Hi,
Here's some comments, on comments.
page 6 -
"Comments SHALL be supported using the /* ... */ syntax to delimit
comments.
Comments are only supported before and after the main query. "
Not allowing comments within queries is counter-intuitive and reduces their usefulness. The accepted model for comments in programming or query languages is that they may occur anywhere. The syntax of a comment is usually defined at the lexographical level, so that they don't form a part of the BNF grammar of the language.
Once comments are included in the BNF, where they occur is necessarily limited. I can only think that this has been done because you wish to preserve comment information in the translation from adql/ s to adql/x (probably a good idea), and so define an xml tag to represent a comment? (I can't verify this, as the link from this document to the associated xml schema is broken )
If this is the case, I don't think using an xml tag to model a comment is the correct approach - a processing instruction would be more suitable, as it then does not occur in the XSD schema, and can be placed anywhere within the adql/x version of the query without complicating the schema.
According to Elliot Rusty Harold's book (can't find reference at the moment), this is an acceptable use of a processing instruction.
--- BTW I think the adql/x schema should be inline as an appendix in the final version of this document - as well as a stand-alone xsd file - as defining this is just as important as defining the adql/s presentation of the language.Received on 2006-07-04Z00:08:26