Re: $(( )): binary/unary VAR/NUM inconsistency

[Robert Elz]

    Date:        Fri, 8 Jul 2022 12:08:38 -0400
    From:        Chet Ramey <chet.ramey@case.edu>
    Message-ID:  <e238f3f9-6aa9-627b-93c3-a5bb673ea650@case.edu>

  | This is where folks like kre are going to argue.

Am I?   I was keeping out of this one!

As long as valid expressions are parsed correctly, I don't think it
really matters all that much what you do with the nonsense.

I don't think it is really helpful to anyone to donate a meaning to
nonsense input, but the input really should not be nonsense in the
first place, so it also isn't all that harmful.

About the only other thing I'd say (more than 30 years too late) is that
even when the ++ and -- operators weren't implemented, the tokenizer
really should always have recognised them - they were always part of
the C arithmetic syntax, and while POSIX doesn't require that shells
implement those operators, that doesn't really excuse failing to know
they exist.   Had that been done (all those years ago) the backward
compat issue never would have arisen, as all those +++++ expressions
would have been rejected from day 1. 

kre