1 |
On Wed, 17 Aug 2011 10:19:06 +0200 |
2 |
Ulrich Mueller <ulm@g.o> wrote: |
3 |
|
4 |
> >>>>> On Wed, 17 Aug 2011, Michał Górny wrote: |
5 |
> |
6 |
> > And isn't a random PROPERTIES value more fragile? If someone uses it |
7 |
> > incorrectly, the results are undefined. With older PMs, results are |
8 |
> > undefined. |
9 |
> |
10 |
> > While having empty SRC_URI and no DEFINED_PHASES guarantees that |
11 |
> > the ebuild won't install a file. That's just per-def, nothing can |
12 |
> > happen. |
13 |
> |
14 |
> Unfortunately, the inverse is not true. There are virtuals that |
15 |
> define phase functions, see virtual/ruby-* for example. |
16 |
|
17 |
And that's another issue in the ebuild. It should be fixed so that |
18 |
phase functions won't be exported in the first place rather than |
19 |
redefining them. This way, PM could assume it doesn't need to execute |
20 |
those at all rather than running dummy code. |
21 |
|
22 |
-- |
23 |
Best regards, |
24 |
Michał Górny |