1 |
I have no strong opinion about this. |
2 |
|
3 |
> 2) EAPI in file extension |
4 |
> - Allows changing global scope and the internal format of the ebuild |
5 |
> a) .ebuild-<eapi> |
6 |
> - ignored by current Portage |
7 |
|
8 |
simple, straightforward but ugly |
9 |
|
10 |
> b) .<eapi>.ebuild |
11 |
> - current Portage does not work with this |
12 |
|
13 |
this only has disadvantages in front of a) |
14 |
|
15 |
> c) .<eapi>.<new extension> |
16 |
> - ignored by current Portage |
17 |
|
18 |
same as a) but maybe less ugly |
19 |
|
20 |
> 3) EAPI in locked down place in the ebuild |
21 |
> - Allows changing global scope |
22 |
> - EAPI can't be changed in an existing ebuild so the PM can trust |
23 |
> the value in the cache |
24 |
|
25 |
This doesn't need a locked down place, just some strict way of setting |
26 |
it, like at most one EAPI="constant_string_without_space" so that grep |
27 |
& friends can catch it. |
28 |
|
29 |
> - Does not allow changing versioning rules unless version becomes a |
30 |
> normal metadata variable |
31 |
|
32 |
I'm not entirely sure about this and would need some real |
33 |
example/argument in order to be convinced |
34 |
|
35 |
> * Needs more accesses to cache as now you don't have to load older |
36 |
> versions if the latest is not masked |
37 |
|
38 |
true but this "more" would need to be measured and opposed to the |
39 |
number of metadata loads in a dep resolution procedure. |
40 |
|
41 |
> a) <new extension> |
42 |
|
43 |
doesn't have the one year wait drawback and doesn't mess with pm |
44 |
implementation details (eapi) with package names that shall represent |
45 |
upstream ones. |
46 |
|
47 |
> b) new subdirectory like ebuilds/ |
48 |
> - we could drop extension all together so don't have to argue about |
49 |
> it any more |
50 |
> - more directory reads to get the list of ebuilds in a repository |
51 |
|
52 |
I see no advantage there vs 3.a) |
53 |
|
54 |
> c) .ebuild in current directory |
55 |
> - needs one year wait |
56 |
|
57 |
That could have been done one year ago and would be done now; I think |
58 |
it's still fine now because I don't expect major behavior changes in |
59 |
the ebuild format within one year. |
60 |
|
61 |
To summarize, I would retain 3.a as best solution, having the "usable |
62 |
right now" advantage of current glep55 and keeping the ebuild's file |
63 |
names (more or less) consistent with upstream ones. |
64 |
|
65 |
Alexis. |