Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
96 ancestors of d8d3e6d04cbb9e30
|
2014-12-12
| ||
| 01:27 | Add new tests to the threadtest4.c program. Fix a long-standing data race in WAL mode for shared-cache. (check-in: d8d3e6d0 user: drh tags: trunk) | |
| 00:52 | Remove the KeyInfo cache (for now - perhaps we will add it back in later - or maybe not since it provides negligible benefit but adds a lot of complexity and thread-safety risk). Add a mutex to ATTACH to deal with a data race. (check-in: 03c443ea user: drh tags: trunk) | |
| 00:40 | Make sure the Btree mutex is held when setting the locking mode and the secure delete flag when attaching a shared-cache database. (Closed-Leaf check-in: 6bef7ede user: drh tags: threadtest4) | |
| 00:26 | Merge the cell overflow page number cache thread race fix from trunk. (check-in: cefad47e user: drh tags: threadtest4) | |
| 00:20 | Fix a bug in the threadtest4.c program. Remove the keyinfo cache as it provides minimal performance improvements, and then only at SQL preparation time, not at runtime, and it has problems with data races in shared-cache mode. We might later add the keyinfo cache back but only enable it when shared-cache mode is off. (check-in: b7489f94 user: drh tags: threadtest4) | |
|
2014-12-11
| ||
| 19:29 | Add the threadtest4.c test program. Not yet working. (check-in: ec3a7446 user: drh tags: threadtest4) | |
| 16:38 | Fix a race condition to do with very large index keys in shared-cache mode. (check-in: fc157dd7 user: dan tags: trunk) | |
| 15:27 | Fix a typo in the documentation for sqlite3_threadsafe(). (check-in: 258e747b user: drh tags: trunk) | |
|
2014-12-10
| ||
| 20:57 | Fix a typo causing a test error in e_walhook.test. (check-in: d9f916ba user: dan tags: trunk) | |
| 20:29 | Add new test file e_walhook.test. (check-in: 2eb6d3e4 user: dan tags: trunk) | |
| 17:34 | Revise mutex handling by the sqlite3_win32_reset_heap() function. (check-in: eacb3b7b user: mistachkin tags: trunk) | |
| 04:58 | Fix typos in the documentation for sqlite3_table_column_metadata(). No changes to code. (check-in: 3528f8dd user: drh tags: trunk) | |
|
2014-12-09
| ||
| 22:24 | Fix the sqlite3_table_column_metadata() routine so that it gives the correct answer for the "rowid" column in a WITHOUT ROWID table. Enhance it so that it can be used to check for the existence of a table by setting the column name parameter to NULL. The routine is now included in the build by default, even without the SQLITE_ENABLE_COLUMN_METADATA compile-time option. (check-in: cf9be419 user: drh tags: trunk) | |
| 20:13 | Add extra tests to e_walckpt.test. (check-in: 84f95810 user: dan tags: trunk) | |
| 19:16 | Lower the default SQLITE_SORTER_PMASZ value back to 10, where it has been for the past couple of releases. Applications that need a larger value can set one. (check-in: 1ba8911c user: drh tags: trunk) | |
| 19:07 | Make the sqlite3_table_column_metadata() interface available by default and without requiring the SQLITE_ENABLE_COLUMN_METADATA compile-time option. Other sqlite3_column_* interfaces that have a run-time penalty even if they are unused still require the SQLITE_ENABLE_COLUMN_METADATA option at compile-time. (check-in: 4f7549ff user: drh tags: trunk) | |
| 15:12 | Add the SQLITE_REVERSE_UNORDERED_SELECTS compile-time option that causes the "PRAGMA reverse_unordered_selects" setting to be on by default. (check-in: 75a803e6 user: drh tags: trunk) | |
| 14:42 | Increase the default minimum PMA size for multi-threaded sorting from 10x the page size to 250x the page size. Provide the SQLITE_SORTER_PMASZ compile-time option to change this default. (check-in: b05340fe user: drh tags: trunk) | |
| 04:26 | Fix harmless compiler warning. (check-in: e97b7a8b user: mistachkin tags: trunk) | |
|
2014-12-08
| ||
| 20:29 | Add extra tests to e_walckpt.test. (check-in: d6832aa2 user: dan tags: trunk) | |
| 20:20 | Add missing mutex calls around a call to sqlite3SchemaGet() within sqlite3_open(). (check-in: 45415899 user: dan tags: trunk) | |
| 18:08 | Add a missing mutex around calls to clear the KeyInfo cache when closing a database connection. (check-in: 7047ce32 user: drh tags: trunk) | |
| 17:45 | Add a missing mutex around calls to clear the KeyInfo cache when closing a database connection. (check-in: 3ddc7e4c user: drh tags: branch-3.8.7) | |
