Merge lp://staging/~camptocamp/account-financial-tools/7.0-add-account_move_batch_validate-lep into lp://staging/~account-core-editors/account-financial-tools/7.0

Just about these lines:

135 + session_hdl = ConnectorSessionHandler(cr.dbname, uid)
136 + with session_hdl.session() as session:

session_hdl.session() will open a new transaction (new cr), and commit the jobs at the end of the context manager.
I'm not sure that's what you want.

My opinion is that if something fails in the main transaction of the wizard, the jobs should not be created. In that case, you would need to create a Session with the current cursor (instead of creating a new cursor).

    session = Session(cr, uid, context=context)

Guewen,

that means that the session is the one we use to create the jobs, not the one the jobs use? In that case, you're right, creating the jobs is in the same transaction as the wizard.

Thanks!

On 01/13/2014 11:55 AM, Leonardo Pistone @ camptocamp wrote:
> Guewen,
>
> that means that the session is the one we use to create the jobs, not the one the jobs use? In that case, you're right, creating the jobs is in the same transaction as the wizard.
>
> Thanks!
>

The session here is the one used for the INSERT of the job. A fresh
session will be created to execute the job, but you don't have to bother
with that.

Guewen,

for the session: ok, fixed.

Another thing: how do you suggest to test that? Maybe as you do in connector/tests/test_job.py? Or maybe on a yaml test I create the wizard and run it, and then check that it created a job?

I suppose I cannot simply wait for the job to be done in the test, I need something better than that )

thanks a lot

Implemented the ability to stop jobs in bulk, by unmarking the lines (same wizard, action='unmark').

Added two tests for the two workflows (mark and post normally, and mark and change your mind).

The date/period filters are not there yet, but I am asking for a review anyway (it should mark/unmark all moves, without filters)

Thanks!

Note: the date and period filters now work.

Hi,

Thanks for the contribs ! A few remarks of mine:

 * In import, use full path : from . import blabla (line 28, 29, 399)
 * In __openerp__.py:
   * please update the tag (update_xml, init_xml, ... are obsolete) use "data" instead
   * In the description, I will very much appreciate you explain here why to use this module , why it has been made so the user can understand what it bring compared to standard OpenERP
 * Add a i18n folder and the .pot file for translation

Otherwise LGTM, thanks for providing tests with it !

joel, done, thanks for reviewing

Test Ok on my side.

Just one comment : the current 'post entried' menu in periodical processing must be hidden if this module is installed

Thanks

Hi,

Thanks for the corrections, LGTM now !

Regards,

Joël

I checked about would happen in a multi-company environment where it is not OK that 'admin' does anything directly.

It turns out that my module does the post with the user taken from the session, which is the one that ran the wizard in the first place. Does that sound correct to you Frédéric?

thanks!

sounds perfect
thanks :)

ok!

Then I 'abstain' to lift my own 'needs information'.

After a suggestion from Frédéric, I hid the old menuitem.

thanks

LGTM

Hi all,

thanks for you reviews, but another question came up just now.

Look at line 181:
i write the job UUID on the move, and then choose only the moves that have no UUIDs. Advantage: we never create many jobs for posting the same move.

Do you think this could be dangerous? Maybe there is a case where we actually need to create a new job to post the same move?

thanks!

I added the last 4 commits after an internal functional review.

It is just a little fix + translations added + view tweaks, so your reviews are not wasted!

thanks

LGTM Thanks !