Both the behavior and a possible interpretation of the man pages (both Solaris pseudo-ksh88 and early ksh93, i.e. the ksh93 man page that came with CDE, to be read together with the dtksh man page) lead me to believe that -o viraw only modifies -o vi, and isn't an editor mode in itself. By contrast, emacs and gmacs are both clearly identified as editing modes, although they only differ by the way they handle ^T. If emacs, gmacs, and vi mode are all off, setting viraw alone does not give one editing capability. In that case, it's just "remembering" that until you turn of [eg]macs mode and turn back on vi mode again, at which time viraw would still be set. Bug or feature? Also, apparently at one point at least, "emacs", "gmacs", and "vi" were all actual editor names (I sure haven't seen gmacs around anytime recently, only emacs and xemacs; but then I'm not a huge fan of any of the *macs editors, although I don't question their power). Anyway, the man page(s) also say that if EDITOR or VISUAL end in "emacs", "gmacs", or "vi", the corresponding command-line editing mode is set. For darn sure there's never been a stand-alone editor called "viraw" (unless someone made one just as a gag). So all in all, I suppose it's a feature, and made sense at the time. Even if it makes less sense now, changing it would probably cause problems with the expectations of established ksh users (as opposed to crossover users of bash or other sh or ksh clones, about which the less said, the better).
At least, that's my read. See also the wild notion I threw out about adding a framework for loadable command-line editors, which would let people do whatever they wanted without burdening the ksh maintainers with the foolishness of editor holy wars. I'd like to think that if such a notion were feasible, the existing editing modes would become builtin special cases, with some new option to enable a loadable command line editor instead. This message posted from opensolaris.org
