> On April 8, 2015, 4:51 a.m., Jun Rao wrote:
> > core/src/test/scala/unit/kafka/server/KafkaConfigTest.scala, line 137
> > <https://reviews.apache.org/r/32953/diff/1/?file=920411#file920411line137>
> >
> >     Do you know why the test passes before? ConfigDef currently seems to 
> > ignore undefined properties, instead of throwing an exception.
> 
> Onur Karaman wrote:
>     KafkaConfig's configDef has:
>     ```
>     .define(AdvertisedListenersProp, STRING, HIGH, AdvertisedListenersDoc, 
> false)
>     ```
>     
>     So AdvertisedListenersProp doesn't require a default and no default is 
> provided.
>     With the typo, no required default, and no provided default, 
> ConfigDef.parse just sets the value associated with "advertised.listeners" to 
> null.
> 
> Gwen Shapira wrote:
>     Actually, this is not why the test passed.
>     
>     The test checks for *invalid* configuration - i.e. we set the 
> configuration and pass if IllegalArgumentException is thrown.
>     In this case, the exception will be thrown if the configuration is 
> invalid because we set the same port twice (what we wanted to test) or if the 
> configuration is invalid because we mistakenly configured a non-existing 
> parameter (what we actually tested).
>     
>     I'm not sure how we want to address this (or if it is even worth 
> addressing) - having different exceptions for different types of invalid 
> configurations seems a too confusing.
>     We can change the test to check the specific text of the exception, but I 
> really wouldn't want changes of messages to break tests.

Got it. Should we use the constants defined in KafkaConfig to void human errors?


- Jun


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/32953/#review79296
-----------------------------------------------------------


On April 8, 2015, 1:54 a.m., Gwen Shapira wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/32953/
> -----------------------------------------------------------
> 
> (Updated April 8, 2015, 1:54 a.m.)
> 
> 
> Review request for kafka.
> 
> 
> Bugs: KAFKA-2104
>     https://issues.apache.org/jira/browse/KAFKA-2104
> 
> 
> Repository: kafka
> 
> 
> Description
> -------
> 
> fixed typo
> 
> 
> Diffs
> -----
> 
>   core/src/test/scala/unit/kafka/server/KafkaConfigTest.scala 
> b3947191735ee149d050bb48db7bc3b2483f2dd7 
> 
> Diff: https://reviews.apache.org/r/32953/diff/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Gwen Shapira
> 
>

Reply via email to