Filed unassigned and ungraded by the #12783 dev (session session_01PfaSTikked61BkcsB5Rn69). Observation-class polish, recorded rather than folded into that card's diff.
Measured
scripts/pm/os-verify-lock.sh:
human_s() {
local s="$1"
if ((s >= 60)); then printf '%ss (%dm%02ds)' "$s" $((s / 60)) $((s % 60)); else printf '%ss' "$s"; fi
}There is no hour branch, so every duration past an hour is rendered in minutes. Read from the live --report on 2026-08-28:
records: 75, spanning 30245s (504m05s)
⇒ it reaches back to its first record and no further: 2026-08-27T16:51:12Z, 30841s (514m01s) ago.
504m05s is 8h24m. 514m01s is 8h34m.
Why it matters
Mild everywhere it was used before, and slightly less mild now. Until PR #12827 the minute-scale figures this formatter produced were mostly hold and wait times, where minutes are the natural unit and the values are small (the long-hold warning fires at 900s = 15m00s, which reads fine).
The two call sites above are different: they are the population's width and floor, and the question a reader brings to them is a calendar question — "does this cover my 19-minute hold from the 26th?" Answering it from 514m01s needs a division the report is otherwise careful to hand over rather than leave as arithmetic (cf. the arrival-depth block, which prints ⇒ waiters already ahead = this minus 1 precisely so nobody has to do it).
⚠️ This is cosmetic, not a correctness defect: the seconds figure is printed alongside and is exact. Nothing is wrong, it is just harder to read than it needs to be.
Candidate shapes (⛔ suggestions, not a ruling)
- Add an hour branch to
human_s past some threshold — one place, every call site benefits. ⚠️ It changes the text of several existing outputs, so the pin tests that quote them need re-reading; the change is presentation-only but it is not zero-diff.
- Leave
human_s alone and let only the two population figures carry an hour rendering. Smaller blast radius, at the cost of two duration formats in one report.
⛔ Note the file has a bash 3.2 floor (pnpm check:bash32-floor), so whichever shape is chosen stays inside plain arithmetic and printf.
Refs
Filed unassigned and ungraded by the #12783 dev (session
session_01PfaSTikked61BkcsB5Rn69). Observation-class polish, recorded rather than folded into that card's diff.Measured
scripts/pm/os-verify-lock.sh:There is no hour branch, so every duration past an hour is rendered in minutes. Read from the live
--reporton 2026-08-28:504m05sis 8h24m.514m01sis 8h34m.Why it matters
Mild everywhere it was used before, and slightly less mild now. Until PR #12827 the minute-scale figures this formatter produced were mostly hold and wait times, where minutes are the natural unit and the values are small (the long-hold warning fires at 900s =
15m00s, which reads fine).The two call sites above are different: they are the population's width and floor, and the question a reader brings to them is a calendar question — "does this cover my 19-minute hold from the 26th?" Answering it from
514m01sneeds a division the report is otherwise careful to hand over rather than leave as arithmetic (cf. the arrival-depth block, which prints⇒ waiters already ahead = this minus 1precisely so nobody has to do it).Candidate shapes (⛔ suggestions, not a ruling)
human_spast some threshold — one place, every call site benefits.human_salone and let only the two population figures carry an hour rendering. Smaller blast radius, at the cost of two duration formats in one report.⛔ Note the file has a bash 3.2 floor (
pnpm check:bash32-floor), so whichever shape is chosen stays inside plain arithmetic andprintf.Refs
os-verify-lockledger lives in the container's /tmp, so--reportcan never answer the question it was built for — "where does hold time go" is scoped to one shift, and says so nowhere #12783 / PR fix(pm): makeos-verify-lock --reportstate the population it is computed over #12827 — where the two calendar-scale call sites came from