Ok then, we'll define identifier as term, and so:
1. enclose identifer in quotes
2. define identifier as term
3. Define term as character string
"LeVan,Ralph" wrote:
> If identifier goes into quotes (which I prefer), then it doesn't need any
> special characteristics. I'd just define identifier as being a term.
>
> Term is just a characterstring. I have no idea how to define that in BNF.
> It used to be that you actually listed the characters that comprised your
> characterset. But with Unicode, that isn't possible. So, there must either
> be some atomic token that means all Unicode characters, or we just punt and
> say that.
>
> Ralph
>
> -----Original Message-----
> From: Ray Denenberg [mailto:[log in to unmask]]
> Sent: Thursday, October 03, 2002 9:58 AM
> To: [log in to unmask]
> Subject: cql: identifer and term
>
> Someone (Alan I think) pointed out that
> "identifer" isn't defined.
>
> Nor is it enclosed in quotes, thus instead of:
> result-set-expression ::= "resultSet =" identifier
>
> should be:
> result-set-expression ::= "resultSet ="
> ""identifier""
>
> Also, term isn't defined.
>
> The identifer should be easy to define, as it
> applies only to: Index prefix, index base name,
> and result set name. Alan suggested:
> "a letter followed by zero or more letters,
> digits, or underscores".
> I'll go with that if nobody complains.
>
> Term, however, is still a mystery. I posted a
> message asking for suggestions about how it's to
> be defined, and got alot of "idunno"s. Should we
> leave it undefined? Define it simply as a
> character string?
>
> So for now I plan to:
> 1. Enclose Identifier in quotes.
> 2. Define identifier as "a letter followed by zero
> or more letters, digits, or underscores".
> 3. Define Term as a character string.
>
> --Ray
|