50 most recent check-ins
|
2026-06-17
| ||
| 20:11 | Small parsing performance improvement by reducing the amount of the Parse object that needs to be initialized. (leaf check-in: 0901b7c2df ... user: drh tags: trunk) | |
| 19:28 | Remove the unnecessary Parse.szOpAlloc field, for small size reduction and performance improvement. (check-in: 9a2a230464 ... user: drh tags: trunk) | |
| 18:39 | Small performance boost in the sqlite3VdbeResolveLabel() routine of the code generator. (check-in: f786f3643e ... user: drh tags: trunk) | |
| 16:57 | Safer conversion from double to 64-bit integer in two places. (check-in: 17c4443829 ... user: drh tags: trunk) | |
| 14:53 | Additional refinements to the large floating-point handling in generate_series(). Follow-up to [5aab8c2cde63f9db]. (check-in: 4247c5d9af ... user: drh tags: trunk) | |
| 14:12 | Fix CLI CSV test cases so that they suppress the CR on windows and thus work cross-platform. Test case changes only - no changes to code. (leaf check-in: 89d9120af9 ... user: drh tags: branch-3.53) | |
| 13:52 | Fix CLI CSV test cases so that they suppress the CR on windows and thus work cross-platform. (check-in: 7a2c96405e ... user: drh tags: trunk) | |
| 13:31 | Fix the OR-to-IN optimization so that it works in cases where there are conflicting collation sequences on the equality constraints within the OR. Bug 2026-06-17T05:04:48Z. (check-in: 984c9b1818 ... user: drh tags: trunk) | |
| 09:46 | Improved detection of when the alternative windows-compatible main() is needed in the sqldiff.exe utility. (check-in: 571846a9c2 ... user: drh tags: branch-3.53) | |
|
2026-06-16
| ||
| 23:59 | Typo fix in a comment used to generate documentation. No changes to code. Forum 2026-06-16T23:18:50Z. (check-in: 7386faef61 ... user: drh tags: trunk) | |
| 18:06 | After opening a non-covering index cursor, pass the corresponding table cursor to the implementation via sqlite3BtreeCursorHint(). (leaf check-in: d48c6d2dcd ... user: dan tags: cursor-hint-tblcsr) | |
| 13:46 | Limit VIEW recursion depth to SQLITE_LIMIT_EXPR_DEPTH to prevent static overflow in malicious schemas with tens of thousands of levels of recursive views. (check-in: 38430aa5ed ... user: drh tags: branch-3.53) | |
| 13:43 | Limit VIEW recursion depth to SQLITE_LIMIT_EXPR_DEPTH to prevent static overflow in malicious schemas with tens of thousands of levels of recursive views. Bug 2026-06-16T04:21:51Z (check-in: 3f3fb9b638 ... user: drh tags: trunk) | |
| 12:02 | Fix instances of potential buffer overflow caused by loading the contents of a stat4 table with an absurd number of samples in a 32-bit build. (check-in: c25ac7af86 ... user: drh tags: branch-3.53) | |
| 11:53 | Fix instances of potential buffer overflow caused by loading the contents of a stat4 table with an absurd number of samples in a 32-bit build. Bug 2026-06-16T11:30:52Z. (check-in: 0b72246732 ... user: dan tags: trunk) | |
| 09:53 | Fix a possible call to memcpy() with a NULL source pointer when the size parameter is zero (which is technically undefined behavior but in practice is completely harmless) in the sha1_query() extension function. Bug 2026-06-16T07:42:23Z. (check-in: 2063926473 ... user: drh tags: trunk) | |
|
2026-06-15
| ||
| 17:39 | Consistent error messages for excess trigger nesting. (check-in: 4fb9f2ffb4 ... user: drh tags: trunk) | |
| 17:27 | Speed up SQL aggregate functions percentile() and median() by using quickselect to find the required values instead of fully sorting the array of values with quicksort. (check-in: 0796d337f6 ... user: dan tags: trunk) | |
