Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
217 check-ins using file src/os_unix.c version c230a7a2
|
2018-07-27
| ||
| 20:01 | Reduce the performance cost of the constant propagation optimization to less than 200,000 cycles. (Closed-Leaf check-in: 865249de user: drh tags: propagate-const-opt) | |
| 18:19 | Performance improvement to sqlite3ExprCollSeq(). With this change, the performance of speed-check.sh is within 400,000 cycles of trunk. (check-in: a5f86f49 user: drh tags: propagate-const-opt) | |
| 18:12 | Faster implementation of sqlite3IsBinary(). (check-in: be05d0db user: drh tags: propagate-const-opt) | |
| 17:51 | Only run the constant propagation optimization on joins since that is the only scenario where it is useful. This saves prepare time for the common case of a simple query. (check-in: 598d6083 user: drh tags: propagate-const-opt) | |
| 16:57 | Constant propagation is now restricted to just the WHERE clause. The mechanism is changed to take affinity and collation into account. This seems to give correct answers. But the search for constant propagation costs 4 million cycles in the speed test. (check-in: 82c67efb user: drh tags: propagate-const-opt) | |
|
2018-07-26
| ||
| 23:54 | Add a test case demonstrating the collation problem with constant propagation. (check-in: 50add839 user: drh tags: propagate-const-opt) | |
| 23:47 | Generalize the constant propagation optimization so that it applies on every WHERE close, not just those that contain a subquery. This then demonstrates that the current implementation is inadequate since it does not take into account collating sequences. (check-in: 57eb2abd user: drh tags: propagate-const-opt) | |
| 21:16 | Initial implementation of the WHERE-clause constant propagation optimization. (check-in: 2fb82ad8 user: drh tags: propagate-const-opt) | |
|
2018-07-25
| ||
| 15:25 | Fix a buffer overread in fts5. (check-in: 0e3de8ab user: dan tags: trunk) | |
| 15:12 | Keep generated opcode values grouped together when required, even when they do not correspond to a token. (check-in: 6ee2950b user: mistachkin tags: trunk) | |
| 07:29 | Fix a typo in the documentation for sqlite3_snapshot_get(). (check-in: cd9713c9 user: dan tags: trunk) | |
|
2018-07-24
| ||
| 22:02 | Do not abort running queries due to a CREATE INDEX statement. Allow them to run to completion before being reprepared. Fix for ticket [c694113e50321afdf9]. (check-in: 2bd59333 user: drh tags: trunk) | |
| 13:57 | Add missing SQLITE_OMIT_WINDOWFUNC check to the json1 extension. (check-in: fe19f823 user: mistachkin tags: trunk) | |
|
2018-07-23
| ||
| 18:53 | Fix typo of 'SQLITE_OMIT_WINDOWFUNC'. (check-in: 8607b84a user: mistachkin tags: trunk) | |
| 18:32 | Attempt to improve the documentation for snapshots. No changes to code. (check-in: 46dd076a user: drh tags: trunk) | |
| 10:57 | Fix the macros used to generate the output of "PRAGMA compile_options" so that they handle valid values for the SQLITE_DEFAULT_LOOKASIDE macro (which contain a "," character). (check-in: 1cc72845 user: dan tags: trunk) | |
|
2018-07-22
| ||
| 21:23 | Fix more harmless compiler warnings seen with MSVC. (check-in: edab5666 user: mistachkin tags: trunk) | |
| 21:08 | Another minor revision to the 'fulltestonly' set of tests. (check-in: 61f8e2a4 user: mistachkin tags: trunk) | |
| 20:24 | Fix harmless compiler warning. (check-in: ead4762d user: mistachkin tags: trunk) | |
| 06:25 | In the Win32 VFS, when truncating a file, unmap it first. (check-in: 21510a66 user: mistachkin tags: trunk) | |
| 00:45 | Remove an unused branch in the FK logic. (Closed-Leaf check-in: 523ff779 user: drh tags: dropViewNoStat) | |
|
2018-07-21
| ||
| 23:15 | In 'resetdb.test', close a database prior to trying to delete it. (check-in: 45137053 user: mistachkin tags: trunk) | |
|
2018-07-20
| ||
| 20:56 | When dropping a view, skip trying to delete from 'sqlite_stat*'. (check-in: 2f5be3a2 user: mistachkin tags: dropViewNoStat) | |
| 19:24 | Change the SQLITE_Stat34 bit of the optimization test-control so that it prevents STAT4 data from being used but allows it to be loaded into the Index objects. This permits STAT4 to be turned on and off on a per-statement basis. (check-in: 489f3caf user: drh tags: trunk) | |
| 18:01 | Experimental change to ANALYZE so that when computing values for sqlite_stat1, it uses the maximum number for the number of rows having equivalent values, rather than the average. (Leaf check-in: 085f17e1 user: drh tags: stat1-max) | |
| 15:44 | New checks in PRAGMA integrity_check to validate the autovacuum settings in the header. (check-in: a4663f09 user: drh tags: trunk) | |
