Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
100 most recent check-ins that include changes to files matching 'src/*'
|
2021-04-16
| ||
| 15:44 | Merge from trunk. (Leaf check-in: d762784a user: larrybr tags: compile_options) | |
| 12:33 | Remove an ALWAYS() from a branch in sqlite3ExprAddCollateToken() which can be false following an OOM. dbsqlfuzz 9e8516bf1e786c84e520ae43141b75b7399f8618. (check-in: 6af4e6d0 user: drh tags: trunk) | |
| 12:13 | Fix an assert() in whereLoopAddOr() that can be wrong if an OOM occurs in an OR constraint on a virtual table if one of the terms of the OR is an IN operator. dbsqlfuzz 5ee7350b40014f5a878c5d6e76913a2d619f5146 (check-in: 5cb40e53 user: drh tags: trunk) | |
| 11:05 | Fix the comment on the implementation of the pi() SQL function. No changes to code. (check-in: 5620bd46 user: drh tags: trunk) | |
| 10:25 | Dial back assumptions on compile option defaults to reflect actual conditionals. (check-in: cb312772 user: larrybr tags: compile_options) | |
| 01:03 | Do not invoke sqlite3ExprAffinity() if there is a possibility that the Expr is incomplete due to a prior OOM. dbsqlfuzz b8a824706914488bd236da51118eb9174ceb870f (check-in: e8a1515b user: drh tags: trunk) | |
|
2021-04-15
| ||
| 23:43 | Merge from (forked) trunk. Handle compile options that differ from default by being 0. (check-in: 6eb83f83 user: larrybr tags: compile_options) | |
| 19:09 | Fix a bad interaction between constant propagation and transitive term handling causing patterns like "WHERE unlikely(t1.x=t1.y) AND t1.y=?" to return non-matching rows. See forum post c38462ab5e. (check-in: 2363a14c user: dan tags: trunk) | |
| 12:56 | Make sure the WhereInfo.pExprMods list is properly cleared when existing sqlite3WhereBegin() early due to an OOM fault. dbsqlfuzz 1247a51318047aba42e7f6991dfa62577cb7a151. (check-in: 0e19af72 user: drh tags: trunk) | |
| 00:26 | Bring pragma compile_options up to date. And make mkctimec.tcl help do so. (check-in: 63ea3e29 user: larrybr tags: compile_options) | |
|
2021-04-14
| ||
| 15:25 | Prevent an assert() from failing if sqlite3_blob_reopen() is called on a blob-handle that has already hit an SQLITE_CORRUPT error. (check-in: b5dc7aba user: dan tags: trunk) | |
| 11:20 | Fix an assert() in the code generator for expressions so that it is valid even after an OOM. (check-in: 427e83eb user: drh tags: trunk) | |
|
2021-04-13
| ||
| 18:39 | Fix a problem with handling expressions like "(col IS NULL AND <expr1>) OR col == NULL" in WHERE clauses. Reported at https://sqlite.org/forum/forumpost/1c4f2fdcc4. (Leaf check-in: e8c7bd34 user: drh tags: branch-3.35) | |
| 13:48 | Do not leave a WITHOUT ROWID table without a primary key index structure due to an OOM error. This prevents downstream troubles in the case of PRAGMA writable_schema=ON. dbsqlfuzz 69fb32cc82d59b4d790881566e3f6c727e616122 (check-in: 608b6644 user: drh tags: trunk) | |
| 13:20 | The SQLITE_NoSchemaError flag should not apply to OOM errors while parsing the schema. (check-in: 9cc484ad user: drh tags: trunk) | |
| 13:01 | Earlier detection empty_result_callbacks in sqlite3InitCallbac() to avoid problems with corrupt schemas an writable_schema=ON. dbsqlfuzz ca4ca43bdb22eab9859de4979aec615af851e111. (check-in: fb305fe0 user: drh tags: trunk) | |
|
2021-04-12
| ||
| 23:18 | Fix a faulty assert() inside sqlite3ExprDup(). (check-in: 59812e7e user: drh tags: trunk) | |
| 22:39 | Fix an assert() in OP_ParseSchema so that it works even after an OOM. (check-in: 79b88353 user: drh tags: trunk) | |
| 17:34 | Add an assert() to show that the SF_Resolved flag does not get set if errors are being suppressed. (check-in: f455c5b0 user: drh tags: trunk) | |
| 16:59 | Fix a segfault that could occur if the ORDER BY clause of a compound SELECT contains a sub-select that uses one or more window functions. (check-in: 5ba15ebb user: dan tags: trunk) | |
