bug-autoconf
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: $SHELL for $ac_install_sh?


From: Ralf Wildenhues
Subject: Re: $SHELL for $ac_install_sh?
Date: Tue, 18 Sep 2007 21:23:09 +0200
User-agent: Mutt/1.5.13 (2006-08-11)

Hello,

FWIW, I've put Karl and Vincent back in Cc:.

* Eric Blake-1 wrote on Tue, Sep 18, 2007 at 04:39:44PM CEST:
> 
> > When a configure script has to fall back on $ac_install_sh for INSTALL
> > or MKDIR_P, how about invoking it via $SHELL, as in the diff below?
> 
> This idea has come up before:
> 
> http://lists.gnu.org/archive/html/bug-autoconf/2006-09/msg00022.html
> 
> There have been a few changes in autoconf's MKDIR_P since then,
> and I haven't analyzed all of the ramifications of making such a change.

At least both older and newer Automake and gettext versions would need
to be checked.  (I'm not saying the change shouldn't be made, at least
one could consider the possibilities of setting it to
  '${SHELL}'" $ac_install_sh"
  '$(SHELL)'" $ac_install_sh"
  "$SHELL $ac_install_sh"

> Would a different approach be adding a config.status hook that
> always does 'chmod +x install-sh' if install-sh was required?

That won't work for read-only source trees.  Of course, at the first
time configure is run, a read-only source tree may even break in other
places, but I'd hate to add more instances like that.  At the least, the
bogus error output of chmod's failure should be suppressed.
(Hmm, do people mount CDs with no-exec flag?  Then using a shell prefix
in some way or other can't be avoided IIRC.)

Karl, for texinfo, why do you even store install-sh in CVS (and why was
it added there without execute permission)?

Cheers,
Ralf




reply via email to

[Prev in Thread] Current Thread [Next in Thread]