[erlang-questions] map elements in defined order

Richard A. O'Keefe ok@REDACTED
Mon Oct 30 01:09:18 CET 2017

On 28/10/17 2:26 AM, Sverker Eriksson wrote:
> Yes. All Erlang terms, including maps, have a globally defined,
> implementation independent, total order.
> The reason, for this "surprising" order of maps, is the alternative of
> using the normal arithmetic order for keys is much worse.
> If we want 1 and 1.0 to be different keys (which we do as we use
> matching (=:=) to lookup keys)
> then we need to order 1 and 1.0, and normal arithmetic term ordering
> does not (1 == 1.0).

The fundamental mistake here is confusing *arithmetic* ordering
with *term* ordering.  In *term* ordering, if X and Y are
behaviourally distinguishable then either X < Y or X > Y should
be true.  Since integers and floats are behaviourally
	1> X = 1 bsl 53.
	2> Y = float(X).
	3> X == Y.
	4> X+1 == Y+1.
it follows that either X < Y or Y > X should be true in
*term* (but not *arithmetic*) order.

Arithmetic order in Erlang has some seriously weird issues, like
you can find numbers X Y such that X - Y is 0.0 but X == Y is false.

> It has been discussed (more and less serious) to expose this
> "map-key-order" with operators like :<, :>, =:<, >:=.

Surely we deserve *some* sane ordering in Erlang?

Prolog had two sets of comparison operators: term ordering and
arithmetic ordering.  There were *reasons* for that.

More information about the erlang-questions mailing list