i haven't followed the entire discussion, but how about adding a method boolean getStripXmlHeader() to Page, returning false by default and overriding it for the default error pages?
the downside to this approach is, that custom users would have to do that for custom error pages. --> add a new flag (or use a free "preserved" one) to store the stripXmlHeader data and call setStripXmlHeader(true) whenever an errorpage is instantiated. wdyt? Gerolf On Dec 28, 2007 12:26 PM, C. Bergström <[EMAIL PROTECTED]> wrote: > > > > > Right now the problem why error pages doesn't work with IE really is > > the comment between <xml declaration and doctype. I've tested it. > > > > > > > > > > > > > We should focus on the problem itself, and that is the > > > > Apache header between the <?xml declaration and doctype which is > what > > > > completely breaks IE. > > > > > > I'm not sure this is true. Anything, including the xml decl, before > > > the doctype makes IE switch into quirks mode. > > Yes, but for our error pages we can either remove the xml declaration > > or not care that it switches IE into quirksmode. Quirksmode is not the > > real problem here. The problem is that right now IE doesn't show the > > error pages at all, quirksmode or not. > > > > > > > > > > > Now I can see two simple solution. Either wraps the header comment > > > > with <wicket:remove> or move the header comment after doctype. > > > > > > I'm not sure this will work. Did you try it already? > > Yes, I've tried both. They both work and IE shows the pages properly. > > > > If we really want the behavior when you add an <xml header> to file > > and then don't want to show it in ouput, we should have a way to > > configure that per file, such as <wicket:stripXmlHeader/> somewhere in > > the markup. > > I definitely don't like the idea of putting 'configuration' in the > markup. However, the idea to configure it per webpage is sound. > >
