|
2026-08-25
| | |
| 18:40 |
|
Improved handling of OOM conditions in the sessions extension.
Bug 2026-08-25T17:04:14Z.
(leaf check-in: ce2aa9f849 ...
user: drh tags: trunk)
|
| 18:28 |
|
Add assert() statements and comments to explain why the return value of sqlite3_column_name() does not need to be checked in fts3.c. Repor 2026-08-25T17:08:22Z.
(check-in: 08b34b1656 ...
user: dan tags: trunk)
|
| 18:26 |
|
Avoid a possible NULL pointer dereference in the QRF sublibrary when
compiled with SQLITE_ENABLE_STMT_SCANSTATUS.
(check-in: 58014b2a2e ...
user: drh tags: trunk)
|
| 18:16 |
|
Improved handling of OOM errors in the Windows implementation of the
sqlite3_stdio.c extension.
Bug 2026-08-25T16:55:53Z.
(check-in: 555489e4c2 ...
user: drh tags: trunk)
|
| 17:59 |
|
Prevent some potential NULL pointer deferences following misuse of the
non-standard amatch extension.
Bug 2026-08-25T16:47:32Z.
(check-in: 99b69371a1 ...
user: drh tags: trunk)
|
| 17:12 |
|
New optimizations for improved performance of UNION with LIMIT 1, with
or without ORDER BY.
(check-in: b5b8ede613 ...
user: drh tags: trunk)
|
| 17:08 |
|
Fix two test cases that do not work when optimizations are disabled.
(closed check-in: e78b72fadf ...
user: drh tags: union-to-union-all)
|
| 16:29 |
|
Make the optimization of UNION with ORDER BY and LIMIT sound by restricing
it to LIMIT 1.
(check-in: 7e6e665d05 ...
user: drh tags: union-to-union-all)
|
| 16:15 |
|
Fix OOM handling in little-used extension file "expert.c". Report 2026-08-25T16:02:31Z.
(check-in: c89f4f7967 ...
user: dan tags: trunk)
|
| 15:49 |
|
Merge the latest trunk enhancements into the union-to-union-all branch.
(check-in: c58631b26a ...
user: drh tags: union-to-union-all)
|
| 14:27 |
|
Update the Lemon documentation with clarifications regarding the
%fallback and %wildcard directives.
(check-in: f3ad1b4394 ...
user: drh tags: trunk)
|
| 14:26 |
|
When an expression that may have a subtype from the WHERE clause of a partial index is substituted for a column value, clear the subtype before the value is used. Report 2026-08-25T05:36:33Z.
(check-in: e6ae6fbd5c ...
user: dan tags: trunk)
|
| 11:50 |
|
Fix test file default.test so that it works with SQLITE_OMIT_ALTERTABLE builds.
(check-in: ff0e3dd094 ...
user: dan tags: trunk)
|
| 11:42 |
|
Avoid coding WHERE constraints too early when using the OR-optimization with OUTER join queries. Report 2026-08-25T02:08:22Z.
(check-in: a3d15e4937 ...
user: dan tags: trunk)
|
| 10:39 |
|
Change error message text in the previous check-in to not have initial caps,
for consistency with other error messages.
(check-in: c4bd2ffcb7 ...
user: drh tags: trunk)
|
|
2026-08-24
| | |
| 21:20 |
|
Prohibit using ALTER TABLE to add or drop, or rename to or from a rowid alias, unless the table is a WITHOUT ROWID table. Report 2026-08-23T15:59:23Z.
(check-in: dfcff5be53 ...
user: dan tags: trunk)
|
| 18:55 |
|
Avoid a technically undefined signed integer overflow in fts5. Report 2026-08-23T16:25:44Z.
(check-in: cde23480be ...
user: dan tags: trunk)
|
| 18:24 |
|
Fix some fts5 cases where all columns are excluded by a column filter (so that no rows match). Report 2026-08-23T16:25:44Z.
