Existing contributors getting welcome messages

[nrc]

e.g. rust-lang/rust#17508 got a comment but the author has previously submitted rust-lang/rust#17286

cc @huonw

My guess would be a mismatch between the Git and GitHub user info, but who knows?

[jdm]

Yeah, it goes by email addresses and https://api.github.com/users/vberger doesn't have one specified, while the PR does use one.

[huonw]

More data:

• Remove unnecessary code from an example rust#17510
• Use cargo run in more places rust#16752

(https://api.github.com/users/MatejLach vs. https://github.com/rust-lang/rust/pull/17510.patch)

I have a feeling the current implementation isn't quite enough for the high traffic Rust repo, at least compared to Servo, e.g. the contributors API endpoint is only returning the top 100 (despite the per_page=400, which isn't quite enough anyway, we currently have 614!), meaning new contributors will get welcome messages until they've landed ~20 commits. One possible way to resolve this would be to store the emails from the git history; but statefulness is sad. :(

[jdm]

Actually I was wrong; it's not email but github-login based, but it's definitely the fact that it's only pulling the top 100 contributors that's messing us up here.

[nrc]

I hope I fixed this by handling pagination properly. I guess there might still be issues with username mismatch. Lets see if that is a problem...

[nrc]

Hopefully fixed by fcffc5d I'll leave this issue open for a while and close it if we get no new instances of this.

[nrc]

Haven't seen any false positives recently, so closing.