|
2026-08-18
| | |
| 00:36 |
|
Fix the include guard on pcache.h, as requested by Report
2026-08-17T20:09:10Z.
(leaf check-in: e53c5fd75f ...
user: drh tags: trunk)
|
| 00:30 |
|
Declare fts5 structure records corrupt if they contain a page number greater than 0x7FFFFF00.
(leaf check-in: 14be580d24 ...
user: drh tags: branch-3.53)
|
| 00:23 |
|
Avoid a problem causing a use-after-free in the fts3 snippet() function.
(check-in: 9a367c177f ...
user: drh tags: branch-3.53)
|
| 00:17 |
|
Further optimizations to jsonbPayloadSize(). The code is now faster than
the first check-in on this branch, but less than 700 bytes larger than trunk
(on x64 with gcc -Os).
(leaf check-in: 870313fea0 ...
user: drh tags: opt-inline-jsonbPayloadSize)
|
|
2026-08-17
| | |
| 21:26 |
|
Declare fts5 structure records corrupt if they contain a page number greater than 0x7FFFFF00. Report 2026-08-17T20:07:33Z.
(check-in: c97a940ceb ...
user: dan tags: trunk)
|
| 15:31 |
|
New -pgo option to test/speedtest.tcl compiles using gcc-style PGO.
(check-in: b460488d05 ...
user: drh tags: trunk)
|
| 13:55 |
|
Avoid a problem causing a use-after-free in the fts3 snippet() function. Report 2026-08-17T09:18:24Z.
(check-in: e4635c4153 ...
user: dan tags: trunk)
|
| 13:25 |
|
Fix harmless missing end-comment.
(leaf check-in: 9de93bd467 ...
user: drh tags: threaded-bytecode-experiment)
|
| 13:07 |
|
Merge recent trunk enhancements into the threaded bytecode branch.
(check-in: 3ba47c84bd ...
user: drh tags: threaded-bytecode-experiment)
|
| 13:02 |
|
Increment the patch level to 5.
(check-in: f7371f41f9 ...
user: drh tags: branch-3.53)
|
| 12:59 |
|
Avoid a buffer overrun when processing a strategically corrupted changeset.
(check-in: a1dd7123d8 ...
user: drh tags: branch-3.53)
|
| 11:52 |
|
Have the fts5term and fts3aux virtual table modules only process constraints on the term column that use binary collation. Report 2026-08-17T10:17:15Z.
(check-in: f26099b7fa ...
user: dan tags: trunk)
|
| 11:31 |
|
Avoid a buffer overrun when processing a strategically corrupted changeset. Report 2026-08-16T10:15:25Z.
(check-in: 3f57024aec ...
user: dan tags: trunk)
|
|
2026-08-16
| | |
| 23:05 |
|
Partial in-lining of jsonbPayloadSize(), happens in only 5 performance critical
places instead of everywhere. This gives 96% of the performance increase with
only 18% of the object code size increase.
(check-in: e934c4a7aa ...
user: drh tags: opt-inline-jsonbPayloadSize)
|
| 19:54 |
|
Add the "Variable Number Bitmap Cache" to speed up parsing of queries that
have a large number of "?NNN" style variables.
(check-in: 2955ac4264 ...
user: drh tags: trunk)
|
| 19:50 |
|
Change the variable name for the variable-number bitmap cache. Make the size
of the cache configurable at compile-time.
(closed check-in: 35c9c6e635 ...
user: drh tags: opt-add-bitmap-scan-VList)
|
| 12:04 |
|
Add a small bitmap to the Parse object recording which variable numbers below 128 are already in Parse.pVList, so that repeated or out-of-order ?NNN parameters avoid a linear scan of the VList on each occurrence. Optimize particularly well on ORM testset.
