| Summary: | No error when parsing untyped variable | ||||||
|---|---|---|---|---|---|---|---|
| Product: | [Modeling] OCL | Reporter: | Wayne <wdiu> | ||||
| Component: | Core | Assignee: | OCL Inbox <mdt-ocl-inbox> | ||||
| Status: | CLOSED WONTFIX | QA Contact: | |||||
| Severity: | normal | ||||||
| Priority: | P3 | CC: | ed | ||||
| Version: | unspecified | ||||||
| Target Milestone: | --- | ||||||
| Hardware: | PC | ||||||
| OS: | Windows XP | ||||||
| Whiteboard: | |||||||
| Attachments: |
|
||||||
|
Description
Wayne
Created attachment 170877 [details]
Patch to fix the problem
Your patch generates an analysis error when a meta-model property has no type; not when an OCL variable has no type. A missing meta-model property type is a meta-model validation error not an OCL analysis error. The role of the analyzer is (now) to report errors that prevent creation of a consistent AST. The role of the validator is (now) to report errors in the AST. So this could be a validation check, but I would need to study the ramifications of OCL duplicating EMF validation. I suspect we get an irritating number of duplicate messages. OCL should diagnose the downstream consequence that some expression node has no type. Please submit a test case if you can demonstrate that it is OCL rather than the meta-model that is wrong. Closing WONTFIXes. Closing all bugs resolved in Indigo. |