Florian CharlierandJulien Castiaux
beaae001f8
and better handle progress requests. In previous versions, it sometimes occurred that multiple onboarding_progress records were created despite our measures against that. Two problems were identified and fixed here, together with a simple mean to handle this error on the client-side. 1. `_sql_constraints` entry was not adequate to prevent multiple records from being created for onboardings which are not to be completed per-company but per database because in our version(s) of PostgreSQL `NULL` values cannot be considered `NOT DISTINCT`, i.e., all NULL values are different from one another so `UNIQUE` is respected when we wouldn't want it. Therefore, we implement here a `UNIQUE` index via the model's init, allowing to catch unique violations when company_id is not set. 2. While we search if a record exists before creating a new one, it could happen that another worker created a record between these steps. As the database is configured (`REPEATABLE_READ`, which applies until the end of the transaction, when we leave the controller), it is very difficult to retrieve the record created by the other process. See additional details in the testing part below. Therefore, we chose to ask the client to perform a new request if desirable. Existing extra records are removed with the related UPG PR script. ### Tests Tests are included to make sure it is not possible to create multiple onboarding_progress records for the same onboarding and company or without any. This includes a non-standard (runbot will not run it) and "database breaking" (we try to clean up our mess but we cannot give any guarantee) concurrent test that reproduce the following scenario: The `/onboarding/<string:route_name>` is called twice in two different workers for a same `route_name`. The corresponding onboarding exists and has no corresponding `onboarding.progress` record yet. Python side, the two workers are oblivious of what's happening in the transaction of the other worker thus any python-side-only unicity constraint would fail to detect any progress created by other workers. This would end up with two progress records created for a single onboarding which is illegal. Task-3101666 closes odoo/odoo#109221 Related: odoo/upgrade#4142 Signed-off-by: Thibault Delavallee (tde) <tde@openerp.com> Co-authored-by: Julien Castiaux <juc@odoo.com>
77 lines
3.4 KiB
Python
77 lines
3.4 KiB
Python
# -*- coding: utf-8 -*-
|
|
# Part of Odoo. See LICENSE file for full copyright and licensing details.
|
|
|
|
from odoo import api, fields, models
|
|
|
|
|
|
ONBOARDING_PROGRESS_STATES = [
|
|
('not_done', 'Not done'),
|
|
('just_done', 'Just done'),
|
|
('done', 'Done'),
|
|
]
|
|
|
|
|
|
class OnboardingProgress(models.Model):
|
|
_name = 'onboarding.progress'
|
|
_description = 'Onboarding Progress Tracker'
|
|
_rec_name = 'onboarding_id'
|
|
|
|
onboarding_state = fields.Selection(
|
|
ONBOARDING_PROGRESS_STATES, string='Onboarding progress', compute='_compute_onboarding_state', store=True)
|
|
is_onboarding_closed = fields.Boolean('Was panel closed?')
|
|
company_id = fields.Many2one('res.company')
|
|
onboarding_id = fields.Many2one(
|
|
'onboarding.onboarding', 'Related onboarding tracked', required=True, ondelete='cascade')
|
|
progress_step_ids = fields.One2many('onboarding.progress.step', 'progress_id', 'Progress Steps Trackers')
|
|
|
|
def init(self):
|
|
# not in _sql_constraint because COALESCE is not supported for PostgreSQL constraint
|
|
self.env.cr.execute("""
|
|
CREATE UNIQUE INDEX IF NOT EXISTS onboarding_progress_onboarding_company_uniq
|
|
ON onboarding_progress (onboarding_id, COALESCE(company_id, 0))
|
|
""")
|
|
|
|
@api.depends('onboarding_id.step_ids', 'progress_step_ids', 'progress_step_ids.step_state')
|
|
def _compute_onboarding_state(self):
|
|
progress_steps_data = self.env['onboarding.progress.step'].read_group(
|
|
[('progress_id', 'in', self.ids), ('step_state', 'in', ['just_done', 'done'])],
|
|
['progress_id'], ['progress_id']
|
|
)
|
|
result = dict((data['progress_id'][0], data['progress_id_count']) for data in progress_steps_data)
|
|
for progress in self:
|
|
progress.onboarding_state = (
|
|
'not_done' if result.get(progress.id, 0) != len(progress.onboarding_id.step_ids)
|
|
else 'done')
|
|
|
|
def action_close(self):
|
|
self.is_onboarding_closed = True
|
|
|
|
def action_toggle_visibility(self):
|
|
for progress in self:
|
|
progress.is_onboarding_closed = not progress.is_onboarding_closed
|
|
|
|
def _get_and_update_onboarding_state(self):
|
|
"""Used to fetch the progress of an onboarding for rendering its panel and is expected to
|
|
be called by the onboarding controller. It also has the responsibility of updating the
|
|
'just_done' states into 'done' so that the 'just_done' states are only rendered once.
|
|
"""
|
|
self.ensure_one()
|
|
onboarding_states_values = {}
|
|
progress_steps_to_consolidate = self.env['onboarding.progress.step']
|
|
|
|
# Iterate over onboarding step_ids and not self.progress_step_ids because 'not_done' steps
|
|
# may not have a progress_step record.
|
|
for step in self.onboarding_id.step_ids:
|
|
step_state = step.current_step_state
|
|
if step_state == 'just_done':
|
|
progress_steps_to_consolidate |= step.current_progress_step_id
|
|
onboarding_states_values[step.id] = step_state
|
|
|
|
progress_steps_to_consolidate.action_consolidate_just_done()
|
|
|
|
if self.is_onboarding_closed:
|
|
onboarding_states_values['onboarding_state'] = 'closed'
|
|
elif self.onboarding_state == 'done':
|
|
onboarding_states_values['onboarding_state'] = 'just_done' if progress_steps_to_consolidate else 'done'
|
|
return onboarding_states_values
|