SQLite

Timeline
Login

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

100 descendants and 100 ancestors of 9f6ea1de5abab0ca28688e257ddf03c66413cf6e

2010-07-06
10:55
Before calling Tcl_ExitThread() in a multi-threaded test, call Tcl_DoOneEvent() as many times as necessary to handle any queued events. (check-in: b3399b40 user: dan tags: trunk)
09:29
Reorder variable declarations in the previous check-in to avoid putting code before declarations when not testing. (check-in: d1fe8ab4 user: drh tags: trunk)
07:36
Ensure the correct error code is returned if an attempt to parse a database schema made by an ATTACH statement fails. (check-in: c2721961 user: dan tags: trunk)
2010-07-05
21:00
Modify the VFS xAccess() method on winNT so that it returns false for an exists test of a zero-length file. This makes the windows VFS work the same as the unix VFS. (check-in: ec35f254 user: drh tags: trunk)
19:13
Simplify the previous commit by removing the pagerCheckForOrDeleteWAL() wrapper. (check-in: a1324d12 user: dan tags: trunk)
19:03
Allocate a buffer containing the full path-name to the associated WAL file when a pager is created. This saves having to construct a new buffer each time a new read-transaction is opened and SQLite checks for the existance of a WAL file. (check-in: 3053a4ad user: dan tags: trunk)
17:43
When rolling back page 1 from a journal, restore the "nReserve" setting in case it has been corrupted by a prior crash. (check-in: c0d124da user: drh tags: trunk)
14:54
Do not report an error if the open-file-count is not as it should be after running a multi-threaded test. The counter instrumentation is not thread-safe. (check-in: ea80b21c user: dan tags: trunk)
12:54
Call Tcl_ExitThread() from within threads created using Tcl_CreateThread(). Not doing so causes notifier related errors on OSX. (check-in: 684eae66 user: dan tags: trunk)
05:54
Add any files specified as part of the QUICKTEST_INCLUDE environment variable to the set of test files run as part of "make test". Release testing uses this trick. (check-in: a40a6e7d user: dan tags: trunk)
05:31
Fix typos in permutations.test. (check-in: b6600ecb user: dan tags: trunk)
2010-07-03
19:08
Do not run the memsubsys1.test script under the memsubsys1 permutation. (check-in: 3a1a8c77 user: dan tags: trunk)
17:13
Remove the unused xRename() method from the sqlite3_vfs object. Add better documentation on the xCurrentTimeInt64() method. (check-in: 51ec0e54 user: drh tags: trunk)
16:37
Further changes to test scripts so that the "inmemory_journal" permutation works. (check-in: 50f2f7df user: dan tags: trunk)
13:59
Fix a bug in the permutations.test script. (check-in: 3b20ad03 user: dan tags: trunk)
13:50
Fix an assert in pager.c. And various test cases that fail with the in-memory journal permutation. (check-in: 622378db user: dan tags: trunk)
13:45
Fix an assert in pager.c. And various test cases that fail with the in-memory journal permutation. (Closed-Leaf check-in: 78fc35ff user: dan tags: mistake)
12:31
Cherry-pick the correct changes out of the recent "mistake" branch while omitting the bugs. (check-in: 48bf3093 user: drh tags: trunk)
12:26
Remove dead code from the pager. (check-in: 7cbe175a user: drh tags: mistake)
12:00
Print the SQLite source_id() string when running speed tests. (check-in: 6d7640ed user: drh tags: mistake)
10:00
Fix an assert() in pager.c added by the previous commit. And various problems with test scripts in autovacuum and in-memory journal mode. (check-in: 62a10101 user: dan tags: mistake)
08:01
Do not attempt to open a WAL file for an in-memory or temporary database. Even if the database header suggests that it is a WAL database. (check-in: 0fd80924 user: dan tags: mistake)
05:56
