Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
200 most recent check-ins using file test/fkey8.test version 48ef829d
|
2022-05-09
| ||
| 21:57 | Update this branch with latest changes from trunk, and the fts5 fix for "BEGIN CCONCURRENT" transactions. (check-in: c0135441 user: dan tags: begin-concurrent-report-wal2) | |
| 14:22 | Ensure that fts5 closes the read-only blob handle used to read the fts index at the end of each xUpdate operation. This doesn't matter for stock SQLite, as the handle will be closed in the subsequent xRollback or xSync call, but prevents "BEGIN CONCURRENT" transactions from being committed on this branch. (check-in: 093b6084 user: dan tags: begin-concurrent-pnu-wal2) | |
|
2022-05-06
| ||
| 17:06 | Merge the 3.38.5 fix into the 3.38 subbranch of reuse-schema. (Leaf check-in: bfccca2b user: drh tags: reuse-schema-3.38) | |
| 16:52 | Merge the 3.38.5 fix into the begin-concurrent-wal2 subbranch for 3.38. (Leaf check-in: 0586b3e9 user: drh tags: begin-concurrent-wal2-3.38) | |
| 15:25 | Version 3.38.5 (Leaf check-in: 78d9c993 user: drh tags: release, version-3.38.5, branch-3.38) | |
| 01:02 | Prevent an infinite loop on SQLITE_ERROR_RETRY when trying to modify a corrupt schema while PRAGMA writeable_schema=ON is active. dbsqlfuzz ded83609f475cc989c7339d45efb5151c1495501 (check-in: 7fcdb760 user: drh tags: branch-3.38) | |
|
2022-05-05
| ||
| 10:04 | Update the version number to 3.38.5 (check-in: bf73b57f user: drh tags: branch-3.38) | |
| 10:03 | Backout check-in [9fb20a21feb8f697] which was incorrect. (check-in: 66f65b32 user: drh tags: branch-3.38) | |
|
2022-05-04
| ||
| 16:46 | Merge version-3.38.4 into the reuse-schema-3.38 subbranch. (check-in: d5b0fd84 user: drh tags: reuse-schema-3.38) | |
| 16:36 | Merge the version 3.38.4 fixes into the begin-concurrent-wal2-3.38 branch. (check-in: 8473718d user: drh tags: begin-concurrent-wal2-3.38) | |
| 15:45 | Version 3.38.4 (check-in: d402f498 user: drh tags: release, version-3.38.4, branch-3.38) | |
| 03:48 | Fix CLI use of wrong allocator for free(). (check-in: 4f83c442 user: larrybr tags: branch-3.38) | |
|
2022-05-03
| ||
| 19:51 | Fix multi-byte column name rendition on .import and display of NULL in columnar modes. (check-in: f863aabe user: larrybr tags: branch-3.38) | |
| 14:26 | Increase the version number to 3.38.4. (check-in: def971a0 user: drh tags: branch-3.38) | |
| 14:25 | Fix the Bloom filter pull-down optimization so that it jumps to the correct place if it encounters a NULL key. Fix for the bug described by forum thread 2482b32700384a0f. (check-in: 3dc9fc2f user: drh tags: branch-3.38) | |
|
2022-04-30
| ||
| 12:55 | Preserve the COLLATE operator on an index on an expression when resolving the use of that expression into a reference to the index. See forum thread 7efabf4b03328e57 for details. (check-in: ef72f9e2 user: drh tags: branch-3.38) | |
|
2022-04-27
| ||
| 13:12 | Merge the version 3.38.3 fixes into the reuse-schema-3.8 subbranch. (check-in: 436d0a58 user: drh tags: reuse-schema-3.38) | |
| 13:00 | Merge version 3.38.3 fixes onto the begin-concurrent-wal2-3.38 branch. (check-in: 5590f0a4 user: drh tags: begin-concurrent-wal2-3.38) | |
| 12:03 | Version 3.38.3 (check-in: 9547e2c3 user: drh tags: release, version-3.38.3, branch-3.38) | |
|
2022-04-26
| ||
| 20:58 | Modify utility function sqlite3SetString() so that it may be used safely on Parse.zErrMsg. Fuzzer test case "crash-1604e5d76c92574e21e437049dab9b672e06b767.txt". (check-in: b9c7d2c1 user: drh tags: branch-3.38) | |
| 19:48 | Cherry-pick various minor fixes from trunk into branch-3.38. (check-in: fece097c user: drh tags: branch-3.38) | |
|
2022-04-25
| ||
| 23:00 | Reverse the order of two tests to stop a harmless initialized value warning from MSAN - reported by OSSFuzz. (check-in: fb7e1a07 user: drh tags: branch-3.38) | |
| 21:31 | Fix a problem with using multiple SQLITE_SUBTYPE function as window functions in a single query. (check-in: 9caca6cb user: drh tags: branch-3.38) | |
| 21:22 | Increase the version number to 3.38.3. (check-in: b113afa2 user: drh tags: branch-3.38) | |
| 21:21 | Fix a problem with automatic indexes introduced by check-in [664b461bb5063d980] and reported by forum post 0d3200f4f3bcd3a3. (check-in: 4fbb9ffe user: drh tags: branch-3.38) | |
|
2022-04-24
| ||
| 20:00 | Get CLI conformed to revised doc for it, and take recent fixes. (check-in: 60e85c7e user: larrybr tags: cli_extension) | |
|
