So this is just a performance concern? I agree that creating a new Mapper
inside of a loop is obviously a bad idea. That's why :
public class StreamsJacksonMapper extends ObjectMapper {
private static final StreamsJacksonMapper INSTANCE = new
StreamsJacksonMapper();
public static StreamsJacksonMapper getInstance(){ return INSTANCE; }
I don't think the fact that bad implementations can get slow is a
reason to make the default behavior worse, to make a call that should
only happen once per jvm a bit faster.
On Tue, Feb 17, 2015 at 2:07 PM, ASF GitHub Bot (JIRA) <[email protected]>
wrote:
>
> [
> https://issues.apache.org/jira/browse/STREAMS-288?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14324789#comment-14324789
> ]
>
> ASF GitHub Bot commented on STREAMS-288:
> ----------------------------------------
>
> Github user smashew commented on the pull request:
>
>
> https://github.com/apache/incubator-streams/pull/187#issuecomment-74742070
>
> If you scan, you should only scan the classpath ONCE! Collect the
> classes that you want to use and never, ever, EVER scan the class-path,
> ever again.
>
>
> > StreamsJacksonModule should not scan for DateTimeFormats by default
> > -------------------------------------------------------------------
> >
> > Key: STREAMS-288
> > URL: https://issues.apache.org/jira/browse/STREAMS-288
> > Project: Streams
> > Issue Type: Bug
> > Reporter: Robert Douglas
> >
> > The StreamsJacksonModule's constructor is called all over the codebase
> and by default, it uses Reflections to scan the classpath for valid
> DateTimeFormats.
> > This is problematic because it will happen multiple times during a
> Stream's runtime, considerably slowing down the execution.
> > Instead, we should default to NOT scanning the classpath, but provide a
> constructor where we can pass in a flag dictating whether or not we want to
> scan.
>
>
>
> --
> This message was sent by Atlassian JIRA
> (v6.3.4#6332)
>
