Mistake in PRINT statement on stream-stream join tutorial

[chuck-confluent]

In this tutorial, there is a mistake in this PRINT statement:

PRINT SHIPPED_ORDERS FROM BEGINNING LIMIT 3;

The PRINT statement takes a topic as input, not a kstream. To use PRINT, the user would have to use the name of the derived topic, or an earlier step would have to use a WITH kafka_topic = 'blah' to set the topic name.

A better approach would probably be to use this instead:

SELECT * FROM SHIPPED_ORDERS EMIT CHANGES LIMIT 3;

[mikebin]

Is this statement not valid? SHIPPED_ORDERS would be the name of the output topic in this case, and the intent of the tutorial seems to be to show what's in the topic.

[chuck-confluent]

I see -- I'm working in Confluent Cloud where the topic is called pksqlc-yo50oSHIPPED_ORDERS

You're right, we lose the intentioned purpose, which is to show that the data is in a kafka topic and consumable by any consumer, not just ksqlDB.

The safe thing to do would be to set the topic name explicitly with a WITH statement. There's already topics _shipped and _orders, so _shipped-orders or something similar would work

[ybyzek]

cc: @bbejeck regarding @chuck-confluent 's comment above -- this an example of one issue KT would have to take into consideration if it presents Confluent Cloud workflows but validates locally.

[chuck-confluent]

On a strange related note, I have never gotten this to work in confluent cloud for this tutorial:

PRINT 'pksqlc-yo50oSHIPPED_ORDERS' FROM BEGINNING;

I know there are messages in the topic because I was able to see them with a SELECT statement on the kstream. I can't print the topic, and when I go to the topic viewer and jump to offset 0 of each partition, I see nothing.

I can print and inspect the messages of other topics, but something weird might be happening if there are very few messages in a topic. This might affect other tutorials in confluent cloud that don't use datagen, but instead use insert into with a small number of input records.

[bbejeck]

IMO this isn't a bug; it's working as expected locally. We definitely need to take into consideration running the tutorials in the cloud vs. locally. But it's important to realize that KT was set up entirely thinking of running locally.