wiki "Migration to version 19.0" - [REM] core: odoo.fields.first

[chienandalu]

odoo.fields.first is removed in [REM] core: odoo.fields.first

An alternative, as proposed in the same commit would be to use next(iter(. For example: fields.first(recordset) would be turned into next(iter(recordset), recordset). (note the fallback parameter in case the recordset is empty that will return the proper empty recordset).

[pedrobaeza]

But I would say records[0] now keeps the prefetch, isn't it?

[chienandalu]

I think the idea is to have the former ergonomy of the method. Toy example:

>>> users = env["res.users"].browse()

>>> users[0]
Traceback (most recent call last):
  File "<console>", line 1, in <module>
  File "/opt/odoo/custom/src/odoo/odoo/orm/models.py", line 6690, in __getitem__
    return self.browse((self._ids[key],))
                        ~~~~~~~~~^^^^^
IndexError: tuple index out of range


>>> next(iter(users), users)
res.users()

[pedrobaeza]

Then do records[:1].

[chienandalu]

Prefetch wise, these are some experiments I've run. In a nutshell, the main difference is that with next(iter we prefetch the whole recordset, while with [:1] only the record we're dumping into the variable... so the conclussion would be... use at your own conveniance 😅

>>> recs = env['res.partner'].search([], limit=10)
>>> a = recs[:1]
>>> print(a._prefetch_ids == recs._prefetch_ids)
False


>>> recs = env['res.partner'].search([], limit=10)
>>> b = next(iter(recs), recs)
>>> print(b._prefetch_ids == recs._prefetch_ids)
True

>>> env.invalidate_all()
>>> recs = env['res.partner'].search([], limit=10)
>>> recs[:1].name   # trigger prefetch
'Acme Corporation'
>>> all(
...     env.cache.contains(rec, rec._fields['name'])
...     for rec in recs
... )
False


>>> env.invalidate_all()
>>> recs = env['res.partner'].search([], limit=10)
>>> next(iter(recs)).name   # trigger prefetch
'Acme Corporation'
>>> all(
...     env.cache.contains(rec, rec._fields['name'])
...     for rec in recs
... )
True