[jira] [Comment Edited] (FLEX-34260) s:datagrid cacheAsBitmap caches an itemRenderer for every item in the dataProvider

["Vitaliy Parfenov (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/FLEX-34260?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14222713#comment-14222713 ] 

Vitaliy Parfenov edited comment on FLEX-34260 at 11/24/14 6:28 AM:
-------------------------------------------------------------------

flex-34260-fdk4.13.png - memory spike with Flex SDK 4.13
flex-34260-fdk_mod.png - same data in spark DataGrid on Flex SDK 4.13 with old DataGrid classes from SDK 4.9.1:
spark.components.gridClasses.DataGridEditor.as
spark.components.gridClasses.GridColumnHeaderGroupLayout.as
spark.components.gridClasses.GridLayout.as
spark.components.DataGrid.as
spark.components.Grid.as
spark.components.GridColumnHeaderGroup.as
spark.skins.spark.DataGridSkin.as
spark.skins.spark.DefaultGridHeaderRenderer.as


was (Author: vsp):
flex-34260-fdk4.13.png - memory spike with Flex SDK 4.13
flex-34260-fdk_mod.png - same data in spark DataGrid on Flex SDK 4.13 with old DataGrid classes from SDK 4.9.1

> s:datagrid cacheAsBitmap caches an itemRenderer for every item in the dataProvider
> ----------------------------------------------------------------------------------
>
>                 Key: FLEX-34260
>                 URL: https://issues.apache.org/jira/browse/FLEX-34260
>             Project: Apache Flex
>          Issue Type: Bug
>          Components: Spark: DataGrid
>    Affects Versions: Apache Flex 4.11.0, Apache Flex 4.12.0
>         Environment: Windows 7, FB 4.7, ie10
>            Reporter: Jeff Spicer
>              Labels: easytest
>         Attachments: ItemRendererCachingBug.fxp, flex-34260-fdk4.13.png, flex-34260-fdk_mod.png
>
>
> setting cacheAsBitmap to true on a spark dataGrid allocates a lot of memory, noticeably when the dataProvider is long.
> My guess is it's allocating enough memory to cache an itemRenderer for each item in the dataprovider, whether it's showing or not.
> Steps to reproduce:
> (fxp attached)
> 1. create a spark dataGrid which has a dataProvider of length 1000
> 2. set cacheAsBitmap to true
> 3. run and profile memory usage
> a dataprovider of length 10,000 shows no significant spike in memory, using sdk 4.6.
> in 4.11/12, 10,000 items can cause !GCOD



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)