The shared cache is designed for loop usage so we must ensure that we fill the
LRU dictionary in the same order as the loop will request attributes.
So we must order the result of the read because the API does not return data in
the same order as the list of ids.
issue9518
review318041002
Some read value may be needed to instantiate other records if it is used in the
context.
Also the test to fill fields dictionary was always false as it tested the field
instance instead of the name.
issue9380
review298171002
In order to have store them efficiently in dictionary, we generate a different
hash for each unsaved record by using the instance id.
issue9240
review295421002
By using a class with slots instead of standard dict, we can reduce a lot the
memory usage to store the record data in the caches and values.
Also we can rely on the data in the cache to exist only for real field which
can be used to create a fast path in ModelStorage.__getattr__.
As now storing data in the cache has no extra cost, we can store data from the
transactional cache in the local. This allow also to have a shorter path in
ModelStorage.__getattr__.
For more memory efficiency, we remove the record from the cache instead of
clearing them. It was better to clear a dict but a record is less expensive to
re-instantiate than clearing.
issue9369
issue9225
review297861002
When a relation domain is not valid, the domain used to set on the exception
must be evaluated with the origin record and not the invalid target record.
We take the first origin record as any of the passed record will generate the
same domain per design.
issue9383
review317531002
If the record is unsaved and the getter is based on on_change_with, we can call
the getter to compute the value.
This value must not be stored in the cache as it may change prior to saving.
issue4193
issue9263
review317291002
This ensures that internal changes that are reverted later in the same
transaction does not trigger an action.
Also we ensure that a record is triggered by the same trigger only
once per transaction.
issue8342
review280971002
We add the flake8 configuration used so we ensure everyone uses the same.
We remove the usage of __all__ for non public API.
When possible, we rationalize the class name according to its __name__ and module.
issue9082
review297061002
On distributed setup, the cache is not synchronized instantaneously so we can
not use the cache if the record was modified by another process since the last
synchronisation.
issue9013
review262841002
It is wrong to instantiate the cache in the __setup__ because it can be run
multiple time in the same process (but with new pooled classes).
issue9019
review253011003
We sent an order to the client for action and xxx2Many so its screen has
always a default order set.
To position a new record on non editable tree, we use the order of the 'id'
column if presents.
issue8880
review276501003
The Many2One and Reference fields are stored in global cache. If they are
already in the cache, the other eager fields should also be in the cache and it
is highly probable that no read will be needed. So it is better to skip the
eager loading.
issue8863
review272541003
Databases are too random about rounding when casting epoch into numerical
value. So it is better to check against inequality of the string.
issue8779
review254571002