patch, 9 partsmake stash apply with --index by default

>
> Since git-stash's inception, it has defaulted to stashing the index but not
> restoring it. This has caused some confusion: change the default as part of Git
> 3.0 to unstash the index, too.
>
> I ran into this myself a while back and did some digging; it appears that other
> have been bit, too (see references in patch 3). Moreover, when git-stash was
> originally written, defaulting with --index was suggested but not implemented
> before the script merged. So this RFC should bring us back towards "less
> confusing," hopefully.
>
> The series is structured as follows:
>
> Patches 1-2: unrelated prep/style commits noticed while working on tests.
> Patch 3: update Documentation/BreakingChanges.adoc. This seems like a natural
>     place to discuss the proposal, so it contains no other changes.
> Patch 4: make it so in builtin/stash.c.
> Patches 5-9: update the impacted tests. Separated out for ease of review. I used
>     a style more like "split the test into 2: one with, one without breaking
>     changes." In retrospect, the diff might be smaller (and the 2 versions of
>     the test easier to compare) if I used "test_has_prereq" in the tests… but at
>     the cost of making the tests harder to follow. Thoughts?
>
> D. Ben Knoble (9):
>   t3903: reduce dependencies on previous tests
>   t3905: remove unneeded blank line
>   BreakingChanges: announce stash {apply,pop} will imply --index
>   stash: restore the index by default when breaking changes are enabled
>   t0450: mark stash documentation as a known discrepancy
>   t3903: adjust stash test to account for --[no-]index with breaking
>     changes
>   t3904: adjust stash -p test to account for index states with breaking
>     changes
>   t3905: adjust stash -u tests for breaking changes
>   t3906: adjust stash submodule tests to account for breaking changes
>
>  Documentation/BreakingChanges.adoc |  11 ++
>  Documentation/git-stash.adoc       |   6 ++
>  builtin/stash.c                    |  38 +++++++
>  t/lib-submodule-update.sh          |  24 ++++-
>  t/t0450/adoc-help-mismatches       |   1 +
>  t/t3903-stash.sh                   | 161 +++++++++++++++++++++++++++--
>  t/t3904-stash-patch.sh             |  14 ++-
>  t/t3905-stash-include-untracked.sh |  40 ++++++-
>  8 files changed, 276 insertions(+), 19 deletions(-)
>
>
> base-commit: 1ee85f0e215f22b0878d0ad4b2445d12bbb63887
> --
> 2.48.1
>

>
> Since git-stash's inception, it has defaulted to stashing the index but not
> restoring it. This has caused some confusion: change the default as part of Git
> 3.0 to unstash the index, too.
>
> I ran into this myself a while back and did some digging; it appears that other
> have been bit, too (see references in patch 3). Moreover, when git-stash was
> originally written, defaulting with --index was suggested but not implemented
> before the script merged. So this RFC should bring us back towards "less
> confusing," hopefully.
>
> The series is structured as follows:
>
> Patches 1-2: unrelated prep/style commits noticed while working on tests.
> Patch 3: update Documentation/BreakingChanges.adoc. This seems like a natural
>     place to discuss the proposal, so it contains no other changes.
> Patch 4: make it so in builtin/stash.c.
> Patches 5-9: update the impacted tests. Separated out for ease of review. I used
>     a style more like "split the test into 2: one with, one without breaking
>     changes." In retrospect, the diff might be smaller (and the 2 versions of
>     the test easier to compare) if I used "test_has_prereq" in the tests… but at
>     the cost of making the tests harder to follow. Thoughts?
>
> D. Ben Knoble (9):
>   t3903: reduce dependencies on previous tests
>   t3905: remove unneeded blank line
>   BreakingChanges: announce stash {apply,pop} will imply --index
>   stash: restore the index by default when breaking changes are enabled
>   t0450: mark stash documentation as a known discrepancy
>   t3903: adjust stash test to account for --[no-]index with breaking
>     changes
>   t3904: adjust stash -p test to account for index states with breaking
>     changes
>   t3905: adjust stash -u tests for breaking changes
>   t3906: adjust stash submodule tests to account for breaking changes
>
>  Documentation/BreakingChanges.adoc |  11 ++
>  Documentation/git-stash.adoc       |   6 ++
>  builtin/stash.c                    |  38 +++++++
>  t/lib-submodule-update.sh          |  24 ++++-
>  t/t0450/adoc-help-mismatches       |   1 +
>  t/t3903-stash.sh                   | 161 +++++++++++++++++++++++++++--
>  t/t3904-stash-patch.sh             |  14 ++-
>  t/t3905-stash-include-untracked.sh |  40 ++++++-
>  8 files changed, 276 insertions(+), 19 deletions(-)
>
>
> base-commit: 1ee85f0e215f22b0878d0ad4b2445d12bbb63887
> --
> 2.48.1
>

> With stash.index=true, git-stash(1) command now tries to reinstate the
> index by default in the "apply" and "pop" modes. Not doing so creates a
> common trap [1], [2]: "git stash apply" is not the reverse of "git stash
> push" because carefully staged indices are lost and have to be manually
> recreated. OTOH, this mode is not always desirable and may create more
> conflicts when applying stashes. As usual, "--no-index" will disable
> this behavior if you set "stash.index".

