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

Re: [ddlm-group] Stakeholders

SIMON WESTRIP wrote:
> At whatever stage CIF2 is presented to 'end users' I think its success 
> or otherwise
> will depend on how well we 'sell it'. For example, when I simply 
> described some of the
> possible changes to a colleague recently, one 'gut reaction' was 'why?', 
> followed by
> 'why not just convert to xml...!' That is, I had to sell the idea that 
> DDLm would
> make life easier for the end user by e.g. deriving missing data, and 
> that CIF2
> would enable us to include far richer content in our CIFs...
I am essentially an end-user here. My answers:

Why not XML? XML is an ugly, overly-verbose mess. It is particularly 
inefficient for tables/arrays of data. It's only value is standardization.

Why DDLm? I have no idea; I'm just here to discuss CIF2.

Why CIF2? I suppose the main reason is that lists are useful. It should 
be possible to use nested loops defined by STAR, but excluded by CIF, 
but a list syntax is more efficient. There are also some situations that 
are not well defined in CIF, such as how to enclose a large string with 
all three types of close-quotes. Unfortunately (in my opinion), the 
current set of changes seems to go beyond addressing the CIF1 shortcomings.

I think it would be easier to sell a conversion of value lists to CSV 
syntax, rather than an increasingly complex and non-standard quoting system.
> 
> I would have no idea of how to sell this to the macromolecular community -
> especially without a working implementation that could demonstrate some 
> benefit?
Apparently, Herbert's group is already doing some CIF2 (or CIF1.5?) 
coding, which might give some good examples, but also is causing 
problems because it is a moving target.

I would be in favor of presenting a tentative CIF2 format, with some 
alternatives, and get more feedback from developers. The number of 
subscribers here is too small. For example, maybe most developers want 
comma-delimited lists.

