SQLite

Timeline
Login

Many hyperlinks are disabled.
Use anonymous login to enable hyperlinks.

100 descendants and 100 ancestors of 4e54e9c0fefe417f94a2d95921acf61b1dda5ca4

2016-03-08
17:59
In memjournal.c, reuse the same sqlite3_file object for both the in-memory phase and the on-disk phase. (check-in: e7fbbdc2 user: drh tags: trunk)
17:44
Modify the memjournal.c code to make it a bit smaller. (Closed-Leaf check-in: d99ac415 user: dan tags: memjournal-exp)
16:35
Merge changes from trunk, especially the SQLITE_DEFAULT_SYNCHRONOUS enhancements. (check-in: 29741941 user: drh tags: apple-osx)
16:07
Drop support for SQLITE_EXTRA_DURABLE. The new SQLITE_DEFAULT_SYNCHRONOUS compile-time option is a more general replacement. (check-in: f6d3156b user: drh tags: trunk)
16:02
Remove an unused goto label in the RBU extension. (check-in: 2179a106 user: drh tags: trunk)
15:52
Add a new row type to RBU (a peer of insert, update and delete) - "delete then insert". (check-in: 1d9468d2 user: dan tags: trunk)
15:47
Add the SQLITE_DEFAULT_SYNCHRONOUS and SQLITE_DEFAULT_WAL_SYNCHRONOUS compile-time options. Automatically switch to the WAL_SYNCHRONOUS setting when first opening a WAL-mode database if the synchronous setting has not been previously set by the application. (check-in: 5a847a67 user: drh tags: trunk)
15:37
Update fts3/4 so that the 'merge=X,Y' command merges at least, instead of exactly, Y segments from a single level. This matches the documentation. 'merge=X,0' is, as it was in 3.11, an error. (check-in: 64b3cb29 user: dan tags: trunk)
15:30
Make the SQLITE_DEFAULT_SYNCHRONOUS and SQLITE_DEFAULT_WAL_SYNCHRONOUS values zero-based to agree with PRAGMA synchronous. (Closed-Leaf check-in: 592d2104 user: drh tags: default-synchronous)
15:14
If SQLITE_DEFAULT_WAL_SYNCHRONOUS is not the same value as SQLITE_DEFAULT_SYNCHRONOUS and the application has not run "PRAGMA synchronous", then set synchronous to the SQLITE_DEFAULT_WAL_SYNCHRONOUS setting when entering WAL mode for the first time. (check-in: 5791407b user: drh tags: default-synchronous)
14:40
Add compile-time options SQLITE_DEFAULT_SYNCHRONOUS and SQLITE_DEFAULT_WAL_SYNCHRONOUS used to specify the default synchronous settings for all database connections. (check-in: 1fefa967 user: drh tags: default-synchronous)
14:16
Fix ATTACH to use the symbolic name PAGER_SYNCHRONOUS_FULL rather than an integer literal. (check-in: c4e192a0 user: drh tags: trunk)
14:01
Merge coverage improvements and comment fixes from trunk. (check-in: 58023bfc user: drh tags: apple-osx)
13:56
Fix comments on pager flag settings to include synchronous=EXTRA. (check-in: 3a65a1fc user: drh tags: trunk)
02:59
Simplified error detection in the xBestIndex processing. (check-in: 82ca2131 user: drh tags: trunk)
01:32
Add the ALLBITS macro as a shorthand for "(Bitmask)(-1)". (check-in: 91bd619d user: drh tags: trunk)
01:11
Add a NEVER() to the code generator. Change the parameter name "mExtra" to "mPrereq" in the query planner, to more closely reflect its meaning. (check-in: 721ae51e user: drh tags: trunk)
00:39
Changes so that some assert()s in the virtual table query planner are correct even following an OOM error. (check-in: 9805f6f8 user: drh tags: trunk)
2016-03-07
21:28
Fix a typo and remove an unused variable from test logic. (check-in: 9f31516c user: drh tags: apple-osx)
20:48
Merge recent enhancements from trunk. (check-in: 84f09f00 user: drh tags: apple-osx)
20:14
Update fts3/4 so that the 'merge=X,0' command merges X pages from all segments of the first level in the fts index that contains 2 or more segments. (check-in: cddf69db user: dan tags: trunk)
19:08
Avoid a NULL pointer dereference following an OOM while generating code for IN operators on virtual tables. (check-in: c9240086 user: drh tags: trunk)
17:49
Merge the virtual table query planner enhancement, the RTREE cost estimate fix, and the statement journal spill delay enhancement from trunk. (check-in: 17fd8f3c user: drh tags: sessions)
