Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.
Bug 318784 - DeclarationGenerator fails for some cases
Summary: DeclarationGenerator fails for some cases
Status: RESOLVED FIXED
Alias: None
Product: CDT
Classification: Tools
Component: cdt-refactoring (show other bugs)
Version: 7.0   Edit
Hardware: PC Windows 7
: P3 normal (vote)
Target Milestone: 7.0.1   Edit
Assignee: Emanuel Graf CLA
QA Contact: Emanuel Graf CLA
URL:
Whiteboard:
Keywords:
Depends on: 318785
Blocks: 318874
  Show dependency tree
 
Reported: 2010-07-03 09:37 EDT by Tomasz Wesolowski CLA
Modified: 2010-07-07 05:23 EDT (History)
0 users

See Also:

Attachments
new declarator generation (20.56 KB, patch)
2010-07-03 17:40 EDT, Tomasz Wesolowski CLA 		no flags Details | Diff
fixed declaration generation (22.75 KB, patch)
2010-07-06 14:17 EDT, Tomasz Wesolowski CLA 		emanuel: iplog+
Details | Diff
test case (860 bytes, patch)
2010-07-07 04:37 EDT, Emanuel Graf CLA 		emanuel: iplog-
Details | Diff
Show Obsolete (1) View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Tomasz Wesolowski CLA 2010-07-03 09:37:25 EDT 
Build Identifier: 

While fixing Extract Constant refactoring, I've found out some problems with my current implementation of DeclarationGenerator - it fails to generate the proper IASTDeclarator for complex cases like nested pointer-to-array-of-pointer-to-sth types.

I'm redesigning the implementation of DeclarationGenerator atm, I hope to have it finished and working today. I hope that my new approach to this problem will allow to extend it with handling function pointers, too.

Update soon.

Reproducible: Always
Comment 1 Tomasz Wesolowski CLA 2010-07-03 17:40:59 EDT 
Created attachment 173362 [details]
new declarator generation

Success!

Well, that's literally my 4th attempt on the generation logic - fortunately, this one seems to be both correct and shortest in terms of code so far :) Wasn't so pleasant to throw away my original code, but at least we have the sophisticated declarations generated correctly.

The good thing is that after rewriting the code it was easy to add support for function pointers, including those returning function pointers. Works like a charm, finally.

See bug 318785 for NPEs after applying the change.
Comment 2 Emanuel Graf CLA 2010-07-06 05:32:59 EDT 
> 
> See bug 318785 for NPEs after applying the change.

The cause for the NPE is the ArrayDeclarator, getName() returns null but according to the javaDoc for this method the method should return an empty name.
Comment 3 Tomasz Wesolowski CLA 2010-07-06 14:17:29 EDT 
Created attachment 173580 [details]
fixed declaration generation

Here we go!

+ added an empty ASTName for every created declarator where there was a null ASTName before
and:
+ fixed PointerOperator order
+ added some support for pointer-to-members
Comment 4 Emanuel Graf CLA 2010-07-07 04:37:43 EDT 
Created attachment 173625 [details]
test case
Comment 5 Emanuel Graf CLA 2010-07-07 05:00:33 EDT 
Fixed in HEAD and 7.0 > 20100707

Thanks Tomasz