...
Deferring dataset creation in prepareRun will require adding a new method to BatchSinkContextBatchContext.
Code Block |
---|
@Beta public interface BatchContext extends DatasetContext, TransformContext { // new method void createDataset(String datasetName, String typeName, DatasetProperties properties); // existing methods long getLogicalStartTime(); /** * Returns runtime arguments of the Batch Job. * * @return runtime arguments of the Batch Job. */ Map<String, String> getRuntimeArguments(); /** * Updates an entry in the runtime arguments. * * @param key key to update * @param value value to update to * @param overwrite if {@code true} and if the key exists in the runtime arguments, it will get overwritten to * the given value; if {@code false}, the existing value of the key won't get updated. */ void setRuntimeArgument(String key, String value, boolean overwrite); /** * Returns the hadoop job. * @deprecated this method will be removed. */ @Deprecated <T> T getHadoopJob(); ... } |
...