stack/agents/filbert/work/queue-87/r6/body.md at dev-console

0 Fork 1

Files

T

jason.woltje b9ea9304fd lead: commit filbert's untracked evidence (273 files: queue-101 queue-102 queue-105 queue-107 queue-108 queue-109 queue-110 queue-112 queue-113 queue-114 queue-115 queue-116 queue-119 queue-120 queue-122 queue-123 queue-125 queue-126 queue-128 queue-129 queue-130 queue-131 queue-132 queue-133 queue-134 queue-135 queue-136 queue-137 queue-139 queue-140 queue-141 queue-142 queue-145 queue-146 queue-147 queue-148 queue-149 queue-150 queue-151 queue-152 queue-153 queue-154 queue-155 queue-156 queue-162 queue-163 queue-165 queue-166 queue-167 queue-169 queue-170 queue-175 queue-177 queue-181 queue-182 queue-184 queue-185 queue-186 queue-187 queue-188 queue-49 queue-64 queue-66 queue-70 queue-71 queue-80 queue-84 queue-85 queue-87 queue-88)

Unclaimed files left in the canonical checkout. Inventory: canonical-cleanup-inventory.md (commit-as-evidence groups). detect-secrets scan clean on 2026-11-02: hits are sha256 pins, git object ids, test sentinels already in tracked tests and unit-file lines.

2026-11-02 15:05:14 -06:00

2.6 KiB

Raw Permalink Blame History

Filbert review, row 87 round 6: approve.

Candidate: commit 80ebe2afc, diff from 232e05449 on the design docs/plans/2026-10-18_containerized-install-design.md and the rows docs/plans/2026-10-18_containerized-install-rows.md. Line numbers are at 80ebe2afc.

No test reads either file (grep -rlF containerized-install packages/*/tests scripts/test-*.sh finds nothing), so under decision 137 nothing here blocks.

My round 5 follow-ups (28881): all four answered

Name refusal. design.md:178-186: exit 3 with a message naming the container and the client's exit code, host.lock given back, and nothing removed that the store's labels didn't list. Exit 3 matches the host's other start refusals (CliError(..., 3) at packages/cli/src/host.mjs:240, :246). Tested at rows.md:489-492.
One spelling. design.md:154-166: the store path is the real path, and it's the source of the bind mount, the label, the leftover filter and hostStatus's filter. The reason it gives is correct: safe() lstats only the leaf (packages/bus/src/store.mjs:100-110), so a symlinked parent gets past it. Tested at rows.md:483-486.
The bus directory. Bus start makes <dataRoot>/bus at 0700 before it resolves the path (design.md:154-158, store.mjs:126). Tested at rows.md:487-488.
The pid tests. The four uses are listed with what each does (:202-205 /proc, :227 SIGKILL, :546-563 SIGSTOP/SIGCONT, :612 IPC lookup), and I checked each against packages/cli/tests/host.test.mjs.

The rest of the round

The entry starts each fork on its first line (boot for the broker, start for the notifier), and connected is true from when the host makes the wrapper, as a forked child's is today. That removes the round 5 wait on the {from, pid} line before the host's first send. The anchors are right: the broker's boot send at host.mjs:217, the notifier fork at :242 and its start at :244. A business without a notifier gets no notifier fork, and the one-fork-exits and exit-code rules now say "if it started" / "each fork it started" (design.md:244-250). Tested at rows.md:480-482.

Doc-only, follow-up

design.md:154 says bus start makes the directory "before anything else touches the engine". Making a directory and resolving a path don't touch the engine, and the endpoint check (the name=rootless probe, design.md:372) presumably comes first. If the order matters, name it, such as "after the endpoint check and before host.lock"; otherwise drop the clause.

Not run by me

Nothing executed. This is a read of the diff and the cited lines in host.mjs, host.test.mjs and store.mjs.

Verdict: approve, with follow-up 1.

2.6 KiB Raw Permalink Blame History

My round 5 follow-ups (28881): all four answered

The rest of the round

Doc-only, follow-up

Not run by me

2.6 KiB

Raw Permalink Blame History