1 |
On 11/4/20 8:18 PM, Alec Warner wrote: |
2 |
> |
3 |
> I disagree. These packages are not installable by default, and must be |
4 |
> unmasked by users, so this tradeoff is one we expect them to make. Are |
5 |
> there practical problems that these packages pose to developers? You |
6 |
> listed a bunch of user problems, but again users are opting into these |
7 |
> problems, presumably. |
8 |
I just managed to install Gentoo yesterday, today I want package x and |
9 |
apparently it's masked? I type emerge --autounmask-write x because the |
10 |
guide told me so without understanding any of the concerns (that are |
11 |
also written in devmanual) listed above. I have no idea what I'm doing, |
12 |
but at least I got the package on stable system... if upstream wasn't so |
13 |
broken currently. |
14 |
|
15 |
Now imagine you're a developer and you need a certain library to develop |
16 |
your software, but the upstream repo has been broken for weeks or |
17 |
months. You couldn't emerge the package, but if there was a snapshot |
18 |
available to a latest working commit, it could save a lot of your time. |
19 |
|
20 |
> |
21 |
> But again, why are we making this a firm policy; as opposed to letting |
22 |
> the maintainer make their decision? |
23 |
Look where maintainer decision got us, majority of these ebuilds being |
24 |
broken, outdated and totally ignored. There aren't that many packages |
25 |
available right now, but the state could still be cleaner. |
26 |
|
27 |
Wouldn't you like the solution offered by mgorny, where there could be a |
28 |
time-limit for these -9999-only packages? |
29 |
|
30 |
> |
31 |
> |
32 |
> We can't guarantee any package to build..so I'm not sure how this is a |
33 |
> practical policy goal. |
34 |
> |
35 |
> -A |
36 |
Sure we can. You test it before you push, then it hits at least two |
37 |
tinderboxes currently. Before stabilization multiple test runs are made. |
38 |
I do trust the current state of Gentoo packages being better and better |
39 |
each day. |
40 |
|
41 |
-- juippis |