Dev - #977
Merged
Merged
Conversation
chore: back-merge v2.0.0 into dev
The canonical published-schema URI root was `https://raes.dev/schemas/`, carried over from the identity cutover in #921. That domain is registered to a third party, as was the root it replaced. Using an uncontrolled domain in `$id` gives up the only property the URI form buys — that the identifier is unambiguously ours to assign — and left issue #908 blocked on provisioning a domain the project cannot obtain. Repoint every published `$id`, the semantic-invariant profile URI, and the concept authority reference at `https://raesystem.github.io/rae/`, which is bound to the repository's own GitHub organisation. `$id` is an identifier and is not required to resolve, so this carries no DNS or hosting obligation; the path matches the org's GitHub Pages layout if the schemas are ever served. Refresh the schema-publication content hashes for the 84 affected schemas and record the correction as an amendment to ADR-096. Also re-pin the CHANGELOG.md identity digest, which went stale on dev when release-please appended the 2.0.0 notes. The appended content introduces no retired identity occurrences; the count is unchanged at 297. BREAKING CHANGE: published schema `$id` values move from `https://raes.dev/schemas/` to `https://raesystem.github.io/rae/schemas/`. Consumers pinning or caching schema identifiers must update them. Refs #908
`check_identity_cutover` pinned CHANGELOG.md by whole-file digest, but release-please rewrites that file on every release, inserting each new section above the existing ones. The pin therefore goes stale at every release by construction. It did at v2.0.0: PR #937 back-merged the release into `dev` with the required `verify` check failing, and the failure was overridden rather than fixed, leaving `dev` red on this gate. Add a `generated-release-history` record class that matches how the file is actually maintained. It pins the classified tail exactly and holds everything written above it to the live-tree rule of zero retired identity occurrences, so newly generated content carries no historical exemption. Editing or truncating the tail still fails closed. This is strictly stricter than the whole-file pin it replaces: the old rule allowed any content anywhere in the file so long as the digest was re-recorded, which is exactly the re-record-and-move-on habit that let the override stand. Manifest schema moves to historical-identity-records/v3. Refs #908
The identity gate holds the generated head of CHANGELOG.md to zero retired identity occurrences, but nothing checked the text that lands there. A feature PR into `dev` is squash-merged, so its title becomes the commit release-please writes into the changelog, and any BREAKING CHANGE footer becomes a changelog entry. A retired name in either surfaced days later as a red release PR, far from the change that caused it. Check both at the point they are authored, in the existing PR title guard that already runs on PRs targeting `dev` from the trusted base ref. Body prose is deliberately not checked. Explaining the retired name is legitimate — this PR's own description does it — and only the title and breaking-change footers reach the changelog. The pattern is duplicated in text form rather than imported, because check_pr_title is stdlib-only by design and check_identity_cutover pulls in PyYAML. A test asserts the two patterns agree on the same samples so they cannot drift. Refs #908
The previous commit checked the PR body for BREAKING CHANGE footers. That is the wrong text. This repository squashes with squash_merge_commit_title=PR_TITLE and squash_merge_commit_message=COMMIT_MESSAGES, so the squash commit body release-please parses is the concatenated commit messages; the PR body never reaches it. Supply the PR's commit messages to the guard from the workflow and check those. The messages are fetched with a read-only token and written to a file rather than passed as an argument, so untrusted PR data is never shell-interpolated. The PR body is still checked, so the rule holds if the squash setting changes. Refs #908
Two CI failures on this branch. `ruff format` rejected tests/test_runtime_contracts.py. The namespace rewrite lengthened a URI past the 120-column limit and I only linted the files I edited by hand, not the ones the rewrite touched. `title-guard` failed with "unrecognized arguments: --commit-messages-file". The workflow file comes from the PR head but the checker is deliberately taken from the base ref, so on the PR that introduces the flag the base copy cannot accept it yet. Probe for support before passing it, matching the bootstrap shape the workflow already uses for a missing checker. Commit-message footer enforcement becomes active from the next PR onward; title enforcement is unaffected now. Refs #908
fix!: repoint published schema namespace to a controlled URI root
The PyPI package metadata carried no trove classifiers, so the Python-versions badge in the README rendered as "missing". Shields reads Programming Language :: Python :: X.Y classifiers and does not fall back to Requires-Python, so declare the supported interpreters explicitly.
…fiers fix: publish Python version classifiers for the raes distribution
Resolves the open dependency alerts on the default branch by moving the locked versions past their patched releases, and raises the direct mcp floor so a fresh resolve cannot fall back to an affected version. Closes #941
fix: upgrade dependencies with known vulnerabilities
…idence test: add formal semantic evidence retest
feat: enforce participant information-flow policy
…on-policies feat: add experiment variation selection policies
feat: add executable behavioral validation probes
feat: migrate participant control compatibility
docs: document participant input and output control
…fication feat: add bounded determinism and replay-consistency verification (ASR-514)
…ation feat: add deterministic trial compilation
security: define participant-relative predicate opacity
fix: replace platform profiles with composable capabilities
9 tasks
fix: reduce behavioral relation validation complexity
9 tasks
fix: reduce claim binding validation complexity
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.
Summary
Related issues
Changes
Test plan
nox -s verifygate passes, or unchecked sessions are listed belowChecklist
CHANGELOG.mdfrom it)Notes for review