git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: [PATCH v3 1/3] worktree: remove "the_repository" from is_current_worktree()

From
Junio C Hamano <gitster@pobox.com>
Date
Mar 26, 2026, 15:48 UTC
Message-ID
<xmqqy0jer3qu.fsf@gitster.g>
In-Reply-To
<5357c0dd53ee123a4ea064412c83983b0be5e400.1774534617.git.phillip.wood@dunelm.org.uk>
Phillip Wood <phillip.wood123@gmail.com> writes:
Show 7 quoted lines
> ... it calls get_workree_git_dir() which also depends on
> "the_repository". This means that even if "wt->path" matches
> "wt->repo->worktree" is_current_worktree(wt) will return false when
> "wt->repo" is not "the_repository". Consequently die_if_checked_out()
> will fail to skip such a worktree when checking if a branch is already
> checked out and may die errounously. Fix this by using the worktree's
> repository instance instead of "the_repository" when comparing gitdirs.

It sounds like the above is something we can write in a test to make sure the code with this fix won't regress in the future. Can we add one?

Show 36 quoted lines
> The use of "the_repository" in is_current_wortree() comes from
> replacing get_git_dir() with repo_get_git_dir() in 246deeac951
> (environment: make `get_git_dir()` accept a repository, 2024-09-12). In
> get_worktree_git_dir() it comes from replacing git_common_path() with
> repo_common_path() in 07242c2a5af (path: drop `git_common_path()`
> in favor of `repo_common_path()`, 2025-02-07). In both cases the
> replacements appear to have been mechanical.
>
> Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk>
> ---
>  worktree.c | 8 ++++----
>  worktree.h | 2 +-
>  2 files changed, 5 insertions(+), 5 deletions(-)
>
> diff --git a/worktree.c b/worktree.c
> index e9ff6e6ef2e..344ad0c031b 100644
> --- a/worktree.c
> +++ b/worktree.c
> @@ -58,7 +58,7 @@ static void add_head_info(struct worktree *wt)
>  
>  static int is_current_worktree(struct worktree *wt)
>  {
> -	char *git_dir = absolute_pathdup(repo_get_git_dir(the_repository));
> +	char *git_dir = absolute_pathdup(repo_get_git_dir(wt->repo));
>  	char *wt_git_dir = get_worktree_git_dir(wt);
>  	int is_current = !fspathcmp(git_dir, absolute_path(wt_git_dir));
>  	free(wt_git_dir);
> @@ -78,7 +78,7 @@ struct worktree *get_worktree_from_repository(struct repository *repo)
>  	wt->is_bare = !repo->worktree;
>  	if (fspathcmp(gitdir, commondir))
>  		wt->id = xstrdup(find_last_dir_sep(gitdir) + 1);
> -	wt->is_current = is_current_worktree(wt);
> +	wt->is_current = true;
>  	add_head_info(wt);
>  
>  	free(gitdir);

I think I found the semantics of get_worktree_from_repository() unclear when we discussed a different patch series, so I may have asked the same question already, but what exactly does it mean to "get worktree from repository", i.e., this function does? A repository can have one or more worktrees attached to it (that was the whole point of introducing the worktree mechanism), so "I have this repository, give me the worktree for it" is not a sensible request. The function's comment in <worktree.h> talks only at the implementation level "construct a worktree struct from repo->gitdir and repo->worktree" as if it is so obvious what the resulting worktree struct means at a higher layer's point of view, which does not help, either.

I am asking the above because I find this unconditional assignment of "true" utterly confusing. Is it because "we created a brand new worktree structure that by definition is current out of a repository structure, so as far as the caller is concerned by definition it is what it is currently working on"?

Stepping back a bit, how do "is_current_worktree(wt)" and "wt->is_current" relate to each other? Here is what happens in the former:

        static int is_current_worktree(struct worktree *wt)
        {
                char *git_dir = absolute_pathdup(repo_get_git_dir(wt->repo));
                char *wt_git_dir = get_worktree_git_dir(wt);
                int is_current = !fspathcmp(git_dir, absolute_path(wt_git_dir));
                free(wt_git_dir);
                free(git_dir);
                return is_current;
        }

and given the definition of get_worktree_git_dir(), which gives either the "common" .git directory for the primary worktree, or the worktree specific .git directory for the secondaries, the is_current being computed here sounds more like "is wt the primary worktree among the ones that are attached to the same repository?"