|
2014-12-06
| ||
| 14:56 | Avoid accessing a single uninitialized byte when moving a rare 3-byte cell from an internal page to a leaf. This was not actually causing a problem, just a valgrind warning. (check-in: 6aeece19 user: dan tags: trunk) | |
| 02:05 | Fix a (harmless) uninitialized variable reference in b-tree balancing for auto-vacuumed tables with overflow pages. (check-in: dd1dd445 user: drh tags: trunk) | |
|
2014-12-05
| ||
| 21:18 | Avoid a potential NULL pointer deference in the sqlite3_stmt_scanstatus() logic. (check-in: 42d44adc user: drh tags: trunk) | |
| 21:04 | Fix a buffer overread that might occur in analyze.c if SQLITE_ENABLE_STAT4 was defined. (check-in: c1ae1268 user: dan tags: trunk) | |
| 20:46 | Add new test file e_walckpt.test. Still some tests to come. (check-in: e4db3db3 user: dan tags: trunk) | |
| 20:43 | Fix a buffer overread that might occur in analyze.c if SQLITE_ENABLE_STAT4 was defined. (check-in: 194c90db user: dan tags: branch-3.8.7) | |
| 19:50 | Make sure the WhereTerm objects are fully zeroed when they are allocated. (check-in: fdb66733 user: drh tags: trunk) | |
| 15:31 | The KeyInfo cache must be cleared before closing the btree, not after. Revised fix for ticket [e4a18565a36884b00edf]. (check-in: 7ed3346e user: drh tags: trunk) | |
| 15:25 | The KeyInfo cache must be cleared before closing the btree, not after. Revised fix for ticket [e4a18565a36884b00edf]. (check-in: 48529508 user: drh tags: branch-3.8.7) | |
| 14:44 | When closing a (shared-cache) database connection, be sure to clear out all KeyInfo objects cached on Index objects. Fix for ticket [e4a18565a36884b00edf]. (check-in: adca7688 user: drh tags: trunk) | |
| 14:34 | Improved comment on the sharedB.test test script. (Closed-Leaf check-in: 71f589e3 user: drh tags: fix-stale-keyinfo-cache) | |
| 14:07 | Test case demonstrating the problem described by ticket [e4a18565a36884b00edf]. (check-in: ffea3e90 user: drh tags: fix-stale-keyinfo-cache) | |
| 05:38 | When closing a (shared-cache) database connection, be sure to clear out all KeyInfo objects cached on Index objects. Proposed fix for ticket [e4a18565a36884b00edf]. (check-in: 651ed97d user: drh tags: fix-stale-keyinfo-cache) | |
| 00:32 | Fix compiler warnings. (check-in: e9955c0e user: drh tags: trunk) | |
| 00:17 | Fix the autoconf and MSVC makefiles, which have been broken for nearly a month. :-( (check-in: 520c2b83 user: drh tags: trunk) | |
|
2014-12-04
| ||
| 21:54 | Make sure that a DISTINCT query with an ORDER BY works correctly even if it uses a descending index. Fix for ticket [c5ea805691bfc4204b1cb9e]. (check-in: 0d3aef97 user: drh tags: trunk) | |
| 20:24 | Performance enhancement for single-table queries with many OR-connected WHERE clause terms and multiple indexes with the same left-most columns. (check-in: 1461d543 user: drh tags: trunk) | |
| 16:27 | If a table is the right operand of a LEFT JOIN, then any column of that table can be NULL even if that column has a NOT NULL constraint. Fix for ticket [6f2222d550f5b0ee7ed]. (check-in: 6f6fcbe4 user: drh tags: trunk) | |
| 15:02 | Clarification of the meaning of the second parameter to the busy-handler callback. No changes to code. (check-in: 1e2bc484 user: drh tags: trunk) | |
| 14:01 | Fix comment typos reported on the mailing list. No changes to code. (check-in: 93a71c9f user: drh tags: trunk) | |
| 04:50 | Add an implementation mark and fix a comment describing the OP_Checkpoint opcode. (check-in: 7475b90c user: drh tags: trunk) | |
|
2014-12-03
| ||
| 19:25 | Add evidence marks and assert()s used as evidence for checkpoint requirements. (check-in: b2da8afc user: drh tags: trunk) | |
| 19:08 | Fix over-length comment lines in sqlite.h.in. No changes to code. (check-in: cbd357fd user: drh tags: trunk) | |
| 18:32 | Simplify the sqlite3_wal_checkpoint() documentation. Add some source code evidence marks. (check-in: 026c44ff user: drh tags: trunk) | |
| 16:30 | Futher tweaks to the sqlite3_wal_checkpoint_v2() documentation. (check-in: 7d284d04 user: drh tags: trunk) | |
| 15:50 | Updates to the documentation for sqlite3_wal_checkpoint_v2() and related interfaces, including adding many requirements marks. (check-in: 1e212d98 user: drh tags: trunk) | |
|
2014-12-02
| ||
| 20:51 | Add support for SQLITE_CHECKPOINT_TRUNCATE. (check-in: edda2b9e user: drh tags: trunk) | |
