| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
| |
The firefox ebuild received fixes for various arm configurations (neon, hardfp/softfp)
but these fixes should also be applied to the other mozilla packages. This commit
removes them from firefox and adds them to mozconfig.
|
| | |
|
| |
|
|
|
|
| |
See local PYTHON_COMPAT setting in _distutils-r1_run_common_phase.
Fixes: 33380ad5e121 ("python*-r1.eclass: Commonize PYTHON_COMPAT processing, cache the result")
|
| |
|
|
| |
setfattr
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
| |
Fix python_gen_useflags() and python_gen_cond_dep() to output correct
flag name when only a single implementation is listed in PYTHON_COMPAT.
In this case, the PYTHON_SINGLE_TARGET flags are not emitted
and PYTHON_TARGETS are used directly instead.
|
| |
|
|
|
|
|
|
|
| |
Introduce a common _python_set_impls function in python-utils-r1.eclass
that validates and processes PYTHON_COMPAT, then stores the result in
_PYTHON_SUPPORTED_IMPLS and _PYTHON_UNSUPPORTED_IMPLS variables. Reuse
those variables in all python-r1 suite eclasses, effectively reducing
code duplication and providing cache for repeated implementation support
checks.
|
| | |
|
| | |
|
| |
|
|
| |
This reverts commit cf00534e64f93c0992faf2584a970f1910afcc5a.
|
| | |
|
| |
|
|
|
|
|
| |
This was only needed in readme.gentoo_create_doc() for a single call
of eshopts_{push,pop}. Replace it by saving the set of options in a
variable. Die if writing the temp file in readme.gentoo_create_doc()
fails.
|
| |
|
|
| |
Gentoo-bug: 568892
|
| |
|
|
|
|
|
|
|
|
|
| |
Modify genpatches logic to delay setting UNIPATCH_LIST_GENPATCHES until
src_unpack(). The value of UNIPATCH_LIST_GENPATCHES depends on state of
'use experimental', therefore it can not be obtained in global scope.
Therefore, handle_genpatches() will only set the URI when called in
global scope, and it will set UNIPATCH_LIST_GENPATCHES on request which
is done during kernel-2_src_unpack().
Fixes: https://bugs.gentoo.org/show_bug.cgi?id=566520
|
| | |
|
| | |
|
| | |
|
| |\ |
|
| | |
| |
| |
| | |
.packlist files and keep/fix them in EAPI=6
|
| | | |
|
| | |
| |
| |
| | |
PERLQAFATAL, eqawarn, and (in EAPI=6) eutils.eclass
|
| | | |
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
parallel", add features
* The variable SRC_TEST used to allow some control over the testing phase (run
it, run tests parallelized, skip it). Named DIST_TEST in EAPI=6 now for
consistency.
* Add functionality for ebuild authors: even if the ebuild specifies that e.g.
tests should not be run or run single-threaded, allow in EAPI=6 to override
that from make.conf or package.env.
* Rewrite of the test phase for EAPI=6. Adds options to control test verbosity
and switch network tests (something that might or might not work depending on
upstream cooperativity).
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
avoid MY_ magic
This eclass is full of magic of the type of "Set MY_xxx before inheriting,
and I'll automatically use it somehow instead of xxx". I think this is majorly
silly since 1) MY_xxx is no proper API, and 2) it was never documented anyway.
So, in EAPI=6 ignore MY_PN, MY_P, and MY_S and use DIST_PN and DIST_P
instead. In addition also replace MODULE_ control variables with their
DIST_ analogue.
The DIST_ prefix is more consistent with upstream terminology since
what we're downloading from CPAN are not actually "modules" (like CGI.pm)
but "module distributions" (like CGI.tar.gz containg a set of modules).
In addition it avoids clashes in case of kernel module build systems. We
can't use PERL_ since this is picked up by perl itself.
|
| | |
| |
| |
| | |
eapply_user)
|
| | | |
|
| | | |
|
| | |
| |
| |
| | |
GENTOO_DEPEND_ON_PERL_SUBSLOT, simplify dependency
|
| | | |
|
| | |
| |
| |
| |
| |
| |
| | |
Use 'cp -R' for distutils-r1_copy_egg_info() as the '-r' option triggers
triggers undesired '-L' behavior wrt symbolic links.
Fixes: https://bugs.gentoo.org/show_bug.cgi?id=568692
|
| | |
| |
| |
| |
| |
| |
| | |
Use 'cp -R' for multibuild_copy_sources() as the '-r' option triggers
triggers undesired '-L' behavior wrt symbolic links.
Fixes: https://bugs.gentoo.org/show_bug.cgi?id=568692
|
| | |
| |
| |
| | |
with ulm via IRC to let easier migration.
|
| |/ |
|
| | |
|
| | |
|
| |
|
|
| |
building, bug #511946
|
| |
|
|
| |
Signed-off-by: Brian Evans <grknight@gentoo.org>
|
| |
|
|
| |
Signed-off-by: Lars Wendler <polynomial-c@gentoo.org>
|
| |
|
|
| |
support and explain the people the changes regarding it stopping to export phase functions (#520094#c6)
|
| |
|
|
| |
support (#520094#c6)
|
| |
|
|
|
|
|
|
|
|
|
|
| |
This eclass is a revision to the previous mysql-multilib to clean up
the duplicated code.
This includes specific dependencies for certain providers.
Provider specific information will be kept in the provider's ebuilds.
Only common items should be placed here and possibly exceptions to those
rules.
Signed-off-by: Brian Evans <grknight@gentoo.org>
|
| |
|
|
|
|
|
| |
We have dev-lang/go provide `go` and `gofmt`, and having gcc include
these in its BINPATH causes gcc-config to clobber the go package with
symlinks. Drop the gcc versions for now (they're copies from the go
package anyways) until we have a better solution.
|
| |
|
|
|
|
|
|
| |
When gcc installs programs into BINPATH, it sometimes links them so that
the $CHOST-xxx and xxx progs are the same. The logic to standardize the
naming in BINPATH then triggers a warning as they're the same file.
Manually delete the target before running the mv to avoid that business.
|
| | |
|
| |
|
|
| |
that EAPI as explained at bug #520094
|
| |
|
|
|
|
| |
This reverts commit 164ed90923c11957a3908a0f56607208232a8133.
bug #520094 needs to be handled first
|
| | |
|
Ian Stakenvicius
Brian Evans
Zac Medico
Anthony G. Basile
Michał Górny
Michael Palimaka
Michael Sterrett
Ulrich Müller
Christoph Junghans
Davide Pesavento
Mike Frysinger
Andreas K. Hüttel
Pacho Ramos
Lars Wendler
GitWeb