| 12:58 | Fix the display of SrcList in the sqlite3TreeView debugging routines. (check-in: 28ecb381 user: drh tags: trunk) | |
| 12:02 | Avoid an assert() failure when a compound scalar sub-select is, due to a "WHERE (sub-select) AND (... OR ...)" clause, coded twice by the code generator. (check-in: a4e3f133 user: dan tags: trunk) | |
| 11:51 | Fix harmless comment typo in main.c. (check-in: 785a553e user: drh tags: trunk) | |
|
2021-04-11
| ||
| 00:11 | Add NEVER() to a branch that is no longer reachable due to check-in [a1d823f6879ce1ac]. (check-in: 5fda21fe user: drh tags: trunk) | |
|
2021-04-10
| ||
| 20:27 | Avoid an uninitialized-memory valgrind error by tightening up corruption detection in internal routine defragmentPage(). (check-in: 8096f7ae user: dan tags: trunk) | |
| 20:21 | Add NEVER() to a branch that became unreachable with check-in [58f36af2271517ab]. (check-in: e07ce463 user: drh tags: trunk) | |
| 15:34 | Avoid invoking sqlite3Atoi64() will a null pointer following an error from ExpandBlob(). (check-in: 780412f2 user: drh tags: trunk) | |
| 14:49 | Fix a problem with handling expressions like "(col IS NULL AND <expr1>) OR col == NULL" in WHERE clauses. Reported at https://sqlite.org/forum/forumpost/1c4f2fdcc4. (check-in: 40852ca8 user: dan tags: trunk) | |
| 13:37 | Refactor NameContext.nErr into nNcErr to avoid confusion with Parse.nErr. Do not abandon sqlite3ResolveExprList() on nNcErr if nErr is still zero as we might have hit a problem with ORDER BY resolution that should be a suppressed error. dbsqlfuzz 41b9dad40919d3549ca7e52d893da81a6dded4ad (check-in: 7d674970 user: drh tags: trunk) | |
|
2021-04-09
| ||
| 22:34 | Remove a NEVER() in a corruption detection conditional in defragmentPage(). dbsqlfuzz 9c61c33802600b2b01dfb09b0386ceb4ade02a19 (check-in: c174f5d3 user: drh tags: trunk) | |
| 22:20 | Remove an ALWAYS() that is actually reachable following an OOM on an UPDATE FROM of a virtual table. dbsqlfuzz aa03237ef7c4a028c7cdaf8bbcde2b62e2bcd36e (check-in: 49eac389 user: drh tags: trunk) | |
| 20:50 | Have the VFS in memdb.c return SQLITE_IOERR_NOMEM instead of SQLITE_NOMEM when an OOM error is encountered. This is required to get the pager module to handle such OOM errors correctly in some cases. (check-in: 09c96b4c user: dan tags: trunk) | |
|
2021-04-08
| ||
| 20:29 | Fix a use-after-free error that could occur when processing "SELECT aggregate(DISTINCT <expr>)..." queries. (check-in: 0e478986 user: dan tags: trunk) | |
| 19:56 | Remove an ALWAYS() that might be false under very unusual circumstances. dbsqlfuzz 300261f469ace7ecc57ed32ea7b0de3ea9d7dbf. Test case in TH3. (check-in: 466f5089 user: drh tags: trunk) | |
| 19:39 | Handle the corruption fixed by the previous commit in a different way so as to also fix dbsqlfuzz crash 753de0a0ac5b25b18f1e4d41e650d3333cdc270c. (check-in: cb27ce25 user: dan tags: trunk) | |
| 15:19 | Handle a special case of corruption that can present if "PRAGMA writable_schema=1" is set. Fix for dbsqlfuzz test case 6229ad63de49e3ba0630aaf0058868f36008bcca. (check-in: 58f36af2 user: dan tags: trunk) | |
| 14:15 | Hardden the filter_over grammar rule against OOM faults. dbsqlfuzz e47c54502a9c36778a5ed553199d5870e2ebd9f2 (check-in: f375f541 user: drh tags: trunk) | |
| 13:52 | Remove an incorrect assert() from sqlite3EndTable(). dbsqlfuzz 4af45bcda779592449b1985433bca90b0e683b9b. (check-in: 38a1085c user: drh tags: trunk) | |
|
2021-04-07
| ||
| 18:17 | Do not allow the use of rowid for views and subqueries unless compiled with -DSQLITE_ALLOW_ROWID_IN_VIEW. (check-in: a2ddb89b user: drh tags: trunk) | |
| 17:46 | Merge latest changes into this branch. (Leaf check-in: 655c2841 user: dan tags: begin-concurrent-report) | |
| 17:32 | Merge latest trunk changes into this branch. (Leaf check-in: 1f3f7e4b user: dan tags: begin-concurrent-pnu) | |
