| 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>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is analogous to tc-getFOO failures for
variables like CC in non-native ABI's.
gcc-version will always do the wrong thing unless
ABI==${DEFAULT_ABI} (unless they have identical
CHOSTS somehow): if crossdev is installed with a matching
CHOST then the version of it's compiler will be returned
(in this case it apparently succeeds but returns an
incorrect result); otherwise, it will fail to find
/etc/env.d/gcc/${CHOST}-config, dump a bunch of garbage
into stdout, and (I surmise) nothing into stdout
(without dying).
This works around it but this is starting to become pretty
onerous and I'm starting to think that it may be time to
begin considering an automated solution to these toolchain*
consumption problems under multilib-build non-native ABI's.
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
qt4-build clears the myconf variable after folding it in,
presumably to prevent one from accidentally applying the
same myconf twice -- yet, that is precisely what we want to
do when we are using it under multilib-build auspices.
Were it not for the fact that we happened to be doing
parallel foreach_abi iteration in this framework, it would
never have worked, due to this -- certainly, that weird
side effect is not one we wish to rely on or even that is
likely to make it upstream.
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
|
|
|
|
| |
some using my-god-its-full-of-quotation-marks.eclass
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|
|
Signed-off-by: Gregory M. Tuner <gmt@be-evil.net>
|