On 04-Jan-01 at 18:18, Alexander Mai ([EMAIL PROTECTED]) wrote:
>
> On Thu, 4 Jan 2001 11:04:32 -0500 (EST), Rick Scott wrote:
>
> >The immediate problem that I see is the following
> >
> >LDADD= -L$(top_builddir)/src/.libs -lXbae \
> > -L$(top_builddir)/../Xm$(lib_version_suffix)/.libs \
> > @MOTIF_LIBS@ -lXm @X_LIBS@ -lXt ${X_PRE_LIBS} -lXext -lX11
> >${X_EXTRA_LIBS}
> >
> >The second line ensures that Xm is picked up from the LessTif source tree,
> >no matter what was specified with --with-motif-libs, which is contained
> >in @MOTIF_LIBS@. Now if LessTif was not compiled, every thing will be fine,
> >because there will not be a libXm.so in the LessTif tree. On second
> >thought, it won't be fine because it will be picking up the wrong headers
> >also.
> >This is one of the _many_ problems with compiling a library that depends
> >on another library that is being built at the same time.
>
> Well, we could and should make all references from Xbae/Xlt to
> any paths above the library topdir _only_ if ... hmm... not sure
> what to write ....
> Insert:
> - if a specific flag given
> - if actually the headers/libs are in ../foo
> - if no -with-motif is given
Which is the biggest part of the problem. There are so many situations it's
almost impossible to anticipate them all. And if you do, it will be so
complicated to configure that you won't get it right. Originally the
--enable-motif options were to locate Motif _for_the_tests_. We now have at
least 3 versions of LessTif, both installed and not installed, vendor supplied
Motif's, OpenMotif installed and not yet installed. That's somewhere around 8
factorial combinations that we are trying to deal with!!! Not to mention all of
the infinite places that _all_ of the above could get installed into.
>
> Also note that my today's changes to Xbae/configure.in, acinclude.m4
> may have broken the configure stuff temporarily!
>
>
> ---
> Alexander Mai
> [EMAIL PROTECTED]
>
>
>
>