| 15:45 | Also prohibit the use of rowid from a subquery. Add the SQLITE_ALLOW_ROWID_IN_VIEW compile-time option to restore legacy behavior in case somebody actually needs it. (check-in: 14b1d56e user: drh tags: no-rowid-on-view) | |
| 15:23 | Merge version 3.35.4 with this branch. (Leaf check-in: 4b4f8791 user: dan tags: reuse-schema) | |
| 15:13 | Update a broken assert() on this branch. (check-in: e69f2989 user: dan tags: reuse-schema) | |
| 13:20 | Unnecessary local variable initialization to fix a harmless compiler warning. (check-in: c22e47c7 user: drh tags: trunk) | |
| 12:36 | Fix an assert in the automatic-index constructor that can be false following an OOM error. dbsqlfuzz ee04d66c627ed5b5301e0ce09e24db47e0f1544e. (check-in: cc6463fe user: drh tags: trunk) | |
| 12:21 | Earlier detection of page 1 on the freelist (which is impossible) and reporting of that as database corruption. dbsqlfuzz 6fe6c504b320be5e3eed66896cb0db28bfb38a93. (check-in: b1bf0d98 user: drh tags: trunk) | |
| 12:02 | Fix a possible crash that could occur when executing a drop column operation on a corrupt database schema. (check-in: c3378494 user: dan tags: trunk) | |
|
2021-04-06
| ||
| 23:29 | Do not apply the optimization that omits DISTINCT if all result terms are part of a UNIQUE index if the index is also a partial index. Fix for the bug reported by forum post 66954e9ece. (check-in: c2f940b0 user: drh tags: trunk) | |
| 21:20 | Modify the drop column code to handle WITHOUT ROWID tables for which a single column appears more than once in the primary key. (check-in: e0d25070 user: dan tags: trunk) | |
| 19:13 | Raise an error if a query tries to access the "rowid" of a view, as views do not have rowids. (check-in: 7a73c616 user: drh tags: no-rowid-on-view) | |
| 18:40 | Restructure a loop in window.c to avoid hitting an assert() following an OOM. dbsqlfuzz test case 6ef74a9659de87e9be3d8694ad062c448aa7ef1d. (check-in: 887c19a1 user: dan tags: trunk) | |
| 18:02 | Update an assert statement that could fail when using "PRAGMA writable_schema = 1" to access a database with a corrupt schema. (check-in: 55cad6f6 user: dan tags: trunk) | |
| 16:42 | Silence an msan warning triggered by a corrupt database by initializing a local variable in sqlite3Insert(). (check-in: f64deece user: dan tags: trunk) | |
| 16:16 | Avoid a use-after-free that may occur when accessing a corrupt database schema with "PRAGMA writable_schema = 1" set. (check-in: a1d823f6 user: dan tags: trunk) | |
| 13:56 | Earlier detection of OOM errors during window function processing. dbsqlfuzz b08676f5a0437552c95d2a5d7ef285f8c451ea35. (check-in: 8e04f52e user: drh tags: trunk) | |
| 13:53 | Handle "RETURNING rowid" clauses attached to INSERTs on views without causing an assert() to fail. Fixes dbsqlfuzz crash 0081f863d7b2002045ac2361879fc80dfebb98f1. (check-in: 9c5b7386 user: dan tags: trunk) | |
| 13:03 | Further improvements (subsequent to [d91450847a3a3a72]) to handling of attempts to delete a journal using the memdb VFS. dbsqlfuzz ac61dab357279a6a531de067f040e9a4828d2df0. (check-in: a0839406 user: drh tags: trunk) | |
| 12:50 | Earlier detection and handling of OOM problems. dbsqlfuzz 39f2963ea5559aa3a16e24e0e3cb42aac85a7371. (check-in: 8d46df73 user: drh tags: trunk) | |
|
2021-04-05
| ||
| 22:42 | Relax an assertion on the sqlite3WhereEnd() routine after dbsqlfuzz found a counter-example. (check-in: e6b591e7 user: drh tags: trunk) | |
| 22:30 | Add a missing CORRUPT_DB to an assert in the OP_ResultRow opcode. (check-in: 82f92d79 user: drh tags: trunk) | |
| 19:05 | Do not allow floating point rounding errors to cause a window function xInverse() function to be invoked before the corresponding xStep() call. (check-in: 7a19fed4 user: dan tags: trunk) | |
| 17:50 | Fix an assert() that can be false when doing an incremental vacuum on a corrupt database file. dbsqlfuzz cced0668cfd4da4eb2382cb9dd26c17c64aaff76. (check-in: 15b80101 user: drh tags: trunk) | |
