• To: xml-dev@l...
• Subject: Re: Deterministic content models???
• From: Richard Tobin <richard@c...>
• Date: Wed, 7 Aug 2002 17:12:50 +0100 (BST)
• Cc:
• In-reply-to: <200208071539.LAA00666@m...>
• Organization: HCRC, University of Edinburgh

In article <200208071539.LAA00666@m...> you write:

>> 42: <!ELEMENT SMARTDOC (HEADER, ((DATA, VIEW*) | (DATA?, VIEW+)),
>> SIGNATURES?)>

>DTDs can't cope with choices of this sort.  This declaration has to be
>loosened to:
>
>	<!ELEMENT SMARTDOC (HEADER, DATA?, VIEW*, SIGNATURES?)>

You can do a bit better than that.  The original declaration allows

 (DATA, VIEW*):
   DATA, no VIEWs
   DATA, some VIEWs
 (DATA?, VIEW+)
   DATA, some VIEWs
   some VIEWS

So use this:

   <!ELEMENT SMARTDOC (HEADER, ((DATA, VIEW*) | VIEW+), SIGNATURES?)>

which allows precisely those cases.

There *are* content models that can't be made deterministic without
allowing undesired combinations, but they're very rare.  The example
usually quoted (I forget who discovered this) is the chess game:

  (WHITE, (BLACK, WHITE)*, BLACK?)

-- Richard

• References:
  • Re: Deterministic content models???
    • From: John Cowan <jcowan@r...>

• Prev by Date: Re: Deterministic content models???
• Next by Date: Re: XML Schema - Enumerated List to Allow Blank Value
• Previous by thread: Re: Deterministic content models???
• Next by thread: Re: Deterministic content models???
• Index(es):
  • Date
  • Thread