Steven Haryanto

NAME

Git::Bunch - Manage gitbunch directory (directory which contain git repos)

VERSION

This document describes version 0.37 of Git::Bunch (from Perl distribution Git-Bunch), released on 2014-06-21.

SYNOPSIS

To check the status of bunch (will do a 'git status' for each git repo inside the bunch and report which repos are 'unclean', e.g. needs commit, has untracked files, etc):

 % gitbunch check ~/repos

To synchronize bunch to another (will do a 'git pull/push' for each git repo, and do an rsync for everything else):

 % gitbunch sync ~/repos /mnt/laptop/repos

DESCRIPTION

A gitbunch or bunch directory is just a term I coined to refer to a directory which contains, well, a bunch of git repositories. It can also contain other stuffs like files and non-git repositories (but they must be dot-dirs). Example:

 repos/            -> a gitbunch dir
   proj1/          -> a git repo
   proj2/          -> ditto
   perl-Git-Bunch/ -> ditto
   ...
   .foo/           -> a non-git dir
   README.txt      -> file

A little bit of history: after git got popular, in 2008 I started using it for software projects, replacing Subversion and Bazaar. Soon, I moved everything to git: notes & writings, Emacs .org agenda files, configuration, even temporary downloads/browser-saved HTML files. Currently, except large media files, all my personal data resides in git repositories. I put them all in ~/repos (and add symlinks to various places for convenience). This setup makes it easy to sync to laptops, backup to disk, etc. Git::Bunch is the library/script I wrote to do this.

See also File::RsyBak, which I wrote to backup everything else.

FUNCTIONS

check_bunch(%args) -> [status, msg, result, meta]

Check status of git repositories inside gitbunch directory.

Will perform a 'git status' for each git repositories inside the bunch and report which repositories are clean/unclean.

Will die if can't chdir into bunch or git repository.

Arguments ('*' denotes required arguments):

  • exclude_files => bool

    Exclude files from processing.

    This only applies to sync_bunch operations. Operations like check_bunch and exec_bunch already ignore these and only operate on git repos.

  • exclude_non_git_dirs => bool

    Exclude non-git dirs from processing.

    This only applies to and sync_bunch operations. Operations like check_bunch and exec_bunch already ignore these and only operate on git repos.

  • exclude_repos => array

    Exclude some repos from processing.

  • exclude_repos_pat => str

    Specify regex pattern of repos to exclude.

  • include_repos => array

    Specific git repos to sync, if not specified all repos in the bunch will be processed.

  • include_repos_pat => str

    Specify regex pattern of repos to include.

  • sort => str (default: "-commit-timestamp")

    Order entries in bunch.

    commit-timestamp (and -commit-timestamp) compares the timestamp of .git/commit-timestamp file in each repo. Repos or dirs not having this file will be processed later. You can touch these .git/commit-timestamp files in your post-commit script, for example. This allows sorting recently committed repos more cheaply (compared to doing git log -1).

  • source* => str

    Directory to check.

Return value:

Returns an enveloped result (an array).

First element (status) is an integer containing HTTP status code (200 means OK, 4xx caller error, 5xx function error). Second element (msg) is a string containing error message, or 'OK' if status is 200. Third element (result) is optional, the actual result. Fourth element (meta) is called result metadata and is optional, a hash that contains extra information.

exec_bunch(%args) -> [status, msg, result, meta]

Execute a command for each repo in the bunch.

For each git repository in the bunch, will chdir to it and execute specified command.

Arguments ('*' denotes required arguments):

  • command* => str

    Command to execute.

  • exclude_files => bool

    Exclude files from processing.

    This only applies to sync_bunch operations. Operations like check_bunch and exec_bunch already ignore these and only operate on git repos.

  • exclude_non_git_dirs => bool

    Exclude non-git dirs from processing.

    This only applies to and sync_bunch operations. Operations like check_bunch and exec_bunch already ignore these and only operate on git repos.

  • exclude_repos => array

    Exclude some repos from processing.

  • exclude_repos_pat => str

    Specify regex pattern of repos to exclude.

  • include_repos => array

    Specific git repos to sync, if not specified all repos in the bunch will be processed.

  • include_repos_pat => str

    Specify regex pattern of repos to include.

  • sort => str (default: "-commit-timestamp")

    Order entries in bunch.

    commit-timestamp (and -commit-timestamp) compares the timestamp of .git/commit-timestamp file in each repo. Repos or dirs not having this file will be processed later. You can touch these .git/commit-timestamp files in your post-commit script, for example. This allows sorting recently committed repos more cheaply (compared to doing git log -1).

  • source* => str

    Directory to check.

Return value:

Returns an enveloped result (an array).

First element (status) is an integer containing HTTP status code (200 means OK, 4xx caller error, 5xx function error). Second element (msg) is a string containing error message, or 'OK' if status is 200. Third element (result) is optional, the actual result. Fourth element (meta) is called result metadata and is optional, a hash that contains extra information.

sync_bunch(%args) -> [status, msg, result, meta]

Synchronize bunch to another bunch.

