[
https://issues.apache.org/jira/browse/KAFKA-391?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Joel Koshy updated KAFKA-391:
-----------------------------
Attachment: KAFKA-391-v2.patch
2.1 - Rebased after the first review, but I will need to rebase again. Unit
tests and system tests pass. This patch applies cleanly to svn revision
1381858 - it would be great if this can be reviewed against that revision.
I can provide an incremental patch if that helps after I rebase.
2.2 - Removed all the equals methods in classes that were using Arrays earlier.
2.3 - Switched fetch request/response to maps. It makes the code a little
cleaner, but I think the improvement was greater with the refactoring on
the producer side. Anyway, this makes the APIs more consistent. One small
observation: previously, fetch requests would throw a
FetchRequestFormatException if the TopicData array for a fetch request
contained the same topic multiple times. Right now we don't do any checks
since we use a map. We can add it to the FetchRequestBuilder, but not sure
if it is required/worth it. Also, I think it previously would allow
fetches from different offsets in the same partition in the same fetch
request. That is no longer allowed, although I don't know why anyone would
need that.
2.4 - Removed the global error code.
Another minor detail: I wonder if it would help to have a case class for
TopicAndPartition. We use (topic, partition) and the associated tuple
addressing all over the place. That would make the Map declarations slightly
clearer, although now we're accustomed to understanding that (String, Int)
must mean (topic, partitionId).
> Producer request and response classes should use maps
> -----------------------------------------------------
>
> Key: KAFKA-391
> URL: https://issues.apache.org/jira/browse/KAFKA-391
> Project: Kafka
> Issue Type: Bug
> Reporter: Joel Koshy
> Assignee: Joel Koshy
> Labels: optimization
> Fix For: 0.8
>
> Attachments: KAFKA-391-draft-r1374069.patch, KAFKA-391-v2.patch
>
>
> Producer response contains two arrays of error codes and offsets - the
> ordering in these arrays correspond to the flattened ordering of the request
> arrays.
> It would be better to switch to maps in the request and response as this
> would make the code clearer and more efficient (right now, linear scans are
> used in handling producer acks).
> We can probably do the same in the fetch request/response.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
