You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Commit 33fb393
Browse filesBrowse the repository at this point in the historyBrowse files
core/task: describe the cancellation fence as what it is today
The fence comments asserted an effect no production reader produces yet: that the
counter separates a cancelled run from the run that resumes afterwards, and that a
write still parked on its path chain publishes after disposal. At this head
cancellationGeneration has three production sites (the field and the two increments
in requestAbort and requestDispose) and no production reader; nine assertions in
src/core/task/__tests__/Task.spec.ts 4786-4851 are what pin it. The three blocks now
say the counter is monotonic, that a future guarded write would compare the value it
captured when queued with the value at publish time, and that no such write exists
today.
Comment-only: the non-comment content of src/core/task/Task.ts is byte-identical
(sha256/16 60377E70665E7EE0 over 5,530 non-comment lines before and after), so the
mutation gate prices 0 counted executable lines for this commit.
Answered on Zoo-Code-Org/Zoo-Code pull request 2396, review thread 4520707090.
0 commit comments