[PATCH] submodule.c: report the submodule that an error occurs in

[PATCH] submodule.c: report the submodule that an error occurs in

[Stefan Beller]

When an error occurs in updating the working tree of a submodule in
submodule_move_head, tell the user which submodule the error occurred in.

The call to read-tree contains a super-prefix, such that the read-tree
will correctly report any path related issues, but some error messages
do not contain a path, for example:

  ~/gerrit$ git checkout --recurse-submodules origin/master
  ~/gerrit$ fatal: failed to unpack tree object 07672f31880ba80300b38492df9d0acfcd6ee00a

Give the hint which submodule has a problem.

Signed-off-by: Stefan Beller <sbeller@google.com>
---
 submodule.c               | 2 +-
 t/lib-submodule-update.sh | 3 ++-
 2 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/submodule.c b/submodule.c
index 939d6870ecd..ebd092a14fd 100644
--- a/submodule.c
+++ b/submodule.c
@@ -1668,7 +1668,7 @@ int submodule_move_head(const char *path,
 	argv_array_push(&cp.args, new_head ? new_head : empty_tree_oid_hex());
 
 	if (run_command(&cp)) {
-		ret = -1;
+		ret = error(_("Submodule '%s' could not be updated."), path);
 		goto out;
 	}
 
diff --git a/t/lib-submodule-update.sh b/t/lib-submodule-update.sh
index 1f38a85371a..e27f5d8541d 100755
--- a/t/lib-submodule-update.sh
+++ b/t/lib-submodule-update.sh
@@ -781,7 +781,8 @@ test_submodule_recursing_with_args_common() {
 		(
 			cd submodule_update &&
 			git branch -t invalid_sub1 origin/invalid_sub1 &&
-			test_must_fail $command invalid_sub1 &&
+			test_must_fail $command invalid_sub1 2>err &&
+			grep sub1 err &&
 			test_superproject_content origin/add_sub1 &&
 			test_submodule_content sub1 origin/add_sub1
 		)
-- 
2.18.0.rc2.346.g013aa6912e-goog

Re: [PATCH] submodule.c: report the submodule that an error occurs in

[Junio C Hamano]

Stefan Beller <sbeller@google.com> writes:

> When an error occurs in updating the working tree of a submodule in
> submodule_move_head, tell the user which submodule the error occurred in.
>
> The call to read-tree contains a super-prefix, such that the read-tree
> will correctly report any path related issues, but some error messages
> do not contain a path, for example:
>
>   ~/gerrit$ git checkout --recurse-submodules origin/master
>   ~/gerrit$ fatal: failed to unpack tree object 07672f31880ba80300b38492df9d0acfcd6ee00a
>
> Give the hint which submodule has a problem.
>
> Signed-off-by: Stefan Beller <sbeller@google.com>
> ---
>  submodule.c               | 2 +-
>  t/lib-submodule-update.sh | 3 ++-
>  2 files changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/submodule.c b/submodule.c
> index 939d6870ecd..ebd092a14fd 100644
> --- a/submodule.c
> +++ b/submodule.c
> @@ -1668,7 +1668,7 @@ int submodule_move_head(const char *path,
>  	argv_array_push(&cp.args, new_head ? new_head : empty_tree_oid_hex());
>  
>  	if (run_command(&cp)) {
> -		ret = -1;
> +		ret = error(_("Submodule '%s' could not be updated."), path);
>  		goto out;
>  	}

OK.  The function uses path in messages elsewhere so we know path
must not be empty and have a reasonable string content at this
point, which makes this a reasonably safe patch to apply ;-)

Thanks.