Change code in OP_Rewind to avoid a valgrind warning following an error in the btree layer. This change does not fix any bug, just a warning. (check-in: 4f62dbcf user: dan tags: mistake)
01:44
Fix a buffer overrun in the where.c. Problem detected by valgrind. (check-in: 15bb6233 user: drh tags: mistake)
2010-07-02
19:49
Changes to os_win.c to have it return same error code as os_unix.c. (check-in: 33b6f069 user: shaneh tags: mistake)
19:36
Omit the OP_JournalMode opcode from the VDBE when SQLITE_OMIT_PRAGMA is defined. (check-in: 565ff65c user: drh tags: mistake)
19:04
Fix a bug in test_vfs.c causing an assert to fail. Changes to test code only. (check-in: 336ce7d2 user: dan tags: mistake)
18:58
Do not run the tests in notify3.test unless the unlock-notify API is available. (check-in: 0d7fd6fe user: dan tags: mistake)
18:44
Take out the incomplete initializer on the constant "dummy" in sqlite3VdbeGetOp(). Add a comment that the MSVC warning there should be ignored. (check-in: 452ccaa9 user: drh tags: mistake)
18:15
Fix a couple more compiler warnings under MSVC. (check-in: 26bc27e3 user: shaneh tags: mistake)
17:10
Fix compiler warnings in the proxy locking code. (check-in: 26c7689c user: drh tags: mistake)
17:07
Remove a couple of C++ style comments from pager.c. (check-in: 4f12e8eb user: drh tags: mistake)
17:05
Fix some warnings when compiling under MSVC. (check-in: 70833877 user: shaneh tags: mistake)
16:36
Add initializers for the xShmXXX() members to an sqlite3_io_methods structure in journal.c. This doesn't fix any real problem, just prevents a compiler warning. (check-in: dafb3577 user: dan tags: trunk)
15:35
Remove an unused call to sqlite3PageBegin() from pager_write(). Replace it with an assert(). (check-in: 01c12782 user: drh tags: trunk)
13:49
Run tkt-9d68c883.test along with the other pager.c coverage tests. Add a test case to pager1.test. (check-in: c6714b0d user: dan tags: trunk)
11:27
Additional test cases to cover branches in pager.c. (check-in: eddfb2b4 user: dan tags: trunk)
01:18
Add a NEVER to sqlite3PagerCommitPhaseTwo() because it is now no longer possible to invoke that function without holding a RESERVED lock. (check-in: 6ae76172 user: drh tags: trunk)
2010-07-01
19:45
If you delete a database file but not its *-wal file, then open the database (thus creating a new empty database file) the *-wal file is automatically deleted. (check-in: 95c69dfb user: drh tags: trunk)
19:01
Add tests to pager1.test and pagerfault.test. (check-in: c6e75950 user: dan tags: trunk)
15:09
Add pager test cases. Change a condition in pager.c to NEVER(). (check-in: a8f6341d user: dan tags: trunk)
2010-06-30
10:36
Add further test cases. Fix an assert() in pager.c. (check-in: 8e65c0e3 user: dan tags: trunk)
04:36
Do not call pager_open_journal() from within PagerBegin() if the connection is in exclusive-access mode. It will be called from within PagerWrite() just as it is for non-exclusive mode anyway. (check-in: cdf2c5c2 user: dan tags: trunk)
04:29
Add coverage test cases. Have sqlite3_backup_step() transform SQLITE_IOERR_NOMEM to SQLITE_NOMEM before returning. (check-in: 5e19bc36 user: dan tags: trunk)
2010-06-29
10:30
Add tests to pager1.test and pagerfault.test. (check-in: 008513ee user: dan tags: trunk)
2010-06-28
19:04
Add extra pager tests. (check-in: 6b7e419d user: dan tags: trunk)
11:23
Fix some errors when compiling with SQLITE_OMIT_WAL. (check-in: 3b68cb9c user: dan tags: trunk)
11:06
Remove debugging code from test script. (check-in: a85ae332 user: dan tags: trunk)
10:15
Currently, if SQLite cannot find a table or index referred to by a query, it reloads the database schema from disk to see if the table or index has been added since the schema was cached in memory. Extend this behaviour to columns (which may have been added using ALTER TABLE) and fix some obscure cases related to tables and indexes (INDEXED BY, DROP TABLE etc.). (check-in: 4932f228 user: dan tags: trunk)
