Title: SQLite Bug Forum: Forum
Domain: sqlite.org
| None | default-src 'self' data:; script-src 'self' 'nonce-129112c8877db759b9be9deae735410eb30e4076cf4c06a7'; style-src 'self' 'unsafe-inline'; img-src * data: |
Links:
| Login | https://sqlite.org/bugs/login |
| ☰ | https://sqlite.org/bugs/sitemap |
| About | https://sqlite.org/bugs/about |
| Threads | https://sqlite.org/bugs/forum |
| Post | https://sqlite.org/bugs/forumnew |
| Subscribe | https://sqlite.org/bugs/register |
| SQLite | https://sqlite.org/ |
| Forum | https://sqlite.org/forum |
| New Thread | https://sqlite.org/bugs/forumnew |
| Timeline | https://sqlite.org/bugs/timeline?ss=v&y=f |
| Tcl test suite cross-test failures: shared-memory leaks, stale callbacks, test-order failures, and path-dependent exclusions | https://sqlite.org/bugs/honeypot |
| Failed assertion in unixOpen for VACUUM INTO with URI and empty modeof | https://sqlite.org/bugs/honeypot |
| Multi-row UPSERT with a BEFORE UPDATE trigger that deletes the conflicting row leaves a rowid out of order | https://sqlite.org/bugs/honeypot |
| Wrong result: new.c can be NULL in a BEFORE UPDATE trigger for generated column c AS (a+b) when only a is set | https://sqlite.org/bugs/honeypot |
| FTS5: totals are cached as valid after a failed load of the averages record | https://sqlite.org/bugs/honeypot |
| A RANGE window frame and its exact mirror image return different results, and each is wrong on a different row | https://sqlite.org/bugs/honeypot |
| Clarification: floating-point rounding at a RANGE PRECEDING boundary | https://sqlite.org/bugs/honeypot |
| CLI: ".recover ''" (empty or "-" option) acts as --ignore-freelist | https://sqlite.org/bugs/honeypot |
| excluded.x lacks the column’s affinity and collation, so DO UPDATE ... WHERE may compare values differently and skip the upda | https://sqlite.org/bugs/honeypot |
| Wrong result: json_each() fullkey does not escape a double quote in a JSON5 single-quoted key | https://sqlite.org/bugs/honeypot |
| Authorizer bypass via ALTER TABLE ADD COLUMN generated column exposes denied column values | https://sqlite.org/bugs/honeypot |
| URI mode=ro does not propagate read-only to ATTACHed databases | https://sqlite.org/bugs/honeypot |
| WHERE-clause push-down into a compound subquery ignores the affinity of the subquery's own column, so rows are lost | https://sqlite.org/bugs/honeypot |
| ALTER TABLE RENAME COLUMN fails when a view refers to the column through a "*" expanded in a subquery or CTE | https://sqlite.org/bugs/honeypot |
| An index on a REAL generated column can make CAST(x AS INT) return a REAL | https://sqlite.org/bugs/honeypot |
| ALTER TABLE RENAME COLUMN silently changes what an existing view returns | https://sqlite.org/bugs/honeypot |
| A COLLATE on a later row-value element makes range scans use the wrong collation on the first index column, omitting rows. | https://sqlite.org/bugs/honeypot |
| In UTF-16BE, a LIKE/`GLOB` prefix ending in U+D7FF gets an incorrect U+FFFD upper bound, admitting nonmatching rows | https://sqlite.org/bugs/honeypot |
| Wrong result: dbstat reports pgsize and pgoffset of another database when the schema is selected per query | https://sqlite.org/bugs/honeypot |
| Wrong result: dbstat WHERE name='T1' COLLATE NOCASE, or joined to a NOCASE column, finds no rows for table t1 | https://sqlite.org/bugs/honeypot |
| PRAGMA integrity_check does not report surplus entries in a partial index | https://sqlite.org/bugs/honeypot |
| Various problems in extensions including RBU, sqlite3_rsync, expert, amatch, vec1 and session. | https://sqlite.org/bugs/honeypot |
| Wrong result: printf('%#010x', 255) returns 12 characters, '0x00000000ff', instead of padding to the width 10 | https://sqlite.org/bugs/honeypot |
| ON DELETE CASCADE chain that stops at the trigger-depth limit can keep its partial deletes in a transaction | https://sqlite.org/bugs/honeypot |
| Wrong result: format('%.0e') rounds 750.0 down to 7e+02 but 7.5 up to 8e+00 | https://sqlite.org/bugs/honeypot |
| ↓ Older... | https://sqlite.org/bugs/forum?x=25&n=25 |
Viewport: width=device-width, initial-scale=1.0