Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
100 check-ins occurring around a9b90aa12eecdd9f.
|
2019-05-10
| ||
| 14:34 | Ensure that the BtShared.nPage value is reset correctly on a SAVEPOINT ROLLBACK. (check-in: e6d5fee8 user: drh tags: trunk) | |
| 14:26 | Treat integer values in window definition ORDER BY clauses as constants, not as references to another expression. (check-in: 7e4809ea user: dan tags: trunk) | |
| 14:03 | Improved comments and extra assert() statements on the OP_Savepoint opcode, just to make the code a little easier to read and reason about. (check-in: e80b2d9f user: drh tags: trunk) | |
| 12:14 | Fix the previous check-in [db9acef14d492121] so that the amalgamation build works for systems for which lack HAVE_ISNAN. (check-in: 3cc55e09 user: drh tags: trunk) | |
| 12:06 | Fix the round() SQL function so that it handles infinities correctly. (check-in: db9acef1 user: drh tags: trunk) | |
|
2019-05-09
| ||
| 18:37 | Fix another small buffer overread in sqlite_dbdata triggered by a corrupt database page. (check-in: 1dfc95b8 user: dan tags: trunk) | |
| 18:33 | Fix a problem in the ".recover" command allowing a circular loop of b-tree pages in a database file to cause an infinite loop. (check-in: 8d2a062e user: dan tags: trunk) | |
| 18:13 | Make the ".schema" command in the CLI resistant to corrupt database files. (check-in: f22c7e22 user: drh tags: trunk) | |
| 17:54 | Fix a crash in the sqlite_dbdata module that could occur if a pointer within an overflow chain in a corrupt database pointed past the end of the db. (check-in: 3eae4e30 user: dan tags: trunk) | |
| 17:10 | Disable the optimization where a REAL value with no fractional part is stored as an INTEGER when the integer uses as much space as the real value it proposes to stand in for (8 bytes). This avoids corner cases of comparing integers against real values that are beyond the resolution of an IEEE 754 double. Fix for ticket [6c1d3febc00b22d457c78c2] (check-in: 9b091527 user: drh tags: trunk) | |
| 16:57 | Avoid long delays that can occur when ".recover"ing data from a database with a corrupt freelist. (check-in: 20f06bf2 user: dan tags: trunk) | |
| 16:22 | Improved debugging output from the OP_MakeRecord opcode. (check-in: ac790729 user: drh tags: trunk) | |
| 15:51 | Fix another corruption related buffer overread in the sqlite_dbdata module. (check-in: 5cd728fb user: dan tags: trunk) | |
| 15:07 | Fix further buffer overreads triggered by passing corrupt records to the sqlite_dbdata module. (check-in: dbc6a9f7 user: dan tags: trunk) | |
| 14:15 | Have ".recover" handle cases where the sqlite_master table contains malformed SQL statements. (check-in: e736da9c user: dan tags: trunk) | |
| 11:45 | Fix a problem with running the ".recover" command on a database that is zero pages in size. (check-in: 47fa6534 user: dan tags: trunk) | |
| 11:34 | Avoid reading outside a record buffer when extracting an SQL value in the sqlite_dbdata virtual table code. (check-in: d99bb0c1 user: dan tags: trunk) | |
| 11:21 | Merge accidental fork in trunk. (check-in: ec937303 user: dan tags: trunk) | |
| 11:19 | Fix a problem in the new code introduced by [658b84d7] causing corruption and other errors to be ignored. (check-in: 7ccf2e7d user: dan tags: trunk) | |
| 11:18 | Improved error handling for the ".open --hexdb" command of the CLI. (check-in: 4fa831d5 user: drh tags: trunk) | |
|
2019-05-08
| ||
