> Date: Tue, 28 Jun 2005 10:35:09 -0500
> From: Eric Lease Morgan <[log in to unmask]>
> SRW/U next steps
Thanks, Eric, both for leading this discussion and for posting the
summary. I have only two comments to make, one technical and one
First point -- technical:
> Learning from OAI's success
> 1. a well-defined problem statement - OAI did and
> does not try to be everything to everybody. It is a
> one-trick pony allowing people to expose and harvest
I will go further and say that an enormous amount of the credit for
OAI's success must go to the specifications, which are contained in:
* a single document, which
* avoids abstractions, and
* is replete with examples
In stark contrast, someone coming to SRW/U has to read _a lot_ of
documentation, and understand a lot of abstractions. We are not too
bad on the examples, but we could do better.
My contention is that much of the perceived complexity, which requires
the multiplicity of documents, is down to SRW (as opposed to SRU). I
draw attention particularly to the affect that the WSDL file for SRW
requires SIX other schema files. Fine for computer scientists, not so
fine for software engineers, even less fine for decision-making
executives. If we focussed exclusively on the URL-based version of
our protocol, things would simplify remarkably. In particular, we
could then write a single document that avoids abstractions and is
replete with examples.
This is what lies behind the question I asked on Day Two of the ZING
Forum: "Why do we need both SRU and SRW?" I found it _very_ telling
that when I asked the follow-up question, "Who needs SRW?", _no-one_
was willing to step forward and say they are using it for anything. I
think Ralph's experiments with Beowulf-clustering ZING servers also
show pretty conclusively that the technical arguments in favour of SRU
are stronger than those for SRW.
(The only person who argued on Day Two that SRW should be retained was
Matthew. I do feel for him, as it has been his thankless task to
maintain the WSDL and schemas for all this time, and it must be hard
to imagine that work being discarded. Still and all, I don't think
that would be reason enough for the ZING initiative to keep carting
around baggage that no-one is actually using.)
So my modest-but-radical proposal is that we dump SRW completely, and
then simplify the documentation of SRU down as far as possible. Also,
we could then give SRU a nicer, non-TLA name, such as SearchURL.
I welcome discussion.
Second point -- organisational:
> "What can we do to make more people aware of the benefits of this
> search and retrieve protocol?" Some of the ideas included:
> * brand it
> * create five-minute Powerpoint show
> * create X profile
> * get grant funding
> * implement a conformance tester
> * make it a formal standard
> * make X more SRW/U services available
> * write "gentle" guides
The fourth of these, "get grant funding", is not really something to
do to make people aware of SRU; it's more of an enabler for the
others. It was I who raised this subject on Day One, due to the
furstration of being a small software house that has lots to give to
this kind of initiative, but no way to fund more involvement -- in
fact, I keep thinking I need to do _less_ on SRU rather than more.
Specifically, given a reasonable level of funding, Index Data people
including myself could do a pretty darned good job of making
presentations at conferences, creating profiles, implementing
conformance testers, raising running SRU services and writing "gentle
guide" documents -- I think we have the track record to prove that.
But we simply cannot plough the time and effort into these activities
that they demand when an organisation our size doesn't have the cash
slopping around to fund such things.
What SRU really needs is for a big, well-funded and highly visible
organisation (e.g. OCLC, ProQuest, Serial Solutions) to get right
behind it and start to invest a modest amount in making it happen. I
am pretty confident that if that happened, then people would step up
to the plate to offer their services in SRU evangelism (and, no, I
don't just mean Index Data people!)
Hope that's thought-provoking.
/o ) \/ Mike Taylor <[log in to unmask]> http://www.miketaylor.org.uk
)_v__/\ "First fork() emulation and now select()? If we are not careful,
in ten years or so NT/W2K/W2010 will be almost as useful as UNIX
was in mid-1980s" -- Jarkko Hietaniemi.
Listen to free demos of soundtrack music for film, TV and radio