Community
Participate
Working Groups
Build 202a. After a failed attempt at renaming an interface from 'Context' to 'IContext', various compilation units' source code was trashed. For example: within JavaDoc * @return the result of the bind as an HTTP response * @see delete() */ became * @return the result of the bind as an HTTP response * @see delIContext */ And there are many more examples of a text replacement being done in very bizzare places.
The failed refactoring mentioned in this bug report was Bug#5015
need a reproducible testcase
- Use Eclipse build 202a with the attached workspace Let me know if you require the complete install. - From the outline view of org.eclipse.webdav.IContext choose to Rename... the interface. - Enter a new name, say, 'IContextDebug', click Next. - There should be two problems noted. - Optionally, click Next to see the walkback reported in Bug#5015 (This trashing bug also appears if you don't cause the walkback) - Click Finish, the rename progresses to completion. - Take a look at org.eclipse.webdav.server.PreconditionCheck.java The rename has changed line 51 from throw new InternalServerError("Tokenizing problem ... to throw new IIContextDebugerverError("Tokenizing problem ... There are, in general, other random places where a bad rename has occurred.
1. i see no attached workspace 2. sounds like the indexer is corrupted - what is reported by search? (search for references to org.eclipse.webdav.IContext) my guess is that matches are incorrectly reported
1. I'm having problems with the Bugzila attachments from here, please see the file in ftp://walleye.oti.com/upload/Bug#5021-workspace.zip until I get it attached. 2. Indeed, the search you requested gives many false hits.
in that case - the bug gets moved to jcore for investigation
Please investigate match positions.
Walleye seems to be down for now. Tim, if you have another ftp server that I can access, please let me know.
Walleye is up, but I cannot see the 'upload' directory. What user/password should I use?
Was able to reproduce. The problem was that the same match set was used by a compilation unit and all the compilation units that were requested during type resolution. Matches were reported agains the wrong compilation unit. Fixed in SNZ's HEAD stream.
