> 1)
> What is "string" in the BNF in Section 3?
I added "The symbol string matches any string." (A string is defined in
section 2.)
> 2)
>
> <key name="NCName">..
> <keyRef name="NCName">..
>
> should read
>
> <key name="QName">..
> <keyRef name="QName">..
Fixed.
> 3)
>
> The BNF in Section 3 does not allow <div> to have more than one <define>.
>
> I think that the RHS of grammarContent should
> be changed as below:
>
> (start | define | <div>grammarContent</div> |
> <include href="anyURI"> includeContent* </include>)*
>
> This is consistent with the schema in appendix A.
>
> The same change is required for includeContent
Note that the reference to includeContent is followed by a * in the
productions for includeContent and grammarContent. So I don't believe
there's a bug there.
The bug is that there should be a * on the reference to grammarContent in
the production for grammarContent.
> 4)
>
> <name>string</name>
>
> should read
>
> <name>QName</name>
Fixed (in both full grammar and relaxng.rng).
> 5)
>
> The last para of Section 3 mentions QName, NCName, and "combine". On
> the other hand, the second para of 4.2 mentions "name", "type",
> "combine", and "name". They are talking about the same places
> in the BNF. (Section 3 says that leading and trailing whitespaces
> are allowere there and 4.2 says that they are removed by the RELAX NG
> Processor.) I prefer the style of 4.2.
I fixed Section 3 to use the style of Section 4.2.
>
> 6)
>
> The para before last in 4.15 should say that this transforamtion
> is repeated until there is one <define> per name.
Fxed.
> Alternatively, we can
> move 4.11 after 4.15.
That would break (or complicate) 4.12 and 4.13.
> 7)
>
> In 4.18, I would like to insert "directly" between "that" and "has a
> notAllowed child element" and
I don't agree with that. I think it's confusing to use "direct" with
"child" in some places and not in others. Readers will wonder how a
"direct child" is different from a "child".
James
