[gdal-dev] Motion: adopt RFC 99: Geometry coordinate precision

[Even Rouault]

Le 07/03/2024 à 20:44, Kurt Schwehr a écrit :
> +0 KurtS. It seems like a good idea, but I worry about 
> unintended consequences, but can't come up with any.

I'm quite confident that as most new features (or sometimes even 
bugfixes), it will cause issues in some scenarios. Hopefully it will 
also help address more scenarios that aren't covered currently than ones 
that it will break :-) The main scenario I can imagine where the feature 
will trigger without users explicitly opting for it is when using 
ogr2ogr from a source that has an advertized precision (so no existing 
dataset). And by default none of the drivers will write coordinate 
precision metadata, unless the user has explictly defined one (through 
the API or the -xyRes switch of ogr2ogr). For example, even if you use 
the GeoJSON driver with RFC7946 output which rounds to 7 decimals by 
default, the xy_coordinate_resolution=1e-7 metadata will *not* be 
written in the file. And the new switch -unsetCoordPrecision switch will 
be a way to go back to the pre 3.9 behavior when dealing with source 
datasets that have an advertized coordinate precision. So hopefully 
backward compatibility should be quite good. A good indicator is that 
none of the existing regression tests had to be modified.

-- 

http://www.spatialys.com
My software is free, but my time generally not.