Daniel,
I just submitted a bug report against libtool 2.4.3 upstream
regarding the broken --program-prefix=g.
My concern with that version would be whether that breakage is limited to
just the build machinery in the
libtool2 package or a more general flaw introduced in the package which
will generate configure files
that break --program-prefix=g elsewhere.
Jack
On Thu, Oct 30, 2014 at 11:17 PM, Daniel Macks <dma...@netspace.org> wrote:
> There's a new upstream version that also has the fix, builds and seems to
> pass tests:
>
> 138 tests behaved as expected. 31 tests were skipped.
> Only change was that the PatchScript was looking for
> libltdl/m4/libtool.m4, which has now moved up a level to m4/libtool.m4.
> However, the installation-time filename munging did not seem to work.
> Despite seemingly the same "transform" related entries in Makefile,
> libtoolize is directly created as "libtoolize" rather than "glibtoolize"
> (which I only noticed because the next InstallScript command failed when
> trying to symlink). It's no problem to switch the sense of linkname and
> targetname (which is real vs link), but clearly something is misbehaving
> and I don't know what.
> Unrelatedly, TestDepends could bump from automake1.11 to automake1.14.
> It's still a circular dependency, but may as well reduce dependence on such
> an old am that isn't used by so many other packages.
> dan
>
>
> On Thu, 30 Oct 2014 22:47:25 -0400, Jack Howarth <
> howarth.at.f...@gmail.com> wrote:
> Max, We need an fixed libtool2 package for Yosemite. The current one
> has the libtool.m4 bug which confused 10.10 with 10.1 and results in shared
> library being linked with "-dynamiclib -flat_namespace" rather than the
> appropriate "-dynamiclib -undefined dynamic_lookup". While this is fixed in
> the new libtool 2.4.3 release, they appear to have
> broken --program-prefix=g in that release.
>
>> Attached is a libtool2-2.4.2-4 packaging which uses the same back port
>> of the fix as home-brew.
>> https://raw.githubusercontent.com/Homebrew/homebrew/master/
>> Library/Formula/libtool.rb
>>
>> Index: libtool2.info
>> ===================================================================
>> RCS file: /cvsroot/fink/dists/10.7/stable/main/finkinfo/devel/lib
>> tool2.info,v
>> retrieving revision 1.5
>> diff -r1.5 libtool2.info
>> 19c19
>> < Revision: 3
>> ---
>> > Revision: 4
>> 26c26,28
>> < Source-MD5: d2f3b7d4627e69e13514a40e72a24d50
>> ---
>> > Source-MD5: d2f3b7d4627e69e13514a40e72a24d50
>> > PatchFile: %n.patch
>> > PatchFile-MD5: da6decbe2f0b7a3572db27c4bddcadb5
>> 29a32
>> > %{default_script}
>>
>> This version passes 'fink -m' and also produces a glib tool which
>> properly links
>> the files in cyrus-sasl2-shlibs on Yosemite. Jack
>> ps The current fink 0.38.2 release now treats the absence of the TWOLEVEL
>> flag from the output of 'toool -hv' on a non-bundle shared library as a
>> marker for the use of -flat_namespace and uses a fatal validation error.
>> The inappropriate use of -dynamiclib -flat_namespace causes major runtime
>> regressions for packages like gmp/mpfr/gmp and libgfortran from FSF gcc.
>> -------------------------
>>
>> ------------------------------------------------------------
>> ------------------
>>
>> -------------------------
>>
>> _______________________________________________
>> Fink-devel mailing list
>> Fink-devel@lists.sourceforge.net
>> List archive:
>> http://news.gmane.org/gmane.os.apple.fink.devel
>> Subscription management:
>> https://lists.sourceforge.net/lists/listinfo/fink-devel
>>
>>
> --
> Daniel Macks
> dma...@netspace.org
>
>
>
------------------------------------------------------------------------------
_______________________________________________
Fink-devel mailing list
Fink-devel@lists.sourceforge.net
List archive:
http://news.gmane.org/gmane.os.apple.fink.devel
Subscription management:
https://lists.sourceforge.net/lists/listinfo/fink-devel
