This was originally introduced to work around some behaviour in the
libtool build, however these days it is actively harmful for a number of
packages, where removing additional arguments when -v is present on the
command line can break ABI detection (notably in CMake packages).
Instead, filter out any references to BUILDLINK_DIR from the libtool
scripts, as that should do the same job.
Retain the ability to run the 'scan' wrapper script, as it can be useful
in certain cases, and is required to support the scan-libtool script
anyway.
incorrectly handle the default wrapper scripts on Cygwin.
libtool bug 14579: Disable ar(1) locking on OSX. It creates more harm
than good.
Bump libtool-base and libtool-fortran.
a) refer 'perl' in their Makefile, or
b) have a directory name of p5-*, or
c) have any dependency on any p5-* package
Like last time, where this caused no complaints.
- Initial support for Go.
- On Mac OS X, try to open .dylib as well as .so in lt_dlopenext.
- sysroot support
- man pages for libtool and libtoolize
- various smaller bug fixes
pkgsrc side: remove support for pre-ELF NetBSD
the f2c frontend. It is not recognized as GCC even though there is a gcc
underneath. Thus, use "-Wl," to introduce linker options.
This unbreaks the build of shared libraries written in Fortran on MirBSD
(exotic combination, yes I know) such as blas.
Bump PKGREVISION.
Changes are many, important for pkgsrc:
- much higher speed (up to 30% faster for the overall build of libX11)
- less magic for detection of the tag
The pkgsrc integration of f2c has been changed to be more sane.
libtool itself now depends on the f2c frontend to use proper
auto-configuration and only fakes up the library. This part is
encapsulated in mk/compiler/f2c.mk.
only scheme. Properly distinguish FreeBSD and DragonFly case to prevent
further confusion. Differs from the patch in the PR as it forces the
correct branch to be used.
Bump revision of libtool-base. Claim ownership.
decision making in libtool. This was lost during the update and broke
the build of shared C++ libraries with SunPro on Solaris. Bump revision
of libtool-base.
Noticed by salo and dmcmahill, tested by dmcmahill.
to make decisions about which compiler is being used (gcc, SunPro, etc).
To make this function "correctly" in the presence of the compiler wrapper
scripts, we need to extract the actual compiler command from the wrapper
frame work. So adapt the section in the configure script which already
attempts to deal with some other compiler wrappers to deal with the pkgsrc
wrappers.
This should address PR pkg/33467 and PR pkg/34581.
Bump PKGREVISION.
of postgresql80-client. According to Ralf Wildenhues from libtool,
this fragment is just broken and the pre-1.5.21 was effectively a NOP.
Bump libtool-base revision.
some highlighted changes:
* Fixes minor /tmp race condition bug.
* Integrated Interix support; no longer requires pkgsrc hackery.
* shlibtool is a wrapper using the (now working) "--tag=disable-static".
* Several other bugs fixed, removing some pkgsrc-local changes.
Fix major problem brought to the surface by buildlink3 on Interix. For C++
or F77 libraries only, the soname did not always make it into the final
binary, resulting in runtime link failures in some cases and runtime linking
with a missing major version in others.
This is unfortunately a flag day for Interix + pkgsrc. Identifying all the
PKGREVISIONs normally requiring a bump is a little too unwieldy. However,
existing binaries not experiencing the runtime link failure will continue
to run, but will be linking against the unversioned ".so" at runtime until
the binaries have been rebuilt.
While here, fix a couple wrong assumptions in the libtool config for
Interix. shlibvar_overrides_runpath should have been "no", and
hardcode_direct should have been "yes".
makes -avoid-version work properly on platforms using $versuffix2 (e.g., OpenBSD,
FreeBSD, DragonFly). Bump PKGREVISION (and will bump LIBTOOL_REQD shortly).
MirOS change by Benny Siegert <bsiegert@mirbsd.org>.
Noted in private mail by Thorsten Glaser <tg@66h.42h.de>.
