Proper attribute propagation in presence of objects#12344
Proper attribute propagation in presence of objects#12344lthls wants to merge 3 commits intoocaml:trunkfrom
Conversation
1162e99 to
9008a00
Compare
|
The test failures: |
|
The "unrelated CI failure" is for test tests/weak-ephe-final/'finaliser_handover.ml' on a |
|
Some comments to help reviewers: What I am trying to do in this PR is to ensure that the extra let bindings introduced by My first attempt was to make sure that all problematic calls were already under another I've got a slightly better solution for modules, where I only insert calls to To sum up, with this PR I've taken the view that it's fine to make more classes use the cache mechanism than strictly needed. If that's fine, we could keep the current version, or push this even further and wrap the whole program (that would simplify the code a lot). If that's not fine, I'll need to know it so that I can work on a proper solution (which is not going to be pretty). @garrigue I suspect that you're the only person which may still know (or care) about that. Do you have an opinion on this issue ? |
Fixes #12232.