Michael Chirico

I think the default should be failure. false positives should be reported and/or `# nolint`'d

lintr itself has been bitten a few times recently from the lack of failure. it's easy to miss the warnings in a review, leading to extra review toil (1) handling...

Thanks Gabor. To clarify I don't think there's any compliance issue going on. This is to satisfy our one version policy: https://opensource.google/documentation/reference/thirdparty/oneversion Ideally we can remove any duplicated code from...

thanks Gabor, that's plenty helpful!! I think the rest is for us to sort out internally

Hey @gaborcsardi, maybe a LICENSE.note would be appropriate? Since utf8 is Apache-licensed. That would be consistent e.g. with what's done in these r-lib repos: - https://github.com/r-lib/later/blob/main/LICENSE.note - https://github.com/r-lib/rlang/blob/main/LICENSE.note - https://github.com/r-lib/vctrs/blob/main/LICENSE.note...

Great! feel free to close if you think this is fully subsumed by that issue (it may or may not be)

Is 974243881c3aefd033659ebeb387d7736b781fe1 a commit I should be able to cherry pick in isolation to see if this fixes our problem? Or should I wait for further updates?

> How are you loading the packages? it's complicated -- our build system is a strange beast. `load_all()` won't be involved though it _is_ possible it's entirely our fault for...

a workaround would also be fine -- what's the right way to use `@inheritParams` on an S4 generic? any examples you could point to?

thanks, I'll investigate more and post any findings.