1 |
On Thu, 12 Feb 2009 09:37:43 -0800 |
2 |
Donnie Berkholz <dberkholz@g.o> wrote: |
3 |
> On 17:21 Thu 12 Feb , Ciaran McCreesh wrote: |
4 |
> > The problem is, without GLEP 55, EAPI isn't known before the ebuild |
5 |
> > is sourced to generate metadata. If someone uses += anywhere that |
6 |
> > older bash looks when sourcing for metadata generation (which is |
7 |
> > not just global scope) |
8 |
> |
9 |
> Where else does it look? |
10 |
|
11 |
It has to be able to parse the file, which means it can get confused by |
12 |
things that appear to older versions to be mismatched brackets, even if |
13 |
they're hidden deep in some function. |
14 |
|
15 |
> > the package manager won't know that the EAPI says that bash-3.1 is |
16 |
> > required for sourcing until after it's already done the sourcing, |
17 |
> > by which point it's too late. |
18 |
> |
19 |
> OK. What could we do about this? GLEP 55 was one suggestion. |
20 |
> |
21 |
> I'm seeing a lot of people shooting down suggestions, and not many |
22 |
> people presenting good, workable solutions. Could we get some more of |
23 |
> those on the table? |
24 |
|
25 |
GLEP 55 *is* the good, workable solution. There still haven't been |
26 |
legitimate any technical objections to it. |
27 |
|
28 |
-- |
29 |
Ciaran McCreesh |