I'm sure this is already known by most of you, but a work-a-round is to wrap the outer-most content that you want sent to the browser in a <ui:composition> tag. Not the cleanest solution, but it does the trick: <?xml junk...> ... <ui:composition> Real content here </ui:composition> ... +1 to making this configurable option. Ken Dan Allen wrote: > Max (Oracle) brought up a critical point at JSF Summit (and earlier in > private discussions). Facelets is passing through the XML declaration in a > template to the browser. This is problematic for certain browsers, to the > point that it can change the rendering behavior of the browser. There needs > to be a way to suppress this. A stopgap solution is to introduce a context > param. A more long term approach is to add an attribute to f:view that > indicates whether the XML declaration should be sent to the client (you may > want it for an atom feed for example). > > I'll let Andy chime in on Max's behalf for follow up, since he has some > use cases he can cite and perhaps further suggestion. > > -Dan > > -- > Dan Allen > Senior Software Engineer, Red Hat | Author of Seam in Action > Registered Linux User #231597 > > http://mojavelinux.com > http://mojavelinux.com/seaminaction > http://www.google.com/profiles/dan.j.allen >

Oh, THAT's what he meant. He tried to describe the problem to me, but our discussion got cut off. I've even run into this with demos - it's actually a pretty big deal, I think.

Fortunately, the declaration is optional, as I understand the XML standard.

As for "some browsers", let's just name it: IE. The XML declaration triggers their "I don't care about your mime-type, I know better than you" code. Though it wouldn't be surprising if it triggered some quirks-mode in other browsers.

Oh, I forgot one use case. Max was saying that they need to use CDATA in their templates, but the CDATA should not be sent to the browser. So there is the same problem with CDATA as with the XML declaration. -Dan On Fri, Dec 11, 2009 at 6:38 PM, Dan Allen <dan.j.allen(a)gmail.com&gt; wrote: > On Fri, Dec 11, 2009 at 6:30 PM, Jim Driscoll <Jim.Driscoll(a)sun.com&gt;wrote: > >> Oh, THAT's what he meant. He tried to describe the problem to me, but >> our discussion got cut off. I've even run into this with demos - it's >> actually a pretty big deal, I think. >> > > +1 > > >> >> Fortunately, the declaration is optional, as I understand the XML >> standard. >> > > Right, but from what I understand from Max, tooling is adding it > aggressively and it becomes a fight between developer and tool to leave it > off. He could probably speak to this more. > > >> >> As for "some browsers", let's just name it: IE. The XML declaration >> triggers their "I don't care about your mime-type, I know better than you" >> code. Though it wouldn't be surprising if it triggered some quirks-mode in >> other browsers. >> > > Yep, pretty much. > > -- > Dan Allen > Senior Software Engineer, Red Hat | Author of Seam in Action > Registered Linux User #231597 > > http://mojavelinux.com > http://mojavelinux.com/seaminaction > http://www.google.com/profiles/dan.j.allen > -- Dan Allen Senior Software Engineer, Red Hat | Author of Seam in Action Registered Linux User #231597 http://mojavelinux.com http://mojavelinux.com/seaminaction http://www.google.com/profiles/dan.j.allen

On 12/11/09 3:45 PM, Dan Allen wrote: > Oh, I forgot one use case. Max was saying that they need to use CDATA in > their templates, but the CDATA should not be sent to the browser. So > there is the same problem with CDATA as with the XML declaration. > I commented directly on the bug, but here's my thoughts. The CDATA case is actually considerably harder than the XML declaration. Typically, you either want to serve the XML declaration application wide, or you don't. Generally, because of IE, you don't. I would argue that we should make that the default case. If IE ever gets it's act together, or folks start using XML v2.0, then we can revisit the decision - that should last us for at least 5 years, anyway. With CDATA, there will be times when you want to serve CDATA blocks to the browser (most notably, to wrap scripts). Apparently, there will also be times when you don't (though the example of a custom script tag is not compelling to me - we handle that in Mojarra's implementation without incident, though it's a painful block of code to read). There will therefore be times when you both want to serve CDATA blocks, and when you do not, possibly in the same page.