marius/Open_Ultra_SAST
1
0
Fork 0
mirror of https://github.com/norandom/OpenUltraSAST.git synced 2026-10-16 03:49:26 +00:00
Code Issues Packages Projects Releases Wiki Activity

test(learning-harness): a flaky harness must not be mistaken for a flaky detector

Browse Source 
The round-zero test decided which materialized side it was looking at by
searching the whole path for a substring, and the random temporary directory
name sometimes contained it. The detector then "found" the bug on the fixed side
too, the pair flipped, and the model comparison failed about one run in eight.

The check is now the directory name alone. This matters beyond the test: a noise
floor exists to measure how much a detector disagrees with itself, and noise
coming from the harness would be recorded as the detector's own and inflate
every family's regression budget.

Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
Claude-Session: https://claude.ai/code/session_01T7XD7x5KonMdmGRkp63EQG
This commit is contained in:
Marius Ciepluch
2026-09-06 11:45:00 +02:00
co-authored by Claude Opus 5
parent ad7c98eb50
commit 346cc703c3
1 changed files with 13 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/test_learning_rounds.py
+13 -2
View File
@@ -55,13 +55,24 @@ def _finding(family: str) -> StaticFinding:
)
def _is_vulnerable_side(root: Path) -> bool:
"""The materialized side, by directory name only.
An earlier version also matched "-v" anywhere in the path, which the random temporary directory name
sometimes contained, so the detector "found" the bug on the fixed side too and the pair flipped. A
flaky harness is indistinguishable from a flaky detector, which is precisely what a noise floor must
not measure.
"""
return root.name == "vuln"
def _steady(root: Path) -> list[StaticFinding]:
return [_finding("injection")] if "-v" in str(root) or root.name == "vuln" else []
return [_finding("injection")] if _is_vulnerable_side(root) else []
def _flaky(counter: list[int]): # type: ignore[no-untyped-def]
def scan(root: Path) -> list[StaticFinding]:
if not ("-v" in str(root) or root.name == "vuln"):
if not _is_vulnerable_side(root):
return []
counter.append(1)
return [_finding("injection")] if len(counter) % 2 else []