Hi ! I had a question regarding the Pulsar plugin....
# pinot-dev
m
Hi ! I had a question regarding the Pulsar plugin. Adding a realtime table to Pinot requires a time column. Natively, a Pulsar message contains 2 temporal metadata fields: the publishTime and the eventTime (but which are not part of the payload). I think it would be interesting that the Pulsar plugin could directly integrate the publishTime and the eventTime as available fields inside the GenericRow in addition to the payload fields, this would allow the direct use of Pulsar messages without adding an additional redundant field in the schema. The evolution should be relatively simple by modifying the "public GenericRow next(GenericRow destination)" method of the PulsarStreamLevelConsumer class (unless i forget something). What do you think? Thank you !
x
cc: @Kartik Khare
k
I added this support recently for kafka
looks like @Navina is extending this to other sources as well - https://github.com/apache/pinot/pull/9224
🙏 1
n
@Mathieu Druart yep. Your use-case will be possible with this PR. It will likely get merged this week. and I can follow-it up with a pulsar extension.
m
thank you @Navina!