[gdal-dev] Driver order for HDF5 versus netCDF

[Even Rouault]

Kurt,


> Looking at the expected behavior at GDAL's head, I see this. I think 
> it's probably fine, but figured I'd ask...

I don't reproduce with master:

$ gdalinfo autotest/gdrivers/data/hdf5/metadata.h5
Driver: HDF5/Hierarchical Data Format Release 5
[...]

$ gdalinfo autotest/gdrivers/data/hdf5/u8be.h5
Driver: HDF5Image/HDF5 Dataset
[...]


>
> Is there a reason for the order to be different for the drivers in the 
> DEFERRED section compared to the normal driver registration? Does the 
> order not matter for DEFERRED?
>
>
No, that doesn't matter since drivers are re-ordered using the order 
defined in frmts/drivers.ini

But in the netCDF vs HDF5 case, this doesn't matter much since the logic 
in netCDFIdentifyFormat() will skip a HDF5 file if "#ifdef HAVE_HDF5" is 
true and GDALGetDriverByName("HDF5") != nullptr

My guess would be that your build of the netCDF driver lacks -DHAVE_HDF5 
. By the way I'm thinking this check might be incorrect. This HAVE_HDF5 
is if libnetcdf has HDF5 support, but in the context of the check if 
GDALGetDriverByName("HDF5") != nullptr, we don't actually care about the 
capabilities of libnetcdf. It is just that if the HDF5 driver is there, 
we should just use it.

Even

-- 
http://www.spatialys.com
My software is free, but my time generally not.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/gdal-dev/attachments/20240905/d4512aff/attachment.htm>