# [PATCH] mv: allow moving nested submodules

4 messages from 2016-04-18 to 2016-04-18. Participants: Stefan Beller, Junio C Hamano.
Thread: https://gitlist.dev/t/42064

## Stefan Beller, 2016-04-18 16:54

Subject: [PATCH] mv: allow moving nested submodules
Message-ID: <1460998489-2155-1-git-send-email-sbeller@google.com>
URL: https://gitlist.dev/e/1460998489-2155-1-git-send-email-sbeller%40google.com

```
When directories are moved using `git mv` all files in the directory
have been just moved, but no further action was taken on them. This
was done by assigning the mode = WORKING_DIRECTORY to the files
inside a moved directory.

submodules however need to update their link to the git directory as
well as updates to the .gitmodules file. By removing the condition of
`mode != INDEX` (the remaining modes are BOTH and WORKING_DIRECTORY) for
the required submodule actions, we perform these for submodules in a
moved directory.

Signed-off-by: Stefan Beller <sbeller@google.com>
---
 builtin/mv.c  |  7 +++++--
 t/t7001-mv.sh | 16 ++++++++++++++++
 2 files changed, 21 insertions(+), 2 deletions(-)

diff --git a/builtin/mv.c b/builtin/mv.c
index aeae855..65fff11 100644
--- a/builtin/mv.c
+++ b/builtin/mv.c
@@ -252,9 +252,12 @@ int cmd_mv(int argc, const char **argv, const char *prefix)
 		int pos;
 		if (show_only || verbose)
 			printf(_("Renaming %s to %s\n"), src, dst);
-		if (!show_only && mode != INDEX) {
-			if (rename(src, dst) < 0 && !ignore_errors)
+		if (!show_only) {
+			if (mode != INDEX &&
+			    rename(src, dst) < 0 &&
+			    !ignore_errors)
 				die_errno(_("renaming '%s' failed"), src);
+
 			if (submodule_gitfile[i]) {
 				if (submodule_gitfile[i] != SUBMODULE_WITH_GITDIR)
 					connect_work_tree_and_git_dir(dst, submodule_gitfile[i]);
diff --git a/t/t7001-mv.sh b/t/t7001-mv.sh
index 4008fae..4a2570e 100755
--- a/t/t7001-mv.sh
+++ b/t/t7001-mv.sh
@@ -292,6 +292,9 @@ test_expect_success 'setup submodule' '
 	echo content >file &&
 	git add file &&
 	git commit -m "added sub and file" &&
+	mkdir -p deep/directory/hierachy &&
+	git submodule add ./. deep/directory/hierachy/sub &&
+	git commit -m "added another submodule" &&
 	git branch submodule
 '
 
@@ -475,4 +478,17 @@ test_expect_success 'mv -k does not accidentally destroy submodules' '
 	git checkout .
 '
 
+test_expect_success 'moving a submodule in nested directories' '
+	(
+		cd deep &&
+		git mv directory ../ &&
+		# git status would fail if the update of linking git dir to
+		# work dir of the submodule failed.
+		git status &&
+		git config -f ../.gitmodules submodule.deep/directory/hierachy/sub.path >../actual &&
+		echo "directory/hierachy/sub" >../expect
+	) &&
+	test_cmp actual expect
+'
+
 test_done
-- 
2.8.0.26.gba39a1b.dirty

```

## Junio C Hamano, 2016-04-18 20:54

Subject: Re: [PATCH] mv: allow moving nested submodules
Message-ID: <xmqqk2ju4ozy.fsf@gitster.mtv.corp.google.com>
URL: https://gitlist.dev/e/xmqqk2ju4ozy.fsf%40gitster.mtv.corp.google.com
In-Reply-To: <1460998489-2155-1-git-send-email-sbeller@google.com>

```
Stefan Beller <sbeller@google.com> writes:

>  		if (show_only || verbose)
>  			printf(_("Renaming %s to %s\n"), src, dst);
> -		if (!show_only && mode != INDEX) {
> -			if (rename(src, dst) < 0 && !ignore_errors)
> +		if (!show_only) {
> +			if (mode != INDEX &&
> +			    rename(src, dst) < 0 &&
> +			    !ignore_errors)
>  				die_errno(_("renaming '%s' failed"), src);
> +

If ignore-errors is set and rename fails, this would fall through
and try to touch this codepath...

>  			if (submodule_gitfile[i]) {
>  				if (submodule_gitfile[i] != SUBMODULE_WITH_GITDIR)
>  					connect_work_tree_and_git_dir(dst, submodule_gitfile[i]);

... but I am not sure if this thing is prepared to cope with such a
case?  src should have been moved to dst but if rename() failed we
wouldn't see what we expect at dst, or would we?

> diff --git a/t/t7001-mv.sh b/t/t7001-mv.sh
> index 4008fae..4a2570e 100755
> --- a/t/t7001-mv.sh
> +++ b/t/t7001-mv.sh
> @@ -292,6 +292,9 @@ test_expect_success 'setup submodule' '
>  	echo content >file &&
>  	git add file &&
>  	git commit -m "added sub and file" &&
> +	mkdir -p deep/directory/hierachy &&
> +	git submodule add ./. deep/directory/hierachy/sub &&
> +	git commit -m "added another submodule" &&
>  	git branch submodule
>  '
>  
> @@ -475,4 +478,17 @@ test_expect_success 'mv -k does not accidentally destroy submodules' '
>  	git checkout .
>  '
>  
> +test_expect_success 'moving a submodule in nested directories' '
> +	(
> +		cd deep &&
> +		git mv directory ../ &&
> +		# git status would fail if the update of linking git dir to
> +		# work dir of the submodule failed.
> +		git status &&
> +		git config -f ../.gitmodules submodule.deep/directory/hierachy/sub.path >../actual &&
> +		echo "directory/hierachy/sub" >../expect
> +	) &&
> +	test_cmp actual expect
> +'
> +
>  test_done

