🚀 go-pugleaf

In article <561@halley.UUCP> rrt@halley.UUCP (Robert Teisberg) writes:
>In article <545@brazos.Rice.edu> preston@titan.rice.edu (Preston Briggs) writes:
>>In Pascal, < doesn't associate.  You can't write a<b<c.
>
>True, you usually can't, but associativity (or lack thereof) is not the
>reason.  In Pascal, the expression parses as:
>
>(a<b)<c
>
>where (a<b) is an expression with a boolean value.  If c is also
>boolean, this is a legitimate (though weird) expression.  If c is not
>boolean, the compiler should complain of a type clash.

Wait.  Check the BNF or railroad tracks.  (Pause while I try
to find a reference!)  In "Algorithms+Data Structures" (all I
can find right away), an Expression is defined:

	Expression ::= SimpleExpression [relation SimpleExpression]

where [x] denotes an optional "x".  So, an expression is a
SimpleExpression, optional followed by a Relation and another
SimpleExpression.  There can't be more than one relation.

Contrast this with the definition of SimpleExpression:

	SimpleExpression ::= [`+'|`-'] Term {AddOp Term}

where {x} denotes 0 or more x's.  Here we see an optional
leading + or -, followed by a Term, followed by a sequence
of AddOp's and Term's.

It's possible that some standard has changed this definition in the past
few years; I haven't kept up.  Try it on your favorite compiler!

	Preston