1 |
On Mon, Mar 23, 2020 at 02:14:06PM -0500, William Hubbs wrote: |
2 |
> On Mon, Mar 23, 2020 at 08:03:47PM +0100, David Seifert wrote: |
3 |
> > On Mon, 2020-03-23 at 14:00 -0500, William Hubbs wrote: |
4 |
> > > On Mon, Mar 23, 2020 at 07:36:13PM +0100, David Seifert wrote: |
5 |
> > > > On Mon, 2020-03-23 at 13:23 -0500, William Hubbs wrote: |
6 |
> > > > > Hey all, |
7 |
> > > > > |
8 |
> > > > > it has been brought to my attention that there have been several |
9 |
> > > > > backward-incompatible changes made to the python eclasses lately. |
10 |
> > > > > |
11 |
> > > > > It is true that everything in ::gentoo has been fixed along with the |
12 |
> > > > > changes to the eclasses; however, when a change like this goes into a |
13 |
> > > > > widely used eclass it breaks overlays with little to no notice; |
14 |
> > > > > especially since we do not require developers to be subscribed to this |
15 |
> > > > > mailing list. |
16 |
> > > > > |
17 |
> > > > > I do agree that overlay authors are on their own to fix things, but we |
18 |
> > > > > need to |
19 |
> > > > > find a way to notify them when a breaking change is going into a widely |
20 |
> > > > > used eclass and give them time to adjust their ebuilds. |
21 |
> > > > > |
22 |
> > > > > If the old way of doing things cannot be supported |
23 |
> > > > > along side the new way the correct path forward is a new version of the |
24 |
> > > > > eclass then a lastrites on the old version. That would give overlay |
25 |
> > > > > authors time to switch to the new eclass. |
26 |
> > > > > |
27 |
> > > > > If the old and new way can be supported in the same code base, a |
28 |
> > > > > reasonable way forward is to allow both ways to exist while ::gentoo is |
29 |
> > > > > migrated to the new code path then do the equivalent of a lastrites for |
30 |
> > > > > the old code path so overlay authors can adjust their ebuilds. |
31 |
> > > > > |
32 |
> > > > > Thoughts? |
33 |
> > > > > |
34 |
> > > > > William |
35 |
> > > > > |
36 |
> > > > |
37 |
> > > > All of this was announced with a 3 month timeout, using the right channels. |
38 |
> > > > Have |
39 |
> > > > you checked all python-related eclasses changes submitted to the ML? In this |
40 |
> > > > case, eqawarn would not have been possible, because the change involved |
41 |
> > > > dereferencing a variable. |
42 |
> > > |
43 |
> > > This is the change that broke us today. |
44 |
> > > |
45 |
> > > https://archives.gentoo.org/gentoo-dev/message/aa45f4f86f9b865eb0fe7344d83a7258 |
46 |
> > > |
47 |
> > > Where is the three month timeout for it? |
48 |
> > > |
49 |
> > > Thanks, |
50 |
> > > |
51 |
> > > William |
52 |
> > > |
53 |
> > |
54 |
> > Oh I thought you meant the python-single-r1 change from a month ago. |
55 |
> |
56 |
> The other one that is being pointed out to me is this one: |
57 |
> |
58 |
> https://archives.gentoo.org/gentoo-dev/message/4bf9f0250115c57779f93817356df871 |
59 |
> |
60 |
> Is that the python-single-r1 change you were talking about? |
61 |
> |
62 |
> If it is, what do we consider the appropriate channels for api change |
63 |
> announcements to be? |
64 |
|
65 |
Sorry about sending too quickly on this message. |
66 |
|
67 |
We have mostly fixed this one by now. |
68 |
|
69 |
Thanks, |
70 |
|
71 |
William |