17:39
Add a new row type to RBU (a peer of insert, update and delete) - "delete then insert". (Closed-Leaf check-in: 169311c8 user: dan tags: rbu-delete-then-insert)
17:37
Add the SQLITE_CONFIG_STMTJRNL_SPILL option for sqlite3_config(). (check-in: b6c42024 user: drh tags: trunk)
2016-03-05
21:32
Fix a problem in the previous commit affecting queries with three or more tables joined together to the right of a LEFT or CROSS JOIN operator. (check-in: d8d89d69 user: dan tags: trunk)
21:07
Change the way SQLite invokes the xBestIndex method of virtual tables so that N-way joins involving virtual tables work as expected. (check-in: 19544438 user: dan tags: trunk)
20:28
Adjust the cost estimates for RTREE such that the expected number of rows and expected cost is halved for each additional constraint. (Closed-Leaf check-in: fd7cd060 user: drh tags: xbestindex-fix)
17:29
Change the way SQLite invokes the xBestIndex method of virtual tables so that N-way joins involving virtual tables work as expected. (check-in: ffc65968 user: dan tags: xbestindex-fix)
15:35
Remove the unused journal.c source file (its function have been subsumed into memjournal.c). Refactor some of the names in memjournal.c. No functional changes. (check-in: 5f2a262d user: drh tags: trunk)
15:03
Work around (incorrect) ASAN warnings in memjournal.c. (check-in: 4de09777 user: drh tags: trunk)
14:19
Fix an integer size mismatch problem in test_bestindex.c (check-in: 2e35eb6b user: drh tags: trunk)
14:00
Fix an incorrect #ifdef on sqlite3LogEstToInt(). (check-in: dca7b233 user: drh tags: trunk)
2016-03-04
21:18
Fix an assert() in sqlite3VarintLen(), even though it is impossible to hit in SQLite due to the way sqlite3VarintLen() is used. (check-in: 251424c5 user: drh tags: trunk)
19:55
Simplify the computation of Index.aAvgEq. (Leaf check-in: c4488730 user: drh tags: analyze-worst-case)
18:45
Merge changes from trunk. (check-in: 5294c977 user: drh tags: analyze-worst-case)
16:42
Merge recent enhancements from trunk. Default page size is 4096. Writes to statement journals are avoided. (check-in: 456df336 user: drh tags: sessions)
14:57
Merge recent enhancements from trunk, and especially the changes that reduce the heap-memory footprint of schemas, and defer opening and writing to statement journals. (check-in: 2f0c195c user: drh tags: apple-osx)
14:43
Defer opening and writing statement journals until the size reaches a threshold (currently 64KiB). (check-in: cb9302cc user: drh tags: trunk)
14:23
Update test cases to taken deferred statement-journal opening into account. (Closed-Leaf check-in: 5b2fe521 user: drh tags: memjournal-exp)
13:08
Merge changes from trunk. (check-in: a87305df user: drh tags: memjournal-exp)
04:01
Change the default cache_size to -2000 (which means 2000*1024 bytes independent of page_size). (check-in: 2682e8e4 user: drh tags: trunk)
03:43
Change the default page size for new database files to 4096 (from 1024). Except, when building the testfixture, preserve the legacy page size. Also fix a comment on SQLITE_MAX_ATTACHED. (check-in: 2e8a9ca9 user: drh tags: trunk)
03:20
Detect the presence of pread()/pwrite() system calls on unix systems and use them if available. (check-in: 82cbf5ad user: drh tags: trunk)
03:14
Further fixes to the pread()/pwrite() enabling logic. (Closed-Leaf check-in: 38fb54e7 user: drh tags: pread)
03:02
Fix the build for cases when pread()/pwrite() are not available. (check-in: 7d67d876 user: drh tags: pread)
02:59
Add the test_bestindex.c file to Makefile.in and Makefile.msc so that "make test" will work again. (check-in: 6ec6b6a9 user: drh tags: trunk)
02:38
Update the configure script to detect pread/pwrite and update os_unix.c to use those routines if they are available. (check-in: 2cffb9e5 user: drh tags: pread)
00:13
Get in-memory subjournals working for builds without SQLITE_ENABLE_ATOMIC_WRITE. (check-in: b5378dce user: drh tags: memjournal-exp)
