Skip to content

feat(erp): integrate bounded runtime observations into main - #659

Merged
zzylol merged 22 commits into
mainfrom
feat/metricsql-calibration-workflow
Sep 11, 2026
Merged

zzylol merged 22 commits into
mainfrom
feat/metricsql-calibration-workflow

Conversation

@zzylol

@zzylol zzylol commented Sep 11, 2026

Copy link
Copy Markdown
Contributor

Runtime ERP evidence is complete on its stack, but PR #638 was merged into its already-merged calibration base rather than the backend main branch. The production observation path therefore still needs one explicit main integration.

This integrates the bounded empirical observation contract, catalog-generation isolation, control-plane fitting, finite-completion publication, and actual Worker feedback path. The data plane emits bounded empirical counts; the control plane owns distribution fitting and Planner translation.

Before this PR: main contained ERP readout calibration but not the runtime observation and hydration path.

After this PR: a finite, durably completed ingest can publish catalog-bound observations which a later normal compilation may use. Stale, malformed, foreign, partial, or unsupported observations fail closed to theoretical sizing or exact execution.

Verification: 18 focused control-plane ERP tests and the actual Worker → feedback → control plane → Planner process test passed on the main-integrated tree with the pinned asap_sketchlib #139 revision. Merge conflict resolution received an independent source review.

This remains a finite-input feedback closure. It does not implement autonomous continuous replanning.

zzylol and others added 22 commits September 10, 2026 20:13
feat(erp): publish bounded materialization observations after finite completion
@zzylol
zzylol merged commit c0928c4 into main Sep 11, 2026
1 check passed
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