> diff --git a/t/lib-submodule-update.sh b/t/lib-submodule-update.sh
> index 1f38a85371a..e27f5d8541d 100755
> --- a/t/lib-submodule-update.sh
> +++ b/t/lib-submodule-update.sh
> @@ -781,7 +781,8 @@ test_submodule_recursing_with_args_common() {
>  		(
>  			cd submodule_update &&
>  			git branch -t invalid_sub1 origin/invalid_sub1 &&
> -			test_must_fail $command invalid_sub1 &&
> +			test_must_fail $command invalid_sub1 2>err &&
> +			grep sub1 err &&
>  			test_superproject_content origin/add_sub1 &&
>  			test_submodule_content sub1 origin/add_sub1
>  		)

Re: [PATCH] submodule.c: report the submodule that an error occurs in

[SZEDER Gábor]

> When an error occurs in updating the working tree of a submodule in
> submodule_move_head, tell the user which submodule the error occurred in.
> 
> The call to read-tree contains a super-prefix, such that the read-tree
> will correctly report any path related issues, but some error messages
> do not contain a path, for example:
> 
>   ~/gerrit$ git checkout --recurse-submodules origin/master
>   ~/gerrit$ fatal: failed to unpack tree object 07672f31880ba80300b38492df9d0acfcd6ee00a
> 
> Give the hint which submodule has a problem.
> 
> Signed-off-by: Stefan Beller <sbeller@google.com>
> ---
>  submodule.c               | 2 +-
>  t/lib-submodule-update.sh | 3 ++-
>  2 files changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/submodule.c b/submodule.c
> index 939d6870ecd..ebd092a14fd 100644
> --- a/submodule.c
> +++ b/submodule.c
> @@ -1668,7 +1668,7 @@ int submodule_move_head(const char *path,
>  	argv_array_push(&cp.args, new_head ? new_head : empty_tree_oid_hex());
>  
>  	if (run_command(&cp)) {
> -		ret = -1;
> +		ret = error(_("Submodule '%s' could not be updated."), path);

This is a translated error message ...

>  		goto out;
>  	}
>  
> diff --git a/t/lib-submodule-update.sh b/t/lib-submodule-update.sh
> index 1f38a85371a..e27f5d8541d 100755
> --- a/t/lib-submodule-update.sh
> +++ b/t/lib-submodule-update.sh
> @@ -781,7 +781,8 @@ test_submodule_recursing_with_args_common() {
>  		(
>  			cd submodule_update &&
>  			git branch -t invalid_sub1 origin/invalid_sub1 &&
> -			test_must_fail $command invalid_sub1 &&
> +			test_must_fail $command invalid_sub1 2>err &&
> +			grep sub1 err &&

... so the test should use 'test_i18ngrep' to check it.

>  			test_superproject_content origin/add_sub1 &&
>  			test_submodule_content sub1 origin/add_sub1
>  		)
> -- 
> 2.18.0.rc2.346.g013aa6912e-goog
> 
> 

Re: [PATCH] submodule.c: report the submodule that an error occurs in

[Junio C Hamano]

SZEDER Gábor <szeder.dev@gmail.com> writes:

>> When an error occurs in updating the working tree of a submodule in
>> submodule_move_head, tell the user which submodule the error occurred in.
>> 
>> The call to read-tree contains a super-prefix, such that the read-tree
>> will correctly report any path related issues, but some error messages
>> do not contain a path, for example:
>> 
>>   ~/gerrit$ git checkout --recurse-submodules origin/master
>>   ~/gerrit$ fatal: failed to unpack tree object 07672f31880ba80300b38492df9d0acfcd6ee00a
>> 
>> Give the hint which submodule has a problem.
>> 
>> Signed-off-by: Stefan Beller <sbeller@google.com>
>> ---
>>  submodule.c               | 2 +-
>>  t/lib-submodule-update.sh | 3 ++-
>>  2 files changed, 3 insertions(+), 2 deletions(-)
>> 
>> diff --git a/submodule.c b/submodule.c
>> index 939d6870ecd..ebd092a14fd 100644
>> --- a/submodule.c
>> +++ b/submodule.c
>> @@ -1668,7 +1668,7 @@ int submodule_move_head(const char *path,
>>  	argv_array_push(&cp.args, new_head ? new_head : empty_tree_oid_hex());
>>  
>>  	if (run_command(&cp)) {
>> -		ret = -1;
>> +		ret = error(_("Submodule '%s' could not be updated."), path);
>
> This is a translated error message ...
>
>>  		goto out;
>>  	}
>>  
>> diff --git a/t/lib-submodule-update.sh b/t/lib-submodule-update.sh
>> index 1f38a85371a..e27f5d8541d 100755
>> --- a/t/lib-submodule-update.sh
>> +++ b/t/lib-submodule-update.sh
>> @@ -781,7 +781,8 @@ test_submodule_recursing_with_args_common() {
>>  		(
>>  			cd submodule_update &&
>>  			git branch -t invalid_sub1 origin/invalid_sub1 &&
>> -			test_must_fail $command invalid_sub1 &&
>> +			test_must_fail $command invalid_sub1 2>err &&
>> +			grep sub1 err &&
>
> ... so the test should use 'test_i18ngrep' to check it.

