feathub
feathub copied to clipboard
alibaba
FeatHub - A stream-batch unified feature store for real-time machine learning
This close #256 ## What is the purpose of the change *(For example: This pull request introduces Flink Processor.)* ## Brief change log *(for example:)* - *Introduces Flink Processor* -...
Thanks for you work, it seems SlidingWindow doesn't support restore state from CheckPoint ``` 2023-10-07 16:22:31 java.io.IOException: Exception while applying AggregateFunction in aggregating state at org.apache.flink.runtime.state.heap.HeapAggregatingState.add(HeapAggregatingState.java:107) at org.apache.flink.streaming.runtime.operators.windowing.WindowOperator.processElement(WindowOperator.java:404) at org.apache.flink.streaming.runtime.tasks.OneInputStreamTask$StreamTaskNetworkOutput.emitRecord(OneInputStreamTask.java:233)...
build-cli-deps.sh ```shell # zip Feathub and its dependencies cd "${CURRENT_DIR}" if [[ -z "${FEATHUB_WHEEL_PATH}" ]]; then # TODO: Install the latest stable version after Feathub released. python -m pip install...
We should also replace timestamp format values such as `epoch` and `epoch_millis` with enums.
Add test and document for built-in function ARRAY.
Feathub should provide document for configurations like `common.timeZone`.
SlidingWindowKeyedProcessFunction.java: ```java // TODO: Register at most one timer at a timestamp regardless of the number of the keys // processed by the operator. while (triggerTime
mysql_registry.py ```python # TODO: add document about the limitations on the length of feature # table names. if len(table.name) > 64: raise FeathubException( "Cannot build or register a descriptor with...
Add general and formal description for concepts like bounded/unbounded stream, max out of orderness, (watermark,) and late data in Feathub documents.