Hi Leo, a first idea would be to remember if validation was executed at all. If not, the state should be kept. I added a simple patch for this to the issue (it works for me but there might be side effects I'm not aware of). Am 27.09.2011 um 16:23 schrieb Leonardo Uribe:
> Hi > > Some weeks ago, I read the description of UIData.encodeBegin() too. > > That description comes from JSF 1.0, so in that time it had sense, but > with JSF 2.0 we have new use cases and that description is becoming > problematic. For example, originally the idea was that if a > conversion/validation error occur, an error message is added so the > model/state must be preserved. > > In JSF 2.0 we have FacesContext.isValidationFailed() method, so one > idea is check this condition in that part, but still it seems to be > insuficient. I suppose a change on the spec should be done in > that part, but I still don't get how it should looks like. There are some > cases like the one described on MYFACES-3326 that requires > something "special", to indicate the model/state should not be > cleared. > > For now the only workaround is create a copy and implement > UIData class and add some code there, just like in tomahawk, but > that does not solve the problem on the spec. > > Suggestions are welcome. > > regards, > > Leonardo Uribe > > 2011/9/27 Michael Kurz <michi.k...@gmx.at> >> >> Hi, >> >> I would like to move the discussion about MYFACES-3326 ([1]) from the issue >> over her (as suggested by Mike). >> >> I don't want to repeat the whole story, but I am very interested if this is >> a bug (probably not as Mojarra behaves the same way), if it is >> underspecified or if I wrongly assumed that this should work like I expected. >> >> Best regards >> Michael >> >> [1]: https://issues.apache.org/jira/browse/MYFACES-3326