This simplifies the use of subqueries by avoiding some costly default
order on the model or the idiotic order='id'. Method _flush_search()
has been adapted accordingly.
Part-of: odoo/odoo#112126
Let's say there is one account with code 431 and another with code 4310,
`ODOO.BALANCE("431,4310", 2022)` counts 4310 twice.
The reason is queries are done one code at a time, one by one.
The account 4310 is counted first when using 431 as a prefix (431%) and once
more with 4310 (4310%).
Now, the query uses all codes all at once.
The original idea was to cache queries as much as we could. If you later add the
formula `ODOO.BALANCE("431", 2022)`, the result would have been cached
because queries were done code by code.
Given the issue it brings, we thing this opmitisation is not worth it and is
therefore removed. It also makes the code client-side simpler.
opw-3144473
closesodoo/odoo#113418
X-original-commit: 07bd67f3a389bde4f33e08d97967dbb70ad5a19f
Signed-off-by: Pierre Rousseau (pro) <pro@odoo.com>
Signed-off-by: Lucas Lefèvre (lul) <lul@odoo.com>
Rémy Voet (ryv)