2010-06-26
22:16
Make walIndexTryHdr() a private function. Fix an issue with SQLITE_MUTEX_NOOP. (check-in: ec65bbd0 user: drh tags: trunk)
21:34
Suppress various compiler warnings. (check-in: e82d008e user: drh tags: trunk)
20:25
Fix two asserts on the scratch allocator to allow for up to two outstanding scratch allocations per thread. (check-in: f149b498 user: drh tags: trunk)
20:00
Suppress a couple uninitialized variable warnings. (check-in: 29571e22 user: drh tags: trunk)
19:03
Add test file tkt-d11f09d36e.test. (check-in: 6e469b63 user: dan tags: trunk)
19:02
Remove a NEVER() in pager that can in fact be reached in obscure failure cases. (check-in: 42b8bd1c user: drh tags: trunk)
17:15
Fix a case where the doNotSyncSpill flag may remain permanently set following an IO error. (check-in: 56c7d111 user: dan tags: trunk)
15:42
When synchronous=NORMAL, use the same journal file format as with synchronous=FULL (i.e. multiple journal headers within the one journal). Fix for [d11f09d36e]. (check-in: 2eaf5ee0 user: dan tags: trunk)
2010-06-25
19:09
Further test cases for pager1.test and pagerfault.test. (check-in: bfd563c4 user: dan tags: trunk)
16:34
Reduce the average (but not maximum) size of the allocations made as part of a checkpoint. (check-in: 4a7fd91b user: dan tags: trunk)
15:16
Reduce the size of the large allocation (approx 8KB for every 4000 frames in the log) that occurs during checkpoint. Use the 'scratch' memory for this allocation instead of the general purpose allocation. (check-in: 29887487 user: dan tags: trunk)
14:17
Fix up a branch in sqlite3ValueFromExpr() so that we can achieve full branch test coverage regardless of whether or not SQLITE_ENABLE_STAT2 is used. (check-in: af471ed7 user: drh tags: trunk)
12:52
Change the name of the shared-memory file on windows from *-wal-index to *-shm, for consistency with unix. (check-in: 5995cb15 user: drh tags: trunk)
11:35
Modify the merge-sort in wal.c so that it does not use recursion. (check-in: daea6c05 user: dan tags: trunk)
2010-06-24
19:16
Add test cases to pager1.test and pagerfault.test. (check-in: 4941e437 user: dan tags: trunk)
18:36
Revert to allowing a cache spill during writes of multiple pages within a single sector as long as the spill does not require a journal sync and a new journal header. (check-in: 7d83fbae user: drh tags: trunk)
17:37
Modify ctime.test to work with SQLITE_THREADSAFE=2. (check-in: c6db3b30 user: dan tags: trunk)
14:52
Disable memory-pressure induced cache spill during savepoint rollback. (check-in: a55eb4c3 user: drh tags: trunk)
13:24
Add a coverage test to pagerfault.test. (check-in: b58db67e user: dan tags: trunk)
10:50
Add coverage tests. Remove a NEVER macro from pager.c, as the condition can now be true in wal mode. (check-in: 7aac9ad6 user: dan tags: trunk)
02:46
Make sure the wal-index reader detects an incorrect version number even if it had to hold a lock in order to read the wal-index. Also, expand and enhance various comments in wal.c. (check-in: 2e6a462c user: drh tags: trunk)
2010-06-23
22:00
Fix and/or improve comments in wal.c. No code changes. (check-in: ee9991be user: drh tags: trunk)
19:27
Add test case for SQLITE_FULL errors to pagerfault.test. Remove a NEVER macro in pager.c that this hits. (check-in: 3e9680c4 user: dan tags: trunk)
17:59
Include the SOURCE_ID value in the log messages output for corruption, misuse, and CANTOPEN errors. (check-in: 1727a81f user: drh tags: trunk)
17:58
Adjust the shared-memory locking range to account for the new version number values in the wal-index header. (check-in: a6dc0df3 user: drh tags: trunk)
