fetch/pull: recurse into submodules when necessary

To be able to access all commits of populated submodules referenced by the
superproject it is sufficient to only then let "git fetch" recurse into a
submodule when the new commits fetched in the superproject record new
commits for it. Having these commits present is extremely useful when
using the "--submodule" option to "git diff" (which is what "git gui" and
"gitk" do since 1.6.6), as all submodule commits needed for creating a
descriptive output can be accessed. Also merging submodule commits (added
in 1.7.3) depends on the submodule commits in question being present to
work. Last but not least this enables disconnected operation when using
submodules, as all commits necessary for a successful "git submodule
update -N" will have been fetched automatically. So we choose this mode as
the default for fetch and pull.

Before a new or changed ref from upstream is updated in update_local_ref()
"git rev-list <new-sha1> --not --branches --remotes" is used to determine
all newly fetched commits. These are then walked and diffed against their
parent(s) to see if a submodule has been changed. If that is the case, its
path is stored to be fetched after the superproject fetch is completed.

Using the "--recurse-submodules" or the "--no-recurse-submodules" option
disables the examination of the fetched refs because the result will be
ignored anyway.

There is currently no infrastructure for storing deleted and new
submodules in the .git directory of the superproject. That's why fetch and
pull for now only fetch submodules that are already checked out and are
not renamed.

In t7403 the "--no-recurse-submodules" argument had to be added to "git
pull" to avoid failure because of the moved upstream submodule repo.

Thanks-to: Jonathan Nieder <jrnieder@gmail.com>
Thanks-to: Heiko Voigt <hvoigt@hvoigt.net>
Signed-off-by: Jens Lehmann <Jens.Lehmann@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

Documentation/fetch-options.txt

@@ -73,6 +73,14 @@ ifndef::git-pull[]
 	Prepend <path> to paths printed in informative messages
 	such as "Fetching submodule foo".  This option is used
 	internally when recursing over submodules.
+
+--recurse-submodules-default=[yes|on-demand]::
+	This option is used internally to temporarily provide a
+	non-negative default value for the --recurse-submodules
+	option.  All other methods of configuring fetch's submodule
+	recursion (such as settings in linkgit:gitmodules[5] and
+	linkgit:git-config[1]) override this option, as does
+	specifying --[no-]recurse-submodules directly.
 endif::git-pull[]
 
 -u::

builtin/fetch.c

@@ -28,12 +28,6 @@ enum {
 	TAGS_SET = 2
 };
 
-enum {
-	RECURSE_SUBMODULES_OFF = 0,
-	RECURSE_SUBMODULES_DEFAULT = 1,
-	RECURSE_SUBMODULES_ON = 2
-};
-
 static int all, append, dry_run, force, keep, multiple, prune, update_head_ok, verbosity;
 static int progress, recurse_submodules = RECURSE_SUBMODULES_DEFAULT;
 static int tags = TAGS_DEFAULT;
@@ -42,6 +36,7 @@ static const char *upload_pack;
 static struct strbuf default_rla = STRBUF_INIT;
 static struct transport *transport;
 static const char *submodule_prefix = "";
+static const char *recurse_submodules_default;
 
 static struct option builtin_fetch_options[] = {
 	OPT__VERBOSITY(&verbosity),
@@ -73,6 +68,9 @@ static struct option builtin_fetch_options[] = {
 		   "deepen history of shallow clone"),
 	{ OPTION_STRING, 0, "submodule-prefix", &submodule_prefix, "DIR",
 		   "prepend this to submodule path output", PARSE_OPT_HIDDEN },
+	{ OPTION_STRING, 0, "recurse-submodules-default",
+		   &recurse_submodules_default, NULL,
+		   "default mode for recursion", PARSE_OPT_HIDDEN },
 	OPT_END()
 };
 
@@ -284,6 +282,9 @@ static int update_local_ref(struct ref *ref,
 		else {
 			msg = "storing head";
 			what = "[new branch]";
+			if ((recurse_submodules != RECURSE_SUBMODULES_OFF) &&
+			    (recurse_submodules != RECURSE_SUBMODULES_ON))
+				check_for_new_submodule_commits(ref->new_sha1);
 		}
 