2022-04-18
| ||
| 13:57 | Improved comment field in the bytecode generated for OP_Column and OP_Rowid. (check-in: 009bbf80 user: drh tags: right-join) | |
| 13:08 | New RIGHT JOIN test cases, one of which is currently failing. (check-in: 9168cbbe user: drh tags: right-join) | |
| 11:22 | Change the row-value initialization subroutine call from within the right-join body subroutine from an issue to an optimization opportunity. (check-in: 829d49f2 user: drh tags: right-join) | |
| 10:26 | Do not allow an empty table bypass to jump outside of a right-join body subroutine. (check-in: 1549dcd2 user: drh tags: right-join) | |
| 10:09 | Fix the sqlite3SrcListAppendList() routine so that it correctly adds the JT_LTORJ attribute to the first SrcItem, if needed. (check-in: 07ed0dca user: drh tags: right-join) | |
| 09:59 | Leave sqlite3ProcessJoin() early if an OOM is encountered in the NATURAL to USING converter. This avoids NULL pointer dereferences further along in the code. (check-in: 0ccb224e user: drh tags: right-join) | |
| 00:57 | Fix to the coalesce() function generation in the resolver. (check-in: 584bd55e user: drh tags: right-join) | |
| 00:04 | When converting a NATURAL JOIN into a JOIN USING, make sure to insert the name correctly into the IdList even if it is a wierd quoted name. (check-in: e62156b6 user: drh tags: right-join) | |
|
2022-04-17
| ||
| 23:46 | Fix to the logic that computes coalesc() functions to resolve USING terms. (check-in: 20388548 user: drh tags: right-join) | |
| 22:58 | Deal with OP_Return bypass in the RIGHT JOIN subroutine by adding extra OP_Returns where needed. (check-in: 71abe564 user: drh tags: right-join) | |
| 21:03 | Get docs caught up with the .shxload change. (check-in: e2196448 user: larrybr tags: cli_extension) | |
| 20:30 | Change OP_Return such that if P3 is 1, the Return is a no-op when the P1 register contains a NULL. (check-in: c9060232 user: drh tags: right-join) | |
| 19:22 | Move shell extension load functionality into its own .shxload command for usage and code simplicity. (check-in: e2c8c88b user: larrybr tags: cli_extension) | |
| 18:46 | Further improvements to USING() processing for RIGHT and FULL JOINs. All currently known issues are now resolved. Performace is improved. (check-in: 9fd3f22e user: drh tags: right-join) | |
| 00:48 | For sqlite3x, rudimentary testing in place as make target shellxtest . More to come. (check-in: b73928be user: larrybr tags: cli_extension) | |
|
2022-04-16
| ||
| 23:38 | New test cases, one of which is failing, indicating a bug that needs fixing. (check-in: bd5fd684 user: drh tags: right-join) | |
| 22:57 | Fix the USING to ON translation so that it works correctly for a sequence of two or more joins where one of the joins to the right-hand side of the list is a RIGHT or FULL JOIN. (check-in: 9ffc2b23 user: drh tags: right-join) | |
| 19:13 | Since the query planner is unable to cope with a LEFT JOIN on the left-hand side of a RIGHT JOIN, detect that situation early and raise a parsing error. This is a temporary measure that needs to be fixed. (check-in: 6d5d6e04 user: drh tags: right-join) | |
| 18:33 | Additional SELECT trace logic: Show the tree after result-set wildcard expansion but before name resolution. (check-in: a7babf75 user: drh tags: right-join) | |
| 13:55 | New test cases. One of the new test cases shows a flaw in the base design of RIGHT JOIN. (check-in: 4c3ce647 user: drh tags: right-join) | |
| 12:40 | RIGHT JOIN USING now appears to work the same as PG-14. Legacy tests pass. (check-in: 5bfb8624 user: drh tags: right-join) | |
|
2022-04-15
| ||
| 21:21 | Get sqlite3x and extensions testing from Makefile setup, and some of wapptest changes done to run same. (a WIP) (check-in: d238fcf4 user: larrybr tags: cli_extension) | |
| 19:53 | Revamp the USING clause processing so that it works (mostly) with RIGHT and FULL JOIN. (check-in: 719c8d92 user: drh tags: right-join) | |
| 19:49 | Avoid deleting substructure of the Expr node in lookupName() until after the error message is generated. (Closed-Leaf check-in: b8e71766 user: drh tags: right-join-using) | |
| 19:27 | Progress toward a working USING for FULL JOIN. (check-in: fed2646a user: drh tags: right-join-using) | |
| 18:30 | When expanding "*" in the result set of a SELECT, do not attach a table name to columns that are in subsequent USING clauses. (check-in: 91530990 user: drh tags: right-join-using) | |
