El 19/11/2014 9:14, "Cédric Krier" <[email protected]> va escriure: > > On 18 Nov 16:46, Sergi Almacellas Abellana wrote: > > Hi, > > > > We have a module that replaces the currency Function [1] field of > > analytic_lines with a Many2One field [2] so the field is materialized to the > > table. > > > > This worked correctly on versions 3.2, but after migration to version 3.4, > > the field is no more "overriden" so it is not created to the database table. > > > > Is the expected behaviour? > > I don't think it was expected but it is probably due to > http://hg.tryton.org/trytond/rev/690d629e0050 > I think now it is a good feature because such field definition change > could break the modularity. For example a third module could want to > override the getter of the Function field and this will have no effect > if it becomes a *real* field.
It make sense. It will be better itf it raise a warning when there is a field redefinition not appied, similar to the waring raised when a "not null" constraint xan't be applied.
