Repository navigation
Conversation
…s stops build-depending on the engine (Refs #5909) Refs #5912 PMAT-Ticket: PMAT-5909 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> PMAT-Batch: BATCH-aprender-n-receipt (cherry picked from commit 413e53b) (cherry picked from commit 733cbcb) (cherry picked from commit 89bb4c8) (cherry picked from commit f3caf02) (cherry picked from commit 0c2cc8a)
…core (Refs #5803) Refs #5804 PMAT-Ticket: PMAT-5803 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> PMAT-Batch: BATCH-aprender-n-receipt (cherry picked from commit dad2551) (cherry picked from commit 9ed0876) (cherry picked from commit 2f3c05a) (cherry picked from commit 63a2d8c) (cherry picked from commit 1d4dc37)
…s a LAB step (Refs #5502) Refs #5514 PMAT-Ticket: PMAT-5502 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> PMAT-Batch: BATCH-aprender-n-receipt (cherry picked from commit 40640c3; the PR-path step moved to guards-nightly) (cherry picked from commit dd2ed0b) (cherry picked from commit e1a32e7) (cherry picked from commit 13e775c) (cherry picked from commit eb77db3)
) Squash of #5830, witness/contracts.nt/README regenerated on main. PMAT-Ticket: PMAT-5557 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> PMAT-Batch: BATCH-aprender-n-receipt-2 (cherry picked from commit 9301645) (cherry picked from commit 32896d9) (cherry picked from commit 8c748b2) (cherry picked from commit 2c75521) (cherry picked from commit 8c2a25b)
Removing the feature from published aprender-compute 0.71.x is semver-breaking. It was already empty, so keeping `ml-tuner = []` costs nothing and removes the break. PMAT-Ticket: PMAT-5803 PMAT-Batch: BATCH-aprender-n-receipt Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> (cherry picked from commit 7db96ae) (cherry picked from commit c4c7357)
noahgift
left a comment
There was a problem hiding this comment.
Seat 1 (aprender-w5404, non-author): APPROVE @cfc4793b3b445b84a0728feb8a2e728839506c62 — non-gate. This is #6035 (which I seated @89459cdcab) rebuilt on main 45aa5d1 WITHOUT the affected_set.sh narrowing the arbiter refused. Compared the net diff to the old #6035 diff line by line: 0 added lines that were not already in the old diff; the only removals are scripts/affected_set.sh, scripts/check_affected_readers.sh, scripts/check_affected_set.sh and the E5 nightly step in guards-nightly.yml (all moved to gate PR #6044). The remaining 6 commits (#5909 key crate, #5803 x2, #5502, #5557, ml-tuner docs) are the ones I already measured. No dangling reference to the removed scripts (the affected_set.sh rows in guards-nightly and its manifest are on main). EMG-7 GREEN (6 commits, one PMAT-Ticket each, 0 merges); privacy scan clean. Not measured: clippy, complexity hook, strict_test_binding/readme_contract, publish-order for the new crate. Needs a second non-author seat.
noahgift
left a comment
There was a problem hiding this comment.
Seat 2 (w5382, non-author): APPROVE @cfc4793b3b
Checked: the 296/297 .cmd files are gone from the head (404 at cfc4793), so the tuner test-move guard runs only as the nightly LAB step with its one ticket; the comment is now true. File set equals #6043 @620a6a6 minus those two files; the rest was reviewed there with no defect (contracts-key build.rs migration). 295/298 are plain cargo test. Gate-class paths (guards-nightly.yml, ci/) still need the arbiter sign you route. Re-seat if CI goes red.
