Definitions for "viewport"

I think there are problems with the definitions of viewport and top-level viewport. They're complicated and pretty confusing in many respects.

These currently read

    viewport: The part of an onscreen view that the user agent is presenting onscreen to the user, such that the user can attend to any part of it without further action (e.g. scrolling). There may be multiple viewports on to the same view (e.g. when a split-screen is used to present the top and bottom of a document simultaneously) and viewports may be nested (e.g. a scrolling frame located within a larger document). A container can hold multiple viewports at the same nesting level. When the viewport is smaller in extent than the content it is presenting, user agents typically provide mechanisms to bring the occluded content into the viewport (e.g. scrollbars).

    top-level viewport: A viewport that is not contained within another user agent viewport.


1. I think they should do more to clarify whether or not a browser's window is a viewport, or whether only the areas within it that render content are viewports. That also affects whether, in a browser that uses tabbed panes to display multiple documents or views of the same document, whether these tabbed panes are top-level viewports. 1.8.13 says "top-level viewports (e.g. windows or tabs)", but since browser windows typically contain user agent user interface as well as rendered content, this only works if viewports also include browser windows; if they don't, then this occurrence would be wrong and need to change.

2. I find the phrase "The part of an onscreen view that the user agent is presenting onscreen to the user" confusing, particularly in light of the definition of "view". That is, is it a portion of the rendered content or presented content source, or is it the pane or bounded region used to display those? It is the pane (or container) that the user resizes, not the content. The "part of an onscreen view that the user agent is presenting onscreen to the user" may only takes up a little space in this pane, but it's the pane which has controls to let the user resize it (e.g. drag handles), not the content inside the pane.

3. It uses the term "container" which is not defined, and I don't think used in a universally-understood sense. The document sometimes uses "container" to mean a component of the user agent user interface, such as a toolbar, and here it seems to mean a component of the user agent user interface such as a browser window, but in 2.1.1 it's used as a synonym for viewport.

4. This definition of viewport is limited only to "onscreen" presentations, but elsewhere don't we talk about non-visual viewports? If we didn't, why would we so often say "graphical viewports"? Views are defined as being "A view can be visual, audio, or tactile", and what would the non-visual ones be presented in if not a viewport?

5. If one viewport has scrolling, then strictly according to the definition any viewport nested inside it would cease to be a viewport once it's scrolled out of view, because the definition says "such that the user can attend to any part of it without further action (e.g. scrolling)". Well, it's scrolled out of view, so you have to use scrolling before you can attend any part of it. (If it's scrolled so it's only partially visible, then only the visible portion would seem to be viewport.

     Thanks,
     Greg