> [1]: https://lore.kernel.org/git/CAPx1GvcxyDDQmCssMjEnt6JoV6qPc5ZUpgPLX3mpUC_4PNYA1w@mail.gmail.com/
> [2]: https://lore.kernel.org/git/c5a811ac-8cd3-c389-ac6d-29020a648c87@gmail.com/
> 
> Signed-off-by: D. Ben Knoble <ben.knoble+github@gmail.com>
> ---
>   Documentation/config/stash.adoc    |   5 +
>   builtin/stash.c                    |   9 +-
>   t/t3903-stash.sh                   | 155 +++++++++++++++++++++++++++++
>   t/t3904-stash-patch.sh             |  11 ++
>   t/t3905-stash-include-untracked.sh |  48 ++++++++-
>   5 files changed, 225 insertions(+), 3 deletions(-)
> 
> diff --git a/Documentation/config/stash.adoc b/Documentation/config/stash.adoc
> index ec1edaeba6..e556105a15 100644
> --- a/Documentation/config/stash.adoc
> +++ b/Documentation/config/stash.adoc
> @@ -1,3 +1,8 @@
> +stash.index::
> +	If this is set to true, `git stash apply` and `git stash pop` will
> +	behave as if `--index` was supplied. Defaults to false. See the
> +	descriptions in linkgit:git-stash[1].
> +
>   stash.showIncludeUntracked::
>   	If this is set to true, the `git stash show` command will show
>   	the untracked files of a stash entry.  Defaults to false. See
> diff --git a/builtin/stash.c b/builtin/stash.c
> index d9b478d1d1..8a0eef3c70 100644
> --- a/builtin/stash.c
> +++ b/builtin/stash.c
> @@ -130,6 +130,7 @@ static struct strbuf stash_index_path = STRBUF_INIT;
>   static int show_stat = 1;
>   static int show_patch;
>   static int show_include_untracked;
> +static int use_index;
>   
>   /*
>    * w_commit is set to the commit containing the working tree
> @@ -662,7 +663,7 @@ static int apply_stash(int argc, const char **argv, const char *prefix,
>   {
>   	int ret = -1;
>   	int quiet = 0;
> -	int index = 0;
> +	int index = use_index;
>   	struct stash_info info = STASH_INFO_INIT;
>   	struct option options[] = {
>   		OPT__QUIET(&quiet, N_("be quiet, only report errors")),
> @@ -759,7 +760,7 @@ static int pop_stash(int argc, const char **argv, const char *prefix,
>   		     struct repository *repo UNUSED)
>   {
>   	int ret = -1;
> -	int index = 0;
> +	int index = use_index;
>   	int quiet = 0;
>   	struct stash_info info = STASH_INFO_INIT;
>   	struct option options[] = {
> @@ -864,6 +865,10 @@ static int git_stash_config(const char *var, const char *value,
>   		show_include_untracked = git_config_bool(var, value);
>   		return 0;
>   	}
> +	if (!strcmp(var, "stash.index")) {
> +		use_index = git_config_bool(var, value);
> +		return 0;
> +	}
>   	return git_diff_basic_config(var, value, ctx, cb);
>   }
>   
> diff --git a/t/t3903-stash.sh b/t/t3903-stash.sh
> index b8936a653b..1d53a94165 100755
> --- a/t/t3903-stash.sh
> +++ b/t/t3903-stash.sh
> @@ -111,6 +111,19 @@ setup_stash()
>   	test 1 = $(git show HEAD:file)
>   '
>   
> +test_expect_success 'apply stashed changes with stash.index' '
> +	test_config stash.index true &&
> +	git reset --hard HEAD^ &&
> +	echo 5 >other-file &&
> +	git add other-file &&
> +	test_tick &&
> +	git commit -m other-file &&
> +	git stash apply &&
> +	test 3 = $(cat file) &&
> +	test 2 = $(git show :file) &&
> +	test 1 = $(git show HEAD:file)
> +'
> +
>   test_expect_success 'apply stashed changes (including index)' '
>   	git reset --hard HEAD^ &&
>   	echo 6 >other-file &&
> @@ -150,6 +163,21 @@ setup_stash()
>   	test 1 = $(git show HEAD:file)
>   '
>   
> +test_expect_success 'drop top stash with stash.index' '
> +	test_config stash.index true &&
> +	git reset --hard &&
> +	git stash list >expected &&
> +	echo 7 >file &&
> +	git stash &&
> +	git stash drop &&
> +	git stash list >actual &&
> +	test_cmp expected actual &&
> +	git stash apply &&
> +	test 3 = $(cat file) &&
> +	test 2 = $(git show :file) &&
> +	test 1 = $(git show HEAD:file)
> +'
> +
>   test_expect_success 'drop middle stash' '
>   	git reset --hard &&
>   	echo 8 >file &&
> @@ -170,6 +198,27 @@ setup_stash()
>   	test 1 = $(git show HEAD:file)
>   '
>   
> +test_expect_success 'drop middle stash with stash.index' '
> +	test_config stash.index true &&
> +	git reset --hard &&
> +	echo 8 >file &&
> +	git stash &&
> +	echo 9 >file &&
> +	git stash &&
> +	git stash drop stash@{1} &&
> +	test 2 = $(git stash list | wc -l) &&
> +	git stash apply &&
> +	test 9 = $(cat file) &&
> +	test 1 = $(git show :file) &&
> +	test 1 = $(git show HEAD:file) &&
> +	git reset --hard &&
> +	git stash drop &&
> +	git stash apply &&
> +	test 3 = $(cat file) &&
> +	test 2 = $(git show :file) &&
> +	test 1 = $(git show HEAD:file)
> +'
> +
>   test_expect_success 'drop middle stash by index' '
>   	git reset --hard &&
>   	echo 8 >file &&
> @@ -236,6 +285,17 @@ setup_stash()
>   	test 0 = $(git stash list | wc -l)
>   '
>   
> +test_expect_success 'stash pop with stash.index' '
> +	test_config stash.index true &&
> +	git reset --hard &&
> +	setup_stash &&
> +	git stash pop &&
> +	test 3 = $(cat file) &&
> +	test 2 = $(git show :file) &&
> +	test 1 = $(git show HEAD:file) &&
> +	test 0 = $(git stash list | wc -l)
> +'
> +
>   cat >expect <<EOF
>   diff --git a/file2 b/file2