| 17:08 | The sqlite3ProcessJoin() routine converts a NATURAL JOIN into a JOIN USING so that henceforth the NATURAL keyword can be ignored. (check-in: 8378e1e0 user: drh tags: right-join) | |
| 15:47 | Enhance the IdList object to exist in a single memory allocation (rather than a separate allocate for the base object and the array of IDs). Also permit an IdList object to store an Expr pointer together with each name. (check-in: 40f3c958 user: drh tags: right-join) | |
| 15:15 | Add the "3" in the name of the sqlite3ProcessJoin() function. (check-in: b925f72b user: drh tags: right-join) | |
|
2022-04-14
| ||
| 20:43 | Record the three known big issues with the current RIGHT JOIN design in the join8 test module, with tests that deliberately fail in order to remind us to fix the issues. (check-in: 2c4ee723 user: drh tags: right-join) | |
| 19:48 | Cherry pick subroutine indentation improvements and the hardening of OP_Gosub from the right-join branch back into trunk. (check-in: 12645f10 user: drh tags: trunk) | |
| 19:05 | Another instance of indenting a subroutine. (check-in: 73f4036b user: drh tags: right-join) | |
| 18:19 | Check for interrupts and invoke the progress handler following a Gosub opcode, to avoid and recover from infinite subroutine loops. (check-in: 647211e0 user: drh tags: right-join) | |
| 16:34 | Rerun the subroutines that compute row-values if necessary from within the RIGHT JOIN body subroutine. (check-in: 9b9038bc user: drh tags: right-join) | |
| 15:55 | Adjust the output formatting of bytecode listings so that subroutines used to implement subqueries are indented one level. (check-in: 079b7b12 user: drh tags: right-join) | |
| 14:58 | Disable autoindexing for the RIGHT JOIN loop. (check-in: 402a89e3 user: drh tags: right-join) | |
| 14:40 | Adjust assert() statements to account for the new always-NULL cursor type added by [4526c5d316508093]. (check-in: 7822faa4 user: drh tags: right-join) | |
| 14:19 | Previous check-in accidentally inverted a test on an assert(). Fix that. (Closed-Leaf check-in: 5ec4f806 user: drh tags: right-join-fix, right-join) | |
| 12:59 | Fix assert() statements associated with artifical null-value cursors created by RIGHT JOIN. (check-in: f5bce5f1 user: drh tags: right-join-fix, right-join) | |
|
2022-04-13
| ||
| 19:00 | Fix a harmless (false-positive) unused variable compiler warning on MSVC. (check-in: 63b04c63 user: drh tags: right-join) | |
| 18:32 | Make the sqlite3TreeViewSrcList() routine a no-op if called with a NULL SrcList object. (check-in: 185d2720 user: drh tags: right-join) | |
| 18:20 | Fix an assert() in OP_Column so that it accounts for the new type of pseudo-cursor that always returns NULL for any column. (check-in: 371ddc97 user: drh tags: right-join) | |
| 17:41 | If the OP_NullRow opcode is given a cursor that has not yet been opened, then go automatically open it to a special pseudo-cursor that allows returns NULL for every column. Used by the new RIGHT JOIN implementation. (check-in: 4526c5d3 user: drh tags: right-join) | |
| 16:46 | Move the explain comment for the RIGHT-JOIN post-processing loop to the verify beginning of the loop - to make the start of the loop clearer to human bytecode readers. (check-in: 7ed2a271 user: drh tags: right-join) | |
| 12:34 | This was the original landing of the RIGHT and FULL OUTER JOIN changes to trunk. It was moved back onto a branch due to problems discovered. (check-in: fa9d206f user: drh tags: right-join) | |
| 12:12 | The rows of a RIGHT JOIN might come out in any arbitrary order. So disable the ORDER-BY/GROUP-BY optimizations if a RIGHT JOIN is involved. (check-in: d168f245 user: drh tags: right-join) | |
| 10:49 | Fix an assert() in btree.c that might not be true for a corrupt database file. (check-in: dbd8e2e4 user: drh tags: trunk) | |
| 03:48 | Extension interface cleanup, and rename MetaCommand to DotCommand (to better match docs) (check-in: f78d7b8b user: larrybr tags: cli_extension) | |
| 01:52 | Add missing VdbeCoverage() macros on new branch byte-code opcodes. (check-in: 218c7167 user: drh tags: right-join) | |
|
2022-04-12
| ||
| 20:20 | Ensure that the JT_LTORJ bit in the SrcItem.fg.jointype is preserved during query flattening. (check-in: 61259050 user: drh tags: right-join) | |
| 18:40 | For the bad join type error message "unknown or unsupported join type" remove the "or unsupported" clause, because we now support all valid join types. (check-in: ab0a0562 user: drh tags: right-join) | |
