Re: [h5md-user] Mandate Variable-length string datatype

[Felix Höfling]

Am 23.09.2013, 18:01 Uhr, schrieb Peter Colberg  
<address@hidden>:

> On Mon, Sep 23, 2013 at 05:45:51PM +0200, Felix Höfling wrote:
> > I don't get your point.
>
> In the C++ mindset, anything can be wrapped, not a problem… ;-)
>
> > Why are readers and writers not interoperable?
>
> If program A writes the string attributes of the h5md group using a
> variable-lengths string datatype, and program B reads the string
> attributes of the h5md group using a fixed-length string datatype,
> the HDF5 library will return an error.
>
> Unless you have a good reason why fixed-length string datatypes
> are important for the storage of the existing H5MD attributes,
> I see no point in provoking these incompatibilities.
>
> Peter

The reader has to be aware of the two different HDF5 string types, of  
course. Both string types are proper HDF5 and I don't see any reason why  
we should restrict to a subtype. As I said, the distinction is a really  
simple task. I believe it consumes more time discussing the issue further  
than fixing the reader.

Felix