>   new file mode 100644
> @@ -328,6 +388,22 @@ setup_stash()
>   	test_must_be_empty output.out
>   '
>   
> +test_expect_success 'pop -q works and is quiet with stash.index' '
> +	# Added file, deleted file, modified file all staged for commit
> +	echo foo >new-file &&
> +	echo test >file &&
> +	git add new-file file &&
> +	git rm other-file &&
> +	git stash &&
> +
> +	test_config stash.index true &&
> +	git stash pop -q >output.out 2>&1 &&
> +	echo test >expect &&
> +	git show :file >actual &&
> +	test_cmp expect actual &&
> +	test_must_be_empty output.out
> +'
> +
>   test_expect_success 'pop -q --index works and is quiet' '
>   	echo foo >file &&
>   	git add file &&
> @@ -1178,6 +1254,19 @@ setup_stash()
>   	test_path_is_file bar
>   '
>   
> +test_expect_success 'stash -- <pathspec> stashes and restores the file with stash.index' '
> +	test_config stash.index true &&
> +	>foo &&
> +	>bar &&
> +	git add foo bar &&
> +	git stash push -- foo &&
> +	test_path_is_file bar &&
> +	test_path_is_missing foo &&
> +	git stash pop --no-index &&
> +	test_path_is_file foo &&
> +	test_path_is_file bar
> +'
> +
>   test_expect_success 'stash -- <pathspec> stashes in subdirectory' '
>   	mkdir sub &&
>   	>foo &&
> @@ -1194,6 +1283,24 @@ setup_stash()
>   	test_path_is_file bar
>   '
>   
> +test_expect_success 'stash -- <pathspec> stashes in subdirectory with stash.index' '
> +	test_config stash.index true &&
> +	rm -r sub &&
> +	mkdir sub &&
> +	>foo &&
> +	>bar &&
> +	git add foo bar &&
> +	(
> +		cd sub &&
> +		git stash push -- ../foo
> +	) &&
> +	test_path_is_file bar &&
> +	test_path_is_missing foo &&
> +	git stash pop --no-index &&
> +	test_path_is_file foo &&
> +	test_path_is_file bar
> +'
> +
>   test_expect_success 'stash with multiple pathspec arguments' '
>   	>foo &&
>   	>bar &&
> @@ -1209,6 +1316,22 @@ setup_stash()
>   	test_path_is_file extra
>   '
>   
> +test_expect_success 'stash with multiple pathspec arguments with stash.index' '
> +	test_config stash.index true &&
> +	>foo &&
> +	>bar &&
> +	>extra &&
> +	git add foo bar extra &&
> +	git stash push -- foo bar &&
> +	test_path_is_missing bar &&
> +	test_path_is_missing foo &&
> +	test_path_is_file extra &&
> +	git stash pop --no-index &&
> +	test_path_is_file foo &&
> +	test_path_is_file bar &&
> +	test_path_is_file extra
> +'
> +
>   test_expect_success 'stash with file including $IFS character' '
>   	>"foo bar" &&
>   	>foo &&
> @@ -1224,6 +1347,22 @@ setup_stash()
>   	test_path_is_file bar
>   '
>   
> +test_expect_success 'stash with file including $IFS character with stash.index' '
> +	test_config stash.index true &&
> +	>"foo bar" &&
> +	>foo &&
> +	>bar &&
> +	git add foo* &&
> +	git stash push -- "foo b*" &&
> +	test_path_is_missing "foo bar" &&
> +	test_path_is_file foo &&
> +	test_path_is_file bar &&
> +	git stash pop --no-index &&
> +	test_path_is_file "foo bar" &&
> +	test_path_is_file foo &&
> +	test_path_is_file bar
> +'
> +
>   test_expect_success 'stash with pathspec matching multiple paths' '
>   	echo original >file &&
>   	echo original >other-file &&
> @@ -1312,6 +1451,22 @@ setup_stash()
>   	test_path_is_file bar
>   '
>   
> +test_expect_success 'stash without verb with pathspec with stash.index' '
> +	test_config stash.index true &&
> +	>"foo bar" &&
> +	>foo &&
> +	>bar &&
> +	git add foo* &&
> +	git stash -- "foo b*" &&
> +	test_path_is_missing "foo bar" &&
> +	test_path_is_file foo &&
> +	test_path_is_file bar &&
> +	git stash pop --no-index &&
> +	test_path_is_file "foo bar" &&
> +	test_path_is_file foo &&
> +	test_path_is_file bar
> +'
> +
>   test_expect_success 'stash -k -- <pathspec> leaves unstaged files intact' '
>   	git reset &&
>   	>foo &&
> diff --git a/t/t3904-stash-patch.sh b/t/t3904-stash-patch.sh
> index ae313e3c70..fe402f6ab5 100755
> --- a/t/t3904-stash-patch.sh
> +++ b/t/t3904-stash-patch.sh
> @@ -42,6 +42,17 @@
>   	verify_state dir/foo work head
>   '
>   
> +test_expect_success 'git stash -p with stash.index' '
> +	test_config stash.index true &&
> +	set_state HEAD HEADfile_work HEADfile_index &&
> +	set_state dir/foo work index &&
> +	test_write_lines y n y | git stash save -p &&
> +	git reset --hard &&
> +	git stash apply &&
> +	verify_state HEAD HEADfile_work HEADfile_index &&
> +	verify_state dir/foo head index
> +'
> +
>   test_expect_success 'git stash -p --no-keep-index' '
>   	set_state HEAD HEADfile_work HEADfile_index &&
>   	set_state bar bar_work bar_index &&
> diff --git a/t/t3905-stash-include-untracked.sh b/t/t3905-stash-include-untracked.sh
> index 7704709054..5407f11030 100755
> --- a/t/t3905-stash-include-untracked.sh
> +++ b/t/t3905-stash-include-untracked.sh
> @@ -7,7 +7,7 @@
>   
>   . ./test-lib.sh
>   
> -test_expect_success 'stash save --include-untracked some dirty working directory' '
> +setup() {
>   	echo 1 >file &&
>   	git add file &&
>   	test_tick &&
> @@ -23,6 +23,10 @@
>   	git stash --include-untracked &&
>   	git diff-files --quiet &&
>   	git diff-index --cached --quiet HEAD