| 19:35 | When attempting to restart a wal file, make any required calls to sqlite3_randomness() before waiting on or checking for wal file readers. This restores the behaviour exhibited by the trunk. (Closed-Leaf check-in: 6ee08769 user: dan tags: checkpoint-truncate) | |
| 19:04 | Add the SQLITE_CHECKPOINT_TRUNCATE option. (check-in: 8e20a434 user: dan tags: checkpoint-truncate) | |
| 16:16 | Convert two unreachable branches into assert() statements. (check-in: 61b31e77 user: drh tags: trunk) | |
| 13:46 | Work around overzealous NULL pointer checking in memcpy() and memset() for some systems. (check-in: 0d04f380 user: drh tags: trunk) | |
|
2014-11-28
| ||
| 13:35 | Remove the ill-designed "-end" option from the command-line shell. Instead, allow multiple SQL or dot-commands as command-line arguments. Any -cmd commands are processed first, followed by other command-line arguments, for backwards compatibility. (check-in: 24fa2e98 user: drh tags: trunk) | |
| 11:54 | Add the -end option to the command-line shell, which forces it to exit after reading prior command-line options (presumably including one or more -cmd options) and without reading standard input. (check-in: b59397b1 user: drh tags: trunk) | |
|
2014-11-27
| ||
| 11:36 | Fix a buffer overread during compilation of CREATE VIRTUAL TABLE statements that featured an explicit database name but no virtual table arguments. For example, "CREATE VIRTUAL TABLE main.ft USING fts4". (check-in: f095cde5 user: dan tags: trunk) | |
| 04:23 | More test cases for the balancer. (check-in: 358ea818 user: drh tags: trunk) | |
| 03:46 | Fix a problem in the new b-tree balancer that was causing corruption of the fragmentation count. (check-in: f242394e user: drh tags: trunk) | |
|
2014-11-25
| ||
| 18:59 | Fix an integer overflow bug in vdbesort.c. (check-in: 62382719 user: dan tags: trunk) | |
|
2014-11-22
| ||
| 21:37 | Always reinitialized the Index.bUnordered and Index.noSkipscan flags before rereading the sqlite_stat1 table, even if SQLITE_ENABLE_STAT4 is defined. (check-in: 1e1221fc user: drh tags: trunk) | |
| 19:52 | Fix an error in the comments from the previous check-in. (check-in: 9660ce54 user: drh tags: trunk) | |
| 18:50 | Deploy heuristics (well-commented) to better estimate how much unindexed terms in the WHERE clause filter the number of output rows from a single table. (check-in: 22165994 user: drh tags: trunk) | |
| 12:22 | Remove a redundant test case (probably a copy/paste error). Add an assert() to where.c to ensure that automatic indexes do not have there output row counts adjusted downward by supplementary constraints. (check-in: eea47933 user: drh tags: trunk) | |
|
2014-11-20
| ||
| 23:21 | Fix a typo in a requirements mark on the abs() SQL function. (check-in: b1e6c02f user: drh tags: trunk) | |
| 23:11 | Fix a benign test error on PRAGMA collation_list introduced by a recent checkin. (check-in: 332cc959 user: drh tags: trunk) | |
| 23:03 | Updates to requirements tags on the mutex documentation. (check-in: fcf8b7e4 user: drh tags: trunk) | |
| 19:22 | Add requirements marks on the built-in collating functions. (check-in: 4b608b62 user: drh tags: trunk) | |
| 15:30 | Ensure that when the number of cells on a page drops to zero that the freelist and fragment counter are both cleared. Also add evidence marks corresponding to file-format documentation. (check-in: ef9fbc08 user: drh tags: trunk) | |
| 02:58 | Fix the encoding of some integers to use the minimum amount of space: -128, -32768, -8388608, -217483648, and -140737488355328. (check-in: 2d7c8da5 user: drh tags: trunk) | |
| 02:18 | Add some requirements marks to the record formatting logic. Comment changes only - the code is unaltered. (check-in: 9a9627e1 user: drh tags: trunk) | |
|
2014-11-19
| ||
| 16:36 | Add new requirements marks associated with the file format documentation. No changes to code. (check-in: 6d00bcca user: drh tags: trunk) | |
| 14:31 | Completely remove an assert() that had previously been commented out. (check-in: 89b3c1c4 user: drh tags: trunk) | |
| 14:05 | Add an ALWAYS on an always-true branch in wal.c. Fix the ANALYZE command so that it resets the "unordered" and "noskipscan" flags on indices when reloading the sqlite_stat1 table. (check-in: 9ed97a85 user: drh tags: trunk) | |
|
2014-11-18
| ||
