autoconf-archive-maintainers | |
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: AX_BOOST_BASE
From: |
Peter Johansson |
Subject: |
Re: AX_BOOST_BASE |
Date: |
Tue, 05 Nov 2013 11:09:33 +1000 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:10.0.12) Gecko/20130108 Thunderbird/10.0.12 |
Hi Marcin,
On 11/05/2013 01:36 AM, Marcin Zalewski wrote:
Peter,
I should start saying I'm neither the writer of AX_BBOST_BASE nor the
maintainer, but your emailed just made me curious.
I get a syntax error from expr.
It's always preferable if you copy & paste the error the error message
you experience.
The problem is that currently the
script tries to figure out the version of boost in BOOST_ROOT from the
directory name in BOOST_ROOT. This works only if boost root is in a
standard directory name (e.g., boost_1_53_0). I have a client who
wants to use his unstaged boost that he has in a non-standard
directory.
Makes sense and I agree looking for version info in 'version.hpp' is
better like you patch does.
The reason I am ending up in this region of code is that there is no
system-wide boost that meets the version requirement. I have
recommended to my client that he use BOOST_ROOT for his unstaged boost
(for some reason my client does not want to install the unstaged
version of his boost library in his home directory).
I asked because you mentioned that you use '--with-boost' and I assumed
you used it with some argument such as '--with-boost=/my/path/to/boost'.
IIUC, using '--with-boost' like that will set $as_boost_path and
test "$ac_boost_path" != ""
on line 162 will be true and you will never get into the else clause
that involves your patch. I realize now that you probably use
'--with-boost' with no argument in which case $ac_boost_path is not set
and you could end up in the the patched code
FWIW, using --with-boost has no effect, i.e., './configure
--with-boost' should be synonymous to './configure'.
So, it seems that the version checking for BOOST_ROOT case makes a lot
of assumptions. An even more problematic issue is the difficulty of
using unstaged boost, but that would be a bigger patch.
I agree. Most of the boost macros would probably benefit from being
revised, to reduce number of assumptions and make the shell code more
portable.
Cheers,
Peter