Repository navigation
Floor batch-3 wall budget: signed raise 1680 → 2100s (namespace flip #7178 cost) - #7204
Merged
Merged
Conversation
…7178 cost) The namespace-only flip #7178 (merged 2026-07-25) makes bare references resolve by containment-walk (#6848 per-entry bare-reference fixpoint), adding ~200s to the affected-set discovery-corpus resolve wall vs the import-scoped basis the prior 1680 was sized on (#7137, all pre-flip). This is a real, attributed cost shift, not a per-PR regression: THREE disjoint post-flip observations exceed 1680 -- #7198 (run 30140810666) 1820001ms, #7188 (run 30142221249) 1859345ms, and the falsifier lane's warm post-flip run (~36min, same signature). Operator-signed raise (briansrls, 2026-07-25): 2100s (35min), sized to the +/-20% fleet host envelope over the observations so a slow-host broad-touch run does not round-trip this row a third time. STOPGAP until #6848 cuts per-entry resolve (dispatched as a lane) and the row dissolves into #7162's derived clamp. Remedy stays diagnose-or-signed-raise, never rerun. Standalone budget PR -- unblocks the fleet (#7198, #7188, and main's next push all red on batch-3 over-budget) in one commit, disjoint from the feature lanes. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
cursor Bot
pushed a commit
that referenced
this pull request
Jul 25, 2026
Bring design_register_lift_parity digest re-pin (#7169), namespace-flip aftermath, and related main landings. Keep this PR's derived FloorBatchClamp authority over main's stopgap static 2100s batch-3 raise (#7204) — the clamp is the mechanism that raise was awaiting. Co-authored-by: Brian Searls <briansrls@users.noreply.github.com>
briansrls
added a commit
that referenced
this pull request
Jul 25, 2026
…dmap_css re-pin, migration-debt ratchet updates)
briansrls
added a commit
that referenced
this pull request
Jul 25, 2026
…ip containment-walk resolve isn't recomputed once-per-entry-per-worker (~200s/run on batch-3 since namespace flip #7178; this is the STOPGAP-dissolving root behind the #7204 budget raise). Root-cause + share the resolve across entr (#7217) * WIP: Floor #6848: memoize the per-entry bare-reference fixpoint so post-flip * WIP: Floor #6848: memoize the per-entry bare-reference fixpoint so post-flip * WIP: Floor #6848: memoize the per-entry bare-reference fixpoint so post-flip * WIP: Floor #6848: memoize the per-entry bare-reference fixpoint so post-flip * WIP: Floor #6848: memoize the per-entry bare-reference fixpoint so post-flip * WIP: Floor #6848: memoize the per-entry bare-reference fixpoint so post-flip * Doc + carrier notes: assembly sub-row dissolution trigger, second cost-shape fix Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> * WIP: Floor #6848: memoize the per-entry bare-reference fixpoint so post-flip * WIP: Floor #6848: memoize the per-entry bare-reference fixpoint so post-flip * Review 42566: caller reads local_names from the export-name index instead of re-folding One extra O(|bindings|) fold per module per pass — the same authority computed twice in the function whose thesis is that the set is ONE derived fact. The Absent arm computes the true value from that same authority, so the structurally-unreachable fallback stays honest rather than fabricating a default. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> * WIP: Floor #6848: memoize the per-entry bare-reference fixpoint so post-flip * Restore the regen-emitted seed clobbered by an auto-commit mid-experiment An auto-commit captured a TEMPORARY revert of v1_compiler_infer.rs (a6a0a90) that I had made only to build a before-arm test binary for the unit-suite comparison. That left the .dag authority carrying the change while the emitted seed carried main's version - regen drift, which the self-host gate would have redded. This restores the regen output that src/v1/04_infer.dag actually emits. No .dag change. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> * WIP: Floor #6848: memoize the per-entry bare-reference fixpoint so post-flip --------- Co-authored-by: Brian Searls <briansrls@gunb.ai> Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
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.
What
One
datarow change:gunbc.ci_spec.gunbc_ci_floor_batch_wall_budget_secondsindex[2](batch 3, the affected-set discovery corpus) 1680 → 2100 seconds, plus the appended operator-signed raise receipt ingunbc_ci_floor_batch_wall_budget_note.Why — attributed, not suspicious
The namespace-only flip (#7178, merged 2026-07-25T02:02) makes bare references resolve by containment-walk (#6848 per-entry bare-reference fixpoint). That adds ~200s to the batch-3 resolve wall fleet-wide versus the import-scoped basis the prior
1680was sized on (#7137, all pre-flip). The cost shift is real and shared, not a per-PR regression — three disjoint post-flip observations exceed 1680:Sizing
1980 clears the two PR observations at ~6.5%, but both are single host placements and this row's own history (1440 sized off three fast-biased observations, redded the majority, re-raised same day) is the cautionary tale. 2100 is sized to the ±20% fleet host envelope so a slow-host broad-touch post-flip run does not round-trip this row a third time.
STOPGAP until #6848 cuts per-entry resolve (dispatched as its own lane now) and the row dissolves into #7162's derived clamp (units × rate absorbs the width axis; the flip's cost becomes a re-signed rate, not a wall). The remedy stays diagnose-or-signed-raise, never rerun.
Scope
Standalone — a single data-row edit, disjoint from the feature lanes. Unblocks the fleet (#7198, #7188, and main's next push all red on
FLOOR-BATCH-OVER-BUDGET batch=3) in one commit. No behavior change beyond the wall ceiling; ceiling-sum 3540 → 3960s remains a paper figure (per-batch maxima never co-occur; the actual floor step stays well under the 55-min step cap).🤖 Generated with Claude Code