| 13:39 | Enhanced detection of problems on the freelist and on overflow list in PRAGMA integrity_check. (check-in: 11e58f5b user: drh tags: trunk) | |
|
2018-07-19
| ||
| 15:27 | Avoid a branch in the commit logic that is unreachable when compiled without SQLITE_ENABLE_BATCH_ATOMIC_WRITE. (check-in: 271b8980 user: drh tags: trunk) | |
| 14:40 | Allow the VACUUM following SQLITE_DBCONFIG_RESET_DATABASE to proceed even if the write-version in the header indicates that the database file is not writable. (check-in: 3dca8b9d user: drh tags: trunk) | |
| 11:44 | Experimental fix for resetting databases that have been deemed read-only due to a corrupt "read-version" header field. (Closed-Leaf check-in: e2394002 user: dan tags: exp-dbreset-fix) | |
|
2018-07-18
| ||
| 19:09 | Add the SQLITE_FCNTL_DATA_VERSION file control (check-in: a5087c5c user: drh tags: trunk) | |
| 17:37 | Fix a harmless compiler warning in the pager. Enhance the docs for sqlite3_changes() and sqlite3_total_changes() to refer to the data_version pragma. (check-in: 4c70ea5b user: drh tags: trunk) | |
|
2018-07-17
| ||
| 14:01 | If an SQLITE_IOERR error is encountered as part of an atomic commit on an F2FS file-system, retry the commit in legacy journal mode. (check-in: 2e0357c2 user: dan tags: trunk) | |
| 13:55 | Fix for builds without SQLITE_ENABLE_BATCH_ATOMIC_WRITE. (Closed-Leaf check-in: b10ec14e user: dan tags: exp-retry-atomic-commit) | |
|
2018-07-16
| ||
| 20:44 | Add new file doc/F2FS.txt, containing notes on the way SQLite uses the F2FS atomic commit feature. (check-in: 59efb1bf user: dan tags: exp-retry-atomic-commit) | |
| 11:32 | Minor simplification to sqlite3RollbackAll(). (check-in: 432fdc22 user: drh tags: trunk) | |
|
2018-07-14
| ||
| 20:25 | If an SQLITE_IOERR error is encountered as part of an atomic commit on an F2FS file-system, retry the commit in legacy journal mode. (check-in: 1c41250f user: dan tags: exp-retry-atomic-commit) | |
|
2018-07-13
| ||
| 20:28 | Remove an unused function declaration from fts5. (check-in: 148d9b61 user: dan tags: trunk) | |
| 19:52 | Add the "categories" option to the unicode61 tokenizer in fts5. (check-in: 80d2b9e6 user: dan tags: trunk) | |
|
2018-07-12
| ||
| 19:14 | Add a test-control to disable the skip-scan optimization. (check-in: 650a3fe0 user: dan tags: trunk) | |
| 11:28 | Add a test case to check that SQLITE_DBCONFIG_RESET_DATABASE can be used with wal mode databases even if there are active readers. (check-in: 6145f5b3 user: dan tags: trunk) | |
|
2018-07-11
| ||
| 13:34 | On an UPSERT when the order of constraint checks is rearranged, make sure that the affinity transformations on the inserted content occur before any of the constraint checks. Fix for ticket [79cad5e4b2e219dd197242e9e5f4e]. (check-in: d120c45f user: drh tags: trunk) | |
| 03:27 | Adjustments to VdbeCoverage macros to deal with byte-code branches that can never be taken in some directions. (check-in: b170c009 user: drh tags: trunk) | |
|
2018-07-10
| ||
| 23:31 | Remove incorrect ALWAYS() macros from the window function logic. (check-in: 94ac51cc user: drh tags: trunk) | |
| 22:24 | Add VdbeModuleComment()s on the three main code generators for window functions. (check-in: f7c239e9 user: drh tags: trunk) | |
| 20:50 | Minor comment changes. Add ALWAYS() macros on some unreachable branches in the xStep() methods of built-in window functions. (check-in: f2057542 user: drh tags: trunk) | |
| 19:48 | Identify Select objects within a single statement using small sequential integers rather than "0x%p". This is more readable and yields the same result on successive runs. (check-in: a7cdc5bc user: drh tags: trunk) | |
| 18:50 | Modify tests in window3.test to be tolerant of rounding errors when comparing floating point values. (check-in: 7c26c443 user: dan tags: trunk) | |
| 17:26 | Fix a problem with using scalar sub-selects in window function queries. (check-in: 687fe532 user: dan tags: trunk) | |
| 17:10 | Further improvements to bytecode branch testing. Fix cases where the macros said a branch could not be taken when in fact it could be. Alter some window function branch coverage macros to indicate that comparison operands cannot be NULL. (check-in: 76e42b70 user: drh tags: trunk) | |
| 16:04 | Enhancements and improved documentation to the byte-code branch coverage testing logic. Provide new macros that allow the code to specify that some branch instructions can never take the NULL path and that the OP_Jump opcode is only interested in equal/not-equal. The SQLITE_TESTCTRL_VDBE_COVERAGE file control callback now works slightly differently (it provides the callback with a bitmask of the branch action, rather than an integer). (check-in: cd2da7e1 user: drh tags: trunk) | |
