Re: Guile test-ffi uses an unportable assumption

[Eli Zaretskii <eliz@gnu.org>]

> From: Andy Wingo <wingo@pobox.com>
> Cc: guile-devel@gnu.org
> Date: Sat, 23 Jul 2016 23:17:58 +0200
> 
> > It assumes that libltdl can only produce a handle for a symbol in the
> > the program itself, as opposed to those loaded from shared libraries.
> > It tries 'strerror'.  This cannot work on MS-Windows, unless the
> > program was linked with -export-dynamic, which is not true for
> > the test program.
> 
> Interesting.  The test program is a Scheme script which looks up
> strerror in the dlopen(NULL).  Libguile is compiled with -export-dynamic
> but the guile binary is not.  Which part needs to be compiled with
> -export-dynamic, do you think?

The guile binary.  Here's what the libltdl documentation has to say
about this:

 -- Function: lt_dlhandle lt_dlopen (const char *FILENAME)
    [...]
     If FILENAME is `NULL' and the program was linked with
     `-export-dynamic' or `-dlopen self', `lt_dlopen' will return a
     handle for the program itself, which can be used to access its
     symbols.
    [...]
     If you use `lt_dlopen (NULL)' to get a HANDLE for the running
     binary, that handle will always be marked as resident, and
     consequently cannot be successfully `lt_dlclose'd.

However, in the case in point even compiling guile with -export-dynamic
won't help, because the function being used by the test, strerror,
comes from the C library, and is not statically linked into the guile
binary.  So I suggest to modify the test to use a function that is
part of the guile binary's own sources.