| 16:20 | Avoid coding unindexed WHERE constraints if an equivalent transitive term has already been coded by an outer loop. (check-in: 644bb77f user: dan tags: trunk) | |
| 13:41 | Do not invoke the xDelete method when converting to PRAGMA journal_mode=OFF if the underlying VFS does not support that method. dbsqlfuzz 39d6af88ef5242f866c4f9b21ede330c5c1e36a8. (check-in: d9145084 user: drh tags: trunk) | |
| 11:39 | As a continuation of [1f66a85b7757754f], ensure that an OOM that occurs while checking the validity of the schema is reported out as SQLITE_NOMEM and not as SQLITE_SCHEMA. (check-in: 5045c8a7 user: drh tags: trunk) | |
|
2021-04-04
| ||
| 23:56 | Fix an assert() in sqlite3BtreeLast() that needs an "|| CORRUPT_DB" term. Dbsqlfuzz case b92b72e4de80b5140c30ab71372ca719b8feb618. (check-in: ad718388 user: drh tags: trunk) | |
| 14:22 | Avoid compile error when SQLITE_OMIT_VIRTUALTABLE defined. (check-in: 0ee0ef47 user: larrybr tags: trunk) | |
| 12:52 | Make shell .parameter feature type-agnostic. Hush harmless compiler warning. (check-in: 35cf295e user: larrybr tags: trunk) | |
|
2021-04-03
| ||
| 23:30 | Improved robustness to OOM conditions in the window function logic. dbsqlfuzz 0c123f7d80b29beaafc8411c12129e46f7ffdac3. (check-in: 35ff7cbf user: drh tags: trunk) | |
| 20:35 | Take care not to change an SQLITE_NOMEM error into SQLITE_SCHEMA. (check-in: 1f66a85b user: drh tags: trunk) | |
| 19:23 | Fix a crash in handling queries of the form "SELECT aggregate(DISTINCT tbl.col) FROM ... LEFT JOIN tbl ...". Fixes a problem introduced by [ef4ac0ddd297bbd3]. (check-in: 0dcf808d user: dan tags: trunk) | |
|
2021-04-02
| ||
| 20:53 | Merge latest trunk changes with this branch. (check-in: a2216259 user: dan tags: begin-concurrent-report) | |
| 20:43 | Correctly capture the error when a RETURNING clause appears on an attempt to UPDATE an eponymous virtual table. dbsqlfuzz 486f791cbe2dc45839310073e71367a1d8ad22dd. (check-in: 778a9a6e user: drh tags: trunk) | |
| 20:40 | Merge latest trunk changes into this branch. (check-in: a7e612ee user: dan tags: begin-concurrent-pnu) | |
| 20:29 | Merge latest trunk changes into this branch. (Leaf check-in: 98420842 user: dan tags: begin-concurrent) | |
| 19:55 | Add experimental SQLITE_FCNTL_EXTERNAL_READER file control. (check-in: e16da5af user: dan tags: trunk) | |
|
2021-04-01
| ||
| 11:30 | When resolving names in the RETURNING clause, do not accept trigger names even within subquires. (check-in: 8541c71b user: drh tags: branch-3.35) | |
|
2021-03-31
| ||
| 23:56 | When resolving names in the RETURNING clause, do not accept trigger names even within subquires. See forum post 34c81d83c9177f46 for context. (check-in: fd4ea3f6 user: drh tags: trunk) | |
| 17:49 | When resolving column names in the RETURNING clause, do not ignore an incorrect table name qualifier. Raise an error instead. (check-in: 3b916924 user: drh tags: branch-3.35) | |
| 17:42 | When resolving column names in the RETURNING clause, do not ignore an incorrect table name qualifier. Raise an error instead. See forum post 85aef8bc01 for context. (check-in: 51d5c50b user: drh tags: trunk) | |
| 13:51 | Defer deletion of expressions that are optimized out by the AND optimizer in the sqlite3ExprAnd() routine until the corresponding Parse object is deleted. This avoids a dangling pointer in AggInfo if sqlite3ExprAnd() is invoked by the push-down optimization. The dangling pointer appears to be harmless in release builds, only showing up in debug builds. Problem found by dbsqlfuzz. (check-in: 77a30f3f user: drh tags: branch-3.35) | |
| 13:41 | Ensure that negative numbers may not be used in frame offset clauses even if they are initially text value. e.g. (RANGE BETWEEN '-1' PRECEDING ...). (check-in: 2d179f58 user: drh tags: branch-3.35) | |