| 07:39 | Fix a harmless warning about comment formatting in the previous check-in. Simplify the ORDER BY dereferencing logic so that it avoids unreachable branches. (check-in: 0f6ec605 user: drh tags: trunk) | |
| 07:25 | Assert that if two functions compare equal in every other way, then they must both have OVER clauses, or neither has an OVER clause. Use this fact to simplify expression comparison. (check-in: 52559ad5 user: drh tags: trunk) | |
| 06:47 | Enhance the sqlite3ExprCompare() routine so that it knows to compare the OVER clause of window functions. (check-in: 0a7649af user: drh tags: trunk) | |
| 06:32 | Enhance the TreeView mechanism so that it shows the window function data structures as part of the abstract syntax tree. (check-in: a2c0e1be user: drh tags: trunk) | |
| 05:11 | New VDBE comments and coverage macros. (check-in: a9a30726 user: drh tags: trunk) | |
|
2018-07-09
| ||
| 22:49 | Fixes for various harmless compiler warnings. (check-in: 5023b1b8 user: drh tags: trunk) | |
| 20:58 | Add a comment to restriction (6) of the push-down optimization. No changes to code. (check-in: 0c8a2f25 user: drh tags: trunk) | |
| 20:41 | Add an ALWAYS() around a branch in sqlite3BtreeSkipNext() that we believe must always be true. (check-in: 12be361a user: drh tags: trunk) | |
| 18:55 | Avoid leaving view-definitions with an incomplete set of column names/types in the in-memory schema if an OOM strikes while allocating the same. (check-in: 1ddbb537 user: dan tags: trunk) | |
| 18:11 | Add a couple of OOM tests. (check-in: 50de2a0d user: dan tags: trunk) | |
| 17:33 | Remove redundant branches in window function processing. (check-in: 8fdaf3f3 user: drh tags: trunk) | |
| 16:24 | Simplification to the grammar rules for window functions. Fix a memory leak that can follow an OOM while parsing a comma-separated list of window definitions. (check-in: a568f9c9 user: drh tags: trunk) | |
| 13:31 | Throw an error if the second argument passed to nth_value() is not a positive integer. (check-in: 1a06e57a user: dan tags: trunk) | |
| 06:51 | Fix a bad assert() in window.c. (check-in: fe8aaf0c user: dan tags: trunk) | |
| 02:37 | Simplification to the logic that detects misuse of the application-defined function creation interfaces. (check-in: 8f21d778 user: drh tags: trunk) | |
| 02:02 | During WITHOUT ROWID table creation, if the sqlite3CreateTable() call fails due to an authorizer denying the INSERT into the sqlite_master table, then detect this fact and abort the CREATE TABLE early, to avoid an assertion fault. (check-in: 164b1641 user: drh tags: trunk) | |
|
2018-07-08
| ||
| 01:02 | Identify specific FuncDef objects for window functions using the pointer to the function name (FuncDef.zName) rather than the pointer to the xStep method. This allows xStep method pointer to be replaced with a single noopStepFunc() procedure, and thus save having lots of different no-op step functions. (check-in: 410e13b0 user: drh tags: trunk) | |
|
2018-07-07
| ||
| 20:55 | Simplifications to the implementation of the sum() SQL function. (check-in: a8b13002 user: drh tags: trunk) | |
| 20:26 | Fix another problem with reusing LSM cursors for range scans. (check-in: ae3fc765 user: dan tags: trunk) | |
| 20:23 | Fix the JSON extension so that it can be compiled separately from the amalgamation. (check-in: 4b8cfe7b user: drh tags: trunk) | |
| 20:13 | Fix an LSM problem caused by using the same cursor for equality and range scans. (check-in: f05bead3 user: dan tags: trunk) | |
| 19:47 | Add ALWAYS() macros on results of sqlite3_aggregate_context() calls in xInverse() implements, since they can never fail. (check-in: fdef2a92 user: drh tags: trunk) | |
| 19:36 | Add an assert() to help verify that OP_AggInverse is never called on an accumulator that has not previously been processed by OP_AggStep. (check-in: 42138891 user: drh tags: trunk) | |
| 17:38 | Add missing VdbeCoverage() macro to window.c. (check-in: 63f4d306 user: dan tags: trunk) | |
| 17:30 | Fix a problem with the handling of NULL values in the min() window function. (check-in: b76f35b0 user: dan tags: trunk) | |
|
2018-07-06
| ||
| 17:19 | Try to improve the error messages for misformed frame specifications in window definitions. (check-in: 927b95a0 user: drh tags: trunk) | |
| 14:31 | Also disallow non-constant expressions in "<expr> PRECEDING" or "<expr> FOLLOWING" clauses. (check-in: a6dffecc user: dan tags: trunk) | |
| 14:15 | Ensure an error is returned if the user specifies an unsupported frame type. (check-in: 0f3f8fcd user: dan tags: trunk) | |
| 13:48 | Prevent "UNBOUNDED FOLLOWING" from being used as the starting boundary of a window-frame. And "UNBOUNDED PRECEDING" from being used as the ending boundary. (check-in: e51fdf66 user: dan tags: trunk) | |