> +}
> +
> +test_expect_success 'stash save --include-untracked some dirty working directory' '
> +	setup
>   '
>   
>   test_expect_success 'stash save --include-untracked cleaned the untracked files' '
> @@ -108,6 +112,32 @@
>   	test_cmp untracked_expect untracked/untracked
>   '
>   
> +test_expect_success 'stash pop after save --include-untracked leaves files untracked again with stash.index' '
> +	git init repo &&
> +	test_when_finished rm -r repo &&
> +	(
> +		cd repo &&
> +		git config stash.index true &&
> +		setup &&
> +		cat >expect <<-EOF &&
> +		MM file
> +		?? HEAD
> +		?? actual
> +		?? expect
> +		?? file2
> +		?? untracked/
> +		EOF
> +
> +		git stash pop &&
> +		git status --porcelain >actual &&
> +		test_cmp expect actual &&
> +		echo 1 >expect_file2 &&
> +		test_cmp expect_file2 file2 &&
> +		echo untracked >untracked_expect &&
> +		test_cmp untracked_expect untracked/untracked
> +	)
> +'
> +
>   test_expect_success 'clean up untracked/ directory to prepare for next tests' '
>   	git clean --force --quiet -d
>   '
> @@ -221,6 +251,22 @@
>   	test_path_is_file bar
>   '
>   
> +test_expect_success 'stash push with $IFS character with stash.index' '
> +	test_config stash.index true &&
> +	>"foo bar" &&
> +	>foo &&
> +	>bar &&
> +	git add foo* &&
> +	git stash push --include-untracked -- "foo b*" &&
> +	test_path_is_missing "foo bar" &&
> +	test_path_is_file foo &&
> +	test_path_is_file bar &&
> +	git stash pop --no-index &&
> +	test_path_is_file "foo bar" &&
> +	test_path_is_file foo &&
> +	test_path_is_file bar
> +'
> +
>   test_expect_success 'stash previously ignored file' '
>   	cat >.gitignore <<-EOF &&
>   	ignored

>
> Hi Ben
>
> On 16/09/2025 01:37, D. Ben Knoble wrote:
> > With stash.index=true, git-stash(1) command now tries to reinstate the
> > index by default in the "apply" and "pop" modes. Not doing so creates a
> > common trap [1], [2]: "git stash apply" is not the reverse of "git stash
> > push" because carefully staged indices are lost and have to be manually
> > recreated. OTOH, this mode is not always desirable and may create more
> > conflicts when applying stashes. As usual, "--no-index" will disable
> > this behavior if you set "stash.index".
>
> I don't have a strong opinion either way on the new config setting but I
> do think we should rationalize the new tests. Assuming we already have
> good coverage for "git stash pop --index" then all we need to do is
> check that "git -c stash.index=true stash pop", "git -c stash.index=true
> stash pop --no-index" and "git -c stash.index=false stash pop --index".
> We don't need an exhaustive list of tests that check the config setting
> in scenarios like "create twos stashes, drop the second one and apply
> the first". Tests like that add no new coverage for the changes in this
> patch and slow the test suite down.

> 
> PS I've left some new t3903 tests as copy-pasta for now to get feedback
> on the rest of the series; there are bits of that file that could use an
> update to the modern style (e.g., not using "test 1 = $(cat file)").
> Since some new tests are substantially similar to old tests that use
> this style, such cleanup is /probably/ warranted but will delay eyeballs
> on the core of this series.

> Published-as: https://github.com/benknoble/git/tree/stash-apply-index
> v1: https://lore.kernel.org/git/20250510183358.36806-1-ben.knoble+github@gmail.com/
> 
> D. Ben Knoble (4):
>    t3903: reduce dependencies on previous tests
>    t3905: remove unneeded blank line
>    stash: refactor private config globals
>    stash: honor stash.index in apply, pop modes
> 
>   Documentation/config/stash.adoc    |   5 +
>   builtin/stash.c                    |  17 ++--
>   t/t3903-stash.sh                   | 158 +++++++++++++++++++++++++++++
>   t/t3904-stash-patch.sh             |  11 ++
>   t/t3905-stash-include-untracked.sh |  49 ++++++++-
>   5 files changed, 232 insertions(+), 8 deletions(-)
> 
> Diff-intervalle contre v1 :
>   1:  1328eb8eac =  1:  1328eb8eac t3903: reduce dependencies on previous tests
>   2:  8ac06ad62d =  2:  8ac06ad62d t3905: remove unneeded blank line
>   3:  c068f1dc0b <  -:  ---------- BreakingChanges: announce stash {apply,pop} will imply --index
>   4:  8caff91c0e <  -:  ---------- stash: restore the index by default when breaking changes are enabled
>   5:  387427bb8c <  -:  ---------- t0450: mark stash documentation as a known discrepancy
>   9:  c72a1fe6ea !  3:  bf0a561ce3 t3906: adjust stash submodule tests to account for breaking changes
>      @@ Metadata
>       Author: D. Ben Knoble <ben.knoble+github@gmail.com>
>       
>        ## Commit message ##
>      -    t3906: adjust stash submodule tests to account for breaking changes
>      +    stash: refactor private config globals
>       
>      -    I cannot explain _why_ this occurs, but it seems that automatically
>      -    unstashing the index from previous commits resolves some known failures
>      -    in t3906 (which are captured by t/lib-submodule-updates.sh).
