- From: Joe D Williams <joedwil@earthlink.net>
- Date: Fri, 29 Jan 2010 15:36:37 -0800
- To: "Leif Halvard Silli" <xn--mlform-iua@xn--mlform-iua.no>
- Cc: "Jonas Sicking" <jonas@sicking.cc>, <public-html@w3.org>
> Of course, what HTML4 doesn't mention is that the object can also be > instantiated by removing the @declare attribute ... Well, then you would loose your reference and all the childs would have to die. @declare changed the tree into a graph. The 'same' child could have any number of parents. I might have been way off, but it was @declare='string' then the object to be rendered was actually coded as '#string' in an <a>. So, the same child could have many parents. The concept is was wild. @declare was independent of @id. I wanted to learn about that new to html thing then because x3d (then vrml) did similar; add a DEF to a node then a duplicate node can be made active anywhere by USE. On the topic of the idea of putting some data in there that is intended to be hidden and not rendered, but just informative for the author or consumer in some hidden way. >>>>>>> The hidden attribute is meant to mark a DOM subtree pruned >>>>>>> from all presentations on all media. I am reaching some due to the <iframe with @sandbox and <sandbox></sandbox> discussion so why is this an attribute? Aren't there enough of them?. I mean we need the right number, but... In looking at all of current browser behaviours isn't there an existing way to cover this use case? Simple typos sometimes cause this. If "hidden" is needed, then why not a way to declare it in advance. It doesn't sound reasonable to expect the UA to do whatever if I add/remove that attr. so why not an element? The head could contain a declaration of this category that says this element and its children represent the kind of data we are discussing in this topic and it is mystuff, so UA don't concern yourself with it. <head> <mystuff [space separated strings representing name.id containers that are to be hidden as quoted above] /> ... </head> <mystuff> is new. Any element name.id in that list is guaranteed to be not making it into the render/dom builder or client viewsource(?). So maybe it is the idea that this authors directive that produces some behavior should be included as an element with values and maybe content that defines some behavior, not as a simple attribute. Or, in the body <hidden> <htmlstuffthatshouldnotbeincluded... > </hidden> That way would make it more apparent that the UA should just skip this entire container of stuff because it is really not pertinent to any application of the user. No client choices on this. Also this construction would make it less likely for an author to abuse this feature thinking it should be live. Sorry, you can't bring it back because it was never really there. Thank You All and Best Regards, Joe
Received on Friday, 29 January 2010 23:38:02 UTC