Discussion List Archives

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

(67) Call for votes on pdCIF and mmCIF

  • To: COMCIFS@iucr.ac.uk
  • Subject: (67) Call for votes on pdCIF and mmCIF
  • From: bm
  • Date: Wed, 21 May 1997 14:23:02 +0100
Dear Colleagues

D61.1. pdCIF categories
-----------------------
The discussions of the past few weeks have reopened the debate on the 
handling of categories under DDL version 1 and specifically in the powder
dictionary. The purpose of this debate at this time has been to clarify the
policy that COMCIFS wishes to pursue in the release of its first two
extension dictionaries. Do we wish to endorse the parallel development of
data files and associated specialist dictionaries in a development
environment that permits two data structure formulations - two dialects, if
you will? Or should we insist on a unified data model across the whole field
of crystallography? I think there is a consensus that the dual formalism
is not ideal; it presents overheads in maintaining the dictionaries and
programming applications to use the data files. On the other hand, Brian
Toby is not convinced that the additional complexity of a fully normalised
relational tables model is appropriate to the needs of the powder community.
If we recall that the Acta publication CIF still uses the DDL1.4 formalism,
and that there is a large body of existing software that will adhere to this
standard for the foreseeable future, we can identify the powder dictionary
as an extension to files of this type, and somewhat in the spirit of the
original lightweight CIF data model.

Given the existing DDL1 data archive and the stated requirements of the
powder practitioners, I believe that endorsing the most recent draft of the
pdCIF dictionary, subject to any required fine tuning to permit a later
migration to DDL2 if required, is a lesser evil than the suppression of the
existing draft and its reworking at a later date.

There is other support for this pragmatic view, including Syd's comments
below.

S> Brian: With a some reservations, I concur with your summary of the _pd_
S> discussions. The simple fact of the matter is that we are too far down 
S> the track to reverse out of some of the _pd_ constructions...and there is 
S> probably too much at stake to trying to do so! We (COMCIFS) should have
S> insisted at the outset if we wanted a direct relationship between tags
S> and categories. Despite some early discussions on this point, it got away
S> from us because we were all busy with other matters while Brian T was
S> beavering away, doing a very good job with these definitions. 
S> 
S> Herbert Bernstein has his finger on the pulse of these cross disciplinary
S> applications better than most (through his development of CIFtbx), and he 
S> clearly prefers a DDL2 consistent approach....because this makes life 
S> much easier for us software bods (well-defined rules always do!). However,
S> he also points out that the use of tag aliases in DDL2 provides a solution
S> for dictionaries and validation software when these rule don't exist.
S> 
S> I think that we have to live with that (imperfect) solution. I vote that 
S> we put aside the category issue, sort out any definition problems or name
S> peculiarities, and accept the current _pd_ definitions asap.         

Given the perceived consensus that COMCIFS will support both DDL models as
appropriate pro tem (the '99  Congress might be a suitable occasion to
review this policy), and with the understanding that the two dialects need
to be properly documented and explained to the community, David Brown feels
the time has come for the formal endorsement of *both* dictionaries, and he
calls for a final vote under the two motions tabulated below. David presents
a timescale of two weeks; but I would encourage you to respond - especially
if you have any reservations - as soon as possible. Even if approval is given
immediately, there is a still a great deal of work to be done to present
the supporting documentation for the approved versions on the web, a task
I would dearly like to see completed in time for the St Louis meeting.

Regards

Brian

                       ******************************
                       TO ALL FULL MEMBERS OF COMCIFS
                       ******************************

	It is now several months since members of Comcifs received copies
of the pdCIF and mmCIF dictionaries and we should therefore be in a
position to vote on the acceptability of these two dictionaries.

	The role of Comcifs is to determine whether the dictionaries
conform to the cif standards.  Since the preparation of the dictionaries
is delegated to members of the community that the dictionary is designed
to serve, and since these member generally do a good job in putting
together definitions that will suit their needs, the role of Comcifs is
primarily to agree that the dictionary does conform to the existing
standards.  Comcifs, of course, sets and, if necessary, changes these
standards, and we have had many discussions in the past to establish our
standards.  Now we are being asked to decide whether these two
dictionaries meet these standards. 

	In the course of reviewing these dictionaries, each of us has
discovered places where the wording can be improved and we rightly point
out these corrections, but the vote in Comcifs is not on the detailed
wording of the dictionaries but on whether the dictionary as a whole
conforms to cif standards.  Even when Comcifs has voted to accept a
dictionary, therefore, there is still a chance for the originator to make
minor corrections up until the time when Brian McMahon lowers the boom
because the final version is about to be made public. 

	The powder dictionary has been the subject of some discussion in
the last week or two, bearing on the question of whether it was
sufficiently structured to meet our current requirements.  There has been
some concern that looseness of the proposed file structure could present
problems in the future if a more structured approach were found to be
necessary, but I have heard no very strong objections apart from those
raised and admirably presented by Paula.  However, few members of Comcifs
seem to be willing to insist on the powder dictionary being rewritten to
conform to an as-yet-unwritten and still ill-defined standard.  If we vote
*not* to accept it, the powder dictionary will require major revisions
which will considerably delay the implementation.

	By comparison with the powder dictionary, there has been little
discussion of the mmCIF dictionary.  This may be because of the difficulty
most of have in just reading our way through it, though if anyone has
serious reservations about the way in which it has been put together, they
have not raised their voices in the discussion. 

	I am therefore calling for two votes.  The first is on the motion:

D67.1 Call for vote of approval of pdCIF version 0.996+
-------------------------------------------------------

	That Comcifs approves the powder cif dictionary as presented,
	subject to editorial corrections.

The second is on the motion:

D67.2 Call for vote of approval of mmCIF version 0.9.01+
--------------------------------------------------------

	That Comcifs approves the macromolecular cif dictionary as
	presented, subject to editorial corrections. 

I propose to adopt the following procedure.  All votes are to be
registered with Brian McMahon during the next 2 weeks.  If you vote
against either of these motions you should supply a rationale, listing
those points on which you have reservations.  In the event that the
majority approve the dictionaries but the acceptance is not unanimous, the
reservations will be circulated to the members of Comcifs for a decision
on whether we wish to continue the discussion in the light of the points
raised or whether we wish to confirm the original majority vote.  If the
majority vote is for rejection, then the discussion will clearly continue
anyway.

	All the full members of Comcifs are expected to register their
votes.  Please do so as soon as possible.  There are several groups
anxious to see these dictionaries approved and the sooner we respond the
more friends we will have out there!

*****************************************************
Dr I.David Brown