Publish exception in lifecycle methods (process*) instead of re-thrown
----------------------------------------------------------------------
Key: MYFACES-3201
URL: https://issues.apache.org/jira/browse/MYFACES-3201
Project: MyFaces Core
Issue Type: Sub-task
Reporter: Martin Kočí
Requirement: "user should see not just a cryptic stack trace, but also the
component that triggered the problem"
Problem in current code is that first exception breaks current phase and
exception in queued without component info.
I think that every lifecycle method (processDecodes, processValidator etc.)
should try catch every exception and publish it for later processing with
exception handler.
Spec does not says it directly but we can find:
"The exception must not be re-thrown. This enables tree traversal to continue
for this lifecycle phase, as in all the other lifecycle phases" from
UIInput.updateModel
"ExceptionHandler is the central point for handling unexpected Exceptions that
are thrown during the Faces lifecycle" from ExceptionHandler javadoc
process* method can silently "do nothing" : UIInput.updateModel does it already.
Publishing event allows handle multiple problem at once: consider buggy
validators/converters -> create more than one exception in queue and coder can
see them at once.
The main parameter of ExceptionQueuedContext is UIComponent and the best place
where component is always known is component itself.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
