Signed-off-by: Carmen Bianca BAKKER <[email protected]>

…calendar

This is the bones of the implementation.

Signed-off-by: Carmen Bianca BAKKER <[email protected]>

Signed-off-by: Carmen Bianca BAKKER <[email protected]>

…atch

This is the real implementation work. With this method implemented, all
other methods correctly get the correct week each time.

Signed-off-by: Carmen Bianca BAKKER <[email protected]>

The epoch date is hidden on the child anyway. Let's just hide it, and
always make sure to get the parent's epoch date. This gets rid of the
complicated computation stuff that won't backport well to v12.

Signed-off-by: Carmen Bianca BAKKER <[email protected]>

…s attendances

The idea here is that the children contain all the logic/attendances,
and the parent is just a holder of children.

Signed-off-by: Carmen Bianca BAKKER <[email protected]>

Signed-off-by: Carmen Bianca BAKKER <[email protected]>

Signed-off-by: Carmen Bianca BAKKER <[email protected]>

Signed-off-by: Carmen Bianca BAKKER <[email protected]>

I need this elsewhere. It _technically_ reduces performance by doing the
same calculation twice, but this module is horrible as pertains to
performance in any case.

Signed-off-by: Carmen Bianca BAKKER <[email protected]>

- `parent_calendar_id` is now `ondelete="cascade"`.
- Renamed `family_calendar_ids` to `multi_week_calendar_ids`.
- Renamed `current_calendar_id` to `current_multi_week_calendar_id`.
- Renamed `_get_calendar` to `_get_multi_week_calendar`.

Signed-off-by: Carmen Bianca BAKKER <[email protected]>

Signed-off-by: Carmen Bianca BAKKER <[email protected]>

- Improved the comment on how week_sequence works.
- Renamed family_size to calendar_count
- Added a comment on _get_multi_week_calendar() returning a 1-item
  recordset.
- Re-optimised the current week calculation.

Signed-off-by: Carmen Bianca BAKKER <[email protected]>

…ndar

Signed-off-by: Carmen Bianca BAKKER <[email protected]>