(check-in: 7c9b711fc8 ...
user: dan tags: trunk)
|
| 17:29 |
|
Ensure that the fts5_vocab.term column behaves as if it has blob affinity for comparisons. Report 2026-08-23T16:25:44Z.
(check-in: a832bc1ef2 ...
user: dan tags: trunk)
|
| 17:08 |
|
Fix a problem with combining index=none and secure-delete modes causing full-text index entries to be sometimes lost when inserts and deletes on the same token are combined in the same transaction. Report 2026-08-23T16:25:44Z.
(check-in: e1f939f473 ...
user: dan tags: trunk)
|
| 16:19 |
|
Fix harmless compiler warning.
(check-in: 0c5f5b37ce ...
user: drh tags: trunk)
|
| 14:41 |
|
When generating code to evaluate a vector IN constraint where the RHS
of the IN operator is a subquery, if the removeUnindexableInClauseTerms()
subroutine changes the order of terms in the IN operator, then do not
allow the modified IN operator to reuse a previously generated evaluation
of the subquery.
(leaf check-in: 2f95f4a873 ...
user: drh tags: branch-3.53)
|
| 14:40 |
|
Ensure that when a large integer value (too large to be losslessly converted to double) is specified as a default value for a column with REAL affinity, the real version of the value is stored in indexes. Report 2026-08-23T15:39:59Z.
(check-in: 2cb349e5b0 ...
user: dan tags: trunk)
|
| 14:30 |
|
When generating code to evaluate a vector IN constraint where the RHS
of the IN operator is a subquery, if the removeUnindexableInClauseTerms()
subroutine changes the order of terms in the IN operator, then do not
allow the modified IN operator to reuse a previously generated evaluation
of the subquery.
Bug 2026-08-23T06:58:36Z.
(check-in: 8ee92456eb ...
user: drh tags: trunk)
|
| 14:22 |
|
Ensure that when a large integer value (too large to be losslessly converted to double) is specified as a default value for a column with REAL affinity, the real version of the value is stored in indexes. Report 2026-08-23T15:39:59Z.
(closed check-in: c494b123f5 ...
user: dan tags: bigint-defaults)
|
| 09:56 |
|
Remove a harmless unused variadic argument in sqlite3_rsync. Bug report 2026-08-24T01:22:51Z.
(check-in: 5231ac428d ...
user: stephan tags: trunk)
|
| 00:08 |
|
The optimization of disabling an ORDER BY term if it is constrained by
an equality comparison with a column from an outer query does not work
if there is an affinity mismatch between the two columns. This is a
follow-on to check-in [16aed5d0c63dcdc2] and a fix for
Bug 2026-08-23T01:02:14Z.
(check-in: 383e42bc10 ...
user: drh tags: trunk)
|
|
2026-08-23
| | |
| 21:41 |
|
Correctly enforce a UNIQUE constraint on a non-PRIMARY KEY column of a table
that has an INTEGER PRIMARY KEY with ON CONFLICT REPLACE and that has
a AFTER DELETE trigger that fires when the REPLACE happens and that trigger
tries to INSERT a new row that violates uniqueness.
Bug 2026-08-22T23:55:15Z.
(check-in: 521c9f5f28 ...
user: drh tags: trunk)
|
|
2026-08-22
| | |
| 19:27 |
|
Enhance sqlite3_bind_int64() so that it never triggers a reprepare if the
value does not actually change.
(check-in: db0cb462aa ...
user: drh tags: trunk)
|
| 17:56 |
|
Fix incorrect termination of varargs argument to pickStr() in
check-in [852972861636d75e] from yesterday.
(check-in: ad7d15323b ...
user: drh tags: trunk)
|
| 16:33 |
|
Remove some unnecessary sqlite3_test_control() optimization control flags
for a small performance increase and size reduction.
(check-in: 2a9cb1e12a ...
user: drh tags: trunk)
|
| 15:56 |
|
Improved use of LIMIT in UNION operators.