| 23:55 | Ensure that the sqlite3BtreeLast() interface initializes the *pRes return to zero if the cursor is already sitting on the last entry of the table. Manuel Rigger found the obscure test case for which this makes a difference. (check-in: ebe4845c user: drh tags: trunk) | |
| 23:53 | This is part of the previous check-in that didn't get committed for some reason - probably because I failed to press the "Save" button on the text editor. (check-in: 51d32173 user: drh tags: trunk) | |
| 21:14 | Add another test case to cover a previously uncovered branch in the RBU module. (check-in: 30392985 user: dan tags: trunk) | |
| 19:55 | Fix another pattern for which the LIKE optimization does not work for a non-TEXT affinity. Case found by Manuel Rigger. (check-in: 74020110 user: drh tags: trunk) | |
| 19:32 | Simplification to the logic underlying PRAGMA case_sensitive_like. (check-in: ef0015fd user: drh tags: trunk) | |
| 19:06 | Avoid an excess register allocation in UPDATE, when possible. This improves speed (slightly) and reduces the code footprint. (check-in: 8658574e user: drh tags: trunk) | |
| 18:49 | Add tests to improve code coverage of the RBU module. (check-in: ecb56b75 user: dan tags: trunk) | |
| 17:27 | Provide the SQLITE_OMIT_CASE_SENSITIVE_LIKE_PRAGMA compile-time option to omit the case_sensitive_like pragma. This change, in combination with documentation changes, is the current solution to ticket [a340eef47b0cad5]. (check-in: eabe7f2d user: drh tags: trunk) | |
| 11:52 | Fix VDBE opcodes OP_SeekLT and OP_SeekLE so that they work on intkey tables with non-numeric text values. Fix for [9cf6c9bb]. (check-in: 658b84d7 user: dan tags: trunk) | |
| 11:42 | Fix VDBE opcodes OP_SeekLT and OP_SeekLE so that they work on intkey tables with non-numeric text values. (Closed-Leaf check-in: a870c196 user: dan tags: tkt-9cf6c9bb) | |
| 04:33 | Remove an ALWAYS() that was previously added by check-in [a0819086] but which turns out can sometimes be false. (check-in: ad8fc5d8 user: drh tags: trunk) | |
| 03:34 | Remove a NEVER() that was added in the previous check-in [a0819086a521f] because it is in fact reachable. (check-in: 74eba255 user: drh tags: trunk) | |
| 00:17 | Earlier detections of errors in the byte-offset-to-cell-content integer at offset 5 in the header of a btree page. (check-in: a0819086 user: drh tags: trunk) | |
|
2019-05-07
| ||
| 20:06 | Generate all records for INSERT or UPDATE prior to running foreign key constraint checks, since the FK checks might modify the datatype of registers used to generate the records. Fix for ticket [e63cbcfd3378afe6980d6]. (check-in: 3c75605b user: drh tags: trunk) | |
| 19:44 | Do not commit an "OR FAIL" statement that causes foriegn key constraint violations. (check-in: 659c551d user: dan tags: trunk) | |
| 19:21 | Add test cases for the fix on this branch. (Closed-Leaf check-in: 2e31abe0 user: dan tags: tkt-e63cbcfd) | |
| 19:13 | On an INSERT or UPDATE, generate the new table record prior to running foreign key checks, in case the foreign key checks changes datatypes on the registers holding column values. Proposed fix for ticket [e63cbcfd3378afe6980d626]. (check-in: 3f1c8051 user: drh tags: tkt-e63cbcfd) | |
| 17:47 | Strive to prevent harmless compiler warnings in GCC 4.8.5. (check-in: 8b6691f6 user: drh tags: trunk) | |
| 16:37 | Optimize the restarting of an RBU vacuum. (check-in: 82062351 user: dan tags: trunk) | |
| 16:28 | Remove some redundant code from sqlite3rbu.c. Add test cases for RBU vacuum. (Closed-Leaf check-in: 7b051698 user: dan tags: rbu-opt) | |
