mirror of
https://github.com/odoo/runbot.git
synced 2025-03-15 23:45:44 +07:00
b177361f20
Not sure why I didn't think about it previously (in #357), but it would make sense for the entire batch creation to be atomic and thus under the same lock. The commit during forward porting also makes a lot less sense: it was a failed early attempt at resolving the problem by hoping we'd win the race with the webhook (commit before the webhook hit). By locking the PRs table to update, we actually resolved it. But since all that happens then is a few updates and then a commit by the cron itself (it commits per batch), it's probably good enough to leave the entire thing under the same lock. This means we lock out other interactions a bit longer, but since the span is still just the forward port of a single batch it should not be too much of an issue outside of post-freeze recovery thingie. |
||
|---|---|---|
| .. | ||
| __init__.py | ||
| forwardport.py | ||
| project_freeze.py | ||
| project.py | ||