Most iCal polling is wasted work
An unchanged calendar still costs a full HTTP round trip and a parse. Conditional requests turn almost all of that traffic into a 304 — and leave only the part that is worth acting on.
- ical
- polling
- architecture
Blog
What we learn running iCal ingestion against other people's calendars: the traffic that turns out to be wasted, the failures worth waiting out, and the ones that are not.
An unchanged calendar still costs a full HTTP round trip and a parse. Conditional requests turn almost all of that traffic into a 304 — and leave only the part that is worth acting on.
A feed larger than the body limit returns the same oversized file on every request. No retry will fix it, and that is a different problem from a poll that failed — so it gets a different webhook.