The clone command creates a new git repository, initialized from
a TFS source tree and fetch all the changesets
Usage: git-tfs clone [options] tfs-url-or-instance-name repository-path <git-repository-path>
ex : git tfs clone http://myTfsServer:8080/tfs/TfsRepository $/ProjectName/ProjectBranch
-h, -H, --help
-V, --version
-d, --debug Show debug output about everything git-tfs does
-i, --tfs-remote, --remote, --id=VALUE
The remote ID of the TFS to interact with
default: default
--template=VALUE Passed to git-init
--shared[=VALUE] Passed to git-init
--autocrlf=VALUE Normalize line endings (default: false)
--ignorecase=VALUE Ignore case in file paths (default: system default)
--bare clone the TFS repository in a bare git repository
--workspace=VALUE set tfs workspace to a specific folder (a shorter path is better!)
--gitignore=VALUE Path toward the .gitignore file which be
committed and used to ignore files
--ignore-regex=VALUE a regex of files to ignore
--except-regex=VALUE a regex of exceptions to ignore-regex
-u, --username=VALUE TFS username
-p, --password=VALUE TFS password
--no-parallel Do not do parallel requests to TFS
--all, --fetch-all
--parents
--authors=VALUE Path to an Authors file to map TFS users to Git users
-l, --with-labels, --fetch-labels
Fetch the labels also when fetching TFS changesets
-x, --export Export metadata
--export-work-item-mapping=VALUE
Path to Work-items mapping export file
--branches=VALUE Strategy to manage branches:
* none: Ignore branches and merge changesets,
fetching only the clone tfs path
* auto:(default) Manage merged changesets and
initialize the merged branches
* all: Manage merged changesets and initialize
all the branches during the clone
--ignore-branches-regex=VALUE
Don't initialize branches that match given regex
--ignore-not-init-branches
Don't initialize additional branches (only use what already was initialized)
--batch-size=VALUE Size of the batch of tfs changesets fetched (-1 for all in one batch)
-c, --changeset=VALUE The changeset to clone from (must be a number)
-t, --up-to=VALUE up-to changeset # (optional, -1 for up to
maximum, must be a number, not prefixed with C)
--resumable if an error occurred, try to continue when you restart clone
with same parameters
To clone all of $/Project1 from your TFS server tfs
into a new directory Project1, do this:
git tfs clone http://tfs:8080/tfs/DefaultCollection $/Project1
Note: Equivalent to cloning with dependency branches (with option --branches=auto) if you are cloning the trunk branch.
To clone from a specific changeset in the history of $/Project1 from your TFS server tfs
into a new directory Project1, do this:
git tfs clone http://tfs:8080/tfs/DefaultCollection $/Project1 -c=126
where 126 is the id of the changeset to clone.
This command will get all the history from this specific changeset. It could be especially useful when you have a huge history and also when the entire history could not be clone due to not supported tfs specificities!
Sometimes, it could be interesting to clone only a branch of a TFS repository (for example to extract only the trunk of your project and manage branches using the branch command.
Suppose you have on TFS:
A <- B <- C <- D <- E $/Project1/Trunk
\
M <- N $/Project1/Branch
Then, do this (the clone will be done in the MyProject1Directory directory):
git tfs clone http://tfs:8080/tfs/DefaultCollection $/Project1/Trunk MyProject1Directory
This command is equivalent to specifying explicitly the option --branches=auto:
git tfs clone http://tfs:8080/tfs/DefaultCollection $/Project1/Trunk MyProject1Directory --branches=auto
Note:
- Since v0.21, git-tfs will also initialize all the branches that have been merged in the trunk during its changeset history and will try to manage all the merge changesets accordingly (See Merge changesets and branches for more details).
- It is highly recommended to clone the root branch ( the branch that has no parents, here
$/Project1/Trunk ) to be able to init the other branches after. If you clone the branch $ /Project1/Branch, you will never able to init the root branch $/Project1/Trunk after. - Some complex branch scenario possible with TFS are actually not supported and the clone could end up with an error.
If that's the case, you will be obliged to clone without branch support and use the option
--branches=none
If you want to clone the trunk without cloning dependency branches (because for example it fails at managing these branches) or you want to clone a child branch instead, you could use the option --branches=none.
Then, do this (the clone will be done in the MyProject1Directory directory):
git tfs clone http://tfs:8080/tfs/DefaultCollection $/Project1/Trunk MyProject1Directory --branches=none
Note: * This is a way to clone that has the more chances to succeed (when you have a complex branch history with some cases not supported by git-tfs). * This is the one to choose when you clone another tfs path that the trunk. * All the merged branches are ignored and merge changesets are treated as normal changesets.
Since version v0.21, when cloning the trunk from TFS, if git-tfs encounters a merge changeset, it initializes and fetches automatically the other branch merged.
Suppose you have on TFS:
A <- B <- C <- D <- E <- X <- Y <- Z $/Project1/Trunk
\ /
M <- N <- O <- P <- Q $/Project1/Branch
When cloning the tfs branch $/Project1/Trunk, after having fetched changesets A to E, git-tfs encountered merge changeset X. When it did, git-tfs also initialized the TFS branch $/Project1/Branch and fetched changesets M to O to be able to create the merge commit X and then continued to fetch changesets Y and Z.