| 02:57 | Add the exprNodeCopy() routine that will safely memcpy() an Expr node that might be a size-reduced node. (Leaf check-in: ab2ba8e7 user: drh tags: expr-node-copy-patch) | |
|
2019-05-06
| ||
| 20:40 | Optimize further cases of restarting an RBU vacuum. (check-in: 6b3261bf user: dan tags: rbu-opt) | |
| 16:15 | Fix a problem with renaming an INTEGER PRIMARY KEY column of a WITHOUT ROWID table using ALTER TABLE. (check-in: 91f701d3 user: dan tags: trunk) | |
|
2019-05-04
| ||
| 20:04 | Optimize some cases of restarting an RBU vacuum. (check-in: cdc09867 user: dan tags: rbu-opt) | |
| 17:32 | Fix the NOT NULL logic in the theorem prover that determines when a partial index can be used. Ticket [5c6955204c392ae763a95]. (check-in: c2e439bc user: drh tags: trunk) | |
| 03:56 | Fix harmless compiler warning seen with MSVC. (check-in: 5862b83e user: mistachkin tags: trunk) | |
| 01:41 | In the sqlite3_value or Mem object, make the MEM_IntReal type completely independent from MEM_Int and MEM_Real. This helps avoid problems when inserting non-float values into a "REAL" column. (check-in: 5a8a23ee user: drh tags: trunk) | |
| 01:29 | New testcase macros to ensure that MEM_IntReal is fully tested. (Closed-Leaf check-in: 8b8ef445 user: drh tags: int-real) | |
|
2019-05-03
| ||
| 21:17 | Add the SQLITE_TESTCTRL_RESULT_INTREAL test-control and use it to create the intreal() SQL function in testfixture. Write a few simple tests to prove this all works. TH3 will hold most of the INTREAL tests, probably. (check-in: c9838731 user: drh tags: int-real) | |
| 19:34 | Ensure that UTF16 strings are properly zero-terminated before returning them in an sqlite3_value_text16() request, even if the string is invalid UTF16 because it was formed from an arbitrary and/or odd-length BLOB. (check-in: 3a16ddf9 user: drh tags: trunk) | |
| 18:50 | Fix a memory-leak/segfault caused by using OP_OpenDup and OP_OpenEphemeral on the same VM cursor. (check-in: a9b90aa1 user: dan tags: trunk) | |
| 17:19 | Fix a problem where self-joins on views that are aggregate queries may return the wrong result. (check-in: 74ef97bf user: dan tags: trunk) | |
| 17:08 | Improved comments on the elements of the array constant used to implement the sqlite3_value_type() interface. (check-in: f73a7de7 user: drh tags: int-real) | |
| 02:41 | Fix the ".open --hexdb" command in the CLI so that it works even with terminal input. (check-in: 9b5d9434 user: drh tags: trunk) | |
|
2019-05-02
| ||
| 21:36 | Make MEM_IntReal a completely independent type, meaning a floating point value stored as an integer. This fixes a problem with arithmetic within arguments to string functions on indexes of expressions. But it is a big change and needs lots of new testcase() macros for MC/DC and so it is initially put on this branch. (check-in: dba836e3 user: drh tags: int-real) | |
| 17:45 | Ensure that the typeof() function always returns SQLITE_FLOAT for floating point values even when the value is stored as an integer to save space. (check-in: 48889530 user: drh tags: trunk) | |
| 17:06 | Add options to wapptest.tcl similar to those supported by releasetest.tcl. Also add the -noui switch, for running without wapp altogether. (check-in: 005a1694 user: dan tags: trunk) | |
| 15:56 | Earlier detection of a database corruption case in balance_nonroot(), to prevent a possible use of an uninitialized variable. (check-in: c509d8a8 user: drh tags: trunk) | |
| 14:15 | Merge enhancements from trunk (check-in: 3a4751a9 user: drh tags: hard-heap-limit) | |
