1848 2005-08-16 10:15:16 +0000 [XQuery] Module Imports 2005-08-31 22:41:07 +0000 1 1 1 Unclassified XPath / XQuery / XSLT XQuery 1.0 Last Call drafts PC Windows XP CLOSED FIXED P2 minor --- 1 mike chamberl public-qt-comments oldest_to_newest 5409 0 mike 2005-08-16 10:15:17 +0000 This is a minor technical point which can probably be handled editorially. We allow a module to do an "import module" on its own namespace. We say that the effect of this is to import all modules for that namespace. This includes the importing module. We say that it's an error if the imported module contains functions or variables that clash with those defined in the importing module. But when a module imports itself, there will always be such a clash. Solution: we should specify that when a module does an "import module" on its own namespace, it should import all modules for that namespace *other than itself*. (We have a bit of a problem with module identity: how do we know whether two URIs identify the same module? But I suggest we leave implementations to sort that one out.) 5411 1 mike 2005-08-16 18:01:38 +0000 While on the topic, it would be useful to clarify whether a module that imports its own namespace is allowed to contain forwards references to its own global variables. You can read the spec either way on this: on the one hand it says forwards references to global variables aren't allowed; on the other hand it says you can access all the global variables declared in a namespace that you've imported. It's a little odd if it isn't allowed: if cyclic references to variables are allowed across modules, why not allow them within a module? It's also difficult to explain (and implement!) that a module has access to all the global variables in an imported namespace except those that are declared in the same module and after the variable reference. 5624 2 chamberl 2005-08-31 21:48:48 +0000 The original bug report, about a module importing its own namespace, was accepted by the Query working group on 31 Aug 2005, and the proposed change has been implemented. I will mark this bug as "fixed". The working group felt that the additional comment about forward references to global variables required further discussion. Mike, it is suggested that you open a separate bug report for this issue. --Don Chamberlin 5629 3 chamberl 2005-08-31 22:41:07 +0000 (Done. The new bug report is number 1928.)