[ https://issues.apache.org/jira/browse/MYFACES-3835?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13851925#comment-13851925 ]
Leonardo Uribe commented on MYFACES-3835: ----------------------------------------- Just for the record, It seems a bug on webkit https://bugs.webkit.org/show_bug.cgi?id=35284 https://bugzilla.gnome.org/show_bug.cgi?id=610802 http://code.google.com/p/chromium/issues/detail?id=36431 There is no relationship with any param of MyFaces. There is similar report to this one here: http://stackoverflow.com/questions/20316773/primefaces-4-0-fileupload-works-with-mojarra-2-2-but-not-myfaces-2-2 Maybe we need a workaround on the javascript part to deal with this chrome bug. > ViewState gets truncated on chrome with richfaces fileupload component > ---------------------------------------------------------------------- > > Key: MYFACES-3835 > URL: https://issues.apache.org/jira/browse/MYFACES-3835 > Project: MyFaces Core > Issue Type: Bug > Affects Versions: 2.1.11, 2.1.13 > Environment: Windows XP, Chrome 31.0.1650.63 m (latest at this > moment), tomcat 7.0.37, client state saving, myfaces 2.1.11, richfaces > 4.3.1.Final > Reporter: Ricardo Tercero Lozano > Attachments: FileUploadBean.class, FileUploadBean.java, > imgUpload-sample.xhtml, web.xml > > > On certain conditions viewstate gets corrupted (truncated). > I've got a page with a richfaces fileupload component. The page works well on > IE7 and Firefox (latest), but not in chrome. Digging into Javascript and Ajax > response I got some extra info about the problem. I don't know why, but a > partial response like: > <?xml version="1.0" encoding="utf-8"?><partial-response><changes><update > id="javax.faces.ViewState"><![CDATA[.... > results in two CDATA sections when handling the response. This is the problem > caused by Google Chrome. Inspecting the JSF.JS library, the line that gets de > updated view state is: > mfInternal.appliedViewState = node.firstChild.nodeValue; > This line is in 'processUpdate' method. When Chrome, for some reason splits > the original CDATA block into two, that line only updates the first section, > obtaining a truncated viewState and ViewExpiredException in next request. > The first CDATA section created by Google Chrome has 300 bytes. Weird, but > searching Google for 'Chrome cdata 300' appears to be a libxml2 problem. -- This message was sent by Atlassian JIRA (v6.1.4#6159)