On Wed, 4 Jun 2003, Thomas Radke wrote:

> > > I'd start without these complexities in the hdf5 module, as you can
> > > do nearly all of this stuff within your dx program.
> >
> > I agree with Richard.  A module is going to be easier to use.
>
> Well, not necessarily. Sure, if people aren't interested in the hyperslabbing
> features then they don't need to care about them. It's up to the import
> module to provide sensible defaults for those parameters so that the module
> does what people expect it to do, without much fiddling around.
>
> But it is essential for us to have the hyperslabbing functionality already
> integrated in the import module (and not having to use the Slab module)
> because our datasets might simply be too large to fit in memory.

Yes, I also think hyperslabbing is essential.

> > This summer we will be working intensly for roughly a month on our
> > package.  We will let you know of our advances.  Please keep us posted
> > too.  Thank you for sharing your ideas!
>
> Thanks to you for this discussion !
>
> Let me finally come back to the question of merging our efforts of developing
> a general HDF5 data import module which fits peoples' needs (for their unigrid
> data at least), and make it eventually available to the main distribution of
> OpenDX. Do you think that's possible ?

I think its badly needed to have "basic" hdf5 in OpenDX. And I believe we
can create a module that fits all our needs, if with some surrounding DX
macros, though. Of course individual modules will fit our individual needs
better, but even building those on a common base will help everyone.

> As I see it the biggest issue here isn't of technical nature (we almost agreed
> on a common file layout) but rather what license should be used for such a
> module.
> For your dxhdf5 package you use a special UoC license, Richard has the 
> Academic
> Free License, and I'm using GPL for the OpenDXutils package.

Well - if I work on the hdf5 module during my work, I need to interrogate
with the german science foundation about the license (I assume the same is
technically true for you, Tomas?) - last time they said only
academic/research use is ok. Personally I'm fine with even LGPL - but
compatibility with the OpenDX license should be important.

> Any chance of finding a solution which works for everyone (not just for us 
> from
> "Good Old Europe") ?

At least people form "Good Old Europe" may meet easily. I know at least 4
persons in Tuebingen using either hdf5 or even Cactus, three of them
involved with the Gravitationswellen Transregio (not me). I think we in
Tuebingen would be able to invite two persons for a week (of course we
need to find about possible license issues again here, as that would be
sponsored by the German science foundation again...).

Richard.

Reply via email to