You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
ci: restart the nightly size history after the libmoq rename - #9071
Restarts Nightly's size history now that moq-dev/moq#8941 renamed the hand-written C archives from moq-c-<target> to libmoq-<target>, and measures the generated C package's moq-c-generated-<target> archives again, which that rename stopped matching. Finishes quest/m2/libmoq-size-history.md.
Changes
Deleted the only retained nightly-size-history artifact (id 14356087495, from Nightly run 44127032469 at cd9a8e9, which predates feat(c): the hand-written C API ships as libmoq, its final line #8941's merge), with the command in sh/size/README.md. Zero such artifacts remain, so tonight's Nightly reports without --history and starts one fresh history.
Measures the generated C package again: sh/size/release.py collects moq-c-generated-* as its own series. feat(c): the hand-written C API ships as libmoq, its final line #8941's libmoq-* replaced the moq-c-* glob that used to match these archives too, and release-c.yml still publishes them on c-v* tags. report.py starts a series for a new key, so they join the fresh history without another restart.
The sizes job in nightly.yml now needs release-c, so the download no longer relies on Release C happening to finish first.
Deletes the quest file. No other file links to it.
The restart drops every series' 30-day baseline: for the next 30 days each night compares only against the previous night.
Decisions
Deleted the artifact before tonight's Nightly ran instead of waiting for a red run to prove the refusal. A local simulation showed the same thing (below), and waiting would only have cost one failed night.
Generated C archives: ✅ track them as their own moq-c-generated-<target> series (rejected: record them as deliberately untracked). They still ship, and quest/m2/README.md keeps the generated C line for after the cut. Raised by Pullfrog's review.
Didn't dispatch a Nightly by hand to confirm: it rebuilds every release target. The next scheduled run is the confirmation. If its Collect and compare sizes step fails, the cause is something else, not the rename.
Validation
Downloaded the deleted artifact first (kept locally) and confirmed its history.json keys name release/moq-c-<target>/… and release/moq-c-generated-<target>/….
with --history set to the retained history: exit 1, artifacts missing or renamed: release/moq-c-… (all 8 old keys), which is the failure the quest predicted.
without history: exit 0, "a new artifact history starts with one report", with the four release/libmoq-<target>/… keys recorded.
gh api …/actions/artifacts?name=nightly-size-history now returns total_count: 0.
Upload names on main match: libmoq.yml uploads libmoq-${{ matrix.target }} and sh/size/release.py collects libmoq-*.
Observed: the two Nightlies after the deletion (44342367698, 44562804083) passed Collect and compare sizes and now retain a two-night history keyed on libmoq-*.
Generated archives: downloaded run 44562804083's four moq-c-generated-* artifacts and its nightly-size-history, ran release.sizes on them (keys release/moq-c-generated-<target>/moq-c-<target>.{tar.gz,zip}, 14 to 22 MB), merged them into that night's sizes.json, and ran report.py with --history on the retained history: exit 0, "No size regression", the four keys listed under "New artifact series", and the written history carries them.
release.test.py's fixture now includes a generated archive; release.test.py, report.test.py, and nightly.test.py pass.
Impact
No public API or wire change. CI only: Nightly's size baseline restarts, and the generated C archives are measured again.
Caused by
moq-dev/moq#8941 - renamed the release artifacts, which the retained history can't follow by design (found by person, planned in the quest before merge), and its libmoq-* pattern dropped the moq-c-generated-* archives the old moq-c-* glob matched (found by reviewer)
Risk: high, it changes .github/workflows/nightly.yml, which is always high: workflows run with the repository's secrets and decide what merges and ships
Score: 9/10, Focused changes add generated C archives with matching fixture coverage and no actionable defects found in static review; tests were not run as instructed.
Rounds: 1; first round 0; findings 0: fixed 0, refuted 0, open 0
Findings: none
Models
Size: XS
Implementer: Claude Code, claude-opus-5-5, unknown effort (arm B)
Implementer: Claude Code, claude-opus-5-5, medium effort (iterate, XS, arm B)
Deleted the one retained nightly-size-history artifact (run 44127032469,
cd9a8e9), which still named the moq-c-<target> and
moq-c-generated-<target> keys, so the next Nightly starts a fresh history.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
kixelated
changed the title
Restart the nightly size history after the libmoq rename
ci: restart the nightly size history after the libmoq rename
Nov 18, 2026
No actionable comments were generated in the recent review.
Walkthrough
The nightly sizes job now depends on release-c. The size artifact patterns include moq-c-generated-*, and the release-size test covers a versioned Windows x86_64 C-generated ZIP. The document describing the size-history restart plan is removed.
Merge Risk: Minimal · up to 3b407
This change makes the nightly size job wait for the C release and track the generated C archives as their own series. No concrete merge-blocking risk was found. The next scheduled Nightly run will confirm the workflow end to end.
