git-clean - Remove untracked files from the working tree
Synopsis
git clean [-d] [-f] [-i] [-n] [-q] [-e <pattern>] [-x | -X] [--] <path>…
Description
Cleans the working tree by recursively removing files that are not under version control, starting from the current directory.
Normally, only files unknown to Git are removed, but if the -x
option is specified, ignored files are also removed. This can, for example, be useful to remove all build products.
If any optional <path>...
arguments are given, only those paths are affected.
Step 1 is to show what will be deleted by using the -n
option:
# Print out the list of files and directories which will be removed (dry run)
git clean -n -d
Clean Step - beware: this will delete files:
# Delete the files from the repository
git clean -f
- To remove directories, run
git clean -f -d
or git clean -fd
- To remove ignored files, run
git clean -f -X
or git clean -fX
- To remove ignored and non-ignored files, run
git clean -f -x
or git clean -fx
Note the case difference on the X
for the two latter commands.
If clean.requireForce
is set to "true" (the default) in your configuration, one needs to specify -f
otherwise nothing will actually happen.
Again see the git-clean
docs for more information.
Options
-f
, --force
If the Git configuration variable clean.requireForce is not set to
false, git clean will refuse to run unless given -f
, -n
or -i
.
-x
Don’t use the standard ignore rules read from .gitignore (per
directory) and $GIT_DIR/info/exclude
, but do still use the ignore
rules given with -e
options. This allows removing all untracked files,
including build products. This can be used (possibly in conjunction
with git reset) to create a pristine working directory to test a clean
build.
-X
Remove only files ignored by Git. This may be useful to rebuild
everything from scratch, but keep manually created files.
-n
, --dry-run
Don’t actually remove anything, just show what would be done.
-d
Remove untracked directories in addition to untracked files. If an
untracked directory is managed by a different Git repository, it is
not removed by default. Use -f
option twice if you really want to
remove such a directory.
Amending the most recent commit message
git commit --amend
will open your editor, allowing you to change the commit message of the most recent commit. Additionally, you can set the commit message directly in the command line with:
git commit --amend -m "New commit message"
…however, this can make multi-line commit messages or small corrections more cumbersome to enter.
Make sure you don't have any working copy changes staged before doing this or they will get committed too. (Unstaged changes will not get committed.)
Changing the message of a commit that you've already pushed to your remote branch
If you've already pushed your commit up to your remote branch, then - after amending your commit locally (as described above) - you'll also need to force push the commit with:
git push <remote> <branch> --force
# Or
git push <remote> <branch> -f
Warning: force-pushing will overwrite the remote branch with the state of your local one. If there are commits on the remote branch that you don't have in your local branch, you will lose those commits.
Warning: be cautious about amending commits that you have already shared with other people. Amending commits essentially rewrites them to have different SHA IDs, which poses a problem if other people have copies of the old commit that you've rewritten. Anyone who has a copy of the old commit will need to synchronize their work with your newly re-written commit, which can sometimes be difficult, so make sure you coordinate with others when attempting to rewrite shared commit history, or just avoid rewriting shared commits altogether.
Perform an interactive rebase
Another option is to use interactive rebase.
This allows you to edit any message you want to update even if it's not the latest message.
In order to do a Git squash, follow these steps:
// n is the number of commits up to the last commit you want to be able to edit
git rebase -i HEAD~n
Once you squash your commits - choose the e/r
for editing the message:
Important note about interactive rebase
When you use git rebase -i HEAD~n
there can be more than n commits. Git will "collect" all the commits in the last n commits, and if there was a merge somewhere in between that range you will see all the commits as well, so the outcome will be n + .
Good tip:
If you have to do it for more than a single branch and you might face conflicts when amending the content, set up git rerere
and let Git resolve those conflicts automatically for you.
Documentation
Best Answer
It's a good idea to have some sort of central repository, because it'll allow for you to share code, but also have a branch somewhere that you can directly generate your builds/export your snapshots from. That server will probably have more than one branch, one of which is thought of as a 'trunk' branch. Any previous releases will have their own branch, and depending on the hierarchy of your team (ie if you are divided into groups with each group working on one aspect of the application) then there may be team or feature based branches, though if you don't work that way that's not necessary.
Of course, because it is distributed each developer will also have their own local repository, to make things nice and fast. Or they can each have multiple repositories, even. For instance a developer who likes to work while commuting may have a repository on his workstation, and another one on his laptop, with branches on his laptop that are 'checked out' from the ones on his workstation. It's up to him. I guess that the 'distributed' part makes this kind of thing a lot easier, because you can commit and even branch while you are away from the network.
If you're transitioning from a non-distributed VCS, then you can just slip straight into the same model as before, because a DVCS is flexible enough to work in the same way. Otherwise, you can just start with a single central repository with a few branches, and it is always trivially easy to create more repositories and branches later.
One last thing is that you still need backups. The fact that various developers each have copies of the same thing adds redundancy, but it is not instead of backups.
The DVCS I use regularly is Bazaar. I have also tried Mercurial.