2016-03-03
21:29
Merge the latest updates from trunk. (check-in: 55c00f71 user: drh tags: memjournal-exp)
21:22
Reduce the amount of heap memory required to store many schemas by storing the column datatype appended to the column name (with a \000 separator), rather than in separate memory allocation. (check-in: 16fbf2e1 user: drh tags: trunk)
20:42
Allow the left-hand side of IN operators on virtual tables to have the aConstraintUsage[].omit flag clear. (check-in: 3eb74342 user: drh tags: trunk)
17:54
Merge 3.11.1 changes into trunk. (check-in: 7abc2dd9 user: drh tags: trunk)
16:17
Version 3.11.1 (Leaf check-in: f047920c user: drh tags: release, branch-3.11, version-3.11.1)
15:16
Increase the default upper bound on scope to 30. (check-in: b1258814 user: drh tags: branch-3.11)
2016-03-02
17:57
Add new test script bestindex1.test to the utf16 permutation. (Closed-Leaf check-in: 5893e972 user: dan tags: vtab-IN-opt)
16:13
Add an extra test for the change on this branch. (check-in: d2d28251 user: dan tags: vtab-IN-opt)
16:01
Enhance test_bestindex.c so that it can be used to test plans generated by xBestIndex. (check-in: 3c15a9bf user: dan tags: trunk)
15:37
Fix a recently introduced problem in permutations.test causing an error when the QUICKTEST_OMIT environment variable is defined. (check-in: 45bc9557 user: dan tags: branch-3.11)
13:26
Fix an error in an fts5 test script. (check-in: b2a03e2d user: dan tags: branch-3.11)
03:28
Allow the left-hand side of IN operators on virtual tables to have the aConstraintUsage[].omit flag clear. (check-in: 1622623c user: drh tags: vtab-IN-opt)
00:58
Add the optional -DSERIES_OMIT_CONSTRAINT_VERIFY=0 option to the series.c extension that implements the generate_series() virtual table. (check-in: 3d9daa92 user: drh tags: trunk)
2016-03-01
22:48
New test cases and infrastructure for testing the xBestIndex method of virtual tables. (check-in: 1d41c161 user: drh tags: trunk)
22:41
Improved debugging output with wheretrace. Fix some typos in test script comments. (check-in: 13a37fd4 user: drh tags: trunk)
18:35
Fix a memory leak in the test code on this branch. (Closed-Leaf check-in: 7a1add56 user: dan tags: test-bestindex)
18:24
Allow test_bestindex.c to set the omit flag for a constraint. (check-in: 759b9d5b user: dan tags: test-bestindex)
18:07
Add test code useful for testing the planners use of teh virtual table xBestIndex() method. (check-in: de034c0d user: dan tags: test-bestindex)
16:56
Do not attempt to run fts5bigtok.test or fts5merge2.test if FTS5 is not compiled in. (check-in: f7ed3739 user: dan tags: branch-3.11)
16:02
Add -DSQLITE_ENABLE_FTS5 to one of the release-test configurations in releasetest.tcl. Run fts5 tests as part of "make test" if SQLITE_ENABLE_FTS5 is defined. (check-in: 3cee9371 user: dan tags: branch-3.11)
15:52
Remove a forgotten "breakpoint" command from an FTS5 test script. (check-in: 61927c8f user: drh tags: branch-3.11)
15:41
Increase the version number to 3.11.1 (check-in: 2dcd2fda user: drh tags: branch-3.11)
15:09
Merge branch-3.11-matchinfo into this branch. (check-in: 42358170 user: dan tags: branch-3.11)
14:31
Change the estimated row counts in stat1 to be one-third worst-case and two-threads average case. (check-in: 21bfd47c user: drh tags: analyze-worst-case)
12:45
Fix test cases to align with the improved stats computation. (check-in: 810967bf user: drh tags: analyze-worst-case)
02:11
Better estimatedCost in the xBestIndex method of the generate_series vtab. (check-in: f2c16094 user: drh tags: trunk)
2016-02-29
23:02
Improvements to the logic for adding the "noskipscan" flag to stat1 entries. (check-in: 421b5b54 user: drh tags: analyze-worst-case)
21:27
The ANALYZE command automatically appends "noskipscan" to sqlite_stat1 entries that have large worst-case repeat estimates but small average repeat estimates. (check-in: 6326ba58 user: drh tags: analyze-worst-case)
