> We used to recurse into submodules, even if they were broken having
> only an objects directory. The child process executed in the submodule
> would fail though if the submodule was broken.
>
> This patch tightens the check upfront, such that we do not need
> to spawn a child process to find out if the submodule is broken.
Thanks for the clear commit message. Also mention which tests currently
exercise this.
> +static struct repository *get_submodule_repo_for(struct repository *r,
> + const struct submodule *sub)
> +{
> + struct repository *ret = xmalloc(sizeof(*ret));
> +
> + if (repo_submodule_init(ret, r, sub)) {
> + /*
> + * No entry in .gitmodules? Technically not a submodule,
> + * but historically we supported repositories that happen to be
> + * in-place where a gitlink is. Keep supporting them.
> + */
> + struct strbuf gitdir = STRBUF_INIT;
> + strbuf_repo_worktree_path(&gitdir, r, "%s/.git", sub->path);
> + if (repo_init(ret, gitdir.buf, NULL)) {
> + strbuf_release(&gitdir);
You should also free ret here.
> + return NULL;
> + }
> + strbuf_release(&gitdir);
> + }
> +
> + return ret;
> +}
The above is the rest of get_submodule_repo_for(), so that we can see
that gitdir is indeed freed.
> - if (!git_dir)
> - git_dir = submodule_git_dir.buf;
> - if (is_directory(git_dir)) {
> + repo = get_submodule_repo_for(spf->r, submodule);
> + if (repo) {
> child_process_init(cp);
> - cp->dir = strbuf_detach(&submodule_path, NULL);
> prepare_submodule_repo_env(&cp->env_array);
> + cp->dir = xstrdup(repo->worktree);
Move the cp->dir one line up for a cleaner diff.
[snip]
> + repo_clear(repo);
> + free(repo);
> ret = 1;
> + } else {
> + /*
> + * An empty directory is normal,
> + * the submodule is not initialized
> + */
> + if (S_ISGITLINK(ce->ce_mode) &&
> + !is_empty_dir(ce->name))
> + die(_("Could not access submodule '%s'"), ce->name);
Previously, a failed fetch would just set spf->result = 1 (in
fetch_finish()), allowing other fetches to still proceed. This sounds
like a better idea to me instead of die-ing outright. (Also remember to
print a warning message - since we no longer spawn a child process to
try a fetch that will fail, we need to print a message ourselves.)
