| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
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>
|
|
|
|
|
|
|
|
|
|
|
| |
Remove dev-db/sqlite, dev-libs/libgcrypt, dev-libs/libgpg-error,
dev-libs/libtasn1, dev-libs/nettle and sys-devel/gettext. Move
package.unmask into the multilibpython-base profile since for unknown
reasons, it is not working right in profiles/. Bump
app-emulation/emul-linux-x86-baselibs to 20131008-r22 and merge upstream
changes.
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
| |
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
|
|
| |
no multilib requirement
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
|
|
|
|
|
|
| |
Don't know how this escaped my attention until now but packages
consuming libopts are going to need a multilib autogen. Since autogen
was putting libopts's .pc file in PREFIX/share (which correctly refers
to libdir=/usr/lib64), I have moved it -- hopefully this is correct and
won't break anything -- to /usr/lib64/pkgconfig.
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>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Slightly tricky multilib-utization, mostly because
I went out of my way to preserve the CXX variable
export across the ABI's in all the phase functions
after configure. (Not sure if it matters -- quite
probably, it doesn't, but erring on the side of not
breaking shit that worked before can't be completely
terrible).
To preserve the variables, I came up with a "trick".
It is an unfortunate design flaw -- one that I'm
contemplating fixing -- of the ehook API that no provision
is made for the explicit management of variable lifetimes.
Due to this, there's no elegant way to make a variable's lifetime
precisely that of the ABI (except to go in and implement that in
an ad-hoc manner depending on what framework you're consuming).
In this case, and probably many similar cases, a sufficient solution is
to rely on the registration-time ordering of hook invocations by
encapsulating code to set up and tear down a variable or set of
variables, and then stack per-abi-pre and per-abi-post registrations
somewhere before any hooks are registered for the relevant phase functions
and at the end of the ebuild.
It's kinda ugly though -- a multiple-registration API would certainly
make it easier on the eyes.
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
|
|
|
| |
I have left a handful of lingering usages that would
just create a merge for me a few commits down the line.
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If emacs is installed but po-mode is not, the an error occurs
during the build. Perhaps that is what the emacs?( app-emacs/po-mode )
in PDEPEND was supposed to roughly capture. However, it doesn't.
For example, although I'm not an emacs user, for some reason, I had the
USE flag set and xemacs installed. This resulted in a repeatably
failing build.
There is probably some more sophisticated solution to this problem that
allows to make the emacs dependency once again conditional; for a start,
though, I'd rather have a repeatable, shitty circumstance than a
non-repeatable pseudo-optimized one.
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>
|