Re: Tagged intervals (Was Branch & bound for not everywhere defined constraints)
> From: "Corliss, George" <george.corliss@xxxxxxxxxxxxx>
> To: Dan Zuras Intervals <intervals08@xxxxxxxxxxxxxx>, "R. Baker Kearfott"
> <rbk@xxxxxxxxxxxxx>, Ralph Baker Kearfott <rbk@xxxxxxxxxxxx>
> CC: "stds-1788@xxxxxxxxxxxxxxxxx" <stds-1788@xxxxxxxxxxxxxxxxx>
> Date: Sat, 5 Sep 2009 07:33:03 -0500
> Subject: Tagged intervals (Was Branch & bound for not everywhere defined
> constraints)
>
> This thread reminds me of a discussion that MIGHT be slightly related:
> Tagged intervals.
>
> Please set aside concerns of execution speed for a moment (yes, that's
> hard for me, too). Also, set aside my frequent calls for KISS.
>
> . . .
>
> In an object-oriented software design, we could define a class
> DecoratedInterval that is a [double, double] and whatever attribute
> tags we wanted.
>
> . . .
>
> George
>
I'd be very happy with this approach. We'd get our KISS
which gets us our standard hardware someday & everyone
could djinn up their own decorations.
Would you also want to define some standard decorations
together with some constraints on their behavior? Or
would decorating an interval take it out of 1788
altogether? I can see arguments both ways.
We would also have to convince all parties involved that
they need not lobby for their favorite features on the
grounds that the decorations are available. That may be
a hard sell.
But I like the simplicity of it all. - Dan