SQLite

Timeline
Login

Many hyperlinks are disabled.
Use anonymous login to enable hyperlinks.

100 check-ins occurring around 24602557fc3295fe.

2011-03-17
01:58
Comment enhancement to better explain the logic in the "x IS NULL" optimization. (check-in: 869f8947 user: drh tags: trunk)
01:53
Backport the "x IS NULL" query planner enhancement of [2353176811f] to the 3.7.2 branch. (check-in: 68daf20d user: drh tags: branch-3.7.2)
01:34
Enhances to the query planner such that "x IS NULL" constraints take the STAT2 statistics into account, just like "x=VALUE" constraints. (check-in: 23531768 user: drh tags: trunk)
2011-03-16
19:59
Add the sqlite3_preupdate_new() API, for retrieving the new.* values from within a pre-update callback. (check-in: 526545c4 user: dan tags: sessions)
18:54
Additional interpretation of flags and constants in the VFS trace output. (check-in: 3e984195 user: drh tags: trunk)
17:05
Add the -vfstrace option to the usage error message in the shell. (check-in: baca45c5 user: drh tags: trunk)
16:56
Add the VFS-trace shim. (check-in: f49a9ef3 user: drh tags: trunk)
09:49
Remove the sqlite3_transaction_hook() API. (check-in: b0015a1c user: dan tags: sessions)
2011-03-15
19:08
Make the "unix-excl" VFS work exactly like "unix" if the database file is read-only. (check-in: d9846834 user: drh tags: trunk)
18:35
Fix an out-of-order variable declaration in shell.c. (check-in: 72570846 user: drh tags: trunk)
16:37
Fix some bugs and other code issues in the session module. (check-in: f2930840 user: dan tags: sessions)
04:45
Allow multiplex file names to be preceeded by prefix of the form ":multiplex:chunksize:maxchunks:" Still work to be done, though it compiles and prefixes are ignored. (check-in: cfa4a2f7 user: shaneh tags: multiplex-enhancements)
02:55
Fix cut-and-paste typo in debugging print statement in winMutexTry(). (check-in: def98fd2 user: shaneh tags: trunk)
2011-03-14
19:49
Fix handling of return values from the conflict handler. Document the conflict handler arguments and return codes in sqlite3session.h. (check-in: cbbb274e user: dan tags: sessions)
13:54
Merge the unix-excl VFS into the trunk. This merge also adds the -vfs option to the command-line shell. (check-in: 3934b004 user: drh tags: trunk)
2011-03-12
18:10
In the "unix-excl" VFS, use the heap for shared memory, since only a single process is able to read or write the database. (Closed-Leaf check-in: a05a6d40 user: drh tags: unix-excl)
17:22
Fix some issues with UPDATE changes in the session module. (check-in: 57862efe user: dan tags: sessions)
17:02
Add the new optional "unix-excl" VFS. This VFS grabs an exclusive lock on the database preventing other processes from accessing it, but continues to allow other database connections from the same process. (check-in: 00051c32 user: drh tags: unix-excl)
05:55
Backport the SQLITE_OMIT_UNIQUE_ENFORCEMENT change from check-in [b86999436e] on the trunk. (check-in: 668b91db user: shaneh tags: branch-3.7.2)
04:58
More tests for SQLITE_OMIT_UNIQUE_ENFORCEMENT and minor change to implementation. (check-in: b8699943 user: shaneh tags: trunk)
2011-03-11
19:05
Add the sqlite3changeset_apply() function. Does not yet handle all cases. (check-in: 2b19be7b user: dan tags: sessions)
16:15
Remove an unused field from the unix sqlite3_file object. (check-in: f957f23a user: drh tags: trunk)
2011-03-10
21:48
Simplification of tests and more added for SQLITE_OMIT_UNIQUE_ENFORCEMENT tests. (check-in: 75a38411 user: shaneh tags: trunk)
21:13
Skip unique constraint enforcement if compiled with SQLITE_OMIT_UNIQUE_ENFORCEMENT. (check-in: ba85bf8c user: shaneh tags: trunk)
03:54
Minor clean-up of previous mem5 allocator fix. (check-in: 36438423 user: shaneh tags: trunk)
2011-03-09
22:09
Backport the OP_Next and OP_Prev for UNIQUE indices patch from checkin [f000c9b2b7] on the trunk. (check-in: 2d55234e user: drh tags: branch-3.7.2)
21:36
Fix issue with mem5 allocator when min request size is larger thatn 2^30. (check-in: d7dae06f user: shaneh tags: trunk)
21:02
