The simbols used for operator dealing with nan in floating point expression are
dificult to learn and remember and is easy to confuse.
Programer must use the reverse logic to find the meaning of the operator.
But with a direct logic this problems, IMHO, could be simpler.
The main idea is add the char '?' when dealing with nans

Current table of Operators:

Operator	Greater		Less Than	Equal		Unordered	Exception	Relation
---------------------------------------------------------------------------------------------------
==			F			F			T			F			no			equal
!=			T			T			F			T			no			unordered, less, or greater
>			T			F			F			F			yes			greater
>=			T			F			T			F			yes			greater or equal
<			F			T			F			F			yes			less
<=			F			T			T			F			yes			less or equal
!<>=		F			F			F			T			no			unordered
<>			T			T			F			F			yes			less or greater
<>=			T			T			T			F			yes			less, equal, or greater
!<=			T			F			F			T			no			unordered or greater
!<			T			F			T			T			no			unordered, greater, or equal
!>=			F			T			F			T			no			unordered or less
!>			F			T			T			T			no			unordered, less, or equal
!<>			F			F			T			T			no			unordered or equal

The simbols to change:

New 		Current		Relation
-----------------------------------------------------
?=			!<>=		unordered
>?			!<=			unordered or greater
>=?			!<			unordered, greater, or equal
<?			!>=			unordered or less
<=?			!>			unordered, less, or equal
==?			!<>			unordered or equal

The new table of Operators:

Operator	Greater		Less Than	Equal		Unordered	Exception	Relation
---------------------------------------------------------------------------------------------------
==			F			F			T			F			no			equal
!=			T			T			F			T			no			unordered, less, or greater
>			T			F			F			F			yes			greater
>=			T			F			T			F			yes			greater or equal
<			F			T			F			F			yes			less
<=			F			T			T			F			yes			less or equal
<>			T			T			F			F			yes			less or greater
<>=			T			T			T			F			yes			less, equal, or greater
?=			F			F			F			T			no			unordered
==?			F			F			T			T			no			equal or unordered
>?			T			F			F			T			no			greater	or unordered
>=?			T			F			T			T			no			greater, equal, or unordered
<?			F			T			F			T			no			less or unordered
<=?			F			T			T			T			no			less, equal, or unordered

The are motivation for this change are:

a) The same operator as integer expressions are used for normal comparison (call
this basic operators):

== != > >= < <=

b) When the operator contains < or > and one of the one of the operands is a NAN
then a exception is raised.
c) To deal with NAN you add the '?' after the normal operator
d) There are two more operators: <> and <>=

This rule is simpler, IMHO.

Comments ?

I know they're a bit complicated. They do match, however, the NCEG proposed extensions to C (which were never adopted). Your idea has merit, I need to think about it.

<bono@u2.art> wrote in message news:bhtfsr$1sgt$1@digitaldaemon.com...
> The simbols used for operator dealing with nan in floating point
expression are
> dificult to learn and remember and is easy to confuse.
> Programer must use the reverse logic to find the meaning of the operator.
> But with a direct logic this problems, IMHO, could be simpler.
> The main idea is add the char '?' when dealing with nans
>
> Current table of Operators:
>
> Operator Greater Less Than Equal Unordered Exception Relation
> --------------------------------------------------------------------------
-------------------------
> == F F T F no equal
> != T T F T no unordered, less, or greater
> > T F F F yes greater
> >= T F T F yes greater or equal
> < F T F F yes less
> <= F T T F yes less or equal
> !<>= F F F T no unordered
> <> T T F F yes less or greater
> <>= T T T F yes less, equal, or greater
> !<= T F F T no unordered or greater
> !< T F T T no unordered, greater, or equal
> !>= F T F T no unordered or less
> !> F T T T no unordered, less, or equal
> !<> F F T T no unordered or equal
>
> The simbols to change:
>
> New Current Relation
> -----------------------------------------------------
> ?= !<>= unordered
> >? !<= unordered or greater
> >=? !< unordered, greater, or equal
> <? !>= unordered or less
> <=? !> unordered, less, or equal
> ==? !<> unordered or equal
>
> The new table of Operators:
>
> Operator Greater Less Than Equal Unordered Exception Relation
> --------------------------------------------------------------------------
-------------------------
> == F F T F no equal
> != T T F T no unordered, less, or greater
> > T F F F yes greater
> >= T F T F yes greater or equal
> < F T F F yes less
> <= F T T F yes less or equal
> <> T T F F yes less or greater
> <>= T T T F yes less, equal, or greater
> ?= F F F T no unordered
> ==? F F T T no equal or unordered
> >? T F F T no greater or unordered
> >=? T F T T no greater, equal, or unordered
> <? F T F T no less or unordered
> <=? F T T T no less, equal, or unordered
>
> The are motivation for this change are:
>
> a) The same operator as integer expressions are used for normal comparison
(call
> this basic operators):
>
> == != > >= < <=
>
> b) When the operator contains < or > and one of the one of the operands is
a NAN
> then a exception is raised.
> c) To deal with NAN you add the '?' after the normal operator
> d) There are two more operators: <> and <>=
>
> This rule is simpler, IMHO.
>
> Comments ?
>
>
>
>