Discussion List Archives

[Date Prev][Date Next][Date Index]

(63) more on pdCIF categories

  • To: COMCIFS@iucr.ac.uk
  • Subject: (63) more on pdCIF categories
  • From: bm
  • Date: Fri, 9 May 1997 20:20:33 +0100
Dear Colleagues

D61.1 pdCIF categories
----------------------
I attach Syd Hall's response to the question of imposing a tighter category
model on the pdCIF dictionary.

S> Brian: I have read through the latest communiques and in a way am relieved
S> that this category business will get further airing with respect to the
S> powder dictionary. 
S> 
S> As you indicated I have had reservations about the lack of correspondence 
S> between some of the datanames and their categories for some time (my mail 
S> goes back to the early days of the pd definitions). I suspect that the 
S> departure of categories from names evolved as the list structures of the
S> powder definitions evolved (divergently)...BT can fill us in on that 
S> history. My reminder to COMCIFS about these departures was so that we
S> understood that they represented a change in philosophy from earlier 
S> dictionaries and a different approach for some software (those using the
S> data name rather than the explicit category definition to check category).
S> 
S> However, I want to emphasise that am not categorically (:->) against the 
S> separation of category and dataname naming conventions, because this is 
S> strictly an "application matter", as opposed to a requirement of the STAR 
S> syntax (see p506, JCICS, 34, 1994 for explanation and examples). By the way 
S> it is also not part of the STAR DDL definitions (JCSCI 35, 819-825, 1995).
S> 
S> On the other hand, the association between "category" and the construction 
S> of datanames is very strongly implied in the original 1991 CIF core paper, 
S> though as far as I can see there is no EXPLICIT instruction that the data
S> name MUST contain the category name in a CIF. Certainly the earlier DDL1
S> CIF dictionaries never departed from this association, and the later DDL2 
S> mmCIF dictionary built this into its dot naming structure.
S> 
S> I continue to support both the DDL1 and DDL2 efforts because they are 
S> appropriate to their applications. A rather pragmatic view perhaps, but we
S> all need to be reminded occasionally that dictionaries are not pedantic 
S> exercises...and existing data handling needs must be addressed with some 
S> urgency. So I too do not have a particular dogma on the dataname/category
S> relationship. I can see advantages and disadvantages to a strict association.
S> Perhaps it would make life simpler to fix this relationship but there are 
S> views that the hierarchical approach of DDL2 is not only one to be considered
S> in data handling, and therefore it MAY be unwise to set the relationship
S> in stone at this stage. Perhaps Nick Spadaccini should comment on this.
S> 
S> This is an important matter and I am pleased that Paula's effort and concern 
S> has brought it to a head!

Have a good weekend,
Brian