| Summary: | wrong merge ? native git reporting conflicts, while egit says "no conflicts" | ||
|---|---|---|---|
| Product: | [Technology] EGit | Reporter: | Matthias Sohn <matthias.sohn> |
| Component: | Core | Assignee: | Christian Halstrick <christian.halstrick> |
| Status: | RESOLVED FIXED | QA Contact: | |
| Severity: | major | ||
| Priority: | P3 | CC: | christian.halstrick, christian.halstrick, lcf, shivamshukla |
| Version: | 0.10.0 | ||
| Target Milestone: | --- | ||
| Hardware: | PC | ||
| OS: | Mac OS X - Carbon (unsup.) | ||
| See Also: | https://bugs.eclipse.org/bugs/show_bug.cgi?id=501513 | ||
| Whiteboard: | |||
| Bug Depends on: | |||
| Bug Blocks: | 501513 | ||
|
Description
Matthias Sohn
Steps: 1. Start two test branches from master. 2. On each of the branches add a bit different version of same method. 3. Switch to master. 4. Merge test branch 1. 5. Merge test branch 2. Result: no conflict reported, two versions of the same method in merged code on master (compilation error). Expected result: conflict reported (using "git gui", as a reference). Eclipse JDT 3.6.1, JGit 0.9.3, EGit 0.9.3. I think there is a bug in JGit's merge algorithm. Before I can fix it there are some discussions about this topic on the mailing list. See here: http://dev.eclipse.org/mhonarc/lists/jgit-dev/msg00752.html . In am confident that we have a solution in the next days. (In reply to comment #2) > I think there is a bug in JGit's merge algorithm. Before I can fix it there > are some discussions about this topic on the mailing list. See here: > http://dev.eclipse.org/mhonarc/lists/jgit-dev/msg00752.html . In am > confident that we have a solution in the next days. any update on this ? I am facing the same problem. Any update on this???? the concrete problem reported in the first comment is solved. When todays jgit merges in the egit repo the commits cb2e630b and 3e494ae2 then it reports conflicts. But of course there may be other situations where E/JGit and native git behave differently. Whether and where conflicts exists is far less clear then many people expect. The question what is the minimal diff between content1 and content2 is cannot be solved always with a single correct answer. And because of that also the conflicts computed may be different based on the diff algorithm the git client uses. But, of course, our goal is to be as near to native git's behavior as possible. If JGit reports no conflicts where native git does I suspect a bug in JGit. If we behave differently than native git please report that and of course it's great if you have a reproduce able example for us. The concrete bug reported in the first command was fixed long ago. Let's close this bug and discuss similar issues (e.g. 501513) separately. |