feat(cli): impact --tests lexical fallback + IDF-weighted pack briefings#453
Merged
Conversation
…efings Empty tests answers from sidecar/disk-cache fast paths now fall through to the full path instead of printing 'No tests found'; the full path backstops missing call edges (namespace calls like xr.where) with whole-word lexical reachability over test bodies, labeled as such. Briefing ranking: IDF term weights, <details> stripping, .attr suffix terms, one name-echo slot.
There was a problem hiding this comment.
inspect review
Triage: 9 entities analyzed | 0 critical, 0 high, 8 medium, 1 low
Verdict: standard_review
Findings (0)
Reviewed by inspect | Entity-level triage found 0 high-risk changes
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Two fixes from live agent-benchmark failures:
impact --tests answered 'No tests found' for xr.where — tests reach it via a module-namespace call (
xr.where(...)), which produces no call edge. Empty tests answers from the sidecar/disk-cache fast paths are now non-authoritative (fall through to the full path), and the full path backstops zero edges with whole-word lexical reachability over test bodies, printed with an explicit '(lexical fallback)' label since it is weaker evidence than a call edge. On the failing case this turns 0 selected tests into a 56-test net (vs 1,900+ for whole-file pytest runs).--pack briefings misaimed on issues whose key term is a common word ("method") or whose tail is an environment dump. Ranking is now IDF-weighted,
<details>blocks are stripped before term extraction, attribute accesses glued to receivers (d2.loc) emit.loc-style suffix terms, and one of three briefing slots goes to the strongest name-echo orient hit (culprits are often NAMED like the issue's API but never mention it in their body). Calibration: 3 of 4 ground-truth issues now put the gold-patch function (or its immediate neighbor) in the briefing, up from 2 of 4.Note:
cwd_resolutiondiff tests fail identically on clean main (pre-existing, unrelated).