FOR NEXT behavior

It seems to me that there are five possible ways in which an
implementation of BASIC might behave in the event of the values in a
FOR statement being 'inconsistent', any of which could be argued to be
'valid'.  In the case of the statement 'FOR i = 3 TO 2' these are:

1.  Don't execute the body of the loop at all, and leave i unchanged.

2.  Don't execute the body of the loop at all, but set i to 3.

3.  Execute the body of the loop once, with i = 3.

4.  Execute the body of the loop twice, with i = 3 and i = 2.

5.  Re****t an error.

Personally the one which I consider to be nearest to the 'common
sense' expectation, and therefore the one I prefer, is (4) i.e. infer
that the step must be negative.  Given that BASIC statements are
supposed to be 'English like', and remembering what the 'B' of BASIC
stands for, this seems best to reflect the 'spirit' of the language.

Failing that, I think my next preference would be (5) - re****t an
error.  Since arriving at that situation is quite likely to be
unintended (e.g. a typo) or indicate sloppy thinking (e.g. failure to
anticipate that it might occur) drawing attention to it seems
sensible.

Despite them representing what the vast majority of 'real' BASIC
implementations do, I don't really feel any of (1) to (3) can be
justified from a 'logical' viewpoint.  (3), which is what BBC BASIC
does, is the simplest to implement and has the fewest unexpected side-
effects but that's about all that can be said for it.  (1) and (2) are
probably quite often 'useful' (that is, they may simplify code) but
they fail the 'common sense' test quite badly.  Also, the very
ambiguity over whether the control variable should be changed
illustrates a flaw.

Richard.
http://www.rtrussell.co.uk/
To reply by email change 'news' to my forename.