kobotoolbox / kpi / 30907077616 / 5

59%

master: 76%

push

github

fix(formbuilder): preserve manual skip logic and validation inputs DEV-2539 (#7347)

### 📣 Summary

Fixed an issue where skip logic and validation criteria entered manually
could be lost after closing and reopening the question settings panel.

### 💭 Notes

In the form builder's hand-code mode for skip logic and validation
criteria, closing the settings panel (clicking ✕) and reopening it would
reset the text field to the value it had when the panel was first
opened. This meant any edits made since opening the panel were silently
discarded — including deleting existing logic. Saving the form at that
point would preserve the stale value rather than the intended one.

The root cause was that `TextArea.render()` always resets the DOM value
to its original construction-time text. On each re-open, `render()` was
called again, clobbering whatever the user had typed. Separately,
`serialize()` read its value directly from the DOM element — which
returns empty string when the element is detached — and fell back to the
last-known non-empty value, making it impossible to clear existing
logic.

Also removed a spurious `survey.trigger('change')` that was firing when
switching to hand-code mode before the user had typed anything, which
caused the save button to show unsaved changes immediately on mode
switch.

### 👀 Preview steps

1. ℹ️ Have a project with at least two questions
2. Open the form builder and click the settings (gear) icon on a
question
3. Go to **Skip Logic**, click "Manually enter your skip logic in
XLSForm code", and type `${q1} != ''`
4. Click ✕ to close the settings panel, then reopen it
5. 🔴 [on main] the textarea is empty or reset to its original value
6. 🟢 [on PR] the textarea shows `${q1} != ''`
7. Clear the textarea completely, close the panel, and save the form
8. 🔴 [on main] the old skip logic value is still saved
9. 🟢 [on PR] the skip logic is cleared
10. Repeat steps 3–9 using **Validation Criteria** instead of Skip Logic
11. 🟢 [on PR] ... (continued)

3846 of 12692 branches covered (30.3%)

17135 of 37629 relevant lines covered (45.54%)

0.46 hits per line