If you don't want to initialize the merged branches automatically (or you can't because your use of TFS is not supported), you could use the option --branches=none to disable it.
Note: To successfully process the merge changeset (or come from a version older than TFS2010), you should have converted all the folders corresponding to a TFS branch to a branch in TFS (even the old deleted branches). To do that, open the 'Source Control Explorer', right click on a folder and choose Branching and Merging -> Convert to Branch.
If you don't know exactly what repository path to clone, see list-remote-branches command to get a list of the existing repositories.
Prerequisite: To use this feature, all your source code folders corresponding to branches
should be converted into branches (a notion introduced by TFS2010).
To change that, you should open 'Source Control Explorer' and then, for each folder corresponding to a branch, right-click on your source folder and select Branching and Merging -> Convert to Branch.
If you want to clone your entire repository with all the branches or that the TFS branches are merged through merge changeset, perhaps you should use the option --branches=all:
git tfs clone http://tfs:8080/tfs/DefaultCollection $/Project1/Trunk --branches=all
All the TFS history (and all the branches) and the merge changesets will consequently be fetched from TFS and created in the git repository.
Note:
- Here again, some complex branch scenario possible with TFS are actually not supported and the clone could end up with an error.
If that's the case, you will be obliged to clone without branch support and use the option
--branches=none
See quick-clone.
Let's say you want to clone $/Project, but you don't want to
clone exes.
git tfs clone --ignore-regex=exe$ http://tfs:8080/tfs/DefaultCollection $/Project1
You could also use the --except-regex parameter to add an exception to the previous rule:
git tfs clone --ignore-regex=exe$ --except-regex=i_want_this.exe$ http://tfs:8080/tfs/DefaultCollection $/Project1
To ignore some files, and prevent theses files to be committed, you could also provide a .gitignore file to git-tfs.
The .gitignore file will be committed as the first commit of the repository and then will be used by git-tfs to ignore all the files
matching one of the regex in the file. You need to give the path toward of an external .gitignore which will be used as a template.
git tfs clone http://tfs:8080/tfs/DefaultCollection $/Project1 --gitignore="c:\path\toward\a\.gitignore"
You could download a .gitignore file for your language or project from the github repository
or generate one for multiple languages using gitignore.io
If the TFS server need an authentication, you could use the --username and --password parameters. If you don't specify theses informations, you will be prompted to enter them. If you use these parameters, the informations, git-tfs will store these informations (in the .git/config file --in clear--) and never prompt you again. If you don't want your password to be saved, don't use these options.
git tfs clone http://tfs:8080/tfs/DefaultCollection $/Project1 -u=DISSRVTFS03\peter.pan -p=wendy
With the parameter --authors, you could specify a file containing all the mapping of the TFS users to the git users. Each line describing a mapping following the syntax:
DISSRVTFS03\peter.pan = Peter Pan <[email protected]>
The clone command will be :
git tfs clone http://tfs:8080/tfs/DefaultCollection $/Project1/Trunk --authors="c:\project_file\authors.txt"
Once the clone is done, the file is store in the .git folder (with the name git-tfs_authors) and used with later fetch. You could overwrite it by specifying another file (or go delete it).
Note: You could use the tf history command to help you find all the Tfs users logins that should be found in the authors.txt file.
tf history $/Project1/Trunk /collection:"http://tfs:8080/tfs/TeamProjectCollectionUrl" /recursive | awk '{print $2}' | tail -n+3 | sort -u > authors.txt
Be aware that if your user logins contain spaces, you will need to use the cut command instead. The parameters of the cut command (column of beginning and column of end) depend on multiple parameters and you surely will have to find them experimentally.
The best way is perhaps to run the command 2 times and look inside the first file generated authors_tmp.txt where the users column began and end.
tf history $/Project1/Trunk /collection:"http://tfs:8080/tfs/TeamProjectCollectionUrl" /recursive > authors_tmp.txt
cat authors_tmp.txt | cut -b 11-28 | tail -n+3 | sort -u > authors.txt
By default, git-tfs use as a Tfs workspace an internal directory and you shouldn't care about ;) But, due to file system limitations, it could be useful to set a custom directory (with a path as short as possible) as a tfs workspace. You could do it with the --workspace parameter:
git tfs clone http://tfs:8080/tfs/DefaultCollection $/Project1 --workspace="c:\ws"
The option --export permit, when fetching changesets and creating commits, to add all the tfs metadatas to the commit message (work items ids, reviewers, ...).
This option could be used with other option --export-work-item-mapping that specify a mapping file to convert old work items ids to new work items ids.
It could be used to migrate sources away from TFSVC. See Migrate from tfs to git for more details.
The option --batch-size permit to specify the number of changesets fetched from tfs at the same time (default:100).
You could use this option to specify smaller batch size if git-tfs use too much memory because some changesets are huge.
This option is saved in the git config file (key git-tfs.batch-size). See config file doc.
Note: this option could also be specified during the fetch.
You can clone all projects by specifying $/ as the tfs-repository path. If you do not specify a git repository name, it will clone into tfs-collection.
It is recommended, especially if the TFS repository is a big one, to run, after a clone :
- a git garbage collect :
git gc - a cleanup :
git tfs cleanup
Feel free also to look at some special use cases: