You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
benchmarks: the opening "Nothing else is cited" is contradicted by four later citations #1140
Found by the approver of #1131 while judging AC16 of that job's spec (docs/rewrite/benchmarks.md must present nothing unmeasured as measured). It blocks nothing: every number it concerns names its own source on its own line, and AC16's numeral rules hold.
What
docs/rewrite/benchmarks.md:8 says, of the document's numbers, "Nothing else is cited." Four later places cite a source other than the committed artifact for a number:
:23 and :199 — the free disk at run time, 8.6 GB, from df -h / (labelled as such);
the closing section (:328-334) — ADR 0003's 267.6 / 44.9 / 12.2 ms, labelled as job 007's context.
Each of those is permitted by the spec's AC16 (rule 6 for #868/#864, rule 7 for ADR 0003, rule 5 for the disk figure) and each names its source where it stands, so no number is presented as something it is not. The opening sentence nevertheless promises more than the document does, and it is the sentence the round-1 and round-3 lenses used as the standard that removed the unsourced multipliers.
Acceptance
The opening paragraph says what is true: the rig's own numbers come from the one command and its artifact, and the gate's thresholds, the tracker's target, the disk reading and job 007's figures are cited where they appear and labelled as not this rig's.
"Nothing else is cited" does not survive in a form a reader can falsify four sections later.
Activity
feat(storage): the benchmark rig, and the first LMDB recording (#1141)