15:55
Add a version number to the wal-index header. If SQLite encounters a version number in either the wal or wal-index files that it does not understand, the operation is abandoned and SQLITE_CANTOPEN returned. (check-in: 8d0f8a7f user: dan tags: trunk)
15:18
Fix the xCurrentTimeInt64 interface on windows. It was off by a factor of 10. (check-in: 51ef43b9 user: drh tags: trunk)
15:04
Simplifications to the pager_delmaster() implementation. (check-in: 8bfbdec6 user: drh tags: trunk)
2010-06-22
21:15
Disable code used only by the codec when the codec is not deployed. (check-in: 2c90276e user: drh tags: trunk)
15:18
Add codec support to wal mode. (check-in: 393741eb user: dan tags: trunk)
14:49
When trying to transition from journal_mode MEMORY to WAL, use OFF as an intermediate journal mode. (check-in: 4775b8f9 user: drh tags: trunk)
13:46
Some changes to test scripts related to codec enabled versions of sqlite. (check-in: 85dd51a7 user: dan tags: trunk)
11:10
Fix problem with handling of -vfs option in tcl interface. (check-in: 8f98e8b5 user: dan tags: trunk)
2010-06-21
21:45
Minor tweaks to get the pager working for corner cases when the SQLITE_IOCAP_UNDELETABLE_WHEN_OPEN flag is set. (check-in: 82b38fca user: drh tags: trunk)
20:19
Fix comment inside the pager_unlock() routine. Always close the master journal file before trying to delete it. (check-in: f820b815 user: drh tags: trunk)
18:29
Add test cases to pager1.test and pager2.test. (check-in: cc9ddae6 user: dan tags: trunk)
12:47
Merge the experimental UNDELETABLE_WHEN_OPEN optimization into the trunk. (check-in: ee0acef1 user: drh tags: trunk)
12:34
Change things so that journal2.test works with ENABLE_ATOMIC_WRITE. (Closed-Leaf check-in: a64d96db user: dan tags: experimental)
07:45
Add further pager tests. (check-in: 4104b175 user: dan tags: experimental)
06:00
Merge latest trunk change. (check-in: f6d26e07 user: dan tags: experimental)
05:40
Fix jrnlmode2.test so that it works on systems where UNDELETABLE_WHEN_OPEN is defined. (check-in: 59be370e user: dan tags: experimental)
2010-06-19
19:06
Fix an assert() failure that could occur if compiling with OMIT_SHARED_CACHE. (check-in: 3e76a9f2 user: dan tags: experimental)
18:12
Change the name of IOCAP_SAFE_DELETE to IOCAP_UNDELETABLE_WHEN_OPEN. Have the xDeviceCharacteristics() method of the win32 VFS return this flag. (check-in: 5a5ff4e3 user: dan tags: experimental)
17:26
Add tests to pager1.test and pagerfault.test. (check-in: f5df83fd user: dan tags: experimental)
11:30
Add tests to pager1.test and pagerfault.test. (check-in: 58c0b5bf user: dan tags: experimental)
2010-06-18
18:59
Add tests to pager1.test. (check-in: 582fca89 user: dan tags: experimental)
16:13
Rearrange some code in OP_JournalMode to avoid incorrect returns from subsequent invocations of "PRAGMA journal_mode" following an IO error. (check-in: ce0a4a9b user: dan tags: experimental)
11:10
Change the implementation of the unix implementation of xAccess() so that it returns 0 (does not exist) to an SQLITE_ACCESS_EXISTS query on a file that exists but is zero bytes in size. (check-in: 077b0e5b user: dan tags: experimental)
2010-06-17
16:44
Merge trunk change [7c3a86b9c7]. (check-in: bd7bc4e0 user: dan tags: experimental)
16:44
A different fix to [fc62af4523]. When changing from journal_mode=PERSIST or TRINCATE to some other rollback mode, delete the journal file only if a RESERVED lock can be obtained on the database file first. (check-in: b9b11855 user: dan tags: experimental)
11:36
Fix bug in journal2.test. (check-in: c1e04f1d user: dan tags: experimental)
10:52
Merge trunk updates with experimental branch. Also add tests to journal2.test. (check-in: 9f6ea1de user: dan tags: experimental)
10:42
