Results audit¶
Generated by scripts/audit_results.py. results/ is never committed; this table is the record of what is on disk. Nothing was deleted. Whether a flagged folder is superseded is the researcher’s decision.
Folder |
Finished runs |
Failed runs |
Code versions (src files changed since) |
Flags |
|---|---|---|---|---|
bench |
0 |
0 |
none |
tooling output (no run.json) |
canary |
0 |
0 |
none |
tooling output (no run.json) |
e0 |
2 |
0 |
cd72d83 (35) |
none |
e0_rerun |
2 |
0 |
cd72d83 (35) |
none |
e1 |
2 |
0 |
2bd12d6 (34) |
none |
e2 |
54 |
0 |
c26aa5f (32) |
none |
e2_smoke |
42 |
0 |
c26aa5f (32) |
smoke |
e3 |
56 |
0 |
f38858e (23) |
none |
e4 |
76 |
0 |
1d302ea (1) |
none |
e4_smoke |
0 |
0 |
none |
smoke; tooling output (no run.json) |
e4_smoke_v2 |
8 |
0 |
0d90105 (1) |
smoke |
e4_smoke_v3 |
38 |
0 |
3c396a6 (1) |
smoke |
li |
9 |
2 |
2bd12d6 (34), fa614c3 (34) |
several versions; has failed runs |
logs |
0 |
0 |
none |
tooling output (no run.json) |
mutation |
0 |
0 |
none |
tooling output (no run.json) |
rerun |
0 |
0 |
none |
tooling output (no run.json) |
resume_check |
1 |
0 |
42da088 (34) |
none |
slicing |
0 |
0 |
none |
tooling output (no run.json) |
The number in brackets is how many files under src/ differ between that code version and HEAD (? when git cannot resolve the version). A run on code that has since changed is not necessarily wrong: the regression tests and the clean-clone reproduction decide that. It is the signal to look.
Flags: dirty = made on an uncommitted tree; smoke = a timing run, never evidence; several versions = more than one code version in the folder; tooling output = no run record (benchmarks, canaries, mutation logs).