[jira] [Comment Edited] (IGNITE-4551) Reconsider cache key/value peer class loading

["Nikolay Tikhonov (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/IGNITE-4551?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16019608#comment-16019608 ] 

Nikolay Tikhonov edited comment on IGNITE-4551 at 5/22/17 2:08 PM:
-------------------------------------------------------------------

* Added into {{CacheDataRow}} new methods {{keyClassLoaderId}}, {{valueClassLoaderId}} and {{deploymentEnabled}};
* Updated {{writeFragmentData}} and {{writeRowData}};
* Updated {{readData}} and {{readFragmentData}};


was (Author: ntikhonov):
* Added into {{CacheDataRow}} new methods {{keyClassLoaderId}}, {{valueClassLoaderId}} and {{deploymentEnabled}}.
* Updated {{writeFragmentData}} and {{writeRowData}}

> Reconsider cache key/value peer class loading
> ---------------------------------------------
>
>                 Key: IGNITE-4551
>                 URL: https://issues.apache.org/jira/browse/IGNITE-4551
>             Project: Ignite
>          Issue Type: Bug
>          Components: cache
>            Reporter: Alexey Goncharuk
>            Assignee: Nikolay Tikhonov
>             Fix For: 2.1
>
>
> In new cache implementation after entry is written in offheap information about key/value classloaders in lost (before classloader ids were stored in swap/offheap see GridCacheMapEntry.swap in 'master').
> Need decide how it should work with new architecture (maybe single type per cache can simplify implementation).



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)