Thanks for the advice Andreas, I'm currently reading "ProGit" (which doesn't mention pull --rebase) at the moment. I guess that pull --rebase would be roughly equivalent to:
- get fetch (to update my cloned repository) - git rebase master Now I assume my local commits would look like they are based on the latest commits made on the master which will give us a cleaner commit history. I will then either push or issue a pull request. Hope I got this right... /Bengt 2012/2/7 Andreas Pieber <[email protected]> > Hey, > > > On 02/06/2012 09:50 PM, Bengt Rodehav wrote: > > Oh yes, I remember I did register my SSH keys on GitHub. I guess that's > what gave me access... > > So in the future I do this: > > > Almost :-) > > > > - Register my user name and email with git config > - Clone the ops4j project directly to my local computer without forking > - Do my local changes, add, and then commit > - git fetch and then git merge before I do... > > it's easier to do a git pull --rebase here. > > > - ...git push > > Is that correct? > > > Basically yes. At OPS4J we follow the karma: "push first and revert if > necessary". Still if you're absolutely unsure about something you can still > create a pull request to be reviewed first. A pull request can also be > merged via the github UI. > > > I guess the above will render an extra "merge" commit if there are any > changes made at ops4j after I created my clone but that is normal behaviour > then. > > > No, as long as you're commits are only local feel free to modify them at > will. Git allows you to modify (e.g. with git commit --amend), squash > (rebase -i) or completely remove/undo commits (git revert). As long as all > your modifications are commited nowhere you can change them as you like. > Therefore the simplest (and "most beautiful") option for ppl not so firm > with git is to follow your described workflow but replace the git fetch/git > merge with git pull --rebase. This will also avoid the additional merge > node. > > @UI: for windows user use tortoise-git; mac user should use the github > app; linux user, well... use the shell ;-) > > I hope this helps. > > Kind regards, > Andreas > > > /Bengt > > 2012/2/6 Harald Wellmann <[email protected]> > >> Am 06.02.2012 21:22, schrieb Bengt Rodehav: >> >> Thanks a lot for your reply Harald, >>> >>> I'm glad I didn't mess things up completely then. I guess we can live >>> with "unknown" for this commit. >>> >>> >> Yeah, now we know it was you ;-) >> >> >> I thought I had to go via my GitHub account since that account is what >>> is given permission to push to ops4j. Or did I misunderstood this - >>> perhaps anyone can push to ops4j projects? >>> >>> >> No, you need the GitHub account and your SSH key or HTTPS password to >> push to GitHub, and you need to be a member of the ops4j organization. >> >> The name and email in the Git commit message is just a string and might >> be anything. >> >> Of course it *should* match your actual email address, and I'd say it >> *has to* when it comes to signed tags. >> >> >> Cheers, >> Harald >> >> >> _______________________________________________ >> general mailing list >> [email protected] >> http://lists.ops4j.org/mailman/listinfo/general >> > > > > _______________________________________________ > general mailing > [email protected]http://lists.ops4j.org/mailman/listinfo/general > > > _______________________________________________ > general mailing list > [email protected] > http://lists.ops4j.org/mailman/listinfo/general > >
_______________________________________________ general mailing list [email protected] http://lists.ops4j.org/mailman/listinfo/general
