- From: Mark Reichert <markr@sirs.com>
- Date: Tue, 7 May 2002 10:43:57 -0400
- To: "Z39.50 LISTSERV" <www-zig@w3.org>
Now that Z39.58 is no longer, there's no one to complain to. You can do whatever you like if you have no need to support what Z39.58 was. ----- Original Message ----- > Okay, I'll just come right out and say it. I hate the idea of double-quotes > inside of strings! I'm sure I'm just being old fashioned, but it feels like > a parsing nightmare. Whatever happened to the good old backslash to escape > things? I have no problems with a proposal that 123?4\56 means that the > five is a literal and not a digit and should be ignored by the question-mark > processor. But I don't like 123?4"5"6. > > Sorry. > > Ralph > > "No sir, I don't like it!" > The Horse from Ren and Stimpy > > > -----Original Message----- > > From: Mark Reichert [mailto:markr@sirs.com] > > Sent: Tuesday, May 07, 2002 7:56 AM > > To: Z39.50 LISTSERV > > Subject: Re: CCL proposal (quotes) > > > > > > I knew I had it around somewhere. Z39.58-1992 is/was clear > > on this matter. > > > > 7.7.2.1 > > > > ... > > When ? is immediately followed by a positive integer, it > > shall be used to > > indicate a limited range of characters to be masked, from > > zero up to and > > including the specified integer.... To search embedded > > numbers, restoration > > marks are required. See Section 7.7.7. > > > > 7.7.7 > > > > In order to use a reserved command word, abbreviation, > > symbol, or operator > > as a search word, double quotation marks, " ", shall be used > > to restore its > > literal meaning.... > > > > FIND 0?10"5" // ten zeroes followed by a five (my example) > > FIND C?"14" // word beginning with C, ending in 14 (from > > Z39.58 appendix) > > > > # has no interaction with digits: Multiple #s shall be used > > to indicate > > that a precise number of characters greater than one and qual > > to the number > > of # symbols are to be masked (7.7.2.2). > > > > The standard never offered an explicit explanation/example of > > restoring ", > > but presumably by 7.7.2.1... > > > > FIND """Some text in quotes""" > > > > There is no mention of the more typical "" escaping. > > > > The portion of a <search_term> that corresponds to restoration is: > > > > {<restoration><word>[<space><word>]...<restoration>} > > > > <restoration> ::= [<space>}<">[<space>] > > <space> ::= < >[< >}... > > <word> ::= {<char>|<var_mask>|<exact_mask>}... > > <var_mask> ::= <?>[<positive_integer>] > > <exact_mask> ::= <#>[<#>]... > > <char> ::= <any_searchable_char> > > <any_searchable_char> ::= any character locally defined as searchable > > > > > > ----- Original Message ----- > > > > > Not making it to the ZIG, someone sent me some private mail > > indicating > > > that Ralph's proposed single digit after '?' change got accepted > > > and possibly no-one mentioned my counter double quotes suggestion. > > > Fair enough, if you don't turn up you have less influence. > > > > > > Just thought I would have a last bash at a compromise with the idea > > > that if the CCL regexp is changing, may as well try and get as many > > > changes in as possible in one hit rather than change it again later. > > > > > > To repeat the problem I currently have with the CCL regexp is that > > > you cannot specify '?' or '#' as literal text (ie, release their > > > special meaning). So even if there is now allowed only to be a > > > single digit after '?', while the spec is being changed is it worth > > > allowing double quotes ('"') to be used to release special chars > > > anyway? This would allow 'find all terms starting with "#"'. > > > At present, you cannot do this with the CCL regexp. Normally > > > regexp's have release mechanism ( \ for regexp-1 I believe). > > > CCL uses " as a release mechanism so seemed the natural thing > > > to use in the CCL regexp (rather than \ which in CCL has no > > > special meaning). > > > > > > It seems an oversight not to allow searching for serial numbers etc > > > using patterns. > > > > > > #41434 > > > #53423 > > > > > > If people have to change their CCL regexp implementation anyway, > > > I would rather do both changes at the same time and make it possible > > > to search for all possible characters. > > > > > > I wonder also if the Z39.58/CCL regexp attribute needs to be renamed > > > to indicate that it no longer conforms to CCL. I don't actually have > > > a copy of Z39.58, but if its anything like the ISO version of CCL > > > the spec is so woolly that it isn't funny! The formal grammar is > > > given by examples only, and the examples contradict themselves > > > in places! (Mind you, the copy I have of ISO8777 is pretty old now > > > so maybe its been improved.) Not stressed, just thought it was the > > > correct time to at least ask the question. > > > > > > Alan > > > -- > > > Alan Kent (mailto:ajk@mds.rmit.edu.au, > > http://www.mds.rmit.edu.au/~ajk/) > > > Project: TeraText Technical Director, InQuirion Pty Ltd > > (www.inquirion.com) > > > Postal: Multimedia Database Systems, RMIT, GPO Box 2476V, > > Melbourne 3001. > > > Where: RMIT MDS, Bld 91, Level 3, 110 Victoria St, Carlton 3053, VIC > > Australia. > > > Phone: +61 3 9925 4114 Reception: +61 3 9925 4099 Fax: > > +61 3 9925 4098 > >
Received on Tuesday, 7 May 2002 10:44:45 UTC