OK here is what happened. One 3/20/05 matzew updated UIComponentBase to *correctly* disallow characters such as ':' in the *component* id. This was a good thing and brings us into compliance with the spec. The inputSuggest component has not been tested since then so that is why we are now having problems.
As Craig mentioned, its permissable to to have a ':' in the *clientId* and you will have that often because of naming characters. The actual id of the component, however, cannot contain such characters. There were a few hidden fields being generated by the component and those were incorrectly based of the clientId instead of component id. Thanks for all of feedback. I'll check in the fix shortly. sean On 7/1/05, Abrams, Howard A <[EMAIL PROTECTED]> wrote: > > > > Read Craig's reply again. It's not that colons are not allowed in *HTML, > it's that they are use by the naming container to separate the container's > id from the component's id. The HTML generated by JSF Renderers will have > colons in it. > > > > > ________________________________ > > > From: Matt Blum [mailto:[EMAIL PROTECTED] > Sent: Friday, July 01, 2005 12:35 PM > To: MyFaces Development; [EMAIL PROTECTED] > Subject: Re: Could use some help debugging sandbox > > > > > But post-initial colons are supposed to be acceptable in XHTML, according to > the spec. > > http://www.w3.org/TR/xhtml1/#C_8 > > So why aren't they permissible in JSF? > > -Matt > > > On 7/1/05, Craig McClanahan <[EMAIL PROTECTED]> wrote: > > On 7/1/05, Grant Smith <[EMAIL PROTECTED]> wrote: > > Hmm.. Does this mean that section 3.2.2 of the spec directly contradicts > > section 3.3.1 ? > > > > Craig ? > > > > There's a difference between the value of the *id* attribute in your > source page (i.e. what 3.1.1 talks about) and the value that gets > generated in the rendered output (which is what 3.2.2 talks about). > The latter value is the *clientId* of the component, not the id, and > therefore can include ":" characters when there is a NamingContainer > involved. > > Note that the reason for having any restrictions in the first place is > based on the fact that, in XHTML environments, the "id" attribute is > declared to be of type ID, which imposes essentially the same set of > restrictions of you want your source page, or the rendered output, to > validate. > > Craig > > > > > > Bruno Aranda wrote: > > > > >Hey, stop, the colon is what the NamingContainer uses, as stated in > > >section 3.2.2 of the spec. > > > > > >'NamingContainer defines a public static final character constant, > > >SEPARATOR_CHAR, that is used to separate components of client > > >identifiers, as well > > >as the components of search expressions used by the findComponent() > method see > > >(Section 3.1.8 "Component Tree Navigation"). The value of this > > >constant must be a > > >colon character (":").' > > > > > >So, better to keep it ;-) > > > > > >Regards, > > > > > >Bruno > > > > > > > > > > > >
