Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | Cut release 1.3.1.1.3.1 | Michael Cahill | 2012-09-25 | 5 | -6/+53 |
| | |||||
* | Merge pull request #336 from wiredtiger/lsm-checkpoint-open | agorrod | 2012-09-24 | 5 | -14/+43 |
|\ | | | | | Bug fixes for LSM open and handle locking | ||||
| * | Clear the btree handle at the beginning of a drop operation. | Michael Cahill | 2012-09-25 | 1 | -0/+3 |
| | | |||||
| * | Another fix to make sure handles are unlocked exactly once if drop fails. | Michael Cahill | 2012-09-24 | 1 | -3/+5 |
| | | |||||
| * | Fix the change to track exclusive handles for drop operations: always log the | Michael Cahill | 2012-09-24 | 2 | -25/+16 |
| | | | | | | | | | | primary first when closing, regardless of what order they appear in the handle list. | ||||
| * | If a drop fails to acquire all of the handle locks it needs, make sure it | Michael Cahill | 2012-09-24 | 1 | -0/+18 |
| | | | | | | | | releases the primary handle lock. | ||||
| * | Have LSM read from checkpoints. In corner cases, we may have an empty chunk ↵ | Michael Cahill | 2012-09-24 | 1 | -2/+13 |
| | | | | | | | | that was never checkpointed. If so, fall back to the ordinary "current" version of that (empty) chunk. | ||||
| * | Remove the hard-coded LSM checkpoint config stack: use the defaults. | Michael Cahill | 2012-09-24 | 1 | -1/+1 |
| | | |||||
| * | Two fixes for LSM, one to ensure that cursors read from a checkpoint | Alex Gorrod | 2012-09-24 | 2 | -2/+6 |
| | | | | | | | | | | if one is available. The other to reduce the number of empty chunks that can be created initially. | ||||
* | | Overflow objects created during reconciliation won't have any flags set, | Keith Bostic | 2012-09-24 | 1 | -0/+2 |
|/ | | | | don't return random bytes in the buffer describing them. | ||||
* | Don't nest auto-commit transactions. | Michael Cahill | 2012-09-24 | 1 | -1/+1 |
| | | | | refs #309 | ||||
* | Merge branch 'master' into develop | Michael Cahill | 2012-09-24 | 0 | -0/+0 |
|\ | |||||
| * | Merge branch 'develop' | Michael Cahill | 2012-09-17 | 314 | -7283/+19446 |
| |\ | |||||
| * \ | Merge branch 'release-1.2' | Michael Cahill | 2012-06-20 | 22 | -55/+207 |
| |\ \ | |||||
| * \ \ | Merge branch 'develop' | Michael Cahill | 2012-06-15 | 19 | -84/+191 |
| |\ \ \ | |||||
| * \ \ \ | Merge branch 'develop' | Michael Cahill | 2012-06-04 | 275 | -6413/+12351 |
| |\ \ \ \ | |||||
| * \ \ \ \ | Merge branch 'develop' | Michael Cahill | 2012-04-26 | 19 | -90/+177 |
| |\ \ \ \ \ | |||||
| * \ \ \ \ \ | Merge branch 'develop' | Michael Cahill | 2012-04-16 | 6 | -6/+37 |
| |\ \ \ \ \ \ | |||||
| * | | | | | | | Merge branches 'develop' and 'master' of github.com:wiredtiger/wiredtiger | Michael Cahill | 2012-04-16 | 0 | -0/+0 |
| | | | | | | | | |||||
* | | | | | | | | Be lazy about allocating auto-commit transactions. | Michael Cahill | 2012-09-24 | 5 | -6/+35 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | In particular, begin the transaction after successfully reading the first page for the operation. This avoids tying down pages unnecessarily when threads wait to read pages in very low memory configurations. refs #309 | ||||
* | | | | | | | | Be careful not to destroy the transaction modify context when evicting. | Michael Cahill | 2012-09-24 | 1 | -0/+7 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | refs #333 | ||||
* | | | | | | | | Fix a leak of the LSM data source handle. | Michael Cahill | 2012-09-24 | 2 | -5/+14 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | refs #333 | ||||
* | | | | | | | | Fix a leak of the in-memory transaction log structure. | Michael Cahill | 2012-09-24 | 1 | -0/+2 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | closes #333 | ||||
* | | | | | | | | Use Keith's new list of valgrind suppressions applied to any Python binary. | Michael Cahill | 2012-09-24 | 1 | -376/+32 |
| | | | | | | | | |||||
* | | | | | | | | Lint for dictionary skip list implementation. | Alex | 2012-09-23 | 1 | -2/+0 |
| | | | | | | | | |||||
* | | | | | | | | Revert last push. It had the wrong set of changes. | Alex | 2012-09-23 | 2 | -179/+0 |
| | | | | | | | | |||||
* | | | | | | | | Lint for dictionary skip lists. | Alex | 2012-09-23 | 2 | -0/+179 |
| | | | | | | | | |||||
* | | | | | | | | Merge pull request #335 from wiredtiger/skips | agorrod | 2012-09-23 | 7 | -162/+317 |
|\ \ \ \ \ \ \ \ | | | | | | | | | | | | | | | | | | | Update dictionary implementation to use skip lists. | ||||
| * | | | | | | | | Add dictionary support to variable-length column store objects, | Keith Bostic | 2012-09-22 | 5 | -89/+138 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | closes #329. | ||||
| * | | | | | | | | Merge branch 'develop' into skips | Keith Bostic | 2012-09-22 | 3 | -9/+11 |
| |\ \ \ \ \ \ \ \ | |||||
