Hi,
At least for the AJAX part I can tell that many details are treated as
implementation details, so the spec purposely leaves them open. On some
details in the jsdocs Roger has even relaxed the wording of the spec to
leave the details to the impl. If we muddle through the specs and do it
our way the MyFaces AJAX and the Mojarra AJAX won't have compatible XML
interfaces (syntactically, yes, but not sematically). This would be bad
in two ways:
1. The way things have been going until now the spec will further evolve
to match the behaviour of Mojarra. They have all right to try out new
features in their impl before writing them into the spec, but it seems
wise to stay close.
2. If the MyFaces Javascript doesn't work with the Mojarra server, the
t:ajax tag (I'm working on it ...) wouldn't run with Mojarra.
Here is how I've been treating this: By testing with Mojarra I've
checked their implementation's behaviour if I wasn't sure how to
interpret the specs. Just running it and checking a test apps behaviour
as well as the transferred XML stream for serveral corner cases
definitely doesn't break any license. I don't think reading their code
is a helpful idea - we'd just repeat the same errors they have made and
aren't we convinced to know it better? ;-)
Best regards,
Ganesh
Matthias Wessendorf schrieb:
isn't the behavior *clear* defined in the spec ?
If so, go the route;
If not, ask the EG on the why ;-)
-Matthias
