Fwd: Re: DDLm, dREL, images and NeXus
Herbert J. Bernstein
yaya at bernstein-plus-sons.com
Tue Dec 16 23:11:20 GMT 2008
In Nexus the same item may appear at multiple points in the tree. In CIF
all uses of the same item need to be gathered together into one table.
In order to handle normalization/denormalization, whatever language
is use has to be able to address an arbitrary file/data_structure, not
just the fairly well-normalized CIF data structures with their neat
table structures. NeXus (and, for that matter, XML) is tree-structured, rather
At 9:55 AM +1100 12/17/08, Doug wrote:
>On Mon, 15 Dec 2008, Herbert J. Bernstein wrote:
>> The most complex transform is the normalization in going from
>> NeXus to CIF, and the denormalization going from CIF to NeXus.
>Actually I don't see how you could "denormalize" the
>audit_athor category *faithfully* back into NeXus by putting multiple
><name>s back within a single NXuser instance, assuming they had common
>affiliation and phone numbers etc.
>As I understand it, within dREL, each time you attempted to set
>the nexus:raw_author.user_name item (say within a ddlm/dREL category method)
>for consecutive <names>, it would overwrite the previous value.
>As an alternative, if you were to just treat <name> as a dREL list
>then a single NeXus <name> element would end up holding every name.
>In this case you could always write every name out individually as
>a full NXuser entry (as it would be in CIF), I would guess.
>But generally, any place where a NeXus element can have a mixture of
>single and multiple subelements, it will probably be incompatible
>with ddlm/dREL/RDBMS design because it breaks the neat table structure.
>comcifs mailing list
>comcifs at iucr.org
Herbert J. Bernstein, Professor of Computer Science
Dowling College, Kramer Science Center, KSC 121
Idle Hour Blvd, Oakdale, NY, 11769
yaya at dowling.edu
More information about the comcifs