20:18
When using a temporary file for a statement journal, store the first 64KiB in memory. If the file grows larger than that, flush it to disk and free the memory. Hardcoding to 64KiB is just an experiment to check that the memjournal.c code works. (check-in: 44b2dc18 user: dan tags: memjournal-exp)
20:00
Fix minor problems with new code in memjournal.c. (check-in: 9fd3f7b9 user: dan tags: memjournal-exp)
18:30
Modify the ANALYZE command to store worst-case statistics in sqlite_stat1, rather thn average case. (check-in: 5a0143c9 user: drh tags: analyze-worst-case)
17:34
Fix a problem in fts5 where a corrupt db could lead to a (huge) buffer overread. (check-in: c9a30e11 user: dan tags: trunk)
17:16
Fix an fts5 problem causing 'optimize' to corrupt the fts index under some circumstances. (check-in: 251d6473 user: dan tags: trunk)
15:53
Reduce the amount of heap required to store many schemas by storing each column datatype appended to the column name, rather than as a separate allocation. (Closed-Leaf check-in: 842b2116 user: drh tags: schema-storage)
13:44
Very minor improvement to the performance and reduction in size to the parser by capturing the name and datatype of table columns in a single grammar rule reduction. (check-in: 4b55c520 user: drh tags: trunk)
13:37
Further simplifications to the grammar. (Closed-Leaf check-in: 519fd03b user: drh tags: schema-storage)
2016-02-27
23:25
Update the parser so that it pulls out the column name and type all in one go, rather than using separate reductions. (check-in: ad3ffe2e user: drh tags: schema-storage)
21:16
Eliminate the need for the Column.zDflt (using Column.pDflt instead) to reduce the amount of memory needed to hold the schema. (check-in: d8c94a46 user: drh tags: trunk)
20:14
Extend the code in memjournal.c so that it subsumes the role of journal.c. And (untested) can flush journal or statement journal files to disk after they grow to a specified size. (check-in: e0b0b431 user: dan tags: memjournal-exp)
19:19
Tighter description of I/O stats in the shell. Show I/O stats on speedtest1.c. (check-in: f681d800 user: drh tags: trunk)
19:03
More compact implementation of PRAGMA onconflict, and some test cases. (Leaf check-in: 3e5d38f5 user: drh tags: pragma-onconflict)
17:16
Experimental "PRAGMA onconflict=FAIL" statement to change the default ON CONFLICT algorithm to something other than ABORT. (check-in: 3a0c347c user: drh tags: pragma-onconflict)
17:12
Enhance the ".stats" command in sqlite3.exe to show one-time stats information if invoked with one argument. Also show /proc/PID/io information if run on Linux. (check-in: 3c36948f user: drh tags: trunk)
14:04
Fix a requirement mark to conform to a typo fix in the documentation. (check-in: d1392c62 user: drh tags: trunk)
14:00
Enhance documentation of sqlite3_snapshot_open() to explain that the database connection must have participated in at least one read operation prior to the beginning of the transaction for which the snapshot is to be opened. Add test cases for this fact. (check-in: 33dd671c user: drh tags: trunk)
00:21
Add 'Replace.exe' to the MSVC clean target. (check-in: c2277fab user: mistachkin tags: trunk)
00:21
Fix typo in ICU error message. (check-in: 00c8fffd user: mistachkin tags: trunk)
00:07
Make sure the Replace.cs file is included in the autoconf tarball. (check-in: b850df48 user: drh tags: trunk)
2016-02-26
23:13
Support generating the '.def' file using the autoconf Makefile for MSVC. (check-in: 3617564c user: mistachkin tags: trunk)
21:20
Fix a potential buffer overflow in the ICU upper() function. (check-in: b8dc1b9f user: drh tags: trunk)
21:03
Further refinements to the MSVC batch build process. (check-in: cf4e4fbd user: mistachkin tags: trunk)
21:01
Further refinements to the MSVC batch build process. (check-in: 4e54e9c0 user: mistachkin tags: branch-3.11)
2016-02-19
19:29
