summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
* Cut release 1.6.41.6.4Alex Gorrod2013-08-214-5/+47
|
* Remove "Memrata" doc link from the top-level navigation.Michael Cahill2013-08-211-1/+0
|
* Cleanup: __wt_update_check() has lost most of its functionality overKeith Bostic2013-08-203-23/+3
| | | | | | time, replace it with calls to __wt_txn_update_check, and remove the calls entirely when there's no WT_UPDATE structure and so no possibility of detecting a collision.
* update to the current Memrata design.Keith Bostic2013-08-201-48/+35
|
* whitespaceMichael Cahill2013-08-201-1/+1
|
* Avoid a race installing an insert list.Michael Cahill2013-08-201-0/+2
|
* lintMichael Cahill2013-08-201-4/+2
|
* Fix logic in WT_CURSOR_NEEDVALUE: only copy values if they are not already ↵Michael Cahill2013-08-201-1/+1
| | | | | | in the WT_ITEM's memory. Found by valgrind reporting "Source and destination overlap in memcpy".
* Fix compiler warning:Alex Gorrod2013-08-201-15/+0
| | | | | ../../../../ext/test/memrata/memrata.c:2612: error: unused variable ‘kvs_options’ ../../../../ext/test/memrata/memrata.c:2612: error: unused variable ‘p’
* [mq]: inslist-raceMichael Cahill2013-08-201-2/+17
|
* Fix build warning in wtperfAlex Gorrod2013-08-201-1/+2
|
* [mq]: lsm-drop-lock-while-closingMichael Cahill2013-08-201-2/+16
|
* Add option to wtperf to populate in batches.Alex Gorrod2013-08-192-3/+23
|
* updateKeith Bostic2013-08-161-0/+1
|
* Lint, whitespace.Keith Bostic2013-08-162-1/+6
|
* Fix build failure.Alex Gorrod2013-08-161-2/+0
|
* Merge pull request #618 from wiredtiger/wtperf-enhancementsAlex Gorrod2013-08-155-285/+730
|\ | | | | Wtperf configuration enhancements.
| * Leave default home directory as WT_TEST so make check works.Alex Gorrod2013-08-161-1/+0
| |
| * Compiler warnings..Alex Gorrod2013-08-161-2/+2
| |
| * Fix compiler warnings.Alex Gorrod2013-08-151-9/+7
| | | | | | | | | | | | | | | | | | | | | | | | ../../../bench/wtperf/wtperf.c: In function ‘connection_reconfigure’: ../../../bench/wtperf/wtperf.c:797:18: error: conversion to ‘size_t’ from ‘long int’ may change the sign of the result [-Werror=sign-conversion] ../../../bench/wtperf/wtperf.c: In function ‘config_opt_file’: ../../../bench/wtperf/wtperf.c:1213:19: error: conversion to ‘size_t’ from ‘long int’ may change the sign of the result [-Werror=sign-conversion] ../../../bench/wtperf/wtperf.c: In function ‘config_opt_usage’: ../../../bench/wtperf/wtperf.c:1348:15: error: conversion to ‘size_t’ from ‘int’ may change the sign of the result [-Werror=sign-conversion] ../../../bench/wtperf/wtperf.c: In function ‘indent_lines’: ../../../bench/wtperf/wtperf.c:1517:16: error: conversion to ‘size_t’ from ‘long int’ may change the sign of the result [-Werror=sign-conversion] cc1: all warnings being treated as errors
| * Style.Susan LoVerso2013-08-141-77/+56
| |
| * formatting. refs #593Don Anderson2013-08-131-3/+2
| |
| * formatting. refs #593Don Anderson2013-08-131-21/+9
| |
| * removed -v option (duplicates -o verbose=N). Fixed typo in message. refs #593Don Anderson2013-08-132-7/+3
| |
| * formatting fixes, typo in comment, removed redundant 'break' stmt. refs #593Don Anderson2013-08-131-4/+2
| |
| * alphabetize forward definitions. refs #593Don Anderson2013-08-131-6/+6
| |
| * added comments to describe DEF_OPT_* usage, reformatted entries. refs #593Don Anderson2013-08-131-26/+33
| |
| * Merge branch 'develop' into wtperf-enhancementsDon Anderson2013-08-130-0/+0
| |\
| * \ Merge branch 'develop' into wtperf-enhancementsMichael Cahill2013-08-1349-1398/+3683
| |\ \ | | | | | | | | | | | | | | | | | | | | | | | | | | | | Conflicts: examples/c/ex_all.c src/cursor/cur_ds.c src/cursor/cur_file.c src/include/wiredtiger.in
| * | | fixed signed/unsigned and 32/64 bit compilation warnings. refs #593Don Anderson2013-08-091-8/+8
| | | |
| * | | Rework the documentation & examples to better reflect "overwrite" isKeith Bostic2013-08-093-32/+80
| | | | | | | | | | | | | | | | | | | | configured on/true by default, and should only be set if you want non-standard behavior.