| 01:41 | The "LIKE Optimization" (which converts a LIKE into a BETWEEN operator that is able to use an index) does not work and cannot be used if the collating sequence of the column is not TEXT and if the pattern starts with a "+" sign. This is another case of ticket [c94369cae9b561b1f996d005]. The new test case was discovered by Manuel Rigger. (check-in: b043a54c user: drh tags: trunk) | |
| 00:52 | Fix an issue (discovered by OSSFuzz) in the enhanced OP_Concat operator from check-in [713caa382cf7dd] earlier today. (check-in: 3e897702 user: drh tags: trunk) | |
|
2019-05-01
| ||
| 19:01 | Add a test case for ticket [ae0f637bddc5290b44669e066a]. (check-in: ece48169 user: drh tags: trunk) | |
| 18:59 | When values have real affinity and are converted into strings for CHECK constraints or index expressions, do the conversions into a real-number format even if the values are stored as integers for efficiency. This appears to fix ticket [ae0f637bddc5290b446]. (check-in: 5997d075 user: drh tags: trunk) | |
| 17:36 | Fix an incompatibility with -DSQLITE_OMIT_LOAD_EXTENSION=1 in dbdata.test. (check-in: a77cd85b user: dan tags: trunk) | |
| 17:32 | Fix a case in wapptest.tcl where a failed test might report 0 errors. (check-in: 2be1ed70 user: dan tags: trunk) | |
| 15:32 | Avoid unwelcomed side effects on the input operands in the OP_Concat operator. Fix for ticket [3be1295b264be2fac49b681] (check-in: 713caa38 user: drh tags: trunk) | |
| 15:25 | Update wapptest.tcl so that it deletes extra files if the "Keep files:" checkbox is clear. Set it by default. (check-in: 09623cc4 user: dan tags: trunk) | |
| 14:41 | In "PRAGMA vdbe_trace" output, show the results of OP_Affinity opcodes. (check-in: 56604bb6 user: drh tags: trunk) | |
| 08:48 | Fix an incompatibility with auto-vacuum mode in new test script recover.test. (check-in: 36dd5b08 user: dan tags: trunk) | |
|
2019-04-30
| ||
| 20:43 | Update wapptest.tcl to use a simpler slave script. And to leave scripts wapptest_configure.sh and wapptest_make.sh in each test directory. (check-in: 07e527d7 user: dan tags: trunk) | |
| 15:36 | Fix a problem allowing a Table object to be deleted from within a call to the xDestroy method of the associated virtual table, causing a use-after-free error. (check-in: 1dbbb010 user: dan tags: trunk) | |
| 14:26 | Fix an error message in the Lemon parser generator. (check-in: b6d7d42b user: drh tags: trunk) | |
| 11:54 | Small performance increase and size reduction in the implementation of the LIKE and GLOB operators. (check-in: f97626f9 user: drh tags: trunk) | |
| 01:08 | Slightly smaller and faster implementation of the OP_MakeRecord opcode. (check-in: 3bdce7ef user: drh tags: trunk) | |
|
2019-04-29
| ||
| 16:44 | Changes to oserror.test so that it works even on systems that allow an unusually large number of file descriptors. (check-in: a27b0b88 user: drh tags: trunk) | |
| 13:48 | Do not de-duplicate columns index columns associated with a WITHOUT ROWID table if the columns have different collating sequences. This is the fix for ticket [3182d3879020ef3b2]. There is one test case added, but most of the tests are done in TH3. (check-in: 1b1dd4d4 user: drh tags: trunk) | |
| 13:30 | Improved header comment and precondition checking for the new isDupColumn() function. (Closed-Leaf check-in: 740d5ff6 user: drh tags: tkt-3182d38790) | |
| 11:41 | Fix a buffer overwrite in shell.c.in (part of the new .recover code). (check-in: 92facbc7 user: dan tags: trunk) | |
