Repository navigation
One file per standard and its debt; the train runs the browser acceptances once per batch; QR-03 takes the middle of three - #174
Merged
Conversation
The standards register, its debt and the parts debt are read as parts by tools/specs/record (Python and JavaScript), which now reads a map by path and a marker at any depth. A generator step's "each" may name files. Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013Asa3LuNYtPDBa4FbSEQUZ
Contributor
|
Merge train notice: the merge train merges this pull request once it is green on current main; don't enable auto-merge. Generated by Claude Code |
…hat board together One by one only when the batch fails, to find which it refuses. A train of several pull requests took over an hour before anything merged. Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013Asa3LuNYtPDBa4FbSEQUZ
The slowest of three failed pull requests at random on shared runners (274 against 250 on #160, main green on the same base). Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013Asa3LuNYtPDBa4FbSEQUZ
…-jbqzew The parts debt stays in app/specs/parts/spec.json: PT-01 is now a LexLean program that reads it there (#142). Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013Asa3LuNYtPDBa4FbSEQUZ
Contributor
|
Merge train: 16389a0: not merged, because it conflicts with main (c26b29a) and the pull requests the train takes before it. Fix it on this branch (merge main first) and the train tests it again. Generated by Claude Code |
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013Asa3LuNYtPDBa4FbSEQUZ
This was referenced Nov 16, 2026
Contributor
|
Live: this change is on the published page, as main The publish job opened the published page in a fresh Chromium and its live checks passed (PG-01, PD-01, PD-02): https://github.com/afflom/me/actions/runs/43919204957 |
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
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.
The train runs the browser acceptances once per batch
Since #164 the train runs the fast gates and the browser acceptances for each ready pull request in turn. The first run that took pull requests (run 43817755958) passed on a runner and took 83 minutes for two pull requests (#142 in 43 minutes, #169 in 39), and nothing merged in that time.
The train now merges every ready pull request that passes the fast gates onto main in order, then runs the browser acceptances once on that tree. If they pass, all of them are accepted. Only if they fail does it go back to main and take them one by one, fast gates and acceptances each, to find which it refuses. What is tested is still exactly the tree that merges.
The change is in the model,
app/specs/pipeline/workflows/merge-train.yml.json. I ran the new loop in a scratch repository with a stubbed gate: three pull requests board, the batch fails, the one-by-one pass accepts two and refuses the one that fails, and a conflicting one is refused once.QR-03 takes the middle of three cold starts
#164 made the labelling time the slowest of three cold starts. On shared runners that failed pull requests at random (#160: 274 against 250, with main green on the same base). The measure is now the middle of the three, so neither one lucky sample nor one slow moment of a runner decides it. The threshold is unchanged.
One file per standard and per standard's debt
Every pull request that registers a standard or changes a standard's debt edited the same two lists:
standardsinapp/specs/standards/spec.jsonandapp/specs/standards/debt.json. Two such pull requests conflict with each other. This is the class #164 closed for guests.app/specs/standards/register/<key>.json(70 files).spec.jsonkeeps the comment,ungovernedandunassigned, and says"standards": {"parts": "register"}.app/specs/standards/debt/<key>.json(24 files).debt.jsonis gone;spec.jsonsays"debt": {"parts": "debt", "as": "map"}.tools/specs/record.pyand its JavaScript twinrecord.mjs, unchanged.test/specs/record.test.mjsholds the two to the same reading for the standards record as well.tools/standards/gate.pyreads the register through that loader.--seedwrites one debt file. On a pull request the base's debt is read per file, or from the base'sdebt.jsonwhere the base still has one.test/page/standards.test.mjsreads the register and debt throughrecord.mjs.eachmay namefiles(one request per file) in place of one file.generators/system'sregisterkeystep uses it; the program itself is unchanged.Standards are now read in key order, not in the order they were appended.
test/standards/requirements.jsonand every generated output are byte-for-byte unchanged.To register a standard now: add
app/specs/standards/register/<key>.json; seed its debt withtools/standards/gate.py --seed <key>, which writesdebt/<key>.json. An open pull request that touched the old lists moves its entries into those files when it next merges main.Not in this change
debt.partsinapp/specs/parts/spec.json, so it stays there.app/specs/oracle/spec.json): next, now that The chat is a conversation log of whole turns, apart from the status line #169 gives find a total order.Checks run on 16389a0 (merged with main 0f9d804)
tools/generators/run.sh --check(GN-01)tools/standards/gate.py --check --base origin/main(ST-01: 379 of 1627 requirements enforced, as on main)tools/handwritten/gate.py --check --base origin/main(25,524 hand-written lines, none added or grown)tools/guests/gen.py check: 20 currentnode --test test/specs/*.test.mjs: 5 pass;test/router/labels.test.mjspasses (176 against 250 here)bash -non the train's script, and the scratch-repository run above.Not run: the new train loop on a runner;
cargo test -p repo-conformanceand the browser acceptances on this head.🤖 Generated with Claude Code
https://claude.ai/code/session_013Asa3LuNYtPDBa4FbSEQUZ