Git submodule for Linux and Windows
git submodule on Linux I might have a
.gitmodules with one or more subsystems checked out from e.g.
/mnt/gitrepos/subsystem1.git (central file based access).
[submodule "subsystem1"] path = subsystem11 url = /mnt/gitrepos/subsystem1.git/
I would like to support also a Windows user who can access the same module repo from
- How do I init/update a git submodule in a working tree after pushing to a bare working directory?
- Add a list of submodules to the git
- GitHub does not let me access Subfolders
- Git submodule head 'reference is not a tree' error
- Bash: Git submodule foreach?
- Setting upstream to a submodule (or how to include a GitHub fork as a submodule)
It is e.g. a Samba share from Linux of
/mnt/gitrepos/ as the share
Is there a way where Git can handle the URL line depending on the OS?
For Windows the
.gitmodules would be
[submodule "subsystem1"] path = subsystem11 url = H:/gitrepos/subsystem1.git/
Thus I would like “something like” this generic code for
.gitmodules (speculative syntax):
[submodule "subsystem1"] path = subsystem11 if Linux url = /mnt/gitrepos/subsystem1.git/ else url = H:/gitrepos/subsystem1.git/
2 Solutions collect form web for “Git submodule for Linux and Windows”
No. Submodules are generally designed for one repository URL that works everywhere, usually network-based (e.g.
git://host/path). There is no mechanism for providing multiple different URLs for one repository.
That said, Git does allow you to customize a submodule’s URL. When submodules are initialized (
git submodule init), the URL from
.gitmodules is copied to your
.git/config file. You can now edit the URLs there before running
git submodule update.
One workaround I’ve considered (but not tried) is to initialize the submodule codebase (“subsystem1.git” in your example) as an independent, local repository. Add a “remote” in the submodule which points to the independent, local version of the repository. Also add a “remote” in the non-submodule (indepent, local) version of your subsystem repository which points to the submodule version. You should then be able to use Samba / Windows / Linux to update the independent (non-submodule) version of subsystem1.git, and then “git fetch independent” from within the submodule to copy the code from the non-submodule repository to the submodule repository.
At the end, your directory structure should look something like:
/mnt/gitrepos/ /mnt/gitrepos/subsystem1.git/ /mnt/subsystem1.git/
The remotes in each of these repositories would look like the following:
/mnt/gitrepos/.git/config would have one remote block: [remote "origin"] (points to your main, non-local repository) /mnt/subsystem1.git/.git/config would have two remote blocks: [remote "origin"] (points to your main, non-local repository) [remote "local_origin"] (points to /mnt/gitrepos/subsystem1.git) /mnt/gitrepos/.git/modules/subsystem1.git/config would also have two remote blocks: [remote "origin"] (points to your main, non-local repository) [remote "independent"] (points to /mnt/subsystem1.git)
Synchronizing your submodule’s code with what’s on the remote server then becomes a multi-step process, since you have to use the “independent” repository as an intermediate repository / staging area.
Your Mileage May Vary… This is just a thing that seems like it should work, not something I’ve tried.