docs(designs): correct T5 — the owed surface and the offline premise (RIG-3532) - #1227
Open
rigel-mintaka wants to merge 1 commit into
Open
rigel-mintaka wants to merge 1 commit into
rigel-mintaka wants to merge 1 commit into
Conversation
rigel-mintaka
added this pull request to stack #1228
September 14, 2026 02:16
|
Compass engineering docs preview: https://compass-comms-rig3532-amend.compass-eng-docs.pages.dev Deployed from Changed pages: |
…(RIG-3532) The frozen coverage record (RIG-3473) prescribes a T5 mechanism the code falsifies on two independent axes. Both were verified against the tree before T5 was implemented (#1225); this amends the record so the next reader is not misled, and records WHY the shipped test differs rather than rewriting the task as if it had always said this. 1. `CountOwedMentions`/`ClearOwedMention` are the wrong durable surface: a home-channel post never writes an `owed_mentions` row, because `RecordOwedMention` requires an @-mention AND an out-of-sweep-set agent, and `InSweepSet` is satisfied by the home-channel disjunct. A test asserting on that count is green while proving nothing. The surface the sweep actually consumes is `UndeliveredMessages`. 2. Posting to a provisioned-but-not-started agent cannot leave a message owed: the post wakes the agent and fresh-starts it against the placement `Provision` recorded, and the waker is wired in the assembly the e2e fixture runs. Offline has to be a despawn boundary. The durable-not-live-tail requirement the record states is unaffected and is why the shipped test reads the store. Co-authored-by: Matt Wilkinson <matt@rigel.build>
rigel-mintaka
force-pushed
the
compass-comms/rig3532-amend-record
branch
from
September 14, 2026 02:45
6dcb807 to
570014c
Compare
rigel-mintaka
marked this pull request as ready for review
September 14, 2026 02:46
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This PR is part of a stack containing 2 PRs:
mainThe frozen coverage record (RIG-3473) prescribes a T5 mechanism the code
falsifies on two independent axes. Both were verified against the tree
before T5 was implemented (#1225); this amends the record so the next
reader is not misled, and records WHY the shipped test differs rather than
rewriting the task as if it had always said this.
CountOwedMentions/ClearOwedMentionare the wrong durable surface: ahome-channel post never writes an
owed_mentionsrow, becauseRecordOwedMentionrequires an @-mention AND an out-of-sweep-set agent,and
InSweepSetis satisfied by the home-channel disjunct. A testasserting on that count is green while proving nothing. The surface the
sweep actually consumes is
UndeliveredMessages.owed: the post wakes the agent and fresh-starts it against the placement
Provisionrecorded, and the waker is wired in the assembly the e2efixture runs. Offline has to be a despawn boundary.
The durable-not-live-tail requirement the record states is unaffected and
is why the shipped test reads the store.
Co-authored-by: Matt Wilkinson matt@rigel.build