On 01.07.2004 10:32, Colin Paul Adams wrote:
You are really nagging on this issue, aren't you? ;-) http://marc.theaimsgroup.com/?l=xml-cocoon-cvs&m=108862151322458&w=4
Drip. Drip. Drip. Look - a hollow's appeared in that stone! (no hole yet, though)
:)
I checked out all the samples - since everyone of them is using the Transitional DTD, that's not much of a test for your claim to adherence to the strict DTD. Inspecting the source code by hand suggests it might well be a valid claim.
Do you believe me that I can configure my serializer so that it outputs the strict document type declaration? ;-)
I used the w3c validator for the tests.
There are one or two of the samples do not completely validate (ignoring the xmlns:fi issue), and the forms-gui one is nothing like (but I think you are already well aware of that).
Cocoon's part of the work were the stylesheets, not the templates - though it should also set a good example. The aggregate field template was a bit more complex and I was lazy then. For the Forms GUI sample it seems the FormsTransformer must be fixed. How I can fix the empty select element without breaking the double-listbox - I don't know.
Have you looked at all the additional comments I made to bug #29854 yesterday? I surmise that this xmlns:fi issue is probably the same bug (or at least, closely related to it - in any case it is produced by the FormsTransformer).
If you refer to the pure additional namespace declaration: no, it is much easier. The elements that have that namespace at the end were copied from the template into the output - and they are always copied with all their namespaces. This is a correct behaviour. To fix it you have to add a namespace clearing XSLT at the end that uses xsl:element instead of xsl:copy.
Visual inspection also shows one other thing that would be a problem for xhtml validation, and that is method="POST" rather than method="post".
Good to know, I have not been aware of this. Maybe I should go one step further today to XHTML 1.0 strict ;-)
Joerg