| 18:04 | Factor out the RIGHT JOIN non-matched row loop from sqlite3WhereEnd(). This reduces the register pressure on that routine and helps it to run faster in the common case where there is no RIGHT JOIN. (check-in: beeecf16 user: drh tags: right-join) | |
| 17:43 | Merge the latest enhancements from trunk into the right-join branch. (check-in: b3e57ba1 user: drh tags: right-join) | |
| 17:02 | Earlier detection of corruption in balance_nonroot(). dbsqlfuzz 9191ade77ebd3b7a3356e074957aa85b0c669d14. (check-in: bff4f083 user: dan tags: trunk) | |
| 14:23 | The multi-index OR optimization does not work for RIGHT join, so disallow it. (check-in: 34c2f7b2 user: drh tags: right-join) | |
| 13:46 | Always explicitly set each table cursor to NullRow before doing the RIGHT-JOIN unmatched row pass. This is a cheap opcode, and it adds an extra layer of defense against incorrect results. (check-in: a3d14e61 user: drh tags: right-join) | |
| 11:02 | Modify utility function sqlite3SetString() so that it may be used safely on Parse.zErrMsg. Fuzzer test case "crash-1604e5d76c92574e21e437049dab9b672e06b767.txt". (check-in: 09c5a4ec user: dan tags: trunk) | |
| 10:06 | Further clarification of Tcl extension intro. (check-in: 29ab6e90 user: larrybr tags: cli_extension) | |
|
2022-04-11
| ||
| 23:08 | Tcl extension doc, minor clarifications and spelling improvements (check-in: 6ccdd9b7 user: larrybr tags: cli_extension) | |
| 23:01 | Fix CLI mishandling of SQL "execute" tokens, go and / (check-in: 1a117b03 user: larrybr tags: cli_extension) | |
| 21:00 | Fix RIGHT JOIN for virtual tables. (check-in: 75a9116e user: drh tags: right-join) | |
| 20:15 | The query flattener must add TK_IF_NULL_ROW opcodes on substituted values that land on the left operand of a RIGHT JOIN, just as it already does for the right operand of a LEFT JOIN. (check-in: 8e02cdf5 user: drh tags: right-join) | |
| 18:54 | New test cases added. (check-in: bdd1499c user: drh tags: right-join) | |
| 18:42 | Fix mis-renditions of help text. (check-in: 1a7a6b6f user: larrybr tags: cli_extension) | |
| 17:27 | Fix handling of "continue" and "break" from inside the loop for the right operand of a RIGHT JOIN. (check-in: b6e773a2 user: drh tags: right-join) | |
| 14:43 | Do not attempt the LEFT JOIN strength reduction optimization on a FULL JOIN. (check-in: 7ef3e99a user: drh tags: right-join) | |
| 14:26 | Show LEFT and RIGHT JOIN processing in the EXPLAIN QUERY PLAN output. (check-in: d91faeff user: drh tags: right-join) | |
| 13:13 | New test cases. (check-in: d5f6791b user: drh tags: right-join) | |
| 12:38 | Make a distinction between (1) WHERE clause constraints, (2) ON/USING constraints on outer joins, and (3) ON/USING clause constraints on inner joins. Formerly, there was no distinctionb between 1 and 3, but RIGHT JOIN needs to know the difference. Make RIGHT JOIN aware of this difference and add test cases. (check-in: 0f6f61c3 user: drh tags: right-join) | |
| 11:59 | Fix some comments that refer to LEFT JOIN that should refer to OUTER JOIN. No changes to code. (check-in: 5be5ede5 user: drh tags: right-join) | |
| 11:48 | Merge the Expr.w.iJoin name change from trunk into the right-join branch. (check-in: 29927926 user: drh tags: right-join) | |
| 11:25 | Rename the Expr.w.iRightJoinTable to just Expr.w.iJoin, so that the words "RightJoin" in the former name do not lead readers to believe that this has something to do with RIGHT JOINs in particular. (check-in: e8c00442 user: drh tags: trunk) | |
| 10:38 | Ensure that the JT_LTORJ flag is preserved when flattening a subquery that is on the left side of a RIGHT JOIN. (check-in: ccb61fb1 user: drh tags: right-join) | |
| 08:07 | Add embedded doc, Intro to Tcl Extension (check-in: 33591833 user: larrybr tags: cli_extension) | |
| 00:54 | Show the JT_LTORJ flag in TreeView debugging output. (check-in: 21eb4491 user: drh tags: right-join) | |
| 00:21 | Do not allow query flattening nor the push-down optimization on the right operand of a RIGHT JOIN. (check-in: 5aa0c9ea user: drh tags: right-join) | |
|
2022-04-10
| ||
| 23:48 | Cannot use an automatic index on the right table of a RIGHT JOIN because automatic indexes must be WHERE_IDX_ONLY, but the RIGHT JOIN post-processing does not know how to work with an index-only scan. (check-in: beb4401d user: drh tags: right-join) | |
