Gabriel Scherer
Gabriel Scherer
(Instead of having `coop_` functions in platform, I think it would be rather natural to have them in sync as a C counterpart of `caml_ml_mutex` which has almost the same...
I pushed a new iteration that takes your remarks into account, and also moves the non-blocking code to sync.h. What do you think?
Done in #13716.
When this PR was submitted I wrote an email to @NickBarnes to ask whether he would be willing to review it. I haven't received a response for now, so I...
I rebased the PR and am planning to merge it if the CI passes.
I observe a test failure of the arm64 ci configuration on the test `weak-ephe-final/weak_array_par.ml`: ``` Fatal error: exception File "weak_array_par.ml", line 16, characters 17-23: Assertion failed ``` The failure happens...
Good point! But the assertion failure I observed is a correctness bug in this test, not just a (possibly-benign) race. This suggests that the TSan-reported race discussed in #14061 is...
Indeed, I agree with the reasoning. (The temporary nature of CI outputs is why I post some details in the PR when observing a failure, at the risk of inconveniencing...
@dbuenzli I use your `git describe` trick in 63184270c5251ea4035f0d5c7c46a6a231ad7261 , now also part of this PR. I still kept the VERSION file because that allows building from the source repository...
I added a `make check-release` target that checks that `VERSION` and `git describe` coincide, and that `NEXT_RELEASE` occurrences are gone from source files.