Joe


> ------------------------------------------------------------------------
> *From:* Herbert J. Bernstein <yaya@bernstein-plus-sons.com>
> *To:* jwest@rcsb.rutgers.edu; Group finalising DDLm and associated 
> dictionaries <ddlm-group@iucr.org>
> *Sent:* Tuesday, 1 December, 2009 12:45:41
> *Subject:* Re: [ddlm-group] Stakeholders
> 
> Dear John,
> 
>  > Circulating a request for comment on community lists carries an
>  > implication impending change.  For PDB users this will be viewed
>  > by many as adding confusion and complexity to a format that is
>  > already viewed as over complicated.
> 
> Isn't the very intention of this effort that there be "impending change"?
> Don't those same users have to see a specification of CIF 2 before it
> comes into use?  Will not the "maximally disruptive" approach adopted
> by this group add "confusion and complexity to a format that is already
> viewed as overly complicated" whenever they learn of this?
> 
> Surely the sooner we start the necessary educational process and
> the sooner we build support infrastructure for what has been made into
> an intrinsicly difficult transition, the better.
> 
> Regards,
>   Herbert
> =====================================================
>   Herbert J. Bernstein, Professor of Computer Science
>     Dowling College, Kramer Science Center, KSC 121
>         Idle Hour Blvd, Oakdale, NY, 11769
> 
>                   +1-631-244-3035
>                   yaya@dowling.edu <mailto:yaya@dowling.edu>
> =====================================================
> 
> On Tue, 1 Dec 2009, John Westbrook wrote:
> 
>  > Hi James and Herb,
>  >
>  > My comments are not directed at the style of organizing a software
>  > engineering effort, rather my concerns related to standardization
>  > and managing change from the perspective of an archive.
>  >
>  > Circulating a request for comment on community lists carries an
>  > implication impending change.  For PDB users this will be viewed
>  > by many as adding confusion and complexity to a format that is
>  > already viewed as over complicated.
>  >
>  > I think that James' suggestion to post this on the cif-developers
>  > list is perhaps the best path forward.
>  >
>  > John
>  >
>  > James Hester wrote:
>  >> Dear all:
>  >>
>  >> Let's remember that we are only talking about the syntax specification
>  >> here, and as such our audience is almost completely restricted to
>  >> software authors.  I would therefore have thought that the
>  >> cif-developers list would be a suitable forum to post a syntax draft.
>  >>
>  >> The process I had envisioned runs like this:
>  >>
>  >> 1. We finalise a syntax specification in this group - I remind you that
>  >> we only have one or two outstanding issues, both of which could be drawn
>  >> to a close with a vote fairly soon as we have had enough time to air our
>  >> opinions;
>  >>
>  >> 2. The draft syntax specification is posted on the IUCr website together
>  >> with an information page describing brief reasons for our choices, and
>  >> directing people to the archive of our discussions;
>  >>
>  >> 3. Feedback is requested from COMCIFS (who will have to approve it) and
>  >> the cif-developers list (where most of the people directly affected are
>  >> located), and any other appropriate list
>  >>
>  >> 4. We discuss and incorporate this feedback into our final draft, which
>  >> we submit to COMCIFS for approval.
>  >>
>  >> On Tue, Dec 1, 2009 at 1:49 AM, Herbert J. Bernstein
>  >> <yaya@bernstein-plus-sons.com <mailto:yaya@bernstein-plus-sons.com> 
> <mailto:yaya@bernstein-plus-sons.com 
> <mailto:yaya@bernstein-plus-sons.com>>> wrote:
>  >>
>  >>    Dear John,
>  >>
>  >>      I understand your point.  You are espousing the classic 
> hierarchical
>  >>    view of software engineering as seen in joint application design and
>  >>    other
>  >>    middle-manager-to-middle-manager design protocols.  I am 
> espousing the
>  >>    Scandinavian method, also called participatory design, in which
>  >>    implementors deal directly with users.
>  >>
>  >>      Both approaches have their passionate adherents.  I suggest the
>  >>    following compromise to allow us to benefit from the virtues of both
>  >>    approaches:
>  >>
>  >>      1.  Whatever the current state of the design of CIF2 and CIF 
> 1.5, the
>  >>    essential user externals issues be summarized in a "one-pager" to be
>  >>    reviewed and approved by this group, hopefully within the next week.
>  >>
>  >>      2.  Then James, in the name of this group, forward that document
>  >>    to the
>  >>    appropriate managerial level contacts for the stakeholders, 
> asking them
>  >>    for their thoughts and comments, say by the beginning of the new 
> year.
>  >>
>  >>      3.  Then, in response to that feedback, that this group try to 
> have a
>  >>    revised one-pager vetted both by this group and by any interested
>  >>    managerial level contacts, say by the beginning of February.
>  >>
>  >>      4.  Then, the revised one-pager be posted to the appropriate lists
>  >>    for a
>  >>    six-week comment period.
>  >>
>  >>    That should then put us in good shape to have something to 
> discuss with
>  >>    people at the ACA meeting in summer 2010.
>  >>
>  >>      Regards,
>  >>        Herbert
>  >>
>  >>    =====================================================
>  >>      Herbert J. Bernstein, Professor of Computer Science
>  >>        Dowling College, Kramer Science Center, KSC 121
>  >>            Idle Hour Blvd, Oakdale, NY, 11769
>  >>
>  >>                      +1-631-244-3035
>  >>                      yaya@dowling.edu <mailto:yaya@dowling.edu> 
> <mailto:yaya@dowling.edu <mailto:yaya@dowling.edu>>
>  >>    =====================================================
>  >>
>  >>    On Mon, 30 Nov 2009, John Westbrook wrote:
>  >>
>  >>    > The wwPDB will provide collective comment on issues of policy with
>  >>    > respect to its use of CIF.  wwPDB should include the BMRB as an
>  >>    > organization as well.
>  >>    >
>  >>    > Comments to lists regarding changes to CIF or mmCIF need to be 
> vetted
>  >>    > carefully by all of these groups prior to any public announcements.
>  >>    > To give the impression that CIF is suddenly changing may be very
>  >>    > detrimental to the entire CIF/mmCIF enterprise and could well 
> reverse
>  >>    > the progress in adoption of this format.
>  >>    >
>  >>    > Regards,
>  >>    >
>  >>    > John
>  >>    >
>  >>    >
>  >>    > Herbert J. Bernstein wrote:
>  >>    >> Dear Colleagues,
>  >>    >>
>  >>    >>    Before this discussion goes much further on any of its 
> threads, I
>  >>    >> would suggest agreeing on a list of stakeholders and 
> consulting them
>  >>    >> on what has been proposed and trying to come up with an external
>  >>    >> user specification that they understand and agree to.  Some of
>  >>    >> the stakeholders that come to mind are:
>  >>    >>
>  >>    >>    1.  The PDB in Rutgers
>  >>    >>    2.  The PDB in Europe
>  >>    >>    3.  The PDB in Japan
>  >>    >>    4.  People to speak for the Powder diffraction community
>  >>    >>    5.  People to speak for the NMR community
>  >>    >>    6.  The IUCr journal operation
>  >>    >>    7.  CCDC
>  >>    >>    8.  The writers of the various structure solution packages that
>  >>    >> write (and in some cases read) CIF files
>  >>    >>    9.  The writers of visualization programs that read (and in 
> some
>  >>    >> cases write) CIF files
>  >>    >>    10.  The synchrotron data collection community.
>  >>    >>    11.  Service crystallographers
>  >>    >>    12.  Diffraction equipment vendors
>  >>    >>
>  >>    >> I expect I have missed some and hope that others will add to
>  >>    this list.
>  >>    >>
>  >>    >> I would suggest we prepare a summary of the current best
>  >>    definition of
>  >>    >> CIF 2 and, if this groups accepts the idea, CIF 1.5, and send 
> it out
>  >>    >> the lists that cover these stakeholders, starting with the PDB
>  >>    and CCP4
>  >>    >> lists, and see what feedback we get.
>  >>    >>
>  >>    >> Regards,
>  >>    >>    Herbert
>  >>    >>
>  >>    >> =====================================================
>  >>    >>  Herbert J. Bernstein, Professor of Computer Science
>  >>    >>    Dowling College, Kramer Science Center, KSC 121
>  >>    >>          Idle Hour Blvd, Oakdale, NY, 11769
_______________________________________________
ddlm-group mailing list
ddlm-group@iucr.org
http://scripts.iucr.org/mailman/listinfo/ddlm-group

Reply to: [list | sender only]