Nguyen Thai Ngoc Duy <pclouds@xxxxxxxxx> wrote: > I think it's a valid use case. Anyone remember why Matthias' patchset > was dropped? > It was last mentioned in > http://article.gmane.org/gmane.comp.version-control.git/43041 > > Junio, Matthias? May I help? Thanks for reminding me, I just did not finish the patch and then there was an exam which took all my time. But I took another look at it and made a new series, including the missing test cases. If you want to help please test it, check the source for errors, make comments etc. This series introduces the GIT_WORK_TREE environment variable (and core.worktree config option) to specify the working tree that should be used with the repository (not for repositories found as .git directory). This allows to separate the repository and working tree. Example use cases: - you don't want to put the repository in the checkout (e.g. the checkout is publicly available but the history shouldn't) - you want to track a read-only directory with git (e.g. track configuration files as normal user that are modified by other people or by yourself as superuser) - there is a directory many people are working in and you can track the changes made by yourself and others without placing the .git directory in the directory - checkout multiple repositories into the same directory - see the mail form nguyen (Message-ID: <fcaeb9bf0705300742u22b54c78vccbc037fb553141f@xxxxxxxxxxxxxx>) The patches are also available from git://igit.ath.cx/~matled/git in branch worktree (I might change this branch later) or as tag worktree1. [PATCH] rev-parse: document --is-inside-git-dir [PATCH] rev-parse: introduce --is-bare-repository [PATCH] test git rev-parse [PATCH] introduce GIT_WORK_TREE to specify the work tree [PATCH] use new semantics of is_bare/inside_git_dir/inside_work_tree [PATCH] extend rev-parse test for --is-inside-work-tree [PATCH] test GIT_WORK_TREE There are also a few things which are not addressed in this series: - The documentation needs updates on what bare means. - is_bare_repository should ignore trailing slashes when guessing if the repository is bare (/path/to/.git is not bare, /path/to/.git/ is bare if core.bare is unspecified). Perhaps we can also set GIT_DIR to the path returned by getcwd which would solve this too. - Aliases using --git-dir and/or --work-tree cause problems (also before this patch). git should probably exec itself if an alias uses --git-dir or --work-tree. - Calls to git_config(git_default_config) can be removed in many places because setup_git_directory_gently always reads the configuration. - git init could set core.worktree if GIT_WORK_TREE is set. - To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