Do not delete the journal file in "PRAGMA journal_mode" commands. This fixes [fc62af4523]. (check-in: 1ec74591 user: dan tags: trunk)
10:24
Add test case for [fc62af4523]. (check-in: cccd32c6 user: dan tags: trunk)
06:19
Merge fix [f80c3f922a] with experimental changes. (check-in: 20133e9c user: dan tags: experimental)
02:13
Bug fix: Only trust the database size number at offset 28 if the change counter at offset 24 matches the version number counter at offset 92. This prevents corruption in the case of two applications writing to the database where one is an older version of SQLite and the other is a newer version. (check-in: f80c3f92 user: drh tags: trunk)
2010-06-16
19:04
Experimental change: On systems where it is not possible to unlink a file while one or more processes has it open (i.e. not unix), avoid closing the journal file each time the database is unlocked and reopening it at the start of each transaction. (check-in: bede8c8a user: dan tags: experimental)
12:30
Add extra test cases to pager1.test. (check-in: ad320957 user: dan tags: trunk)
10:55
Fix a memory leak that can occur in os_unix.c if an IO error occurs within the xUnlock method. (check-in: 6c5c04ee user: dan tags: trunk)
2010-06-15
19:07
Rationalize a common pattern in tcl test cases into proc do_multiclient_test. (check-in: efe44564 user: dan tags: trunk)
18:00
Fix a problem introduced into lock2.test by the previous commit. (check-in: c1c9f6fa user: dan tags: trunk)
17:44
Add test file pager1.test, containing tests of inter-process locking in non-wal mode. (check-in: 6e43eed9 user: dan tags: trunk)
14:21
Run extra iterations in wal3.test to ensure test coverage. (check-in: ea80cbe5 user: dan tags: trunk)
14:06
Merge in [2c5e48a485]. Merge in [2c5e48a485]. Merge in [2c5e48a485]. (check-in: 03c0f4fd user: dan tags: trunk)
13:56
Changes to the way tcl test scripts work. No changes to production code. (Closed-Leaf check-in: 2c5e48a4 user: dan tags: experimental)
12:09
Minor tweaks to the interface documentation - adding hyperlinks. (check-in: 07b2fdd1 user: drh tags: trunk)
11:21
Updates to comments and testcase() macros in wal.c. (check-in: 4d90cc0b user: drh tags: trunk)
2010-06-14
18:06
Fix an error in the walfault.test script introduced by the previous check-in. (check-in: a2b9374f user: drh tags: trunk)
18:01
Move the xShmMap method to in between xShmLock and xShmBarrier, since it seems to fit in there logically. (check-in: 58dfd83d user: drh tags: trunk)
17:09
Merge the experimental shared-memory mmap-by-chunk changes into the trunk. (check-in: f295e7ed user: drh tags: trunk)
16:16
Add the new xShmMap (formerly xShmPage) to os_win.c. (check-in: 13e7a824 user: dan tags: experimental)
14:07
Remove xShmGet/Size/Release from the sqlite3_vfs structure. Change the name of xShmPage to xShmMap. Remove some code that is now unused from os_unix.c and some of the test VFS implementations. (check-in: fc0cabc1 user: dan tags: experimental)
11:49
Update some comments in wal.c. No code changes. (check-in: 1ce9c92b user: dan tags: experimental)
11:18
Change the interface to internal function walGetHash() to make it easier to follow. (check-in: 5e8e2e97 user: dan tags: experimental)
10:30
Add the xShmPage method to the "crash" vfs in test6.c. (check-in: 1008f536 user: dan tags: experimental)
07:53
Add some fault-injection tests to improve coverage. (check-in: 37b26d12 user: dan tags: experimental)
2010-06-12
12:02
Fix some problems with handling IO errors on the experimental branch. (check-in: eade8bc2 user: dan tags: experimental)
2010-06-11
19:04
Experimental change to the xShmXXX parts of the VFS interface. (check-in: ca68472d user: dan tags: experimental)
17:01
Refactor and simplify the logic used to change journalmode. (check-in: 95cc3f6f user: drh tags: trunk)
