| 86.9 minutes ago |
A single-element IN list uses an explicit COLLATE on its element, contrary to the documentation and to a two-element list | no replies | Open |
| 10.5 hours ago |
LEFT JOIN with `x = NULL AND y IN (…)` on a two-column index returns a matched row as well as the NULL-extended one | 2 posts spanning 4.2 hours | Open |
| 10.5 hours ago |
In a UTF-16 database, a blob over 4000 bytes read from a table's overflow pages is converted as UTF-8 | 2 posts spanning 11.8 hours | Resolved |
| 19.3 hours ago |
ROLLBACK TO of a savepoint that dropped a table commits a stale root page in an incremental-vacuum database | 3 posts spanning 87.7 minutes | Resolved |
| 27.3 hours ago |
Wrong query result in RIGHT JOIN | 4 posts spanning 7.8 hours | Resolved |
| 2.6 days ago |
DELETE on a WITHOUT ROWID table leaves an unaccounted byte on an interior page; integrity_check flags it, later DELETE fails | 3 posts spanning 4.9 hours | Resolved |
| 2.7 days ago |
opfs-sahpool forgets a journal whose delete failed, so the hot-journal fix later rolls it back over newer commits | 2 posts spanning 2.6 hours | Resolved |
| 2.8 days ago |
ALTER COLUMN SET NOT NULL on a column that is already NOT NULL drops its ON CONFLICT clause and constraint name | 2 posts spanning 59.2 minutes | Resolved |
| 2.9 days ago |
heap-buffer-overflow in sqlite3SchemaToIndex() using an eponymous virtual table after sqlite3_deserialize() | 3 posts spanning 1.9 days | Resolved |
| 2.9 days ago |
ADD COLUMN ... TEXT DEFAULT 1e2: rows that existed before the ALTER read '1e2', new rows read '100.0' | no replies | Open |
| 3.0 days ago |
"datatype mismatch" on the rowid in INSERT/UPDATE OR REPLACE does not undo the statement's earlier rows inside a transaction | 2 posts spanning 6.2 hours | Resolved |
| 3.1 days ago |
Multi-row VALUES is exempt from SQLITE_LIMIT_COMPOUND_SELECT; the UNION ALL fallback crashes during error cleanup | 2 posts spanning 30.7 hours | Resolved |
| 3.1 days ago |
sqlite3_preupdate_new() reports SQLITE_INTEGER for a whole-number REAL column value on INSERT | 2 posts spanning 19.2 hours | Resolved |
| 3.1 days ago |
ALTER TABLE ADD CONSTRAINT ... CHECK fails with "constraint failed" when the expression is NULL for an existing row | 2 posts spanning 3.3 hours | Resolved |
| 3.1 days ago |
Nothing bounds CTE expansion: linear SQL costs exponential time and memory | 2 posts spanning 29.1 hours | Resolved |
| 3.2 days ago |
A chain of CTEs escapes the nesting-depth limit and overflows the stack | 2 posts spanning 28.6 hours | Resolved |
| 3.7 days ago |
A COLLATE chain escapes SQLITE_LIMIT_EXPR_DEPTH and overflows the stack | 2 posts spanning 17.1 hours | Resolved |
| 3.8 days ago |
char() and unistr() emit ill-formed UTF-8 (CESU-8 surrogates) for code points U+D800..U+DFFF | 2 posts spanning 19.6 hours | Resolved |
| 3.9 days ago |
Regression since 3.52.0: .dump and .mode insert/json write infinity as Inf, so the dump cannot be restored | 2 posts spanning 10.8 hours | Resolved |
| 3.9 days ago |
geopoly_regular(): N is truncated to 32 bits before the cap, so N=2^31 gives NULL and N=2^32+3 gives a triangle | 2 posts spanning 15.6 hours | Resolved |
| 3.9 days ago |
geopoly: vertex count truncated to 24 bits, producing unreadable blobs with no error | 2 posts spanning 10.6 hours | Resolved |
| 4.1 days ago |
PRAGMA busy_timeout=N with N >= 2^31 sets the timeout to 0, so a "wait longer" request becomes "never wait" | 2 posts spanning 10.6 hours | Resolved |
| 5.1 days ago |
opfs-sahpool: xCheckReservedLock always reports a lock, so hot journals are never rolled back after a crash | 2 posts spanning 2.8 hours | Open |
| 6.0 days ago |
PRAGMA integrity_check fails on every contentless FTS4 table since 3.44.0 | 2 posts spanning 2.2 hours | Resolved |
| 6.1 days ago |
Benign data races in SQLITE_DEBUG-only code are reported by ThreadSanitizer (testcase(), staticMutexes nRef, sqlite3ParserTra | 2 posts spanning 1.6 hours | Resolved |
| ↓ Older... |
