Re: Unary operator after binary operator: legal or not?

[Adam Beneschan <adam@irvine.com>]

On Jul 30, 3:52 pm, "Jeffrey R. Carter"
<spam.jrcarter....@acm.nospam.org> wrote:
> Here's something confusing I encountered:
>
> with Ada.Strings.Unbounded;
>
> function Xyz (Left, Right : in Ada.Strings.Unbounded.Unbounded_String)
> return String is
>     function "+" (Right : in Ada.Strings.Unbounded.Unbounded_String)
>     return String renames Ada.Strings.Unbounded.To_String;
> begin -- Xyz
>     if +Left <= +Right then
>        return +Left & +Right;
>        --            ^ Error reported here.
>     end if;
> end Xyz;
>
> A compiler reports "missing operand", referencing the space after the "&".
>
> Why then does it accept the comparison in the previous line? Both are
>
> +Left [binary operator] +Right

I think that was a deliberate decision.  See the BNF in 4.4.  The
operands of "and", "and then", "or", "or else", "xor", "in", or
relational operators are <simple_expression>.  The definition of a
<simple_expression> is

  [unary_adding_operator] term {binary_adding_operator term}

A unary_adding_operator is part of the syntax of <simple_expression>
but not of <term>, so the consequences is that the right operator of a
binary_adding_operator (or a multiplying_operator, or "**" or "abs" or
"not") can't start with a unary adding operator, unless you
parenthesize it.  But the right operand of a relational operator or
one of the logical operators I listed above *can* start with a unary
adding operator.

                         -- Adam