Use the arguments to NMAKE consistently in the MSVC batch build tool. (check-in: 480e5049 user: mistachkin tags: branch-3.11)
19:03
Enhance clean target in the MSVC makefiles. (check-in: 66a9ad4a user: mistachkin tags: branch-3.11)
05:07
Modify the MSVC batch build tool to support the MSVC autoconf Makefile. (check-in: 72b10e5e user: mistachkin tags: branch-3.11)
00:47
Further improvements to the MSVC batch build tool. (check-in: 90dd6eb1 user: mistachkin tags: branch-3.11)
2016-02-18
21:00
Correct the install location for the UWP VSIX package. (check-in: b47d0302 user: mistachkin tags: branch-3.11)
2016-02-15
17:29
Version 3.11.0 (check-in: 3d862f20 user: drh tags: trunk, release, version-3.11.0)
15:06
Make sure the codec is invoked when overwriting existing frames of the WAL file. Yikes! (check-in: f694e60a user: drh tags: trunk)
2016-02-13
18:54
Provide Sqlite3_SafeInit() and Sqlite3_SafeUnload() entry points on the TCL interface, but have the always return TCL_ERROR, because the non-standard TCL builds on Macs require this. (check-in: 37ec3015 user: drh tags: trunk)
2016-02-12
18:48
Fix a potential buffer overread provoked by invalid utf-8 in fts5. (check-in: a049fbbd user: dan tags: trunk)
17:56
Fix a fairly obscure buffer overread in fts5. (check-in: 13058020 user: dan tags: trunk)
17:30
Fix a documentation typo. No changes to code. (check-in: d9c98587 user: drh tags: trunk)
17:27
Enhance mptester so that it takes great care to delete its test database prior to start-up, including checking the return code from unlink() and retrying after a delay if unlink() fails. Hopefully this will clear intermittant startup problems on Windows. (check-in: 6ea84ec1 user: drh tags: trunk)
05:19
Naming updates for Universal Windows Platform. (check-in: 717c1fc4 user: mistachkin tags: trunk)
00:13
Remove an unused parameter from whereSortingCost(). (check-in: b5d77199 user: drh tags: trunk)
2016-02-11
22:41
Automatically disable FTS3 and FTS4 when building with SQLITE_OMIT_VIRTUALTABLE. (check-in: 0beb32d2 user: drh tags: trunk)
21:59
Enhance the MSVC makefile to enable building 'testfixture' fully from source code. (check-in: f183e059 user: mistachkin tags: trunk)
21:55
Remove the unexplained extra 3.0 cost factor for the B-tree sorting that is done to implement DISTINCT in the absence of an index to help. (check-in: 52571991 user: drh tags: trunk)
21:53
Remove the unused fts5BlobCompare() routine. (check-in: defc762d user: drh tags: trunk)
21:38
Prevent 'expanded command line X too long' errors during the batch 'for' loops used for copying files. (Closed-Leaf check-in: 67ee9b5a user: mistachkin tags: testFixtureSrc)
21:28
Enhance the MSVC makefile to enable building 'testfixture' fully from source code. (check-in: 54ff3a26 user: mistachkin tags: testFixtureSrc)
18:18
Fix another unused variable warning in fts5 code. (check-in: 61b4c120 user: dan tags: trunk)
18:08
Avoid a buffer overread when reading a corrupt fts5 structure record. (check-in: facbc424 user: dan tags: trunk)
17:01
Handle parser stack overflow when parsing fts5 query expressions. Fix some compiler warnings in fts5 code. (check-in: bc3f7900 user: dan tags: trunk)
15:37
Add JSON1 and FTS5 to the set of extensions subject to close compiler warning analysis. Fix some warnings in each. More (harmless) warnings still exist in FTS5. (check-in: cfe2eb88 user: drh tags: trunk)
2016-02-10
21:45
In the MSVC makefile, include the working directory so that it can pickup build generated files (e.g. sqlite3.h, etc). (check-in: e9e6000b user: mistachkin tags: trunk)
19:10
Add testcase() macros to the CHECK constraint avoidance logic. Avoid creating an unused VDBE label during CHECK constraint code generation. (check-in: 970881be user: drh tags: trunk)
18:24
Fix to previous check-in: Make sure CHECK constraints involving the ROWID are not ignored when the ROWID changes. (check-in: 7782cb1d user: drh tags: trunk)
16:52
