>There's a programming construct I've thought of, and I've been searching on
>for it for a while, and haven't found any ideas on why it has/hasn't been
>implemented in any language.
>
>It's fairly obvious...

> What might be interesting would be the capability of manipulating
> order relationship statements in much the same that one can
> manipulate boolean relationships. There is probably a whole body
> of theory out there that I don't know anything about. As an
> example, consider triplets of the form (x,y,v) where v is the
> order relationship between x and y. Then if we have two triplets
> (x,y,u) and (y,z,v) and we know that u=v then we also know that
> (x,z,v).

> On Thu, 17 Jul 2008 11:57:34 -0400, "S James S Stapleton"
> osu.edu> wrote:

>>There's a programming construct I've thought of, and I've been searching on
>>for it for a while, and haven't found any ideas on why it has/hasn't been
>>implemented in any language.

> Then there is the question of what the comparison yields.  In C
> strcmp yields -1, 0, and 1. However this is a hack.  Decent
> languages have a native boolean type with values true and false;
> one could consider having a native three valued type with values
> lt, eq, and gt.

> This is sort of interesting, to me anyway, but I perceive it as a
> language design issue. Accordingly, I've added comp.lang.misc to
> the newsgroups.

> ... Decent
> languages have a native boolean type with values true and false;
> one could consider having a native three valued type with values
> lt, eq, and gt.

> On Jul 18, 8:38

> On Thu, 17 Jul 2008 11:57:34 -0400, "S James S Stapleton"
> osu.edu> wrote:
>

>>There's a programming construct I've thought of, and I've been searching

>>for it for a while, and haven't found any ideas on why it has/hasn't been
>>implemented in any language.
>>
>>It's fairly obvious, so I'm sure there's a reason, could someone point me

>>the right direction?
>>
>>
>>The construct: NIF
>>Rationale: Numerical IF, used for comparisons. Many functions, such as
>>strcmp in C, as an example, have negative values for less than, 0 for
>>greater than, and 1 for equal. Three options, however, you can either use
>>two options with an if (if/else), or have many options for a switch - all ...

>On Jul 18, 8:38=A0pm, c...@tiac.net (Richard Harter) wrote:

>> On Thu, 17 Jul 2008 11:57:34 -0400, "S James S Stapleton"
>> osu.edu> wrote:

>>>There's a programming construct I've thought...

>cri@tiac.net (Richard Harter) writes:
>

>> This is sort of interesting, to me anyway, but I perceive it as a
>> language design issue. Accordingly, I've added comp.lang.misc to
>> the newsgroups.

>

>> ... Decent
>> languages have a native boolean type with values true and false;
>> one could consider having a native three valued type with values
>> lt, eq, and gt.

>
>Or LT, GT and EQ in Haskell. I don't see why the type needs to be
>native unless, by that, you mean that any built-in types must have an
>operator that produces the correct values.

>
> No, the compare should return one of the three. To get le, ne,
> or ge, do a boolean negation, e.g., in C, le is
> !(compare(x,y)==gt).

> No offense - I'm curious, but I don't see what value this offers. With a
> function like strcmp(), you wan't either a true/false result or you want a
> specific one out of three result. I.e., are they equal or not? Or, how
> are
> they different?

> It is "nifty" that they combined two methods of comparison
> into a single function. So, I don't understand why you'd need two or
> three
> of three results. But, if you did, you just save the result to a
> temporary
> variable or "flag" and use another if() or 'else if()'... Also, compilers
> optimize switch() very well.