[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: install-sh defect?
Re: install-sh defect?
Thu, 10 Apr 2008 22:09:22 -0600
Thunderbird 18.104.22.168 (Windows/20080213)
Eric Blake wrote:
According to John Calcote on 4/10/2008 3:00 PM:
| you can use if you wish. I believe this to be an inaccurate statement,
| as no install-sh script actually gets installed with the autoconf
| package (at least not lately).
It's there, as autoconf-2.62/build-aux/install-sh. You are correct that
we don't install it anywhere, because automake does a better job at that;
but by virtue of the fact that you have the autoconf package, you have
access to the script.
Well, you're right - it is indeed in the Autoconf package, but it
doesn't get installed into /usr/local/share, so it's not really
available for use by Autoconf users. This problem is especially
noticeable when you've installed Autoconf via an RPM, or APT package
file, where you don't really have access to the original source anymore.
Of course, I've installed into /usr/local via the source tar.gz, so I
can get it from there if I want...but this is probably a bit more
primitive than you intended.
| 2. autoreconf --install is supposed to install missing files, or
| rather pass appropriate options to the tools that it runs to install
| missing files. This works fine with automake, which supports
| --add-missing, but autoconf has no such option to install missing
| files (even if it had one to install).
Yes, it is 'automake --add-missing' which automatically installs
install-sh, and 'autoreconf --install' relies on automake to do this. If
you don't use automake on your project, then the installation of
install-sh is not automatic, and there is nothing autoreconf will do
I think it should, and I'd be happy to take a stab at a patch to make it
I suppose so, but not one that has bothered me enough to write a patch
(mainly because I always use automake). What bothers me more is that we
don't have a decent way of enforcing, at 'autoconf' time, that install-sh
exists and has executable permissions; it takes 'configure' to test its
existence, and too many people have distributed packages with install-sh
accidentally lacking execute permissions; but still not to the point that
I've done anything about it. Care to write a patch yourself? Or maybe
you could think of autoconf's behavior as free advertising for
I'll take a stab at this patch also.