docs(obs): mark the gateway prerequisite satisfied and T2 superseded (RIG-2867) - #1273
Open
rigel-mintaka wants to merge 5 commits into
Open
rigel-mintaka wants to merge 5 commits into
rigel-mintaka wants to merge 5 commits into
Conversation
…(RIG-2867) Slice T6 of the frozen Compass Server LLM Gateway record: the #656 observability-architecture record's OMP-gateway prerequisite is satisfied, and its T2 usage RPCs are superseded by that record's UsageService. Both edits are additive, per the design-freeze rule — the frozen prose is untouched and the PREREQUISITE checkbox stays unticked, since the box tracks a task rather than a dependency. The annotation cites the gateway record under docs/designs/server/, not the docs/designs/platform/ path the tracker issue named; it moved in the docs reorg, and a stale path would send readers nowhere. Co-authored-by: Matt Wilkinson <matt@rigel.build>
|
Merging to
After your PR is submitted to the merge queue, this comment will be automatically updated with its status. If the PR fails, failure details will also be posted here |
|
Compass engineering docs preview: https://compass-obs-rig-2867-t6-hand.compass-eng-docs.pages.dev Deployed from Changed pages: |
…G-2867)
The first pass annotated the PREREQUISITE entry and the T2 prose section, but
the record asserts the gateway is undesigned in three further places, so it
contradicted itself: the Source paragraph ("NO design record yet"), the Plan
prerequisite, and the Tasks preamble ("the undesigned OMP-gateway-into-Server
prerequisite").
The load-bearing one is the Tasks checklist. An agent executing Track A reads
the task entries, not the prose sections, so it would have reached T2's
instruction to add compass.v1 usage RPCs and never seen the supersession note
living 100 lines up. T2's entry now carries it inline.
Still additive: 12 inserted lines, no frozen sentence touched, and the
PREREQUISITE checkbox stays unticked since the box tracks a task here rather
than a dependency met elsewhere. Also wrapped the first pass's two annotations,
which ran to 153 and 113 columns against the file's 65-72.
Co-authored-by: Matt Wilkinson <matt@rigel.build>
Review caught that the previous pass was still incomplete in two ways. First, the Track A list heading — "Track A — blocked on the OMP-gateway prerequisite:" — sits directly above the checklist and was un-annotated, with the note 14 lines up, past the whole Track B list. The note now sits under the heading it scopes, and the Tasks preamble keeps one of its own. Second, the frozen T6 spec names three dependency facts and this slice carried two. The missing pair mattered more than the count suggests: annotating T2 as superseded while T3 still read "Consumes: T2's generated RPCs" pointed an executing agent at a surface the record now says will not exist. T1's Consumes now names TokenUsageEvent as the event shape, and T3's names the UsageService RPCs it reads instead of T2's. Still additive: 18 inserted lines, no frozen sentence touched. Co-authored-by: Matt Wilkinson <matt@rigel.build>
The T1/T2/T3 interface annotations read as rival bullets next to the ones they qualify — T3 in particular had two Consumes lines, the original naming T2's RPCs and mine naming UsageService, with nothing telling a reader which wins. Each now opens with Correction or Since, so it reads as qualifying the bullet above rather than competing with it. T1's is Since, not Correction: its prerequisite line was never wrong, only unmet at the time of writing. T2 keeps the exact phrase the frozen T6 spec asks for, "superseded by RIG-1715 UsageService", verified still present. Co-authored-by: Matt Wilkinson <matt@rigel.build>
…(RIG-2867) The previous commit's own annotation over-claimed. It said T1 and T3 are both executable, but T3 reads UsageService.GetUsageSeries and GetProviderQuota, which the gateway record's T4 has not shipped — that task is still unticked. It matters because this record defines executable narrowly, as depending only on shipped code, so the claim was false against its own vocabulary. T1 is executable now; T3 also waits on that T4. Also finished what the previous commit claimed to finish. T3's Produces still rendered "T2's time-series JSON" one line below the Consumes fix — the same dangling pointer, same four-line block. T3's correction now covers both. T1's Produces mentions "the T2 read RPCs" and is deliberately left alone: it names the read layer to claim store-swap invariance, not to send anyone to build a surface, so it does not dangle. Co-authored-by: Matt Wilkinson <matt@rigel.build>
rigel-mintaka
marked this pull request as ready for review
September 17, 2026 03:58
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.
Slice T6 of the frozen Compass Server LLM Gateway record: the #656
observability-architecture record's OMP-gateway prerequisite is satisfied, and
its T2 usage RPCs are superseded by that record's UsageService.
Both edits are additive, per the design-freeze rule — the frozen prose is
untouched and the PREREQUISITE checkbox stays unticked, since the box tracks a
task rather than a dependency.
The annotation cites the gateway record under docs/designs/server/, not the
docs/designs/platform/ path the tracker issue named; it moved in the docs
reorg, and a stale path would send readers nowhere.
Co-authored-by: Matt Wilkinson matt@rigel.build