> > The order is defined (see the schema for exact details). > > We modified the schema to add <import> anyway so I don't think moving it > further up adds any more problems.
I thought that was covered by the <any..> in the spec schema so you're not "breaking" the spec by including the <import.xxx> where you currently have it. > It seems to me that knowing the types for any complex properties is needed > > before parsing the files (.module, .fragment etc). Is the handling of > > complex properties being discussed in the specifiactions group? > > > > It is under discussion, yes. There is a proposal in for a recursive > composition model and as part of that some changes have been proposed > for how properties are defined and set. To my knowledge the inclusion of > SDO types or WSDL definitions has not yet been discussed. > > Whether you need to define the types in advance depends on whether the > parser is able to hand imports during the parse process. I think > requiring the user to do it in advance add inconvenience that can be > avoided. XML-Schema and WDSL both allow imports/includes as part of > their definition and I think SCDL should be similarly self-contained. > This makes the parser a little more complex, but that is our problem not > the user's. Agreed but (you'll like this...) because we use SDO to load/parse the assembly artifacts we'll have to do a pre-parse to pull out the types. Hmmm... why don't we write our own loader using... um, er. Stax? Anyho, I have no objection to extending the spec schema and if we need to alter the schema we need to feed that back to the spec. Maybe it's a <types> <schema...> as in wsdl or an extensible <import.xxxxx > -- Pete
