RE: XML Syntax Issues / root element

> [  I wish the wiki could do color-coded changes in context.  :-(  ]

Right, but the diff feature is already quite helpful:
http://www.w3.org/2005/rules/wg/wiki/Arch/XML_Syntax_Issues_2?action=dif
f&rev2=26&rev1=25

> 1.  Under "Is the XML root element 'rdf:RDF'?", I think you ...

I think there is no big difference between XML editors and XML
validators here -- if you meant dialect URIs are only *added*,
then also XML editors *could* find them at the other place.

> Why are you using "type" (eg rif:type) instead of xsi:type there?

For uniformity reasons: rif:type is more general than xsi:type.
For example, we also have rif:type="rif:local".

> More on that -- Using attributes on Const vs. using subclass of
> Const is a style issue, as I understand it.

By analogy, your

<Var id="var_x" />

is preferable to something like

<Var>
  <identifier>var_x</identifier>
</Var>

-- Harold


-----Original Message-----
From: Sandro Hawke [mailto:sandro@w3.org] 
Sent: Sunday, September 23, 2007 9:27 PM
To: Boley, Harold
Cc: public-rif-wg@w3.org
Subject: XML Syntax Issues / root element


I'm thinking about all the changes you just made to 
http://www.w3.org/2005/rules/wg/wiki/Arch/XML_Syntax_Issues_2

[  I wish the wiki could do color-coded changes in context.  :-(  ]

Two quick questions:

1.  Under "Is the XML root element 'rdf:RDF'?", I think you
    misunderstood the third option.  The third option, "use something
    dialect-specific", would mean, for instance, using "rif:BLD" as the
    XML root element (for BLD documents).

    I think we've always had consensus that the dialect-of-authoring
    will always be identified in the document, so the advantage you
added
    (allowing implementation to select the appropriate tools) is not
    *added* by making the dialect name be the root element; it was there
    all along.    I think the only advantage of making it the root
    element is that XML tools that know nothing about RIF (eg
    schema-driven editors) might make better use of it that way, but
    that was the reason I already listed.

2.  (this is really a comment on BLD)

    <arg><Const type="xsd:long">49</Const></arg>

    Why are you using "type" (eg rif:type) instead of xsi:type there?
    Is that on purpose?

    More on that -- Using attributes on Const vs. using subclass of
    Const is a style issue, as I understand it.  What's the reason for
    doing it this way?  (I guess it corresponds to the terse approach in
    the formal syntax?)

In general, it's looking like that issues list needs to be refactored
again.  Your proposal uses attributes and uses this implied "name"
property.  Those probably need to be called out as specific issues.  Are
dialect designers to be free to put things in attributes when they want,
or is "type" somehow fundamental and shared across all dialects?

      -- Sandro

Received on Monday, 24 September 2007 01:13:26 UTC