a+ b performs (juxtaposition) multiplication, not addition

[djb]

The following program

export Executable run(args:String...) :() = do

a = 3 b = 10 println ("a=" a ", b=" b) c = a+ b println ("a+ b=" c) d = a + b println ("a +b=" d) e = a+b println ("a+b=" e)

end

produces the output

a=3, b=10 a+ b=30 a +b=13 a+b=13

It appears that the expression

a+ b

is doing multiplication due to juxtaposition of the "expression" (a+) and (b). However, I cannot find a reference to a postfix + operator in the LRM. Sorry if it is there and I missed it. However, if intentional, this does feel counterintuitive.

I'm using code from Subversion, updated todat (Oct 31, 2007)

[djb]

source program

[djb]

sorry for the unformatted source in the original report. Use the attached source.

[dr2chase]

It's highly likely that this is an artifact of bug #16 -- "a+ b" is probably parsing as

(juxt  (postfix+ a) b)

but then the postfix+ is identical to prefix+, which is identity.

[jmaessen]

Bug #16 is indeed part of the problem. However, it is at best dubious to parse in this way; if we have a known lower-precedence operator used in prefix or postfix form within a juxtaposition, we probably ought to reject. The interesting question is how we distinguish, eg, prefix + and - from (say) postfix ! so that we can still write k! (n-k)! / n! and the like.

[sukyoungryu]

In Fortress, "a +b" is a static error and "a+ b" is parsed as a loose juxtaposition of "a+" and "b" and "a+" is parsed as a postfix operator application. The asymmetry in handling "a +b" and "a+ b" is intentional because it is often to use an operator as both infix and prefix but not as both infix and postfix. For example, "-" is often used as both infix and prefix operator.

Now, this ticket is reduced to Ticket #16. I'm closing this ticket.