XML element replacement

with-attributes.xml

<?xml version="1.0" encoding="UTF-8"?>

<document>

    <meta>
        
        <title replaced-by="1">Some text</title>
        <full-title id="1">Some text</full-title>
        
    </meta>

    <content>

        <p>Some <italic replaced-by="2"><em id="2">inline</em></italic> content</p>

        <image replaced-by="3">
            <src>http://www.example.com/image.jpg</src>
        </image>

        <figure id="3">
            <src>http://www.example.com/image.jpg</src>
        </figure>

    </content>

</document>

without-attributes.xml

<?xml version="1.0" encoding="UTF-8"?>

<document>

    <meta>
        
        <title>Some text</title>
        <full-title>Some text</full-title>
        
    </meta>

    <content>

        <p>Some <italic><em>inline</em></italic> content</p>

        <image>
            <src>http://www.example.com/image.jpg</src>
        </image>

        <figure>
            <src>http://www.example.com/image.jpg</src>
        </figure>

    </content>

</document>

These are the three types of changes I can imagine:

1. Replacing an element in a data context.
2. Replacing an inline element.
3. Replacing a 'block' element.

2 and 3 look unclear without the attribute. The spec is currently saying that you have to just ignore things you don't recognise, but this is hard to do for Libero-namespaced code as we should support any Libero-namespaced things, which covers both.

This attributes usage seems reasonable and clear.

Being a pointer from old to new, it supports the joining case but not the splitting case, I think:

<figure replaced-by="4?">
<image>
<src>http://www.example.com/image.jpg</src>
</image>
<caption>Lorem ipsum</caption>
</figure>

<image id="4a">
<src>http://www.example.com/image.jpg</src>
</image>
<caption id="4b">Lorem ipsum</caption>

(tag names are hypothetical)

Can use IDREFS rather than IDREF to allow referencing multiple IDs in that case:

<figure replaced-by="4a 4b">
<image>
<src>http://www.example.com/image.jpg</src>
</image>
<caption>Lorem ipsum</caption>
</figure>

<image id="4a">
<src>http://www.example.com/image.jpg</src>
</image>
<caption id="4b">Lorem ipsum</caption>

Wonder about bi-directional?

<?xml version="1.0" encoding="UTF-8"?>

<document>

    <meta>
        
        <title replaced-by="1" id="i">Some text</title>
        <full-title id="1" replaces="i">Some text</full-title>
        
    </meta>

    <content>

        <p>Some <italic replaced-by="2" id="ii"><em id="2" replaces="ii">inline</em></italic> content</p>

        <image replaced-by="3" id="iii">
            <src>http://www.example.com/image.jpg</src>
        </image>

        <figure id="3" replaces="iii">
            <src>http://www.example.com/image.jpg</src>
        </figure>

        <figure replaced-by="4a 4b" id="iv">
            <image>
                <src>http://www.example.com/image.jpg</src>
            </image>
            <caption>Lorem ipsum</caption>
        </figure>

        <image id="4a" replaces="iv">
            <src>http://www.example.com/image.jpg</src>
        </image>
        <caption id="4b" replaces="iv">Lorem ipsum</caption>

    </content>

</document>