Omit unnecessary OP_Next and OP_Prev operators when uniqueness constraints guarantee that the code will only make one pass through the loop. (check-in: f000c9b2 user: drh tags: trunk)
11:17
Add a function to the session extension invert a changeset. (check-in: 75d5dff7 user: dan tags: sessions)
11:04
Updates to the OS/2 implementation from Rich Walsh. (check-in: dc46156a user: drh tags: trunk)
2011-03-08
19:22
Add start of sessions feature. (check-in: 269a81a3 user: dan tags: sessions)
16:39
Merge the syscall-override changes into trunk. (check-in: 36d79e6f user: drh tags: trunk)
14:05
Unix errno integer can vary from one system to the next. So do not depend on specific errno values in test cases. (check-in: ddb747d3 user: drh tags: trunk)
13:06
Fix two compiler warnings. No functional code changes. (check-in: c829868a user: drh tags: trunk)
02:38
Fix additional cases of possible signed integer overflow, especially with regard to negation. (check-in: 2d5800bd user: drh tags: trunk)
2011-03-07
16:12
Backport a fix from [6b0cabd017ed25] (version 3.6.16 - 2009-03-05) to the 3.6.1 branch, a fix that prevents a null pointer deref if one of the column access routines such as sqlite3_column_int() is called with a NULL first parameter. (check-in: df343640 user: drh tags: branch-3.6.1)
2011-03-06
21:54
Another minor simplification brought to light by clang. (check-in: 3bfbf026 user: drh tags: trunk)
21:28
Remove dead code identified by the clang static analyzer. (check-in: 01a79d5a user: drh tags: trunk)
2011-03-05
21:41
Simplifications to the overflow-free multiplier. Also remove some commented-out code that was left in that subroutine by mistake on the previous check-in. (check-in: 55fc25fd user: drh tags: trunk)
20:59
Fix all known instances of signed-integer overflow. Within SQL expressions, integer overflow now forces coercion to floating point. The shift operators work with any integer right-hand operand with negative values reversing the direction of the shift. (check-in: abf21394 user: drh tags: trunk)
13:54
Fix an instance of signed arithmetic overflow and an one bit-shift overflow. Mark six other signed arithmetic overflow locations that need fixing. (check-in: 04abab71 user: drh tags: trunk)
2011-03-04
15:43
Handle EINTR errors from open(). (Closed-Leaf check-in: a7d176b2 user: drh tags: syscall-override)
01:23
Backport the query planner enhancement of [952f5e8c69904] to the 3.7.2 branch. (check-in: 440d9956 user: drh tags: branch-3.7.2)
00:56
Do a better job of choosing the join table order when the tables having very different numbers of rows. (check-in: 952f5e8c user: drh tags: trunk)
2011-03-03
20:06
Add the experimental sqlite3_transaction_hook() API. (check-in: 093d8cd8 user: dan tags: sessions)
2011-03-02
22:07
Fix quoting of the result in rtreeB.test. (check-in: c6532b35 user: drh tags: trunk)
19:06
Add additional VFS methods to retrieve system call pointers and to get a list of all changeable system calls. (check-in: 38558363 user: drh tags: syscall-override)
18:01
Add more system calls to the set that can be overridden in os_unix.c. Also merge in recent fixes from trunk. (check-in: 80fac2a6 user: drh tags: syscall-override)
17:54
Fix bugs in [7b6e30e6a7] that only show up on Mac. (check-in: ec55e8c6 user: drh tags: trunk)
15:44
Make sure the rtreenode() testing and analysis routine in the RTREE extension can handle 64-bit rowids. This fix is in response to a message on the mailing list. (check-in: 24602557 user: drh tags: trunk)
15:09
Proof-of-concept prototype for the proposed xSetSystemCall extension method on the VFS. (check-in: 92b5a76a user: drh tags: syscall-override)
02:08
Log all error from close() in os_unix.c to sqlite3_log() but do not attempt to report errors back up to the application. Update the unix error logging to put the most important information earlier in the message. (check-in: 7b6e30e6 user: drh tags: trunk)
2011-03-01
18:42
Add the experimental sqlite3_preupdate_hook() API. (check-in: 6145d7b8 user: dan tags: sessions)
