- 12 1月, 2015 9 次提交
-
-
由 Robert Metzger 提交于
-
由 Gyula Fora 提交于
[FLINK-1380] [streaming] Updated SplitDataStream to extend DataStream to get rid of selectAll method for splits
-
由 Ufuk Celebi 提交于
This closes #254.
-
由 Aljoscha Krettek 提交于
This closes #293
-
由 Aljoscha Krettek 提交于
-
由 Aljoscha Krettek 提交于
Before, something like: def f[T: TypeInformation](data: DataSet[T]) = { val tpe = createTypeInformation[(T, Seq[T])] println("Type: " + tpe) } f(Seq(1.0f, 2.0f) would fail because the type extractor could not re-use existing TypeInformation for nested types.
-
由 Stephan Ewen 提交于
-
由 mbalassi 提交于
-
由 mbalassi 提交于
This closes #279
-
- 10 1月, 2015 1 次提交
-
-
由 Stephan Ewen 提交于
-
- 09 1月, 2015 3 次提交
-
-
由 Ufuk Celebi 提交于
-
由 Stephan Ewen 提交于
-
由 Stephan Ewen 提交于
-
- 08 1月, 2015 16 次提交
-
-
由 Robert Metzger 提交于
This closes #268
-
由 Robert Metzger 提交于
-
由 Robert Metzger 提交于
-
由 Robert Metzger 提交于
to HadoopFileSystem wrapper, which supports all subclasses of org.apache.hadoop.fs.FileSystem. This allows us to let users use all file systems with support for HDFS. The change has been tested with Tachyon, Google Cloud Storage Hadoop Adapter and HDFS. The change also cleans up the Hadoop dependency exclusions.
-
由 twalthr 提交于
Also contains fixes by sewen@apache.org - Make MissingTypeInfo optional in TypeExtractor (by default still throws exception) - Simplified deferred evaluation of type dependend code by making evaluations lazy - Add call location function names to MissingTypeInfo error messages. - Improvements on other error messages. This closes #270
-
由 Gyula Fora 提交于
-
由 Gyula Fora 提交于
-
由 mbalassi 提交于
Organized imports for streaming scala examples Added template parameter for scala streaming iterate Minor fixes in streaming examples
-
由 Stephan Ewen 提交于
-
由 Stephan Ewen 提交于
-
由 Stephan Ewen 提交于
- The map driver now also supports this - In the merge iterator (merges sorted runs from external sort), we now always use the non-reusing code path, because the reusing codepath here implies in all cases additional instances to be held concurrently, and copy between elements, which voids the benefits of reusing elements. - For many utility iterators (in test cases), consolidates the logic between the two variants of the "next()" functions (one calls the other, where possible) - Eliminates a few copies between elements in the non-reusing parts (where possible) - Removes unused variables in the non-reusing variants (mainly serializers previously used to create instance or copy between instances) - Remove some unused types - Improves generic type safety (fewer raw types)
-
由 Aljoscha Krettek 提交于
This closes #259
-
由 Aljoscha Krettek 提交于
-
由 Aljoscha Krettek 提交于
This is in preparation for configurable object-reuse mode. We previously referred to this as mutable object vs. mutable object safe mode or some such thing.
-
由 Kostas Tzoumas 提交于
This closes #285
-
由 Max 提交于
This closes #286
-
- 07 1月, 2015 3 次提交
-
-
由 Aljoscha Krettek 提交于
-
由 Aljoscha Krettek 提交于
-
由 Till Rohrmann 提交于
[FLINK-1363] [runtime tests] Fix race condition in ExecutionVertexCancelTest.testSendCancelAndReceiveFail This closes #288
-
- 06 1月, 2015 8 次提交
-
-
由 Stephan Ewen 提交于
-
由 Stephan Ewen 提交于
[FLINK-1259] [apis] Added comments to interface and docs that filters must not mutate date elements.
-
由 Stephan Ewen 提交于
-
由 carbone 提交于
-
由 Gyula Fora 提交于
-
由 Gyula Fora 提交于
-
由 Gyula Fora 提交于
-
由 Gyula Fora 提交于
-