| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This logic is highly experimental, but it does work for me
( though I don't do what ago does with LD because its too spicy )
This set of changes patches acelib so that all gcc/cc/ar calls
can be overloaded externally, though this somewhat defeats the point
of most of these files, as some of them only differed in wether they
said "cc" or "gcc".
The downside is I *had* to rework how the elibc_glibc block worked,
because as it was *conditional*, and patched lines adjacent to the lines
changed in -toolchain.patch, the existing patch could *only* be adapted
to work dependent on the -toolchain patch.
Instead, the essential variable overrides are performed externally
via `make` arguments, as this has the same outcome.
Any logic I didn't fully understand I left in place.
And I had to get clever with names, because acelib already has an
internal variable called "CC", and its value .... contains then entire
set of compile parameters for the compiler!
Also, the use of these TARGET_ seems to be less prone to conflict with
values that do similar things in perl space, where doing something like
LD="something-ld" can be fatal, as perl space doesn't expect LD to be an
actaul LD, but a CCLD, and passes flags LD doesn't support.
But there will *probably* be a few bugs in this approach, I'm no Make
wizard.
Bug: https://bugs.gentoo.org/723152
Package-Manager: Portage-2.3.99, Repoman-2.3.22
Signed-off-by: Kent Fredric <kentnl@gentoo.org>
|
Kent Fredric
GitWeb