Just a few quick notes, i will be away today ... To implement XHTML2 will affect many parts of our operation, all the way through to the output stages.
It would be ideal if we can develop that in parallel so that existing work continues to function, balanced with the need to get this implemented ASAP so that we don't continue to make workarounds. In some ways our existing sitemap pipelines need to be rewritten. They have grown topsy turvy over the years. Not the whole sitemap, just central parts like the "skinit" and the map:aggregate stuff which carries around unnecessary info like the whole skinconf. On the other hand, i am concerned that it is going to be difficult to re-address some of the problems that have been solved. I am not discussing much about views here, basically because it is ahead of my comprehension at the moment, though i am improving. I know that XHTML2 will affect that. They are both to be part of the core and should be invisible. We don't even need to give "views" a name, because that is just how the machinery will operate. So i do think that we need to take the actual design of views into account. Thanks to the others for their various visions. I reckon that we do need to redevelop the main internal parts, not because it is needed by xhmtl2 or views, but because there must be a simpler way to build all that structure. Doing that new structurer using xhtml2 seems ideal. Using *.xhtml2 map:match patterns, will (i hope) enable the internal xdoc stuff to continue. Although perhaps we do need a switch in forrest.properties to start with. If we can leave views out of the initial implementation then that would be good, i.e. one step at a time. Perhaps the initial pipeline will just place the menu and tab information at the bottom/top of the page. Start with xhtml2 and html input plugins, an internal plugin, some filter plugins to add menu and tab content nuggets, then a raw html output plugin with a simple CSS. Concentrate on enabling SourceTypeAction and Locationmap stuff. Then add views to this, one piece at a time. That is what i want to resolve in our IRC discussion, what is the best way to approach this. -David