| * | | There are four possible states for the WT_PAGE.dsk memory: the WT_PAGEKeith Bostic2013-08-098-23/+21
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | disk-image is mapped into memory (unmap on page discard), is in allocated memory (free on page discard), is in memory belonging to a caller (ignore on page discard), or isn't set at all (ignore on page discard). We've been trying to figure out which based on the state of WT_PAGE.dsk and a single flag, and that's not enough. Use two explicit flags, one for mapped memory and one for allocated memory, and neither flag is set for the latter two states.
| * | | When we changed the meaning of the overwrite configuration, we missedKeith Bostic2013-08-095-20/+49
| | | | | | | | | | | | | | | | | | | | the wt utility's load command. Change the flag's name, add tests for the flag to the test suite, make a note in the upgrading docs.
| * | | whitespaceMichael Cahill2013-08-091-1/+1
| | | |
| * | | Rename cursor flags KEY/VALUE_INT/EXT to clarify that they track whether the ↵Michael Cahill2013-08-0910-52/+52
| | | | | | | | | | | | | | | | key/value points into the tree or out of the tree.
| * | | Make a copy of the key after a successful remove.Michael Cahill2013-08-091-5/+8
| | | | | | | | | | | | | | | | refs #607
| * | | Don't clear the key after a cursor remove.Michael Cahill2013-08-092-15/+17
| | | | | | | | | | | | | | | | refs #607
| * | | Revert to switching to the APP flags when making a copy of the key/value. ↵Michael Cahill2013-08-092-16/+27
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This is required to support the following sequence: - set_key - KEY_APP flag set - search - success, RET flags set - remove - key copied, RET flag stays set - deadlock, RET flags cleared - retry remove - fails: neither KEY_APP or KEY_RET is set refs #607
| * | | spelling typoKeith Bostic2013-08-091-1/+1
| | | |
| * | | cur_file.c:268: warning: '__value_data' may be used uninitialized in this ↵Michael Cahill2013-08-091-12/+5
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | function cur_file.c:268: warning: '__key_data' may be used uninitialized in this function cur_file.c:268: warning: '__value_size' may be used uninitialized in this function cur_file.c:268: warning: '__key_size' may be used uninitialized in this function cur_file.c:268: warning: '__recno' may be used uninitialized in this function
| * | | Clear "key set" flags in LSM if searches fail.Michael Cahill2013-08-091-7/+10
| | | |
| * | | Restore the copy of the key/value as part of entering the API: the reason isKeith Bostic2013-08-093-5/+20
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | that when a cursor's key or value are pointing at internal memory (which is what the RET flag indicates), that memory is only safe while the cursor keeps a page referenced. As soon as the cursor releases the page (which happens at the beginning of most operations), it could be evicted and the memory could become invalid. Make the test for whether or not a WT_ITEM.data field references inside the WT_ITEM's allocated memory a bit more complicated: overflow items will have data fields that point inside the allocated memory, not at the start of it.
| * | | Wrap the btree cursor operations with code that saves/restores the cursorKeith Bostic2013-08-098-89/+110
| | | | | | | | | | | | | | | | | | | | key/data information, move the handling of WT_CURSTD_{KEY,VALUE}_{APP,RET} flags out of the btree code and into the cursor code. Reference #607.
| * | | The "loadtext" command can reasonably create the database.Keith Bostic2013-08-091-10/+13
| | | | | | | | | | | | | | | | | | | | Add the "loadtext" command to the help/usage output. Sort the command parsing switch statement alphabetically.
| * | | We weren't testing that a valid key format with an invalid value formatKeith Bostic2013-08-092-1/+6
| | | | | | | | | | | | | | | | failed. Add code to test and enhance test_schema02 to check.
| * | | Make wtperf compile without an installed WiredTiger tree.Alex Gorrod2013-08-091-0/+2
| | | |
| * | | Remove flagval for random, random_range implies this.Don Anderson2013-08-021-12/+25
| | | | | | | | | | | | | | | | | | | | Expand multiline usage messages to format nicely. refs #593
| * | | Reorder options in alphabetical order.Don Anderson2013-08-021-10/+6
| | | | | | | | | | | | | | | | | | | | Remove flagval for random, random_range implies this. refs #593
| * | | Added a help description to the definition of each option.Don Anderson2013-08-012-140/+116
| | | | | | | | | | | | | | | | | | | | | | | | Added config_opt_usage() that prints all options/default values/descriptions. Removed command line options that are available via -o/-O. refs #593
| * | | Use -o {options} when possible, as many wtperf opts are going away. refs #593Don Anderson2013-08-011-9/+9
| | | |
View Lorry Controller Status