Re: [Fink-devel] Updated HDF5 packages in my experimental area

[Alexander H. <ale...@gm...>]

Alexander Hansen wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> The update is motivated by my desire to try to package NetCDF 4, which
> requires HDF5 1.8 .
>
> The files are at:
> http://www.finkdeveloper.net/svn/users/akh/experimental/netcdf4/finkinfo/
>
> The "hdf5" packages are an update of our existing packaging from
> version 1.6.5 to 1.6.8 .
> The "hdf5-18" packages corresponds to a newer upstream version with a
> different API and library install_name.   Due to overlap of one of the
> libraries, I decided to put the headers and libraries in a private
> directory; hdf5-shlibs and hdf5-18-shlibs coexist as should be the
> case, and the main packages (which are the headers) Conflict/Replace.
> The user executables stay in "hdf5-bin".  The fortran bindings had to
> go into a separate package.
>
> Try them out and see if everything's OK.  In particular, I'd like to
> make sure that the update to 'hdf5' doesn't cause any problems, since
> that's intended to be able just to update our existing package.
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.9 (Darwin)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
>
> iEYEARECAAYFAkltAtcACgkQB8UpO3rKjQ94swCeIK39JeY5eZr2Gq3Ww4UPekgV
> ADMAnA+tmzqjHUP9DfKBncHd8XEm4JmL
> =k4zT
> -----END PGP SIGNATURE-----
>
>   
I've gone ahead and added these to the unstable tree.  I've kept the
prior hdf5-1.6.5 around, too, in case of any problems, since I know how
science people get when they lose access to their data.

The major difference between what was committed and what was in my
experimental is that I decided to have the user binaries from hdf5-18 be
in a "hdf5-18-bin" splitoff instead.


-- 
Alexander Hansen
Fink User Liaison