Re: bug in m4/mkdirp.m4

[Alexandre Duret-Lutz]

>>> "PO" == Peter O'Gorman <address@hidden> writes:

 PO> I guess I'll try this list, since mails and pings to bug-automake get no 
 PO> response...

 PO> <http://lists.gnu.org/archive/html/bug-automake/2005-07/msg00141.html>

 PO> install-sh, if it loses its execute bit causes install to fail.


 PO> 2005-??-??  Peter O'Gorman  <address@hidden>
 PO> * lib/am/distdir.am: Remove $(SHELL) when calling install_sh
 PO> * m4/strip.m4: Ditto.
 PO> * m4/install-sh.m4: Add $(SHELL) to the definition of install_sh

Thanks for the patch.  Can you explain how the x bit vanished?
This looks like the bug we should fix first.

Tom once said he wasn't sure why all those invocations through
$(SHELL) were necessary, that maybe they were relic from some
distant past (something with cygnus?) and no longer required
today.  Recently, I've been tempted to remove them all, because
I believe these scripts are simple enough to work with any
/bin/sh, and presently configure can set @SHELL@ to a shell that
would cause the scripts to fail (like zsh, which requires some
initialization soap none of the scripts has).  Of course the
latter point is not a real argument, because we could add the
soap, and anyway if @address@hidden then make will run the Makefile
fragments using zsh, and this should fail at some point too,
perhaps before starting any of the scripts.

So I'm neither for or against $(SHELL), but I'd like to be sure
why it's needed.
-- 
Alexandre Duret-Lutz