| 17:14 | New Windows makefile targets for common EXEs that omit the ".exe" suffix, so that when muscle memory kicks in and we type "make sqlite3" on windows, it still works. (check-in: 1152463a66 ... user: drh tags: trunk) | |
| 17:13 | Improve comments in percentSort(). (closed check-in: 9b43500f51 ... user: dan tags: pecentile-partial-sort) | |
| 16:49 | Enhance SQLITE_LIMIT_TRIGGER_DEPTH so that it also limits a chain of distinct triggers during code generation. (check-in: b22709dcb3 ... user: drh tags: branch-3.53) | |
| 16:46 | Fix a problem sometimes causing an UPSERT to incorrectly convert an excluded.* column from type REAL to INT in cases where doing so does not cause data loss. (check-in: 5d344e0288 ... user: drh tags: branch-3.53) | |
| 16:45 | Enhance SQLITE_LIMIT_TRIGGER_DEPTH so that it also limits a chain of distinct triggers during code generation. Bug 2026-06-15T06:44:07Z. (check-in: 0b5378678e ... user: drh tags: trunk) | |
| 16:27 | Improvements to handling of inequality constraints on generate_series().value against very large floating-point values. (check-in: e10ff7cd09 ... user: drh tags: branch-3.53) | |
| 15:48 | Change the PACKAGE_BUGREPORT config option point to /bugs instead of /forum. (check-in: ca15fedede ... user: stephan tags: trunk) | |
| 15:38 | Fix a problem sometimes causing an UPSERT to incorrectly convert an excluded.* column from type REAL to INT in cases where doing so does not cause data loss. Bug 2026-06-15T09:08:12Z. (check-in: 71334f7ec9 ... user: dan tags: trunk) | |
| 15:27 | Improvements to handling of inequality constraints on generate_series().value against very large floating-point values. Bug 2026-06-15T04:37:48Z. (check-in: 5aab8c2cde ... user: drh tags: trunk) | |
| 15:24 | Additional tweaks and test cases for generate_series() constraints using very large floating-point numbers. (closed check-in: 50d611404a ... user: drh tags: series-boundry-cases) | |
| 14:16 | Better behavior for generate_series() in cases where constraints involve floating point number that are near the minimum and maximum 64-bit integer values. (check-in: 236ef21789 ... user: drh tags: series-boundry-cases) | |
| 13:53 | Avoid a possible buffer overrun in 32-bit builds of fts5 that could occur when processing corrupt records. (check-in: 3a4667ce36 ... user: drh tags: branch-3.53) | |
| 13:39 | Avoid a possible buffer overrun in 32-bit builds of fts5 that could occur when processing corrupt records. Bug 2026-06-14T12:21:15Z. (check-in: 0fa3345fe4 ... user: dan tags: trunk) | |
| 12:36 | Revert the native impl of kvvfs's xOpen() (as distinct from the JS impl which the wasm build uses) to the historical db name restrictions of 'local' or 'session', failing with SQLITE_CANTOPEN if passed another name. [ec866b04d088e53b] overhauled support for kvvfs db names in JS but it turns out that the native impl still relies on those names in order to match journals to databases. Correct a related too-lenient assert() in the WASM pieces. (check-in: 4b3ec30c63 ... user: stephan tags: trunk) | |
| 12:22 | Revert the native impl of kvvfs's xOpen() (as distinct from the JS impl which the wasm build uses) to the historical db name restrictions of 'local' or 'session', failing with SQLITE_CANTOPEN if passed another name. [ec866b04d088e53b] overhauled support for kvvfs db names in JS but it turns out that the native impl still relies on those names in order to match journals to databases. Correct a related too-lenient assert() in the WASM pieces. Edit: mistake: checked in an incomplete in-between test... (closed check-in: fabcad0917 ... user: stephan tags: mistake) | |
