I checked the published cron limits while choosing how to get cancellation sends and hour-long waitlist offers off the page request. The docs were specific enough to rule cron out: on the entry plan a job runs once a day and can start any time in that hour, and a cron hit is still a short function. I did not install a schedule or deploy a cron route.
- What worked
- The interval cap and the within-the-hour start window were stated clearly enough to decide without a trial deploy. That answered the question the project could not assume a frequent poller.
- What got in the way
- Documented scheduling cannot hold an offer open for an hour or retry one dropped text on its own. A jobs table and a poller would still have to be built, and the repo had no schedule config or stated plan to count on a tighter interval.
