On Fri, 30 Nov 2001, Peter Crowther wrote:

> > From: Lex Spoon [mailto:lex at cc.gatech.edu]
> [...]
> > Overall, Multiple-Windows seems like a safer default.
> 
> I agree, but maybe it should be exactly that --- a default, with navigation
> buttons appearing on browsers if a developer decides to change that default?

I read this and it made me think what I came to think you said below. :)

> 
> A more interesting question is: what should the interface do to support
> *both* behaviours?  For example, sometimes I know how to get where I'm going
> and want to navigate through several layers quickly and without retaining
> intermediate context (senders and implementors are typical).  I'd be quite
> happy to shift-click to have the system present the next layer in the same
> window (or, for simplicity, to throw away the existing window and replace
> it).  At other times I'm exploring, and really appreciate the ability to
> retain context in the windows.

But I'm not longer sure that I think you said what I thought above.

One quick solution to the "history vs. multi windows" debate is to
implement the history with multiple windows (hey, they're just objects,
and in morphic, just morphs!). Instead of staggering new windows, stack
them exactly (better, make the original window set the bounds and
"contain" all the subwindows). Maybe have a bookmorph navigator up top.

Or just use a bookmorph.

Hmm. A quick dork reveals that I don't know much about bookmorphs, being
unable to drag a window into one :) or even resize it.

In any case, that would certainly be a simple to implement history
mechanism that would work across Browsers of various sorts.

Cheers,
Bijan Parsia.