chip@chipfraser.co.uk
bbf413d48d
Make code compatible with jna 3.5.2 as well as 4.4.0
2017-11-27 14:01:12 +00:00
Rob Austin
93a642054e
updated the bom to the latest map
2017-11-17 15:01:31 +00:00
epickrram
d297d1e4ef
update bom version, fix usage of Memory methods
2017-11-08 14:05:31 +00:00
Peter Lawrey
4b8b232f0d
chronicle-bom 1.14.28-SNAPSHOT for all modules
2017-07-01 10:27:43 +01:00
Marc Strapetz
ffab7840dd
BigSegmentHeader: system property to specify LOCK_TIMEOUT_SECONDS
...
The default of 60 seconds may be too long if using Chronicle-Map in a
way that prevents real interprocess dead locks, i.e.
InterProcessDeadLockException can only be raised if a previous
Chronicle-Map process has been killed while holding a lock. In this
case, a timeout of e.g 5 seconds (or less) is sufficient to detect the
problem and possibly decide to invoke a recovery.
2017-06-06 19:42:43 +02:00
Marc Strapetz
957288bd5f
BigSegmentHeader: comment fixed
2017-06-06 19:38:33 +02:00
leventov
3e903775f9
Revert method and field reordering
2017-06-04 12:33:41 -05:00
Peter Lawrey
ed8361b5cf
Add a fix for enums being written as marshallables.
2017-06-04 14:01:48 +01:00
leventov
3291a43c95
Use ThreadHints.onSpinWait() in BigSegmentHeader
2017-03-06 00:10:01 -06:00
leventov
7f3cca96d2
Add ChronicleHashCorruption listening interface, and Beta annotation
2017-02-26 21:02:34 -06:00
leventov
c90e065e71
Ensure compatibility with 3.12 and Chronicle Maps created with Integer key, add ChronicleMap3_12IntegerKeyCompatibilityTest
2017-02-19 17:20:09 -06:00
leventov
d31d974e2e
Update to chronicle-algorithms 1.1.8 with a bug fix and add NegativeIntegerKeyTest
2017-02-18 15:02:52 -06:00
leventov
d1fd590be1
In methods which return contexts, wrap all code after context is obtained in try-catch and try to close the context if any exception occurred, to prevent context leaks
2016-12-27 15:38:52 +07:00
leventov
842064dca5
Clean serializers and ThreadLocal fields of VanillaChronicleMap on close()
2016-12-27 15:50:24 +08:00
leventov
fa70e216bd
Improvements to ChronicleMap resource closing subsystem, making it more robust in exotic scenarios (OutOfMemoryError, StackOverflowError) and be able to close after incorrect attempt to close when a context is open in the same thread
2016-12-26 14:27:32 +08:00
leventov
dad7313bf4
Remove unused VanillaChronicleHash.doClose()
2016-12-11 17:02:24 -06:00
leventov
fafc50f89d
Close wrapped value after each iteration in MapSegmentIteration
2016-12-10 21:03:20 -06:00
leventov
e0b534af23
Move all ChronicleMap closing logic into independent ChronicleHashResources. Reference contexts from ThreadLocals via ContextHolders and clear references to contexts on closing ChronicleMap. Make some closing procedures more paranoid than they were. Cache ChronicleMap's identityString and always return the same String from toIdentityString(). Add Throwables.returnOrSuppress().
