50 most recent check-ins
|
2026-11-12
| ||
| 00:10 | Do not misidentify dot-commands as database files on the command-line parsing of the CLI. Also fix a help typo. (leaf check-in: 3c48444aff ... user: drh tags: trunk) | |
|
2026-11-11
| ||
| 23:27 | Add the "posting rules for agents" section to AGENTS.md (check-in: 5fed53b458 ... user: drh tags: trunk) | |
| 23:05 | Improvements to the AGENTS.md file. (check-in: d27e488899 ... user: drh tags: trunk) | |
| 20:51 | Fix bugs in and improve the interface to the ".check" command of the CLI. (check-in: 1a00748caf ... user: drh tags: trunk) | |
| 19:39 | Ensure a statement that uses REPLACE conflict handling uses a statement journal if a later index is on an expression or is a partial index, in case evaluating that expression or WHERE clause raises an error after the conflicting row has been deleted. Report 2026-11-04T20:23:28Z. (check-in: 75757ed3a5 ... user: dan tags: trunk) | |
| 19:11 | Ensure a statement is marked as may-abort if it uses an aggregate or window function, or if it reads from a virtual table. Reports 2026-11-04T20:27:20Z and 2026-11-04T20:31:13Z. (check-in: c9c3554918 ... user: dan tags: trunk) | |
| 15:30 | Take out ".mode test" added as part of [85e50de017d28abc] since it is really just an alias for ".mode quote". Update recent test cases to use "quote" instead of "test". (check-in: 509c350654 ... user: drh tags: trunk) | |
| 14:58 | If an error during a batch-atomic-write (F2FS) commit causes SQLite to fall back to a regular rollback mode commit, ensure that the db file is synced to disk before any data is written to disk. Report 2026-11-10T07:50:48Z. (check-in: 75ad951b1b ... user: dan tags: trunk) | |
| 14:43 | Minor typo fix. (check-in: 1e2c11c32d ... user: drh tags: trunk) | |
| 14:38 | Add the doc/writing-test-cases.md document. Improvements to AGENTS.md. (check-in: ea4c56756f ... user: drh tags: trunk) | |
| 13:17 | Still more hardening against smuggled shell command attacks. (check-in: 8981033053 ... user: drh tags: trunk) | |
| 13:09 | A 64-bit WASM fix for [1b7b50a6d601c]. (check-in: d380b494c9 ... user: stephan tags: trunk) | |
| 13:04 | Simplify the WASM on-dispose notifications by eliminating the intermediary struct. It turns out that we can overwrite a static function pointer from JS without it. (check-in: 1b7b50a6d6 ... user: stephan tags: trunk) | |
| 12:32 | Further strengthening of the new ".safe" command in the CLI. (check-in: d486e99790 ... user: drh tags: trunk) | |
| 11:39 | Add the ".safe" command to the CLI and enhance the ".recover" command so that the script it generates includes the ".safe" command near the top. Prior check-in [d684ccc9ec4e41d8] is sufficient to block the attack described by bug 2026-11-06T04:16:54Z and CVE-2026-119586. This check-in simply adds an additional layer of defense. (check-in: f2be7c2a98 ... user: drh tags: trunk) | |
| 03:00 | Further minor tweaks to AGENTS.md (check-in: 27f322ba50 ... user: drh tags: trunk) | |
| 02:57 | Updates to AGENTS.md based on recent experience. (check-in: e80beb913f ... user: drh tags: trunk) | |
| 01:04 | Fix CAST(TO NUMERIC) so that it correctly understand the entire prefix in all circumstances. Bug 2026-11-04T20:34:18Z (check-in: e390985460 ... user: drh tags: trunk) | |
| 00:43 | Fix a boundary case in substr(). Bug 2026-11-04T20:19:36Z (check-in: 4ddf1761c0 ... user: drh tags: trunk) | |
|
2026-11-10
| ||
| 23:49 | Fix corner-case issues in date/time functions as described by bugs 2026-11-05T22:10:45Z, 2026-11-04T00:49:18Z, and 2026-11-04T00:48:32Z. (check-in: 1cbe4599b0 ... user: drh tags: trunk) | |
| 22:58 | Do not allow timezones on a HH:MM:SS offset modifier. (closed check-in: 4b7348afb4 ... user: drh tags: date-time-bugs) | |
| 22:00 | Clean up excess commentary in the initial prototype fix. (check-in: 80e3541043 ... user: drh tags: date-time-bugs) | |
| 21:42 | In the CLI and the ".mode" command, clamp various parameters at the limit of their range and do not allow integer truncation, as robots seem to not like integer truncation behavior in user interfaces. Bug 2026-11-05T22:26:14Z (check-in: a6250888ec ... user: drh tags: trunk) | |
| 21:10 | Enhance QRF to allow columnar modes to wrap at a control character. Bug 2026-11-05T22:27:00Z (check-in: 6982393888 ... user: drh tags: trunk) | |
| 20:46 | Fix NULL outputs for the "html" output mode in the CLI. Bug 2026-11-05T22:27:46Z (check-in: f827be82ea ... user: drh tags: trunk) | |
| 20:42 | Doc-only updates in the JS/WASM pieces. (check-in: 363e7a9f71 ... user: stephan tags: trunk) | |
| 20:31 | When processing immediate FK constraints, do not assume a CASCADE or SET NULL action resolves all constraints if there are BEFORE triggers that will fire on the child table, as a RAISE(IGNORE) may intervene. Report 2026-11-04T20:21:09Z. (check-in: 0c42b4821a ... user: dan tags: trunk) | |
