1 |
On Sun, 6 Nov 2016 11:19:02 +0100 |
2 |
Gerhard Bräunlich <wippbox@×××.net> wrote: |
3 |
|
4 |
> Dear gentoo devs |
5 |
> In August I reported the following bug: |
6 |
> https://bugs.gentoo.org/show_bug.cgi?id=590404 |
7 |
> James Le Cuirot suggested to automatically use the --with-sysroot |
8 |
> configure switch when cross-compiling ( |
9 |
> checking for its presence in --help like for --docdir). |
10 |
> |
11 |
> I proposed a patch in the above bug report for phase-helpers.sh. However |
12 |
> as the current version of phase-helpers.sh calls " |
13 |
> ___eapi_econf_passes_", I suppose that we should add a corresponding |
14 |
> ___eapi_econf_passes_--with-sysroot function in eapi.sh. |
15 |
> There I am not sure which EAPIs are allowed to be modified. |
16 |
> |
17 |
> What do you think? |
18 |
|
19 |
Sorry for not replying to the bug report recently, I'm battling on many |
20 |
fronts. That function would not make sense unless --with-sysroot is |
21 |
introduced into PMS. We could do that but many existing ebuilds would |
22 |
fail to cross-compile and with hundreds of ebuilds still sitting at |
23 |
EAPI 0, we could be waiting a long time for fixes. I noted that while |
24 |
PMS does say which options you must pass, it doesn't say that you |
25 |
cannot pass additional ones. I've tested it a lot and I've never seen |
26 |
--with-sysroot break any ebuilds, while it fixes very many, so I |
27 |
believe it would be safe to pass this option on all EAPIs. We would |
28 |
probably only pass it when SYSROOT!=/ anyway so it wouldn't affect many |
29 |
users. This is more or less what you did in your own patch. We are |
30 |
discussing exactly what influence SYSROOT should have on Portage in |
31 |
bug #573306 so that we can cement this into EAPI 7 but I believe we can |
32 |
still rely on SYSROOT in earlier EAPIs when it's set to the same value |
33 |
as ROOT. |
34 |
|
35 |
-- |
36 |
James Le Cuirot (chewi) |
37 |
Gentoo Linux Developer |