Omit unnecessary CHECK constraints in UPDATE statements, when none of the columns referenced in the CHECK constraint are modified. (check-in: 02fbdbc7 user: drh tags: trunk)
16:03
Omit NOT NULL checks on unchanging columns in an UPDATE. (check-in: 6a3aaedf user: drh tags: trunk)
13:36
When generating the snapshot-tarball, truncate the date/time in the name to 12 significant digits (YYYYMMDDhhmm) omitting the seconds and fractional seconds. (check-in: 604f7775 user: drh tags: trunk)
13:17
Updates to the autoconf tarball README.txt file. (check-in: a3e911e3 user: drh tags: trunk)
2016-02-09
22:54
Update the snapshot tarball name to use the date (ISO8601 format) instead of the SHA1 hash in its name. (check-in: 070ec66f user: drh tags: trunk)
22:39
Add the new "snapshot-tarball" target to the unix makefiles, for building a tarball similar to "amalgamation-tarball" but named by the SHA1 hash of the current check-out rather than by the version number. (check-in: 4f360b2a user: drh tags: trunk)
22:00
Update the MSVC autoconf makefile with the recent changes from the primary one. (check-in: c7242aa3 user: mistachkin tags: trunk)
20:37
Fix the makefiles in the autoconf-tarball so that they build the command-line shell with SQLITE_ENABLE_EXPLAIN_COMMENTS. (check-in: 1135ad01 user: drh tags: trunk)
20:31
In the shell, use the appropriate string constant instead of a literal new-line. (check-in: 61951b15 user: mistachkin tags: trunk)
20:11
Enhance the command-line shell to be in auto-explain mode by default. It is no longer necessary to use the ".explain" command to put the shell into a mode where the EXPLAIN output is formatted nicely. That now happens automatically. (check-in: 751915cb user: drh tags: trunk)
20:04
Improved auto-detection of EXPLAIN output in the shell. (Closed-Leaf check-in: 6c6d7a6e user: drh tags: auto-explain)
18:39
Add auto-explain mode to the command-line shell. Default on. Auto-explain tries to automatically detect EXPLAIN queries and format them appropriately. (check-in: 1d62aa6b user: drh tags: auto-explain)
18:28
Fix MSVC makefile options that enable control-flow guard. (check-in: 51b6823f user: mistachkin tags: trunk)
17:12
Fix the top-level makefiles to always include SQLITE_ENABLE_EXPLAIN_COMMENTS when building the shell. Still need to fix main.mk and the amalgamation tarball. (check-in: 99228239 user: drh tags: trunk)
16:09
Code simplification: ALTER TABLE ADD COLUMN always upgrades the file_format to 4 if is not there already. No need to upgrade to only 2 or 3 since format 4 has now been supported for over 10 years. (check-in: e1d8ec85 user: drh tags: trunk)
02:12
Make sure every co-routines has its own set of temporary registers and does not share temporaries, since a co-routine might expect the content of a temporary register to be preserved across an OP_Yield. Proposed fix for ticket [d06a25c84454a]. (check-in: ca72be86 user: drh tags: trunk)
2016-02-08
20:45
Fix spelling error in MSVC makefile comments. (check-in: 6eab74c9 user: mistachkin tags: trunk)
20:40
More work on Windows 10 SDK integration. (check-in: 1e563c6e user: mistachkin tags: trunk)
20:14
Fix a signed/unsigned comparison warning in fts5. (check-in: 85eb05dd user: dan tags: trunk)
20:06
Fix a compiler warning when compiling without SQLITE_ENABLE_FTS3_TOKENIZER. (check-in: 6926f28c user: drh tags: trunk)
19:40
Disable the two-argument form of the fts3_tokenizer() SQL function unless the library is built with -DSQLITE_ENABLE_FTS3_TOKENIZER. (check-in: e0eb217a user: dan tags: trunk)
19:36
Changes to help the tokenizer run about 33% faster. (check-in: a050e6f0 user: drh tags: trunk)
19:15
Add code to get the tokenizer character-class logic working for EBCDIC. (Closed-Leaf check-in: 04f7da77 user: drh tags: tokenizer-char-class)
03:23
Faster keywordCode() implementation by taking advantage of the fact that the input is always pure ASCII alphabetic and underscore and that the keyword table is always upper-case. (check-in: ff406b97 user: drh tags: tokenizer-char-class)
