Conversation
zzylol
changed the title
zzylol
marked this pull request as draft
September 28, 2026 02:28
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Deferred follow-up. Not required for the current #728 → #742 → #775 synthetic-selection and data-plane execution milestone. Online ERP feedback and runtime replanning are explicitly out of scope for that milestone.
Prevent malformed resource measurements from becoming valid candidate prices, and separate resource arithmetic from ranking.
Before this PR: batch CPU normalization silently truncated mismatched user/system sample arrays; negative raw components could pass when their sum was positive, and infinite work could produce a zero per-operation cost.
After this PR: normalization rejects those inputs, preserves paired repetitions and units, and keeps missing resources unknown. A public ERP test checks update/read/merge multiplicities and retained byte-seconds with record identity.
Validation: the new regression failed before the fix (seven failing cases); all nine exporter tests and the Rust
resource_measurementsintegration test pass afterward. Coefficients in the tests are synthetic; real measurements and provenance remain a Level 3 requirement.Base: accuracy-evidence tests. Next: #759 measured selection and execution audit.