<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#FFFFFF">
On 13-12-13 00:24, Paul Ramsey wrote:<br>
<blockquote cite="mid:etPan.52aa45a9.2ae8944a.6320@Butterfly.local"
type="cite">
<style>body{font-family:Helvetica,Arial;font-size:13px}</style>The
GeoPackage document actually specifies a representation for POINT
EMPTY, "In GeoPackages these points SHALL be encoded as a Point
where each coordinate value is set to an IEEE-754 quiet NaN
value”.</blockquote>
I'm to blame for this one. It's specified in the GeoPackage spec
because it's not specified in SFSQL or SQLMM and I didn't want to
leave this as an unspecified part of the encoding.<br>
<br>
The reason I pushed for the nan based encoding is that it provides
an encoding of 'point empty' that does not lose information. The
alternative of encoding it as null or multipoint empty changes the
semantics of the object. If you stick with the point wkb type then
the only option for special values is special double values.
point(nan nan) seemed like a reasonable approximation of point
empty; a point where the coordinates are not valid numbers.<br>
<br>
If you think this is a bad idea or causes real difficulties in
implementation please let me know, I'll bring it up in the
geopackage swg then. Alternative solutions are also welcome of
course.<br>
<br>
Pepijn<br>
</body>
</html>