| 19:56 | Expand AGENTS.md and add guidance for AI-assisted bug investigation and coordination. Documentation drafted by Codex from the SQLite investigations. (leaf check-in: 769df2de39 ... user: lehenbauer tags: agent-guidance) | |
| 19:16 | Put an ALWAYS() on an always-true condition in [c41e45677d75bf49]. (check-in: 885008f7ff ... user: drh tags: trunk) | |
| 19:10 | Process all immediate FK constraints before all deferred ones. Otherwise, if there is no statement journal and a deferred FK constraint is violated before an immediate one is, the database handle is left in an inconsistent state. Report 2026-11-04T20:20:23Z. (check-in: 43c40c7961 ... user: dan tags: trunk) | |
| 18:38 | Within an EXISTS(...) subquery, do not ignore DISTINCT if there is an OFFSET clause. Report 2026-11-04T01:01:41Z. (check-in: c41e45677d ... user: dan tags: trunk) | |
| 17:16 | Do not propagate a constant gleaned from a no affinity WHERE clause term to the RHS of another WHERE clause comparison that uses TEXT affinity, just as is already done for the LHS. Report 2026-11-04T01:00:08Z. (check-in: 42fbc72bcf ... user: dan tags: trunk) | |
| 15:54 | Ensure that expressions like "x IS (y IN ())" are equivalent to "x IS 0", not "x IS FALSE". Reports 2026-11-10T13:17:50Z and 2026-11-05T22:18:29Z. (check-in: 3b28941b47 ... user: dan tags: trunk) | |
| 15:51 | Enhance QRF to do a better job of restricting the title width when there are accented and double-width characters, with and without ellipsis. Bug 2026-11-05T22:29:41Z. (check-in: a3738b9e3d ... user: drh tags: trunk) | |
| 15:09 | If the RHS of a vector IN() expression is a sub-query with a LIMIT clause, do not allow the sub-query to be split into multiple separate queries each returning a single column, as such queries might return different rows than the overall query. Report 2026-11-04T00:59:22Z. (check-in: 177d350eb8 ... user: dan tags: trunk) | |
| 14:11 | Remove a stray, obsolete bit of Big Endian test code. Forum report 2026-11-05T12:25:12Z. (check-in: b674507ef7 ... user: stephan tags: trunk) | |
| 13:19 | Fix another adverse interaction between outer join strength reduction and query flattening, as reported by bug 2026-11-08T01:01:03Z. Also add more test cases for the previous strength-reduction-flattening fix at [8d213d8a1dc2a1b6]. (check-in: 89066ca37c ... user: drh tags: trunk) | |
| 12:52 | Fix harmless compiler warning in [ab6f3665f960aa12]. (check-in: 84b0327ef2 ... user: drh tags: trunk) | |
| 12:29 | Avoid unnecessary scanning of subqueries in the previous check-in. Move the sqlite3SelectWalkFail() routine from expr.c to walker.c for aesthetics. (check-in: ff612271ea ... user: drh tags: trunk) | |
| 12:21 | Fix an adverse interaction between outer-join strength reduction and query flattening. Bug 2026-11-08T07:03:28Z (check-in: 8d213d8a1d ... user: drh tags: trunk) | |
| 11:18 | Only intepret "WHERE x=y" as allowing the substitution of "x" for "y" in other parts of the WHERE clause if the collation sequence used for "x=y" is either BINARY or else matches the collation sequence used where the substitution takes place. Report 2026-11-04T00:57:49Z. (check-in: 11076b8c60 ... user: dan tags: trunk) | |
| 10:38 | Enhance the TEXT→REAL converter so that it works even for ridiculous AI-generated floating point literals. Forum thread 2026-11-09T04:18:04Z. (check-in: a10f9df37b ... user: drh tags: trunk) | |
| 01:13 | Omit some test cases that require virtual tables when virtual tables are not available. (check-in: 2b0e0e3d59 ... user: drh tags: trunk) | |
|
2026-11-09
| ||
| 23:49 | Fix duplicate test numbers in a test script. No code changes. (check-in: d84818861d ... user: drh tags: trunk) | |
| 21:20 | More commentary on the operation of the capable() function that is built into the CLI. (check-in: 7d156a5fd3 ... user: drh tags: trunk) | |
| 20:49 | Avoid truncating declared column types that end in "always". Report 2026-11-04T00:47:45Z. (check-in: ab6f3665f9 ... user: dan tags: trunk) | |
| 18:44 | Have the rtree code catch SQLITE_TOOBIG and other errors in sqlite3_bind_blob(). Report 2026-11-03T18:57:10Z. (check-in: 24525ea163 ... user: dan tags: trunk) | |
| 18:13 | Ensure that affinities are applied to the excluded.* values in a DO UPDATE clause associated with an IPK confict. Report 2026-11-03T02:17:21Z. (check-in: ed9e465405 ... user: dan tags: trunk) | |
| 17:42 | Update README.md and AGENTS.md to try to make it clearer that bug reports should go to the bugs forum and not to the main forum. (check-in: 192e4ee8f8 ... user: drh tags: trunk) | |
| 17:36 | When a VALUES clause is part of a compound SELECT, do not ignore collation sequences attached to rows or SELECTs to the right of the first VALUES row. Report 2026-11-03T02:16:35Z. (check-in: 4d8d52d654 ... user: dan tags: trunk) | |