>      +    A subsequent commit will access a new config variable in the stash
>      +    subcommand implementations, which requires the variables to be declared
>      +    before the relevant functions. Prep with a pure refactoring change to
>      +    consolidate config-related globals with the rest of the globals.
>       
>      -    In particular:
>      -    - 'replace tracked file with submodule creates empty directory' succeeds
>      -      with breaking changes;
>      -    - all KNOWN_FAILURE_NOFF_MERGE_DOESNT_CREATE_EMPTY_SUBMODULE_DIR tests
>      -      succeed with breaking changes;
>      -    - all KNOWN_FAILURE_STASH_DOES_IGNORE_SUBMODULE_CHANGES suceed with
>      -      breaking changes.
>      +    Best-viewed-with: --color-moved
>       
>      - ## t/lib-submodule-update.sh ##
>      -@@ t/lib-submodule-update.sh: test_submodule_switch_common ()
>      - 	'
>      - 	# Replacing a tracked file with a submodule produces an empty
>      - 	# directory ...
>      --	test_expect_$RESULT "$command: replace tracked file with submodule creates empty directory" '
>      -+	test_expect_$RESULT !WITH_BREAKING_CHANGES "$command: replace tracked file with submodule creates empty directory" '
>      -+		prolog &&
>      -+		reset_work_tree_to replace_sub1_with_file &&
>      -+		(
>      -+			cd submodule_update &&
>      -+			git branch -t replace_file_with_sub1 origin/replace_file_with_sub1 &&
>      -+			$command replace_file_with_sub1 &&
>      -+			test_superproject_content origin/replace_file_with_sub1 &&
>      -+			test_dir_is_empty sub1 &&
>      -+			git submodule update --init --recursive &&
>      -+			test_submodule_content sub1 origin/replace_file_with_sub1
>      -+		)
>      -+	'
>      -+	# (unless we automatically unstash the index!)
>      -+	test_expect_success WITH_BREAKING_CHANGES "$command: replace tracked file with submodule creates empty directory" '
>      - 		prolog &&
>      - 		reset_work_tree_to replace_sub1_with_file &&
>      - 		(
>      -@@ t/lib-submodule-update.sh: test_submodule_switch_common ()
>      - 	'
>      - 	# ... as does removing a directory with tracked files with a
>      - 	# submodule.
>      --	if test "$KNOWN_FAILURE_NOFF_MERGE_DOESNT_CREATE_EMPTY_SUBMODULE_DIR" = 1
>      -+	if ! test_have_prereq WITH_BREAKING_CHANGES &&
>      -+		test "$KNOWN_FAILURE_NOFF_MERGE_DOESNT_CREATE_EMPTY_SUBMODULE_DIR" = 1
>      - 	then
>      - 		# Non fast-forward merges fail with "Directory sub1 doesn't
>      - 		# exist. sub1" because the empty submodule directory is not
>      -@@ t/lib-submodule-update.sh: test_submodule_switch_common ()
>      - 	'
>      + ## builtin/stash.c ##
>      +@@ builtin/stash.c: static const char * const git_stash_save_usage[] = {
>      + static const char ref_stash[] = "refs/stash";
>      + static struct strbuf stash_index_path = STRBUF_INIT;
>        
>      - 	######################## Disappearing submodule #######################
>      --	# Removing a submodule doesn't remove its work tree ...
>      --	if test "$KNOWN_FAILURE_STASH_DOES_IGNORE_SUBMODULE_CHANGES" = 1
>      -+	# Removing a submodule doesn't remove its work tree (unless stash applies the index!) ...
>      -+	if ! test_have_prereq WITH_BREAKING_CHANGES &&
>      -+		test "$KNOWN_FAILURE_STASH_DOES_IGNORE_SUBMODULE_CHANGES" = 1
>      - 	then
>      - 		RESULT="failure"
>      - 	else
>      ++static int show_stat = 1;
>      ++static int show_patch;
>      ++static int show_include_untracked;
>      ++
>      + /*
>      +  * w_commit is set to the commit containing the working tree
>      +  * b_commit is set to the base commit
>      +@@ builtin/stash.c: static int list_stash(int argc, const char **argv, const char *prefix,
>      + 	return run_command(&cp);
>      + }
>      +
>      +-static int show_stat = 1;
>      +-static int show_patch;
>      +-static int show_include_untracked;
>      +-
>      + static int git_stash_config(const char *var, const char *value,
>      + 			    const struct config_context *ctx, void *cb)
>      + {
>   6:  0a12983c00 !  4:  585e124467 t3903: adjust stash test to account for --[no-]index with breaking changes
>      @@ Metadata
>       Author: D. Ben Knoble <ben.knoble+github@gmail.com>
>       
>        ## Commit message ##
>      -    t3903: adjust stash test to account for --[no-]index with breaking changes
>      +    stash: honor stash.index in apply, pop modes
>       
>      -    A few tests check the results of the index after applying a stash; with
>      -    breaking changes from previous commits that automatically restore the
>      -    stashed index, the expected values are wrong.
