Re: [h5md-user] Developer's Meeting?

[Peter Colberg]

On Tue, Sep 17, 2013 at 12:41:50PM +0200, Konrad Hinsen wrote:
> Indeed! What's still missing though is a specification of the unit
> strings. The current documentation merely says that "nm" is an allowed
> unit specification. A complete and general specification is not
> trivial, but as a first step we could provide an explicit list of
> units that is a bit longer than just one.

How about specifying an H5MD module "units/si" for this purpose? That
module could constrain the format of units, e.g., to the format of
the udunits [1] package suggested by you earlier.

[1] 
http://www.unidata.ucar.edu/software/udunits/udunits-2/udunits2.html#Database

> For verifying if H5MD is ready, and for demonstrating to the world
> that it is, it would be good to have a repository of small example
> trajectories. I'd be happy to contribute some protein examples.

This is a good idea!

I recently gave a talk on git-annex [2], a tool to manage large data
using git without checking them into git. We could provide links to
the example data files in the h5md repository pointing to a few web
servers. That would allow convient partial or full retrieval of
example trajectories.

I am not sure how much storage we have on Savannah. Who of you could
provide a mirror for these example files? What maximum size are we
limiting the files to?

[2] http://git-annex.branchable.com/

Peter