The <map-layer> can contain sequences of <map-extent>, <map-feature> and <map-tile> content. This allows the user to turn on/off the layer, and it allows them to turn on/off the <map-extent> as sub-layer elements in the layer control. However, this means that the <map-feature> and <map-tile> elements can't be independently accessed i.e. they're not accessible as sub-layers, and that is an a11y failure I believe.

Should we allow the <map-extent> to have an alternate content model that allows <map-feature>, <map-tile> and maybe <map-extent>? That brings up how deep might it be allowed to nest?

Currently, when a <map-link> element loads a map document, it only supports <map-feature> and <map-tile> and I don't think it would be smart to load and process <map-extent>, due to potentially infinite nesting. Maybe we could make the same injunction wrt <map-extent> i.e. it can't contain another <map-extent>. That would solve the problem of content not being under control of the layer control.

On the other hand, if the nesting is "manual" i.e. not processed blindly by loading a link, maybe allowing some nesting of <map-extent> would be ok? How many levels of nesting is reasonable: 0, 1, 2, 3, n? Maybe we could start off with 0 and see how it works out and prototype our way from there.

wdyt?