Library writes that took in part are partial, read back #41
Loading…
Reference in a new issue
No description provided.
Delete branch "vinny/library-partial"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Part of #23.
When CMO refuses some units but takes others, the tester's
missions.assign_unitsanswers withokfalse,orderedtrue, arefusedlist and anexpectfor what took. CMOnet recorded that as rejected, saying "no requested effect was accepted", and released the units' reservations while the assignment stood.What changes:
library/runtime.py:refusal(result)recognises a write that took in part. The read-back observation carries its refusal.RuntimePort._library_write: sends such a result on to its read-back, like any accepted write.partial. It has one evidenced completed effect, one rejected effect with CMO's refusal, andreconcile_before_retryadvice. When the read-back doesn't show it, the call isoutcome_unknown. Partial calls refresh the observation families like completed ones.Tests:
tests/library/test_library.pyassigns two units where CMO refuses one. The call comes outpartial, its read-back ran, and no reservation is left.Still to do: the tester tools that
error()after an earlier step took are to answer this same way, in a tester change and a later library sync.🤖 Generated with Claude Code