- 09 1月, 2018 10 次提交
-
-
由 Till Rohrmann 提交于
The flip6 build profile only runs the Flip-6 related test cases. Moreover, all Flip-6 related test cases are excluded when not running the flip6 build profile. This should reduce testing time when adding more and more Flip-6 test cases. Include flink-test-utils-junit in all submodules to make the Category marker interfaces Flip6 and OldAndFlip6 available This closes #4889.
-
由 zentol 提交于
This closes #5263.
-
由 zentol 提交于
-
由 yew1eb 提交于
This closes #5213.
-
由 okumin 提交于
* Fix a scala example which is using a wrong variable * Remove partitioning descriptions * partitioning parameters are already removed from IterativeStream#closeWith/DateStream#iterate * https://github.com/apache/flink/pull/988 * https://github.com/apache/flink/pull/4655 This closes #5249.
-
由 wenlong.lwl 提交于
This closes #5122.
-
由 zentol 提交于
This closes #5162.
-
由 Nico Kruber 提交于
This closes #5175.
-
由 Raycee 提交于
This closes #5180.
-
由 zentol 提交于
This closes #5258.
-
- 08 1月, 2018 15 次提交
-
-
由 Nico Kruber 提交于
This closes #5255.
-
由 Piotr Nowojski 提交于
-
由 Piotr Nowojski 提交于
-
由 Piotr Nowojski 提交于
-
由 Piotr Nowojski 提交于
While Buffer class is used in multithreaded context it requires synchronisation. Previously it was miss-leading and unclear, suggesting that RecordSerializer should take into account synchronisation of the Buffer that's holding. With NotThreadSafe BufferBuilder there is now clear separation between single-threaded writing/creating a BufferBuilder and multithreaded Buffer handling/retaining/recycling. This increases throughput of network stack by factor of 2, because previously method getMemorySegment() was called twice per record and it is a synchronized method on recycleLock, while RecordSerializer is sole owner of the Buffer at this point, so synchronisation is not needed.
-
由 Piotr Nowojski 提交于
-
由 Piotr Nowojski 提交于
-
由 Piotr Nowojski 提交于
-
由 Piotr Nowojski 提交于
-
由 Piotr Nowojski 提交于
-
由 Piotr Nowojski 提交于
-
由 Piotr Nowojski 提交于
-
由 Piotr Nowojski 提交于
-
由 Zhijiang 提交于
-
由 Zhijiang 提交于
-
- 06 1月, 2018 15 次提交
-
-
由 Tzu-Li (Gordon) Tai 提交于
Previously, the testScaleUp() test was taking too much resources and causing test resources to be terminated before the test could finish. This commit lowers the intensity of the test, while still retaining the verified behaviour (i.e., when restoring the Kafka consumer with higher parallelism and more Kafka partitions). This closes #5201.
-
由 Piotr Nowojski 提交于
This closes #5193.
-
由 Piotr Nowojski 提交于
[FLINK-8268][streaming][tests] Improve TwoPhaseCommitSinkFunctionTest stability by using custom in memory storage
-
由 Tzu-Li (Gordon) Tai 提交于
-
由 Tzu-Li (Gordon) Tai 提交于
The previous link was referencing a non-existent constructor signature.
-
由 Tzu-Li (Gordon) Tai 提交于
FlinkKafkaProducer010Configuration is the return type of the deprecated writeToKafkaWithTimestamp factory methods. Therefore, the class should also be deprecated as well.
-
由 Tzu-Li (Gordon) Tai 提交于
This commit moves deprecated factory methods of the FlinkKafkaProducer010 behind regular constructors, for better navigation and readability of the code. This closes #5179.
-
由 Tzu-Li (Gordon) Tai 提交于
-
由 Tzu-Li (Gordon) Tai 提交于
-
由 Tzu-Li (Gordon) Tai 提交于
This closes #5121.
-
由 Ankit Parashar 提交于
-
由 Eron Wright 提交于
-
由 Stephan Ewen 提交于
This makes it easier for users and contributors to figure out how to create local file paths in way that works cross operating systems.
-
由 Stephan Ewen 提交于
-
由 Stephan Ewen 提交于
The input stream from ByteStreamStateHandle did not overwrite the 'read(byte[], int, int)' method, meaning that bulk byte reads resulted in many individual byte accesses. Additionally, this change avoids accessing the data array through an outer class, but instead adds a reference directly to the input stream class, avoiding one hop per access. That also allows a more restricted access level on the fields, which may additionally help the jitter in some cases.
-