• Subject: spilt lists in DDL 1.4
  • From: Brian Toby <brian.toby@xxxxxxxx>
  • Date: Tue, 21 Jun 2005 12:24:52 -0400
I have been following loosely the discussions on _atom_site_aniso_label 
and want to offer an observation. The fact is that it is common for 
people to treat only a small fraction of the atoms in a structure with 
an anisotropic ADP model and for those people it can make sense to have 
loop of Uij values separate from the main atom loop. For other uses, it 
makes sense to have a single loop that includes both, where the 
connection between items becomes implicit. When category rules were 
retrofitted into CIF via DDL 1.4, the list reference and list key 
concepts were used as ways to treat this dual usage, though this did 
not entirely please either the data normalization proponents or their 
opponent(s).

There are a number of other examples in CIF (at least in pdCIF) where 
data relationships are reorganized because data collection practices or 
structure model descriptions differ for various types of experiments.

The more formality that CIF places on relationships between related 
data items, the greater chance that someone will come up with a 
crystallographically useful situation that breaks that relationship. To 
treat the new paradigm, CIF will have to catch up (fast!) and define an 
entirely new dictionary populated with previously-defined data items, 
but having different relationships defined or define a entirely new set 
of data items with definitions identical with existing items. In the 
short term, data will be lost from the publication stream, because the 
standard cannot accommodate it. Further, when there are too many ways 
for the same information to be included in a CIF, the value of having a 
standardized format diminishes because it becomes increasing difficult 
for software to use that information.

IMHO, go gently with formalizing data relationships.

Brian Toby

_______________________________________________
cif-developers mailing list
cif-developers@iucr.org
http://scripts.iucr.org/mailman/listinfo/cif-developers

Reply to: [list | sender only]