[ZEPPELIN-6536] Stabilize notebook keyboard shortcut e2e editor seeding#5304
[ZEPPELIN-6536] Stabilize notebook keyboard shortcut e2e editor seeding#5304voidmatcha wants to merge 3 commits into
Conversation
|
Ran the keyboard-shortcut spec 10x in parallel on top of this PR's head (690a62b), all browser projects in anonymous mode, retries=0. All 10 passed: https://github.com/voidmatcha/zeppelin/actions/runs/29462418270 The verification branch is the PR head plus a temporary 10x matrix workflow, nothing else. Auth mode is covered by the regular CI run on the same commit, where both playwright jobs passed: https://github.com/voidmatcha/zeppelin/actions/runs/29351333647 |
519c453 to
039c686
Compare
|
Added 039c686: Running repeated flake loops on my fork, the one remaining failure was webkit-only and hit this suite's setup — |
What is this PR for?
This PR stabilizes the notebook keyboard-shortcut e2e suite by avoiding keyboard-driven fixture seeding in the Monaco editor.
In the Firefox Playwright job, setup could trigger Monaco suggestions while typing fixture content. That let Enter accept a suggestion before the shortcut assertion ran, corrupting expected text, for example
linebecominglicense.The helper now seeds fixture text through the editor textarea without shortcut-style key events. The shortcut assertions still use real keyboard events.
This also re-enables the clone-content shortcut coverage now that #5254 has been merged. The skip is no longer needed, and keeping it would leave that shortcut path without regression coverage.
What type of PR is it?
Bug Fix
What is the Jira issue?
https://issues.apache.org/jira/browse/ZEPPELIN-6536
How should this be tested?
Screenshots (if appropriate)
N/A
Questions: