- 14 2月, 2020 2 次提交
-
-
由 Chesnay Schepler 提交于
-
由 Chesnay Schepler 提交于
-
- 13 2月, 2020 21 次提交
-
-
由 Chesnay Schepler 提交于
Some runners may attempt to run this class as a test.
-
由 Chesnay Schepler 提交于
-
由 wooplevip 提交于
format This closes #10697
-
由 Chesnay Schepler 提交于
-
由 Timo Walther 提交于
Further develops the stack based on BridgingSqlFunction and the new type inference. Similar to FLINK-15487, we don't test every data type combination yet. This closes #11034.
-
由 Timo Walther 提交于
Refactors the code generation around table functions such that return data type and argument data type are read from the function at the same location instead of reading each in a separate code generator. This adds an additional runtime collector that is only responsible for result conversion.
-
由 Chesnay Schepler 提交于
-
由 Chesnay Schepler 提交于
-
由 Chesnay Schepler 提交于
-
由 Aljoscha Krettek 提交于
-
由 Aljoscha Krettek 提交于
-
由 Aljoscha Krettek 提交于
Apparently, search engines penalize META refresh links and this gets around it
-
由 Nico Kruber 提交于
-
由 Nico Kruber 提交于
When not serving docs and just building them, e.g. via `./build_docs.sh`, we spawn sub-processes for each language build instead of running them single-threaded (jekyll currently does not support parallel builds).
-
由 Nico Kruber 提交于
-
由 Chesnay Schepler 提交于
-
由 Chesnay Schepler 提交于
The google mirror for some reason does not contain 2.5.9.
-
由 Robert Metzger 提交于
This closes #11067
-
由 HuangXingBo 提交于
-
由 huangxingbo 提交于
This closes #11078.
-
由 vinoyang 提交于
-
- 12 2月, 2020 11 次提交
-
-
由 Chesnay Schepler 提交于
-
由 Chesnay Schepler 提交于
With jakarta EE having made their first releases projects may start transitioning to the re-branded versions. The package names are identical (for now?), so we should be fine as long as we ensure that at least one version is on the classpath. For the time being we stay on the "original" dependencies, but we may have to switch eventually.
-
由 Yu Li 提交于
-
由 Victor Wong 提交于
-
由 Jingsong Lee 提交于
This closes #11055
-
由 Zhenghua Gao 提交于
This bridges the JDBC table source to new type system to support precisions of types. In the meanwhile, this also adds a validation to check unsupported types and precisions by various dialects to avoid surprising results. This closes #10745
-
由 wangxlong 提交于
-
由 Steve Whelan 提交于
-
由 wangxlong 提交于
This can make the exception message more readable which can be thrown by Calcite instead of code generator. This closes #11058
-
由 felixzheng 提交于
This closes #10956 .
-
由 Dawid Wysakowicz 提交于
-
- 11 2月, 2020 6 次提交
-
-
由 Gary Yao 提交于
This closes #10997.
-
由 Zhijiang 提交于
-
由 Roman Khachatryan 提交于
Intermediate tasks (not source/sink) don't acquire lock anymore (as they use single-threaded model). SourceStreamTask still has to have the lock to conform to the public API. MockStreamTask has the lock for compatibility with the existing tests.
-
由 Roman Khachatryan 提交于
After a3b6be905f lock is not managed by operators so the test aren't relevant anymore.
-
由 Roman Khachatryan 提交于
After b4aa6e lock is acquired for every mailbox action if needed.
-
由 Zhijiang 提交于
[FLINK-15914][checkpointing][metrics] Miss the checkpoint related metrics for the case of two inputs When the StreamTwoInputSelectableProcessor was introduced before, it forgot adding the checkpoint related metrics in the constructor. But it did not cause any problems, because only the StreamTwoInputProcessor actually worked before. After StreamTwoInputProcessor is replaced by StreamTwoInputSelectableProcessor as now, this bug is exposed and we will not see the checkpoint related metrics for the case of two inputs. The solution is to add these metrics while constructing the CheckpointBarrierHandler.
-