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 acc49b9
Browse filesBrowse the repository at this point in the historyBrowse files
test(utils): refuse a captured target that sits under another lock
The reject path of the captured-target check had no row: the handover test moves the alias
A to B and stays there, so the recomputed key always agreed with the moved lock and the check
never fired. Add the A to B to A case the review asks for: A is captured, B is captured as
the publish target while A is held, and the alias is back at A before the key is
recomputed, so the loop breaks on its first pass and only the captured-target check
stands between this write and the commit rename. It refuses, before the merge read
and before staging, both targets keep their bytes, and A is the only lock taken and
the lock released. At blob e45bdc1, the head before the check read the captured
target, the same row resolves instead of rejecting.
0 commit comments