From what I understand, Git doesn't really need to track file rename/move/copy operations, so what's the real purpose
of git mv? The man page isn't particularly descriptive...
Is it obsolete? Is it an internal command, not meant to be used by regular users?
git mv oldname newname
is just shorthand for:
mv oldname newname
git add newname
git rm oldname
i.e. it updates the index for both old and new paths automatically.
mv oldname newname; git add newname; git rm oldname, it will also go wrong for git mv oldname newname (see this answer). –
Samaniego git mv is slightly different from the mv oldname newname; git add newname; git rm oldname, in that if you made changes to the file before git mving it, those changes won't be staged until you git add the new file. –
Samaniego git mv? –
Dow From the official GitFaq:
Git has a rename command
git mv, but that is just a convenience. The effect is indistinguishable from removing the file and adding another with different name and the same content
Git is just trying to guess for you what you are trying to do. It is making every attempt to preserve unbroken history. Of course, it is not perfect. So git mv allows you to be explicit with your intention and to avoid some errors.
Consider this example. Starting with an empty repo,
git init
echo "First" >a
echo "Second" >b
git add *
git commit -m "initial commit"
mv a c
mv b a
git status
Result:
# On branch master
# Changes not staged for commit:
# (use "git add/rm <file>..." to update what will be committed)
# (use "git checkout -- <file>..." to discard changes in working directory)
#
# modified: a
# deleted: b
#
# Untracked files:
# (use "git add <file>..." to include in what will be committed)
#
# c
no changes added to commit (use "git add" and/or "git commit -a")
Autodetection failed :( Or did it?
$ git add *
$ git commit -m "change"
$ git log c
commit 0c5425be1121c20cc45df04734398dfbac689c39
Author: Sergey Orshanskiy <*****@gmail.com>
Date: Sat Oct 12 00:24:56 2013 -0400
change
and then
$ git log --follow c
Author: Sergey Orshanskiy <*****@gmail.com>
Date: Sat Oct 12 00:24:56 2013 -0400
change
commit 50c2a4604a27be2a1f4b95399d5e0f96c3dbf70a
Author: Sergey Orshanskiy <*****@gmail.com>
Date: Sat Oct 12 00:24:45 2013 -0400
initial commit
Now try instead (remember to delete the .git folder when experimenting):
git init
echo "First" >a
echo "Second" >b
git add *
git commit -m "initial commit"
git mv a c
git status
So far so good:
# On branch master
# Changes to be committed:
# (use "git reset HEAD <file>..." to unstage)
#
# renamed: a -> c
git mv b a
git status
Now, nobody is perfect:
# On branch master
# Changes to be committed:
# (use "git reset HEAD <file>..." to unstage)
#
# modified: a
# deleted: b
# new file: c
#
Really? But of course...
git add *
git commit -m "change"
git log c
git log --follow c
...and the result is the same as above: only --follow shows the full history.
Now, be careful with renaming, as either option can still produce weird effects. Example:
git init
echo "First" >a
git add a
git commit -m "initial a"
echo "Second" >b
git add b
git commit -m "initial b"
git mv a c
git commit -m "first move"
git mv b a
git commit -m "second move"
git log --follow a
commit 81b80f5690deec1864ebff294f875980216a059d
Author: Sergey Orshanskiy <*****@gmail.com>
Date: Sat Oct 12 00:35:58 2013 -0400
second move
commit f284fba9dc8455295b1abdaae9cc6ee941b66e7f
Author: Sergey Orshanskiy <*****@gmail.com>
Date: Sat Oct 12 00:34:54 2013 -0400
initial b
Contrast it with:
git init
echo "First" >a
git add a
git commit -m "initial a"
echo "Second" >b
git add b
git commit -m "initial b"
git mv a c
git mv b a
git commit -m "both moves at the same time"
git log --follow a
Result:
commit 84bf29b01f32ea6b746857e0d8401654c4413ecd
Author: Sergey Orshanskiy <*****@gmail.com>
Date: Sat Oct 12 00:37:13 2013 -0400
both moves at the same time
commit ec0de3c5358758ffda462913f6e6294731400455
Author: Sergey Orshanskiy <*****@gmail.com>
Date: Sat Oct 12 00:36:52 2013 -0400
initial a
Ups... Now the history is going back to initial a instead of initial b, which is wrong. So when we did two moves at a time, Git became confused and did not track the changes properly. By the way, in my experiments the same happened when I deleted/created files instead of using git mv. Proceed with care; you've been warned...
As @Charles says, git mv is a shorthand.
The real question here is "Other version control systems (eg. Subversion and Perforce) treat file renames specially. Why doesn't Git?"
Linus explains at http://permalink.gmane.org/gmane.comp.version-control.git/217 with characteristic tact:
Please stop this "track files" crap. Git tracks exactly what matters, namely "collections of files". Nothing else is relevant, and even thinking that it is relevant only limits your world-view. Notice how the notion of CVS "annotate" always inevitably ends up limiting how people use it. I think it's a totally useless piece of crap, and I've described something that I think is a million times more useful, and it all fell out exactly because I'm not limiting my thinking to the wrong model of the world.
There's a niche case where git mv remains very useful: when you want to change the casing of a file name on a case-insensitive file system. Both APFS (mac) and NTFS (windows) are, by default, case-insensitive (but case-preserving).
greg.kindel mentions this in a comment on CB Bailey's answer.
Suppose you are working on a mac and have a file Mytest.txt managed by git. You want to change the file name to MyTest.txt.
