W3C home > Mailing lists > Public > www-zig@w3.org > May 2002

Re: CCL proposal (quotes)

From: Alan Kent <ajk@mds.rmit.edu.au>
Date: Wed, 8 May 2002 11:28:54 +1000
To: "Z39.50 LISTSERV" <www-zig@w3.org>
Message-ID: <20020508112854.I18271@io.mds.rmit.edu.au>
On Tue, May 07, 2002 at 07:55:41AM -0400, Mark Reichert wrote:
> I knew I had it around somewhere.  Z39.58-1992 is/was clear on this matter.

To me this means the documentation for the Z39.58 regexp attribute
already in Bib-1 is incomplete - the textual description of the pattern
has ommitted the Z39.58 documented support for quotes for releasing.

Note: I don't like \ to release things in a Z39.58 regexp. We have
a CCL parser as according to the now obsolete spec. Introducing \
now would be a non-backwards compatible change to our implementation
that is not inline with Z39.58. (We actually allow users to enable
special \ processing as a preprocess phase to the CCL parsing to
allow UTF-8 etc sequences to be embedded into the CCL query. This
did not require a change to CCL - its a preprocess of user input
before its given to the CCL parser.)

I therefore propose that the Z39.58 regexp attribute as currently
in Z39.50 be corrected to be made conformant with the CCL spec as
Mark quoted in has mail (that is, allow quotes in it). This seems
the semantically correct solution.

Ralph for his non-Z39.58-conformant (but arguably more human friendly)
regexp definition can use a new attribute with a new definition.
Since almost no-one (we do!) supports the current Z39.58 regexp
attribute, introducing a new attribute would not be that much of
a problem would it?

But I think its bad precedent to say "well this attribute was intended
to support this standard, but the standard was not very friendly so
we changed its meaning afterwards to be non-conformant."

Note: I am not picking on Ralph here - at the end of the day it really
does not matter to me that much since almost no-one supports the attribute
(at least that appears to be on this list). So I can change our attribute
easily too, but trying to be objective I would say the current Z39.58
regexp attribute appears to be in error with the spec (so fix it),
and if someone wants a new pattern, fine, introduce a new pattern attribute
for the new pattern rather than change the existing definition.
I don't think personal likes and dislikes (including my own) should
influence the adoption of standards.

Mind you, my objectivity is in question since I am probably the only
other person who has implemented the Z39.58 regexp attribute, and Ralph's
proposed \ for releasing would cause me problems! :-) :-) :-)

If a new 'Ralph' pattern was introduced, we would probably have a go
at supporting it in our server (it would be so similar to our existing
pattern support that it should not be too hard to add into our server).


ps: What humans type in by the way does not have to directly correspond
to what is sent in a regex. Ralph's client could always convert a?45 or
a?4\5 automatically into a?4"5" for sending down the wire.
Received on Tuesday, 7 May 2002 21:29:28 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:26:04 UTC