| 13:25 | Remove some bad assert() statements from the implementations of window functions percent_rank() and cume_dist(). (check-in: 443f0c28 user: dan tags: trunk) | |
| 07:42 | Return an error if DISTINCT is used with a window-function (e.g. "count(DISTINCT <expr>) OVER (...)"). (check-in: d59bcc8e user: dan tags: trunk) | |
|
2018-07-05
| ||
| 21:22 | Use separate opcodes, OP_AggValue and OP_AggInverse, for the new callbacks associated with Window Functions, for improved readability of EXPLAIN output. (check-in: fa653805 user: drh tags: trunk) | |
| 20:33 | Update the recipe for resetting a database using SQLITE_DBCONFIG_RESET_DATABASE. (check-in: c43dd23f user: dan tags: trunk) | |
| 20:05 | Get the json_group_array() and json_group_object() SQL functions working as window functions. (check-in: 916cdc83 user: drh tags: trunk) | |
| 18:34 | Return an error if a "RANGE" window-frame uses "<expr> PRECEDING" or "<expr> FOLLOWING". (check-in: 786c87ba user: dan tags: trunk) | |
| 18:19 | Fix a problem in the xInverse method of window-function group_concat(1). (check-in: 73391283 user: dan tags: trunk) | |
| 17:35 | Fix the .dump command in the command-line shell so that it does not show extraneous SELECT statements when ".echo on" is enabled. (check-in: 11763cac user: drh tags: trunk) | |
| 17:16 | In wal mode, if a "BEGIN EXCLUSIVE" command (or any other command that upgrades from no transaction directly to a write transaction) hits an SQLITE_BUSY_SNAPSHOT error, change the error code to SQLITE_BUSY to indicate to the caller that the condition may be transient. (check-in: e6108047 user: dan tags: trunk) | |
| 17:03 | Fix a typo inside an assert() statement introduced by the previous commit. (Closed-Leaf check-in: e3357728 user: dan tags: exp-busy-snapshot-fix) | |
| 15:46 | In wal mode, if a "BEGIN EXCLUSIVE" command (or any other command that upgrades from no transaction directly to a write transaction) hits an SQLITE_BUSY_SNAPSHOT error, change the error code to SQLITE_BUSY to indicate to the caller that the condition may be transient. (check-in: 221ff63e user: dan tags: exp-busy-snapshot-fix) | |
|
2018-07-04
| ||
| 14:28 | Experimental planner change to avoid a skip-scan if a regular index scan on the same index columns can be done instead. (Leaf check-in: 32924446 user: dan tags: exp-avoid-expensive-skipscan) | |
|
2018-07-03
| ||
| 20:17 | Test that a race condition can cause a "BEGIN EXCLUSIVE" to return SQLITE_BUSY_SNAPSHOT in wal mode. (check-in: 5a12db75 user: dan tags: trunk) | |
|
2018-07-02
| ||
| 17:45 | Fix a problem in the xInverse callback for the built-in sum() window function. (check-in: b6563647 user: dan tags: trunk) | |
| 17:14 | Fix a segfault caused by having identical window functions in the select-list and ORDER BY clause of a SELECT statement. (check-in: 693b4350 user: dan tags: trunk) | |
| 15:03 | Fix a crash caused by a LIKE pattern that consists of a single escape character. Problem found by OSSFuzz. (check-in: bb9bfc3a user: dan tags: trunk) | |
| 12:07 | Fix a segfault caused by invoking a regular aggregate as a window-function. And some problems with count(*) when used as a window-function. (check-in: 4f3c8a82 user: dan tags: trunk) | |
|
2018-07-01
| ||
| 16:05 | Quick patch to the Lemon parser template to avoid an array overread reported by OSSFuzz. A proper fix involves enhancements to the table generators in Lemon to make the overread impossible. That fix will take longer to implement. The current check-in is a stop-gap. (check-in: 3f6730be user: drh tags: trunk) | |
|
2018-06-30
| ||
| 20:26 | Avoid a warning in fts5.c due to not using generated routine sqlite3Fts5ParserFallback(). (check-in: 65ff5144 user: dan tags: trunk) | |
| 20:15 | Merge implementation of SQL window functions from the exp-window-functions into trunk. (check-in: 17198a12 user: dan tags: trunk) | |
| 20:00 | Fix a minor problem in the code for determining whether or not an SQL statement is SQLITE_TOOBIG. (Closed-Leaf check-in: 763e6c9e user: dan tags: exp-window-functions) | |
| 19:12 | Merge all changes from the weak-fallback branch except those related to the weak-fallback feature itself. (check-in: aad718fb user: dan tags: exp-window-functions) | |
| 18:54 | Have the tokenizer handle fallback for tokens "OVER" and "FILTER" in the same way as it does for "WINDOW". (Leaf check-in: 12d819e1 user: dan tags: weak-fallback) | |
|
2018-06-29
| ||
| 20:43 | Further performance related tweaks for sqlite3RunParser(). (check-in: 5eb47765 user: dan tags: weak-fallback) | |
| 20:21 | Further tweaks to sqlite3RunParser(). (check-in: eef61ffa user: dan tags: weak-fallback) | |
