Re: subdir builds & $objdir/$output_objdir

[Gary V. Vaughan]

Hi Bob!

Bob Friesenhahn wrote:
> In my opinion, libtool 2.0 should not be released unless it is capable 
> of building libraries in subdirectories.

I don't think this is a 2.0 blocker since it isn't a regression with
respect to 1.5.10.  I think it should be fixed asap, and you can even
add it as a KNOWN BUG in README if you like.

2.0 is, IMHO, already a much better build than 1.5.10, and the sooner
we can get it out there the better.

Once the new tests I added to HEAD are working, we should port the fixes
to branch-2-0... either for 2.0 if we are fast enough, or 2.1 if not. 
(Peter, they still fail for me btw, even after your patch).

>  Currently, the rules for a 
> number of compilers/systems use $objdir or $output_objdir as part of the 
> rule.  This causes build failure since $objdir and $output_objdir are 
> just ".libs" and the .libs subdirectory does not exist.
>
> For example, under Digital Unix (OSF), this fragment is generated (and 
> fails):
>
>   -update_registry .libs/so_locations -o 
> Magick++/lib/.libs/libGraphicsMagick++.so.2.0.0
>
> This is due to use of
>
>   -update_registry \${objdir}/so_locations -o \$lib
>
> in the archive_expsym_cmds rule.
>
> The failure under IRIX is similar except that $output_objdir is used to 
> refer to the .libs sub-directory (with similar results).
>
> If there is a replacement variable which refers to the actual object 
> target directory, then it may be used to replace $objdir/$output_objdir 
> in the linkage rules.  Or perhaps, the definition of $output_objdir can 
> be fixed, and it can be used consistently in rules which refer to the 
> .libs directory.

Fixing output_objdir looks to be the sanest solution from here.

Cheers,
        Gary.
--
Gary V. Vaughan      ())_.  address@hidden,gnu.org}
Research Scientist   ( '/   http://tkd.kicks-ass.net
GNU Hacker           / )=   http://www.gnu.org/software/libtool
Technical Author   `(_~)_   http://sources.redhat.com/autobook

signature.asc
Description: OpenPGP digital signature