Do not warn about missing change ID in tarball builds #145780
Conversation
|
|
rustbot
added
S-waiting-on-review
|
👍 @bors r=lqd rollup |
bors
added
S-waiting-on-bors
Do not warn about missing change ID in tarball builds Fixes: rust-lang#145778 r? `@jieyouxu`
Rollup of 7 pull requests Successful merges: - #144452 (std/sys/fd: Relax `READ_LIMIT` on Darwin) - #145307 (Fix `LazyLock` poison panic message) - #145515 (Optimize `char::encode_utf8`) - #145540 (interpret/allocation: get_range on ProvenanceMap) - #145774 (Remove default opts from config) - #145780 (Do not warn about missing change ID in tarball builds) - #145781 (Remove profile section from Clippy) r? `@ghost` `@rustbot` modify labels: rollup
| @@ -198,14 +198,16 @@ fn check_version(config: &Config) -> Option<String> { | |||
| Some(ChangeId::Id(id)) if id == latest_change_id => return None, | |||
| Some(ChangeId::Ignore) => return None, | |||
| Some(ChangeId::Id(id)) => id, | |||
| None => { | |||
| // The warning is only useful for development, not release tarballs | |||
There was a problem hiding this comment.
That's a pretty strong assertion, which I'm not sure distros agree with, at least Archlinux uses the change-id in their build, and they do update it for every new version.
Same for debian, config.toml.in.
Furthermore users can just put ignore if they don't want it.
|
Hm, I think I agree with Urgau's assessment re. this is used by distros. Maybe we should just print a warning or document this somewhere if someone wants to disable change-id for dist? @bors r- (unsettled design question) |
bors
added
S-waiting-on-author
|
As I wrote here, I also think that distros do this, but I don't think it's common enough that we have to warn them if it's unset. So I would keep the implementation, and just update the source code comment. WDYT? |
|
I don't feel too strongly about this, but I have a weak preference towards being more "upfront". I.e. for users trying to produce distributions, seeing the warning is arguably a good thing -- I think that if they don't want to care about |
|
As in, my first reaction seeing the report in #145778 is apart from the fatal not a git repo error, seeing the |
|
I think that a "clean build" from sources should ideally produce as few warnings as possible. That being said, I also see your point. I'll leave it up to you. If you don't find this worthy, then let's close both this PR and the original issue. |
|
I have a weak preference towards "not-a-bug", though we can probably improve the docs around it. |
I agree in the general sense, but |
|
Ok, fair enough! |
rustbot
removed
the
S-waiting-on-author
Fixes: #145778
r? @jieyouxu