[IMP] rename_models: rename the models in the reference fields

[MiquelRForgeFlow]

This new method renames the models in the reference fields.

Also fixes _change_reference_refs_sql method.

[pedrobaeza]

Why an extra method instead of being part of existing rename_models? All should be done in the same method IMO.

[pedrobaeza]

Please put the other fix in a separate PR

[MiquelRForgeFlow]

Why an extra method instead of being part of existing rename_models?

Because the need of the env. rename_models doesn't use env but the new method needs it.

Maybe we can expand rename_models signature like:

def rename_models(cr, model_spec, env=False):

what do you think?

[pedrobaeza]

But I think we can avoid the use of the env. Let's search directly on ir_model_fields table possible fields of ttype='reference', and then perform the SQL with the change.

[MiquelRForgeFlow]

But how I obtain the table?

[pedrobaeza]

Yeah, good point... Now you know why I always want to pass env although no initial use. I don't like too much the option of adding that extra "optional" argument. Or at least not only. It's better to build a new environment for cr (we can do it the same as we do on migrate decorator) if not provided via that argument.

[MiquelRForgeFlow]

I think this method should remain separated and not merged into the rename_models method.
Reason: rename_models is called in pre-migration and covers all cases without using the env. Instead, rename_models_reference should be called in end-migration, because it's only in end-migration when all models are in the env, and thus, only in end-migration will you obtain any model table, by the same way that merge_records usually is called in end-migration.

[MiquelRForgeFlow]

I have updated the PR. Now the rename_models method make use of it. And also it can be called independently in the end-migration with the environment.

[MiquelRForgeFlow]

instead of being part of existing rename_models? All should be done in the same method IMO.

Now it's handled this way. Very simple.