I agree with not controlling the UIViewRoot -Matthias
On Tue, Sep 21, 2010 at 12:44 PM, MAX STARETS <max.star...@oracle.com> wrote: > Hi Andrew, > > I think I would rather require <tr:document> or <trh:body> (vs. tr:html) > We already say in our documentation that one of these tags is required for > PPR to work. > > Max > > On 9/21/2010 3:09 PM, Andrew Robinson wrote: >> >> Trinidad IMO should not be controlling the UIViewRoot, that is for the >> container to create. If we overrode it, it could also cause problems >> with portals. I think it best that we handle this in tr:document and >> trh:html as they should be the root elements of a Trinidad page. I am >> concerned that if we override the view root it will cause other issues >> or implementation problems. >> >> On Tue, Sep 21, 2010 at 12:44 PM, Pavitra Subramaniam >> <pavitra.subraman...@oracle.com> wrote: >> >>> >>> Hello Andrew, >>> >>> One comment regarding your changes in: >>> >>> +public abstract class UIXDocumentTemplate >>> + extends UIXComponentBase >>> >>> Overall your fixes look fine to me, but I was wondering if rather than >>> adding the override methods to the new class UIXDocument class, if >>> instead >>> this should be moved out of the UIXDocument and into a UIViewRoot >>> subclass. >>> For a different issue (of re-entrant calls to visitTree not properly >>> suspending and restoring context) that I encountered, I had a similar fix >>> in >>> mind but after discussing with Blake he proposed that adding it to >>> UIViewRoot subclass may be a better option. >>> >>> Thanks >>> Pavitra >>> >>> On 9/21/2010 9:43 AM, Andrew Robinson wrote: >>> >>>> >>>> Please see JIRA: >>>> >>>> https://issues.apache.org/jira/browse/TRINIDAD-1919 >>>> >>>> I want to propose adding a set of APIs to Trinidad2 that will allow >>>> component authors to suspend changes and resume changes before and >>>> after an invokeOnComponent or visitTree call. Currently, component >>>> context is not reverted during these calls, and bugs may result. The >>>> proposed changes would allow components to tear down and set back up >>>> their context over one of these calls. >>>> >>>> I have attached a way to reproduce the issue and my proposed patch to >>>> the bug. I'll wait a few days for feedback before going ahead and >>>> committing it. >>>> >>>> Thank you, >>>> Andrew >>>> >>>> > -- Matthias Wessendorf blog: http://matthiaswessendorf.wordpress.com/ sessions: http://www.slideshare.net/mwessendorf twitter: http://twitter.com/mwessendorf