Repository navigation
Batch registered research CI without dropping package coverage - #2505
Merged
Mister-Meeseeks merged 2 commits intoOct 16, 2026
Merged
Conversation
4 tasks done
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
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
A shared replay-runner or workflow change selects all 450 registered packages on the current main branch. The existing single job matrix exceeds GitHub's 256-job matrix limit before any package can run. This addresses the concrete limitation recorded by EdDiboi in the handoff.
Add an opt-in bounded matrix output and a reusable batch workflow. The current 450-package selection becomes two complete batches [256,194]. Outer batches run one at a time; inner package jobs retain
max-parallel: 4, a fresh runner/checkout/environment per package, the 30-minute job limit, exact dependency checks and committed-file drift checks. Both workflow paths invalidate all packages. The legacy flat planner CLI is unchanged. More than 256 batches fails explicitly instead of truncating coverage.Validation: the required research tooling suite passed 69 tests with 3 pre-existing Linux-only skips on CPython 3.12.14. A subsequently added real grouped-CLI regression and all five focused batch tests passed. Boundary checks cover 0, 1, 255, 256, 257, 450 and 65,536 packages, with 65,537 rejected, invalid batch sizes, exact order/coverage and unchanged legacy output. Actual all-package CLI output contains every one of 450 packages exactly once. No hosted full-matrix success or mathematical cold replay is claimed.
GitHub documents the 256-job matrix limit and matrix jobs calling reusable workflows. The workflow design retains isolation rather than combining packages into one Python environment.
Task: portable-replay expertise/handoff override of complex-mechanisms seed c4a2795b534ea90dede453dea434600f. Current source main 0fdd7d8, priority revision 4; original documentation lead stopped as exact overlap with #2214. Complete PR/Discussion overlap checked, with freshness in the local receipt. Credit EdDiboi's recorded limitation and the existing registered-package runner/workflow authors. New work by ywz65 with substantial OpenAI Codex assistance, Apache-2.0. No selected frontier or frozen scientific input changes.
The 65,536-case test is a grouping boundary, not a hosted scalability claim. GitHub output-size and other execution limits still apply; the current 450-package matrix is below the job-output limit.