accumulo-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Sean Busbey (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (ACCUMULO-2487) Value implementation provides conflicting statements
Date Mon, 17 Mar 2014 20:36:43 GMT

    [ https://issues.apache.org/jira/browse/ACCUMULO-2487?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13938341#comment-13938341
] 

Sean Busbey commented on ACCUMULO-2487:
---------------------------------------

We can just use a static immutable zero-sized sequence. that way the readfields isn't wasting
an array allocation. or atleast there's only 1 per JVM instance, and I believe we already
have one sitting around.

In general, I favor javadocs over implementation. Specifically in the case of teh public API
it's unreasonable to expect downstream dev to read code.

> Value implementation provides conflicting statements
> ----------------------------------------------------
>
>                 Key: ACCUMULO-2487
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-2487
>             Project: Accumulo
>          Issue Type: Bug
>          Components: client
>            Reporter: Mike Drob
>            Priority: Critical
>             Fix For: 1.7.0
>
>
> The javadoc for the no-arg constructor for {{Value}} states that it "Creates a zero-size
sequence." However, the implementation of get will error in this case.
> {code}
> public byte[] get() {
>     if (this.value == null) {
>       throw new IllegalStateException("Uninitialized. Null constructor " + "called w/o
accompanying readFields invocation");
>     }
> {code}
> Either we need to change the javadoc to be more explicit or change the behaviour of various
accessors in the class. I would consider both solutions to be breaking of the API contract
since we are changing what clients can expect from us.



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Mime
View raw message