Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.

Bug 265365

Summary:

Generated switch statements can be incorrectly ordered.

Product:

[Modeling] EMF

Reporter:

James Bruck <bruck.james>

Component:

Tools

Assignee:

Dave Steinberg <davidms>

Status:

VERIFIED FIXED

QA Contact:

Severity:

normal

Priority:

CC:

davidms, Ed.Merks

Version:

2.5.0

Target Milestone:

---

Hardware:

OS:

Windows XP

Whiteboard:

Bug Depends on:

Bug Blocks:

265703

Attachments:

Description	Flags
Fix for switch class ordering	none

Description James Bruck

2009-02-18 15:39:04 EST

Consider an example where we have:
Class A that extends B and C.
B extends  Z.
C extends Y.
Y extends Z.

The generated switch for A would be:   B, C, Z, Y.  But the more appropriate order would be B, C, Y, Z since Z is more general than Y.

The algorithm that determines the switch order seems to be breadth first as opposed to considering inheritance depth.

This is a particular problem for the UMLSwitch casePort for example.

Comment 1 Dave Steinberg

2009-02-18 15:45:36 EST

Breadth first *does* of course consider inheritance depth. It just considers depth of the first encountered inheritance path, not the *longest* path. I'm working on a patch.

Comment 2 Dave Steinberg

2009-02-18 18:03:08 EST

Created attachment 126091 [details]
Fix for switch class ordering

Here's a patch that should fix the ordering of supertype cases.

Comment 3 Ed Merks

2009-02-20 12:16:22 EST

Looks good.

Comment 4 Dave Steinberg

2009-02-20 13:01:35 EST

The fix is in CVS for EMF 2.5.

Comment 5 Dave Steinberg

2009-03-17 15:40:47 EDT

Fix available in HEAD: 2.5.0.I200902241800.

Comment 6 Dave Steinberg

2009-06-25 02:51:24 EDT

Fix available in HEAD: 2.5.0 (R200906151043).