pull: optionally rebase submodules (remote submodule changes only)

Teach pull to optionally update submodules when '--recurse-submodules' is provided. This will teach pull to run 'submodule update --rebase' when the '--recurse-submodules' and '--rebase' flags are given under specific circumstances. On a rebase workflow: ===================== 1. Both sides change the submodule ------------------------------ Let's assume the following history in a submodule: H---I---J---K---L local branch \ M---N---O---P remote branch and the following in the superproject (recorded submodule in parens): A(H)---B(I)---F(K)---G(L) local branch \ C(N)---D(N)---E(P) remote branch In an ideal world this would rebase the submodule and rewrite the submodule pointers that the superproject points at such that the superproject looks like A(H)---B(I) F(K')---G(L') rebased branch \ / C(N)---D(N)---E(P) remote branch and the submodule as: J---K---L (old dangeling tip) / H---I J'---K'---L' rebased branch \ / M---N---O---P remote branch And if a conflict arises in the submodule the superproject rebase would stop at that commit at which the submodule conflict occurs. Currently a "pull --rebase" in the superproject produces a merge conflict as the submodule pointer changes are conflicting and cannot be resolved. 2. Local submodule changes only ----------------------- Assuming histories as above, except that the remote branch would not contain submodule changes, then a result as A(H)---B(I) F(K)---G(L) rebased branch \ / C(I)---D(I)---E(I) remote branch is desire-able. This is what currently happens in rebase. If the recursive flag is given, the ideal git would produce a superproject as: A(H)---B(I) F(K')---G(L') rebased branch (incl. sub rebase!) \ / C(I)---D(I)---E(I) remote branch and the submodule as: J---K---L (old dangeling tip) / H---I J'---K'---L' locally rebased branch \ / M---N---O---P advanced branch This patch doesn't address this issue, however a test is added that this fails up front. 3. Remote submodule changes only ---------------------- Assuming histories as in (1) except that the local superproject branch would not have touched the submodule the rebase already works out in the superproject with no conflicts: A(H)---B(I) F(P)---G(P) rebased branch (no sub changes) \ / C(N)---D(N)---E(P) remote branch The recurse flag as presented in this patch would additionally update the submodule as: H---I J'---K'---L' rebased branch \ / M---N---O---P remote branch As neither J, K, L nor J', K', L' are referred to from the superproject, no rewriting of the superproject commits is required. Conclusion for 'pull --rebase --recursive' ----------------------------------------- If there are no local superproject changes it is sufficient to call "submodule update --rebase" as this produces the desired results. In case of conflicts, the behavior is the same as in 'submodule update --recursive' which is assumed to be sane. This patch implements (3) only. On a merge workflow: ==================== We'll start off with the same underlying DAG as in (1) in the rebase workflow. So in an ideal world a 'pull --merge --recursive' would produce this: H---I---J---K---L----X \ / M---N---O---P with X as the new merge-commit in the submodule and the superproject as: A(H)---B(I)---F(K)---G(L)---Y(X) \ / C(N)---D(N)---E(P) However modifying the submodules on the fly is not supported in git-merge such that Y(X) is not easy to produce in a single patch. In fact git-merge doesn't know about submodules at all. However when at least one side does not contain commits touching the submodule at all, then we do not need to perform the merge for the submodule but a fast-forward can be done via checking out either L or P in the submodule. This strategy is implemented in 68d03e4a6e (Implement automatic fast-forward merge for submodules, 2010-07-07) already, so to align with the rebase behavior we need to also update the worktree of the submodule. Signed-off-by: Brandon Williams <bmwill@google.com> Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>

Stefan Beller committed Jun 23, 2017 at 12:13 UTC a6d7eb2c7a6a402a938824bcf1c5f331dd1a06bb
5 files changed +157 -16
Documentation/git-pull.txt
+6 -6
@@ -86,12 +86,12 @@ OPTIONS
86
87 --[no-]recurse-submodules[=yes|on-demand|no]::
88 This option controls if new commits of all populated submodules should
89 - be fetched too (see linkgit:git-config[1] and linkgit:gitmodules[5]).
90 - That might be necessary to get the data needed for merging submodule
91 - commits, a feature Git learned in 1.7.3. Notice that the result of a
92 - merge will not be checked out in the submodule, "git submodule update"
93 - has to be called afterwards to bring the work tree up to date with the
94 - merge result.
89 + be fetched and updated, too (see linkgit:git-config[1] and
90 + linkgit:gitmodules[5]).
91 ++
92 +If the checkout is done via rebase, local submodule commits are rebased as well.
93 ++
94 +If the update is done via merge, the submodule conflicts are resolved and checked out.
