Bugzilla – Full Text Bug Listing
| Summary: | while cloning git repository jgit.git on Android 2.2 | ||||||
|---|---|---|---|---|---|---|---|
| Product: | [Technology] JGit | Reporter: | Venky <Venkatesh.swdev> | ||||
| Component: | JGit | Assignee: | Christian Halstrick <christian.halstrick> | ||||
| Status: | RESOLVED WORKSFORME | QA Contact: | |||||
| Severity: | major | ||||||
| Priority: | P3 | CC: | caniszczyk, christian.halstrick, dugueperoux.esteban, Venkatesh.swdev, vladimir | ||||
| Version: | unspecified | Keywords: | api, core | ||||
| Target Milestone: | --- | ||||||
| Hardware: | Other | ||||||
| OS: | other | ||||||
| Whiteboard: | |||||||
| Attachments: |
|
||||||
|
Description
Venky
Can you try with the latest nightly build from our p2 repo? http://download.eclipse.org/egit/updates-nightly *** Bug 339869 has been marked as a duplicate of this bug. *** Note that I'm seeing this using the latest egit nightly as of Mar 14 on win32 with the Oracle JDK. remote git repo (via git+ssh) was created by git 1.7.4.1, 1.7.0.1 on mingw32 can pull it fine. Cloning from the command line using jgit 0.10.1 works fine, so this looks like a regression. (Can't change the bug summary, but this seems to be general, not Android related) I tried this out. I wrote a small android app which tries to clone your repo and then print info's about the commits. I tested the app on my LG-P500 with android 2.2 ... and it works fine for me. I attached the app here and I shared the sources for it here: git://github.com/chalstrick/GitLog.git . Does this app work for you? Created attachment 191879 [details]
android app which clones your repo
Since I can reproduce this on the desktop, is there any information that I can provide? Happy to get jgit in the debugger or similar at the point where the exception is thrown if it would help. (The repo's internal unfortunately.) It is not yet clear to me how exactly you reproduce the problem. You once wrote that to reproduce the problem you "Clone git repository https://github.com/venkateshswdev/git_test.git on android using code" . That's what I did, right? I wrote a android app which used the JGit lib to clone exactly the mentioned repo. I guess I misunderstood your use-case. What "exactly" do you do to reproduce the problem? I think you have me confused with the original reporter of this bug :-) I filed bug 339869 and was pointed to this one on irc which seemed very similar -- I duped 339869 against this, and then realized I couldn't edit the bug summary to reflect that it's not just android. For my use case, I am just cloning a repo on a win32 host using sun's java version 1.6.0_23. Attempting to clone with both egit/jgit 0.11.3 and 0.12.0.201103170911 fail with the same exception above. Other repos work fine. However, all repos work fine with jgit 0.10.1, both via egit and via jgit command line, which makes me think that there's a regression somewhere. (Can't use egit 0.10.1 though, because there's a file rename bug on windows in that version.) Note that I don't see this error with the original reporter's repo though (just tried), so perhaps the bugs aren't identical, just have a similar symptom. Ahhh, now I understand. That explains. I would suggest to reopen 339869 and to close this (if the original auther has no objections). I'll do that and then try to solve your problem on 339869. (In reply to comment #0) > Reproducible: Always > > Steps to Reproduce: > 1. Clone git repository https://github.com/venkateshswdev/git_test.git on > android using code Since I can't reproduce this problem anymore on android I would like to close it now. If I would have access to your repo then I would run a "git bisect" in order to find out when the problem was introduced. Maybe you can do? In order to do that you just must be able to compile jgit with maven from command line as described in our Contributors guide[1]. Then I would - cd <yourJGitDir> - git bisect start origin/master v0.10.1 -- - git bisect run sh -c "rm -fr <aTempFolder>; mvn -Dmaven.test.skip=true package || exit 125; java -Xmx512m -jar org.eclipse.jgit.pgm/target/jgit-cli.jar clone <yourRepoURL> C:/tmp/a || exit 127; rm -fr <aTempFolder>" This will do a binary search on the jgit states between v0.10.1 and origin/master. Each inspected state is first build and then we try to clone the repo with the just build jgit commandline tool. In the end it should tell you that commit which broke the clone. [1] http://wiki.eclipse.org/EGit/Contributor_Guide (In reply to comment #13) forget this last comment on git bisect. This should go to the other bug 339869. > If I would have access to your repo then I would run a "git bisect" in order to > find out when the problem was introduced. Maybe you can do? In order to do that > you just must be able to compile jgit with maven from command line as described > in our Contributors guide[1]. > > Then I would > - cd <yourJGitDir> > - git bisect start origin/master v0.10.1 -- > - git bisect run sh -c "rm -fr <aTempFolder>; mvn -Dmaven.test.skip=true > package || exit 125; java -Xmx512m -jar > org.eclipse.jgit.pgm/target/jgit-cli.jar clone <yourRepoURL> C:/tmp/a || exit > 127; rm -fr <aTempFolder>" > > This will do a binary search on the jgit states between v0.10.1 and > origin/master. Each inspected state is first build and then we try to clone the > repo with the just build jgit commandline tool. > > In the end it should tell you that commit which broke the clone. > > [1] http://wiki.eclipse.org/EGit/Contributor_Guide |