Even Rouault
Even Rouault
you may want to use https://github.com/airbusgeo/cogger
> and, modifying that with vrt: doesn't fix it because the behaviour is in the NetCDF driver not in VRT This is also an option, so "?oo=ASSUME_LONGLAT=YES". But this won't...
The biggest hurdle will be GDAL's autotest suite, that relies on not-throwing in a likely big number of places. I guess the reasonable thing to do would be to have...
> That may help with the process of converting. It's heavily used in https://github.com/OSGeo/gdal/pull/7475
Assigning milestone 4.0 for that one. In PR #7452, we set some grounds to make that happen
> I am wondering if @rouault has any further thoughts on this? I'm afraid I don't understand what Spack does here, but one general point of caution/advice: the behavior of...
> if I pass an empty array to the rasterio `.read()` method : this smells a bit like a reference counting issue. Maybe related to https://cython.readthedocs.io/en/latest/src/userguide/language_basics.html#python-objects-as-parameters-and-return-values and calls like https://github.com/rasterio/rasterio/blob/5db194fe99c566479858f71baa0b7d22e560dd22/rasterio/_io.pyx#L653,...
Perhaps try adding sys.getrefcount(img) and sys.getrefcount(mask) to compare if there are differences between the rio and gdal paths? In a non-leaking scenario, I would anticipate the counter to be just...
> what would you think about me adding such a workflow to the GDAL repo? sure
should be fixed per https://github.com/OSGeo/gdal/pull/8402