(check-in: 306f8f3ccb ...
user: jeffchen tags: opt-add-bitmap-scan-VList)
|
| 11:52 |
|
Mark jsonbPayloadSize() with SQLITE_OPT_INLINE, so that each JSONB node-header decode site absorbs a specialized copy in amalgamation builds. Reduces JSON testset time at a cost of about 9-10KB of code space in speed builds
(check-in: 6d8e8474c5 ...
user: jeffchen tags: opt-inline-jsonbPayloadSize)
|
| 10:52 |
|
When reading a frame from a WAL file where the WAL pagesize and the database
pagesize disagree, be sure to fully initialize the page that is read.
(check-in: 4fde87b2a0 ...
user: drh tags: branch-3.53)
|
| 10:49 |
|
When reading a frame from a WAL file where the WAL pagesize and the database
pagesize disagree, be sure to fully initialize the page that is read.
Report 2026-08-17T15:54:41Z.
(check-in: f9b4ee3695 ...
user: drh tags: trunk)
|
| 06:00 |
|
Merge trunk into the tcl-init branch.
(leaf check-in: 327a61921d ...
user: stephan tags: tcl-init)
|
| 05:19 |
|
Add -DBUILD_sqlite to the T.compile.tcl makefile definition for the sake of Unix-on-Windows environments as a partial resolution to forum post fe9e99eb27.
(check-in: aff3e43ed5 ...
user: stephan tags: trunk)
|
|
2026-08-15
| | |
| 14:24 |
|
Fix the fts5vocab virtual table so that columns always behave as if they have blob affinity. Report 2026-08-15T07:03:55Z.
(check-in: db05b203f4 ...
user: dan tags: trunk)
|
| 12:35 |
|
In the CLI, fix the internal modeDup() routine so that it correctly handles
zIFmt and zFpFmt.
(check-in: aa5a515490 ...
user: drh tags: trunk)
|
| 12:22 |
|
Fix typos. Turn off -fpfmt and -ifmt using "auto" instead of "default".
(check-in: 530b55da56 ...
user: drh tags: trunk)
|
| 10:10 |
|
Refactor "realfmt"→"fpfmt" and "intfmt"→"ifmt".
(check-in: 17277813d1 ...
user: drh tags: trunk)
|
|
2026-08-14
| | |
| 18:49 |
|
Remove redundant offsetof definition from qrf.c.
(check-in: 5d230cafc6 ...
user: drh tags: trunk)
|
| 17:31 |
|
Enhance QRF with options to supply format strings used to render integers
and floating point numbers. Make those options accessible in the
shell using the ".mode --intfmt" and ".mode --realfmt" respectively.
Also make those options available to the "format" method of the sqlite3
object in the TCL interface.
(check-in: 61b0157686 ...
user: drh tags: trunk)
|
| 17:22 |
|
Add support for integer and float-point format strings to the
"format" method of the sqlite3 object in the TCL interface.
(closed check-in: 49778c334a ...
user: drh tags: qrf-num-format)
|
| 16:58 |
|
We don't actually need to increment the version number of QRF as
version 2 has not yet been released. Updates to the QRF README.md
file.
(check-in: 3ccc25a06f ...
user: drh tags: qrf-num-format)
|
| 16:34 |
|
Fix an off-by-one bug in --intfmt. Add some test cases.
(check-in: 4e86fc1922 ...
user: drh tags: qrf-num-format)
|
| 15:57 |
|
Initial code for an enhancement to QRF that allows for app-defined format
strings used to render integer and floating-point values.
(check-in: 067083e973 ...
user: drh tags: qrf-num-format)
|
| 15:17 |
|
Build portability improvements for buildtclext.tcl, from bug forum post 2026-08-13T19:40:39Z.
(check-in: 2530b5ada6 ...
user: stephan tags: trunk)
|
| 14:03 |
|
Adjust the bytecode generated for RIGHT JOIN so that it checks for duplicate
RHS rows when the LHS is NULL.
Bug 2026-08-14T04:56:02Z
(check-in: f4f57ca539 ...
user: drh tags: trunk)
|
| 13:07 |
|
Adjust the bytecode so that in a RIGHT OUTER JOIN, whenever a row from the
RHS table is added to the output set, the "match" index is updated to avoid
duplicate rows.
