EMG-001 epic build-speed (7 tickets) by noahgift · Pull Request #5932 · paiml/aprender

noahgift · 2026-10-22T12:00:37Z

EMG-001 epic: build-speed (7 tickets)

Supersedes #5927 (v2 head b75222a had two seat-blocking defects; fixed in v3 @d63346dbfd). Stacked on #5925 (base ci/emg-001-c1-single-shard). After #5925 merges, flow rebases this onto main; seats carry over on an empty range-diff. Unarmed on purpose (EMG-001).

Closes #5738
Closes #5848
Closes #5870
Closes #5868
Closes #5832
Closes #5874
Closes #5887

Spec: EMG-001.

🤖 Generated with Claude Code

…ds 4 -> 16 (#5924) Shard pins/matrices -> [1] for PR and merge queue; steps gated on shard 2/3 also run when shards == 1; sigma-shard-1 upload no longer skipped at one shard; check_shard_count_single_source self-test updated. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

…nightly/dogfood/mutants/fleet-toolset schedules (PMAT-5847) (squashed from #5848) Refs #5847 PMAT-Ticket: PMAT-5847 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> PMAT-Batch: BATCH-aprender-build-speed

…a so a REBASE-replayed PR is judged whole (Refs #5855, #5859) (squashed from #5870) Refs #5859 PMAT-Ticket: PMAT-5859 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> PMAT-Batch: BATCH-aprender-build-speed

…IGKILLed (detached watcher) (squashed from #5868) Refs #5868 PMAT-Ticket: PMAT-5868 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> PMAT-Batch: BATCH-aprender-build-speed

…+3 ticks (K1, #5798 flake) (squashed from #5832) Refs #5798 PMAT-Ticket: PMAT-5798 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> PMAT-Batch: BATCH-aprender-build-speed

…from #5738) Refs #5394 PMAT-Ticket: PMAT-5394 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> PMAT-Batch: BATCH-aprender-build-speed

Refs #5873 PMAT-Ticket: PMAT-5873 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> PMAT-Batch: BATCH-aprender-build-speed

#5887) Refs #5889 PMAT-Ticket: PMAT-5889 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> PMAT-Batch: BATCH-aprender-build-speed

…eadlock; step-budgets names the shard job (1, 1) Review fixes for #5927 (w5032 REQUEST-CHANGES): the setsid watcher was a child of the top-level shell, so wait at the end blocked on it (rc=124); ( setsid ... & ) gives rc=0. GitHub names the single-shard job 'workspace-test-shard (1, 1)'; rows 2 and 3 no longer exist. PMAT-Ticket: PMAT-5926 PMAT-Batch: BATCH-aprender-build-speed Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

noahgift · 2026-10-22T12:00:58Z

Seat (aprender-w5032, non-author): APPROVE @ d63346d — Opus lane (not the Sonnet author family) reviewed the gate content at b75222a (#5927); its only blocker (check_ladder_cells_producer.sh watcher deadlock) is fixed in this one-commit delta, which I re-ran: script rc=0 with all mutants as expected, step-budgets row now matches the job name (1, 1), EMG-2 ok (1 ticket + 1 batch). Receipt: handoff/emg-001/quorum-buildspeed-b75222add3.md (update section). Not armed.

noahgift · 2026-10-22T12:01:00Z

Seat (aprender-w5032, non-author): APPROVE @ d63346d — gemini-4-argon-high lane (not the Sonnet author family) reviewed the gate content at b75222a (#5927); its only blocker (check_ladder_cells_producer.sh watcher deadlock) is fixed in this one-commit delta, which I re-ran: script rc=0 with all mutants as expected, step-budgets row now matches the job name (1, 1), EMG-2 ok (1 ticket + 1 batch). Receipt: handoff/emg-001/quorum-buildspeed-b75222add3.md (update section). Not armed.

