Slackbot
12/06/2023, 3:28 PMDavid K
12/06/2023, 4:08 PMUmar Cheema
12/06/2023, 4:12 PMZsolt Kovacs
12/06/2023, 4:21 PMDavid K
12/06/2023, 4:55 PMZsolt Kovacs
12/06/2023, 5:04 PMDavid K
12/06/2023, 5:17 PMUmar Cheema
12/06/2023, 11:25 PMConsumePulsarRecord line 266-7 instead of reading the schema from RecordReaderFactory as in
schema = getSchema(flowFile, readerFactory, data);
we do something like
if (lastMessage == null) {
flowFile = session.create();
flowFile = session.putAllAttributes(flowFile, currentAttributes);
schema = new SimpleRecordSchema(
new String(msg.getReaderSchema().get().getSchemaInfo().getSchema()),
"avro",
SchemaIdentifier.EMPTY
);
rawOut = session.write(flowFile);
writer = getRecordWriter(writerFactory, schema, rawOut, flowFile);
By making the reader property optional in the processor, the AVROSetWriter can continue without interruption, allowing the processor to operate based on embedded schemas. I have created a prototype that correctly generates a flow file with this modification. Do you anticipate any potential problems with this approach?David K
12/06/2023, 11:40 PMmsg.getReaderSchema() is not null, etc. Test cases will have to be added for the different schema types in pulsar, e.g. KeyValue, GenericObject, primitives, etc.David K
12/06/2023, 11:44 PMUmar Cheema
12/13/2023, 9:35 PMZsolt Kovacs
12/27/2023, 12:38 PMDavid K
12/27/2023, 4:42 PMZsolt Kovacs
12/27/2023, 4:49 PMUmar Cheema
01/05/2024, 6:19 PMDavid K
01/08/2024, 6:52 PMUmar Cheema
01/08/2024, 6:52 PM