2016-12-10 21:03:20 -06:00
leventov
9e88392db6
Rename ChronicleHashResourceReleaser to ChronicleHashResources
2016-12-10 21:03:20 -06:00
leventov
37d82f290a
Review Exception throwing and handling throughtput the library, create specific InterProcessDeadLockException for Chronicle Map deadlocks
2016-11-19 19:14:53 -06:00
leventov
5ae69c3035
Fix #103 (a typo)
2016-11-12 15:49:08 -06:00
leventov
d76f838adb
Add ability to configure name for all ChronicleMaps and ChronicleSets via ChronicleHashBuilder.name(). Add ChronicleHash.toIdentityString()
2016-11-12 15:43:27 -06:00
leventov
e7b2836e2a
Impove comments in VanillaChronicleHash
2016-11-05 17:23:48 -06:00
leventov
406bd59848
Remove VanillaChronicleHash from OnExitHook on close()
2016-11-05 17:19:07 -06:00
leventov
9723fcd689
Do VanillaChronicleHash.msync() only when ChronicleHash is persisted
2016-11-05 15:08:54 -06:00
leventov
9a2bb31b66
Add ChronicleHash.offHeapMemoryUsed() method; Centralized management of all direct memory and file resources of ChronicleMap in one Cleaner
2016-11-05 15:08:45 -06:00
leventov
578c6638ba
Add Throwables utilities, inspired by Guava
2016-10-30 13:47:34 -06:00
leventov
a2c8ac4866
Try to close alive ChronicleMaps on JVM shutdown
2016-10-02 18:42:39 +03:00
leventov
1825fc61ee
More correctly formulated reasoning of loadFence() in ThreadLocalState.unlockContextLocally()
2016-10-01 22:06:14 +03:00
leventov
223b2d4329
Prevent leaks of ChronicleMaps though thread local contexts and supplimental OldDeletedEntriesCleanupThread, add MemoryLeakTest
2016-10-01 20:33:16 +03:00
leventov
126f091113
Add missing private constructor to CanonicalRandomAccessFiles to prohibit instantiation of this class
2016-10-01 20:33:16 +03:00
leventov
b4258fa263
Don't use synrhonized on a public object (the map itself) in ReplicatedChronicleMap.close()
2016-10-01 20:33:16 +03:00
leventov
ec919b94eb
More clarifying comments in ThreadLocalState.closeContext()
2016-10-01 20:33:16 +03:00
leventov
95ca197444
Move ThreadLocalState.iterationContextLockedInThisThread after static members in this class
2016-10-01 20:33:16 +03:00
leventov
9a5d6fa50e
Address the issue with ChronicleHash accessors throwing AssertionError instead of IllegalStateException after ChronicleHash is closed
2016-08-05 23:31:09 +07:00
leventov
ba8227ade1
Use more specific ChronicleHashClosedException instead of IllegalStateException, when methods of ChronicleHash are called after hash.close(). Actualize ChronicleHash.close() javadoc.
2016-08-05 23:22:51 +07:00
leventov
f9edc314bb
Optimize VanillaChronicleMap.longSize(): don't acquire and close context for each segment, and make it atomic against concurrent map.close()
2016-08-05 23:20:50 +07:00
leventov
4bcdcdb8ce
Update license information
2016-07-22 15:05:04 +07:00
leventov
33daa3bb9b
Minor optimization in HashLookupSearch.nextPos()
2016-07-17 23:36:00 +07:00
leventov
7ada09a8dc
Make VanillaChronicleHash.close() to wait until concurrent threads accessing the hash finish their work and don't permit new threads to come while close() is in progress and after that
2016-07-17 23:25:49 +07:00
leventov
b3c1f7b1e2
Expose Thread owner() in ChainingInterface
2016-07-17 23:25:46 +07:00
Peter Lawrey
a230c08e89
Cut blank lines
2016-07-07 20:33:40 +01:00
leventov
4cffff652f
Update to newer Core, Bytes & Wire dependency versions; make all enum marshallers ordinary classes with static INSTANCE fields, because of difficulties with Wire serialization of enums
2016-06-24 17:49:34 +07:00
Peter Lawrey
ceab61fd85
Rename Not Ready to Not Complete. Support reading Not COmplete messages unless being read from shared memory.
2016-04-12 11:39:23 +01:00
leventov
e677f07d46
Use Data.equivalent() method in KeySearch.keyEquals() (because it could be specialized to perform more efficiently), and switch arguments in Data.bytesEquivalent() usages, so that the first Data instance is more likely to optimized for comparison
2016-03-04 21:20:12 +07:00
leventov
f2ccfca801
Specify query operations
2016-02-28 21:33:31 +07:00
leventov
dc4b578bef
Lock the segment on the write level right away in HashQuery.doRemove()
2016-02-28 21:32:11 +07:00
leventov
e37de00202
Reduce write lock scope as much as possible in relocation and remove operations
2016-02-26 20:55:47 +07:00
leventov
ed4589a1f4
Removed unused code from VanillaChronicleHash.basicRecover() method
2016-02-24 16:44:36 +07:00
leventov
820573a684
Add method CompactOffHeapLinearHashTable.readEntryVolatile() method and use it during key search
2016-02-20 16:38:59 +07:00