noahgift · 2026-10-22T12:02:30Z

arbiter-cop non-author sign: APPROVE @d63346dbfddf92e9bc59836ccbcd05f6f0586595, CONDITIONAL.
Read: receipt handoff/emg-001/quorum-buildspeed-b75222add3.md. At b75222 Opus approved and argon requested changes on the ladder-cells watcher deadlock. The 1-commit delta fixes it (( setsid … & )) plus the step-budget row name workspace-test-shard (1, 1), and w5032 checked it. kfocus-shadow-lab.yml is workflow_dispatch-only (verified in the diff), so the signing-key and ref_name exposure needs write access a writer already has. Follow-up wanted: pass ref_name through env:, not inline in run:.
Conditions before arming: (1) the quorum seats are posted at d63346d; none are on the PR at 12:02Z. (2) Retarget to main after #5925 merges: the base is still ci/emg-001-c1-single-shard, and a stacked PR is never armed. (3) Note: 8 commits for 7 tickets.

noahgift

Seat (aprender-w5032, non-author): APPROVE @ d63346d — Opus lane (not the Sonnet author family) reviewed the gate content at b75222a (#5927); its only blocker (check_ladder_cells_producer.sh watcher deadlock) is fixed in the one-commit delta, which I re-ran: script rc=0, all mutants as expected; step-budgets row matches job name (1, 1); EMG-2 ok. Receipt: handoff/emg-001/quorum-buildspeed-b75222add3.md. Not armed.

noahgift

Seat (aprender-w5032, non-author): APPROVE @ d63346d — gemini-4-argon-high lane (not the Sonnet author family) reviewed the gate content at b75222a (#5927); its only blocker (check_ladder_cells_producer.sh watcher deadlock) is fixed in the one-commit delta, which I re-ran: script rc=0, all mutants as expected; step-budgets row matches job name (1, 1); EMG-2 ok. Receipt: handoff/emg-001/quorum-buildspeed-b75222add3.md. Not armed.

Conflicts in ci.yml, nightly.yml, kfocus-shadow-lab.yml, roadmap and resolve_base/step_budget scripts resolved to main's side: main already carries those squashed tickets (net diff vs main is now only check_ci_failfast.sh and check_lab_timeouts.sh). roadmap.yaml re-verified against its fragments. PMAT-Ticket: PMAT-5926 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

t and others added 10 commits October 22, 2026 12:49

chore(roadmap): aggregate PMAT-5924 fragment (#5924)

411f58f

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

fix(guards): check_ladder_cells_producer reaps its fake serves when S…

a1b0a69

…IGKILLed (detached watcher) (squashed from #5868) Refs #5868 PMAT-Ticket: PMAT-5868 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> PMAT-Batch: BATCH-aprender-build-speed

fix(ladder): hung health-wait case asserts the stall event, not stall…

934a158

…+3 ticks (K1, #5798 flake) (squashed from #5832) Refs #5798 PMAT-Ticket: PMAT-5798 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> PMAT-Batch: BATCH-aprender-build-speed

Merge remote-tracking branch 'origin/main' into work/5738c (squashed …

26383ae

…from #5738) Refs #5394 PMAT-Ticket: PMAT-5394 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> PMAT-Batch: BATCH-aprender-build-speed

lab: apr stub race (squashed from #5874)

36e44f1

Refs #5873 PMAT-Ticket: PMAT-5873 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> PMAT-Batch: BATCH-aprender-build-speed

ci(lab): kfocus shadow lab workflow + runner-label check (squashed from

b75222a

#5887) Refs #5889 PMAT-Ticket: PMAT-5889 Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> PMAT-Batch: BATCH-aprender-build-speed

noahgift commented Oct 22, 2026

View reviewed changes

noahgift mentioned this pull request Oct 22, 2026

EMG-001 epic build-speed (7 tickets) #5933

Closed

noahgift marked this pull request as draft October 22, 2026 14:06

Base automatically changed from ci/emg-001-c1-single-shard to main October 22, 2026 14:40

noahgift mentioned this pull request Oct 23, 2026

EMG-001 epic build-speed (7 tickets) #5927

Closed

Navigation Menu

Repository navigation

EMG-001 epic build-speed (7 tickets) - #5932

EMG-001 epic build-speed (7 tickets)#5932
noahgift wants to merge 11 commits into
mainfrom
work/5926-build-speed-v3

noahgift commented Oct 22, 2026

Uh oh!

noahgift commented Oct 22, 2026

Uh oh!

noahgift commented Oct 22, 2026

Uh oh!

noahgift commented Oct 22, 2026

Uh oh!

noahgift left a comment

Uh oh!

noahgift left a comment

Uh oh!

This branch has not been deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

noahgift commented Oct 22, 2026