Pre-merge checks: 4
Check name
Status
Explanation
Title check
The title "ci: restart the nightly size history after the libmoq rename" matches the main change. The PR deletes the retained nightly-size-history artifact to restart the size baseline after the libmo…
Description check
The description links the rename in #8941 to the restart of the Nightly size history. It also covers the deleted artifact, the sh/size/release.py change for moq-c-generated-* archives, the …
Linked Issues check
Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check
Check skipped because no linked issues were found for this pull request.
Full details: Title check
Explanation
The title "ci: restart the nightly size history after the libmoq rename" matches the main change. The PR deletes the retained nightly-size-history artifact to restart the size baseline after the libmoq rename, and it adds generated C archive collection and a release-c dependency in nightly.yml.
Full details: Description check
Explanation
The description links the rename in #8941 to the restart of the Nightly size history. It also covers the deleted artifact, the sh/size/release.py change for moq-c-generated-* archives, the nightly.yml dependency on release-c, the deleted quest file, and validation steps. It relates directly to the changeset.
Finishing Touches
Simplify code
Commit to this branch
Create a new PR
Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.
File: No configuration file from this repository applied
Commit: Unavailable
Layers: Organization UI
Review profile: CHILL
Plan: Advanced
Run ID: 8d6161b7-9830-4666-9a22-53e718d4ab06
Commits
Reviewing files that changed from the base of the PR and between f1eecd3 and 3b4077f.
Files selected for processing: 4
.github/workflows/nightly.yml
quest/m2/libmoq-size-history.md
sh/size/release.py
sh/size/release.test.py
Files with no reviewable changes: 1
quest/m2/libmoq-size-history.md
Included review availability: This review used your included allowance. Your plan provides up to 4 included reviews per hour; 0 remain after this review.
The reason will be displayed to describe this comment to others. Learn more.
Important
The restart also permanently stops measuring the four moq-c-generated-<target> archives, which still ship. #8941 replaced the moq-c-* glob in sh/size/release.py with libmoq-*, and that old glob was what matched the generated C package's artifacts that release-c.yml keeps publishing on c-v* tags. Fresh history means report.py's "missing or renamed" refusal can never name them again, so this is the cheapest moment to restore the pattern.
Approach: sound — the history can't follow a rename by design, and deleting the retained artifact is the documented restart in sh/size/README.md, backed by a local simulation instead of a wasted red run. The gap is that the same restart silently retires a series that is still shipping.
Reviewed changes
Quest file deleted — quest/m2/libmoq-size-history.md recorded the merged plan and its waiting state; nothing else in the repo links to it, so removing it finishes the quest.
CI artifact deleted (outside the diff, verified) — actions/artifacts?name=nightly-size-history now returns total_count: 0, and sh/size/nightly.sh therefore passes no --history, so the next Nightly writes a one-row history as described.
Rename coverage checked — libmoq.yml uploads libmoq-${{ matrix.target }} and sh/size/release.py collects libmoq-*, matching the post-#8941 upload names.
⚠️ The restart drops the moq-c-generated-* series, which is still a shipping artifact
Last night's report (run 44127032469, bench-size artifact) carries 8 C keys: four size/release/moq-c-<target>/… and four size/release/moq-c-generated-<target>/…. The second four were only ever collected because moq-c-* matched them, and no pattern in ARTIFACTS matches them now, while release-c.yml still publishes the c/moq package on c-v* tags. Restarting the history is what makes that loss unobservable: compare() refuses removed keys, but only keys the retained history knows about.
Technical details
# Generated C package's size series disappears with the restart## Affected sites-`sh/size/release.py:13` — `ARTIFACTS` has no pattern matching `moq-c-generated-*`; `libmoq-*` replaced the `moq-c-*` glob that used to match both the hand-written and generated uploads.
-`.github/workflows/release-c.yml:124` — still uploads `moq-c-generated-${{ matrix.target }}`, and its `release` job publishes those archives on `c-v*` tags.
-`.github/workflows/nightly.yml:778` — the `sizes` job downloads artifacts from `release-c` but does not list it in `needs`.
-`sh/size/report.py:40` — the "artifacts missing or renamed" refusal only compares against keys the history contains; with an empty history it cannot report a dropped series.
## Required outcome- Either the four `moq-c-generated-<target>` archives come back under comparison, or the PR records that the generated package is deliberately no longer size-tracked (including that it still ships).
- The choice must be made before the fresh history is written, because a pattern added after tonight's run starts the series from that later night.
## Suggested approach (optional)- Add `"moq-c-generated-*"` to `ARTIFACTS` so the first fresh report records it alongside `libmoq-*`; `sizes()` already keys on the artifact directory, so there is no duplicate-key clash with `libmoq-*`.
- Adding `release-c` to the `sizes` job's `needs` would remove the reliance on that job happening to finish before the download, which currently works only by timing.
## Open questions for the human (optional)- Is dropping the generated package's measurement intended, given `quest/m2/README.md` keeps "the generated C line" for after the cut?
#8941's libmoq-* pattern replaced the moq-c-* glob that also matched
release-c.yml's moq-c-generated-<target> uploads, which still ship.