```

## Junio C Hamano, 2016-04-18 21:13

Subject: Re: [PATCH] mv: allow moving nested submodules
Message-ID: <xmqqfuui4o4e.fsf@gitster.mtv.corp.google.com>
URL: https://gitlist.dev/e/xmqqfuui4o4e.fsf%40gitster.mtv.corp.google.com
In-Reply-To: <xmqqk2ju4ozy.fsf@gitster.mtv.corp.google.com>

```
Junio C Hamano <gitster@pobox.com> writes:

> If ignore-errors is set and rename fails, this would fall through
> and try to touch this codepath...
>
>>  			if (submodule_gitfile[i]) {
>>  				if (submodule_gitfile[i] != SUBMODULE_WITH_GITDIR)
>>  					connect_work_tree_and_git_dir(dst, submodule_gitfile[i]);
>
> ... but I am not sure if this thing is prepared to cope with such a
> case?  src should have been moved to dst but if rename() failed we
> wouldn't see what we expect at dst, or would we?

In other words, I was wondering if this part should read more like
this:

 builtin/mv.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/builtin/mv.c b/builtin/mv.c
index aeae855..37ed0fc 100644
--- a/builtin/mv.c
+++ b/builtin/mv.c
@@ -252,9 +252,14 @@ int cmd_mv(int argc, const char **argv, const char *prefix)
 		int pos;
 		if (show_only || verbose)
 			printf(_("Renaming %s to %s\n"), src, dst);
-		if (!show_only && mode != INDEX) {
-			if (rename(src, dst) < 0 && !ignore_errors)
+		if (show_only)
+			;
+		else {
+			if (mode != INDEX && rename(src, dst) < 0) {
+				if (ignore_errors)
+					continue;
 				die_errno(_("renaming '%s' failed"), src);
+			}
 			if (submodule_gitfile[i]) {
 				if (submodule_gitfile[i] != SUBMODULE_WITH_GITDIR)
 					connect_work_tree_and_git_dir(dst, submodule_gitfile[i]);

```

## Stefan Beller, 2016-04-18 21:26

Subject: Re: [PATCH] mv: allow moving nested submodules
Message-ID: <CAGZ79kbVTUzKmSa9KjEJyDuRCtW5rygQmSYMpssLUKYoO1ooSw@mail.gmail.com>
URL: https://gitlist.dev/e/CAGZ79kbVTUzKmSa9KjEJyDuRCtW5rygQmSYMpssLUKYoO1ooSw%40mail.gmail.com
In-Reply-To: <xmqqfuui4o4e.fsf@gitster.mtv.corp.google.com>

```
On Mon, Apr 18, 2016 at 2:13 PM, Junio C Hamano <gitster@pobox.com> wrote:
> Junio C Hamano <gitster@pobox.com> writes:
>
>> If ignore-errors is set and rename fails, this would fall through
>> and try to touch this codepath...
>>
>>>                      if (submodule_gitfile[i]) {
>>>                              if (submodule_gitfile[i] != SUBMODULE_WITH_GITDIR)
>>>                                      connect_work_tree_and_git_dir(dst, submodule_gitfile[i]);
>>
>> ... but I am not sure if this thing is prepared to cope with such a
>> case?  src should have been moved to dst but if rename() failed we
>> wouldn't see what we expect at dst, or would we?
>
> In other words, I was wondering if this part should read more like
> this:
>
>  builtin/mv.c | 9 +++++++--
>  1 file changed, 7 insertions(+), 2 deletions(-)
>
> diff --git a/builtin/mv.c b/builtin/mv.c
> index aeae855..37ed0fc 100644
> --- a/builtin/mv.c
> +++ b/builtin/mv.c
> @@ -252,9 +252,14 @@ int cmd_mv(int argc, const char **argv, const char *prefix)
>                 int pos;
>                 if (show_only || verbose)
>                         printf(_("Renaming %s to %s\n"), src, dst);
> -               if (!show_only && mode != INDEX) {
> -                       if (rename(src, dst) < 0 && !ignore_errors)
> +               if (show_only)
> +                       ;
> +               else {
> +                       if (mode != INDEX && rename(src, dst) < 0) {

I agree until here.


> +                               if (ignore_errors)
> +                                       continue;
>                                 die_errno(_("renaming '%s' failed"), src);

This I thought would be better as:

    if (!ignore_errors)
        die_errno(...);

and not continue, but continuing is the right thing I would expect.

Speaking of which, connect_work_tree_and_git_dir as well as
update_path_in_gitmodules need to learn about the ignore_errors
flag, too.  You would expect them to not die at the faintest problem,
but rather honor the promise of "Skip move or rename actions which
would lead to an error condition."

Thanks for a starting pointer for a new patch!
Stefan

> +                       }
>                         if (submodule_gitfile[i]) {
>                                 if (submodule_gitfile[i] != SUBMODULE_WITH_GITDIR)
>                                         connect_work_tree_and_git_dir(dst, submodule_gitfile[i]);

```