| 23:01 | Revisit [f84b2beca7197589]: disallow co-routine implementations of tables that are to the left of a RIGHT JOIN, to avoid other complications. (check-in: cf00ebfc user: drh tags: right-join) | |
| 20:28 | Minor improvements to the sqlite3SrcListShiftJoinType() routine. This started out to be an effort to convert RIGHT JOIN to LEFT JOIN if the join was on the first pair of relations, but that messes up the "*" expansion and so won't work. (check-in: a48902c7 user: drh tags: right-join) | |
| 19:51 | When the left-most table of a RIGHT JOIN is implemented as a co-routine, make sure all its columns are flushed to NULL when it finishes so that they appear to be NULL during the RIGHT JOIN post-processing. (check-in: f84b2bec user: drh tags: right-join) | |
| 17:14 | More test cases and bug fixes. (check-in: 140e97fd user: drh tags: right-join) | |
| 16:13 | More RIGHT JOIN test cases and a bug fix. (check-in: 19e8ad69 user: drh tags: right-join) | |
| 09:17 | Create parallel C and C++ test/demo shell extensions, with header tweaks to make that work. Fix a bug exposed when multiple shell extensions were loaded. (check-in: 2596e7c4 user: larrybr tags: cli_extension) | |
|
2022-04-09
| ||
| 20:11 | Add a simple test case. (It is not difficult to create additional test cases that assert, at this point.) (check-in: f2201d5d user: drh tags: right-join) | |
| 19:39 | Sync w/trunk, for .import fix. (check-in: 861ab023 user: larrybr tags: cli_extension) | |
| 18:51 | Fix .import bug reported at https://sqlite.org/forum/forumpost/14db09d7e765b819 . zAutoColumn made to deliver characters, not bytes. (check-in: 21e96600 user: larrybr tags: trunk) | |
| 18:48 | Bug fixes. A basic FULL OUTER JOIN now works. (check-in: 34bbeeb7 user: drh tags: right-join) | |
| 14:57 | Get tclshext made with Tk, optionally, and make unknown work as in tclsh. (check-in: 43eb311e user: larrybr tags: cli_extension) | |
| 14:48 | Add byte-code that computes unmatched rows on the right table of a RIGHT JOIN. Compiles, and the code looks semi-reasonable, but still does not run. Incremental check-in. (check-in: 2db5a498 user: drh tags: right-join) | |
| 12:27 | The interior of the RIGHT JOIN loop is now a subroutine. (check-in: 549f5a7e user: drh tags: right-join) | |
| 03:16 | A few bits and bobs of code generation toward getting RIGHT JOIN to work. Much more remains to do. (check-in: 55b45431 user: drh tags: right-join) | |
| 03:12 | Merge the sqlite3WhereMalloc() routine from trunk into the right-join branch. (check-in: b1b3845a user: drh tags: right-join) | |
| 03:12 | Fix TCL extension build issue. (check-in: 0b3bb0e7 user: larrybr tags: cli_extension) | |
| 03:06 | The sqlite3WhereMalloc() routine allocates memory that is automatically deleted when the corresponding WhereInfo object is destroyed. (check-in: f237e1d8 user: drh tags: trunk) | |
| 02:31 | For shell extensions, generate a single-file header. For TCL extension, provide Makefile target, and make Tk opt-in rather than default. (a WIP) (check-in: 2cd49df5 user: larrybr tags: cli_extension) | |
|
2022-04-08
| ||
| 19:20 | Preliminary code to support RIGHT JOIN. Everything seems to work, except that the code to compute the unmatched rows for the RIGHT JOIN has not yet been added, so the result of a RIGHT JOIN is currently the same as an INNER JOIN. (check-in: 415abd67 user: drh tags: right-join) | |
| 17:01 | Enhance the sqlite_dbpage fix at check-in 642a0b4752743 from about a month ago such that it still takes a transaction on all attached databases, but it only starts a read transaction for read-only operations, rather than starting a write transaction for everything. (check-in: 8efd61e8 user: drh tags: trunk) | |
| 15:11 | Improved explanation and commentary on the sqlite3JoinType() routine. One of the error messages changes slightly, but besides that everything works the same. (check-in: 48f2e5a1 user: drh tags: trunk) | |
| 12:50 | Grammar rules for a hypothetical FOR EACH ROW statement. No new keywords required. However, the sqlite3_complete() machine had to be reworked because of the fact that FOR EACH ROW can contain internal semicolon tokens. (Leaf check-in: 32ae9ee7 user: drh tags: foreachrow-statement) | |
| 10:35 | Tokenizer and grammar rules sufficient for the MERGE command. (Leaf check-in: 2deee326 user: drh tags: merge-statement) | |
| 04:24 | Sync w/trunk, including all CLI updates (check-in: ad8b48d6 user: larrybr tags: cli_extension) | |
| 03:40 | For shell, silence some clang warnings, and cure one bona-fide gripe. (check-in: 52be8668 user: larrybr tags: cli_extension) | |
