KAFKA-2622: Add Time logical type for Copycat.#285
KAFKA-2622: Add Time logical type for Copycat.#285ewencp wants to merge 2 commits intoapache:trunkfrom
Conversation
|
@guozhangwang Quick follow up to the logical types patch. I realized we can't avoid adding a Time type for now since we need it for SQL types in the JDBC connector. Should be a quick review since this is basically the same as the Date type, just zeroing out the date fields instead of the time fields. |
|
Actually, hold off on review. I realized there's an issue with validating values when you use Struct.put() that I still need to resolve. |
|
Ok, updated to handle logical types in validation, which is really only happening in Struct.put() for now. Should really be ready for review now, @guozhangwang |
|
LGTM. |
There was a problem hiding this comment.
Is the claim in the exception correct? We should have have non-zero values for some fields.
There was a problem hiding this comment.
Right. It should be "any date field"
…on (apache#285) TICKET = LIKAFKA-40747 LI_DESCRIPTION = A potential reason for NoOffsetForPartitionException is the transition of FetchState for partitions to any state other than INITIALIZING (e.g. offset_reset). However, there are not enough logs on the client side to see if that was the case. Add logs in the exception message to track the actual fetch state of partitions when a NoOffsetForPartitionException is thrown. (cherry picked from commit aaf8c8d)
…on (apache#285) TICKET = LIKAFKA-40747 LI_DESCRIPTION = A potential reason for NoOffsetForPartitionException is the transition of FetchState for partitions to any state other than INITIALIZING (e.g. offset_reset). However, there are not enough logs on the client side to see if that was the case. Add logs in the exception message to track the actual fetch state of partitions when a NoOffsetForPartitionException is thrown. (cherry picked from commit aaf8c8d)
…on (apache#285) TICKET = LIKAFKA-40747 LI_DESCRIPTION = A potential reason for NoOffsetForPartitionException is the transition of FetchState for partitions to any state other than INITIALIZING (e.g. offset_reset). However, there are not enough logs on the client side to see if that was the case. Add logs in the exception message to track the actual fetch state of partitions when a NoOffsetForPartitionException is thrown. (cherry picked from commit aaf8c8d)
No description provided.