| 19:54 | Improve on the previous checkin. Still a bit slow. (check-in: c1fb41aa user: dan tags: weak-fallback) | |
| 17:44 | Instead of using a lemon %fallback directive, have the tokenizer try to figure out whether an instance of "WINDOW" should be TK_WINDOW or TK_ID. (check-in: 022079cb user: dan tags: weak-fallback) | |
|
2018-06-28
| ||
| 20:05 | Modifications to parse.y to better support backwards compatibility for the "window" keyword. (check-in: 7c4b879b user: dan tags: weak-fallback) | |
| 03:38 | Introduce the concept of "weak fallback" in the parser. A weak fallback only occurs if a syntax error would result otherwise. (check-in: c41d7079 user: drh tags: weak-fallback) | |
|
2018-06-27
| ||
| 20:24 | Add missing VdbeCoverage() and VdbeCoverageNeverTaken() macros to window.c. (check-in: 4383cb68 user: dan tags: exp-window-functions) | |
| 19:48 | Avoid redundant ORDER BY operations when rewriting SELECT statements that contain window functions. (check-in: 336de43a user: dan tags: exp-window-functions) | |
|
2018-06-26
| ||
| 20:19 | Merge latest trunk changes. (check-in: d9f814b4 user: dan tags: exp-window-functions) | |
|
2018-06-25
| ||
| 20:34 | Add documentation comment for sqlite3_create_window_function(). And further tests. (check-in: 3f2705b9 user: dan tags: exp-window-functions) | |
| 11:42 | Fix another problem that could cause a crash when a window function was used in a view. (check-in: e954145a user: dan tags: exp-window-functions) | |
|
2018-06-23
| ||
| 19:29 | Fix problems with using window-functions in correlated sub-queries. (check-in: 3e23cfc8 user: dan tags: exp-window-functions) | |
| 16:26 | Fix a problem with using LIMIT in window-function queries. (check-in: c1abd2dd user: dan tags: exp-window-functions) | |
| 07:59 | Fix a problem with using a window-function SELECT as a FROM clause sub-query in some circumstances. (check-in: 11d73339 user: dan tags: exp-window-functions) | |
|
2018-06-22
| ||
| 20:51 | Omit all window-function related code when building with SQLITE_OMIT_WINDOWFUNC. (check-in: 5f04b016 user: dan tags: exp-window-functions) | |
| 17:57 | Merge latest trunk changes. (check-in: ebe65b23 user: dan tags: exp-window-functions) | |
|
2018-06-21
| ||
| 23:53 | Improved context for error_log message coming from sqlite3_prepare(). (check-in: fea7ade6 user: drh tags: trunk) | |
| 23:43 | Improved context for error_log message coming from sqlite3_prepare(). (Leaf check-in: 64df1189 user: drh tags: ieee754) | |
| 19:20 | Fix a problem with handling single row partitions in the percent_rank() window function. (check-in: b84fbf16 user: dan tags: exp-window-functions) | |
|
2018-06-20
| ||
| 09:23 | Add tests to improve coverage of code in window.c. Fix a problem with "SELECT row_number() OVER ()". (check-in: f41b6b73 user: dan tags: exp-window-functions) | |
|
2018-06-19
| ||
| 21:15 | Make the ieee754() a built-in function. Add the --ieee754 option to the ".dump" command in the shell to cause it to output floating-point values as ieee754() calls, so that the floating point number will be reproduced exactly. Not sure yet that we want to go this way, but it is one option. (check-in: 7fa8f16e user: drh tags: ieee754) | |
| 19:15 | Merge latest trunk changes with this branch. (check-in: 6ad0e64b user: dan tags: exp-window-functions) | |
| 18:11 | Fix a problem caused by a sub-query containing a window function in the FROM clause of a query that itself uses a window function. (check-in: f4b1b6f8 user: dan tags: exp-window-functions) | |
| 17:13 | Fix a problem with using min() or max() as a window function. (check-in: 801074ce user: dan tags: exp-window-functions) | |
| 11:15 | Minor change to the input grammar to make the parser tables slightly smaller. (check-in: 320fa69e user: drh tags: trunk) | |
|
2018-06-18
| ||
| 20:34 | Fix problems with using window functions in CREATE VIEW statements. (check-in: 943bccd2 user: dan tags: exp-window-functions) | |
| 20:08 | Add test case for the fix in the previous commit. (check-in: 39434262 user: dan tags: trunk) | |
| 19:09 | Fix typo in the 'normalize' extension. (check-in: 0c9163ea user: mistachkin tags: trunk) | |
| 17:36 | Ensure that all four callbacks are provided when registering a window function (otherwise SQLITE_MISUSE is returned). (check-in: 5720dcd8 user: dan tags: exp-window-functions) | |
| 16:55 | Add new API function sqlite3_create_window_function(), for creating new aggregate window functions. (check-in: da03fb43 user: dan tags: exp-window-functions) | |
|
2018-06-15
| ||
| 20:46 | Add extra OOM test. (check-in: ac251f72 user: dan tags: exp-window-functions) | |
