JSF Cross Browser Compatibility

Question

In reading about JSF I find no promise for cross browser compatibility. Am I to assume that this has been missed because
i. JSF is a specification and not an implementation and I should be looking for the implementation to make that promise.
ii. it's basically to hard to promise this.
iii. I didn't look hard enough.
I am guessing it's a (i) with the likely hood of (ii) to follow. But I would be interested on any feed back that anyone cares to add. For example, a implementation that is trying to promise this.

Answer 1

In my opinion, the JSF specification (or better, any specification) can only refer to other specifications, but never to a concrete software product. Consider a sentence like:
        JSF 1.0 is compatible with IEWhat could that mean? Some possibilities:
(1)     As it happened, until now, we did not stumble over a
        faces application that broke when run on our Faces
        reference implementation (version 1_1_01) in a
        WSAD 5.0 container in combination with IE 5.5
        with patch 12345 on Windows NT4 SP6 patch 98765.
        (To breake means to produce a result not legitimized
        by the spec, or to produce no result when one is
        expected, at least not in a reasonable time.
        Breakdowns caused by power failure or other
        reasons outside the scope of the spec to be ignored)Well, not worth any further comment.
(2)     Implementations of JSF 1.0 Spec will not brake
        on any faces application when run in a
        WSAD 5.0 container in combination with IE 5.5
        with patch 12345 on Windows NT4 SP6 patch 98765
        (To breake means ...)That is ok... if you have just this combination of components and will never change. Keep in mind that (2) does not promise
anything if you use IE5.5 patch 12345a (and the saying is, there have been several patches to IE ;-)
(3)     Implementations of JSF 1.0 Spec will not brake
        on any faces application when run in any container
        in combination with IE 5.5
        with patch 12345 on Windows NT4 SP6 patch 98765.
        (To breake means ...)Same as (2)
(4)     Implementations of JSF 1.0 Spec will not brake
        on any faces application when run in any container
        in combination with IE 6.0 on any platform with
        any patches.
        (To breake means ...)Maybe what is intended, just impossible to promise. First, such a bold statement about some software can not be proven.
Second, even if you have gained good confidence by thorow testing, who knows which future changes Microsoft will apply to IE 6.0 (via patches, which offen not only correct bugs, but also introduce new and maybe incompatible features)
(5)     Implementations of JSF 1.0 Spec will not brake
        on any faces application when run in any JSP 1.2
        or higher container in combination any broser
        which supports:
           - HTML 4.0.1
           - JavaScript 1.3
           - CSS 1.0
        where IE 6.0 is a browser which supports these
        three standards on any platform with any patches.
        (To breake means ...)It's getting better. But why, for heavens sake, should a JSF specification make a statement about the compliance of IE with HTML 4.0.1, JavaScript 1.3 and CSS 1.0? I think it would be the job of Microsoft to declare the compatibilty of their products with standards (if any applies, whatsoever).
Therefore, lets get rid of these and we have:
(6)     Implementations of JSF 1.0 Spec will not brake
        on any faces application when run in any JSP 1.2
        or higher container in combination any broser
        which supports:
           - HTML 4.0.1
           - JavaScript 1.3
           - CSS 1.0
        (To breake means ...)Nearly finished, just a little bit wordy. It is (nearly) obvious what 'to breake' means. And that it is common wisdom that software breakes, not some specification (Or, as Bertrand Meyer put it, Bubbles don crash. Specifications may be contradictory or fail to describe an algorithm, but never crash). Common wisdom removed gives:
(7)     Implementations of JSF 1.0 need a JSP 1.2 container
        and a browser capable of HTML 4.0.1, JavaScript 1.3
        and CSS 1.0That's it. Specs exist for two reasons. First, to be discussed and thought about. Second, to be implemented. (To the supporters of agile methods: of course you can reverse the sequence ;-). Therefore, specs should be as short and concise as possible, omitting any useless blabla. If the specs are sufficently precise and of moderate complexity, we can even try to prove the compatibility of the specifications. 'COMPATIBLE WITH IE' is advertisement, not tech talk.
Conclusion: Mixing up specifications and implementaions is generally a bad thing. Exception: All sentences where a specification and a implementation are mentioned together should have the form:
         Implementation X does [not] conform to specification Y
---------------------------------------------------------------------------------------------------------I have a follow up question, just inspired by yours. It does not especially concern JSF, but web pages in general.
Some time ago, I stumbled over quite a few web sites with big red eye catchers:
OPTIMIZED FOR INTERNET EXPLORER!!!
And really, they did not look good on Netscape Navigator. Now, my question is, what do you think when you hit such a page? My first thoughts (it's long ago) have been:
    - Oh, nice designers. They really care for the majority of people,
      my neighbor uses an IE.
    - They just skiped the work for other browsers and sell this as an
      optimization. And I didn't get it. Very clever.
    - Poor Guys. Probably they had no other browser at hand.
    - Obviously they had publish this tremendously important information
      under pressure, no time to also optimize for others browsers.Wolfgang Oellinger