W3C home > Mailing lists > Public > public-rif-wg@w3.org > June 2008

RE: [BLD] XML syntax for the slots

From: Boley, Harold <Harold.Boley@nrc-cnrc.gc.ca>
Date: Fri, 13 Jun 2008 11:38:31 -0400
Message-ID: <E4D07AB09F5F044299333C8D0FEB45E904FFE15D@nrccenexb1.nrc.ca>
To: "Christian de Sainte Marie" <csma@ilog.fr>
Cc: "RIF WG" <public-rif-wg@w3.org>

Christian,

> Why, in an UNITERM, the argument's name is in a sub-element (<Name>), 
> whereas the slot key is not?

It's also in a sub-element since TERM is just an abbreviation for
<Const>...</Const> or <Var>...</Var> or <Expr>...</Expr> or
<External>...</External>.

E.g., in this UNITERM we have the <Name> sub-element:

<Expr>
...
  <slot ordered="yes">
    <Name>&cpt;author</Name>
    <Var>Author</Var>
  </slot>
. . .
</Expr>

and in this Frame we have the <Const> sub-element:

<Frame>
...
  <slot ordered="yes">
    <Const type="&rif;iri">&cpt;buyer</Const>
    <Var>Buyer</Var>
  </slot>
. . .
</Frame>

In UNITERMs, if we'd never allow a <Name> in filler position,
then slot-child order could be reconstructed from the classes
<Name> vs. non-<Name>. But it would not be a clean thing to do.

In Frames, slot-child order could not be reconstructed anyway,
since there is an arbitrary TERM in both positions.

> A side question is: if we keep it as it is (that is, the argument
names 
> in UNITERMs are in <Name> sub-elements, do we still need the content
of 
> the UNITERM slots to be ordered? That is, do we still need the 
> rif:ordered attribute to be "yes"?

Well, <slot> and <args> also validate if you omit the ordered
attribute -- since these elements ALWAYS have ordered="yes",
the XSD uses a fixed "yes" value for their ordered attribute:
<xs:attribute name="ordered" type="xs:string" fixed="yes"/>

Best,
Harold


-----Original Message-----
From: Christian de Sainte Marie [mailto:csma@ilog.fr] 
Sent: June 13, 2008 4:30 AM
To: Boley, Harold
Cc: RIF WG
Subject: Re: [BLD] XML syntax for the slots

Harold,

Boley, Harold wrote:
> 
> on purpose.
>
> [...] For example, unlike the [argrument names] of named-argument
UNITERMs,
> the [slot keys] of frames can be complex expressions.

Yes, I understand that. My question was more trivial than that :-)

Why, in an UNITERM, the argument's name is in a sub-element (<Name>), 
whereas the slot key is not?

Why not have, in an UNITERM:
   <slot rif:ordered="yes">
      unicodestring
      TERM
   </slot>

and in a Frame:
   <slot rif:ordered="yes">
      TERM
      TERM
   </slot>

A side question is: if we keep it as it is (that is, the argument names 
in UNITERMs are in <Name> sub-elements, do we still need the content of 
the UNITERM slots to be ordered? That is, do we still need the 
rif:ordered attribute to be "yes"?

Cheers,

Christian

Csma wrote:
> in an UNITERM:
>   <slot ordered="yes">
>      <Name>unicodestring1</Name>
>      filler1'
>   </slot>
> 
> and in a Frame:
>   <slot ordered="yes">
>      key1'
>      filler1'
>   </slot>
> 
> Is that on purpose, or is it just oversight?
Received on Friday, 13 June 2008 15:39:14 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 18:33:49 GMT