| 03:20 | Revamp and simplify shell help extension interface, and provide for scripted shell extensions to be seen in .help output. (check-in: aa785473 user: larrybr tags: cli_extension) | |
|
2022-04-07
| ||
| 20:45 | Fix two unreachable branches introduced by the recent sqlite3TriggersExist() optimization. (check-in: 1b5475d2 user: drh tags: trunk) | |
| 19:06 | Enhance the EXPLAIN output formatting in the CLI so that the subroutine that implements the loop body for the multi-index OR optimization is indented. (check-in: c2965fea user: drh tags: trunk) | |
| 18:17 | The ".testctrl optimizations 0x400000" command disables the generation of OP_ReleaseReg opcodes. OP_ReleaseReg opcodes are usually only generated for SQLITE_DEBUG builds and are used to verify that registers are descoped propertly. But they can get in the way of code understanding when studying bytecode dumps. So this new optimization setting is provided to temporarily turn OP_ReleaseReg opcodes off. (check-in: fa527672 user: drh tags: trunk) | |
| 14:13 | Fix the parsing of C-style comments in Lemon, as reported by forum post b6edc69548. This has no affect on SQLite itself. (check-in: 201569e0 user: drh tags: trunk) | |
| 14:03 | Optimization to sqlite3TriggersExist() saves over 700K CPU cycles. (check-in: 5043a350 user: drh tags: trunk) | |
| 13:48 | Avoid compiler warnings about the new sqlite3Show() debugging routines begin "defined but not used". (check-in: 47ddc269 user: drh tags: trunk) | |
| 12:10 | Omit an unnecessary initialization in tokenExpr(). (check-in: 8f6ae686 user: drh tags: trunk) | |
| 10:11 | Fix harmless compiler warnings. (check-in: 61095b5b user: drh tags: trunk) | |
| 01:11 | Improved technique for parsing the ON and USING clauses of a join is faster and uses less memory. (check-in: 158156a3 user: drh tags: trunk) | |
|
2022-04-06
| ||
| 19:46 | Faster parsing of the FROM clause in joins for the common case where there is no INDEXED BY clause. (check-in: 848b7a0e user: drh tags: trunk) | |
| 18:30 | Improvements to the display of AST for DML statements. (check-in: 84c239a0 user: drh tags: trunk) | |
| 17:07 | Get shell.c past clang preprocessor, silently. (check-in: 588c3512 user: larrybr tags: cli_extension) | |
| 16:47 | Add sqlite3x target to other makefiles. (check-in: efc9f08d user: larrybr tags: cli_extension) | |
| 16:19 | Add sqlite3x build target. Cure CLI build warning. Tweak CLI test. (check-in: 45736d46 user: larrybr tags: cli_extension) | |
| 15:41 | Attempt to show triggers in the TreeView output from DELETE, INSERT, and UPDATE statements. (check-in: b0939d6f user: drh tags: trunk) | |
| 12:54 | Corrections and updates to the header comment describing the TriggerStep object. No changes to code. (check-in: abb34c08 user: drh tags: trunk) | |
| 12:25 | Add the sqlite3Show() family of debugging interfaces under SQLITE_DEBUG. No changes to deliverable builds. Rename SQLITE_ENABLE_SELECTTRACE to SQLITE_ENABLE_TREETRACE in ctime.c. (check-in: bc33168c user: drh tags: trunk) | |
| 11:08 | Rename debugging defines and variables from SELECTTRACE to TREETRACE (and similar) since the functionality has how expanded to include data structures beyond SELECT statements. Should not affect deliverable builds. (check-in: 393fa32e user: drh tags: trunk) | |
| 10:37 | Add additional tree display routines for DELETE and UPDATE. No changes to deliverable code. (check-in: fbd288ff user: drh tags: trunk) | |
| 00:29 | Add new diagnostic "sqlite3TreeView" routines for IdList, Upsert, and for INSERT statements. This is all debugging code. There are no changes to release builds. (check-in: f3084122 user: drh tags: trunk) | |
|
2022-04-04
| ||
| 20:20 | Small performance and size improvement to OP_Found. (check-in: 81587a18 user: drh tags: trunk) | |
| 19:58 | Tag an always-taken branch using ALWAYS(). (check-in: ad8bc3d7 user: drh tags: trunk) | |
| 19:43 | Optimizations to the OP_Found opcode save about 600K cycles in speedtest1. (check-in: 5c3357ad user: drh tags: trunk) | |
| 18:17 | Use unpacked keys for OP_Found in foreign key processing. (check-in: 328dc766 user: drh tags: trunk) | |
| 17:27 | For shell extension writers, reduce boilerplate (mimicing SQLITE_EXTENSION_INIT# macros) (check-in: 76120813 user: larrybr tags: cli_extension) | |
| 15:26 | Fix two minor problems in the JSON functions associated with JSON Paths for object labels that require quoting. (check-in: 4a1553b5 user: drh tags: branch-3.38) | |
