1 |
On Fri, Feb 4, 2022 at 11:37 PM Ulrich Mueller <ulm@g.o> wrote: |
2 |
> |
3 |
> >>>>> On Mon, 31 Jan 2022, Marek Szuba wrote: |
4 |
> |
5 |
> > The next Gentoo Council meeting will take place on 2022-02-13 |
6 |
> > (Sunday), starting 19:00 UTC. |
7 |
> |
8 |
> > Please reply to this e-mail with proposed agenda items. If you |
9 |
> > suggested anything between the last council meeting and now, you are |
10 |
> > encouraged to raise the matter again in this thread and include |
11 |
> > link(s) to your earlier communication. |
12 |
> |
13 |
> As I had announced in the January meeting, I'd ask the Council to |
14 |
> pre-approve the list of features for EAPI 9, as listed here: |
15 |
> |
16 |
> https://wiki.gentoo.org/wiki/Future_EAPI/EAPI_9_tentative_features |
17 |
> |
18 |
> - Eclass revisions [1] |
19 |
> - EAPI of profiles defaults to repository EAPI [2] |
20 |
> - Allow comments in profile parent files [3] |
21 |
> - econf: Ensure proper end of string in configure --help output [4] |
22 |
> |
23 |
> This will be an EAPI with few new features, and its motivation is mainly |
24 |
> to have eclass revisions. The second and third feature will affect only |
25 |
> profiles, and the fourth feature is effectively a bug fix. |
26 |
> |
27 |
> There were also ideas about no longer exporting A (or alternatively, any |
28 |
> variables) to the ebuild environment [5], but nobody has come forward |
29 |
> with a concrete proposal yet. So, IMHO it will need more discussion and |
30 |
> won't be ready for EAPI 9; developers who have asked for the feature may |
31 |
> consider leading that discussion. IIUC, these would be the TeX and Go |
32 |
> maintainers. |
33 |
|
34 |
I think the issue with packages with large SRC_URI has very little to |
35 |
do with ${A} (which I think has a straightforward fix) and more to do |
36 |
with the 2nd and 3rd order effects. |
37 |
|
38 |
Large manifests (because we have to checksum all the files and verify them.) |
39 |
Very long computations (because we have to verify all the files.) |
40 |
More mirror load (because these packages often pin on git commit, |
41 |
instead of on released versions, or simply release very often, meaning |
42 |
we are required to mirror 20-30 versions of a package instead of a |
43 |
lower number.) |
44 |
|
45 |
Solving A is easy, solving the other problems seems to involve tougher |
46 |
tradeoffs. |
47 |
|
48 |
-A |
49 |
|
50 |
> |
51 |
> Ulrich |
52 |
> |
53 |
> [1] https://bugs.gentoo.org/806592 |
54 |
> [2] https://bugs.gentoo.org/806181 |
55 |
> [3] https://bugs.gentoo.org/470094 |
56 |
> [4] https://bugs.gentoo.org/815169 |
57 |
> [5] https://bugs.gentoo.org/721088 |