[Ifeffit] Re: Ifeffit Digest, Vol 22, Issue 1

[Matt Newville]

Bruce,

I want to go back a step.  First, you asked what commonly used
crystals athena should know about.  I think the answer should be
"None".  Athena should not be the tool to convert mono steps to
energy.  It should not know the thermal expansion curve for Si.  
The beamline should supply values in units of energy or wavelength
or provide software to convert to such units.

Then the discussion got to supporting data formats from different
beamlines.  I don't disagree with your desired formatting rules.
For what it's worth, I think we (that is, those on this email
list) are as qualified to set such a standard as the IXS (or at
the very least, we're approximately a quorum of any such group of
the IXS).  Here's a proposed draft 0.1 for a data format standard
for beamline collection software:

   The data can be read and used by Athena without modification.

Whether any such standard is actually used is a different matter.
At the APS there are at least six different beamline data formats
used for XAFS (there might be more than that: I'm pretty sure
sectors 4, 5, 10, 12, 13, and 20 each use different formats).  At
least 3 of these pass the athena standard.  One of those formats
is spec, which does not meet this standard.  I believe spec is
used widely at many other synchrotrons.  At the APS, some data is
saved in HDF format.  HDF may become more common as EPICS seems to
be moving this way.  I don't know that any EXAFS data is currently
saved only as HDF, but I wouldn't be surprised if happens. If I'm
not mistaken, SSRL saves data in a binary format. It can be
converted to ASCII, but that still does not meet the 'read by
athena' rule.  I believe we had a recent example where data from
SRS was close, but also did not quite meet this standard.

But: I'm not aware of any beamline saving data in mono steps.  
AFAIK, only the old Lytle data is in this format.  We should not
do anything to encourage this very bad practice. For the record,
it's bad because it assumes you know:  1) steps per degree, 2)
value of steps at angle=0, and 3) lattice spacing of crystal.  If
you don't know all three values, you have nothing.

So, I agree it would be foolish to try to support every beamline
format, and not likely to be successful.  I also think putting in
the lattice constant for YB66 or polynomial coefficients for the
thermal expansion curve of Si is a similarly doomed.

--Matt