02:30
Demonstrate a much faster sqlite3GetToken() routine by using a lookup table to map initial token characters into a character class. This check-in does not work for EBCDIC. More optimization needed. (check-in: 9115baa1 user: drh tags: tokenizer-char-class)
2016-02-07
20:39
Fix harmless compiler warning in 'srcck1' tool and add it to the clean targets. (check-in: 852a529a user: mistachkin tags: trunk)
00:08
Add the sourcetest target to Makefile.msc. (check-in: ab269e72 user: drh tags: trunk)
2016-02-06
22:32
Add a utility program that looks for assert(), NEVER(), ALWAYS(), and testcase() macros that have side-effects, and reports errors when they are found. Also fix a bug that this utility detected as it was being tested. (check-in: b0b4624f user: drh tags: trunk)
19:48
Make sure variable declarations occur at the beginning of blocks, even with SQLITE_DEBUG enabled. (check-in: 2f7778e6 user: drh tags: trunk)
14:14
Fix an assert() to have a test instead of a side effect (check-in: a2952231 user: pdr tags: trunk)
2016-02-05
21:09
Add tests for and remove unreachable branches from fts5 in order to restore test coverage. (check-in: 22589018 user: dan tags: trunk)
19:40
More work on Windows 10 SDK integration. (Closed-Leaf check-in: ebace2c9 user: mistachkin tags: win10sdk)
19:18
Further streamlining of fts5 prefix query code. (check-in: ca91bd8a user: dan tags: trunk)
17:49
Make sure the "bak.db" database file does not actually exist before starting the "quota.test" testing. (check-in: 1cac6c45 user: drh tags: trunk)
16:46
From test scripts, remove stray "breakpoint" commands and blank lines at the ends of files. (check-in: a6b35273 user: drh tags: trunk)
14:11
Improvements to the way out-of-memory conditions are handled, resulting in an almost 1% performance gain in the commmon case where OOM errors do not occur. (check-in: b082538d user: drh tags: trunk)
13:38
Add the slightly faster sqlite3DbMallocRawNN(db,n) routine for the majority cases where db is guaranteed to be not NULL. (Closed-Leaf check-in: 0a802e96 user: drh tags: oom-handling)
04:55
Improve ERRORLEVEL handling in the MSVC batch build tool sub-routine 'fn_UnsetVariable'. (check-in: fe123ab4 user: mistachkin tags: trunk)
04:50
OOM failures on sqlite3_errmsg16() in an app-defined function are no longer benign. (check-in: 9efb223f user: drh tags: oom-handling)
04:39
Another small tweak to the MSVC batch build tool. (check-in: 76718bbc user: mistachkin tags: trunk)
02:50
Minor simplification of schema error reporting. (check-in: a8846aee user: drh tags: oom-handling)
02:42
Further performance improvements that fall out of the mallocFailed changes. (check-in: d94416dd user: drh tags: oom-handling)
01:55
Improvements to the way that OOM errors are processed. (check-in: c3ef0347 user: drh tags: oom-handling)
2016-02-04
19:50
Temporarily back out the 0.5% performance improvement from check-in [632071bac5ff32]. Need a more elaborate solution that works with reentrant virtual tables and SQL functions. (check-in: 42736fb0 user: drh tags: trunk)
19:45
Further improve performance of unindexed fts5 prefix queries. (check-in: c9c6457d user: dan tags: trunk)
17:31
Avoid running some particularly time-consuming tests as part of veryquick.test. (check-in: f465944b user: dan tags: trunk)
11:48
Remove unnecessary sets of db->mallocFailed. (check-in: b787165b user: drh tags: trunk)
11:15
Use sqlite3_malloc64() instead of sqlite3_malloc() in the spellfix extension. (check-in: 634d008c user: drh tags: trunk)
10:28
Escape control characters in JSON. Fix for ticket [ad2559db380abf8]. (check-in: 4f1b5229 user: drh tags: trunk)
09:48
Refinements to synchronous logic: (1) Use PAGER_SYNCHRONOUS_FULL rather than the corresponding magic number. (2) Honor SQLITE_NO_SYNC on xDelete calls with sync (3) Count xDelete syncs during testing (4) Fix #ifs on SQLITE_EXTRA_DURABLE so that directory syncs on journal unlink are off by default. (check-in: e3157cb5 user: drh tags: trunk)
