1 |
W dniu śro, 07.02.2018 o godzinie 14∶18 +0100, użytkownik Ulrich Mueller |
2 |
napisał: |
3 |
> > > > > > On Wed, 07 Feb 2018, Michał Górny wrote: |
4 |
> > > -# @FUNCTION: bzr_src_prepare |
5 |
> > > -# @DESCRIPTION: |
6 |
> > > -# Default src_prepare(), calls bzr_bootstrap. |
7 |
> > > -bzr_src_prepare() { |
8 |
> > > - bzr_bootstrap |
9 |
> > > } |
10 |
> > Hmm, unless I'm mistaken, this can cause another definition |
11 |
> > of src_prepare to start applying to ebuilds. |
12 |
> |
13 |
> That's right, but wasn't relying on inherit order considered a QA |
14 |
> violation? In other words, shouldn't an ebuild define an explicit |
15 |
> phase function if it inherits more than one eclass exporting that |
16 |
> function? |
17 |
|
18 |
Not that I know of. However, I was actually more worried about ebuilds |
19 |
whose authors didn't even notice it. |
20 |
|
21 |
> > You can submit a PR with this change to get md5-cache with exported |
22 |
> > phase data suitable for comparison. |
23 |
> > Or... given the popularity of the eclass, you can check by hand ;-P. |
24 |
> |
25 |
> Done so for the Gentoo repo (before posting the patch). Or rather, |
26 |
> I have checked that WORKDIR has identical contents after the prepare |
27 |
> phase. |
28 |
|
29 |
Good enough for me. |
30 |
|
31 |
-- |
32 |
Best regards, |
33 |
Michał Górny |