| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
| |
During multilib_get_enabled_abi_pairs, we employ the use helper to
determine whether a particular flag is enabled. All well and good,
except that this crashes with a QA thingy (maybe only on developer
profiles?) when that flag is not in IUSE, which, notably, is not
a requirement; indeed, we have a feature in there to create precisely
this circumstance by design.
Easy solution: use use_if_iuse instead.
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
|
|
| |
an error)
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
|
|
| |
multilibutized versions
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
A few mistakes spotted in multilib-build.eclass are fixed here:
Use the wrapper variable more
Add comments to the very long if/else/for/etc clauses in
multilib_prepare_wrappers which are starting to get pretty hard to read.
Fix a thinko in multilib_prepare_wrappers which broke ^wrapped headers.
Add the missing multilib_get_native_abi which was referred to but never
actually implemented as a replacement for the old multilib_get_best_abi
function.
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This function is incredibly difficult to implement correctly. For all
intents and purposes, without employing extremely aggressive hacks,
it simply is not possible to nest multilib abi iterations in gentoo.
Anyhow, my only remaining use of it was pretty stupid anyhow, and
causing bugs too -- the creation of the multilib executable
wrappers for @wrapped executables. Instead of the fairly insane
approach of nested iteration, I now simply build the wrapper
during native abi subphase execution.
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
|
|
|
|
| |
The updated format of multilib-build variants in multibuild broke the
native ABI constants in @wrapped executables. This restores them to
their former glory by trimming the extra junk out.
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
|
|
| |
MULTILIB_CHOST_TOOLS merging
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
|
|
| |
upstream now handles better)
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
|
|
| |
multilib_get_enabled_abis
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Upstream has a clean approach to AUTOTOOLS_IN_SOURCE_BUILD
support in autotools-multilib which is either a much-simplified port
of mine or a parallel implementation of the feature.
Either way, mine was pretty much a relic of incremental hacking with
no particular benefit above what upstream was doing.
Subtle differences between my implementation and upstream's was also
causing a compile failure problem in media-gfx/exiv2. This patch makes
our overlay implementation work just like upstream's, hopefully fixing
that without breaking anything else.
|
|
|
|
|
|
| |
gnome2-multilib_gdk_pixbuf_update
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This adds support for MULTILIB_CHOST_TOOLS-based wrappers for
gtk-query-immodules-{2,3}.0 along-side the old
MULTILIB_WRAPPED_EXECUTABLES-based naming conventions to
gnome2-multilib_query_immodules{,_gtk3}. This should provide
a clean migration path for any programs using the old names.
This is part of a broader plan to migrate all
MULTILIB_WRAPPED_EXECUTABLES usage to MULTILIB_CHOST_TOOLS, since,
it is clearly retarded (UK residents: please read 'mentally
handicapped' or whatever you find tolerably inoffensive) to have both
of these features around serving exactly the same purpose.
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
|
|
|
| |
Pypy doesn't support CFLAGS fetching in python-utils-r1, so don't
bother checking for multilib; our hacks won't work for it anyhow.
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Here we merge upstream (nb: these two eclasses, and the ebuilds in
dev-db/mysql are following the mysql overlay, not gentoo-x86 like most
other stuff in this overlay) deltas.
Something changed (I suspect, something in my workstation's kernel
configuration that was breaking systemtap entirely, causing mysql
to steer clear of it and, until now, thereby hiding this bug on my
test box)
So make sure to explicitly disable systemtap in the minimal build (which
will also be used for non-best-ABI's.
Otherwise, this is 99% robo-merge with no intersting new content.
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|