2010-06-10
16:25
Adding original art for the new SQLite logo. (check-in: af353bd8 user: drh tags: trunk)
14:07
Add the SQLITE_TESTCTRL_PGHDRSZ verb for sqlite3_test_control() used to get the size of pcache headers for testing purposes. (check-in: 5d694f04 user: drh tags: trunk)
10:51
Update the test VFS in test_journal.c so that it implements the xCurrentTimeInt64 method. (check-in: ec3b4a7e user: drh tags: trunk)
06:53
Store the MemPage structure in memory following, instead of preceding, the page data for cached pages. This reduces the likelihood of a corrupt database page image causing SQLite to read past the end of a buffer. (check-in: 0ce42e76 user: dan tags: trunk)
01:07
Remove a condition that must always be true. Replace it with an assert(). (check-in: 7abcd54e user: drh tags: trunk)
2010-06-09
19:01
Fix test script problems causing errors in the "inmemory_journal" permutations. (check-in: ea05b7f9 user: dan tags: trunk)
16:58
When restarting (wrapping) a log file, set all unused aReadMark[] slots to READMARK_NOT_USED instead of 0. Setting them to 0 does not cause a problem, but may cause SQLite to obtain and release a few more file locks than would otherwise be necessary. (check-in: 0797b10c user: dan tags: trunk)
15:47
Fix for ticket [f973c7ac31]. (check-in: 6eb058dd user: dan tags: trunk)
14:45
Shift the meaning of aReadMark[] back so that +1 offset from mxFrame is removed. Add the new READMARK_NOT_USED value (0xffffffff) instead of zero to signal an aReadMark[] that is not in use. (check-in: c576827d user: drh tags: trunk)
11:28
Simpler fix for the race condition also fixed by [7c102c7b5f] (check-in: 3c2de820 user: dan tags: trunk)
2010-06-08
15:50
Close database opened by tester.tcl when it is sourced in all.test. Because test scripts are now run in slave interpreters, this connection was not being closed by the first script run as it was previously. (check-in: b072e9f6 user: dan tags: trunk)
15:16
Changes to wal3.test to support small default cache size settings. And builds that create autovacuum databases by default. (check-in: ac4cc257 user: dan tags: trunk)
2010-06-07
19:26
In os_unix.c, call munmap() to unmap a shared-memory region prior to closing the associated file descriptor. (check-in: 95cc976f user: dan tags: trunk)
17:47
Change all.test, quick.test and permutations.test so that they use a separate interpreter for each test file. (check-in: 7c33eb5a user: dan tags: trunk)
14:28
Refactor some of the global variables and commands used by tester.tcl. (check-in: c2edf8e1 user: dan tags: trunk)
06:11
Add a couple of extra coverage tests for wal.c. (check-in: cfe60254 user: dan tags: trunk)
2010-06-05
19:18
Add further test cases for the logic in sqlite3WalBeginReadTransaction(). (check-in: a49713db user: dan tags: trunk)
18:34
Clarify the purpose of a test for a race-condition in sqlite3WalBeginReadTransaction(). (check-in: c041c6a9 user: dan tags: trunk)
18:12
Remove a condition from sqlite3WalRead() that is unreachable as of the changes to clear entries out of the wal-index hash tables on transaction or savepoint rollback. (check-in: 39420473 user: dan tags: trunk)
14:42
Mark a condition in wal.c as ALWAYS(). (check-in: 3fe0cc78 user: dan tags: trunk)
11:53
Add extra coverage test cases for wal.c. No changes to production code. (check-in: f9d4ae0e user: dan tags: trunk)
2010-06-04
18:38
Clarify an assert in sqlite3WalExclusiveMode(). (check-in: 25585069 user: dan tags: trunk)
17:16
Remove an unnecessary branch from wal.c. (check-in: 8e54786c user: dan tags: trunk)
15:59
Add an English language error message to corresponding to SQLITE_PROTOCOL. "locking protocol". (check-in: ca327e32 user: dan tags: trunk)
12:22
Add test for the code that detects an inconsistent pair of wal-index headers to wal2.test. (check-in: 157feba1 user: dan tags: trunk)
