Interesting...

Are you looking at the schema 1.1 draft? I was answering using my memor=
y of
schema 1.0. Seems this is a change between 1.0 and 1.1. I don't recall
off-hand whether this was intended. Need to dig up the history and poss=
ibly
talk to other members of the WG.

In 1.0, bullet 4 of "Element Declaration Properties Correct" refers to
"Type Derivation OK (Complex) (=A73.4.6)". Neither constraint mentions =
the
{prohibited substitutions} property of the base type (type of the
{substitution group affiliation}). So in 1.0, your example schema is va=
lid.
(But during instance validation, the constraint "Substitution Group OK
(Transitive)" is used, which does take into account "block" on the base=

type.)

But in 1.1, bullet 4 refers to the definition of "validly substitutable=
",
which takes {prohibited substitutions} of the base type into account. S=
o in
1.1, your example schema is invalid.

Thanks,
Sandy Gao
XML Technologies, IBM Canada
Editor, W3C XML Schema WG
Member, W3C SML WG
(1-905) 413-3255 T/L 313-3255



                                                                       =
    
             Tobias Koenig                                             =
    
             <tobias.koenig@tr                                         =
    
             olltech.com>                                              =
 To 
             Sent by:                  xmlschema-dev@w...            =
    
             xmlschema-dev-req                                         =
 cc 
             uest@w...                                               =
    
                                                                   Subj=
ect 
                                       Re: block attribute and         =
    
             07/01/2009 09:54          substitution groups             =
    
             AM                                                        =
    
                                                                       =
    
                                                                       =
    
                                                                       =
    
                                                                       =
    
                                                                       =
    





On Wednesday 07 January 2009 15:36:05 Sandy Gao wrote:
Hej Sandy,

> The "validly substitutable" rule is checked when the instance is
validated,
> and not when the schema is built.
Ok, but why is it listed in rule 4) of Element Declaration Properties
Correct
which (as far as the 'Declaration' in the name suggests) should be test=
ed
when the schema is built?

Ciao,
Tobias

=