Dennis Byrne wrote:
I wrote a unit test comparing the myfaces TLDs(html and core)
and those of the RI. There are more than 100 disrepencies.
Most of these are attributes that are spelled differently,
attributes that are in the RI but missing in MyFaces, and
attributes that are in MyFaces but not the RI.
I have always been under the impression that "h" and "f" are
supposed to implement only those attributes found in the spec
and that anything else must go in something like tomahawk -
so that end users can avoid being locked in to a particular
JSF implementation and to help tools vendors.
For example, if the spec is silent on an attribute, and
Implementation A has added it to the html or core TLDs, is
this an issue for Implementation A, Implementation B, or
neither? Furthermore, is the answer to this the same across
the entire JSR process? Can Hibernate put @Filter (not in
JSR 220) feature under javax.persistence, rather than
org.hibernate.annotations ?
I would agree with you that the attributes defined in the TLD for the
Sun RI and MyFaces should be identical.
I'm sure these are just accidental differences. If you can provide a
list of the differences (and hopefully the tools you used to create this
list) I'm willing to put in some time to clean this up (unless someone
wants to claim that differences here are ok).
I'm surprised that these differences weren't picked up by the Sun TCK;
if the TCK isn't checking for TLD differences then I think it should be.
While on this topic, it might be nice to try to run clirr
(clirr.sourceforge.net) comparing the sun RI with myfaces-api.jar to
look for API differences (rather than TLD differences).
Cheers,
Simon
- Re: ri compatibility Simon Kitching
-