| 19:01 | Fix another problem in lead()/lag(). And some errors that could occur following OOM faults. (check-in: fadd4dc1 user: dan tags: exp-window-functions) | |
| 16:10 | Fix a bug in the lead() and lag() window functions causing them to fail when used in queries featuring multiple window functions. (check-in: 3839fb18 user: dan tags: exp-window-functions) | |
|
2018-06-14
| ||
| 20:52 | Fix a problem with handling of statements containing two or more different windows. (check-in: 567e09ef user: dan tags: exp-window-functions) | |
| 19:06 | Fix problem with window functions min() and max() when used with a PARTITION clause and a frame starting point other than "UNBOUNDED PRECEDING". (check-in: 43eb1e75 user: dan tags: exp-window-functions) | |
| 14:30 | Merge latest trunk changes into this branch. (check-in: 5cf5f180 user: dan tags: exp-window-functions) | |
| 14:27 | Improve comments and code legibility in new file window.c. (check-in: bb915854 user: dan tags: exp-window-functions) | |
|
2018-06-13
| ||
| 20:29 | Fix problems with "RANGE BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING" window frames. (check-in: c34f31db user: dan tags: exp-window-functions) | |
| 17:19 | Output infinity as 1e999 in the ".dump" command of the command-line shell. (check-in: ee431d55 user: drh tags: trunk) | |
| 16:52 | Pad out the sqlite3_value structure to be a multiple of 8 bytes. (check-in: f76dc33b user: drh tags: trunk) | |
| 16:08 | Fix some test case numbers in expert1.test. (check-in: b90c3c16 user: dan tags: trunk) | |
| 11:41 | Unless SQLITE_OMIT_AUTOINIT is defined, allow sqlite3_vfs_unregister() to be called before sqlite3_initialize(). (check-in: ab6c1ae4 user: dan tags: trunk) | |
| 11:11 | Check that malloc() has not already failed when invoking the xUpdate() method of a virtual table implementation. This fixes an assert() failure in FTS5 that could occur following an OOM condition. (check-in: e9a7e4d5 user: dan tags: trunk) | |
| 02:20 | Add the unused SQLITE_CANTOPEN_DIRTYWAL result code. This code was used in some historical versions on the apple-osx branch but was removed by check-in [27e20d699872b2b8]. Restore it so that old code that actually references that result code will still compile. (check-in: 9f40383e user: drh tags: trunk) | |
|
2018-06-12
| ||
| 20:53 | Fix another issue to do with window-functions in aggregate queries. (check-in: 6413e38a user: dan tags: exp-window-functions) | |
| 19:35 | Documentation update: clarify that sqlite3_errcode() and related interfaces do not themselves modify the error code. (check-in: 858fc52b user: drh tags: trunk) | |
| 19:22 | Documentation updates: clarify the behavior of sqlite3_column and sqlite3_value interfaces following an OOM error. (check-in: 428c581e user: drh tags: trunk) | |
| 18:40 | Fix some problems with using window-functions in aggregate queries. (check-in: fe7081e0 user: dan tags: exp-window-functions) | |
| 13:52 | Improvements to SCopy correctness tracking when SQLITE_DEBUG is enabled. (check-in: b2973f23 user: drh tags: trunk) | |
| 13:16 | Fix the "Synopsis" on the OP_BitNot opcode. (check-in: acd78df6 user: drh tags: trunk) | |
|
2018-06-11
| ||
| 21:33 | Improvements to the --enable and --disable options for the amalgamation tarball. (check-in: 02e337e0 user: drh tags: trunk) | |
| 20:50 | Clarify the relationship between a Window object and its associated Expr. (check-in: 0cd55e98 user: dan tags: exp-window-functions) | |
| 19:47 | Fix various --enable and --disable options on the top-level configure script. (check-in: 6fd7e8ce user: drh tags: trunk) | |
| 18:16 | Fix handling of window frames containing negative number of rows. e.g. "ROWS x PRECEDING AND y PRECEDING" where (x<y). (check-in: b6d9c7ed user: dan tags: exp-window-functions) | |
| 18:06 | Fix the column cache invalidation logic in the code for ROWID uniqueness constraint checking in the INSERT command. This fixes ticket [c2432ef9089ee73bd]. (check-in: 0b485a57 user: drh tags: trunk) | |
| 17:35 | Add the OP_SetTabCol and OP_VerifyTabCol opcodes, only when compiling with SQLITE_DEBUG, to do run-time verification of the column cache. (check-in: b37614a3 user: drh tags: trunk) | |
| 13:10 | Strengthen the sqlite3VdbeMemAboutToChange() run-time verification mechanism to better detect missed calls to AboutToChange(). (check-in: 793e9422 user: drh tags: trunk) | |
| 11:19 | Fix a typon in main.mk. (check-in: e74f86f2 user: dan tags: exp-window-functions) | |
| 01:30 | Always initialize the WhereClause.hasOr field that was added by check-in [292724ffc4]. Error detected by OSSFuzz. (check-in: 9faf4171 user: drh tags: trunk) | |
|
2018-06-10
| ||
| 07:42 | Update Makefile.msc to include window.c in the build. (check-in: 16db7384 user: dan tags: exp-window-functions) | |
