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

Re: the dictionary merging protocol

• Subject: Re: the dictionary merging protocol
• From: Doug du Boulay <ddb@xxxxxxxxxxxxxxxxxxxxxxx>
• Date: Tue, 16 Jul 2002 03:12:30 +0100 (BST)

On Mon, 15 Jul 2002 20:43, Herbert wrote:
>
> This goal cannot be achieved without invaliding some existing CIFs.
> When an author creates a core CIF for publication, it is valid for him
> to include items in the PUBL_MANUSCRIPT_INCL category, to specify tags
> being used in that particular CIF which are not being drawn from the
> standard publication request lists or, more importantly, even from the
> core dictionary.

Probably those CIFs don't even state that they are valid against 
any dictionary, much less conformant. The assumption that they are 
valid against a crystallographic core dictionary seems kind of short sighted
if not egocentric (with the benefit of hindsight). Classify them as CIF1.0 
and forget about them, or alternatively, fix them to make them conformant 
to something...

>
> As it says in the category description:
>
>     Data items in the PUBL_MANUSCRIPT_INCL category allow
>     the authors of a manuscript submitted for publication to list
>     data names that should be added to the standard request list
>     employed by journal printing software. Although these fields are
>     primarily intended to identify CIF data items that the author
>     wishes to include in a published paper, they can also be used
>     to identify data names created so that non-CIF items can be
>     included in the publication. Note that *_item names MUST be
>     enclosed in single quotes.
>
> One might attempt to solve the problem by insisting that all authors
> making use of this category provide a complete new dictionary merging
> the core CIF dictionary with their proposed additions, but, in

Yes exactly, but  I wasn't thinking of a complete new merged dictionary, I was
thinking of a published, referencible, mini dictionary containing only the 
subset of new definitions that it wishes to iether replace, or modify or 
append with respect to its hierarchical parent (e.g. core), containing 
explicit references to those parent categories and data items that are to be 
modified. In attempting to validate etc. those references would have to be 
followed, by the validating software, and again merged, but in this case at 
least it would be explicitly specified, forever, in fine detail, rather than 
on some long forgotten command line.


> order to ensure consistency of the use of tags from the "real"
> core CIF dictionary, one would then have to somehow perform a
> difference operation between these dictionaries, producing what
> is effectively a mini-dictionary for the proposed layered extensions,

I dont think that would be necessary because one would be built up
(subclass) from the other, through explicit dictionary/data item references.

Any software that wants to validate already has to to be written to
cope with possible dictionary overlaying, so this proposal does not 
increase the already existing demands on the the software developer.
Of course authors of software written to accomodate the existing
overlay protocol would be inconvenienced somewhat (sorry about), 
but IMHO I think it could aid the future development and expansion of 
dictionaries and dictionary driven software.

anyway, 
best wishes
Doug

Reply to: [list | sender only]

• Prev by Date: Re: the dictionary merging protocol
• Next by Date: Re: the dictionary merging protocol
• Prev by thread: Re: the dictionary merging protocol
• Next by thread: Re: the dictionary merging protocol
• Index(es):
  • Date
  • Thread