There was an issue with the computed `read_group` `__range` when grouping on
the same date/datetime field on multiple granularities (i.e. month, week).
Since the range was stored with the field_name as a key, the last evaluated
range would override the previous ones.
Impacted Versions:
- master
(- exists since 15.0 but it does not impact the user directly so it has been
decided to fix this only in master, since the API is modified)
Steps to reproduce:
1. Open a list view and group by a date field with at least 2 granularities
2. Open the chrome debugger (network) and check a web_read_group rpc preview
3. Find the web_read_group for groups related to one of the largest
granularities and check the `__range`
Current behavior:
- `__range = {field_name: false}`
Expected behavior:
- `__range = {field_name: {from: range_start, to: range_end}`
Explanation
Since the smaller granularities are evaluated last, and the condition to update
`__range` is related to the field_name and not the granularity, the range is
always overriden by the smaller granularities (even if their value is False)
when grouping on the same field with multiple granularities.
Furthermore, there is a conceptual problem with the current solution: it does
not allow to store multiple ranges when the read_group is not lazy and when
grouping on the same field with multiple granularities.
Therefore, the proposed solution is to use the full groupby keys in the
`__range` to allow storing multiple ranges depending on granularity. The keys
in `__range` would thus match the group value keys and allow more flexibility
if a domain must be forged from the group(s) range(s).
Task-2894519
closesodoo/odoo#95193
Signed-off-by: Thibault Delavallee (tde) <tde@openerp.com>
abd-msyukyu-odoo
40dda26350