openupgrade-drivers team mailing list archive

[Stefan Rijnhart <stefan@xxxxxxxx>]

On 06/17/2013 06:42 PM, Ondrej Kuznik wrote:
> Currently, none of the migration scripts have been tested beyond being
> able to upgrade an empty database and support the dependent modules
> doing the same.

Hi Ondrej,

thanks in advance for sharing your work, albeit unfinished. I am sure it 
will be of great help to other developers!

> In the process, there has been a bit of functionality missing and before
> I publish a repository, I'd like to know whether my approach to adding
> it was appropriate and where the code should belong to before I publish
> the rest as a bzr branch(es):
>
> 1. Regarding the partner_address -> partner move, the base module has
> created an extra column to facilitate the migration for everyone using
> partner_address, but there is no code to call to perform it. The code I
> came up with I as follows:
>
> def partner_address_to_partner(cr, table, field):
>      cr.execute("""
>          UPDATE %(table)s
>          SET %(field)s = a.openupgrade_7_migrated_to_partner_id
>          FROM %(table)s t
>          JOIN res_partner_address a
>          ON t.%(field)s = a.id
>          WHERE t.%(field)s is not null""" % { 'table': table,
>                                               'field': field })
>
> To make it available from other modules, given that this is v7 specific,
> where should it go and how will they be able to access it?

For version specific API, maybe have a specific module such as 
openerp/openupgrade/openupgrade70.py (just an example, I am sure someone 
can come up with a better naming convention). How does that sound to you?

> 2. There were a couple of many2one fields that have changed to
> many2many. I could not find any code in openerp.openupgrade.openupgrade
> to do this so we came up with this:
>
> def many2one2many2many(cr, model, table, field, source_field=None):
>      if source_field is None:
>          source_field = openupgrade.get_legacy_name(field)
>
>      cr.execute('SELECT id, %(field)s '
>                 'FROM %(table)s '
>                 'WHERE %(field)s is not null' % {
>                     'table': table,
>                     'field': source_field,
>                     })
>      for row in cr.fetchall():
>          model.write(cr, SUPERUSER_ID, row[0], {field: [(4, row[1])]})
>
> Do you think it should go alongside the other tools in that module?

Yes, that would be a very good addition!

Thanks,
Stefan.


--
Therp - Maatwerk in open ontwikkeling

Stefan Rijnhart - Ontwerp en implementatie

mail: stefan@xxxxxxxx
tel: +31 (0) 614478606
http://therp.nl
https://twitter.com/therp_stefan