Re: [PATCH 09/18] setup: stop using `the_repository` in `setup_work_tree()`
- From
Patrick Steinhardt <ps@pks.im>
- Date
- Apr 20, 2026, 07:06 UTC
- Message-ID
- <aeXQcDF-wHT4wEG-@pks.im>
- In-Reply-To
- <CABPp-BHOvDaWXXDnpPsoDTe-En7T4LUaVqcJZ5Jst8zaQ-mrTg@mail.gmail.com>
On Fri, Apr 17, 2026 at 06:11:48PM -0700, Elijah Newren wrote:
Show 52 quoted lines
> On Mon, Mar 30, 2026 at 6:30 AM Patrick Steinhardt <ps@pks.im> wrote:
> >
> > Stop using `the_repository` in `setup_work_tree()` and instead accept
> > the repository as a parameter. The injection of `the_repository` is thus
> > bumped one level higher, where callers now pass it in explicitly.
> >
> > Note that the function tracks bogus worktree configuration via a global
> > variable. If we have bogus configuration, and if later on some caller
> > tries to setup a worktree, then we'll die instead.
> >
> > Of course, tracking this as a global variable doesn't make sense anymore
> > now that we can set up worktrees for arbitrary repositories. Move the
> > variable into `struct repository` instead.
>
> Okay, we don't want a global, makes sense.
>
> > Signed-off-by: Patrick Steinhardt <ps@pks.im>
>
> > diff --git a/setup.c b/setup.c
> > index dca32addae..64a030c6cd 100644
> > --- a/setup.c
> > +++ b/setup.c
> > @@ -26,7 +26,6 @@
> > #include "trace2.h"
> > #include "worktree.h"
> >
> > -static int work_tree_config_is_bogus;
>
> So, we removed this global....
>
> > enum allowed_bare_repo {
> > ALLOWED_BARE_REPO_EXPLICIT = 0,
> > ALLOWED_BARE_REPO_ALL,
> > @@ -485,7 +484,7 @@ int is_inside_work_tree(struct repository *repo)
> > return is_inside_dir(strbuf_realpath(&buf, worktree, 1));
> > }
> >
> > -void setup_work_tree(void)
> > +void setup_work_tree(struct repository *repo)
> > {
> > const char *work_tree;
> > static int initialized = 0;
>
> ...but we left this one...
>
> > @@ -493,10 +492,10 @@ void setup_work_tree(void)
> > if (initialized)
> > return;
>
> ...and because of this other global, we still can only set up one work
> tree in the process anyway. Should 'initialized' also be moved to a
> field within repo?Good catch. I thought I fixed this instance in a later commit or in the next patch series, but I seemingly don't. Will adapt, thanks!
Patrick