You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@aurora.apache.org by "Mark Chu-Carroll (JIRA)" <ji...@apache.org> on 2014/01/16 21:49:19 UTC
[jira] [Commented] (AURORA-42) Modify client side code to
distinguish between api objects
[ https://issues.apache.org/jira/browse/AURORA-42?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13873914#comment-13873914 ]
Mark Chu-Carroll commented on AURORA-42:
----------------------------------------
Review: https://reviews.apache.org/r/16984/
> Modify client side code to distinguish between api objects
> ----------------------------------------------------------
>
> Key: AURORA-42
> URL: https://issues.apache.org/jira/browse/AURORA-42
> Project: Aurora
> Issue Type: Task
> Components: Client
> Reporter: Mark Chu-Carroll
> Assignee: Mark Chu-Carroll
>
> In client work, especially writing tests, the term "scheduler" is overloaded in the code, and this makes diagnosing and debugging failures difficult.
> An aurora API object contains a property named "scheduler", which is a reference to a SchedulerProxy. A SchedulerProxy, in turn, contains a property named "scheduler", which is actually a reference to a SchedulerClient.
> As a result, code frequently reads "foo. scheduler.scheduler.something"; in tests, when you see a "scheduler" parameter, it's either a SchedulerProxy or a SchedulerClient, with no good way to distinguish. This is especially problematic when you see a test failure: foo. scheduler didn't get an expected call; what is foo.scheduler?
> To clear this mess up, we should get rid of the "scheduler" fields, and rename them to make clear what they are: api contains a field "scheduler_proxy"; SchedulerProxy contains a field "scheduler_client".
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)