stock_lot: Add lots traceability

issue10792

[reviewbot, 2021-09-24 13:53:16]

flake8 OK
URL: https://codereview.tryton.org/381061002

[pokoli, 2021-09-29 09:57:53]

Add tests and some fixes

[reviewbot, 2021-09-29 10:03:59]

flake8 OK
URL: https://codereview.tryton.org/381061002

[albert, 2021-10-19 10:45:06]

https://codereview.tryton.org/381061002/diff/350701002/stock.py
File stock.py (right):

https://codereview.tryton.org/381061002/diff/350701002/stock.py#newcode763
stock.py:763: source.id, target.id, source.production_input))
I think it should be grouped by source.lot and target.lot instead of source.id
and target.id

[pokoli, 2021-10-20 15:58:33]

Group by lots

[reviewbot, 2021-10-20 15:59:43]

flake8 OK
URL: https://codereview.tryton.org/381061002

[pokoli, 2021-11-02 09:44:49]

Update to tip

[reviewbot, 2021-11-02 09:49:24]

flake8 OK
URL: https://codereview.tryton.org/381061002

[ced, 2021-11-21 22:12:10]

https://codereview.tryton.org/381061002/diff/355711002/stock.py
File stock.py (right):

https://codereview.tryton.org/381061002/diff/355711002/stock.py#newcode663
stock.py:663: target = fields.Many2One('stock.lot', "Target")
I'm wondering if it is not better to use up and down.
At least I find "target" odd.

https://codereview.tryton.org/381061002/diff/355711002/stock.py#newcode688
stock.py:688: 'get_related_lots')
Why using function fields instead of Many2Many field?

https://codereview.tryton.org/381061002/diff/355711002/stock.py#newcode701
stock.py:701: return [x.id for x in self.search([
It is not good for performance to make search query in instance getter.

https://codereview.tryton.org/381061002/diff/355711002/stock.py#newcode708
stock.py:708: "Lot Empty Link"
I do not really understand the point of this.

I'm wondering if we should not have the all feature only depending on production
module.

https://codereview.tryton.org/381061002/diff/355711002/stock.py#newcode758
stock.py:758: sql_pairing(Min(source.id), Min(target.id)).as_('id'),
I would prefer to avoid using sql_pairing on big table like stock.move.

I'm wondering if we should not make a table_query that depend on a lot in
context. This could probably reduce the complexity of building an unique id.

[pokoli, 2021-12-14 16:50:00]

Fix remarks

[pokoli, 2021-12-14 16:50:05]

https://codereview.tryton.org/381061002/diff/355711002/stock.py
File stock.py (right):

https://codereview.tryton.org/381061002/diff/355711002/stock.py#newcode663
stock.py:663: target = fields.Many2One('stock.lot', "Target")
On 2021/11/21 22:12:10, ced wrote:
> I'm wondering if it is not better to use up and down.
> At least I find "target" odd.

Done.

https://codereview.tryton.org/381061002/diff/355711002/stock.py#newcode688
stock.py:688: 'get_related_lots')
On 2021/11/21 22:12:10, ced wrote:
> Why using function fields instead of Many2Many field?

Because there is no link between stock.lot.link records. The unique link is the
source and the target field and this forces the usage of the table query.

https://codereview.tryton.org/381061002/diff/355711002/stock.py#newcode701
stock.py:701: return [x.id for x in self.search([
On 2021/11/21 22:12:10, ced wrote:
> It is not good for performance to make search query in instance getter.

How can do it better? Using an sql query is not possible because there is no
real table.

https://codereview.tryton.org/381061002/diff/355711002/stock.py#newcode708
stock.py:708: "Lot Empty Link"
On 2021/11/21 22:12:10, ced wrote:
> I do not really understand the point of this.
> 
> I'm wondering if we should not have the all feature only depending on
production
> module.

Ok, I removed the empty link and just make the code dependent on the production
module.

https://codereview.tryton.org/381061002/diff/355711002/stock.py#newcode758
stock.py:758: sql_pairing(Min(source.id), Min(target.id)).as_('id'),
On 2021/11/21 22:12:09, ced wrote:
> I would prefer to avoid using sql_pairing on big table like stock.move.
> 
> I'm wondering if we should not make a table_query that depend on a lot in
> context. This could probably reduce the complexity of building an unique id.

Done.

[reviewbot, 2021-12-14 17:00:42]

checks OK
URL: https://codereview.tryton.org/381061002

[ced, 2021-12-14 17:08:37]

https://codereview.tryton.org/381061002/diff/355711002/stock.py
File stock.py (right):

https://codereview.tryton.org/381061002/diff/355711002/stock.py#newcode688
stock.py:688: 'get_related_lots')
On 2021/12/14 16:50:05, pokoli wrote:
> On 2021/11/21 22:12:10, ced wrote:
> > Why using function fields instead of Many2Many field?
> 
> Because there is no link between stock.lot.link records. The unique link is
the
> source and the target field and this forces the usage of the table query.

I guess it is possible to create a Model with a table_query which can be used as
Many2Many relation table.

https://codereview.tryton.org/381061002/diff/355711002/stock.py#newcode701
stock.py:701: return [x.id for x in self.search([
On 2021/12/14 16:50:05, pokoli wrote:
> On 2021/11/21 22:12:10, ced wrote:
> > It is not good for performance to make search query in instance getter.
> 
> How can do it better? Using an sql query is not possible because there is no
> real table. 

You could define a table_query.