Re: GOTO considered necessary (reworked)

[dewar@merv.cs.nyu.edu (Robert Dewar)]

Jeff Carter says

<<My point on all cases was that goto introduces an increased likelihood
of error on modification, and all of these cases had an alternative that
was as clear as the goto version, but eliminated that increased chance
of error. Of course, at some point clarity becomes a taste issue, so you
may claim that my versions are not as clear as the versions with goto,
IYNSHO.>>


This is exactly the kind of absolute statement that I regard as fanatic.

It is NOT the case that goto always introduces an increased likelihood
of error on modification, and it is NOT the case that the alternative
to using goto always has the same clarity.

Indeed the whole point is to identify cases, like the <<CONTINUE>> case
where goto is easier and clear from a maintenance point of view.

And it is of course the case that the general case of CONTINUE of
course requires Boolean variables to be introduced to eliminate the
gotos (this is as I mentioned before, a well known fact, well treated
in the literature).