283543f2c9899f

patch-ad
Description: Text document