 		r = s_update_ref(msg, ref, 0);
@@ -299,6 +300,9 @@ static int update_local_ref(struct ref *ref,
 		strcpy(quickref, find_unique_abbrev(current->object.sha1, DEFAULT_ABBREV));
 		strcat(quickref, "..");
 		strcat(quickref, find_unique_abbrev(ref->new_sha1, DEFAULT_ABBREV));
+		if ((recurse_submodules != RECURSE_SUBMODULES_OFF) &&
+		    (recurse_submodules != RECURSE_SUBMODULES_ON))
+			check_for_new_submodule_commits(ref->new_sha1);
 		r = s_update_ref("fast-forward", ref, 1);
 		sprintf(display, "%c %-*s %-*s -> %s%s", r ? '!' : ' ',
 			TRANSPORT_SUMMARY_WIDTH, quickref, REFCOL_WIDTH, remote,
@@ -310,6 +314,9 @@ static int update_local_ref(struct ref *ref,
 		strcpy(quickref, find_unique_abbrev(current->object.sha1, DEFAULT_ABBREV));
 		strcat(quickref, "...");
 		strcat(quickref, find_unique_abbrev(ref->new_sha1, DEFAULT_ABBREV));
+		if ((recurse_submodules != RECURSE_SUBMODULES_OFF) &&
+		    (recurse_submodules != RECURSE_SUBMODULES_ON))
+			check_for_new_submodule_commits(ref->new_sha1);
 		r = s_update_ref("forced-update", ref, 1);
 		sprintf(display, "%c %-*s %-*s -> %s  (%s)", r ? '!' : '+',
 			TRANSPORT_SUMMARY_WIDTH, quickref, REFCOL_WIDTH, remote,
@@ -949,9 +956,10 @@ int cmd_fetch(int argc, const char **argv, const char *prefix)
 	if (!result && (recurse_submodules != RECURSE_SUBMODULES_OFF)) {
 		const char *options[10];
 		int num_options = 0;
-		/* Set recursion as default when we already are recursing */
-		if (submodule_prefix[0])
-			set_config_fetch_recurse_submodules(1);
+		if (recurse_submodules_default) {
+			int arg = parse_fetch_recurse_submodules_arg("--recurse-submodules-default", recurse_submodules_default);
+			set_config_fetch_recurse_submodules(arg);
+		}
 		gitmodules_config();
 		git_config(submodule_config, NULL);
 		add_options_to_argv(&num_options, options);

submodule.c

@@ -12,7 +12,8 @@
 struct string_list config_name_for_path;
 struct string_list config_fetch_recurse_submodules_for_name;
 struct string_list config_ignore_for_name;
-static int config_fetch_recurse_submodules;
+static int config_fetch_recurse_submodules = RECURSE_SUBMODULES_ON_DEMAND;
+struct string_list changed_submodule_paths;
 
 static int add_submodule_odb(const char *path)
 {
@@ -152,6 +153,20 @@ void handle_ignore_submodules_arg(struct diff_options *diffopt,
 		die("bad --ignore-submodules argument: %s", arg);
 }
 
+int parse_fetch_recurse_submodules_arg(const char *opt, const char *arg)
+{
+	switch (git_config_maybe_bool(opt, arg)) {
+	case 1:
+		return RECURSE_SUBMODULES_ON;
+	case 0:
+		return RECURSE_SUBMODULES_OFF;
+	default:
+		if (!strcmp(arg, "on-demand"))
+			return RECURSE_SUBMODULES_ON_DEMAND;
+		die("bad %s argument: %s", opt, arg);
+	}
+}
+
 void show_submodule_summary(FILE *f, const char *path,
 		unsigned char one[20], unsigned char two[20],
 		unsigned dirty_submodule,
@@ -248,27 +263,95 @@ void set_config_fetch_recurse_submodules(int value)
 	config_fetch_recurse_submodules = value;
 }
 
+static void submodule_collect_changed_cb(struct diff_queue_struct *q,
+					 struct diff_options *options,
+					 void *data)
+{
+	int i;
+	for (i = 0; i < q->nr; i++) {
+		struct diff_filepair *p = q->queue[i];
+		if (!S_ISGITLINK(p->two->mode))
+			continue;
+
+		if (S_ISGITLINK(p->one->mode)) {
+			/* NEEDSWORK: We should honor the name configured in
+			 * the .gitmodules file of the commit we are examining
+			 * here to be able to correctly follow submodules
+			 * being moved around. */
+			struct string_list_item *path;
+			path = unsorted_string_list_lookup(&changed_submodule_paths, p->two->path);
+			if (!path)
+				string_list_append(&changed_submodule_paths, xstrdup(p->two->path));
+		} else {
+			/* Submodule is new or was moved here */
+			/* NEEDSWORK: When the .git directories of submodules
+			 * live inside the superprojects .git directory some
+			 * day we should fetch new submodules directly into
+			 * that location too when config or options request
+			 * that so they can be checked out from there. */
+			continue;
+		}
+	}
+}
+
+void check_for_new_submodule_commits(unsigned char new_sha1[20])
+{
+	struct rev_info rev;
+	struct commit *commit;
+	const char *argv[] = {NULL, NULL, "--not", "--all", NULL};
+	int argc = ARRAY_SIZE(argv) - 1;
+
+	init_revisions(&rev, NULL);
+	argv[1] = xstrdup(sha1_to_hex(new_sha1));
+	setup_revisions(argc, argv, &rev, NULL);
+	if (prepare_revision_walk(&rev))
+		die("revision walk setup failed");
+
+	/*
+	 * Collect all submodules (whether checked out or not) for which new
+	 * commits have been recorded upstream in "changed_submodule_paths".
+	 */
+	while ((commit = get_revision(&rev))) {
+		struct commit_list *parent = commit->parents;
+		while (parent) {
+			struct diff_options diff_opts;
+			diff_setup(&diff_opts);
+			diff_opts.output_format |= DIFF_FORMAT_CALLBACK;
+			diff_opts.format_callback = submodule_collect_changed_cb;
+			if (diff_setup_done(&diff_opts) < 0)
+				die("diff_setup_done failed");
+			diff_tree_sha1(parent->item->object.sha1, commit->object.sha1, "", &diff_opts);
+			diffcore_std(&diff_opts);
+			diff_flush(&diff_opts);
+			parent = parent->next;
+		}
+	}
+	free((char *)argv[1]);
+}
+
 int fetch_populated_submodules(int num_options, const char **options,
 			       const char *prefix, int ignore_config,
 			       int quiet)
 {
-	int i, result = 0, argc = 0;
+	int i, result = 0, argc = 0, default_argc;
 	struct child_process cp;
 	const char **argv;
 	struct string_list_item *name_for_path;
 	const char *work_tree = get_git_work_tree();
 	if (!work_tree)
-		return 0;
+		goto out;
 
