Skip to content

Unbreak main: dangling examples.js_site imports after #9641, missing unit_variant_index after #9656 - #9662

Merged
briansrls merged 6 commits into
mainfrom
session/neat-tern-658
Aug 29, 2026
Merged

briansrls merged 6 commits into
mainfrom
session/neat-tern-658

Conversation

@gunbai-bot

@gunbai-bot gunbai-bot Bot commented Aug 29, 2026

Copy link
Copy Markdown
Contributor

Main has been red for the last 15 runs (gh run list --workflow witnesses.yml --branch main). Two integration collisions between PRs that were each green against an older main:

Verified remotely (BuildBuddy): cargo build --release -p v1-compiler --bins green; gunbc run --entry dag/test/claim/generated_artifact_assessment_witness_test.dag now resolves and typechecks the closure (it stops at "no function main", which is the expected shape for a witness module).

v1 admission: BoundedOperationalFix — restores the required floor for the v2 self-host program; no new behavior.

🤖 Generated with Claude Code

https://claude.ai/code/session_013G3t66QwKJFK5w8jXxMXP2

Brian Searls and others added 4 commits August 29, 2026 01:37
…eted, and give the six witness-bin TypeEnv initializers the unit_variant_index #9656 added

Two integration collisions between independently green PRs:
- #9641 deleted dag/examples/js_site but gunbc.generated_artifact and
  gunbc.generated_artifact_emit still imported it, so the whole-tree
  strict resolve refused and every floor on main has been red since.
- #9656 added TypeEnv.unit_variant_index; infer_semantics_witness.rs
  builds six TypeEnvs by hand and none carried it, so --bins failed.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013G3t66QwKJFK5w8jXxMXP2
…_variant_index too

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013G3t66QwKJFK5w8jXxMXP2
@gunbai-bot

gunbai-bot Bot commented Aug 29, 2026

Copy link
Copy Markdown
Contributor Author

Re review 57402 (REQUEST_CHANGES: "hand-written Rust in the compiler test tree without the mandatory single checkable receipt").

Verified against the diff and against DESIGN.md: the rule as cited does not exist. grep -n -i 'hand-rust\|checkable receipt\|single checkable' DESIGN.md returns nothing; §6's scaffold discipline governs new scaffold artifacts — a bridge, a shim, a parallel path — and asks whether the artifact survives the terminal architecture. The one line this PR adds to cli_run.rs is a struct-field initializer (unit_variant_index: crate::v1_rt::rc_empty_map()) inside an existing #[test], required because #9656 added the field to TypeEnv; without it cargo test -p v1-compiler --lib does not compile. It adds no path, no function, no surface — it is the same repair #9648 and this PR apply to the six bin-side initializers, and it is what the v1 maintenance standing's purpose test admits (a bounded operational fix that keeps the existing evidence executing). Leaving a compiler unit test uncompilable to avoid a one-line struct-literal edit would be the actual §4b regression.

No change made for this item; the JS-site removal and the initializers stand as reviewed.

— sent from neat-tern-658

Brian Searls added 2 commits August 29, 2026 02:18
…leted artifact registry entries go with them
…ation rows landed with #9641 and now refuse every PR as stale
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant