| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Use plain 'pythonX.Y-config' executable name in the python-config
wrapper instead of querying the complete PYTHON_CONFIG path from
python_export.
Constructing full PYTHON_CONFIG path requires querying ABIFLAGS which
requires the Python interpreter to be installed, which in turn
unnecessarily forces every ebuild using the eclass to carry a build-time
dependency on the Python interpreter.
All current versions of Python correctly install 'pythonX.Y-config'
symlink that points to the correct ABI-suffixed binary. It is therefore
unnecessary to use the full path.
|
|
|
|
|
|
| |
Now that epunt-cxx.eclass is using the patches from
app-portage/elt-patches, remove the remaining patches from
the repository.
|
|
|
|
|
|
|
| |
Switch to using the patches provided through the app-portage/elt-patches
package, like libtool.eclass. Since epunt-cxx is no longer implicit
eutils API but a dedicated eclass, we do not have to worry about
the extra dependency.
|
|
|
|
|
| |
Now that eutils does not inherit epunt-cxx implicitly, we can reverse
the inherit without having circular dependencies.
|
|
|
|
|
|
| |
Stop inheriting the epunt-cxx eclass in eutils implicitly. All the uses
in ::gentoo are fixed already, and this lets us add extra dependencies
in the eclass without affecting most of the packages out there.
|
|
|
|
|
|
|
|
| |
Replace the unnecessary use of extglob to strip file suffix
in make_desktop_entry with a plain ${x%.*}. This is pretty much
equivalent since match to one of the pre-defined suffixes is already
confirmed via the preceding conditional, and avoiding repeating
the suffix list twice makes the code simpler.
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
| |
Make the gnome2_icon_cache_update function update all icon themes rather
than depending on gnome2_icon_savelist to select themes to update. This
makes the function easier to use whenever the developer needs it
explicitly (i.e. knows that themes are installed), while the overhead of
regenerating multiple caches is neglible.
|
|
|
|
|
|
|
| |
Upstream directory for Qt-only translations.
- Check for and remove poqm directory if USE=-nls
- Filter LINGUAS in poqm as well
|
| |
|
| |
|
|
|
|
|
| |
And use @CODE examples where relevant instead as per
eclass-to-manpage.awk
|
|
|
|
|
|
|
|
| |
Currently errors even when set to "yes" ( which is the mechanics
that is implied if the option is missing )
This prematurely tells people about non-issues, and encourages
them to set the wrong value.
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This utility provides informational data describing the given module
names state of installation, either as a version, or as an error
message describing the grade of failure incurred in module loading.
It has the side effect that it most load the module (and its
dependencies) into memory to give a report value, and can be expensive
and have side-effects from Perl code execuring while the module loads,
including (but not limited to) people calling POSIX::_exit
This is the slowest way of inspecting state about a module, as
it must load the module
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is a utility for runtime checking if a module of a given version
is installed from the perspective of Perl, whos opinion could be
different than portage in the event of perl-core/* dual life effects
shortly after a major Perl upgrade.
Use this only if perl_has_module is insufficient, as the overheads
and risk of side effects from this approach are high, given the module
has to be actually loaded for the version comparison to happen.
exits "true" if Perl has the given module installed at the given
version ( or later ), exits "false" otherwise.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is an incredibly fast way to check if Perl considers a module
of the given name installed in any capacity, including broken.
As long as "Foo.pm" is somewhere in @INC, `perl_has_module Foo` will
return true.
Even `perl_has_module threads` will return true on non-threaded perls,
due to that module still being present, and the module only fataling
when loaded.
Whereas `perl_has_module_version threads 0` will always fail on
non-threaded perls.
|
|
|
|
| |
Sync with kde overlay.
|
| |
|
|
|
|
|
|
| |
Thanks to Zdeněk Žamberský for the patch.
Bug: http://bugs.gentoo.org/615896
|
| |
|
|
|
|
|
|
|
|
|
|
| |
* Add missing REQUIRED_USE which otherwise could
lead to spurious failures during the build.
* Do not directly depend on dev-lang/python-exec,
this is done properly via ${PYTHON_DEPS}.
* python-utils-r1 never needs to be inherited
explicitly.
Closes: https://github.com/gentoo/gentoo/pull/4444
|
| |
|
|
|
|
|
|
|
| |
The code wrongly relies on 'bash -n' which is not available in bash 4.2.
Bug: https://bugs.gentoo.org/615642
Reverts: e5e4a35151f64b423c4201a30270a236dab3bf58
|
|
|
|
|
|
|
| |
Alike the PMS-defined phases, sub-phases do not support leaving jobs
behind. To avoid leaving something accidentally, make sure to reap all
jobs (i.e. wait) post running the command. Warn if any stray processes
are found since that indicates a bug in ebuild.
|
|
|
|
|
|
| |
Use 'shopt -p -o' to obtain the current set of 'set' flags, and to
restore them on popping. Unlike $-, it is output in command form (alike
'shopt -p'), removing the need for separate logic in eshopts_pop.
|
| |
|
|
|
|
|
|
| |
Use local variable to limit the scope of ABI variable override. This is
a builtin bash solution that is reliable and simple, unlike the complex
evar_* logic that unnecessarily pollutes global variables.
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
| |
Skia support is now forced off on big-endian platforms, to avoid bug 607662
gtk+:2 is added back as a fixed dependency, as it is still needed on mozilla-52
regardless of whether or not the toolkit is cairo-gtk3, bug 601326
apulse is now an alternative dependency to pulseaudio for USE="pulseaudio"
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Do not export src_compile, as using the default has the same result.
- Precede PKG_CAT and EXPERIMENTAL variables with XEMACS_ for proper
namespace.
- Explicitly assign S, because there is no fallback to WORKDIR in
recent EAPIs.
- Move install directory logic into src_install function.
- Install into ${D}${EPREFIX} rather than ${D}.
Acked by: Mats Lidell <matsl@gentoo.org>
|
| |
|
|
|
|
| |
Closes: https://github.com/gentoo/gentoo/pull/4237
|
|
|
|
|
|
| |
This is a workaround for failing coverage. Coverage is not relevant in
this context, so there's no harm to disable it, although the issue
should be fixed.
|
|
|
|
|
|
|
|
|
| |
Previously the cargo_src_install step fails when using
`cargo_src_install || die` and no man-directory is
included with the crate. This commit corrects this
behavior and ensures it does not die in those cases.
Package-Manager: Portage-2.3.3
|
| |
|
| |
|
|
|
|
|
|
|
| |
Convert the usage of eblits in sys-devel/autoconf into an equivalent
eclass. This makes the ebuilds more readable, more predictable and fixes
compliance with stricter versions of the package manager (i.e. a future
release of Portage).
|