 	if (!the_index.initialized)
 		if (read_cache() < 0)
 			die("index file corrupt");
 
-	/* 4: "fetch" (options) "--submodule-prefix" prefix NULL */
-	argv = xcalloc(num_options + 4, sizeof(const char *));
+	/* 6: "fetch" (options) --recurse-submodules-default default "--submodule-prefix" prefix NULL */
+	argv = xcalloc(num_options + 6, sizeof(const char *));
 	argv[argc++] = "fetch";
 	for (i = 0; i < num_options; i++)
 		argv[argc++] = options[i];
+	argv[argc++] = "--recurse-submodules-default";
+	default_argc = argc++;
 	argv[argc++] = "--submodule-prefix";
 
 	memset(&cp, 0, sizeof(cp));
@@ -282,7 +365,7 @@ int fetch_populated_submodules(int num_options, const char **options,
 		struct strbuf submodule_git_dir = STRBUF_INIT;
 		struct strbuf submodule_prefix = STRBUF_INIT;
 		struct cache_entry *ce = active_cache[i];
-		const char *git_dir, *name;
+		const char *git_dir, *name, *default_argv;
 
 		if (!S_ISGITLINK(ce->ce_mode))
 			continue;
@@ -292,15 +375,21 @@ int fetch_populated_submodules(int num_options, const char **options,
 		if (name_for_path)
 			name = name_for_path->util;
 
+		default_argv = "yes";
 		if (!ignore_config) {
 			struct string_list_item *fetch_recurse_submodules_option;
 			fetch_recurse_submodules_option = unsorted_string_list_lookup(&config_fetch_recurse_submodules_for_name, name);
 			if (fetch_recurse_submodules_option) {
 				if (!fetch_recurse_submodules_option->util)
 					continue;
 			} else {
-				if (!config_fetch_recurse_submodules)
+				if (config_fetch_recurse_submodules == RECURSE_SUBMODULES_OFF)
 					continue;
+				if (config_fetch_recurse_submodules == RECURSE_SUBMODULES_ON_DEMAND) {
+					if (!unsorted_string_list_lookup(&changed_submodule_paths, ce->name))
+						continue;
+					default_argv = "on-demand";
+				}
 			}
 		}
 
@@ -314,6 +403,7 @@ int fetch_populated_submodules(int num_options, const char **options,
 			if (!quiet)
 				printf("Fetching submodule %s%s\n", prefix, ce->name);
 			cp.dir = submodule_path.buf;
+			argv[default_argc] = default_argv;
 			argv[argc] = submodule_prefix.buf;
 			if (run_command(&cp))
 				result = 1;
@@ -323,6 +413,8 @@ int fetch_populated_submodules(int num_options, const char **options,
 		strbuf_release(&submodule_prefix);
 	}
 	free(argv);
+out:
+	string_list_clear(&changed_submodule_paths, 1);
 	return result;
 }
 

submodule.h

@@ -3,17 +3,26 @@
 
 struct diff_options;
 
+enum {
+	RECURSE_SUBMODULES_ON_DEMAND = -1,
+	RECURSE_SUBMODULES_OFF = 0,
+	RECURSE_SUBMODULES_DEFAULT = 1,
+	RECURSE_SUBMODULES_ON = 2
+};
+
 void set_diffopt_flags_from_submodule_config(struct diff_options *diffopt,
 		const char *path);
 int submodule_config(const char *var, const char *value, void *cb);
 void gitmodules_config();
 int parse_submodule_config_option(const char *var, const char *value);
 void handle_ignore_submodules_arg(struct diff_options *diffopt, const char *);
+int parse_fetch_recurse_submodules_arg(const char *opt, const char *arg);
 void show_submodule_summary(FILE *f, const char *path,
 		unsigned char one[20], unsigned char two[20],
 		unsigned dirty_submodule,
 		const char *del, const char *add, const char *reset);
 void set_config_fetch_recurse_submodules(int value);
+void check_for_new_submodule_commits(unsigned char new_sha1[20]);
 int fetch_populated_submodules(int num_options, const char **options,
 			       const char *prefix, int ignore_config,
 			       int quiet);