1 |
Ciaran McCreesh wrote: |
2 |
> On Mon, 24 Dec 2007 10:52:53 +0000 |
3 |
> Steve Long <slong@××××××××××××××××××.uk> wrote: |
4 |
>> Ciaran McCreesh wrote: |
5 |
>> > On Mon, 24 Dec 2007 06:03:12 +0000 |
6 |
>> > Steve Long <slong@××××××××××××××××××.uk> wrote: |
7 |
>> >> * Set the EAPI inside the ebuild in a way that makes it easy to |
8 |
>> >> fetch it This is ok as atm only EAPI=1 is in the tree, so there is |
9 |
>> >> no backward compatibility issue. |
10 |
>> > |
11 |
>> > It's both a backwards and a forwards compatibility issue. |
12 |
>> > |
13 |
>> Yeah, so forwards into the future where it's impossible to maintain |
14 |
>> this format (er..) there'll be another type of ebuild for your |
15 |
>> purported long-term solution. |
16 |
> |
17 |
> Hopefully that'll be EAPI 2, and hopefully we're talking months rather |
18 |
> than years. |
19 |
> |
20 |
Whatever. EAPI="2" works fine with current software. Could you tell me why |
21 |
you're so hot on export'ing EAPI? I thought it was only relevant, |
22 |
environment-wise, to the ebuild and subshells. What is the use case for |
23 |
exporting it externally? |
24 |
|
25 |
> * Eclasses modifying EAPI. Doesn't scale across multiple eclasses, nor |
26 |
> across EAPIs removing features, nor EAPIs changing eclasses. |
27 |
> |
28 |
I don't see what's wrong with EAPI (if set, otherwise implicitly whatever |
29 |
the ebuild sets, or 0 if not set there) only applying to the file it's |
30 |
declared in. |
31 |
|
32 |
Since we can't remove eclasses, it would be useful to continue allowing them |
33 |
to examine the EAPI variable for future compatibility. I'm sure there are |
34 |
other restrictions imposed by this "enhancement" which make maintenance |
35 |
more difficult for no benefit to the maintainer. |
36 |
|
37 |
|
38 |
-- |
39 |
gentoo-dev@g.o mailing list |