11:56
If an attempt to sync the database file as part of a checkpoint fails, do not update the shared "nBackfill" variable. Otherwise, another process could wrap the log and overwrite content before it is synced into the database. (check-in: b813233d user: dan tags: trunk)
10:37
Fix a problem where an SQLITE_BUSY in the checkpoint code was being treated as an IO error (abandoning, instead of just limiting, the checkpoint). (check-in: 02c4040c user: dan tags: trunk)
2010-06-03
19:10
Fix another problem in test_vfs.c. (check-in: df7d5989 user: dan tags: trunk)
18:20
Have test_vfs.c simulate IO errors in xShmLock. (check-in: fcbf7cf1 user: dan tags: trunk)
18:02
Performance fix for winShmClose(). (check-in: ed7774de user: drh tags: trunk)
16:58
Add extra tests for removing elements from wal-index hash tables as part of a rollback. (check-in: af3e598a user: dan tags: trunk)
12:35
Delay opening the sub-journal until SQLite actually needs to write data to it. (check-in: c43deb33 user: dan tags: trunk)
12:09
Remove global variables when compiled with SQLITE_OMIT_WSD (check-in: dd10a547 user: drh tags: trunk)
09:25
Enhancements to test_vfs.c and walfault.test. (check-in: ac0de2f3 user: dan tags: trunk)
09:17
If an error (OOM or SQLITE_FULL error) occurs while executing an SQL statement and a statement-transaction is automatically rolled back as a result, if a second error occurs during the statement rollback do a full transaction rollback instead. Otherwise the client can be left with an inconsistent cache. This can affect both WAL and rollback modes. (check-in: eb80ddc6 user: dan tags: trunk)
09:01
If a malloc fails while allocating a savepoint object at the pager level, do not try to roll that savepoint back later on. (check-in: 91cb08ff user: dan tags: trunk)
2010-06-02
20:38
When a lock fails due to an I/O error (not an SQLITE_BUSY) then the checkpoint should fail. (check-in: c6eb0a27 user: drh tags: trunk)
18:59
Fix a problem with rolling back to a savepoint opened before the writer decided to wrap the log file. (check-in: 6b4aed6a user: dan tags: trunk)
17:15
If an error occurs while writing frames to the log to commit a transaction, do not mark the pages as clean in the cache. Otherwise, the subsequent rollback does not roll them back (leaving the client to continue with a cache that makes it appear that the transaction was committed - inconsistent with the database on disk). (check-in: cb571c1b user: dan tags: trunk)
16:40
Added implementation of vfslog_time() for Windows testing. (check-in: b3109f85 user: shaneh tags: trunk)
16:39
Updated wal2.test for windows testing. (check-in: c3606f39 user: shaneh tags: trunk)
15:59
Make sure the directory used by os_unix.c for shared-memory files is always the same, regardless of individual client permissions and environment variable settings. (check-in: f97bd520 user: drh tags: trunk)
15:43
When walTryBeginRead() encounters an I/O error trying to set a lock (as opposed to SQLITE_BUSY) be sure to propagate that error back up the call stack. (check-in: aa2c2b67 user: drh tags: trunk)
14:45
Changes to make WAL more robust against SHM locking failures and OOM errors. (check-in: ebf40413 user: drh tags: trunk)
14:43
In the rowhash.test, make sure global variables are cleared prior to use. (check-in: 28efe0a4 user: drh tags: trunk)
12:58
Undo one of the "const" markers from the previous check-in because the value needs to be changable, even if it is not actually ever changed in the current implementation. (check-in: c24ac356 user: drh tags: trunk)
05:53
Add some 'const' markers to static data that is really constant. (check-in: e7073e23 user: dan tags: trunk)
2010-06-01
21:02