Thanks for being a careful reviewer, as always.

Will tweak locally to skip one round-trip.

Re: [PATCH] submodule.c: report the submodule that an error occurs in

[Stefan Beller]

On Mon, Jun 25, 2018 at 8:58 AM Junio C Hamano <gitster@pobox.com> wrote:
>
> SZEDER Gábor <szeder.dev@gmail.com> writes:
>
> >> When an error occurs in updating the working tree of a submodule in
> >> submodule_move_head, tell the user which submodule the error occurred in.
> >>
> >> The call to read-tree contains a super-prefix, such that the read-tree
> >> will correctly report any path related issues, but some error messages
> >> do not contain a path, for example:
> >>
> >>   ~/gerrit$ git checkout --recurse-submodules origin/master
> >>   ~/gerrit$ fatal: failed to unpack tree object 07672f31880ba80300b38492df9d0acfcd6ee00a
> >>
> >> Give the hint which submodule has a problem.
> >>
> >> Signed-off-by: Stefan Beller <sbeller@google.com>
> >> ---
> >>  submodule.c               | 2 +-
> >>  t/lib-submodule-update.sh | 3 ++-
> >>  2 files changed, 3 insertions(+), 2 deletions(-)
> >>
> >> diff --git a/submodule.c b/submodule.c
> >> index 939d6870ecd..ebd092a14fd 100644
> >> --- a/submodule.c
> >> +++ b/submodule.c
> >> @@ -1668,7 +1668,7 @@ int submodule_move_head(const char *path,
> >>      argv_array_push(&cp.args, new_head ? new_head : empty_tree_oid_hex());
> >>
> >>      if (run_command(&cp)) {
> >> -            ret = -1;
> >> +            ret = error(_("Submodule '%s' could not be updated."), path);
> >
> > This is a translated error message ...
> >
> >>              goto out;
> >>      }
> >>
> >> diff --git a/t/lib-submodule-update.sh b/t/lib-submodule-update.sh
> >> index 1f38a85371a..e27f5d8541d 100755
> >> --- a/t/lib-submodule-update.sh
> >> +++ b/t/lib-submodule-update.sh
> >> @@ -781,7 +781,8 @@ test_submodule_recursing_with_args_common() {
> >>              (
> >>                      cd submodule_update &&
> >>                      git branch -t invalid_sub1 origin/invalid_sub1 &&
> >> -                    test_must_fail $command invalid_sub1 &&
> >> +                    test_must_fail $command invalid_sub1 2>err &&
> >> +                    grep sub1 err &&
> >
> > ... so the test should use 'test_i18ngrep' to check it.
>
> Thanks for being a careful reviewer, as always.
>
> Will tweak locally to skip one round-trip.

Thanks for tweaking locally. I reviewed your queued patch and
it looks good to me. Thanks!

I remember deliberately not using the i18n grep as we grep for
a part that should stay the same in all languages. However I
forgot that details of the test suite implementation (it doesn't
have a real translation, so it doesn't matter that this particular
piece of the message will show up in all translations)

Sorry for my mishap in thinking there.

Thanks,
Stefan