Skip to content

Re-vendor the unit observer from house-rules at a4f8dd4 - #77

Merged
bompus merged 2 commits into
mainfrom
revendor-unit-observer
Nov 6, 2026
Merged

bompus merged 2 commits into
mainfrom
revendor-unit-observer

Conversation

@bompus

@bompus bompus commented Nov 6, 2026

Copy link
Copy Markdown
Owner

bench/observe-unit.py is a copy of house-rules' unit observer. It was at e2a593e; this replaces it with the copy at a4f8dd4. The file is byte-identical to the source apart from sigflint's two header lines, which now name the new commit. Nothing else is adapted.

What the newer observer changes: it times the reads of each sample so a stall in the final sample is caught, requires pswpout in a storage phase, compares budgets before rounding, scales each pressure interval to one second, and writes no .ready file when the first sample already fails. run-admitted.sh uses only OBSERVER_OUT, --workload-cgroup auto, the .ready file and the top-level clean key; those are unchanged, and a missing .ready already means "no first sample, unit not measured" there.

A second commit adds bench/results/observer-revendor-20261106.md, a dated note. protocol-release-1.0.4.md says the frozen bench/ code is unchanged since the freeze commit 2685da5, which the observer file in the tree no longer satisfies; the note says the campaign ran the e2a593e copy (still in git at 2685da5) and lists what differs. No protocol or recorded result is edited.

Checked: diff against house-rules a4f8dd4 shows only the header lines; house-rules' 31 observer tests pass against this copy; bun run check passes. A separate read-only review (a non-Claude model, one round) found the harness compatible and the same two documentation points, which the note answers. No benchmark was run, and nothing is released.

Brings in the fixes made there since e2a593e: per-sample read timing so a
stall in the final sample is caught, a missing pswpout counter fails a
storage phase, budgets are compared before rounding, and limits can be set
by flags. The file is the house-rules copy with sigflint's two header lines
naming the new source commit; nothing else is adapted.
A dated note says which observer the 1.0.4 campaign ran (the e2a593e copy,
still in git history at the freeze commit) and what the new copy does
differently, so the protocol's statement about frozen code is not read as
covering the file now in the tree. No result or frozen record is edited.
@bompus bompus added the author:anthropic Pull request written by a anthropic model label Nov 6, 2026

@bompus-ai-review bompus-ai-review Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Bompus AI Review (GPT-6.1 Terra, max reasoning; SWE-2 Max, max reasoning): 3 finding(s).

Comment thread bench/observe-unit.py Resolved
Comment thread bench/observe-unit.py Resolved
Comment thread bench/observe-unit.py Resolved
@bompus bompus closed this Nov 6, 2026
@bompus
bompus merged commit 6a7e7ac into main Nov 6, 2026
7 checks passed
@bompus
bompus deleted the revendor-unit-observer branch November 6, 2026 22:36
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

author:anthropic Pull request written by a anthropic model

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant