Welcome to the Splat alpha!

The example you gave looks like RBTools 0.5, in which case you may also want to add --parent=HEAD^

With RBTools 0.6, you should be able to run `rbt post HEAD` and it should do the right thing.

• Status:

  +	UserError

I have just tried with RBTools 0.6. The change is no longer HEAD, but let's call it abc123.

$ rbt post abc123

This gives the same result.

$ rbt post abc123^ abc123

This gives the same result.

$ rbt post --parent=abc123^ abc123

This gives the same result.

Any other suggestions?

Can you run with --debug and post the output?

• Status:

  -	UserError
  +	NeedInfo

Here we go. Note that def456 is the (only) parent of abc123, and ghi789 is the commit before the git mv commit. It seems odd to pass --no-renames and -M together.

$ rbt post --debug --parent=abc123^ abc123
>>> RBTools 0.6
>>> Python 2.7.5 (default, Aug 25 2013, 00:04:04)
[GCC 4.2.1 Compatible Apple LLVM 5.0 (clang-500.0.68)]
>>> Running on Darwin-13.1.0-x86_64-i386-64bit
>>> Home = /Users/foo
>>> Current directory = /Users/foo/repo
>>> Checking for a Subversion repository...
>>> Running: svn info --non-interactive
>>> Command exited with rc 1: ['svn', 'info', '--non-interactive']
svn: E155007: '/Users/foo/repo' is not a working copy
---
>>> Checking for a Git repository...
>>> Running: git rev-parse --git-dir
>>> Running: git config core.bare
>>> Running: git rev-parse --show-toplevel
>>> Running: git symbolic-ref -q HEAD
>>> Running: git config --get branch.master.merge
>>> Running: git config --get branch.master.remote
>>> Running: git config --get remote.origin.url
>>> repository info: Path: git@github.example.com:example/project.git, Base path: , Supports changesets: False
>>> Making HTTP GET request to https://rb.example.com/api/
>>> Making HTTP GET request to http://rb.example.com/api/info/
>>> Running: git rev-parse abc123
>>> Running: git rev-parse abc123^
>>> Running: git branch -r --contains def456
>>> Running: git merge-base def456 origin/master
>>> Running: git diff --no-color --full-index --ignore-submodules --no-renames def456..abc123 -M
>>> Running: git diff --no-color --full-index --ignore-submodules --no-renames ghi789..def456 -M
>>> Running: git log --reverse --pretty=format:%s%n%n%b ^ghi789 abc123
>>> Making HTTP GET request to http://rb.example.com/api/review-requests/
>>> Making HTTP POST request to http://rb.example.com/api/review-requests/
>>> Making HTTP GET request to http://rb.example.com/api/review-requests/123/diffs/
>>> Making HTTP POST request to http://rb.example.com/api/review-requests/123/diffs/
>>> Making HTTP GET request to http://rb.example.com/api/review-requests/123/draft/
>>> Making HTTP PUT request to http://rb.example.com/api/review-requests/123/draft/
Review request #123 posted.

http://rb.example.com/r/123/
http://rb.example.com/r/123/diff/

So the diffs there look correct--it's generating a diff between def456 and abc123 (the one that gets shown), and then a parent diff between ghi789 and def456. Can you look in the database and verify that the parent diff is present?

I'll fix the --no-renames nonsense in another change.

Sorry, I don't have access to the database.

Without deeper info, I don't think we can make any more progress on this. Posting changes with parent diffs is quite well tested, especially with the new syntax in rbtools 0.6, so I think it's probably some user error somewhere.

• Status:

  -	NeedInfo
  +	UserError