The historical ten-minute data (wind_gust_direction, wind_gust_speed, and solar) is published way earlier than all other historical data (0:50 UTC vs ~ 9:00 UTC), leaving us with an eight-hour span where we have almost-empty historical records for yesterday (UTC) and run into degraded fallback performance.
A possible work-around is to delay parsing of the ten-minute data until the hourly data is available as well. The delay should only apply to historical sources, not current / synop / forecast.
The historical ten-minute data (
wind_gust_direction
,wind_gust_speed
, andsolar
) is published way earlier than all other historical data (0:50 UTC vs ~ 9:00 UTC), leaving us with an eight-hour span where we have almost-emptyhistorical
records for yesterday (UTC) and run into degraded fallback performance.A possible work-around is to delay parsing of the ten-minute data until the hourly data is available as well. The delay should only apply to
historical
sources, notcurrent
/synop
/forecast
.See also: https://github.com/jdemaeyer/brightsky/issues/149#issuecomment-1587046924