Re: Attribute Architecture proposal

I think Mike and Rob have agreed on format/structure becoming string/words
in the Utility attribute set (as distinct from the complete proposal
I put up verbatim).

Sleeping on it overnight, I think adding a new 'type' to Class 1 might
be a bit more extreme than people may be willing to bear, so I am
changing tack slightly and think maybe its less impact to move the
All/Any/Adj Words attributes into expansion/interpretation. I have
grabbed a copy of the Utility attribute set definition from the LOC
web site and had a go at marking up some new text. Changes are in
green and red (with strikeout).

    http://www.mds.rmit.edu.au/~ajk/z39.50/util.html

I found a few other attributes had crept in from the time I downloaded
it last time (which might have been a long time ago). In particular,
Left/RightTruncateEachWord and LeftAnchored. There is no definition
associated with these attributes and they overlap with the new stuff
being proposed, so I deleted them in my proposal. Note: I have also
said that 'Left/Right trunc on character boundary' would be applied to each
word if format/structure is 'Words', which is why I removed those
attributes. Maybe LeftAnchored is still relevant, if someone can
define what it means (eg: for All Words, one of them must be at the
start, for Any Words one of the matching words must be at the start,
or maybe its only relevant for Adjacent Words - but I would like to
remove it unless there is a definition put forward that meets a known need.)
Note: there is no RightAnchored, so I am guessing its intended only
for word based queries. Also note that it does overlap with
a value of type string with a 'right truncate at word boundary'.
(In my implementation they would be implemented differently however
as one would hit the 'string' index and the other would hit the 'word'
index.)

I welcome any comments on the proposed text. My fundamental goal is to
get Any/All/Adj out of format/structure and replace it with String/Words.
The finer details I am open to suggestions on the best way to do.

Rob, I am a bit out of touch with CQL - is what I have written up going
to be easily compatible with CQL? Any suggested changes to make it
more compatible?

Thanks
Alan

Received on Tuesday, 8 July 2003 22:02:44 UTC