Gah, this is a lot of code. @arrbee, to the review-mobile.

@vmg Yeah, this is stupidly large. Let me break some of this stuff out into separate PRs.

So, here's what's still missing:

• There's some hacky stuff that needs to get cleaned up (see TODOs)
• - xxdiff and writing conflicting files does not respect EOL settings
• - No detection for dirty index/workdir
• - I want to spike dumping diff_tree and merging the results of two normal diffs. I thought a function like unpack_trees() would be useful, but I'm now less convinced
• - No rename detection, I've done a few spikes that aren't integrated here yet
• - The merge_opts / merge_tree_opts stuff is terrible. I'm thinking that there should be no high-level git_merge() command and callers should call git_merge_setup() / do the diff / deal with conflicts themselves. Need some feedback on this.
• - Fastforward check is wrong (way too naive.)
• - Can't merge two commits that don't have a common ancestor

Rebased!

What's the status of this pull request? When can we expect it to hit the development branch?

@AnuiCuda Thanks for your interest. This is ready for code review and will need updates based on that review. Getting close!

Sweet! Can't wait for this! I was about to go dig into the code to figure out what needed to be done when I tripped into this request. Nice work @ethomson!

@ethomson: Looks like we're having umask issues again...

Indeed. #1940 has the fix.

On Nov 5, 2013, at 5:58 AM, Vicent Martí notifications@github.com wrote:

@ethomson: Looks like we're having umask issues again...

—
Reply to this email directly or view it on GitHub.

Sounds good. Let's go for the rebase then. :)

@vmg Yup, rebased!

What could possibly go wrong.