|
2018-06-09
| ||
| 20:52 | Fix a typo in the amalgamation autoconf file. (check-in: de0857f3 user: drh tags: trunk) | |
| 18:09 | Merge recent trunk changes with this branch. (check-in: c71f2359 user: dan tags: exp-window-functions) | |
| 17:58 | Update the amalgamation build script to include window.c. (check-in: 21d2f4a6 user: dan tags: exp-window-functions) | |
| 17:43 | Add support for FILTER clause on aggregate window functions. (check-in: ceaf798e user: dan tags: exp-window-functions) | |
| 16:49 | Slightly smaller and faster code by encapsulating wal-index hash table location information in a separate WalHashLoc object rather than passing around the various elements as separate variables. (check-in: 538a365b user: drh tags: trunk) | |
| 14:13 | Improved comments an presentation for the recent IN operator decision improvement. (check-in: 31e480f6 user: drh tags: trunk) | |
| 02:49 | Performance improvement to sqlite3WhereExprUsage(). (check-in: fd093413 user: drh tags: trunk) | |
| 01:12 | Compute the bitmask of indexed columns for each index once when the Index objecct is constructed, instead of recomputing it every time it is needed. (check-in: d735872e user: drh tags: trunk) | |
| 00:09 | Avoid invoking the whereLoopAddOr() routine in the query planner if there are no OR operators in the WHERE clause, thus speeding up query planning slightly. (check-in: 292724ff user: drh tags: trunk) | |
|
2018-06-08
| ||
| 23:23 | When the query planner has the opportunity to use an IN operater constraint on a term of an index other than the left-most term, use the estimated number of elements on the right-hand side of the IN operator to determine if makes sense to use the IN operator with index lookups, or to just do a scan over the range of the table identified by the index terms to the left. Only do this if sqlite_stat1 measurements are available as otherwise the performance estimates will not be accurate enough to discern the best plan. Bias the decision slightly in favor of using index lookups on each element of the IN operator. (check-in: 2cbbabdf user: drh tags: trunk) | |
| 21:21 | Only choose to scan an IN operator rather than use an index if we have real STAT1 data to suggest it is advantageous. (Closed-Leaf check-in: 30e87466 user: drh tags: in-scan-vs-index) | |
| 20:58 | Add support for the WINDOW clause. (check-in: 19c983b5 user: dan tags: exp-window-functions) | |
| 19:54 | Merge the btreeNext() assertion bug fix from trunk. (check-in: 11bd66e0 user: drh tags: in-scan-vs-index) | |
| 19:13 | Fix an assert() that can be false for a corrupt database and a strange query that uses a recursive SQL function to delete content from a corrupt database file while it is being queried. (check-in: 99057383 user: drh tags: trunk) | |
| 18:22 | Consider doing a partial table scan to fulfill an IN operator rather than using an index. Try to pick the plan with the lowest cost. (check-in: 1fa40a78 user: drh tags: in-scan-vs-index) | |
| 16:11 | Do not flatten sub-queries that contain window functions. (check-in: 236cb75b user: dan tags: exp-window-functions) | |
| 11:45 | Fixes to allow group_concat() to be used as a window function. (check-in: 89bbc9ba user: dan tags: exp-window-functions) | |
|
2018-06-07
| ||
| 20:35 | Merge latest trunk changes with this branch. (check-in: 25102203 user: dan tags: exp-window-functions) | |
| 20:08 | Add window functions lag() and lead(). (check-in: ef342070 user: dan tags: exp-window-functions) | |
| 18:13 | The IN-early-out optimization: When doing a look-up on a multi-column index and an IN operator is used on a column other than the left-most column, then if no rows match against the first IN value, check to make sure there exist rows that match the columns to the right before continuing with the next IN value. (check-in: 09fffbdf user: drh tags: trunk) | |
| 18:01 | Fix the assert()s in the byte-code engine that prove that cursors are unidirectional. (Closed-Leaf check-in: 4b0b4e14 user: drh tags: multikey-opt-idea) | |
| 17:45 | Add support for window function first_value(). (check-in: 060b2640 user: dan tags: exp-window-functions) | |
| 17:32 | Remove the NextIfOpen and PrevIfOpen opcodes which are no longer needed when the IN-early-out optimization is working. (check-in: 439c8162 user: drh tags: multikey-opt-idea) | |
| 16:07 | Test cases. (check-in: 085e8637 user: drh tags: multikey-opt-idea) | |
| 15:54 | Fix problems with the nth_value() function. (check-in: 63002b9a user: dan tags: exp-window-functions) | |
| 15:28 | Merge the ".stat/.eqp" CLI fix from trunk. (check-in: a91cad33 user: drh tags: multikey-opt-idea) | |
