Beam YAML 1d: align the day label with an actual 86,400-second window
A fixed UTC day requires 86,400 seconds; a configuration label alone does not establish that duration.
Before calling it a day
Reading a configuration string is different from checking the executed window duration. A fixed UTC day requires comparing the actual start-to-end difference with 86,400 seconds. Give Truthound runner-produced boundaries rather than a label. Showing which events enter each window in the first demonstration makes the unit issue visible to operators. [1, 2, 3]
Two distinct YAML findings
Beam YAML #40902 reports a d suffix multiplier of 12 hours instead of 24 hours and a separate duplicate-field check. This article addresses only the fixed-day duration. Passing that check does not establish a repair of duplicate-field handling. It analyzes the code report without claiming a corrected run in an installed runner. [1]
Show boundaries and events together
Record the fixed UTC 24-hour requirement, then extract window id, start and end from the runner output. Display those boundaries alongside test events and withhold a 43,200-second window from publication as a daily aggregate. Equal durations still require separate checks of start alignment and boundary membership. [1, 3]
A calculation that exposes half a day
The illustrative contract is exactly 24 hours in UTC: 24×60×60=86,400 seconds. A 12×60×60=43,200-second window is half that duration. An observed 86,400 satisfies the length condition; 43,200 fails and holds the daily label and aggregate. This is a defined comparison, not a Beam execution result, and does not use a local-calendar or daylight-saving definition. [1, 3]
The check between settings and results
Truthound Core Python SDK 3.3.2 can compare prepared window_duration_seconds and fixed_day_seconds with ColumnPairValidator using relationship="==". The operator extracts boundaries, converts units and separates missing observations. The SDK does not repair the YAML parser or define local calendar days. [1, 2, 3]
Repeat the comparison with explicit units
An alternative is a controlled run using verified explicit seconds and the same events, followed by checking the exact version carrying the upstream change. If the requirement is the calendar day between local midnights, daylight saving may make 86,400 seconds inappropriate. Define that timezone-specific calendar contract first. [1, 3]
Application conditions
Declare a fixed UTC 24-hour contract and extract actual window boundaries from the runner result.
Alternatives
Compare a verified explicit-seconds setting and confirm the version carrying the upstream change.
Limitations
Truthound does not repair the YAML parser or define every local calendar day as 86,400 seconds.
Evidence
- Event checked
- Subject checked
- Modified
