[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Autoconf: Expansion of installation directory variables
From: |
Hans-Bernhard Broeker |
Subject: |
Re: Autoconf: Expansion of installation directory variables |
Date: |
4 Jan 2002 14:37:04 GMT |
Brad Hards <address@hidden> wrote:
> One thing that did strike me as a bit strange (probable bug) is the way in
> which directories are expanded. I expected that my Makefile would contain
> fully expanded variables, not relative variables.
That could have been done, but would make some advanced usages hard to
achieve. E.g., in the existing scheme, you can
make install prefix=/some/where exec_prefix=/some/where/some_platform
to install the package in a different place than it was configured
for. Comes in very handy at times, esp. if you're using a package
manager like GNU stow.
> No problem, except that I don't need prefix or exec_prefix for
> anything. So I hadn't written them into Makefile.in, and the INSTALL
> stuff ended up as /sbin, /lib, and /include (which is not what my
> configure script says will happen, since it firmly believes in the
> prefixes).
In a nutshell: a Makefile.in without the lines
address@hidden@
address@hidden@
is buggy. If the autoconf docs didnt point that out clearly, that
would IMHO constitute a documentation bug.
Looking around a bit I found it is documented to some extent, but in a
place you probably wouldn't look: it's in the "Porting from autoconf
1" section:
info autoconf Upgrading "Changed Makefiles"
Of course, the canonical GNU answer would be: use automake to generate
your Makefile.in. It knows about this and lots of other fine details
already.
--
Hans-Bernhard Broeker (address@hidden)
Even if all the snow were burnt, ashes would remain.