For each git repository in the bunch, will perform a 'git pull/push' for each branch. If repository in destination doesn't exist, it will be rsync-ed first from source. When 'git pull' fails, will exit to let you fix the problem manually.

For all other non-git repos, will simply synchronize by one-way rsync.

Arguments ('*' denotes required arguments):

  • backup => bool

    Whether doing backup to target.

    This setting lets you express that you want to perform synchronizing to a backup target, and that you do not do work on the target. Thus, you do not care about uncommitted or untracked files/dirs in the target repos (might happen if you also do periodic copying of repos to backup using cp/rsync). When this setting is turned on, the function will first do a git clean -f -d (to delete untracked files/dirs) and then git checkout . (to discard all uncommitted changes). This setting will also implicitly turn on create_bare setting (unless that setting has been explicitly enabled/disabled).

  • create_bare_target => bool

    Whether to create bare git repo when target does not exist.

    When target repo does not exist, gitbunch can either copy the source repo using rsync (the default, if this setting is undefined), or it can create target repo with git init --bare (if this setting is set to 1), or it can create target repo with git init (if this setting is set to 0).

    Bare git repositories contain only contents of the .git folder inside the directory and no working copies of your source files.

    Creating bare repos are apt for backup purposes since they are more space-efficient.

    Non-repos will still be copied/rsync-ed.

  • delete_branch => bool (default: 0)

    Whether to delete branches in dest repos not existing in source repos.

  • exclude_files => bool

    Exclude files from processing.

    This only applies to sync_bunch operations. Operations like check_bunch and exec_bunch already ignore these and only operate on git repos.

  • exclude_non_git_dirs => bool

    Exclude non-git dirs from processing.

    This only applies to and sync_bunch operations. Operations like check_bunch and exec_bunch already ignore these and only operate on git repos.

  • exclude_repos => array

    Exclude some repos from processing.

  • exclude_repos_pat => str

    Specify regex pattern of repos to exclude.

  • include_repos => array

    Specific git repos to sync, if not specified all repos in the bunch will be processed.

  • include_repos_pat => str

    Specify regex pattern of repos to include.

  • rsync_opt_maintain_ownership => bool (default: 0)

    Whether or not, when rsync-ing from source, we use -a (= -rlptgoD) or -rlptD (-a minus -go).

    Sometimes using -a results in failure to preserve permission modes on sshfs-mounted filesystem, while -rlptD succeeds, so by default we don't maintain ownership. If you need to maintain ownership (e.g. you run as root and the repos are not owned by root), turn this option on.

  • sort => str (default: "-commit-timestamp")

    Order entries in bunch.

    commit-timestamp (and -commit-timestamp) compares the timestamp of .git/commit-timestamp file in each repo. Repos or dirs not having this file will be processed later. You can touch these .git/commit-timestamp files in your post-commit script, for example. This allows sorting recently committed repos more cheaply (compared to doing git log -1).

  • source* => str

    Directory to check.

  • target* => str

    Destination bunch.

Return value:

Returns an enveloped result (an array).

First element (status) is an integer containing HTTP status code (200 means OK, 4xx caller error, 5xx function error). Second element (msg) is a string containing error message, or 'OK' if status is 200. Third element (result) is optional, the actual result. Fourth element (meta) is called result metadata and is optional, a hash that contains extra information.

FAQ

TODO

  • Can't handle bare source repos

  • Sync-ing to bare repos (e.g. with --backup) still produces error messages

    Although the sync process succeeds, the error messages (like fatal: Not a git repository) might be confusing.

SEE ALSO

mr, http://joeyh.name/code/mr/ . You probably want to use this instead. mr supports other control version software aside from git, doesn't restrict you to put all your repos in one directory, supports more operations, and has been developed since 2007. Had I known about mr, I probably wouldn't have started Git::Bunch. On the other hand, Git::Bunch is simpler (I think), doesn't require any config file, and can copy/sync files/directories not under source control. I mainly use Git::Bunch to quickly: 1) check whether there are any of my repositories which have uncommitted changes; 2) synchronize (pull/push) to other locations. I put all my data in one big gitbunch directory; I find it simpler. Git::Bunch works for me and I use it daily.

HOMEPAGE

Please visit the project's homepage at https://metacpan.org/release/Git-Bunch.

SOURCE

Source repository is at https://github.com/sharyanto/perl-Git-Bunch.

BUGS

Please report any bugs or feature requests on the bugtracker website https://rt.cpan.org/Public/Dist/Display.html?Name=Git-Bunch

When submitting a bug or request, please include a test-file or a patch to an existing test-file that illustrates the bug or desired feature.

AUTHOR

Steven Haryanto <stevenharyanto@gmail.com>

COPYRIGHT AND LICENSE

This software is copyright (c) 2014 by Steven Haryanto.

This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself.




Hosting generously
sponsored by Bytemark