[rfc-i] draft-xml2rfc-21 error in rnc

cabo at tzi.org (Carsten Bormann) Wed, 02 September 2015 05:03 UTC

From: "cabo at tzi.org"
Date: Wed, 02 Sep 2015 07:03:11 +0200
Subject: [rfc-i] draft-xml2rfc-21 error in rnc
In-Reply-To: <EDB64F32-55E6-461C-B820-D71E7C88B5CE@vpnc.org>
References: <20150831103117.GA11731@miek.nl> <FB9DAEB0-A385-49D0-AE96-30F16B6AF52D@vpnc.org> <55E4DEAD.7040306@tzi.org> <EDB64F32-55E6-461C-B820-D71E7C88B5CE@vpnc.org>
Message-ID: <55E6830F.20200@tzi.org>

My Relax-NG fu may not be strong enough to lead you out of this swamp...

A first step might be to use

  element ns1:*

instead of

  element *

In the long run, I'd expect a full SVG grammar to appear here.

Gr??e, Carsten



Paul Hoffman wrote:
> On 31 Aug 2015, at 16:09, Carsten Bormann wrote:
> 
>> The problem appears to be this snippet:
>>
>> anything =
>>   element * {
>>     attribute * { text }*,
>>     anything*
>>   }
>>
>> This essentially says, all elements have all attributes with text
>> structure.
>> This of course then conflicts with all attributes that are not defined
>> to have text structure.
> 
> I'm weak on this level of RNG. The RNG that generated this RNC is:
> 
>   <define name="anything">
>     <element>
>       <anyName/>
>       <zeroOrMore>
>         <attribute>
>           <anyName/>
>         </attribute>
>       </zeroOrMore>
>       <zeroOrMore>
>         <ref name="anything"/>
>       </zeroOrMore>
>     </element>
>   </define>
> 
> The "anything" element appears as a child of only one element, <svg>.
> The idea is that <svg> can have any element under it, not just the ones
> we define in this spec. Clues for how this should be better stated are
> appreciated.
> 
> --Paul Hoffman
>