On 26/01/12 13:10 +0100, Bertrand Chenal wrote:
> Le Thu, 26 Jan 2012 11:28:21 +0100,
> Cédric Krier <[email protected]> a écrit :
> 
> > On 25/01/12 14:43 +0100, Bertrand Chenal wrote:
> > > Le Wed, 25 Jan 2012 13:45:25 +0100,
> > > Cédric Krier <[email protected]> a écrit :
> > > >   Also I think such behavior will be very resource consuming.
> > > 
> > > I'm not sure but I think it's just a matter of changing 
> > > 
> > >  GROUP BY to_location, product
> > > 
> > > to
> > > 
> > >  GROUP BY to_location, product, lot
> > > 
> > > in the products_by_location method, except that it must be done
> > > through inheritance. 
> > 
> > Indeed it will be ressource consuming because the stock period works
> > at product level and not at lot.
> > And I don't think caching per lot is a smart design choise, the number
> > of row in such table will increase very fast.
> > 
> 
> I agree that caching per location and per lot is not a good idea.
> So, new proposal:
> 
> a naive product_by_lot function will be slow, so we need caching. As
> lots are a bit like temporary locations,

If so, why not use location instead of lot?

> caching per period makes no
> sense. But we can flag 'empty' lots, I.E. lots for which all products
> have been sent to customers (or consumed by a production). While the
> number of lot gets bigger and bigger as the time goes, the number of
> non-empty lot will stay nearly stable.
> 
> This means that an efficient product_by_lot is doable: we only
> have to consider the moves that are linked to an non-empty lot.
> And for the same price we can do a "product_and_location_by_lot".
> 
> Now, the question of how to integrate this with shipments and
> assignment will depends of the use case.

-- 
Cédric Krier

B2CK SPRL
Rue de Rotterdam, 4
4000 Liège
Belgium
Tel: +32 472 54 46 59
Email/Jabber: [email protected]
Website: http://www.b2ck.com/

Attachment: pgpOrBqgs5ZaR.pgp
Description: PGP signature

Reply via email to