mirror of
https://github.com/ZoneMinder/zoneminder.git
synced 2026-11-06 20:53:00 -05:00
Isaac ConnorandClaude Opus 5
db4dd75e14
submitTab() resubmits contentForm without saving, so the view re-renders for a
value that changes which fields it should show. It did
form.attr('tab', tab), which sets an attribute named "tab" on the <form>
element and has no effect whatever: an attribute on a form is not a submitted
value. The view reads $_REQUEST['tab'], which comes from the hidden field.
Writing it correctly would have been worse than leaving it broken. The value
it was being handed, data-tab-name, is the tab the page loaded on, while the
hidden field now holds the tab that is on screen. Assigning the attribute's
value would have undone that tracking and sent anyone who had switched tabs
back to the one they arrived on -- the bug this looks like it would fix. So
the tab is not written here at all, and the field is left to do its job.
Drop data-tab-name from the Method select, since nothing reads it now. An
attribute nothing reads is a trap: the next person here would reasonably
assume it was load bearing.
The submit stays on $j(...).submit() rather than the element's own submit(),
which is worth saying out loud in a comment because the difference is
invisible: jQuery triggers contentForm's submit handler, which validates the
form and blanks the password fields, and a native submit skips both.
Noted, not changed: skin.js also binds submitTab to ".pillList a" on click,
and no markup anywhere in the tree carries that class, so it matches nothing.
It was the only caller that would have wanted data-tab-name's meaning. Left
alone as dead but harmless, and in a file every view loads.
Tests: tests/js/monitor-tab-submit.test.js, 9 checks. skin.js and monitor.js
are loaded into one sandbox, as a page has them, because the behaviour needs
both: monitor.js tracks the tab into the hidden field and skin.js submits it.
The cases are built so that the loaded tab and the showing tab differ, since
any test where they agree would pass whether or not the bug is fixed.
Verified the checks fail when the old attr call is restored (2), when the
naive fix of writing data-tab-name into the field is applied (4), and when
the submit is changed to a native one (7).
Ran: tests/js/monitor-tab-submit.test.js (9 passed), monitor-tab-url.test.js
(14 passed), the rest of tests/js and tests/php, npx eslint . (clean),
utils/check-homoglyphs.py (clean). test_bandwidth_clamp.php fails on this box
for an unrelated reason: it cannot read /etc/zm/conf.d/03custom.conf and the
permission warning lands in its captured output.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JpiSWBmtQkR5bcgpHWY4ME