Note that the website already does this sort of time zone conversion in several places, like timestamps on when games start/end, chat messages, tournament start times
1 Like
That is exactly what I had in mind.
I think the existing conversions are all from UTC to local, which is relatively easy.
Re-localizing a local recurring event like “2:30am Pacific time” is more of a challenge that common timezone libraries usually avoid. For example, what do you do on the day in spring when that time doesn’t exist? Or in the fall, when it happens twice?
If we do it right, though, this would be a great feature!
3 Likes
I think that the backend for various OGS functionalities is just handled all in UTC, for things like the regularly scheduled tournaments, the beginning/end of weekend pauses.
One wrinkle is that some regions observe daylight savings time, while others do not, and the beginning/end of these periods vary around the world. One way to handle this is to just fix the pause schedule according to UTC at the back end (while still translating to local time for user views), which would mean that it would shift (relative to local time) during DST transitions, in regions that observe these.
7 Likes
So the test tournament we nominally held in “Atlantic time zones” finished a little while back: Fast correspondence - Atlantic time zones
The tournament ran with Fischer time, 6 hours / move. There were 13 participants, 3 disqualified for timeout, 1 player deleted their account (hopefully not because of the tournament :)).
Some stats:
- 16 total games, 1611 moves
- 10th percentile move 40 seconds, median 42 minutes, 90th percentile 10 hours
- Per-game maximum move times:
['69.4h', '5.2h', '13.4h', '8.7h', '21.9h', '4.2h', '0.8h', '20.5h', '104.8h', '12.8h', '97.9h', '42.9h', '75.9h', '206.5h', '13.7h', '53.3h'] - Per-game 90th percentile move times:
['14.7h', '5.2h', '5.7h', '3.9h', '7.2h', '4.2h', '0.3h', '1.3h', '50.7h', '7.2h', '18.3h', '23.9h', '9.7h', '21.2h', '4.0h', '22.3h'] - Per-game average move times:
['5.2h', '5.2h', '1.4h', '1.5h', '2.4h', '4.2h', '0.1h', '0.7h', '21.7h', '1.9h', '7.1h', '9.7h', '4.9h', '11.4h', '1.4h', '7.3h']
The games ending in timeout were those with index 5, 8 and 11 (average move times 4.2h, 21.7h and 9.7h). I think the 11.4h one was one the ones that was using a bunch of vacation that to not have timed out.
IDK how big a difference it made that “please be somewhat close to the Atlantic ocean” was explicit in the tournament title and text, but IMO the 6h/move didn’t work poorly at all. Looking at the move timings, I think 4h/move would have worked for most of these pairings too.