Re: [PATCHv6 5/7] worktree: add function to check if worktrees are in use

From: Stefan Beller <sbeller@google.com>
Date: Dec 8, 2016, 19:32 UTC
Message-ID: <CAGZ79kbNAZ81JqO+UEmKJC06oDRKWjxaF3Cb6jpZ3Xxo=des8g@mail.gmail.com>
In-Reply-To: <CACsJy8DMhC7Umnqf+-ViOBKn46Emtan4647=MDdLoYXVNB0=6A@mail.gmail.com>

On Thu, Dec 8, 2016 at 2:51 AM, Duy Nguyen <pclouds@gmail.com> wrote:

Show 17 quoted lines

> On Thu, Dec 8, 2016 at 5:40 PM, Duy Nguyen <pclouds@gmail.com> wrote:
>> Alternatively, we could add a new flag to get_worktrees() to tell it
>> to return all worktrees if there is a least one linked worktree, or
>> return NULL if there's only main worktree. I'm not sure if this is
>> clever or very stupid.
>
> No, this may be better. Add a flag to say "returns linked worktrees
> _only_". Which means when you're in a "normal" repo, get_worktrees()
> with this flag returns NULL. When you're in a multiple-worktree repo,
> it returns all linked worktrees (no main worktree). I think I might
> have a use for this flag in addition to this uses_worktrees() here.
> uses_worktrees() look quite simple with that flag
>
> int uses_worktrees(void)
> {
>     struct worktree **worktrees = get_worktrees(WT_LINKED_ONLY);
>     int retval = worktrees != NULL;

I am interested in the submodule case however, where we already return NULL e.g. when the submodule git dir cannot be found. Actually that would work out fine as well:

    /* NOTE on accuracy of result, hence not exposed. */
    static worktree **submodule_get_worktrees(const char *path, unsigned flags)
    ..

    int submodule_uses_worktrees(const char *path)
    {
        struct worktree **worktrees = submodule_get_worktrees(path,
WT_LINKED_ONLY);
        int retval = worktrees != NULL;
        free_worktrees(worktrees);
        return retval;
    }

Thanks for that inspiration!