On 7 September 2010 15:50, Jonathan Rochkind <[log in to unmask]> wrote:
> Mike Taylor wrote:
>>
>> No, there is no such validator.  That is because the formal
>> specification of CQL was developed separately from, and subsequent to,
>> the actual implementation.  It also has several bugs in it.  All in
>> all, I would advice ignoring the formal specification completely.
>
> Um, in favor of what? If not the formal specification, what else could one
> use to determine what is and is not legal CQL?

I am not necessarily advocating this approach, but the most common way
of producing implementations has been to port CQL-Java to other
languages. So that's an option.

> I get that you are not happy with the formal specification, and think it
> should have been otherwise.  But unless you want to try to develop your own
> and convince people to use your own instead of the current one, I'm not sure
> what good it does to tell people to ignore the specification.

You're right, I should publish an alternative.

Not likely to happen right now, though.

-- Mike.


>
> Jonathan
>
>