Date: Wed, 18 Jul 2001 08:33:20 -0400 (EDT)
From: Bob Bruninga
Subject: Re: [aprsspec] Area Objects


On Tue, 17 Jul 2001, Curt Mills, WE7U wrote:

> Area Object Questions:
> ----------------------
> Perhaps circles/ellipses are defined with the lat/long point being
> the center of the object?  The upper left corner point of the box
> the circle/ellipse would fit inside?

Yes, that is correct

> Same for rectangles: is the lat/long point the [LOWER]right-angle
> corner, or perhaps the upper left corner of the rectangle it would fit
> inside?

Yes.  Lower right.  The offsets then define the distance back to the upper
left.

> 2) What sort of triangle?  

The triangle is just a specialized LINE.  Draw the line using the rules
for LINES.  Then make it into an isosceles triangle symetrical about the
vertical axis.  Thus you can only draw isosceles triangles and only with
the base down...

> End result:  I wouldn't be able to draw a triangle of any other
> rotation.  Imagine a search where I knew the direction of travel
> to be northerly due to tracks being found, but I couldn't draw
> a triangle on the search area that angled upwards...

Use the LINE with a WIDTH.  Im not sure that any of the other authors
implemented LINE WIDTHS, but APRSdos did.  This way you can draw a
rectangle at any angle, showing a corridor or other such shape...

> 3) For circles:  Does it have to be a circle in lat/long, or could it
> be a more useful circle with a constant radius (distance-wise)?

Constant radius.  APRSdos always scales a map so that there is no
distortion based on latitude.  Thus, a circle is a circle...

> spec says lat/long, which isn't as useful as say a circle of 0.5
> miles radius.  The main reason is that it will look like an ellipse
> instead of a circle most places in the world.

A 50 mile circle is an elipse on any mercator projection map.  That is why
APRSdos scales all map screens to latitude to eliminate that distortion.
Most other authors have not done that.

> 4) Unless circles and ellipses are defined differently, why have
> circles at all?  Ellipses can be circles.

Actually, we never did elipses.  The only definitions in APRS are
circles, lines boxes and triangles.  There was no interest in doing
elipses..

> One of the shareware APRS programs appears to have incorrect sizing
> for the area objects, and they also go the wrong direction from the
> corner point w.r.t. what the spec says.

Yep, I have been fighting this for years.  There seems to be
little interest in fixing it...

Bob
----------------------------------------------------------------------


Date: Thu, 16 Aug 2001 09:31:04 -0700 (PDT)
From: Olivier Calle
Subject: [aprsspec] Area Object Sizing


There was a thread on this topic in late June that started with my
confusion in trying to decode area objects by reading the spec.
After discussion in that thread with Steve Dimse, we concluded that the
spec was a bit misleading and came up with the following equation:
offset_in_degrees = (offset_in_packet / 100)^2

Later, I communicated privately with Bob because this equation still
seemed to create the wrong size areas.  He sent me some example objects,
describing their sizes (i.e.: rectangle 20miles by 30miles), for me to
compare.  I then was sure that the above equation is not correct.
Bob then sent me a code snippet and I believe I found the correct equation
and a possible explanation of the confusion:
offset_in_degrees = (offset_in_packet^2) / 1500

Using that equation gave me areas with sizes the same as the example
objects that Bob had sent me.  I believe the confusion may have come from
the comment near that section of code that talks about 100ths of a degree
squared.

Comments?

-- 
Olivier Calle