W3C home > Mailing lists > Public > public-exi@w3.org > February 2016

Re: AW: XHTML representation

From: Brett Zamir <brettz9@yahoo.com>
Date: Thu, 18 Feb 2016 18:29:41 -0700
To: "Peintner, Daniel (ext)" <daniel.peintner.ext@siemens.com>, "public-exi@w3.org" <public-exi@w3.org>
Message-ID: <56C67005.8060908@yahoo.com>
Hi Daniel,

On 2/18/2016 5:58 AM, Peintner, Daniel (ext) wrote:
> I do understand the desire of having a readable format without CSS and/or XSLT and I think it is a valid approach.
> Besides I think your approach suffers in a similar way as the current solution does. Our ultimate goal is to have an XML schema (or respectively EXI grammars) that describes the expected content. Assuming we have such a schema, EXI offers even better processing&compaction results.
> In your mapping the element <dt> defines the key while <dd> is meant to hold the value. Correct?

FYI, I've added a generic XML Schema for JHTML now at 

In order to provide as succinct a format as possible, I've chosen to 
represent strings without needing a parent element (unless the JSON is 
merely a string, in which case, the single element of the document will 
be a span element).

Due to limitations with XML Schema, however, I believe it is not 
possible, at least in XML Schema 1.0, to indicate that the elements (in 
our case, dd or li) must only allow for either text strings or child 
elements (in our case, text or i elements containing 
null/booleans/numbers). The current schema therefore must (mistakenly) 
allow for mixed content containing both text and elements.

With the assert element in XML Schema 1.1, however, I think it would be 
possible to use XPath to confirm that either text alone is present or 
elements without text are present but not both. I think the condition 
could be done with something like this:

     <xs:assert test="dd[not(./text() and *)]" />

(The online validators I have found are either 1.0 or apparently rely on 
Java being in the browser and no JavaScript libraries advertising 1.1 

(One other limitation of JHTML...In order to maintain my design choices 
such as to avoid hidden attributes, and due to the removal from the CSS3 
Selectors spec of the :contains() pseudo-class, there is currently no 
way to distinctively style null, boolean, or numbers from one another 
(all represented as the brief <i>), but that could easily be done, e.g., 
with XSLT adding some attributes.)
> A "person" example could be the following:
> <dl itemscope="" itemtype="http://brett-zamir.me/ns/microdata/json-as-html/2" xmlns="http:/www.w3.org/1999/xhtml">
>      <dt>id</dt>
>      <dd><i>1</i></dd>
>      <dt>lastname</dt>
>      <dd>Peintner</dd>
>      <dt>firstname</dt>
>      <dd>Daniel</dd>
> </dl>
> Let’s assume one would like to restrict the content in a way so that it consist of a key “id” with a number, a required “lastname” typed as string and an optional “firstname” typed as string.
> Please state me wrong but I think it is not possible to restrict the above XML instance in such a way by simply using XML schema constructs.
> Do you see a possibility?
In your specific case, you could simplify my schema of course, but not 
avoid the one shortcoming mentioned above for my generic schema 
regarding mixed content (if that's the issue you were referencing):

<?xml version="1.0"?>

<xs:attributeGroup name="rootAtts">
    <xs:attribute name="itemscope" type="xs:string" use="required" 
     <xs:attribute name="itemtype" type="xs:string" use="required" 

<xs:simpleType name="decimal">
     <xs:restriction base="xs:decimal"/>

<xs:element name="dl">
         <xs:sequence minOccurs="3" maxOccurs="3">
             <xs:element name="dt" type="xs:string"/>
             <xs:element name="dd">
                 <xs:complexType mixed="true">
                         <xs:element name="i" type="decimal" minOccurs="0"/>
         <xs:attributeGroup ref="rootAtts"/>


Best wishes,
Received on Friday, 19 February 2016 01:30:24 UTC

This archive was generated by hypermail 2.3.1 : Friday, 19 February 2016 01:30:25 UTC