| 15:15 | When constructing the JSON Path for the "fullpath" column of the json_tree() and json_each() table-valued functions, be sure to quote object labels where necessary. (check-in: 0fbbe788 user: drh tags: trunk) | |
| 14:24 | Fix the JSON Path parser so that it will accept zero-length object labels. Forum thread c082aeab43. (check-in: 84fe95d2 user: drh tags: trunk) | |
| 11:38 | Performance optimization and stronger assert()s in the comparison opcodes. (check-in: e0305e64 user: drh tags: trunk) | |
| 06:33 | For TCL extension: Adjust provided "gui" command, and document it. For shell: Separate shell variables from binding parameters, mainly so they live longer, in the shell DB. Add .vars dot command to reflect this separation, and specialized for shell variables. Much code shuffling to share code between .parameters and .vars commands. (check-in: fa492ff5 user: larrybr tags: cli_extension) | |
| 01:12 | Only invoke sqlite3VdbeClearObject() from a single location, so that the compiler is more likely to in-line the code. Performance increase and size reduction. (check-in: c6947a96 user: drh tags: trunk) | |
|
2022-04-03
| ||
| 22:35 | Performance optimization in the sqlite3VdbeHalt() routine. (check-in: 9564d72a user: drh tags: trunk) | |
| 20:39 | Performance optimization and size reduction in the OP_Halt opcode. (check-in: 21948e6e user: drh tags: trunk) | |
| 19:13 | Omit the Vdbe.runOnlyOnce flag (simplifying the prepared statement implementation) and accomplish the same result by adding an "OP_Expire 1 1" opcode to prepared statements that would normally have runOnlyOnce set. (check-in: 6e20e1c4 user: drh tags: trunk) | |
| 10:42 | Performance optimization and slight size reduction in the OP_Transaction opcode. (check-in: 7bee8c19 user: drh tags: trunk) | |
|
2022-04-02
| ||
| 22:47 | Expand the getVarint32() macro in a few places, as the C-compiler seems to be able to optimize better when that macro is expanded manually. (check-in: cd4fe34b user: drh tags: trunk) | |
| 20:08 | Fix harmless compiler warnings. (check-in: 8a3a3486 user: drh tags: trunk) | |
| 19:21 | Ensure that a zero-blob does not cause in invocation of memcpy() with a zero length and a NULL source pointer. (check-in: 35441eb5 user: drh tags: trunk) | |
| 15:19 | Minor loop optimization in OP_MakeRecord. (check-in: 4350983b user: drh tags: trunk) | |
| 14:30 | In-line the sqlite3VdbeSerialPut() routine into the OP_MakeRecord opcode. This allows some duplicate comparisons to be omitted, resulting in a size reduction and performance increase. (check-in: 6f4d6f21 user: drh tags: trunk) | |
|
2022-04-01
| ||
| 21:01 | The putVarint32() macro does not optimize well. So expand it into in-line code in places were performance is an issue. (check-in: 390c239e user: drh tags: trunk) | |
| 20:39 | Loop optimization in the OP_MakeRecord opcode. (check-in: 47a61fbd user: drh tags: trunk) | |
| 20:19 | It is not necessary to de-ephermeralize the output registers in the OP_ResultRow opcode. Omit that step for a size reduction and performance increase. (check-in: 8a07745a user: drh tags: trunk) | |
| 19:13 | There is no need for sqlite3_step() to check for an OOM condition prior to starting up. (check-in: 44be7f46 user: drh tags: trunk) | |
| 19:04 | Omit the Vdbe.doingRerun field for a slight size reduction and performance gain. (check-in: e93297a9 user: drh tags: trunk) | |
| 18:45 | New macro ROUND8P() which works like ROUND8() but assumes that the input is already a multiple of the size of a pointer. It becomes a no-op for 64-bit machines, giving a small size reduction and speed boost. (check-in: d126f304 user: drh tags: trunk) | |
| 18:19 | Provide and use a version of sqlite3VdbeFreeCursor() that guarantees the cursor pointer is not NULL. This saves a few bytes of code space and a few CPU cycles. (check-in: cb5e6f8e user: drh tags: trunk) | |
| 17:23 | Remove an unnecessary branch from sqlite3VdbeReset(). (check-in: a7d79560 user: drh tags: trunk) | |
| 17:01 | Refactor the Vdbe.iVdbeMagic field into Vdbe.eVdbeState. Split the RUNNING state into separate RUNNING and READY. This gives a size reduction and performance increase. (check-in: 5a50a42f user: drh tags: trunk) | |
| 16:22 | Avoid a harmless uninitialized variable warning in PRAGMA incremental_vacuum. (check-in: 6d976f90 user: drh tags: trunk) | |
