[synth coordinator — raised by the maintainer after the v0.69.0 tag]
The CHANGELOG is hand-written prose about a typed artifact set that rivet can diff. rivet diff emits exactly what a release note wants: what was added, what changed, and what the change did to the trace graph's own health. v0.69 shipped without running it.
Measured with the CI-pinned rivet 0.37.0, v0.68.0 -> v0.69.0:
9 added, 0 removed, 0 modified, 575 unchanged
0 new errors, 0 resolved errors, 22 NEW WARNINGS, 0 resolved
The nine additions are the release's nine artifacts with their titles — a derived table of contents no human wrote. The 22 warnings are the finding, in three classes, none surfaced by any v0.69 gate:
- All nine IDs cannot be commit-trailer references. "trailers need an uppercase-alphanumeric prefix and an all-digit suffix, e.g. REQ-001; rename it to trace this artifact through commits" — so no
RQ-NN artifact is traceable through commits at all.
req-type: process is not in the schema's allowed set ["functional","performance","interface","constraint","safety"] — 4 of v0.69's 9, and 53 occurrences repo-wide. A long-standing convention the schema never accepted.- All nine lack an incoming
verifies link, so the right side of the V is open for every artifact this release shipped.
Class 1 is the same root cause as #1334, found independently by a second tool. status_evidence's delivery window matches zero commits because RQ-69-SUBTRACT (#242): … doesn't fit its conventional-commit regex; rivet says the same IDs can't be trailer references. One naming decision, two traceability mechanisms silently disabled. They should be scoped together.
A version note that is itself a finding. My first run used the rivet on PATH — ~/.cargo/bin/rivet, 0.32.0 — which is exactly the PATH shadow #1236 pins and #1308 reports on the runners. The artifact delta is identical on both versions, but they disagree about validate: 0.32.0 says 0 broken cross-refs where 0.37.0 says cross-refs NOT CHECKED — externals failed to load. The older version's 0 was the vacuous reading; the 0.37.0 upgrade landed in v0.69 (#1324) and made it honest.
Scoped as RQ-70-RIVETNOTES in the v0.70 plan (#1336).
[synth coordinator — raised by the maintainer after the v0.69.0 tag]
The CHANGELOG is hand-written prose about a typed artifact set that rivet can diff.
rivet diffemits exactly what a release note wants: what was added, what changed, and what the change did to the trace graph's own health. v0.69 shipped without running it.Measured with the CI-pinned rivet 0.37.0,
v0.68.0 -> v0.69.0:The nine additions are the release's nine artifacts with their titles — a derived table of contents no human wrote. The 22 warnings are the finding, in three classes, none surfaced by any v0.69 gate:
RQ-NNartifact is traceable through commits at all.req-type: processis not in the schema's allowed set["functional","performance","interface","constraint","safety"]— 4 of v0.69's 9, and 53 occurrences repo-wide. A long-standing convention the schema never accepted.verifieslink, so the right side of the V is open for every artifact this release shipped.Class 1 is the same root cause as #1334, found independently by a second tool.
status_evidence's delivery window matches zero commits becauseRQ-69-SUBTRACT (#242): …doesn't fit its conventional-commit regex; rivet says the same IDs can't be trailer references. One naming decision, two traceability mechanisms silently disabled. They should be scoped together.A version note that is itself a finding. My first run used the
riveton PATH —~/.cargo/bin/rivet, 0.32.0 — which is exactly the PATH shadow #1236 pins and #1308 reports on the runners. The artifact delta is identical on both versions, but they disagree aboutvalidate: 0.32.0 says0 broken cross-refswhere 0.37.0 sayscross-refs NOT CHECKED — externals failed to load. The older version's0was the vacuous reading; the 0.37.0 upgrade landed in v0.69 (#1324) and made it honest.Scoped as
RQ-70-RIVETNOTESin the v0.70 plan (#1336).