Use the built-in float datatypes to implement geometric types
This patch makes the geometric operators and functions use the exported
function of the float4/float8 datatypes. The main reason of doing so is
to check for underflow and overflow, and to handle NaNs consciously.
The float datatypes consider NaNs values to be equal and greater than
all non-NaN values. This change considers NaNs equal only for equality
operators. The placement operators, contains, overlaps, left/right of
etc. continue to return false when NaNs are involved. We don't need
to worry about them being considered greater than any-NaN because there
aren't any basic comparison operators like less/greater than for the
geometric datatypes.
The changes may be summarised as:
* Check for underflow, overflow and division by zero
* Consider NaN values to be equal
* Return NULL when the distance is NaN for all closest point operators
* Favour not-NaN over NaN where it makes sense
The patch also replaces all occurrences of "double" as "float8". They
are the same, but were used inconsistently in the same file.
Author: Emre Hasegeli
Reviewed-by: Kyotaro Horiguchi, Tomas Vondra
Discussion:
https://www.postgresql.org/message-id/CAE2gYzxF7-5djV6-cEvqQu-fNsnt%3DEqbOURx7ZDg%2BVv6ZMTWbg%40mail.gmail.com
Branch
------
master
Details
-------
https://git.postgresql.org/pg/commitdiff/c4c34008854654279ec30067d72fc5d174d2f42f
Modified Files
--------------
src/backend/access/gist/gistproc.c | 100 ++++----
src/backend/utils/adt/geo_ops.c | 501 ++++++++++++++++++++-----------------
src/backend/utils/adt/geo_spgist.c | 28 +--
src/include/utils/geo_decls.h | 19 +-
4 files changed, 343 insertions(+), 305 deletions(-)