>      +    With stash.index=true, git-stash(1) command now tries to reinstate the
>      +    index by default in the "apply" and "pop" modes. Not doing so creates a
>      +    common trap [1], [2]: "git stash apply" is not the reverse of "git stash
>      +    push" because carefully staged indices are lost and have to be manually
>      +    recreated. OTOH, this mode is not always desirable and may create more
>      +    conflicts when applying stashes. As usual, "--no-index" will disable
>      +    this behavior if you set "stash.index".
>       
>      -    A few of the relevant tests check the restoration of <pathspec>s; with
>      -    the aforementioned breaking changes, things get more interesting. In
>      -    particular, if we "git stash push -- foo" but have "bar" in the index,
>      -    then when applying the stash we get a conflict: "bar" was not removed
>      -    from the index by the stash, but it was included in the recorded index
>      -    in the stash. In those cases, apply the stash with "--no-index" (which
>      -    would be the required user behavior).
>      +    [1]: https://lore.kernel.org/git/CAPx1GvcxyDDQmCssMjEnt6JoV6qPc5ZUpgPLX3mpUC_4PNYA1w@mail.gmail.com/
>      +    [2]: https://lore.kernel.org/git/c5a811ac-8cd3-c389-ac6d-29020a648c87@gmail.com/
>       
>      -    Signed-off-by: D. Ben Knoble <ben.knoble+github@gmail.com>
>      + ## Documentation/config/stash.adoc ##
>      +@@
>      ++stash.index::
>      ++	If this is set to true, `git stash apply` and `git stash pop` will
>      ++	behave as if `--index` was supplied. Defaults to false. See the
>      ++	descriptions in linkgit:git-stash[1].
>      ++
>      + stash.showIncludeUntracked::
>      + 	If this is set to true, the `git stash show` command will show
>      + 	the untracked files of a stash entry.  Defaults to false. See
>       
>      -
>      - ## Notes ##
>      -    It looks like the pathspec filtering is not applied to the stashed
>      -    index; should it be?
>      + ## builtin/stash.c ##
>      +@@ builtin/stash.c: static struct strbuf stash_index_path = STRBUF_INIT;
>      + static int show_stat = 1;
>      + static int show_patch;
>      + static int show_include_untracked;
>      ++static int use_index;
>      +
>      + /*
>      +  * w_commit is set to the commit containing the working tree
>      +@@ builtin/stash.c: static int apply_stash(int argc, const char **argv, const char *prefix,
>      + {
>      + 	int ret = -1;
>      + 	int quiet = 0;
>      +-	int index = 0;
>      ++	int index = use_index;
>      + 	struct stash_info info = STASH_INFO_INIT;
>      + 	struct option options[] = {
>      + 		OPT__QUIET(&quiet, N_("be quiet, only report errors")),
>      +@@ builtin/stash.c: static int pop_stash(int argc, const char **argv, const char *prefix,
>      + 		     struct repository *repo UNUSED)
>      + {
>      + 	int ret = -1;
>      +-	int index = 0;
>      ++	int index = use_index;
>      + 	int quiet = 0;
>      + 	struct stash_info info = STASH_INFO_INIT;
>      + 	struct option options[] = {
>      +@@ builtin/stash.c: static int git_stash_config(const char *var, const char *value,
>      + 		show_include_untracked = git_config_bool(var, value);
>      + 		return 0;
>      + 	}
>      ++	if (!strcmp(var, "stash.index")) {
>      ++		use_index = git_config_bool(var, value);
>      ++		return 0;
>      ++	}
>      + 	return git_diff_basic_config(var, value, ctx, cb);
>      + }
>      +
>       
>        ## t/t3903-stash.sh ##
>      -@@ t/t3903-stash.sh: setup_stash()
>      - 	test_cmp expect file
>      - '
>      -
>      --test_expect_success 'apply stashed changes' '
>      -+test_expect_success !WITH_BREAKING_CHANGES 'apply stashed changes' '
>      - 	git reset --hard &&
>      - 	echo 5 >other-file &&
>      - 	git add other-file &&
>       @@ t/t3903-stash.sh: setup_stash()
>        	test 1 = $(git show HEAD:file)
>        '
>        
>      -+test_expect_success WITH_BREAKING_CHANGES 'apply stashed changes' '
>      -+	git reset --hard &&
>      ++test_expect_success 'apply stashed changes with stash.index' '
>      ++	test_config stash.index true &&
>      ++	git reset --hard HEAD^ &&
>       +	echo 5 >other-file &&
>       +	git add other-file &&
>       +	test_tick &&
>      @@ t/t3903-stash.sh: setup_stash()
>        test_expect_success 'apply stashed changes (including index)' '
>        	git reset --hard HEAD^ &&
>        	echo 6 >other-file &&
>      -@@ t/t3903-stash.sh: setup_stash()
>      - 	test_must_fail git stash drop --foo
>      - '
>      -
>      --test_expect_success 'drop top stash' '
>      -+test_expect_success !WITH_BREAKING_CHANGES 'drop top stash' '
>      - 	git reset --hard &&
>      - 	git stash list >expected &&
>      - 	echo 7 >file &&
>       @@ t/t3903-stash.sh: setup_stash()
>        	test 1 = $(git show HEAD:file)
>        '
>        
>      --test_expect_success 'drop middle stash' '
>      -+test_expect_success WITH_BREAKING_CHANGES 'drop top stash' '
