David has found a few interesting things about the FactoryFinder implementation. I think we can all agree that the memory leak can be fixed, I have mixed feelings about the use of synchronization here. The code is invoked from many places in the app and I am mostly concerned about performance.
If FactoryFinder.setFactory() is only invoked from the context startup listener, or the servlet's init method, do we need synchronization? Dennis Byrne On 3/13/07, David Jencks (JIRA) <dev@myfaces.apache.org> wrote:
[ https://issues.apache.org/jira/browse/MYFACES-1558?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel] David Jencks updated MYFACES-1558: ---------------------------------- Attachment: MYFACES-1558-jsf11.patch Here's a patch for the jsf11 branch. I believe its the same as the patch for jsf 1.2 branch with generics removed, but I spent less time on it than the first one. The entire jsf11 project builds fine with this. > FactoryFinder not thread safe, and has a memory leak > ---------------------------------------------------- > > Key: MYFACES-1558 > URL: https://issues.apache.org/jira/browse/MYFACES-1558 > Project: MyFaces Core > Issue Type: Bug > Components: General > Affects Versions: 1.2.0-SNAPSHOT > Reporter: David Jencks > Assigned To: Dennis Byrne > Attachments: MYFACES-1558-jsf11.patch, MYFACES-1558.patch > > > FactoryFinder has insufficient synchronization to be thread safe, and has a memory leak in _registeredFactoryNames. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
-- Dennis Byrne