You will note, tho, that many of the GUI forms of systems are really
built out of clever interfaces on top of command-line stuff. Much of
the catchy World Wide Web stuff (the PARC map of the world, the
various archie servers) and even commercial stuff (GENIE "Aladdin",
Compuserve "CIM") are at their core really only just someone taking a
command-line interface and making it into a GUI interface.
Dan Farmer had an interesting "works in progress" paper at usenix in
Cincinnati where he talked about taking the "G" out of "GUI" and
separating the design cleanly between "Graphical" and "User
Interface". His key comment was that GUIs tend to be built where the
graphical stuff has seeped quite deeply into the underlying code, and
thus it's hard to pull off the top layer and add on a new or different
look.
For on-line documents the similar observation would be that we are
making a lot of assumptions about the viewer that people are using
when we go to design a text to be read on the screen, and if those
assumptions are wrong the result it going to be icky. Consider all of
the "texinfo" screens aimed at a 24x80 (or at best at 60x80) vt100
style screen and a nicely laid out typeset manual; is the proper
rendering into World Wide Web more book-like or more Info-like?
Or should it be both? Do you expect people to read and browse with
the pageup/pagedown keys or the left/right/up/down arrows, and how
does the availability of bitmap fonts change that?
--Ed