| 15:23 | Avoid using a prepared statement for ".stats on" after it has been closed by the ".eqp full" logic. Fix for ticket [7be932dfa60a8a6b3b26bcf76]. (check-in: bb87c054 user: drh tags: trunk) | |
| 14:59 | Add the WHERE_IN_EARLYOUT flag and use it to clarify the logic of this optimization. (check-in: 522f1eac user: drh tags: multikey-opt-idea) | |
| 14:32 | Merge recent trunk enhancements. (check-in: e9d7bf4f user: drh tags: multikey-opt-idea) | |
|
2018-06-06
| ||
| 23:31 | Change a comma into a logically equivalent but semantically clearer semicolon. (check-in: 71f97f0f user: drh tags: trunk) | |
| 20:55 | More space and performance enhancements to sqlite3VdbeRecordCompare(). (check-in: 83a60ff0 user: drh tags: trunk) | |
| 20:51 | Add implementation of nth_value() window function. (check-in: eb1fb420 user: dan tags: exp-window-functions) | |
| 20:29 | Small performance optimization in sqlite3VdbeRecordCompareWithSkip() for the common case where the comparison is equal. (check-in: 1e616e25 user: drh tags: trunk) | |
| 19:48 | Rearrange the order of some checks in the integrity_check pragma for a very slight performance gain. (check-in: 4b853f02 user: drh tags: trunk) | |
| 18:50 | Fix the sqlite3BeginTrans() calls within the snapshot extension. (check-in: 1fef7ad2 user: drh tags: trunk) | |
| 17:45 | Another minor optimization to OP_Transaction. (check-in: d80077ae user: drh tags: trunk) | |
| 16:28 | Change sqlite3BtreeBeginTrans() to return the BTREE_SCHEMA_COOKIE, for a small speed improvement when starting new transactions. (check-in: a10662aa user: drh tags: trunk) | |
| 01:18 | Fix the ".archive" command in the CLI (and the corresponding -A command-line option) so that it silently ignores filenames that contain "../" in their names. This prevents the "Zip Slip" attack. (check-in: 27291f2d user: drh tags: trunk) | |
| 00:36 | Attempt to optimize the resolveExprStep() routine by mapping Expr opcodes into a subset of opcodes that resolveExprStep() cares about and only calling the routine when those opcodes are encountered. Is slightly faster, but it seems like the very slight performance bump is not worth the added complexity. Saved on a dead-end branch for historical reference. (Closed-Leaf check-in: bad4d7a0 user: drh tags: dead-end) | |
|
2018-06-05
| ||
| 23:51 | Update the version number to 3.25.0 for the next development cycle. (check-in: 7598236c user: drh tags: trunk) | |
| 23:21 | Avoid unnecessary loads of columns in an aggregate query that are not within an aggregate function and that are not part of the GROUP BY clause. (check-in: e15e1006 user: drh tags: trunk) | |
| 20:45 | Add the OP_IfNoHope and OP_SeekHit opcodes used to reduce the number of unnecessary sqlite3BtreeMovetoUnpacked() calls when checking for an early exit on IN-operator loops. Futher optimizations are likely possible here. (check-in: 6bf251af user: drh tags: multikey-opt-idea) | |
| 20:04 | Calculate non-aggregate expressions in the SELECT list of an aggregate query that does not use min() or max() once per group, instead of once per row visited. (Closed-Leaf check-in: dce2dfbe user: dan tags: exp-agg-opt) | |
| 16:16 | Add implementation of last_value() window function. (check-in: 2493ce1a user: dan tags: exp-window-functions) | |
| 15:16 | Use an OP_NotFound opcode to cancel futile IN operators early. The current implementation is suboptimal because it always runs teh OP_NotFound. This still needs to be enhanced to only do the OP_NotFound if no results have been seen on the current loop. (check-in: 87a9fc50 user: drh tags: multikey-opt-idea) | |
| 13:54 | Merge trunk changes. (check-in: 047295c5 user: drh tags: multikey-opt-idea) | |
| 13:43 | Update and correct the documentation on the OP_OpenRead, OP_OpenWrite, and OP_ReopenIdx opcodes. No code changes other than the addition of an assert(). (check-in: 8a0b730d user: drh tags: trunk) | |
|
2018-06-04
| ||
| 19:24 | Version 3.24.0 (check-in: c7ee0833 user: drh tags: trunk, release, version-3.24.0) | |
| 18:55 | Add implementation of window function ntile(). (check-in: 3f093f60 user: dan tags: exp-window-functions) | |
| 13:27 | Merge recent changes from trunk. (check-in: f8df2a8e user: drh tags: multikey-opt-idea) | |
| 08:28 | Merge latest trunk changes into this branch. (check-in: 83d6416a user: dan tags: exp-window-functions) | |
|
2018-06-02
| ||
| 19:14 | Avoid using a misaligned pointer. (check-in: 1ecb3aa1 user: drh tags: trunk) | |
| 19:01 | Fix the usage of posix_fallocate() so that it correctly responds to EINVAL returns. (check-in: ce2c3e7a user: drh tags: trunk) | |
| 14:37 | Proposed fix for the use of posix_fallocate() so that it handles EINVAL returns correctly. (Closed-Leaf check-in: ab3a5539 user: drh tags: einval-from-fallocate) | |