[gdal-dev] Call for 4.0 ideas

[Robert Coup]

Hi Joaquim,

There have been other similar changes... for example, GDAL added an
aspatial extension[1] to GeoPackage v1.0 to support non-geo tables... the
GeoPackage standard added support for such tables in v1.2; so our extension
is no longer needed: GDAL still reads it, but no longer writes it [2].
We've done similar things in other drivers, and obviously removed entire
drivers before.

Every line of code has some test and maintenance burden... Are there netcdf
readers that *don't* support crs_wkt and *only* support spatial_ref?

@Even is it worth a pass through other drivers to swap default option
values or remove obsolete options with 4.0? And do another deprecation/cull
through unused drivers?

Rob :)

[1] https://gdal.org/drivers/vector/geopackage_aspatial.html
[2]
https://gdal.org/drivers/vector/gpkg.html#drivers/vector/gpkg-lco-ASPATIAL_VARIANT


On Thu, 21 Sept 2023 at 15:31, Joaquim Manuel Freire Luís via gdal-dev <
gdal-dev at lists.osgeo.org> wrote:

> Even, I understand that. Specially because not all nc files are created
> with GDAL 😊
> But I have more troubles to understand why breaking code that use
> spatial_ref  for decades without any real need. Sorry, and I know it's not
> decided yet, but it looks so much the Linux culture of *easy breaking*.
>
> Joaquim
>
> -----Original Message-----
> From: Even Rouault <even.rouault at spatialys.com>
> Sent: Thursday, September 21, 2023 12:33 AM
> To: Joaquim Manuel Freire Luís <jluis at ualg.pt>; gdal-dev at lists.osgeo.org
> Subject: Re: [gdal-dev] Call for 4.0 ideas
>
>
> Le 21/09/2023 à 01:12, Joaquim Manuel Freire Luís a écrit :
> > Remove spatial_ref WKT export in netCDF driver (GDAL 4) #4712
> >
> > Would this mean all codes that use spatial_ref would be broken????!!!!
>
> yes, spatial_ref is AFAIK a GDAL specific thing that predates the
> introduction of crs_wkt in the CF conventions (cf
> https://github.com/cf-convention/cf-conventions/issues/222,
> https://github.com/opendatacube/datacube-core/issues/837 where this is
> also discussed). Recent GDAL versions write both crs_wkt and spatial_ref as
> a transition step. This would complete the transition by removing writing
> spatial_ref (presumably we'd still accept it on the reading side for older
> datasets). Code that only checks for spatial_ref should be adapted to deal
> with crs_wkt, since it is the standardized attribute, independently on
> whether GDAL discontinues writing spatial_ref or not.
>
> --
> http://www.spatialys.com
> My software is free, but my time generally not.
>
> _______________________________________________
> gdal-dev mailing list
> gdal-dev at lists.osgeo.org
> https://lists.osgeo.org/mailman/listinfo/gdal-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/gdal-dev/attachments/20230921/3e2d8f35/attachment.htm>