The sizes job now waits on release-c instead of relying on timing.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Fixed in 3b4077f: sh/size/release.py now collects moq-c-generated-* as its own series, and the sizes job in nightly.yml needs release-c so the download no longer relies on timing.
The fresh history had already started (Nightlies on Nov 18 and 19 passed with libmoq-* keys only), but report.py starts a series for a new key, so no second restart is needed. Simulated it on run 44562804083's real moq-c-generated-* artifacts against that night's retained history: exit 0, "No size regression", and the four keys listed as new series. Details under Validation in the PR body.
The reason will be displayed to describe this comment to others. Learn more.
✅ No new issues found.
Reviewed changes — the delta since the prior pullfrog review (6ffcc68..3b4077f), which resolves the earlier finding about the dropped generated C series.
Generated C archives measured again — sh/size/release.pyARTIFACTS now lists moq-c-generated-*, matching the moq-c-generated-${{ matrix.target }} uploads in release-c.yml, so they join the fresh history as their own series.
sizes job depends on release-c — nightly.yml adds release-c (defined at line 727) to needs, removing the reliance on timing.
Test fixture extended — release.test.py includes a generated archive and its expected key.
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
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.
Restarts Nightly's size history now that moq-dev/moq#8941 renamed the hand-written C archives from
moq-c-<target>tolibmoq-<target>, and measures the generated C package'smoq-c-generated-<target>archives again, which that rename stopped matching. Finishesquest/m2/libmoq-size-history.md.Changes
nightly-size-historyartifact (id 14356087495, from Nightly run 44127032469 at cd9a8e9, which predates feat(c): the hand-written C API ships as libmoq, its final line #8941's merge), with the command insh/size/README.md. Zero such artifacts remain, so tonight's Nightly reports without--historyand starts one fresh history.sh/size/release.pycollectsmoq-c-generated-*as its own series. feat(c): the hand-written C API ships as libmoq, its final line #8941'slibmoq-*replaced themoq-c-*glob that used to match these archives too, andrelease-c.ymlstill publishes them onc-v*tags.report.pystarts a series for a new key, so they join the fresh history without another restart.sizesjob innightly.ymlnow needsrelease-c, so the download no longer relies on Release C happening to finish first.The restart drops every series' 30-day baseline: for the next 30 days each night compares only against the previous night.
Decisions
moq-c-generated-<target>series (rejected: record them as deliberately untracked). They still ship, andquest/m2/README.mdkeeps the generated C line for after the cut. Raised by Pullfrog's review.Collect and compare sizesstep fails, the cause is something else, not the rename.Validation
history.jsonkeys namerelease/moq-c-<target>/…andrelease/moq-c-generated-<target>/….sizes.jsoninto the post-feat(c): the hand-written C API ships as libmoq, its final line #8941 shape (moq-c-*becamelibmoq-*,moq-c-generated-*dropped) and ransh/size/report.pyon it:--historyset to the retained history: exit 1,artifacts missing or renamed: release/moq-c-…(all 8 old keys), which is the failure the quest predicted.release/libmoq-<target>/…keys recorded.gh api …/actions/artifacts?name=nightly-size-historynow returnstotal_count: 0.libmoq.ymluploadslibmoq-${{ matrix.target }}andsh/size/release.pycollectslibmoq-*.Collect and compare sizesand now retain a two-night history keyed onlibmoq-*.moq-c-generated-*artifacts and itsnightly-size-history, ranrelease.sizeson them (keysrelease/moq-c-generated-<target>/moq-c-<target>.{tar.gz,zip}, 14 to 22 MB), merged them into that night'ssizes.json, and ranreport.pywith--historyon the retained history: exit 0, "No size regression", the four keys listed under "New artifact series", and the written history carries them.release.test.py's fixture now includes a generated archive;release.test.py,report.test.py, andnightly.test.pypass.Impact
No public API or wire change. CI only: Nightly's size baseline restarts, and the generated C archives are measured again.
Caused by
libmoq-*pattern dropped themoq-c-generated-*archives the oldmoq-c-*glob matched (found by reviewer)Follow-ups
None.
Review
Rewritten by
quest reviewafter each round.Head: 3b4077f
Base: origin/main at 075130c
Risk: high,
it changes .github/workflows/nightly.yml, which is always high: workflows run with the repository's secrets and decide what merges and shipsClassified: high by the always-high paths
Reviewer: Codex, gpt-6.1-sol, medium effort
Session:
01a1f59d-d440-7d20-a9c8-cde4a112158c, round 1Score: 9/10,
Focused changes add generated C archives with matching fixture coverage and no actionable defects found in static review; tests were not run as instructed.Rounds: 1; first round 0; findings 0: fixed 0, refuted 0, open 0
Findings: none
Models
Size: XS
Implementer: Claude Code, claude-opus-5-5, unknown effort (arm B)
Implementer: Claude Code, claude-opus-5-5, medium effort (iterate, XS, arm B)
(Written by Claude Opus 5.5 with medium)