libtool
[Top][All Lists]
Advanced

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

Re: libtool related build questions


From: Christopher Hulbert
Subject: Re: libtool related build questions
Date: Thu, 14 Dec 2006 15:04:25 -0500

On 12/14/06, Ralf Wildenhues <address@hidden> wrote:
* Christopher Hulbert wrote on Thu, Dec 14, 2006 at 08:11:20PM CET:
> On 12/14/06, Bob Rossi <address@hidden> wrote:
> >
> >> >   *** Warning: This system can not link to static lib archive
> >/home/bobbybrasko/log4cxx/apr-util/src/prefixdir/lib/libaprutil.la.
> >> >   *** I have the capability to make that library automatically link in
> >when
> >> >   *** you link to this library.  But I can only do this if you have a
> >> >   *** shared version of the library, which you do not appear to have.
> >>
> >> They are sometimes bogus.  Nobody has gotten around to fixing that yet.
>
> I used to get these a lot.

They are very very likely _not_ bogus in your case, and your case very
very likely has nothing whatsoever to do with Bob's.  Linking shared
libraries against static ones is a bad idea in general, and most often a
hack.

IIRC the warning was bogusly output sometimes when the "static" library
really was a convenience archive.  And there are cases on w32 systems
where you simply have to link against static libraries in some cases.

> I have not removed this, so I'm not sure if the most recent libtool
> still issues these warnings and refuses to build the shared libraries.

Wait until you move from you GNU Linux/x86 to a x86_64 one, for example
(I'm guessing here; please correct me if wrong), to see your setup blow
up.

It works fine on x86, x86_64, and win32. I have not tried win64, OS X,
solaris, etc. The message is not "bogus" in the way you define that
term, but to _some_ extent it is "bogus" because the message clearly
says the system cannot link against a static library which is flat
wrong. It may be <insert your degree here> ill advised to do so, but
it at least appears to work and without any problems (at least none
I've found in the last year or two). I'm not saying this is an
endorsed way of performing the operation, but I find it better than
the alternative of delivering 4 shared libraries so a user can call a
single function from Matlab.

Just my thoughts though,
Chris


Hope that helps.

Cheers,
Ralf





reply via email to

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