2011-02-25
03:25
Comment out some code in os_unix.c that only runs on MacOSX with SQLITE_ENABLE_LOCKING_STYLE. (check-in: 4e50b036 user: drh tags: trunk)
2011-02-24
21:10
Fix a typo in a comment. No changes to code. (check-in: af475618 user: drh tags: trunk)
2011-02-23
22:39
Reserve a range of bits in the SQLITE_OPEN_xxxx bit vector for future expansion. (check-in: 9b904654 user: drh tags: trunk)
14:33
Automatically retry system calls that fail with EINTR. This is a backport of the changes from [b9d29ea385bafc] and [af9ba2a6d2c379]. (Leaf check-in: 8609a15d user: drh tags: branch-3.7.4)
14:05
Backport the os_unix.c error logging enhancements from check-in [01076528a43b61a]. (check-in: a4333b15 user: drh tags: branch-3.7.4)
14:00
Fix a typo in the robust_flock() macro for systems without EINTR. (check-in: af9ba2a6 user: drh tags: trunk)
13:53
The robust_flock() fix that accidently included some unrelated, though harmless changes. I should follow my own checklist! (Closed-Leaf check-in: e701efbd user: drh tags: mistake)
13:33
Retry selected system calls on unix when they fail with EINTR. (check-in: b9d29ea3 user: drh tags: trunk)
2011-02-22
03:34
When a stale schema-cookie is seen, expire only the one statement that encountered the bad cookie, not every statement on the database connection. Ticket [b72787b1a7cea1f] (check-in: 1bca0a7e user: drh tags: trunk)
2011-02-21
17:49
Fix a problem with "EXPLAIN QUERY PLAN SELECT count(*) FROM tbl". (check-in: 9f9f3288 user: dan tags: trunk)
11:46
Have os_unix.c call sqlite3_log() following errors in certain system calls. (check-in: 01076528 user: dan tags: trunk)
2011-02-20
21:03
Pull in the fix to STAT2 processing from check-in [70a3d81742f]. (check-in: 692aafb1 user: drh tags: branch-3.7.4)
03:40
Do not report the database corruption if the the db size header field is greater than the file size on disk unless the two change-counter header fields are identical. Fix for ticket [89b8c9ac54]. Backport by cherrypick of [00c4596f0b270]. (check-in: e2616004 user: drh tags: branch-3.7.4)
03:32
Make sure the change-counter and SQLite-version fields of the header are set correctly even after vacuuming. This is a backport of changes [0be92a7576] and [04fa1e1690] to address ticket [5d863f876ee]. (check-in: 442be135 user: drh tags: branch-3.7.4)
03:27
Do not raise an SQLITE_CORRUPT error in Recoverymode if the database size in the header is larger than the physical file size. This is a cherrypick of checkin [114640d920e16c8] (check-in: 7701b077 user: drh tags: branch-3.7.4)
03:22
Backport the sqlite3.h generator fix so that it works with newer versions of Fossil. See [3513bf6ee090d9b] for the original. (check-in: 29597a71 user: drh tags: branch-3.7.4)
03:20
Backport the changes of checkin [cf86affcb7d308949] ("Make wal_checkpoint a no-op if a prior checkpoint has already copied all WAL content into the database.") to the 3.7.4 release. (check-in: e6e540ab user: drh tags: branch-3.7.4)
03:11
Backport the SQLITE_PROTOCOL fix and the extra defensive measure to version 3.7.4. (check-in: bcc22c4b user: drh tags: branch-3.7.4)
2011-02-19
23:18
An SQLITE_PROTOCOL error counts as a locking error for verification purposes. (check-in: e87d499a user: drh tags: trunk)
17:02
Change a testcase() added by the previous checkin into an assert(). (check-in: 6f3dad32 user: drh tags: trunk)
16:51
Add testcase macros to verify that all return values from walTryBeginRead() are tested. (check-in: 262b6fca user: drh tags: trunk)
15:22
Changes to make WAL more robust against SQLITE_PROTOCOL errors. (check-in: fd578a32 user: drh tags: trunk)
14:19
Changes which attempt to address an obscure SQLITE_PROTOCOL error. (Leaf check-in: 1725aa75 user: drh tags: wal-trace-375)
2011-02-18
17:23
Fix the sqlite3ValueFromExpr() routine so that it returns SQLITE_NOMEM on an OOM when trying to extract a NULL. (check-in: 1061e94f user: drh tags: trunk)
2011-02-17
15:58
Remove a no-op code path from sqlite3ExprIsInteger(). Replace it with an assert() that proves it always does nothing. (check-in: 7af66d1b user: drh tags: trunk)
