[gdal-dev] compile error on windows
even.rouault at spatialys.com
Tue May 3 11:11:53 PDT 2016
Le mardi 03 mai 2016 19:57:45, Jeff McKenna a écrit :
> I'm also very ok with changing the default to set
> H5_BUILT_AS_DYNAMIC_LIB only - I'll just edit the makefile before
> building. Not an issue for me.
I'm not keen to modify things without a clear understanding of what is
required in which context.
> Since no one is recording these notes from this long discussion in the
> buildhints wiki, I'm +1 to change to this new setting in makefile.vc;
> then I'll make a note on the buildhints wiki and make sure the MS4W
> process is updated.
I agree. Ideally we would need a table :
HDF5 version | HDF5 cmake build options | GDAL required compilation flags
> > On 2016-05-03 1:36 PM, Even Rouault wrote:
> >> If you have the opportunity, it would be good if you could check if
> >> defining
> >> both _HDF5USEDLL_ and H5_BUILT_AS_DYNAMIC_LIB doesn't affect
> >> negatively your
> >> builds. Ideally, we'd like to have a default set of compilation flags
> >> that
> >> works out of the box for (almost) everybody.
> > It will depend how HDF5 was compiled (the cmake settings). For example,
> > I build the static lib of HDF5, and therefore in GDAL's nmake.opt I
> > point to "libhdf5.lib". In my case I therefore do not need to set
> > H5_BUILT_AS_DYNAMIC_LIB
> > If I try to set both _HDF5USEDLL_ and H5_BUILT_AS_DYNAMIC_LIB I get 61
> > unresolved symbols in the linker.
> > Now are you understanding my grey hair comment Even? :)
> > I feel these notes are important to document in the buildhints wiki.
> > -jeff
> gdal-dev mailing list
> gdal-dev at lists.osgeo.org
Spatialys - Geospatial professional services
More information about the gdal-dev