[ 
https://issues.apache.org/jira/browse/ASTERIXDB-1951?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16060457#comment-16060457
 ] 

ASF subversion and git services commented on ASTERIXDB-1951:
------------------------------------------------------------

Commit fe00b1af941382c3f47737e9b0d5a9b667365dc7 in asterixdb's branch 
refs/heads/master from [~dlychagin-cb]
[ https://git-wip-us.apache.org/repos/asf?p=asterixdb.git;h=fe00b1a ]

[ASTERIXDB-1951][FUN] Cleanup type constructor functions

- user model changes: yes

1) primitive type constructor functions now accept input values
   of their respective types and return them as is
2) removed type constructor function for 'null' type

- storage format changes: no
- interface changes: no

Change-Id: I4a6627bfcc302b2621df8d7837c15434970b9d3a
Reviewed-on: https://asterix-gerrit.ics.uci.edu/1846
Sonar-Qube: Jenkins <[email protected]>
Tested-by: Jenkins <[email protected]>
BAD: Jenkins <[email protected]>
Integration-Tests: Jenkins <[email protected]>
Reviewed-by: Yingyi Bu <[email protected]>


> Cleanup type constructor functions
> ----------------------------------
>
>                 Key: ASTERIXDB-1951
>                 URL: https://issues.apache.org/jira/browse/ASTERIXDB-1951
>             Project: Apache AsterixDB
>          Issue Type: Improvement
>          Components: FUN - Functions
>            Reporter: Dmitry Lychagin
>
> 1) Currently primitive type constructor functions behavior is inconsistent 
> when input value has the same type as the constructed type.
> Constructor functions for 'string' and 'binary' types allow input value to be 
> 'string'/'binary' and just return it as is, while other functions raise type 
> mismatch exception.
> We should make sure that all constructor functions for primitive types pass 
> through values of their respective types.
> 2) Current implementation for 'null' type constructor function is incorrect. 
> It returns 'missing' instead of 'null'. Moreover it's unclear whether this 
> function is necessary at all, because
> - we already have a keyword for creating 'null's 
> - the function only accepts one string value: 'null', while raising error for 
> all other strings. 
> The proposal is to remove this type constructor function from the function 
> library



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Reply via email to