delight [11:16 AM]
technically speaking we used to load each encounter (expensive) once, then add that data to user and area totals. thus we carried user and area totals from start to finish.
[11:20]
now we start on first area, do all encounters in that area, forget about it and move to the next, then do the same for participations. this way we don’t need to remember everything, because each pass is self-contained. however, there’s up to ten participants in an encounter, so we have to load each encounter up to 11 times (once for area, once for each participant). and remember loading the encounter is the time-guzzler.
[11:22]
tl;di: fixed a server crash but as a tradeoff made aggregate statistics update less frequently?
merforga
commented
6 years ago
Something about pee
delight [11:16 AM] technically speaking we used to load each encounter (expensive) once, then add that data to user and area totals. thus we carried user and area totals from start to finish.
[11:20] now we start on first area, do all encounters in that area, forget about it and move to the next, then do the same for participations. this way we don’t need to remember everything, because each pass is self-contained. however, there’s up to ten participants in an encounter, so we have to load each encounter up to 11 times (once for area, once for each participant). and remember loading the encounter is the time-guzzler.
[11:22] tl;di: fixed a server crash but as a tradeoff made aggregate statistics update less frequently?