Re: Bookmarking the frames.....
David Seibert wrote:
> > [i wrote]
> > I don't think you need the identity of every frame, nor do you need
> > every URL involved. You only need the initial URL, plus the ID's
> > URLs for the frames that have changed. All other information -
> > including dimensions/positions/relationships - can be derived from
> > URLs.
> The other information could, in principle, be derived from the URLs.
> In practice, it's much easier if the browser saves that information
> when it bookmarks the document, as the information is available then.
The most efficient way to reproduce the window state would be as you
propose. It wouldn't be the most efficient way of storing or
publishing it. I was trying for the most efficient way to represent the
state as an extended URL that could be recognized by any browser. It
seems to me that representing the frames as array elements and URLs in
an extended address fragment is the most compact representation of the
window state. A non-supporting browser would simply load the initial
document - no worse than now. You are right that additional http calls
would be necessary, but most framed windows are not so complicated as
to make this a big issue. The UA need only track the last document
loaded into any frame. Because of the hierarchical nature of frames,
that last document loaded into a frame will also change the state of
any nested frames. If the UA knows that frames is changed, there
would be no reason to retrieve the initial document. There is really
far less overhead to this scheme than initially appears.
The notation I suggested definitely needs refinement.