This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
Throughout our specifications, we use the term "undefined", without defining the term, to mean that a particular variable or property has no value and that any attempt to access the value is an error. Notably, we use this for the context item and other parts of the focus, and for the typed value of an element with element-only content. This meaning of the term "undefined" is not intuitive, and it runs contrary to usage in some other standards where it is used to mean something akin to our "implementation-dependent". Some readers have been known to mistakenly read it as meaning "not defined in this specification". If we are going to use the term, we should define what we mean by it. Preferably, we should use a different term, for example the XSD specifications use "absent". I would suggest this as a change to be made across the 2.1/1.1 specifications. I have raised it against XPath 2.0 for want of a more suitable locus.
I prefer to keep the term "undefined" and define it, linking to the definition. Also, there is no one error code for accessing something that is undefined. Where it is an error, we provide such an error code. Jonathan
Action A-384-04 I propose that where we use the word "undefined" we link it to the following definition: [Definition: In certain situations a value is said to be *undefined* (for example, the value of the context item, or the typed value of an element node). This term indicates that the property in question has no value and that any attempt to use its value results in an error.] I will leave it to editors to decide where the definition should be inserted.
I understand that at the recent F2F meeting it was decided to add the required definition of "undefined" to other documents besides XPath and XQuery. I just thought I'd record here that Serialization never uses the term.
The XSL WG accepted MKay's proposal to resolve this bug by accepting the wording stated in Comment #2 below.