[
https://issues.apache.org/jira/browse/WINK-115?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12738018#action_12738018
]
Michael Elman commented on WINK-115:
------------------------------------
The solution looks very good.
Just few really small points regarding the implementation:
* The key for the providersCache map can be Class instead of String. So there
won't be need for a String className = cls.getName() call. Since we are talking
about efficiency here, this call is redundant and the hashCode/equals of Class
should be faster, then of String.
* The key for the inner map can be MediaType instead of String, so there is no
need to build the cacheKey.
I'm just curious: does it really makes 10% performance improvement?
> Add caching to MediaTypeMap.getProvidersbyMediaType()
> ------------------------------------------------------
>
> Key: WINK-115
> URL: https://issues.apache.org/jira/browse/WINK-115
> Project: Wink
> Issue Type: Improvement
> Components: Common
> Affects Versions: 0.1
> Reporter: Nick Gallardo
> Assignee: Nick Gallardo
> Fix For: 0.2
>
> Attachments: WINK-115.patch
>
>
> Calls to MediaTypeMap.getProvidersByMediaType() are expensive, and happen a
> multiple times per invocation. Profiling the runtime shows this as one of
> the major time blocks in successive get() calls.
> Once we've done the work to define a mapping between the MediaType and the
> parameter class, we can cache that content so we don't have to do the same
> level of processing again. One way to cache would be to build a Matrix of
> MediaTypes to Class type mappings. So, for every MediaType, we can lookup a
> particular parameter class type and find the cached list of providers that
> apply to that type.
> Caching this content results in at least 10% performance improvement.
> Working on a patch for this one. Still trying to workout the caching scheme
> exactly.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
