Do you have a clean way to list all the files that ever existed in specified branch?
thanks
A:
You can run git-log --name-status, which echoes something like:
commit afdbbaf52ab24ef7ce1daaf75f3aaf18c4d2fee0
Author: Your Name <[email protected]>
Date: Tue Aug 12 13:28:34 2008 -0700
Added test file.
A test
Then extract files added:
git-log --name-status | sed -ne 's/^A[^u]//p' | sort -u
strager
2009-02-12 21:29:46
+4
A:
Variation on Strager's:
git log --pretty=format: --name-status | cut -f2- | sort -u
Edit:
Thanks to Jakub for teaching me a bit more in the comments:
git log --pretty=format: --name-only --diff-filter=A | sort -
Shorter pipeline, more opportunity to git to get things right.
Dustin
2009-02-12 21:39:18
Hmm, I guess yours is superior. +1. =]
strager
2009-02-12 21:44:14
whao, awesome, and fast!thank you!
elmarco
2009-02-12 21:48:44
Hmmmm. using "uniq -u" may be faster and more memory efficient than "sort -u" if you don't actually want the output sorted. Could make a difference on big repos.
Pat Notz
2009-02-12 22:18:08
@Pat: uniq requires sorted input, so you have to use sort
Jakub Narębski
2009-02-13 12:36:07
@Dustlin: Add --diff-filter=A option (list only added files). Current version (without sed filtering only added files) would fail if you have enabled rename detection and have renames in history. I think you can then use --name-only instead of --name-status and remove 'cut -f2-' from pipeline.
Jakub Narębski
2009-02-13 12:39:44