You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
test: update assertions to match post-260512 production behavior
Four test assertions had silently drifted out of sync with intentional
production changes since the c5x / live-mode work landed in May:
- test_toolbar / suite/TestToolbar — testToolbarHasAllButtons asserted
12 children, but 260512-hrn added a Follow uitoggletool to
FastSenseToolbar between the Live and Metadata buttons. Bump to 13
and update the inline button enumeration comment.
- test_time_range_selector — Case 5 asserted that setDataRange(0, 200)
with a contained selection of [20, 80] would rescale to [40, 160].
260512-live-mode-companion-adhoc-tail-spike inverted that contract:
when the new range is a strict superset of the current selection
(the live-tick growth path) the selection is preserved verbatim so
the user's window doesn't drift on every 1 s data extension. Split
Case 5 into 5a (preserve when contained) and 5b (rescale when range
contracts to exclude selection); both branches now match the actual
setDataRange logic.
- test_dashboard_preview_envelope — Case 6 expected
numel(xCenters) == 200 for a 500-sample widget. minmax_core_mex's
nb-bump (260512-live-mode-companion-adhoc-tail-spike) computes
nb_eff = floor(n / bucketSize), so requesting 200 buckets on 500
samples actually produces 250 envelope columns. Loosen to a range
assertion [200, n] and update the FastSenseWidget.getPreviewSeries
unpacker to derive the bucket count from the output length rather
than rejecting any size that isn't exactly 2*nBucketsEff.
- test_mex_parity — local minmax_core_matlab helper was an old-style
even-bucket reference that didn't include the nb-bump or the
260512-c5x tail anchor. Mirror both into the helper so the MEX vs
pure-MATLAB parity check actually compares apples to apples.
All four tests pass on both Octave 11.1 and MATLAB R2025a after these
changes; getPreviewSeries change is invisible on inputs where the MEX
output was already 2*nBucketsEff.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
0 commit comments