| * | | | | | | | | | Change the WT_CELL_VALUE_COPY cell format to not, by default, include | Keith Bostic | 2012-09-22 | 2 | -8/+20 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | the WT_CELL_64V flag, we'll need that for the RLE in order to support dictionaries on column-store pages. Reference #329. | ||||
| * | | | | | | | | | Merge branch 'develop' into skips | Keith Bostic | 2012-09-21 | 5 | -25/+44 |
| |\ \ \ \ \ \ \ \ \ | |||||
| * | | | | | | | | | | Pick a random number for the size of the dictionary. | Keith Bostic | 2012-09-21 | 1 | -1/+1 |
| | | | | | | | | | | | |||||
| * | | | | | | | | | | Merge branch 'develop' into skips | Keith Bostic | 2012-09-21 | 42 | -539/+814 |
| |\ \ \ \ \ \ \ \ \ \ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Conflicts: src/btree/rec_write.c | ||||
| * | | | | | | | | | | | Support large row-store reconciliation dictionaries: add a skiplist | Keith Bostic | 2012-09-21 | 4 | -83/+175 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | as the indexing mechanism. | ||||
* | | | | | | | | | | | | Don't re-allocate if the memory needed is exactly the right amount, | Keith Bostic | 2012-09-23 | 1 | -1/+1 |
| |_|_|/ / / / / / / / |/| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | the underlying realloc calls checks and if you're asking for what you already have, it asserts. | ||||
* | | | | | | | | | | | minor cleanup of a comment. | Keith Bostic | 2012-09-22 | 1 | -3/+3 |
| | | | | | | | | | | | |||||
* | | | | | | | | | | | The change in 1f4c1aeb58d73d18bb24bdbfdd4a56f7f1f8bac1 is relevant to | Keith Bostic | 2012-09-21 | 2 | -6/+8 |
| |_|/ / / / / / / / |/| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | row-store truncate, not just column-store as well -- back out that change, instead, explicitly set cursor.compare to 0 in the calling truncation code. | ||||
* | | | | | | | | | | Fix a memory leak where a page's replacement address wasn't being free'd | Keith Bostic | 2012-09-21 | 2 | -11/+23 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | in the case where the address didn't get pushed into the parent's WT_REF structure, requires clearing the address when the address does get moved to the parent. Fix and close #334. | ||||
* | | | | | | | | | | Fixes for issues uncovered by autocommit. | Michael Cahill | 2012-09-21 | 3 | -14/+21 |
| |/ / / / / / / / |/| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | With lots of concurrent threads generating non-trivial snapshots, a bug meant that we could overflow the allocated snapshot array if we raced getting a snapshot. Update the read context when committing an auto-commit transaction, so the changes are visible to subsequent reads. refs #309 | ||||
* | | | | | | | | | Merge pull request #331 from wiredtiger/autocommit | Michael Cahill | 2012-09-20 | 18 | -175/+203 |
|\ \ \ \ \ \ \ \ \ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Move auto-commit to the API boundary, cleanup some transaction internals. closes #309 | ||||
| * | | | | | | | | | Deal with errors during auto-commit transactions, update the snapshot of | Michael Cahill | 2012-09-21 | 2 | -2/+13 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | read-committed sessions to see the results of auto-commit operations. refs #309 | ||||
| * | | | | | | | | | Add autocommit to WT_SESSION::truncate. | Michael Cahill | 2012-09-21 | 4 | -35/+56 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | refs #310 | ||||
| * | | | | | | | | | Ignore transactions if the database is configured without transaction support. | Michael Cahill | 2012-09-21 | 1 | -0/+9 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | refs #309 | ||||
| * | | | | | | | | | Only rollback changes that were successfully installed. | Michael Cahill | 2012-09-21 | 2 | -21/+24 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | refs #309 | ||||
| * | | | | | | | | | Tweak page sizes in test/format for small caches. | Michael Cahill | 2012-09-19 | 1 | -2/+5 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Make sure at least 10 leaf pages fit in cache: anything smaller is really pathological. | ||||
| * | | | | | | | | | Implement auto-commit of transactions at the API. | Michael Cahill | 2012-09-19 | 13 | -149/+130 |
| |/ / / / / / / / | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This simplifies code that tried to get the effect of auto-commit internally. closes #309, refs #310 | ||||
* | | | | | | | | | Merge pull request #332 from wiredtiger/bulk-checkpoint | Michael Cahill | 2012-09-20 | 27 | -370/+615 |
|\ \ \ \ \ \ \ \ \ | |/ / / / / / / / |/| | | | | | | | | | | | | | | | | | Change bulk load so it doesn't block checkpoints. closes #321 | ||||
| * | | | | | | | | Clarify that only newly created objects may be bulk-loaded, empty | Keith Bostic | 2012-09-20 | 2 | -7/+8 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | isn't enough. | ||||
| * | | | | | | | | Merge branch 'develop' into bulk-checkpoint | Keith Bostic | 2012-09-20 | 3 | -20/+33 |
| |\ \ \ \ \ \ \ \ | |/ / / / / / / / |/| | | | | | | | |