GTRepository enumerateCommitsBeginningAtSha: not enumerating through all commits about objective-git

[joshaber]

I don't see anything obviously wrong with that snippet and that's certainly not the intended behavior. Sounds like a bug. I'm using this method in a project but in my case I'm always passing nil for the sha (which means start from HEAD). You could try that as a sanity check to see if that does what you'd expect.

[tclem]

Yeah, the code looks right. Can you post some stats from # commits objective-git returns vs. how many you get when you do a
git log --oneline | wc -l

[indragiek]

Thanks for the replies guys. I tried passing nil and the problem still persists. However, oddly the output of git log --oneline | wc -l is the same as the # of commits that I get using objective-git (which is definitely not all the commits). This seems to indicate that its something to do with git rather than libgit2 or objective-git, but I'm not sure what :-/

[tclem]

Both git log and the the rev walker (which is what you are using when you call enumerateCommitsBeginningAtSha:) are walking the DAG starting at the sha you specify or HEAD if you pass nil and following the first parent links. This will never give you the total number of commits in a repository if you have branches with divergent work. Could this be what you are seeing?

Otherwise, can you tell me how you know you are missing commits? What are those commits? Are they reachable from the HEAD?

[indragiek]

Sorry for the waste of time here guys. All the commits were in fact being enumerated, albeit in the wrong order (because I had forgotten to specify the GTEnumeratorOptionsTimeSort sorting option) so the commits weren't in the place I expected them in. Can't believe I missed something this trivial ;)