In article <[EMAIL PROTECTED]>, [EMAIL PROTECTED] (George J. Carrette) writes: > In article <206a8q$[EMAIL PROTECTED]> , [EMAIL PROTECTED] (Brian Harvey) writes: >> An absolutely perfect implementation would be cleverer about what >> to do with typed-ahead characters when the mode switches, but >> in such interactive programs, the computer is usually faster than >> the person typing, anyway, so it rarely comes up. > Really? So how come 20% of the time I log into a Unix timesharing > system, even a state-of-the-art-multi-cpu SPARC system, it looses the > first few characters of my password, and even echoing some after > having typed my username? > > Welcome to THE WORLD > Login as 'new' if you do not have an account > > login: gjc > dPassword: > Login incorrect > > This has been happening to me for at least 15 years. I remember that > at one time in the early 1980's the famous Chris Terman (he did the > PCC port that SUN Microsystems used to get off the ground) tried to > fix it but gave up when it seemed to require major restructuring. This seems to be a good counter-example to the statement in Appendix D (I/O Semantics) that "user' is a _strict_, but otherwise arbitrary, function modelling the user." (Of course, a semantics for user' is left as an open problem for the cognitive science community.) --eric ----------------------------------------------------------------------------- "The commas in writing S-expressions may be omitted. This is an accident." --Lisp 1 Programmer's Manual, March 1960.
