MattBelle commented on PR #28979: URL: https://github.com/apache/flink/pull/28979#issuecomment-6218014195
> Hey @MattBelle @dianfu I was having a look here and wondering why he have added `isNotNan` additionally to `isNan`? I think `isNan` is useful and I can see that probably in a lot of cases `isNotNan` will be used, but I'm a bit torn here between offering `isNotNan` or settling only on one function which can still serve the use case like `NOT IS_NAN` or `not(isNan(...))`. What are your opinions on this? Logically I agree with you, but as I noted [here](https://github.com/apache/flink/pull/28979#discussion_r3856334151) `IS_NOT_NULL` was already a thing before this change. From a user perspective I'd expect `NaN` operations to mirror `Null` operations so if we remove the negative for `Nan` I think we should do the same for `Null` which would be a breaking change. The argument could be made that `IS_NOT_NULL` is tech debt that shouldn't be brought forward to `NaN`, but if we went that route I 'd expect a warning or deprecation notice be tied to the `Null` operator. I'm obviously new to this project so feel free to correct my analysis if I'm missing historical context or something. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
