TW

But `socket.makefile` returns a python `file` object, not a OS-level FD (`int`).

@PhrozenByte btw, windows support is coming, @RayyanAnsari works on it (no remote repos yet though).

About `Location` specifier for `--repo`: ``` socket:///abs/path/to/borg.sock::/abs/path/to/borg.repo ``` Is that a good, usual syntax or are there better ideas? We need to transport socket path and repo path via that...

Thanks for the feedback (I had similar ideas/concerns, but just wanted something simple to get going and then refine after feedback). The reason why I chose double-slash + socket_abs_path is...

A more radical approach could be splitting that url into `--remote=SERVER_URL --repo=local_repo_path`.

@PhrozenByte about stderr: yeah, the borg serve function does not really use it, but the borg repository code emits logging messages on stderr (which usually go either directly to e.g....

about socket file **in** the repo directory: that would limit operations to existing repos, we can't have the socket file in a repo dir that does not yet exist. so...

For most of the logic, we want to use `platformdirs` package, they have a `user_runtime_dir()` function: https://github.com/platformdirs/platformdirs/blob/main/src/platformdirs/unix.py#L155 Misc. fs helpers already exist there: https://github.com/borgbackup/borg/blob/master/src/borg/helpers/fs.py Note: `platformdirs` does not special case...

**socket file permissions:** due to `constants.UMASK_DEFAULT == 0o0077`, the socket file has `0o700` permissions (user only, no permissions for group). Would that be a problem? Usually we have these scenarios:...

@PhrozenByte could you file a bug at `platformdirs`?