- From: Erik Johnson <erikj@epicor.com>
- Date: Tue, 24 Jan 2006 07:14:00 -0800
- To: <public-xsd-databinding@w3.org>
- Message-ID: <BF9C5B9528B9C246BC41C7B988C49C87018C4E23@slate.americas.epicor.net>
* Title: Using PSVI Features in Schemas * Description: see below * Target: "Basic"? * Proposal: Explicitly identify these features and specify usage Advise developers to not assume callers have rich PSVI implemented Most toolkits contain schema validation tools that by now have excellent compliance with the XML Schema specification. XML Schema validation tools take an Infoset and XML schema set as input and emits a post-schema-validated-Infoset (PSVI). The problem is that type generators and/or serializers do not actually use proper schema validation mechanisms. Instead, type generators/serializers emulate selected XML Schema features - like xs:default - meant to be managed by the validation process. There are several issues I think this working group should consider: First, is it even wise for a service description to contain use PSVI features in response messages? In the case of xs:default, it seems impolite to return elements with empty content to expect the caller to fulfill the default value requirements. Second, are there other XML Schema features that are meant to impact the PSVI? Perhaps this working group should list such features and describe how type serializers/generators should emulate the intent of the Schema spec. Finally, should the working group tell developers to avoid Schema features that presume the caller has a schema validation capability? Erik Johnson Epicor Software Corp.
Received on Tuesday, 24 January 2006 15:14:13 UTC