paleolimbot commented on code in PR #240:
URL: https://github.com/apache/parquet-format/pull/240#discussion_r1607000607
##########
src/main/thrift/parquet.thrift:
##########
@@ -982,6 +1054,23 @@ union ColumnOrder {
* `-0.0` should be written into the min statistics field.
*/
1: TypeDefinedOrder TYPE_ORDER;
+
+ /**
+ * The order only applies to GEOMETRY logical type.
+ *
+ * Please note that geometry objects cannot be compared directly. This order
aims to
+ * provide an approach to build a bounding box for geometry objects in the
same page
+ * or column chunk.
+ *
+ * In this order, all 2D geometries are regarded as a collection of
coordinate (x, y).
+ * For example, POINT has one coordinate, LINESTRING has two coordinates,
and POLYGON
+ * might have three or more coordinates. A bounding box is the combination
of x_min,
+ * x_max, y_min, and y_max of all coordinates from all geometry values. For
simplexty,
+ * min_value field in the Statistics/ColumnIndex is encoded as the
concatenation of
+ * PLAIN-encoded DOUBLE-typed x_min and y_min values. Similarly, max_value
field is
+ * encoded as the concatenation of PLAIN-encoded DOUBLE-typed x_max and
y_max values.
Review Comment:
I think the suitability of a bbox vs. a covering is orthogonal to the number
of values...a single bounding box is not able to represent things like Alaska
or Fiji (which straddle the antimeridian) or Antarctica (which covers a pole).
There are workarounds, of course: you can use two bounding boxes in some cases,
use some conventions for bounding boxes that cover a pole, specify that if
min_lat > max_lat that the bounding box straddles the antimeridian, etc.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