>      ++test_expect_success 'drop top stash with stash.index' '
>      ++	test_config stash.index true &&
>       +	git reset --hard &&
>       +	git stash list >expected &&
>       +	echo 7 >file &&
>      @@ t/t3903-stash.sh: setup_stash()
>       +	test 1 = $(git show HEAD:file)
>       +'
>       +
>      -+test_expect_success !WITH_BREAKING_CHANGES 'drop middle stash' '
>      + test_expect_success 'drop middle stash' '
>        	git reset --hard &&
>        	echo 8 >file &&
>      - 	git stash &&
>       @@ t/t3903-stash.sh: setup_stash()
>        	test 1 = $(git show HEAD:file)
>        '
>        
>      --test_expect_success 'drop middle stash by index' '
>      -+test_expect_success WITH_BREAKING_CHANGES 'drop middle stash' '
>      ++test_expect_success 'drop middle stash with stash.index' '
>      ++	test_config stash.index true &&
>       +	git reset --hard &&
>       +	echo 8 >file &&
>       +	git stash &&
>      @@ t/t3903-stash.sh: setup_stash()
>       +	test 1 = $(git show HEAD:file)
>       +'
>       +
>      -+test_expect_success !WITH_BREAKING_CHANGES 'drop middle stash by index' '
>      + test_expect_success 'drop middle stash by index' '
>        	git reset --hard &&
>        	echo 8 >file &&
>      - 	git stash &&
>      -@@ t/t3903-stash.sh: setup_stash()
>      - 	test_cmp expect actual
>      - '
>      -
>      --test_expect_success 'stash pop' '
>      -+test_expect_success !WITH_BREAKING_CHANGES 'stash pop' '
>      - 	git reset --hard &&
>      - 	git stash pop &&
>      - 	test 3 = $(cat file) &&
>       @@ t/t3903-stash.sh: setup_stash()
>        	test 0 = $(git stash list | wc -l)
>        '
>        
>      -+test_expect_success WITH_BREAKING_CHANGES 'stash pop' '
>      ++test_expect_success 'stash pop with stash.index' '
>      ++	test_config stash.index true &&
>       +	git reset --hard &&
>      ++	setup_stash &&
>       +	git stash pop &&
>       +	test 3 = $(cat file) &&
>       +	test 2 = $(git show :file) &&
>      @@ t/t3903-stash.sh: setup_stash()
>        	test_must_be_empty output.out
>        '
>        
>      --test_expect_success 'pop -q works and is quiet' '
>      -+test_expect_success !WITH_BREAKING_CHANGES 'pop -q works and is quiet' '
>      - 	git stash pop -q >output.out 2>&1 &&
>      - 	echo bar >expect &&
>      - 	git show :file >actual &&
>      -@@ t/t3903-stash.sh: setup_stash()
>      - 	test_must_be_empty output.out
>      - '
>      -
>      -+test_expect_success WITH_BREAKING_CHANGES 'pop -q works and is quiet' '
>      ++test_expect_success 'pop -q works and is quiet with stash.index' '
>      ++	# Added file, deleted file, modified file all staged for commit
>      ++	echo foo >new-file &&
>      ++	echo test >file &&
>      ++	git add new-file file &&
>      ++	git rm other-file &&
>      ++	git stash &&
>      ++
>      ++	test_config stash.index true &&
>       +	git stash pop -q >output.out 2>&1 &&
>       +	echo test >expect &&
>       +	git show :file >actual &&
>      @@ t/t3903-stash.sh: setup_stash()
>        test_expect_success 'pop -q --index works and is quiet' '
>        	echo foo >file &&
>        	git add file &&
>      -@@ t/t3903-stash.sh: setup_stash()
>      - 	test_cmp expect actual
>      - '
>      -
>      --test_expect_success 'stash -- <pathspec> stashes and restores the file' '
>      -+test_expect_success !WITH_BREAKING_CHANGES 'stash -- <pathspec> stashes and restores the file' '
>      - 	>foo &&
>      - 	>bar &&
>      - 	git add foo bar &&
>       @@ t/t3903-stash.sh: setup_stash()
>        	test_path_is_file bar
>        '
>        
>      --test_expect_success 'stash -- <pathspec> stashes in subdirectory' '
>      -+test_expect_success WITH_BREAKING_CHANGES 'stash -- <pathspec> stashes and restores the file' '
>      ++test_expect_success 'stash -- <pathspec> stashes and restores the file with stash.index' '
>      ++	test_config stash.index true &&
>       +	>foo &&
>       +	>bar &&
>       +	git add foo bar &&
>      @@ t/t3903-stash.sh: setup_stash()
>       +	test_path_is_file bar
>       +'
>       +
>      -+test_expect_success !WITH_BREAKING_CHANGES 'stash -- <pathspec> stashes in subdirectory' '
>      + test_expect_success 'stash -- <pathspec> stashes in subdirectory' '
>        	mkdir sub &&
>        	>foo &&
>      - 	>bar &&
>       @@ t/t3903-stash.sh: setup_stash()
>        	test_path_is_file bar
>        '
>        
>      --test_expect_success 'stash with multiple pathspec arguments' '
>      -+test_expect_success WITH_BREAKING_CHANGES 'stash -- <pathspec> stashes in subdirectory' '
>      ++test_expect_success 'stash -- <pathspec> stashes in subdirectory with stash.index' '
>      ++	test_config stash.index true &&
>      ++	rm -r sub &&
>       +	mkdir sub &&
>       +	>foo &&
>       +	>bar &&
>      @@ t/t3903-stash.sh: setup_stash()
>       +	test_path_is_file bar
>       +'
>       +
>      -+test_expect_success !WITH_BREAKING_CHANGES 'stash with multiple pathspec arguments' '
>      + test_expect_success 'stash with multiple pathspec arguments' '