95
96 Options related to merging
97 ~~~~~~~~~~~~~~~~~~~~~~~~~~
builtin/pull.c
+63 -10
@@ -15,6 +15,8 @@
15 #include "dir.h"
16 #include "refs.h"
17 #include "revision.h"
18 +#include "submodule.h"
19 +#include "submodule-config.h"
20 #include "tempfile.h"
21 #include "lockfile.h"
22 #include "wt-status.h"
@@ -77,6 +79,7 @@ static const char * const pull_usage[] = {
79 /* Shared options */
80 static int opt_verbosity;
81 static char *opt_progress;
82 +static int recurse_submodules = RECURSE_SUBMODULES_DEFAULT;
83
84 /* Options passed to git-merge or git-rebase */
85 static enum rebase_type opt_rebase = -1;
@@ -101,7 +104,6 @@ static char *opt_upload_pack;
104 static int opt_force;
105 static char *opt_tags;
106 static char *opt_prune;
104 -static char *opt_recurse_submodules;
107 static char *max_children;
108 static int opt_dry_run;
109 static char *opt_keep;
@@ -116,6 +118,10 @@ static struct option pull_options[] = {
118 OPT_PASSTHRU(0, "progress", &opt_progress, NULL,
119 N_("force progress reporting"),
120 PARSE_OPT_NOARG),
121 + { OPTION_CALLBACK, 0, "recurse-submodules",
122 + &recurse_submodules, N_("on-demand"),
123 + N_("control for recursive fetching of submodules"),
124 + PARSE_OPT_OPTARG, option_fetch_parse_recurse_submodules },
125
126 /* Options passed to git-merge or git-rebase */
127 OPT_GROUP(N_("Options related to merging")),
@@ -187,10 +193,6 @@ static struct option pull_options[] = {
193 OPT_PASSTHRU('p', "prune", &opt_prune, NULL,
194 N_("prune remote-tracking branches no longer on remote"),
195 PARSE_OPT_NOARG),
190 - OPT_PASSTHRU(0, "recurse-submodules", &opt_recurse_submodules,
191 - N_("on-demand"),
192 - N_("control recursive fetching of submodules"),
193 - PARSE_OPT_OPTARG),
196 OPT_PASSTHRU('j', "jobs", &max_children, N_("n"),
197 N_("number of submodules pulled in parallel"),
198 PARSE_OPT_OPTARG),
@@ -483,8 +485,20 @@ static int run_fetch(const char *repo, const char **refspecs)
485 argv_array_push(&args, opt_tags);
486 if (opt_prune)
487 argv_array_push(&args, opt_prune);
486 - if (opt_recurse_submodules)
487 - argv_array_push(&args, opt_recurse_submodules);
488 + if (recurse_submodules != RECURSE_SUBMODULES_DEFAULT)
489 + switch (recurse_submodules) {
490 + case RECURSE_SUBMODULES_ON:
491 + argv_array_push(&args, "--recurse-submodules=on");
492 + break;
493 + case RECURSE_SUBMODULES_OFF:
494 + argv_array_push(&args, "--recurse-submodules=no");
495 + break;
496 + case RECURSE_SUBMODULES_ON_DEMAND:
497 + argv_array_push(&args, "--recurse-submodules=on-demand");
498 + break;
499 + default:
500 + BUG("submodule recursion option not understood");
501 + }
502 if (max_children)
503 argv_array_push(&args, max_children);
504 if (opt_dry_run)
@@ -531,6 +545,30 @@ static int pull_into_void(const struct object_id *merge_head,
545 return 0;
546 }
547
548 +static int rebase_submodules(void)
549 +{
550 + struct child_process cp = CHILD_PROCESS_INIT;
551 +
552 + cp.git_cmd = 1;
553 + cp.no_stdin = 1;
554 + argv_array_pushl(&cp.args, "submodule", "update",
555 + "--recursive", "--rebase", NULL);
556 +
557 + return run_command(&cp);
558 +}
559 +
560 +static int update_submodules(void)
561 +{
562 + struct child_process cp = CHILD_PROCESS_INIT;
563 +
564 + cp.git_cmd = 1;
565 + cp.no_stdin = 1;
566 + argv_array_pushl(&cp.args, "submodule", "update",
567 + "--recursive", "--checkout", NULL);
568 +
569 + return run_command(&cp);
570 +}
571 +
572 /**
573 * Runs git-merge, returning its exit status.
