Re: [PATCH v9 2/4] branch: let delete_branches() name where a branch landed
- From
Junio C Hamano <gitster@pobox.com>
- Date
- Nov 20, 2026, 19:32 UTC
- Message-ID
- <xmqqeccfpacc.fsf@gitster.g>
- In-Reply-To
- <c4513f9ec1315c34b7a75b95bf50bba59043c274.1795184664.git.gitgitgadget@gmail.com>
"Harald Nordgren via GitGitGadget" <gitgitgadget@gmail.com> writes:
Show 24 quoted lines
> @@ -357,15 +359,33 @@ static int delete_branches(int argc, const char **argv, int kinds,
> for_each_string_list_item(item, &refs_to_delete) {
> char *describe_ref = item->util;
> char *name = item->string;
> + struct commit *landed = landed_commits ?
> + strmap_get(landed_commits, name + branch_name_pos) : NULL;
> + const char *landed_abbrev = landed ?
> + repo_find_unique_abbrev(the_repository,
> + &landed->object.oid,
> + DEFAULT_ABBREV) : NULL;
> +
> if (flags & DELETE_BRANCH_DRY_RUN) {
> - if (!(flags & DELETE_BRANCH_QUIET))
> + if (flags & DELETE_BRANCH_QUIET)
> + ;
> + else if (landed)
> + printf(_("Would delete branch %s (was %s, landed as %s).\n"),
> + name + branch_name_pos, describe_ref,
> + landed_abbrev);
> + else
> printf(remote_branch
> ? _("Would delete remote-tracking branch %s (was %s).\n")
> : _("Would delete branch %s (was %s).\n"),
> name + branch_name_pos, describe_ref);The use of an empty "if" statement body is a clean way to express that nothing will be emitted under "--quiet" without making the code too deeply nested. Nicely done.
This is not a new issue, but this makes me wonder if giving the "--dry-run" and the "--quiet" options together makes sense. Would users really gain much information from "--quiet" when they are so paranoid as to want a preview of what would happen with "--dry-run"?
Show 12 quoted lines
> } else if (!refs_ref_exists(get_main_ref_store(the_repository), name)) {
> char *refname = name + branch_name_pos;
> - if (!(flags & DELETE_BRANCH_QUIET))
> + if (flags & DELETE_BRANCH_QUIET)
> + ;
> + else if (landed)
> + printf(_("Deleted branch %s (was %s, landed as %s).\n"),
> + refname, describe_ref, landed_abbrev);
> + else
> printf(remote_branch
> ? _("Deleted remote-tracking branch %s (was %s).\n")
> : _("Deleted branch %s (was %s).\n"),Likewise.
I wonder if it makes the code structure cleaner to follow if we swap the order of checks between "are we doing dry-run" (which is on the outside in the current code structure) and "what is the reason we remove this branch" (which is the inner if/elseif cascade) by having these different message templates (currently a pair of three-message sets) in an array and choosing which set to use based on "--dry-run". There are of course a few things that are done only in the non-dry-run mode (like removing the branch configuration and the branch itself). In other words, a code structure like this
const char *msg_templates[2][3] = {
{ "Would delete branch %s ...", ..., ..., },
{ "Deleted branch %s ...", ..., ..., },
};
const char *use_msg[3] = msg_templates[!dry_run]; for-each-branch-to-be-deleted {
compute if branch has landed;
if (quiet)
;
else if (landed)
printf(_(use_msg[0]), ...);
else if (remote_branch)
printf(_(use_msg[1]), ...);
else
printf(_(use_msg[2]), ...);
if (!dry_run)
remove branch and its configuration;
else
what is only done in dry-run mode (if exists)
}
may make it easier to follow for readers?