| 13:31 | Defer deletion of expressions that are optimized out by the AND optimizer in the sqlite3ExprAnd() routine until the corresponding Parse object is deleted. This avoids a dangling pointer in AggInfo if sqlite3ExprAnd() is invoked by the push-down optimization. The dangling pointer appears to be harmless in release builds, only showing up in debug builds. Problem found by dbsqlfuzz. (check-in: c36b4358 user: drh tags: trunk) | |
| 11:31 | Ensure that negative numbers may not be used in frame offset clauses even if they are initially text value. e.g. (RANGE BETWEEN '-1' PRECEDING ...). (check-in: 8b681b27 user: dan tags: trunk) | |
|
2021-03-30
| ||
| 01:57 | Raise an error if a term of the form "TABLE.*" appears in the RETURNING clause, as SQLite does not (yet) know how to handle that. Ticket [132994c8b1063bfb]. (check-in: 257e16f7 user: drh tags: branch-3.35) | |
| 01:52 | Raise an error if a term of the form "TABLE.*" appears in the RETURNING clause, as SQLite does not (yet) know how to handle that. Ticket [132994c8b1063bfb]. (check-in: 3039bcaf user: drh tags: trunk) | |
|
2021-03-29
| ||
| 21:16 | Do not do the EXISTS-to-IN transformation if the sub-select has LIMIT clause. Ticket [de7db14784a08053] (check-in: 3bf8a357 user: dan tags: branch-3.35) | |
| 20:28 | Do not do the EXISTS-to-IN transformation if the sub-select has LIMIT clause. (check-in: d07e246f user: dan tags: trunk) | |
| 20:01 | Simplify the comparison opcodes in the bytecode engine, for a performance improvement. (check-in: f2af5868 user: drh tags: trunk) | |
| 19:47 | Fix VDBE coverage macros. (Closed-Leaf check-in: ebe100de user: drh tags: compare-opcode-opt) | |
| 18:53 | Omit the SQLITE_STOREP2 and SQLITE_KEEPNULL options from the comparison opcodes, allowing them to run faster. This required refactoring the vector comparison logic, which in turn required changing OP_ElseNotEq into OP_ElseEq. (check-in: 380b4605 user: drh tags: compare-opcode-opt) | |
| 14:40 | Add the OP_ZeroOrNull opcode and use it to compute boolean values for scalar comparisons, rather than the SQLITE_STOREP2 parameter to the comparison opcode. (check-in: 93781b6f user: drh tags: compare-opcode-opt) | |
| 13:47 | The comparison opcodes (ex: OP_Eq) now set the iCompare flag so that the result of comparison can be used by subsequent OP_Jump or OP_ElseNotEq opcodes. (check-in: bd00df8f user: drh tags: compare-opcode-opt) | |
|
2021-03-28
| ||
| 23:37 | Alternative implementation of the comparison opcode speed-up of check-in [4a8805d9a66dc888] that should pass muster with UBSAN. (check-in: afb18f64 user: drh tags: trunk) | |
|
2021-03-27
| ||
| 20:45 | Performance improvement in the comparison opcodes by rewriting an array lookup using pointer arithmetic. This makes gcc 5.4.0 builds with -Os both faster and smaller. Though I suspect that in newer compilers with better optimizers in will probably make no difference at all in the generated machine code. Later: Even though the new code is perfectly valid, and generates correct machine code that is both smaller and faster, UBSAN complains about it in OSSFuzz. So we will take it out. So here was have an example of OSSFuzz coercing a widely used application to become bigger and slower. (Leaf check-in: 4a8805d9 user: drh tags: gcc-opt) | |
| 16:21 | For the sqlite3_bind_text16 TCL binding used for testing, ensure that there are at least 3 terminating zeros, so that there will always be a \u0000 character even if the original byte sequence is an odd number of bytes. (check-in: c23d092f user: drh tags: trunk) | |
|
2021-03-26
| ||
| 23:59 | In the sqlite3_bind_text and sqlite3_bind_text16 TCL commands of the test suite, ensure that the string is zero-terminated if the size argument is negative. (check-in: f472fd64 user: drh tags: trunk) | |
| 14:04 | Merge up all the 3.35.3 changes into the begin-concurrent-pnu-wal2 branch. (Leaf check-in: 4d425e64 user: drh tags: begin-concurrent-pnu-wal2) | |
| 13:58 | Merge the 3.35.3 enhancements into the wal2 branch. (Leaf check-in: a1b8d9d2 user: drh tags: wal2) | |