Re: fix libtool test

[Stepan Kasal]

Hello Ralf,
sorry for returning to that so late.

> * Stepan Kasal wrote on Thu, Nov 26, 2009 at 02:05:17AM CET:
> > -AT_CHECK([libtoolize], 0, [stdout], ignore)
> > +AT_CHECK([libtoolize --install], 0, [stdout], ignore)

> libtoolize --install is a 2.2.x-ism, so unless you take care to skip the
> test for earlier Libtool versions (1.5.x are still in use) this will
> cause a failure for them.

would  "libtoolize --install || libtoolize" make sense here?

> >  # Some broken libtoolize fail to install a functional config.guess.
> > -AT_CHECK([./config.guess || exit 77], [], [ignore], [ignore])
> > +AT_CHECK([./config.guess], [], [ignore], [ignore])
> 
> I don't understand that.  Either the original line or drop it
> completely?

Well, the question is: should the test fail if the installed
libtoolize is broken or not?

The original line caused skip if config.guess did not exist.

If the test should be skipped if config.guess is installed,
but it does not work, then perhaps this should read:

: # Some broken libtoolize fail to install a functional config.guess.
: AT_CHECK([test -f config.guess], [], [], [])
: AT_CHECK([./config.guess || exit 77], [], [ignore], [ignore])

What do you think?

Cheers,
        Stepan