(closed check-in: 9d39b0973b ...
user: drh tags: 2026-08-14T04:56:02Z)
|
| 10:57 |
|
Fix the fts4aux virtual table so that it consistently behaves as if its columns have blob affinity. Report 2026-08-14T07:27:28Z.
(check-in: d5b57f76b6 ...
user: dan tags: trunk)
|
|
2026-08-13
| | |
| 18:55 |
|
Try to make test/testrunner.tcl work on other BSD variants. This change
is untested as I don't have access to any of those BSDs.
Bug 2026-08-13T17:55:42Z.
(check-in: c406ce3f1e ...
user: drh tags: trunk)
|
| 18:50 |
|
Disable the seekscan optimization for IN(...) expressions when the RHS is a SELECT that returns more than one column. This was previously only possible if the sqlite_stat1 table was populated with inconsistent information, and was malfunctioning in some circumstances. Report 2026-08-13T08:45:47Z.
(check-in: 601615fd00 ...
user: dan tags: trunk)
|
| 18:37 |
|
Disable the seekscan optimization for IN(...) expressions when the RHS is a SELECT that returns more than one column. This was previously only possible if the sqlite_stat1 table was populated with inconsistent information, and was malfunctioning in some circumstances. Report 2026-08-13T08:45:47Z.
(closed check-in: 9544d46a6d ...
user: dan tags: seekscan-fix)
|
| 17:49 |
|
Disable misuse testing unless the ENABLE_MISUSE environment variable
is set.
Bug 2026-08-13T13:27:50Z
(check-in: 273e1bcc2c ...
user: drh tags: trunk)
|
| 16:10 |
|
The OP_Affinity opcode with argument E should not try to preserve IntReal
but should cast the value to a pure Real.
Bug 2026-08-13T08:12:13Z.
(check-in: 5a4e6920af ...
user: drh tags: trunk)
|
| 15:16 |
|
Ensure that if a seek-scan is used for an expression like (x, y) IN (SELECT a, b FROM ...) then the SELECT delivers rows in order of (a, b).
(check-in: c9661d928e ...
user: dan tags: seekscan-fix)
|
| 14:05 |
|
Log the OOM stack trace rather than printing on stderr.
(leaf check-in: b8aa7d02ca ...
user: drh tags: oom-backtrace)
|
| 11:29 |
|
Fix a problem in the fts4aux virtual table causing it to sometimes return spurious SQLITE_CORRUPT errors for fts4 tables with more than 1 column. Report 2026-08-13T05:56:36Z.
(check-in: 06f3f9f93b ...
user: dan tags: trunk)
|
| 00:21 |
|
Disable recently added test cases for immutable=1 when running
the inmemory_journal permutation.
(check-in: 198e602053 ...
user: drh tags: trunk)
|
|
2026-08-12
| | |
| 20:26 |
|
Fix an MSVC compiler warning in the amatch.c extension.
(check-in: 43b23132a1 ...
user: drh tags: trunk)
|
| 20:17 |
|
Simplification and size reduction for sqlite3GetVarint(). Add and use a new
routine sqlite3VarintValue() that works similarly but returns the value
rather than the number of bytes consumed, for a performance improvement.
Improvements inspired by [48132f815508b6e1].
(check-in: 2d7ac2095e ...
user: drh tags: trunk)
|
| 20:04 |
|
Simplify the implementation of sqlite3GetVarint().
(closed check-in: d3dfaeb8b9 ...
user: drh tags: varint-decode)
|
| 19:42 |
|
Create a new varint decoder that does not need to return the size of the
varint, and use that in sqlite3BtreeTableMoveto(). Omit the pointless
getVarint and putVarint macros and use the actual function names instead.
(check-in: f845ede54c ...
user: drh tags: varint-decode)
|
| 15:55 |
|
Compile with -DHAVE_BACKTRACE to get a backtrace() dump to stderr on
the first OOM of each statement.
(check-in: 6d9c85d4dd ...
user: drh tags: oom-backtrace)
|
| | |