fix: scope merge auto-recovery state per experiment - #1440
Open
XianBW wants to merge 1 commit into
Open
Conversation
Replace process-wide threshold mutations with an experiment-level merge-phase marker so parallel traces cannot leak recovery state across iterations. Add regression coverage for all merge generators. Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Description
Replace merge-phase mutations of the process-wide
DS_RD_SETTING.coding_fail_reanalyze_thresholdandDS_RD_SETTING.consecutive_errorsvalues with an explicitis_merge_phasemarker on each generatedDSExperiment.DataScienceRDLoop.record()now skips auto-recovery only for the merge-phase experiment being recorded. All four merge generation paths set this marker, including the parallel multi-trace router.Regression tests cover the synchronous merge generators, parallel merge generation, unchanged global thresholds, and the default non-merge state.
Motivation and Context
Fixes #1431.
The previous implementation permanently changed a shared settings singleton. That disabled auto-recovery for later iterations and could leak across concurrent traces. A temporary context manager around
ExpGen.gen()would restore values beforerecord()performs the recovery check, so the state instead travels with the experiment through its full iteration lifecycle.How Has This Been Tested?
Targeted tests were added under
test/data_science/test_merge_auto_recovery.py. Local execution was omitted at the request of the submitter.Screenshots of Test Results (if appropriate):
Types of changes
📚 Documentation preview 📚: https://RDAgent--1440.org.readthedocs.build/en/1440/