You could try:
$ mv Mytest.txt MyTest.txt
overwrite MyTest.txt? (y/n [n]) y
$ git status
On branch master
Your branch is up to date with 'origin/master'.
nothing to commit, working tree clean
Oh dear. Git doesn't acknowledge there's been any change to the file.
You could work around this with by renaming the file completely then renaming it back:
$ mv Mytest.txt temp.txt
$ git rm Mytest.txt
rm 'Mytest.txt'
$ mv temp.txt MyTest.txt
$ git add MyTest.txt
$ git status
On branch master
Your branch is up to date with 'origin/master'.
Changes to be committed:
(use "git reset HEAD <file>..." to unstage)
renamed: Mytest.txt -> MyTest.txt
Hurray!
Or you could save yourself all that bother by using git mv:
$ git mv Mytest.txt MyTest.txt
$ git status
On branch master
Your branch is up to date with 'origin/master'.
Changes to be committed:
(use "git reset HEAD <file>..." to unstage)
renamed: Mytest.txt -> MyTest.txt
git mv moves the file, updating the index to record the replaced file path, as well as updating any affected git submodules. Unlike a manual move, it also detects case-only renames that would not otherwise be detected as a change by git.
It is similar (though not identical) in behavior to moving the file externally to git, removing the old path from the index using git rm, and adding the new one to the index using git add.
This question has a lot of great partial answers. This answer is an attempt to combine them into a single cohesive answer. Additionally, one thing not called out by any of the other answers is the fact that the man page actually does mostly answer the question, but it's perhaps less obvious than it could be.
Three different effects are called out in the man page:
The file, directory, or symlink is moved in the filesystem:
git-mv - Move or rename a file, a directory, or a symlink
The index is updated, adding the new path and removing the previous one:
The index is updated after successful completion, but the change must still be committed.
Moved submodules are updated to work at the new location:
Moving a submodule using a gitfile (which means they were cloned with a Git version 1.7.8 or newer) will update the gitfile and core.worktree setting to make the submodule work in the new location. It also will attempt to update the submodule.<name>.path setting in the gitmodules(5) file and stage that file (unless -n is used).
As mentioned in this answer, git mv is very similar to moving the file, adding the new path to the index, and removing the previous path from the index:
mv oldname newname
git add newname
git rm oldname
However, as this answer points out, git mv is not strictly identical to this in behavior. Moving the file via git mv adds the new path to the index, but not any modified content in the file. Using the three individual commands, on the other hand, adds the entire file to the index, including any modified content. This could be relevant when using a workflow which patches the index, rather than adding all changes in the file.
Additionally, as mentioned in this answer and this comment, git mv has the added benefit of handling case-only renames on file systems that are case-insensitive but case-preserving, as is often the case in current macOS and Windows file systems. For example, in such systems, git would not detect that the file name has changed after moving a file via mv Mytest.txt MyTest.txt, whereas using git mv Mytest.txt MyTest.txt would successfully update its name.
There's another use I have for git mv not mentioned above.
Since discovering git add -p (git add's patch mode; see http://git-scm.com/docs/git-add), I like to use it to review changes as I add them to the index. Thus my workflow becomes (1) work on code, (2) review and add to index, (3) commit.
How does git mv fit in? If moving a file directly then using git rm and git add, all changes get added to the index, and using git diff to view changes is less easy (before committing). Using git mv, however, adds the new path to the index but not changes made to the file, thus allowing git diff and git add -p to work as usual.
There is a 2023 discussion advocating for Git to record renames explicitly.
(Note: still a discussion, certainly not yet a feature).
It includes a comment regarding git mv:
git show :0:oldname > newname git add newname git rm --cached oldname mv oldname newname ```Basically, a move but also preserving staged/unstaged contents.
But yes. How git handles renames can sometimes be a bit of a PITA.
And:
git add file1
commit -m "Commit 1"
echo 'Changed' > file1
git mv file1 file2
git status
Which yields:
Changes to be committed:
(use "git restore --staged <file>..." to unstage)
renamed: file1 -> file2
Changes not staged for commit:
(use "git add <file>..." to update what will be committed)
(use "git restore <file>..." to discard changes in working directory)
modified: file2
Maybe git mv has changed since these answers were posted, so I will update briefly. In my view, git mv is not accurately described as short hand for:
# not accurate: #
mv oldname newname
git add newname
git rm oldname
I use git mv frequently for two reasons that have not been described in previous answers:
Moving large directory structures, where I have mixed content of both tracked and untracked files. Both tracked and untracked files will move, and retain their tracking/untracking status
Moving files and directories that are large, I have always assumed that git mv will reduce the size of the repository DB history size. This is because moving/renaming a file is indexation/reference delta. I have not verified this assumption, but it seems logical.
git rm step, which stages the deletion of the file under the old name; once you do that, you will see that git status shows the file as renamed, just as it did with git mv. Other version control systems do treat files or directories as fundamental units, and track renames; git simply stores snapshots of the repository, and calculates everything else as needed. –
Varga git init; echo 'hello' > test; git add test; git commit -m initial; git cat-file -p HEAD^{tree}; cp test test2; git add test2; git commit -m duplicated; git cat-file -p HEAD^{tree}; git rm test; git commit -m deleted; git cat-file -p HEAD^{tree}; git diff HEAD^^ HEAD You can even see the objects created with ls .git/objects/* - there will be 7 objects: 3 commits, 3 trees, and 1 blob, no matter how many duplicates you commit. –
Varga © 2022 - 2024 — McMap. All rights reserved.