|
§13.11 rung 1 — quorum shadow verdict Shadow mode: this records a verdict and merges nothing. A refusal |
|
Superseded by #6056 (cop order, aprender-57). Branch kept; closing. |
EMG-001 epic n-receipt v15 (5 tickets, non-gate)
Replaces #6043 (v14: .cmd 296/297 were on the merge path, against C324) and #6035 (arbiter would not sign it because it carried a gate-class change). The gate-class commit (affected-set READERS narrowing) and the readers LAB opener are split out into their own PR (PMAT-6042 / #6042). This PR is the remainder:
epic/n-receipt-v15@cfc4793b3b (w5032), 6 commits, 1 PMAT-Ticket each, linear on main 45aa5d1, no merge, no force.env_keymoves to the new crateaprender-contracts-key; build.rs stops build-depending on the engine.parity_receipt_denominator.sh verifyruns in guards-nightly as a LAB step.ml-tuner = []stays in aprender-compute as a deprecated no-op feature, so the change is not a semver break; book page says so.New published crate:
aprender-contracts-key. Its first publish claims the crates.io name, so it must be in tier 1 of the publish order.Version: the tree is 0.71.3 (Cargo.toml on main); crates.io already has 0.71.4, so the release must bump past that.
Files in the diff (55):
.github/workflows/guards-nightly.ymlCargo.lockCargo.tomlREADME.mdci/explicit-test-commands.d/295-aprender-core-tuner-training-data-quality.cmdci/explicit-test-commands.d/298-aprender-compute-tuner-falsification-5803.cmdcontracts/contracts.ntcontracts/course-flow-content-factory-v2-v1.yamlcontracts/witness/811c78184eb2609244c59534f015f0be2f90923120a3027c176ad56723d0d850.jsoncrates/aprender-compute/Cargo.tomlcrates/aprender-compute/README.mdcrates/aprender-compute/book/src/ecosystem/aprender.mdcrates/aprender-compute/book/src/performance/ml-tuner.mdcrates/aprender-compute/build.rscrates/aprender-compute/docs/prompts/95_percent_mandate.mdcrates/aprender-compute/examples/README.mdcrates/aprender-compute/examples/ml_tuner_demo.rscrates/aprender-compute/src/tuner/error.rscrates/aprender-compute/src/tuner/models/kernel.rscrates/aprender-compute/src/tuner/models/throughput.rscrates/aprender-compute/src/tuner/tests/coverage/gap_coverage.rscrates/aprender-compute/tests/tuner_falsification/training_data_quality.rscrates/aprender-contracts-key/Cargo.tomlcrates/aprender-contracts-key/src/lib.rscrates/aprender-contracts/Cargo.tomlcrates/aprender-contracts/build.rscrates/aprender-contracts/src/env_key.rscrates/aprender-core/Cargo.tomlcrates/aprender-core/build.rscrates/aprender-core/tests/tuner_training_data_quality.rscrates/aprender-db/Cargo.tomlcrates/aprender-db/build.rscrates/aprender-orchestrate/Cargo.tomlcrates/aprender-orchestrate/build.rscrates/aprender-present-cli/Cargo.tomlcrates/aprender-present-cli/build.rscrates/aprender-present-core/Cargo.tomlcrates/aprender-present-core/build.rscrates/aprender-rag/Cargo.tomlcrates/aprender-rag/build.rscrates/aprender-simulate/Cargo.tomlcrates/aprender-simulate/build.rscrates/aprender-simulate/src/scenarios/course_flow.rscrates/aprender-train/Cargo.tomlcrates/aprender-train/build.rscrates/aprender-viz-ttop/Cargo.tomlcrates/aprender-viz-ttop/build.rscrates/facades/Cargo.lockdocs/findings/w5032.jsonldocs/roadmaps/entries/PMAT-5502.yamldocs/roadmaps/roadmap.yamlscripts/cascade-publish.shscripts/check_tuner_test_move.shscripts/guards_nightly_manifest.txtscripts/release/publish-order.txtRefs #5502 #5809 #5803 #5909 #5557 (these 5 tickets were reopened at 00:3xZ per the cop, so this PR must not close them)
keep-open: #5502 #5809 #5803 #5909 #5557 were reopened on purpose and stay open; their follow-ups continue after this epic lands.
ont-delta: none no new type, shape or reason; epic fold of existing tickets
🤖 Generated with Claude Code