> A little history on vi (and vi historians can correct
> me here), the intent was that the command interface
> needed to use keys that existed on every keyboard. All
> the letters are obviously there. By not using
> backspace or delete or any of the other stuff you
> don't find on every keyboard (we are pre-dating the PC
> era here). That way, problems like you mention
> wouldn't happen. =


Sort of, but that wasn't the motivation, exactly.  Vi was first written =

for the ADM-3A "dumb" terminal.  The arrow symbols on that terminal =

were on the fronts of the HJKL keys, accessed via a special shift key.  =

But the author of vi didn't like using the special shift, so he made =

the letters act for the arrows in command mode.

> However, later versions of vi incorporated more keys
> so they work if your mapping is right. If it's wrong,
> just fall back on the way vi was "meant" to be used!

In insert mode, vi always obeyed whatever character was designated as =

"erase."  Typically, that was ^H, but many people used ^?.  This did =

not carry over to command mode.

Support for terminals other than the 3A came a little later, with the =

advent of termcap.  If I remember correctly, termcap was already in use =

when the first outside release of vi was made, with 2BSD (1BSD was just =

the Pascal system; 2BSD included Pascal and other userland tools - I =

don't recall if it included a kernel or not; 3BSD was the first =

complete Unix system BSD release, it was for the PDP-11; 4BSD was a =

complete system release for the VAX, and eventually other virtual =

memory machines.  At the beginning, all of them - with the possible =

exception of 1BSD - required a Unix source license from AT&T).

	--Ed