Git experts, I want to pull remote branch with specified commit id, how to do it? Below command can get remote branch $git pull remote refs/heads/$branch_name Below command doesn't work $git pull remote objects/$commit_id Thanks, Emily --
You need to fetch it first, and then merge the commit you want. The tools operating the fetching protocol only use refs, so if you want to fetch (or pull) a specific version that has neither a tag nor a branch head pointing to it, you'll have to write a new tool for that. The end-result of the following command will be, barring side-effects in the remote-tracking branches, identical to what you're trying to do though: git fetch remote && git merge $commit_id -- Andreas Ericsson andreas.ericsson@op5.se OP5 AB www.op5.se Tel: +46 8-230225 Fax: +46 8-230231 --
Andreas, I tried your method, it works. Thank you very much ! Emily --
You can't, and that is so by design. Consider this: You accidentally push a branch with confidential data to a public repository. You notice it early, and quickly delete the branch using 'git push the-repo :refs/heads/that-branch'. At this time the objects with the confidential data are still lingering in the public repository. But with the current behavior noone can access them even if the SHA1 happens to be known. -- Hannes --
Might a repack (perhaps an automatic one) put the object in a pack (perhaps in a delta chain) that can be fetched through another ref? -Brad --
Doesn't this line of reasoning only apply to the ssh and git transports? (ie, the file and rsync transport would retrieve it regardless) --
You are right. Http and rsync would happily ship the object. -- Hannes --
No, assuming that-branch was the only ref that contained the objects. Even if the repack happens before the branch is deleted, the objects that were *only* in that-branch will not be sent out. -- Hannes --