| 11:52 | Fix a problem that could cause fts5 to return incorrect answers when performing prefix queries on tokens with synonyms. (check-in: 4f21004f9b ... user: drh tags: branch-3.53) | |
| 11:47 | Fix a problem that could cause fts5 to return incorrect answers when performing prefix queries on tokens with synonyms. Bug 2026-06-14T11:11:06Z. (check-in: 323d87541a ... user: dan tags: trunk) | |
| 11:03 | Regardless of whether the platform is 32 or 64-bits, Use 64-bit arithmetic to calculate buffer sizes in the fts3_snippet(). Bug 2026-06-14T09:19:52Z. (check-in: 79dc1949b2 ... user: dan tags: trunk) | |
| 10:16 | Fix rounding behavior in the decimal extension when all leading digits are 9 and there is not fractional component. (check-in: ae568e059b ... user: drh tags: branch-3.53) | |
| 10:13 | Fix rounding behavior in the decimal extension when all leading digits are 9 and there is not fractional component. Bug 2026-06-15T03:49:42Z. (check-in: c6a3b18ee9 ... user: drh tags: trunk) | |
| 01:24 | Prevent array overflow when decoding a corrupted value from the underlying key/value store. Bug 2026-06-14T21:58:12Z (check-in: 732c8f81b5 ... user: drh tags: trunk) | |
|
2026-06-14
| ||
| 22:01 | Remove the unused, untested, experiemental "blobio.c" extension, to avoid confusing AIs. Bug 2026-06-14T15:12:50Z (check-in: 45918d5e97 ... user: drh tags: trunk) | |
| 20:10 | Do not allow ridiculous "columns=N" values in the (unused) csv virtual table in the extensions folder. Bug 2026-06-14T15:22:47Z (check-in: ecae4a2ddf ... user: drh tags: trunk) | |
|
2026-06-13
| ||
| 20:16 | Speed up SQL aggregate functions percentile() and median() by avoiding a full sort of the array of arguments. (check-in: 9232b5f21d ... user: dan tags: pecentile-partial-sort) | |
| 18:17 | Make the --header option to the CLI sticky, so that it does not get turned off by subsequence --csv or similar mode change options. (check-in: 5019306601 ... user: drh tags: branch-3.53) | |
| 18:04 | Make the --header option to the CLI sticky, so that it does not get turned off by subsequence --csv or similar mode change options. Bug 2026-06-13T11:49:18Z (check-in: b19cef1862 ... user: drh tags: trunk) | |
|
2026-06-12
| ||
| 18:38 | When reading a super-journal name from a journal file, allocate a new buffer rather than using Pager.pTmpSpace. This prevents a buffer overrun that could occur when using a VFS with a large sqlite3_vfs.mxPathname value with a database with a small page size. (check-in: aa756c8038 ... user: drh tags: branch-3.53) | |
| 18:35 | When reading a super-journal name from a journal file, allocate a new buffer rather than using Pager.pTmpSpace. This prevents a buffer overrun that could occur when using a VFS with a large sqlite3_vfs.mxPathname value with a database with a small page size. (check-in: 7f71859841 ... user: dan tags: trunk) | |
| 17:32 | Fix errors in comments on this branch. Also prefer SQLITE_NOMEM_BKPT to SQLITE_NOMEM. (closed check-in: ac17669e84 ... user: dan tags: super-journal-malloc) | |
| 17:16 | Fix a memory leak that could follow an OOM error in the new code on this branch. (check-in: d4774a8987 ... user: dan tags: super-journal-malloc) | |
| 16:47 | Improved documentation for sqlite3_deserialize(). (check-in: f15d076820 ... user: drh tags: trunk) | |
| 16:26 | Do not use shared locks on SHM files when the filename is a long DOS-device path. Only used shared locks for UNC paths. Do not confuse the long-DOS-device syntax with UNC paths. (check-in: a4691489f2 ... user: drh tags: trunk) | |