Print

Print


Mike Taylor wrote:

>>There isn't a mapping between CCL and CQL, but if you'd like to write
>>one up, then I'm sure that it would be appreciated by many :)
>>    
>>
>>>CCL is very poorly documented - you can use it in many OPACs in
>>>some way but libraries seem to hide it from the user. If you send
>>>me a PDF of ISO 8777 I can try to summarize the differences and
>>>commonalities with CQL.
>>>      
>>>
>>Unfortuantely, ISO8777 doesn't even describe the language that lucidly 
>>-- the stnadard predates context-free grammars (at least I hope it does 
>>-- otherwise, I don't know what their excuse would be).
>>    
>>
>
>I think that excuse is pretty unlikely to stick.  According to the
>Wikipedia article on BNF,
>	http://en.wikipedia.org/wiki/Backus-Naur_form
>it was invented as part of creating the rules for Algol 60.  So
>presumably it goes back to the 1950s.  Whereas the most recent version
>of ISO 8777 is dated 1993, according to the ISO store:
>	http://www.iso.org/iso/en/CatalogueDetailPage.CatalogueDetail?CSNUMBER=16192
>(It would be interesting to know the date of the first edition,
>though.)
>  
>
Yep.. and the ISO standard is based on other standards that go further
back... I've heard rumors that it was originally a Danish specification,
which would be particularly ironic since one of the fathers of Algol and
BNF (Peter Naur) was a Dane.. still legendary around the institute of
Computer Science in Copenhagen for his draconian approach to grading papers.

>Adam put quite a bit of effort into capturing the essence of the
>language in a BNF notation.. that might actually be the easiest
>place to start. You'll find it at
>http://www.indexdata.dk/yaz/doc/tools.tkl#CCL
>  
>
>
>(The opening sentence of this section is my favourite in the whole of
>the YAZ manual, and a classic example of phlegmatic Danish
>understatement: "Not all users enjoy typing in prefix query structures
>and numerical attribute values" :-)
>  
>
We do enjoy the occasional, carefully chosen understatement.

>But to summarise the relationship between CCL and CQL: the intention
>when CQL was being design (initially by Ralph, and subsequently by the
>SRW editorial board as a whole) was that it should be pretty much a
>superset of CCL, and that it therefore sets out to provide
>_additional_ syntax rather than _alternative_ syntax.  So what you
>would hope to find is that most CCL queries are also CQL queries with
>the same interpretation, but that you go on to say much more in CQL.
>  
>
AFAIK, one of the crucial differences (unless this was changed in CQL?)
is that the CCL spec   doesn't require quotes around multi-word terms...
this makes the grammar and parsing of CCL queries somewhat more challenging.

--Sebastian

-- 
Sebastian Hammer, Index Data (US)
[log in to unmask]   www.indexdata.com
Ph: (603) 209-6853