The shared-memory used by WAL on linux now really is shared memory in /dev/shm. On other unix flavors, the file is in a temporary directory rather than in the same directory as the database. (check-in: fc18c4aa user: drh tags: trunk)
19:15
Changes to the way faults are injected into xShmXXX VFS calls. (check-in: 716d99f3 user: dan tags: trunk)
17:46
Change the OOM and IO error test cases in walfault.test so that each test case runs both types of error simulation. (check-in: b627e153 user: dan tags: trunk)
15:44
Delay the decision to restart the log file until data is actually ready to be written to the log file (instead of at the start of a write transaction). (check-in: b1abfaaf user: dan tags: trunk)
15:24
The incremental checkpoint feature is not perfect yet, but it is working well enough to merge it into the trunk. (check-in: 1d3e569e user: drh tags: trunk)
14:30
Merge with [15abbc3416]. (Closed-Leaf check-in: f4b9003a user: dan tags: wal-incr-ckpt)
14:12
If an attempt to get a read-lock on the WAL fails with SQLITE_BUSY_RECOVER, call the busy-handler at the btree level. (check-in: ce644965 user: dan tags: wal-incr-ckpt)
13:17
Fix a bug introduced by the previous check-in but only seen if SQLITE_ENABLE_EXPENSIVE_ASSERTS is turned on. (check-in: 9c9ec899 user: drh tags: wal-incr-ckpt)
12:58
Fix an off-by-one boundary-value issue in walCleanupHash(). (check-in: f039552e user: drh tags: wal-incr-ckpt)
11:08
Do not hold the shared-memory mutex in os_unix.c if returning NULL. (check-in: 9622dd46 user: dan tags: wal-incr-ckpt)
10:44
If the checkpoint fails to obtain an exclusive lock on one of the read-lock bytes, do not consider this an error. (check-in: 9e95e357 user: dan tags: wal-incr-ckpt)
07:51
Fixes to the test cases in wal2.test. (check-in: cd5fbcbc user: dan tags: wal-incr-ckpt)
01:08
Comment edits and cleanup in wal.c. No functional code changes. (check-in: e8e666ab user: drh tags: wal-incr-ckpt)
00:28
Attempt to get the filectrl.test script running. (check-in: e46a8f2b user: drh tags: wal-incr-ckpt)
00:03
Fix os_unix.c so that it will compile and build on a Mac. (check-in: bc707c83 user: drh tags: wal-incr-ckpt)
2010-05-31
20:28
Fix issues with locking_mode=EXCLUSIVE in WAL. (check-in: 8deba0ce user: drh tags: wal-incr-ckpt)
18:24
Add an "isInit" field in the wal-index header that must be non-zero for a valid header. Use this to detect an uninitialized wal-index. (check-in: a16fde19 user: drh tags: wal-incr-ckpt)
16:56
Fix an inconsistent #ifdef in wal.c. Fix os_unix.c so that it does not allow moving an SHM lock directly exclusive to shared without going through unlocked. (check-in: 552658da user: drh tags: wal-incr-ckpt)
16:41
Avoid dropping the checkpoint lock after a recovery run as a precursor to a checkpoint operation. (check-in: cc25cfa0 user: dan tags: wal-incr-ckpt)
16:17
Zero the checkpoint header as the last step of successful WAL recovery. Avoid an unnecessary lock/unlock in WalBeginReadTransaction. (check-in: db3509c5 user: dan tags: wal-incr-ckpt)
16:10
Get the new xShmLock interface design working on os_win.c. (check-in: 149a7082 user: drh tags: wal-incr-ckpt)
14:39
Make sure WAL alway requests enough shared-memory space. (check-in: 138f1283 user: drh tags: wal-incr-ckpt)
14:28
Remove the tests that enable tracing based on the existance of the "vdbe_*" files when in SQLITE_DEBUG mode. (check-in: cdbb4e7c user: drh tags: wal-incr-ckpt)
14:18
Update another test case in wal.test. (check-in: 99fde69e user: dan tags: wal-incr-ckpt)
13:11
Change WAL test cases to account for the improved concurrency in the new checkpoint logic. (check-in: 2d6f2485 user: drh tags: wal-incr-ckpt)
12:30
Do not fail a checkpoint just because active readers prevent backfill. (check-in: 9aa4243e user: drh tags: wal-incr-ckpt)