1 |
On Thursday 07 October 2004 08:33, Ned Ludd wrote: |
2 |
> On Wed, 2004-10-06 at 18:53, Jason Stubbs wrote: |
3 |
> > On Thursday 07 October 2004 06:29, Michael Tindal wrote: |
4 |
> > > On Wed, 2004-10-06 at 19:57 +0900, Jason Stubbs wrote: |
5 |
> > > > On Wednesday 06 October 2004 12:09, Michael Tindal wrote: |
6 |
> > > > > The format is something like |
7 |
> > > > > category/pkgname |
8 |
> > > > > CFLAGS="...";LDFLAGS="...";USE="...";FEATURES="..." |
9 |
> > > > |
10 |
> > > > Unless you like things not compiling due to missing dependencies, |
11 |
> > > > don't use USE in this file. Same deal with ACCEPT_KEYWORDS and |
12 |
> > > > anything else that could possibly affect dependencies (which is |
13 |
> > > > nothing that I can think of at this stage). |
14 |
> > > |
15 |
> > > That wasn't really the point of the exercise (USE and the like). |
16 |
> > > Portage already has files for that, so using this for that is |
17 |
> > > pointless. It was more for CFLAGS and LDFLAGS and such, and the example |
18 |
> > > I gave was off the top of my head. |
19 |
> > |
20 |
> > No problem. I was mostly just pointing it out for the onlookers and to |
21 |
> > preempt invalid bug reports. Also worth pointing out is that most |
22 |
> > FEATURES would not be honoured as well, as bashrc is only used on the |
23 |
> > bash side of things whereas the python side of portage handles most of |
24 |
> > them. |
25 |
> |
26 |
> Portage learns about FEATURES after it's sourced the ebuild in the |
27 |
> depend phase right? |
28 |
> If so then it would seem that those functions-FEATURES/USE which have |
29 |
> todo with depgraph creation could be exported in the 'depend' phase from |
30 |
> a bashrc. Is my logic incorrect? |
31 |
|
32 |
The depend phase is only ran to build the cache. If the cache is current, it |
33 |
is read directly. If not, it is built and then read directly. The FEATURES |
34 |
that are enabled are built from the configuration (ie emerge info) and |
35 |
whatever the package might have in its RESTRICT. |
36 |
|
37 |
Actually, I was wrong about ACCEPT_KEYWORDS above. It would simply be ignored |
38 |
- unless ebuilds are actually reading that variable. As far as I know, |
39 |
ebuilds only test the ARCH variable. |
40 |
|
41 |
Regards, |
42 |
Jason Stubbs |
43 |
|
44 |
-- |
45 |
gentoo-dev@g.o mailing list |