2016-02-03
22:14
In the VDBE loop, only check for OOM errors at jumps rather than after every opcode, for about a 0.5% performance increase. (check-in: 632071ba user: drh tags: trunk)
20:04
Improve performance of fts5 prefix queries on detail=col tables. (check-in: ca11f46d user: dan tags: trunk)
19:52
Fix markup errors in comments used to generate the documentation - specifically in the documentation on the OP_Seek opcode. (check-in: ef252bc4 user: drh tags: trunk)
19:20
Add "PRAGMA synchronous=EXTRA" which syncs the directory containing the rollback journal when the rollback journal is deleted in order to commit a transaction in DELETE mode. (check-in: af924018 user: drh tags: trunk)
01:55
Enhance the internal sqlite3VdbeAddOpList() interface to automatically update jump destinations. Use this feature to simplify the AUTOINCREMENT code generator. (check-in: ae8b9d2e user: drh tags: trunk)
2016-02-02
21:19
Add tests to restore full coverage of fts5 code. (check-in: 063755c8 user: dan tags: trunk)
17:40
Enhance the performance of fts5 AND and OR queries. (check-in: 62ea9e5a user: dan tags: trunk)
02:04
Enhance the comment on the sqlite3_index_constraint object to bring attention to the fact than iColumn field can be negative for a rowid. (check-in: d8b7b199 user: drh tags: trunk)
2016-02-01
21:48
Change the OP_SetCookie instruction to write the literal P3 value, not the value in register P3. (check-in: 6d7d4703 user: drh tags: trunk)
20:12
Improve performance of fts5 queries. (check-in: 2334e882 user: dan tags: trunk)
17:20
Number VDBE opcodes starting with 0 instead of 1, as this obviates the lower-bound test on "switch(opcode){...}", making the code smaller and faster. (check-in: 4c9222f7 user: drh tags: trunk)
16:36
Update walcrash.test to ensure that, during a particular test, enough data is written to cause SQLite to sync the wal file 14 times. (check-in: 5d7c0928 user: dan tags: trunk)
13:58
Fix a problem causing the OR/covering-index optimization to be disabled if compile time parameter SQLITE_MAX_ATTACHED were set to greater than 30. (check-in: a17712bf user: dan tags: trunk)
13:21
Minor simplification to the OP_AutoCommit opcode. Fix some code comments related to transaction control. (check-in: a9b6a067 user: drh tags: trunk)
00:21
Fix an assert() that might fail on a corrupt database. (check-in: 22eaced7 user: drh tags: trunk)
2016-01-30
21:09
Fix new test cases in stat.test so that they work with -DSQLITE_DEFAULT_AUTOVACUUM=1 builds. (check-in: a2810cf6 user: dan tags: trunk)
19:16
Performance enhancement for fts5 column filter queries on detail=full tables. (check-in: 13fb4aa6 user: dan tags: trunk)
16:59
Merge the implementation of OP_IdxRowid and OP_Seek so that OP_Seek no longer requires the rowid register and a separate OP_IdxRowid call. Shorter and faster prepared statements result. (check-in: 9bec50a1 user: drh tags: trunk)
15:52
Only honor the register subtype value if the MEM_Subtype flag is set. Revised and improved fix for ticket [f45ac567eaa9f9]. (check-in: 1f4c667f user: drh tags: trunk)
15:46
A different way to clear the subtype on VDBE registers when the value of the register is overwritten with new content. To fix [[f45ac567eaa9f93]. (Closed-Leaf check-in: 2e9fb129 user: dan tags: clear-subtype-flag)
14:17
Output subtype information on register traces for PRAGMA vdbe_trace. (check-in: 96b78020 user: drh tags: trunk)
13:32
Add the BTREE_FORDELETE and BTREE_AUXDELETE flags to the b-tree layer interface and use them. Add assert() statement to verify that they are correct. (check-in: 85c46704 user: drh tags: trunk)
12:50
Simplification and size reduction to the printf logic. Remove the bFlags parameter from sqlite3VXPrintf() and sqlite3XPrintf(). Use sqlite3XPrintf() instead of sqlite3_snprintf() for rendering P4 values in EXPLAIN output. (check-in: 0bdb41c4 user: drh tags: trunk)