Skip to content

Conversation

@RalfJung
Copy link
Member

This is the stdsimd side of rust-lang/rust#58129.

(Having stdsimd in a separate repo means that what would usually be 1 PR turns into 3...)

@gnzlbg
Copy link
Contributor

gnzlbg commented Feb 11, 2019

@RalfJung wait a sec, how did that PR get merged without updating stdsimd ? stdsimd should have failed to build as part of libcore because of this -.-

@gnzlbg
Copy link
Contributor

gnzlbg commented Feb 11, 2019

FWIW, there is a possible 1 PR approach. First you update stdsimd upstream to point to your own fork, then you fix the fork as part of your PR, and then you merge that.

Then you send a PR here, which we merge, and we take care of updating stdsimd back to this repo master for you :)

@RalfJung
Copy link
Member Author

I left a libcore-private copy of fn into_inner which stdsimd uses (and which caused a bors failure because on non-x86 platforms that function is unused :/ ).

First you update stdsimd upstream to point to your own fork

You mean as in, change the rustc branch it tests again?
That'll be "fun" when two people want to do it at the same time. ;)

@RalfJung RalfJung closed this Feb 11, 2019
@RalfJung RalfJung reopened this Feb 11, 2019
@gnzlbg
Copy link
Contributor

gnzlbg commented Feb 11, 2019

You mean as in, change the rustc branch it tests again?
That'll be "fun" when two people want to do it at the same time. ;)

If you ping me or alex we can prevent two people from doing it at the same time.

There is an alternative 2 PR method, which has the same problem if two people do it at the same time (depending on the changes). You could update stdsimd in your PR to stdsimd/master with your fix, if you send a PR here before that breaks stdsimd master temporarily. If your PR is reviewed and ready to merge, we could also do just that.

The safest method is the 3 PR approach, but if you ping us we can tell you if there are others trying to do the same right now or not :P

@gnzlbg
Copy link
Contributor

gnzlbg commented Feb 11, 2019

FYI @RalfJung windows is broken, so as long as travis passes everything should be ok.

@RalfJung
Copy link
Member Author

The Android and macOS build failure seem to be just timeouts?

@gnzlbg
Copy link
Contributor

gnzlbg commented Feb 11, 2019

I've restarted them.

@alexcrichton alexcrichton merged commit c546604 into rust-lang:master Feb 11, 2019
@RalfJung RalfJung deleted the into-initialized branch April 14, 2019 08:06
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants