Re: cross-compiling, RPATH, and DESTDIR vs. AC_ARG_WITH

[Ralf Wildenhues]

* Ray Lehtiniemi wrote on Sat, Jan 22, 2005 at 11:14:42PM CET:
> On Sat, Jan 22, 2005 at 09:49:14AM +0100, Ralf Wildenhues wrote:
> > 
> > * Ray Lehtiniemi wrote on Tue, Jan 11, 2005 at 08:52:47PM CET:
> > 
> > Please consider updating to a more recent Libtool.
> > It will not solve your problem, though.
> 
> for cross-compiling, is it recommended to follow the 1.5 series or
> the 1.9 series? (or 2.1?)

Rather the newest.  If some bugs turn out to need substantial rewrite to
fix, I am not the one to backport.  Simple backports are less
problematic.

> > Libtool does not yet support cross-compiling very well.  We would
> > like to change that, though.
> 
> i've got a few cross-projects for which i was hoping to rely upon
> libtool, so i'm interested in filing bug reports and following up on
> them.

Great!

> i don't see a bugzilla on the home page... is this mailing list the
> primary vehicle for tracking issues?

Well, there is the bug-libtool list for bugs in released versions,
libtool for general discussion, and libtool-patches for development (all
patches go through there).  If you want to work with unreleased
versions, the last one is the one to go.
(BTW, I just posted a patch regarding cross-compiles there.)

There is also the savannah bug/patch tracking system, but developers
seem to rather like the mailing lists.

> i don't normally run libtool explicitly... i just run autoreconf,
> configure, and make...  how and where would i invoke libtool directly
> to produce this output?

Well, just look what `make' invokes.  Then do that same thing yourself.

> instead, i have pared down my example to a minimal testcase and
> attached a gzipped tarball for you to examine.  it turns out cross-compiling
> isn't a factor, and the bug shows itself even on a native build, so
> it should be fairly easy to reproduce on your end.
> 
> please extract the attached tarball and have a look at the 'RUNME'
> script in the top level dir.   that's the minimal test case i could
> come up with to exhibit the faulty RPATH.

Thanks.  Will look at that eventually (unless someone else beats me to
it).

Regards,
Ralf