574 */
@@ -862,6 +900,11 @@ int cmd_pull(int argc, const char **argv, const char *prefix)
900 die(_("Cannot rebase onto multiple branches."));
901
902 if (opt_rebase) {
903 + int ret = 0;
904 + if ((recurse_submodules == RECURSE_SUBMODULES_ON ||
905 + recurse_submodules == RECURSE_SUBMODULES_ON_DEMAND) &&
906 + submodule_touches_in_range(&rebase_fork_point, &curr_head))
907 + die(_("cannot rebase with locally recorded submodule modifications"));
908 if (!autostash) {
909 struct commit_list *list = NULL;
910 struct commit *merge_head, *head;
@@ -872,11 +915,21 @@ int cmd_pull(int argc, const char **argv, const char *prefix)
915 if (is_descendant_of(merge_head, list)) {
916 /* we can fast-forward this without invoking rebase */
917 opt_ff = "--ff-only";
875 - return run_merge();
918 + ret = run_merge();
919 }
920 }
878 - return run_rebase(&curr_head, merge_heads.oid, &rebase_fork_point);
921 + ret = run_rebase(&curr_head, merge_heads.oid, &rebase_fork_point);
922 +
923 + if (!ret && (recurse_submodules == RECURSE_SUBMODULES_ON ||
924 + recurse_submodules == RECURSE_SUBMODULES_ON_DEMAND))
925 + ret = rebase_submodules();
926 +
927 + return ret;
928 } else {
880 - return run_merge();
929 + int ret = run_merge();
930 + if (!ret && (recurse_submodules == RECURSE_SUBMODULES_ON ||
931 + recurse_submodules == RECURSE_SUBMODULES_ON_DEMAND))
932 + ret = update_submodules();
933 + return ret;
934 }
935 }
submodule.c
+26
@@ -1126,6 +1126,32 @@ static void calculate_changed_submodule_paths(void)
1126 initialized_fetch_ref_tips = 0;
1127 }
1128
1129 +int submodule_touches_in_range(struct object_id *excl_oid,
1130 + struct object_id *incl_oid)
1131 +{
1132 + struct string_list subs = STRING_LIST_INIT_DUP;
1133 + struct argv_array args = ARGV_ARRAY_INIT;
1134 + int ret;
1135 +
1136 + gitmodules_config();
1137 + /* No need to check if there are no submodules configured */
1138 + if (!submodule_from_path(NULL, NULL))
1139 + return 0;
1140 +
1141 + argv_array_push(&args, "--"); /* args[0] program name */
1142 + argv_array_push(&args, oid_to_hex(incl_oid));
1143 + argv_array_push(&args, "--not");
1144 + argv_array_push(&args, oid_to_hex(excl_oid));
1145 +
1146 + collect_changed_submodules(&subs, &args);
1147 + ret = subs.nr;
1148 +
1149 + argv_array_clear(&args);
1150 +
1151 + free_submodules_oids(&subs);
1152 + return ret;
1153 +}
1154 +
1155 struct submodule_parallel_fetch {
1156 int count;
1157 struct argv_array args;
submodule.h
+4
@@ -97,6 +97,10 @@ extern int merge_submodule(struct object_id *result, const char *path,
97 const struct object_id *base,
98 const struct object_id *a,
99 const struct object_id *b, int search);
100 +
101 +/* Checks if there are submodule changes in a..b. */
102 +extern int submodule_touches_in_range(struct object_id *a,
103 + struct object_id *b);
104 extern int find_unpushed_submodules(struct oid_array *commits,
105 const char *remotes_name,
106 struct string_list *needs_pushing);
t/t5572-pull-submodule.sh
+58
@@ -42,4 +42,62 @@ KNOWN_FAILURE_NOFF_MERGE_DOESNT_CREATE_EMPTY_SUBMODULE_DIR=1
42 KNOWN_FAILURE_NOFF_MERGE_ATTEMPTS_TO_MERGE_REMOVED_SUBMODULE_FILES=1
43 test_submodule_switch "git_pull_noff"
44
45 +test_expect_success 'pull --recurse-submodule setup' '
46 + test_create_repo child &&
47 + test_commit -C child bar &&
48 +
49 + test_create_repo parent &&
50 + test_commit -C child foo &&
51 +
52 + git -C parent submodule add ../child sub &&
53 + git -C parent commit -m "add submodule" &&
54 +
55 + git clone --recurse-submodules parent super
56 +'
57 +
58 +test_expect_success 'recursive pull updates working tree' '
59 + test_commit -C child merge_strategy &&
60 + git -C parent submodule update --remote &&
61 + git -C parent add sub &&
62 + git -C parent commit -m "update submodule" &&
63 +
64 + git -C super pull --no-rebase --recurse-submodules &&
65 + test_path_is_file super/sub/merge_strategy.t
66 +'
67 +
68 +test_expect_success 'recursive rebasing pull' '
69 + # change upstream
70 + test_commit -C child rebase_strategy &&
71 + git -C parent submodule update --remote &&
72 + git -C parent add sub &&
73 + git -C parent commit -m "update submodule" &&
74 +
75 + # also have local commits
76 + test_commit -C super/sub local_stuff &&
77 +
78 + git -C super pull --rebase --recurse-submodules &&
79 + test_path_is_file super/sub/rebase_strategy.t &&
80 + test_path_is_file super/sub/local_stuff.t
81 +'
82 +
83 +test_expect_success 'pull rebase recursing fails with conflicts' '
84 +
85 + # local changes in submodule recorded in superproject:
86 + test_commit -C super/sub local_stuff_2 &&
87 + git -C super add sub &&
88 + git -C super commit -m "local update submodule" &&
89 +
90 + # and in the remote as well:
91 + test_commit -C child important_upstream_work &&
92 + git -C parent submodule update --remote &&
93 + git -C parent add sub &&
94 + git -C parent commit -m "remote update submodule" &&
95 +
96 + # Unfortunately we fail here, despite no conflict in the
97 + # submodule itself, but the merge strategy in submodules
98 + # does not support rebase:
99 + test_must_fail git -C super pull --rebase --recurse-submodules 2>err &&
100 + test_i18ngrep "locally recorded submodule modifications" err
101 +'
102 +
103 test_done