When I run git blame on a file (using msysgit) I always get the following sort of printout:
00000000 (Not Committed Yet 2011-01-09 11:21:30 +0200 1) package co
00000000 (Not Committed Yet 2011-01-09 11:21:30 +0200 2) {
00000000 (Not Committed Yet 2011-01-09 11:21:30 +0200 3) impor
00000000 (Not Committed Yet 2011-01-09 11:21:30 +0200 4) impor
00000000 (Not Committed Yet 2011-01-09 11:21:30 +0200 5) impor
00000000 (Not Committed Yet 2011-01-09 11:21:30 +0200 6) impor
00000000 (Not Committed Yet 2011-01-09 11:21:30 +0200 7) impor
i.e. it shows all lines as Not Yet Committed.
I tried this on many files, that have many commits - always the same results. I also tried using relative/full path, but it seems to make no difference.
When I try to use TortoiseGit's blame it always shows every line as being last committed at the first c开发者_开发技巧ommit:
even thought, as I've said, there are actually tens of commits in the history of these files..
Ideas?
Edit - More Info
- Git blame works fine on GitHub, where this repo is hosted.
- It also works fine if I clone it to a linux machine and do the blame there
- It seems that only on msysgit this doesn't work
git blame file.txt
blames the version of file.txt in your working copy. If file.txt has Windows-newlines (CRLF) in the repo and you have core.autocrlf = true
, then every line of file.txt will be considered different and will be reported by git blame
as not yet committed.
The reason why git blame <my_branch>
(or even better git blame HEAD
, which works no matter what branch you're on) works, is that it doesn't blame the working copy version so there's no potential for lines not yet being committed.
Found the solution - very weird.
If I run this:
git blame file.txt
The history is broken, as posted above.
If I do this instead:
git blame my_branch file.txt
It works!
This is very weird, because AFAICS the usage doesn't require a branch name:
$ git blame
usage: git blame [options] [rev-opts] [rev] [--] file
Starting git 2.0.1 (June 25th, 2014), git blame should stop reporting all those "Not Yet Committed" lines.
See commit 4d4813a (26 Apr 2014) by brian m. carlson (bk2204
).
(Merged by Junio C Hamano -- gitster
-- in commit e934c67, 06 Jun 2014)
blame
: correctly handle files regardless ofautocrlf
If a file contained
CRLF
line endings in a repository withcore.autocrlf=input
, then blame always marked lines as "Not Committed Yet
", even if they were unmodified.
Don't attempt to convert the line endings when creating the fake commit so that blame works correctly regardless of theautocrlf
setting.
radon8472 adds in the comments:
the answer was
git blame -w
(iggnore whitespace).
The reason was that thetextconv
for my blamed file was working with windows linebreaks, andgit blame
seems to not like it, as mentioned in kusma's answer.
Another possibility: case-sensitive filename typo
I had the same problem with git blame file.txt, then realized that I'd made a case-sensitive filename typo with file.txt
Changed it to File.txt (for example), and I got the expected results w/o having to specify my_branch: git blame File.txt
精彩评论