Yes I think Ralph's explanation of his view of
indexing helps.  We need a few more answers before
I can construct a concrete proposal. First some
preliminary observations:

1. srw indexes must map to Z39.50 attribute
combinations.  I understand the philosophical
arguments against this, but as far as I'm
concerned it's not an option. If we can't tell
people how to map an SRW search to Z39.50 then we
will lose much of the basis for srw's existence.

2. We need clear, quality dc index definitions (in
terms of Z39.50 attribute combinations).  The bath
definitions are easy, we just point to the profile
(and so for bath we don't need to provide
mappings).  We need to get the DC definitions as
good as possible, before we make them public, even
as draft for review.

3. If we accept Ralph (and Rob K.) 's view of
string and word indexes, I don't really see that
this presents major limitations for those who take
the different (I.e. Rob S.'s) view.  If you want
to accept only single-word terms, fine.

I think a solution is to provide yet an additional
mapping in the bib-1 case: Ralph began this
discussion by observing that I'd proposed a
structure attribute value of 2, when what he
really wants is some sort of wordList value (the
to-be-proposed adjacentWords).   So lets include
both mappings. Explain will tell which are
supported.

But Ralph, for the AA case, we have
Left-truncation-on-word-boundary and
Right-truncation-on-word-boundary. How come you
didn't raise an objection? (Didn't notice that?)

Following along this path, we would add an AA
mapping with format/Structure adjacentWords.

Regarding yesterday's suggestion about the two
trucation attributes (which as Ralph observes
should be truncation, not structure -- I forgot
there was a bib-1 truncation attribute -- I've
been looking too much at AA sets, where there
isn't such a thing)  depending on how the above
question(s) resolve(s) we may not need them.
Same for  word masking.

Need responses to these points before I can
construct a concrete proposal.


--Ray