[ https://issues.apache.org/jira/browse/MYFACES-3795?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13795780#comment-13795780 ]
Leonardo Uribe commented on MYFACES-3795: ----------------------------------------- It seems the bug was introduced in MYFACES-2777. It is clear that map should be in FacesContext attribute map. I'll fix that. > ErrorPageWriter cause memory leak in visitedFacetCount map > ---------------------------------------------------------- > > Key: MYFACES-3795 > URL: https://issues.apache.org/jira/browse/MYFACES-3795 > Project: MyFaces Core > Issue Type: Bug > Affects Versions: 2.1.11 > Environment: Linux > Reporter: Tomasz Szlek > > We have noticed that there is a memory leak in class > org.apache.myfaces.renderkit.ErrorPageWriter, map: visitedFacetCount. > This memory leak occurs when <ui:debug/> is used. > Description of problem: > Ui debug component tree page use visitedFacetCount as a temporary source of > UI components in order to generate html page based on such components. > This generator use visitor : ExtendedComponentTreeVisitCallback class. > Inside method visit we have following lines : > 972: _incrementVisitedFacetCount(parent); //put compnent into > visitedFacetCount map > ... > 1131: _removeVisitedFacetCount(parent); // remove visited component from > visitedFacetCount map. > Unfortunatelly visited component is removed from visitedFacetCount map only > if there are no children for component. > Thus it looks like components are never removed properly from > visitedFacetCount. > Possible solution: > Parrent componet should be removed from visitedFacetCount when it back from > recursion. -- This message was sent by Atlassian JIRA (v6.1#6144)