But given that there are API functions with "main" in their names, like is_main_worktree() and internal function get_main_worktree(), what I said apparently is not the intention of whoever built the worktree subsystem. I am still confused... X-<.

Show 22 quoted lines
> @@ -229,9 +229,9 @@ char *get_worktree_git_dir(const struct worktree *wt)
>  	if (!wt)
>  		return xstrdup(repo_get_git_dir(the_repository));
>  	else if (!wt->id)
> -		return xstrdup(repo_get_common_dir(the_repository));
> +		return xstrdup(repo_get_common_dir(wt->repo));
>  	else
> -		return repo_common_path(the_repository, "worktrees/%s", wt->id);
> +		return repo_common_path(wt->repo, "worktrees/%s", wt->id);
>  }
>  
>  static struct worktree *find_worktree_by_suffix(struct worktree **list,
> diff --git a/worktree.h b/worktree.h
> index e450d1a3317..94ae58db973 100644
> --- a/worktree.h
> +++ b/worktree.h
> @@ -16,7 +16,7 @@ struct worktree {
>  	struct object_id head_oid;
>  	int is_detached;
>  	int is_bare;
> -	int is_current;
> +	int is_current;		/* does `path` match `repo->worktree` */

This new comment also talks only at the lowest implementation level. What significance does it have to the upper layer that calls into the API if `path` matches or does not match `repo->worktree` is totally unclear, at least to me.

>  	int lock_reason_valid; /* private */
>  	int prune_reason_valid; /* private */
>  };
Thanks.
Previous: Phillip WoodNext: Phillip Wood
Message 22 of 27 in “worktree: stop using "the_repository" in is_current_worktree()”
  1. 0/3 worktree: stop using "the_repository" in is_current_worktree()Phillip Wood, Mar 13, 2026
  2. 1/3 worktree: remove "the_repository" from is_current_worktree()Phillip Wood, Mar 13, 2026
  3. 2/3 worktree add: stop reading ".git/HEAD"Phillip Wood, Mar 13, 2026
  4. Junio C HamanoMar 13, 2026
  5. 3/3 worktree: reject NULL worktree in get_worktree_git_dir()Phillip Wood, Mar 13, 2026
  6. Junio C HamanoMar 13, 2026
  7. Phillip WoodMar 14, 2026
  8. 0/3 worktree: stop using "the_repository" in is_current_worktree()Phillip Wood, Mar 15, 2026
  9. 1/3 worktree: remove "the_repository" from is_current_worktree()Phillip Wood, Mar 15, 2026
  10. Patrick SteinhardtMar 16, 2026
  11. Phillip WoodMar 16, 2026
  12. Phillip WoodMar 17, 2026
  13. Shreyansh PaliwalMar 23, 2026
  14. Phillip WoodMar 23, 2026
  15. Shreyansh PaliwalMar 23, 2026
  16. 2/3 worktree add: stop reading ".git/HEAD"Phillip Wood, Mar 15, 2026
  17. Patrick SteinhardtMar 16, 2026
  18. 3/3 worktree: reject NULL worktree in get_worktree_git_dir()Phillip Wood, Mar 15, 2026
  19. Junio C HamanoMar 15, 2026
  20. 0/3 worktree: stop using "the_repository" in is_current_worktree()Phillip Wood, Mar 26, 2026
  21. 1/3 worktree: remove "the_repository" from is_current_worktree()Phillip Wood, Mar 26, 2026
  22. Junio C HamanoMar 26, 2026
  23. Phillip WoodMar 27, 2026
  24. Junio C HamanoMar 27, 2026
  25. Phillip WoodApr 2, 2026
  26. 2/3 worktree add: stop reading ".git/HEAD"Phillip Wood, Mar 26, 2026
  27. 3/3 worktree: reject NULL worktree in get_worktree_git_dir()Phillip Wood, Mar 26, 2026

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.