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
{{ message }}
Repository navigation
Commit 1cbc189
Browse filesBrowse the repository at this point in the historyBrowse files
fix(map): while a rescan paints the saved index's tree under a query in the highlight box, the map is drawn undimmed and its footer is the map's own line - K03 took the older scan's answer down and left the query with none, and the overlay then judged "ext:mov,mkv" by the bare-word rule, which matches no name: every rectangle was dimmed and the footer read "0 matches" (ninth browser pass M01)
A grammar query with no answer for the tree on screen (queryMode 'grammar',
matchedPaths null) now dims nothing and counts nothing: renderSearchOverlay
draws no scrim, updateTmStatus writes the map's own line ("N items of 4.0 KB
or more drawn · 9.5 GB in this folder · from the saved index") and, drawn as
circles or Voronoi cells, renderCellsStatus writes the shapes' line. The box
still says the search runs once the scan finishes, and the scan that lands is
asked once and its answer draws the highlights and the count, as K03 had it.
drawTreemap's painting passes are untouched.
treeScanKey runs the page's own query bookkeeping, overlay and both footers
through the rescan, as rectangles and as circles. frontendContract's status
line fixture now gives its answered grammar query the paths it matched, as
the page always does beside a total.
Mutants M01-a (overlay rule reverted), b (footer rule reverted), c (shapes'
footer rule reverted), d (an answered grammar query treated as waiting), g
(the footer treating every grammar query as waiting), f (the match count kept
when the query waits), k1 (K03 keeping the older matches), k2 (K03 not saying
what the search waits for), r (the landing not asking again), each RED,
restored by SHA-256.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
test('while a rescan paints the saved index\'s tree under a query in the highlight box, the map is drawn undimmed and its footer is the map\'s own line — never "0 matches": the query waits for the scan, and the box says so — and the scan that lands is asked once and dims the map again (ninth browser pass M01)',async()=>{
992
+
// corpus scanned, `ext:mov,mkv` answered (266 matches), then Scan pressed again: the saved index's tree was painted
993
+
// while the scan ran, and K03 took the older scan's matches down, leaving the query in the box with no answer. The
994
+
// overlay then judged "ext:mov,mkv" by the bare-word rule, which matches no name: every rectangle got the scrim, and
995
+
// the footer read "0 matches for “ext:mov,mkv” · 9.5 GB in this folder", with no "· from the saved index".
assert.deepEqual([t.queryMode,t.matchedPaths,t.matchTotal],['grammar',null,null],`${drawn}, ${when}: control: the query waits, the older answer gone (K03)`);
1022
+
assert.equal(c.messages.at(-1),'The search runs on a finished scan — it runs once the scan finishes.',`${drawn}, ${when}: control: the box says the search waits for the scan`);
1023
+
assert.deepEqual(c.dimmed,[],`${drawn}, ${when}: nothing is dimmed while the query waits`);
1024
+
assert.equal(t.matches,0,`${drawn}, ${when}: and nothing is counted as a match`);
1025
+
assert.equal(footer(),cells ? '2 shapes · 2 without children · 9.5 GB total' : '2 items of 4.0 KB or more drawn · 9.5 GB in this folder · from the saved index',
1026
+
`${drawn}, ${when}: the footer is the map's own line, with no match count`);
1027
+
}
1028
+
// The scan lands (finishScan re-enters the view): the query is asked once, of it — nothing was asked of the running
1029
+
// scan — and its answer dims the map again and is counted under it.
[['ext:mov,mkv'],['scan-old','scan-new'],answered,'1 of 266 matches for “ext:mov,mkv” · 9.5 GB in this folder'],
1036
+
`${drawn}: the scan that lands is asked the query, once, and its answer dims the map and is counted again`);
1037
+
}
1038
+
});
1039
+
941
1040
test('the highlight box\'s query asked of the older scan during a rescan is not installed over the new tree when it answers after the rescan landed — the new scan is asked — and a scan of the same folder landing stops it on the server',async()=>{
0 commit comments