| 21:54 | Adding the "noskipscan" token to an sqlite_stat1.stat field prevents an index for being used with the skip-scan algorithm. (check-in: 4461bf04 user: drh tags: trunk) | |
| 21:45 | Allow an automatic index to be used if the only uses of declared indexes for the same loop use the skip-scan algorithm. (check-in: c52f7971 user: drh tags: trunk) | |
| 20:49 | Merge in all the other ROLLBACK fixes from the branch-3.8.7 branch. I don't know why I was doing them one-by-one. (check-in: 296b0c73 user: drh tags: trunk) | |
| 20:22 | Fix a bug in the sqlite3TripAllCursors() routine that prevents it from reporting errors. It is unknown at this time whether or not this omission can result in any incorrect result in an actual query. (check-in: 2896f264 user: drh tags: trunk) | |
| 20:16 | Update a couple of test cases to account for the fact that ROLLBACK does not always abort all running SELECT statements. (check-in: abccda76 user: drh tags: trunk) | |
| 12:28 | Increment the version number to 3.8.7.2 (check-in: 945a9e68 user: drh tags: branch-3.8.7) | |
| 02:44 | Add an ALWAYS() to an always-true conditional in the WAL rollback logic. (check-in: c5eae8a6 user: drh tags: branch-3.8.7) | |
|
2014-11-17
| ||
| 20:33 | Remove code from sqlite3BtreeKeySize() made unreachable by the previous check-in. (check-in: 57c4aa98 user: drh tags: branch-3.8.7) | |
| 19:44 | Avoid calling sqlite3BtreeKeysize() on a b-tree cursor in SKIPNEXT or SKIPPREV state. Cherrypick of [54e7d3fcb1]. (check-in: 2f2ecb99 user: dan tags: branch-3.8.7) | |
| 19:42 | Avoid calling sqlite3BtreeKeysize() on a b-tree cursor in SKIPNEXT or SKIPPREV state. (check-in: 54e7d3fc user: dan tags: trunk) | |
| 19:25 | Improved comments on the BtCursor.skipNext field. No changes to code. (check-in: e956e7db user: drh tags: branch-3.8.7) | |
| 17:13 | When a SELECT statement is terminated by a ROLLBACK TO operation, make the error message be "abort due to ROLLBACK" rather than "callback requested query abort". (check-in: 34fc4a08 user: drh tags: branch-3.8.7) | |
| 15:32 | Fix a bug in the sqlite3TripAllCursors() routine that prevents it from reporting errors. It is unknown at this time whether or not this omission can result in any incorrect result in an actual query. (check-in: 42588207 user: drh tags: branch-3.8.7) | |
| 15:22 | Update a couple of test cases to account for the fact that ROLLBACK does not always abort all running SELECT statements. (check-in: eba171e9 user: dan tags: branch-3.8.7) | |
|
2014-11-15
| ||
| 19:08 | Adding the "noskipscan" token to an sqlite_stat1.stat field prevents an index for being used with the skip-scan algorithm. (Closed-Leaf check-in: 00fe0950 user: drh tags: noskipscan-token) | |
|
2014-11-14
| ||
| 19:34 | Consider using an automatic-index for a scan even if there exists a possible skip-scan that uses one or more "=" operators. (Closed-Leaf check-in: 93642a65 user: dan tags: experimental-autoindex-fix) | |
| 15:42 | Do not automatically remove the DISTINCT keyword from "a IN (SELECT DISTINCT ...)" expressions. Fix for [db87229497]. (check-in: 98457a57 user: drh tags: branch-3.8.7) | |
| 15:28 | Do not automatically remove the DISTINCT keyword from "a IN (SELECT DISTINCT ...)" expressions. Fix for [db87229497]. (check-in: 55e453aa user: dan tags: trunk) | |
|
2014-11-13
| ||
| 14:30 | Modify the documentation for sqlite3_backup_init() to indicate that it will fail if there is already a read or read-write transaction open on the destination database. (check-in: ef03a203 user: dan tags: trunk) | |
| 14:18 | Have calls to sqlite3_backup_init() fail if there is already a read or read-write transaction open on the destination database. (check-in: 169b5505 user: dan tags: trunk) | |
| 13:42 | When a transaction or savepoint rollback occurs, save the positions of all open read-cursors so that they can be restored following the rollback operation. Cherry-pick of check-in [dd03a2802f3f27] (check-in: 402780a9 user: drh tags: branch-3.8.7) | |
|
2014-11-12
| ||
| 17:45 | Add further tests for rollback operations in the presence of ongoing selects. (check-in: eaf3aae0 user: dan tags: trunk) | |
| 14:56 | When a transaction or savepoint rollback occurs, save the positions of all open read-cursors so that they can be restored following the rollback operation. (check-in: dd03a280 user: dan tags: trunk) | |