| 11:27 | Fix a stack overflow that could occur when renaming a table that has a trigger containing a window function invocation that itself contains a specific syntax error. (check-in: c621fc66 user: dan tags: trunk) | |
|
2019-04-28
| ||
| 19:27 | Take collating sequence into account when removing redundant columns from indexes on WITHOUT ROWID tables. This is the first proof-of-concept fix for ticket [3182d3879020ef3]. More testing needed. (check-in: b34fa5bf user: drh tags: tkt-3182d38790) | |
|
2019-04-27
| ||
| 20:39 | Add the shardvtab virtual table that uses the new cost estimation functions. (Leaf check-in: 9404300a user: drh tags: cost-est) | |
| 20:30 | Add the ".recover" command to the shell tool. For recovering as much data as possible from corrupt databases. (check-in: 50fe4845 user: dan tags: trunk) | |
| 20:16 | Fix a minor typo in a comment. No changes to code. (check-in: 95209072 user: drh tags: trunk) | |
| 20:15 | Fix building the shell with SQLITE_OMIT_VIRTUAL_TABLE. And without SQLITE_ENABLE_DBPAGE_VTAB. (Closed-Leaf check-in: 425d708c user: dan tags: dbdata) | |
| 19:36 | Add comments and fix formatting issues in new code in shell.c.in. (check-in: b91d819b user: dan tags: dbdata) | |
| 18:47 | Add the "--lost-and-found" option to the ".recover" command. For setting the name of the orphaned rows table. (check-in: 67bb88e2 user: dan tags: dbdata) | |
| 15:35 | Fix a problem in the .recover command with recovering WITHOUT ROWID tables where the PK columns are not the leftmost in the CREATE TABLE statement. (check-in: 91df4b8e user: dan tags: dbdata) | |
|
2019-04-26
| ||
| 21:11 | Have .recover store all orphaned rows in a single table, with extra columns to indicate the orphaned page and sub-tree they were discovered within. (check-in: 7221f6e3 user: dan tags: dbdata) | |
| 17:20 | An experimental interface for retrieving the estimated cost and estimated number of output rows for a query. (check-in: 1b25fa10 user: drh tags: cost-est) | |
| 17:08 | Omit tests of the LIKE optimization in like3.test when SQLITE_ENABLE_ICU is defined. (check-in: af53c41a user: dan tags: trunk) | |
| 15:40 | Fix another problem with database freelist handling in the ".recover" command. (check-in: bee2652a user: dan tags: dbdata) | |
| 15:14 | Fix a locking-page related problem with the ".recover" command. (check-in: afdae104 user: dan tags: dbdata) | |
|
2019-04-25
| ||
| 20:06 | Merge latest trunk changes into this branch. (check-in: 1da302d8 user: dan tags: dbdata) | |
| 19:23 | Unless the "--freelist-corrupt" option is specified, do not have the .recover command attempt to recover data from pages that are on the database free-list. (check-in: 8d2f52bb user: dan tags: dbdata) | |
| 18:15 | Add the sqlite3_hard_heap_limit64() interface and the corresponding "PRAGMA hard_heap_limit=N" command. (check-in: b0ccef61 user: drh tags: hard-heap-limit) | |
| 16:20 | Fix a bug preventing .recover from working on databases where the final page of the db is corrupt. (check-in: 959bbd11 user: dan tags: dbdata) | |
|
2019-04-24
| ||
| 20:48 | Improve the performance of the .recover command. (check-in: a5076831 user: dan tags: dbdata) | |
| 17:04 | New test cases in test/fuzzdata8.db. (check-in: 7be6222c user: drh tags: trunk) | |
| 16:13 | Fix another instance in fts3 where a corrupt record can cause a buffer overflow. (check-in: e1724f1d user: dan tags: trunk) | |
| 15:57 | Fix a problem in fts5 where a corrupt position list could lead to a buffer overwrite. (check-in: 516ca894 user: dan tags: trunk) | |