autoconf-archive-maintainers
[Top][All Lists]
Advanced

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

Re: [patch #8254] AX_BOOST_BASE: fix lib detection on multi-arch systems


From: Peter Johansson
Subject: Re: [patch #8254] AX_BOOST_BASE: fix lib detection on multi-arch systems
Date: Mon, 16 Dec 2013 09:22:27 +1000
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:10.0.12) Gecko/20130108 Thunderbird/10.0.12

Hi Cory,

On 12/13/2013 07:51 PM, Cory Fields wrote:
(Subscribed to the list to reply, sorry for the formatting kludge here)

Libtool has code to figure out which dirs are in linkers search path. Perhaps 
we should borrow that code and let the linker tell us where to look for boost 
rather than patching for every packager.

I don't mind this patch. On the contrary, but I'm just sharing what I though 
might be a better idea. Any thoughts?
Sure, that would be much more reasonable than the current approach by
the module. A more comprehensive fix, though, would require more time
than I am able to spend on this. As far as I can tell, this patch
guards against the only big change distro packagers are likely to make
any time soon, so it should suffice for us.

Fwiw, I've done an implementation of what you mentioned above before,
which pretty much consists of invoking the linker with some dump
option (-Wl,--print-map, -Wl,--trace, etc) and grepping the output, or
probing the resulting object. I found it to be pretty sensitive to the
compiler/linker used. Maybe libtool has a saner approach?

Yeah, I will look into the libtool code in more detail. It won't be on this side of the holidays though.

Seems to me a simpler route would be to avoid trying to find the path
entirely, and just link-test with -lboost_foo, -lboost_foo_mt, etc,
until one hits. I assume there's some need for the path that precludes
this more typical approach?

I was working on a patch like that early this year when I realized the library can be named all kind of funny things including compiler build triplets in the name. The combinations were just too many to enumerate, so I guess that's why original author chose the wild-card solution.



Thanks for the quick and helpful response!

Thanks for catching this and taking the time to submit a patch.

Cheers,
Peter



reply via email to

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