13:52
Remove an assert() that was made redundant by the previous checkin. (check-in: 21db7191 user: drh tags: trunk)
13:33
Add an ALWAYS() around a always-true test in where.c. (check-in: 8123283e user: drh tags: trunk)
2011-02-16
23:32
Fix harmless compiler warnings in the query planner. (check-in: 31fc4ba6 user: drh tags: trunk)
01:23
Add "do_not_use_codec" logic to some of the TCL based test cases that read/write directly to the DB. (check-in: edd27669 user: shaneh tags: trunk)
2011-02-14
03:49
Skip some tests in capi3e.test when compiled with SQLITE_OMIT_UTF16. (check-in: b04304b9 user: shaneh tags: trunk)
2011-02-12
14:23
Fix the expected output on tests so that it corresponds to the new query planner results. All of veryquick.test is now passing with SQLITE_ENABLE_STAT2. (check-in: f2a8b5cc user: drh tags: branch-3.7.2)
05:34
Fix problems in the backport, reducing the number of errors in the TCL tests to just a few dozen. Most of the remaining errors seem to be real and desirable changes of behavior. (check-in: 9d2b0af2 user: drh tags: branch-3.7.2)
01:59
This is the beginning of an attempt to backport recent query planner enhancements to version 3.7.2. The code in this version builds and runs and seems to give correct answers, but it generates suboptimal query plans and hence many of the test cases fail. The test script gives up after 1000 errors. (check-in: e72cf118 user: drh tags: branch-3.7.2)
2011-02-11
22:54
Add a NEVER() around a test that is believed to always be false. (check-in: f7e2ea33 user: drh tags: trunk)
20:52
Skip flattening if subquery has LIMIT and outer query is DISTINCT. Fix for ticket [752e1646fc]. (check-in: 55973999 user: shaneh tags: trunk)
06:59
Fix a bug in the new WHERE-clause processing that tries to use an index to resolve IS NOT NULL constraints when SQLITE_ENABLE_STAT2 is defined. The bug could cause memory overruns and segfaults. The bug was new to the code and has not appeared in an official release. Found during structural testing. (check-in: a5c36b9f user: drh tags: trunk)
03:56
Allow an index paired with an IS NULL constraint to be used for sorting under the condition that the index be treated as a non-unique index. (check-in: d78949fc user: drh tags: trunk)
02:43
Disable unused NULL tests when SQLITE_ENABLE_STAT2 is not in use. (check-in: 5ecd1178 user: drh tags: trunk)
2011-02-10
21:08
Add .testctrl option to CLI. (check-in: f85afa0e user: shaneh tags: trunk)
18:56
Split the documentation for sqlite3_wal_checkpoint_v2() and its constants onto separate pages. No changes to code. (check-in: 0ef8ffd1 user: drh tags: trunk)
17:46
Prevent a segfault when automatic indices try to use a column with an unknown collating function. Ticket [77aa3b1e6592582e38605d36]. This check-in also removes some stray \r characters unrelated to the problem. (check-in: f01030a0 user: drh tags: trunk)
01:49
This is a version of the SQLite 3.7.5 release with Apple's changes for MacOS. (check-in: 55d2e55b user: drh tags: apple-osx)
00:08
Refactor the cost function in the query planner. Give extra cost (thus reduce likelihood of selection) to full table scans. (check-in: 878da276 user: drh tags: trunk)
2011-02-09
19:55
Make sure code *compiles* with each OMIT and ENABLE option. Mostly changes to test modules. (check-in: 7cc515ed user: shaneh tags: trunk)
18:19
Do not report corruption if the the db size header field is greater than the file size on disk unless the two change-counter header fields are identical. Fix for ticket [89b8c9ac54]. (check-in: 00c4596f user: dan tags: trunk)
15:25
Update Makefile.in for fts3_aux changes. (check-in: 38b7cb33 user: shaneh tags: trunk)
03:04
Use macros to define the relative costs of search and seek operations when computing costs in the query planner. Current constants seems wrong and need to be fixed, but doing so will alter test results. Need more experimentation to determine accurate relative costs. (check-in: 5f2ec44b user: drh tags: trunk)