Re: [xml-dev] The Goals of XML at 25, and the one thing that XML now nee

[Arjun Ray <arayq2@gmail.com>]

[Default] On Mon, 19 Jul 2021 22:21:32 +1000, Rick Jelliffe
<rjelliffe@allette.com.au> wrote:

| For example, say we added to XML simple typing by delimiters like this
| 
|  <a b="xyz"   c=123 d=false  e=R23456 >...
| 
| where b is a string, c is a number, d is boolean and e is a symbol.

Sorry, attributes are (best used) for metadata.  They should not be
used to analyse wholes into parts.

A thread from long ago on this very subject (and an answer to Roger
C's question upthread):

http://lists.xml.org/archives/xml-dev/200205/msg01027.html 

Compare:

    <a>
        <b type="string">xyz</b>
        <c type="number">123</c>
        <d type="boolean">false<d>
        <e type="symbol">R23456</e>
    </a>

Or even:

    <a>
        <string item="a">xyz</string>
        <number item="b">123</number>
        <boolean item="c">false</boolean>
        <symbol item="e">R23456</symbol>
    </a>

Note also that the 'type' attribute could be elided if the text
content were autoparsed in exactly the same "smart" fashion as called
for in the original proposal.  

Putting data values - which quite often need metadata to interpret
correctly - into attibutes, is jumping into a narrow tight box and
pulling the cover shut over oneself.

The more general point is that XML, being text in the first instance,
always has to be interpreted into some other domain (number, boolean,
farglebarp, whatever), for which metadata of some form or another is
indispensible.  That was the point of notations too (which have died
on the vine in the XML ecosphere).