3115: git diff must implicitly use --full-index when generating diffs
- Incomplete
- Review Board
| yane****@gmai***** (Google Code) (Is this you? Claim this profile.) | |
| Dec. 11, 2013 |
What version are you running? Reviewboard: 1.7.14 (local install)/1.8-alpha-0 (reviews.reviewboard.org) RBTools: 0.5.2-dev What's the URL of the page containing the problem? ^/r/new/ What steps will reproduce the problem? 1. git clone git@github.com:reviewboard/reviewboard.git 2. Hack on some code. 3. post-review/rbt post/create a new review via ^/r/new. What is the expected output? What do you see instead? Expected output: The diff should be posted if all of my credentials are ok; it is posted at least with svn. Actual output: What I see instead is the the web UI gets stuck when I don't use --full-index when posting the diff (in fact the web UI gets stuck when I try creating a CR on reviews.reviewboard.org), and similarly I run into problems when using rbt post/post-review (but at least it times out and dies), so I assume that the problem is server side, not client side. What operating system are you using? What browser? $ uname -a CYGWIN_NT-6.1 ZL00757 1.7.25(0.270/5/3) 2013-08-31 20:37 x86_64 Cygwin Mozilla/5.0 (Windows NT 6.1; WOW64; rv:21.0) Gecko/20100101 Firefox/21.0
This appears to be a known issue: http://comments.gmane.org/gmane.comp.kde.devel.general/64319 The code in RBTools explicitly spells out --full-index in the git make_diff method (rbtools/clients/git.py): 289 290 def make_diff(self, ancestor, commit=""): 291 """ 292 Performs a diff on a particular branch range. 293 """ ... 310 elif self.type == "git": 311 cmdline = [self.git, "diff", "--no-color", "--full-index", 312 "--no-ext-diff", "--ignore-submodules", "--no-renames", 313 rev_range] Would be nice if this was in the FAQ at the very least.
I'm not sure about the 1.7.x case, but the new review request paths have largely changed in the version we're working on (which we're running on reviews.reviewboard.org). There's a known bug that we just fixed (but haven't deployed to our server yet) where the validation error doesn't get back to you. It should be working on 1.7.x, though. We'll see if something needs to be backported. I wouldn't expect it to "spin" in any way. Is there no error you see?
-
+ NeedInfo
-
- NeedInfo + Incomplete