Thursday, November 13, 2025
7 changes · 18.0
Resolved issues and error corrections
The system now correctly ignores the Raspberry Pi 5’s built-in serial port, which was being shown by mistake on newer Raspberry Pi OS versions. This prevents an internal device from appearing as if it were an available serial connection, reducing confusion and setup errors.
Original PR description
The serial interface previously included a filter to not include the built-in serial port on the Pi 5, however this filter is now broken in Raspberry Pi OS Trixie. To ensure the filter works in all versions, we now explicitly look for the device `/dev/ttyAMA10` and ignore it. --- I confirm I have signed the CLA and read the PR guidelines at www.odoo.com/submit-pr Forward-Port-Of: odoo/odoo#235074
This update corrects how the system reads a configuration setting used during journal setup for ISO 20022 payments. It helps ensure the right settings are available when needed, reducing the chance of errors in payment configuration.
This change makes drag-and-drop test steps finish cleanly before a test ends. It helps prevent flaky test results and makes the test suite more reliable.
Original PR description
Since drag sequences are automatically canceled at the end of tests, 'cancel' or 'drop' calls should be properly awaited before the end of a test. This commit ensures that these actions are properly finished before a test ends. Community: https://github.com/odoo/odoo/pull/235359
This change improves automated web tests by ensuring unfinished drag actions are properly canceled at the end of each test. It prevents stray asynchronous errors from appearing after tests complete, making test results more reliable and easier to trust.
Original PR description
Before this commit, when initiating an unfinished drag sequence (i.e. calling `drag` without `drop` or `cancel`), the drag sequence was ended by destroying the component, which could throw asynchronous errors after the test finished. One possible fix would be to wait for these errors upon ending a test, but this process is handled by Owl and through a possible chain of promises, which would have to be accounted for by an arbitrary amount of 'animationFrame' or 'advanceTime' calls. Instead, the `contains...drag` and `contains...dragAndDrop` now handle a single drag sequence at a time, with a systematic cleanup after each test in case a drag sequence has been left pending. In such cases, the sequence is canceled. Enterprise: https://github.com/odoo/enterprise/pull/99369 --- I confirm I have signed the CLA and read the PR guidelines at www.odoo.com/submit-pr
This change fixes an occasional test failure in the web interface by waiting for the error dialog to actually appear before checking for it. It makes automated testing more reliable and helps prevent flaky build results.
Original PR description
Before this commit, the test sometimes failed because we didn't wait enough before checking the presence of the error dialog. The `unhandledrejection` event being thrown asynchronously, simply waiting for an animation frame isn't enough. We can only wait for the dialog to be displayed. runbot error~234017 Description of the issue/feature this PR addresses: Current behavior before PR: Desired behavior after PR is merged: --- I confirm I have signed the CLA and read the PR guidelines at www.odoo.com/submit-pr
This fix makes interactive tours handle warning steps correctly when going backward. If a previous step is only meant to be skipped, the tour will no longer stop on it or briefly focus the wrong place before moving on. This makes tour navigation more reliable and avoids confusing behavior during guided flows.
Original PR description
Before this commit, the backward wasn't ignoring the warn steps. So, if the backward go to the previous step (warn's one) and the trigger is on the page, the tour interactive put the cursor there. But the step is then ignored and go back the step you came from. Now, the warn's steps are ignored. --- I confirm I have signed the CLA and read the PR guidelines at www.odoo.com/submit-pr
This change makes several editor tests more reliable by giving each test its own time allowance instead of sharing one across all of them. It helps prevent false failures on slower test servers, improving confidence in the quality of future releases.
Original PR description
Split the test timer between the four tests rather than applying
a single timer over all of them, for when the runbot is slower.
runbot-233975