| 15:31 | Have the sqlite3_context object carry the encoding for the prepared statement that it represents, so that sqlite3_result() and similar can set the encoding according to the prepared statement, even if the database encoding has changed. dbsqlfuzz c409b10d0a6bccf78ab00f47e1d29d42ee5b3565 (check-in: d4e19314 user: drh tags: trunk) | |
|
2022-03-31
| ||
| 21:15 | Split out the RUN state into separate READY and RUN states. (Closed-Leaf check-in: d698826b user: drh tags: vdbe-state-refactor) | |
| 20:04 | Rename Vdbe.iVdbeMagic to eVdbeState. Remove unnecessary states. This is a preliminary step toward splitting RUN_STATE out into several other states. (check-in: ff91191d user: drh tags: vdbe-state-refactor) | |
| 16:09 | Fix harmless compiler warnings. (check-in: b916087a user: drh tags: trunk) | |
| 11:12 | Refinement to [3f9887d4a58cbfdb]: The SQLITE_CONSTRAINT error should only be returned if the prepared statement came from sqlite3_prepare_v2(). The legacy sqlite3_prepare() is documented to return only SQLITE_ERROR. (check-in: f2d9262e user: drh tags: trunk) | |
| 04:05 | Make CLI build after an incomplete rename completed. (check-in: 4f11e639 user: larrybr tags: cli_extension) | |
| 03:45 | For TCL extension: Add Tk and means to run it, optionally. Implement ScriptSupport interface. For shell: Get scripting support thru an object interface like the others. Rename OutMode* to Export*. For both: Provide a way to pass arguments to an extension upon load and get them into TCL's argv. (check-in: d2446e50 user: larrybr tags: cli_extension) | |
|
2022-03-30
| ||
| 17:56 | Faster and slightly smaller implementation of sqlite3Step(). (check-in: c6901a8c user: drh tags: trunk) | |
| 17:36 | Small size reduction and performance increase in sqlite3FinishCoding(). (check-in: 9248ce50 user: drh tags: trunk) | |
| 11:12 | Incorporate fix noted as absent by forum post https://sqlite.org/forum/forumpost/aefd62fcc7ebd8b6 (check-in: 336851b1 user: larrybr tags: branch-3.38) | |
|
2022-03-29
| ||
| 22:57 | In setResultStrOrError(), if the input string pointer is NULL and hence the value gets set to an SQL NULL, then the Mem.enc field is uninitialized. So do not read it. This is a harmless OSSFuzz/ASAN found problem. (check-in: 47d0b1c4 user: drh tags: trunk) | |
| 21:11 | For TCL extension: Cleanup TCL library interaction and interpreter management. Group code into more purpose oriented functions. Add .unknown dot command, delegating to TCL commands with "dot" names. Drop planned .eval new dot command. For shell: Implement undocumented .unknown dot command, doing little. Make dispatcher call .unknown's implementation for unknown dot command. Fix a command lookup bug exposed by above change to unknown handling. Add .eval dot command. Add options to .x command aimed at control of errors in command sequences. (a WIP) Make dispatcher report ambiguous dot command lookups, even with no extension. (check-in: 7616a6f4 user: larrybr tags: cli_extension) | |
| 20:50 | Ensure that sqlite3VdbeMemSetStr() always leaves the value in a consistent state even if there is an error. (check-in: c86ab289 user: drh tags: trunk) | |
| 19:19 | Reverse the order of two tests to stop a harmless initialized value warning from MSAN - reported by OSSFuzz. (check-in: 800b8fda user: drh tags: trunk) | |
| 13:16 | Avoid unnecessary calls to balance() from sqlite3BtreeDelete(). (check-in: d0966d1b user: drh tags: trunk) | |
| 01:43 | Fix the sqlite3_result_xxxxx() routines so that they all check for and perform any necessary text encoding conversions and check for oversize strings and BLOBs. Thus those checks can be done where they are most efficient and avoided in cases like OP_Function where they are more expensive. (check-in: d50b162b user: drh tags: trunk) | |
|
2022-03-28
| ||
| 21:42 | TCL extension polished and more smoothly integrated. A couple of bugs related to takedown fixed. Help for added features. Code cleaned up and commented. (check-in: 4ee7df11 user: larrybr tags: cli_extension) | |
| 18:34 | Further size and performance optimizations for sqlite3VdbeMemSetStr(). (check-in: 310a3e10 user: drh tags: trunk) | |
| 17:34 | Slightly smaller and faster sqlite3VdbeMemSetStr(). (check-in: 06928e74 user: drh tags: trunk) | |
| 15:06 | Performance increase in the loop that frees an opcode array. (check-in: 4ccebb86 user: drh tags: trunk) | |
| 14:56 | Another small performance gain by using sqlite3DbFreeNN(). (check-in: daa924af user: drh tags: trunk) | |