Re: FYI: sh.test needs the correct path to ltmain.sh [libtool--release--2.0--patch-11]

[Gary V. Vaughan]

Hallo Ralf!

Ralf Wildenhues wrote:
> This, however (yes, really) breaks things.  Of course this patch is not
> at fault, rather it exposes a couple of things:
> 
> | sh.test: ===  Running sh.test
> | /tmp/libtool-2.0/libtool/tests/../config/ltmain.sh:141:if test 
> "x$as_myself" = x; then
> | ../libtoolize:126:if test "x$as_myself" = x; then
> | use `test "X..."' instead of `test "x..."'
> | /tmp/libtool-2.0/libtool/tests/../config/ltmain.sh:134:  test -r 
> "$as_dir/$0" && as_myself=$as_dir/$0 && break
> | ../libtoolize:119:  test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
> | assignments on the same line as a `break' or `continue' may have no effect
> | FAIL: sh.test
> 
> Both failures come from autoconf/lib/m4sugar/m4sh.m4 (expansion of
> AS_SHELL_SANITIZE) and fail for CVS Autoconf, but should fail for
> 2.59 as well.  While it may or may not be desirable to change things
> there, it's undesirable to force users to upgrade Autoconf once more
> before Libtool-2.0 (let alone thinking out loud about a new Autoconf
> release).

Agreed.

> So sh.test ought to be relaxed:  My suggestion'd be to just disable
> the first test now, and use the following workaround for the second:
> Allow the line if it contains a `&'.  Just a simple-minded idea.

The first test is dumb, lets remove it altogether.

The second test is more or less the same as the assign.test that we
ditched already.  Lets remove that one too.

> OK to apply to branch-2-0 and HEAD?

With out right deletion of the offending tests: Yes, please.

Cheers,
        Gary.
-- 
Gary V. Vaughan      ())_.  address@hidden,gnu.org}
Research Scientist   ( '/   http://tkd.kicks-ass.net
GNU Hacker           / )=   http://www.gnu.org/software/libtool
Technical Author   `(_~)_   http://sources.redhat.com/autobook

signature.asc
Description: OpenPGP digital signature