It has been a nightmare to get along with the duration "month" and "year" as their absolute time interval is variable, unlike "day", "hour", etc. (issue #838, #511, #530)
The attempt to fix issue #838 (2bc11f1) seems a promising approach to deal with the variable durations when an absolute datetime is specified as a reference point, which is always the case for SpineOpt models.
It feels that the approach could also help in the rolling-related issues (#511, #530)?
It has been a nightmare to get along with the duration "month" and "year" as their absolute time interval is variable, unlike "day", "hour", etc. (issue #838, #511, #530)
The attempt to fix issue #838 (2bc11f1) seems a promising approach to deal with the variable durations when an absolute datetime is specified as a reference point, which is always the case for SpineOpt models.
It feels that the approach could also help in the rolling-related issues (#511, #530)?