(check-in: 7304e5aeaf ...
user: drh tags: union-to-union-all)
|
| 14:41 |
|
New optimization bit SQLITE_UnionLimit.
(check-in: 5890dbe54d ...
user: drh tags: union-to-union-all)
|
| 11:10 |
|
Improved testability.
(check-in: e8e1ce6dae ...
user: drh tags: union-to-union-all)
|
| 10:39 |
|
Add an optimization that converts UNION into UNION ALL if there is
a LIMIT 1 and no ORDER BY.
(check-in: fb83742dd1 ...
user: drh tags: union-to-union-all)
|
| 09:33 |
|
Fix check-in [468e730036edac22] so that it works with DESC indexes.
Bug 2026-08-22T07:01:45Z.
(check-in: d1f4fa43f3 ...
user: drh tags: trunk)
|
| 01:21 |
|
A couple of fixes, one of them important, for the previous check-in.
(check-in: 3a01099c45 ...
user: drh tags: trunk)
|
|
2026-08-21
| | |
| 20:32 |
|
Faster implementation of sqlite3_bind_int64(). Reduce the need to
reprepare when replacing one integer binding with a different integer
value is many cases.
(check-in: 98edd4ac22 ...
user: drh tags: trunk)
|
| 20:21 |
|
Refactor. 2M cycles faster on x64.
(closed check-in: 2b3e73200b ...
user: drh tags: reduce-reprepares)
|
| 14:04 |
|
In the CLI, finer-grain control of what status information is displayed via
the ".stats" command. Add the ability to reset peak measurements after
each display.
(check-in: 8529728616 ...
user: drh tags: trunk)
|
| 00:31 |
|
Fix a harmless compiler warning in the non-deliverable normalize.c
extension. Bug 2026-08-20T23:33:57Z.
(check-in: 11a73d18be ...
user: drh tags: trunk)
|
|
2026-08-20
| | |
| 16:47 |
|
Fix a faulty assert() in lemon, as reported at
Forum post 2026-08-20T14:18:08Z.
(check-in: 0672f6f074 ...
user: drh tags: trunk)
|
| 15:59 |
|
Remove a literal 0x0d character from a sessions test script, using
char(0x0d) in its place, to avoid false-positive mixed-line-ending
warnings from Fossil.
(check-in: d0e4714aff ...
user: drh tags: trunk)
|
| 14:43 |
|
Merge the latest trunk enhancements into the reuse-schema branch.
(leaf check-in: 9d4f4c989a ...
user: drh tags: reuse-schema)
|
| 14:32 |
|
Merge the latest trunk enhancements into the bedrock branch via wal2.
(leaf check-in: 82abbf1d03 ...
user: drh tags: bedrock)
|
| 14:21 |
|
Merge the latest trunk enhancements into wal2 branch.
(leaf check-in: 7257cc1b70 ...
user: drh tags: wal2)
|
| 14:14 |
|
Merge the latest trunk enhancements into the begin-concurrent branch.
(leaf check-in: 6095ade60d ...
user: drh tags: begin-concurrent)
|
| 14:05 |
|
Merge the latest trunk enhancements into the reduce-reprepares branch.
(check-in: 62ad88fe1c ...
user: drh tags: reduce-reprepares)
|
| 13:25 |
|
Performance enhancements (inlining and partial inlining) in the
jsonbPayloadSize() routine of the JSON logic saves a little more than
10 million cycles (0.6%) on the standard benchmark (gcc 13.3, -Os, mint linux)
while using less than 700 extra bytes of code space.
(check-in: 3bd5456e50 ...
user: drh tags: trunk)
|
| 13:12 |
|
Use testcase() macros to prove coverage in jsonbPayloadSizeWide()
(closed check-in: cff777eeb9 ...
user: drh tags: opt-inline-jsonbPayloadSize)
|
| | |