[NeXus-committee] Questionaire Reminder
V. Armando Sole
sole at esrf.fr
Thu Sep 20 07:24:24 BST 2012
Hi Ray,
On 20.09.2012 07:53, Ray Osborn wrote:
> I think Armando's summary highlights a communication problem that we
> are having with data analysis developers. They don't see the
> advantages over just using HDF5,
Agreed, but why should be the NIAC who decides what is to be put and
where?
Until the respective communities do not see the need to standardize,
any action is going to imply an opposite reaction...
> but the whole point is that HDF5
> files are only self-describing if there are established conventions
> for what to put in them, that are valuable even if they are not
> strict
> rules.
> Perhaps people writing their own data analysis code don't care,
> because they know what their own files contain,
Here we disagree.
We do not know what the files of Diamond, Soleil or ALBA contain, but
that does not prevent users of those facilities from using our HDF5
analysis codes.
The beauty of HDF5 is that if an analysis program needs a certain
parameter, and that parameter is in the file, the user can tell where to
find it, save his choice for the future and (s)he is done. Definitions,
dictionaries or however we want to call them, can be built "on the fly".
From the analysis point of view, the important thing is "what is needed
for the analysis" not how or where and we do not think it corresponds to
the NIAC to decide.
> but somehow we have to
> persuade people that NeXus is not just another API.
>
I can only say we do not need it to write NeXus compliant files nor to
read them.
In addition, we want our analysis codes to be able to work on non-NeXus
compliant HDF5 files.
> I don't know how this fits into the NIAC agenda, but it's been an
> important issue at the APS, where some have complained about needing
> to install a new API.
>
Well, obviously I cannot talk about APS. Just one consideration: Why
not to install the API and let people use it or not? If they see an
advantage they will use it, if not they will not do so.
Best regards,
Armando
More information about the NeXus-committee
mailing list