Discussion List Archives

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

Re: Problems with CIF BNF

> In general, it is better to think of CIF as a series of line-records 
> where trailing spaces is insignificant (i.e. Fortran I/O).

Just to round off this discussion (for now) on the list, I should
note that Joe is following up some of these technical points off-list
with Syd, Nick and others.

A general comment that I would make is that CIF is a living, if
imperfect, system, and that many suggested improvements - albeit
more elegant and/or efficient - could easily break existing
applications.  The software list on the current CIF pages is
woefully out of date (mea culpa), but I'm in the process of
revising it (and starting to rework the entire CIF site), so
you will be seeing much more up-to-date content in the
next few weeks.

That's not to argue for complete ossification, of course. In
the near future the CIF site will host a proposal for a
development of DDL that will formalise methods-based
dictionary-driven validation. Implementation will require
some changes to the STAR BNF, and that could set the scene
for introducing evolutionary changes to the CIF format.
However, users will be reassured to know that the existing
CIF format will remain as a supported archival format; and
it's important to look at providing the communities with
tools to migrate existing CIFs into any DDLm-inspired
modifications that might evolve.

Much of what is projected in DDLm goes beyond any existing
XML schema-based validation, and it's important for
software developers to concentrate on the potential
of methods-based validation, rather than getting hung up on
the detail of the CIF format.

Brian

Reply to: [list | sender only]