Re: “Consensus” grammar(s)

At least this rules out the "too complex to do easily" objections.

I think you should make this a PR that we can vote on in preparation for v1 that resolves all of these issues/proposals.

_________________
Tomos Hillman
eXpertML Ltd
+44 7793 242058
On 23 Apr 2022, 15:46 +0100, Norm Tovey-Walsh <norm@saxonica.com>, wrote:
> Hi folks,
>
> I was hoping we’d get closure on the consensus proposals before the
> weekend, but that didn’t happen. I think there is support for
>
> * Changing ~ to !
> * Using = and | exclusively
> * A version declaration
> * Using + for insertions
>
> No objections have been made, and each proposal has received at least a
> few nods of support.
>
> I wondered what that might look like, so I wrote up a grammar for it. It
> looks a bit different on the surface, because the delimiters have
> changed, but is otherwise very similar.
>
> I was pleased that it was easy to make the prolog entirely optional. If
> you don’t specify a version declaration, the XML serialization of an
> ixml grammar is unchanged.
>
> For fun, I also extended it to include the namespaces proposal. Again,
> the prolog is entirely optional. If you don’t declare a version or any
> namespaces, the XML serialization of a grammar is unchanged from before.
> If you don’t want or care about namespaces, you can just completely
> ignore them.
>
> Namespace support only adds a few new nonterminals. I made some
> aesthetic choices that we could quibble over. A grammar with a version
> and namespaces under my example grammar looks like this:
>
> ixml version "1.0" .
> default namespace = "http://docbook.org/ns/docbook" .
> namespace mml = "http://www.w3.org/1998/Math/MathML" .
>
> mml:foo = 'foo' .
>
> and serializes like this:
>
> <ixml>
> <prolog>
> <version>1.0</version>
> <namespace uri="http://docbook.org/ns/docbook"/>
> <namespace prefix="mml" uri="http://www.w3.org/1998/Math/MathML"/>
> </prolog>
> <rule name="mml:foo">
> <alt>
> <literal string="foo"/>
> </alt>
> </rule>
> </ixml>
>
> I didn’t try to make it syntactically impossible to specify more than
> one default namespace. I think that’s easier handled as a semantic
> constraint, like the actual prefixes used in nonterminals.
>
> (If you want to play with the grammars with NineML, you’ll have to “back
> out” the change from “~” to “!” since the NineML parser expects “~”.)
>
>
>
> Be seeing you,
> norm
>
> --
> Norm Tovey-Walsh
> Saxonica