>        	>foo &&
>        	>bar &&
>      - 	>extra &&
>       @@ t/t3903-stash.sh: setup_stash()
>        	test_path_is_file extra
>        '
>        
>      --test_expect_success 'stash with file including $IFS character' '
>      -+test_expect_success WITH_BREAKING_CHANGES 'stash with multiple pathspec arguments' '
>      ++test_expect_success 'stash with multiple pathspec arguments with stash.index' '
>      ++	test_config stash.index true &&
>       +	>foo &&
>       +	>bar &&
>       +	>extra &&
>      @@ t/t3903-stash.sh: setup_stash()
>       +	test_path_is_file extra
>       +'
>       +
>      -+test_expect_success !WITH_BREAKING_CHANGES 'stash with file including $IFS character' '
>      + test_expect_success 'stash with file including $IFS character' '
>        	>"foo bar" &&
>        	>foo &&
>      - 	>bar &&
>       @@ t/t3903-stash.sh: setup_stash()
>        	test_path_is_file bar
>        '
>        
>      -+test_expect_success WITH_BREAKING_CHANGES 'stash with file including $IFS character' '
>      ++test_expect_success 'stash with file including $IFS character with stash.index' '
>      ++	test_config stash.index true &&
>       +	>"foo bar" &&
>       +	>foo &&
>       +	>bar &&
>      @@ t/t3903-stash.sh: setup_stash()
>        test_expect_success 'stash with pathspec matching multiple paths' '
>        	echo original >file &&
>        	echo original >other-file &&
>      -@@ t/t3903-stash.sh: setup_stash()
>      - 	test_path_is_file untracked
>      - '
>      -
>      --test_expect_success 'stash without verb with pathspec' '
>      -+test_expect_success !WITH_BREAKING_CHANGES 'stash without verb with pathspec' '
>      - 	>"foo bar" &&
>      - 	>foo &&
>      - 	>bar &&
>       @@ t/t3903-stash.sh: setup_stash()
>        	test_path_is_file bar
>        '
>        
>      -+test_expect_success WITH_BREAKING_CHANGES 'stash without verb with pathspec' '
>      ++test_expect_success 'stash without verb with pathspec with stash.index' '
>      ++	test_config stash.index true &&
>       +	>"foo bar" &&
>       +	>foo &&
>       +	>bar &&
>      @@ t/t3903-stash.sh: setup_stash()
>        test_expect_success 'stash -k -- <pathspec> leaves unstaged files intact' '
>        	git reset &&
>        	>foo &&
>      +
>      + ## t/t3904-stash-patch.sh ##
>      +@@
>      + 	verify_state dir/foo work head
>      + '
>      +
>      ++test_expect_success 'git stash -p with stash.index' '
>      ++	test_config stash.index true &&
>      ++	set_state HEAD HEADfile_work HEADfile_index &&
>      ++	set_state dir/foo work index &&
>      ++	test_write_lines y n y | git stash save -p &&
>      ++	git reset --hard &&
>      ++	git stash apply &&
>      ++	verify_state HEAD HEADfile_work HEADfile_index &&
>      ++	verify_state dir/foo head index
>      ++'
>      ++
>      + test_expect_success 'git stash -p --no-keep-index' '
>      + 	set_state HEAD HEADfile_work HEADfile_index &&
>      + 	set_state bar bar_work bar_index &&
>      +
>      + ## t/t3905-stash-include-untracked.sh ##
>      +@@
>      +
>      + . ./test-lib.sh
>      +
>      +-test_expect_success 'stash save --include-untracked some dirty working directory' '
>      ++setup() {
>      + 	echo 1 >file &&
>      + 	git add file &&
>      + 	test_tick &&
>      +@@
>      + 	git stash --include-untracked &&
>      + 	git diff-files --quiet &&
>      + 	git diff-index --cached --quiet HEAD
>      ++}
>      ++
>      ++test_expect_success 'stash save --include-untracked some dirty working directory' '
>      ++	setup
>      + '
>      +
>      + test_expect_success 'stash save --include-untracked cleaned the untracked files' '
>      +@@
>      + 	test_cmp untracked_expect untracked/untracked
>      + '
>      +
>      ++test_expect_success 'stash pop after save --include-untracked leaves files untracked again with stash.index' '
>      ++	git init repo &&
>      ++	test_when_finished rm -r repo &&
>      ++	(
>      ++		cd repo &&
>      ++		git config stash.index true &&
>      ++		setup &&
>      ++		cat >expect <<-EOF &&
>      ++		MM file
>      ++		?? HEAD
>      ++		?? actual
>      ++		?? expect
>      ++		?? file2
>      ++		?? untracked/
>      ++		EOF
>      ++
>      ++		git stash pop &&
>      ++		git status --porcelain >actual &&
>      ++		test_cmp expect actual &&
>      ++		echo 1 >expect_file2 &&
>      ++		test_cmp expect_file2 file2 &&
>      ++		echo untracked >untracked_expect &&
>      ++		test_cmp untracked_expect untracked/untracked
>      ++	)
>      ++'
>      ++
>      + test_expect_success 'clean up untracked/ directory to prepare for next tests' '
>      + 	git clean --force --quiet -d
>      + '
>      +@@
>      + 	test_path_is_file bar
>      + '
>      +
>      ++test_expect_success 'stash push with $IFS character with stash.index' '
>      ++	test_config stash.index true &&
>      ++	>"foo bar" &&
>      ++	>foo &&
>      ++	>bar &&
>      ++	git add foo* &&
>      ++	git stash push --include-untracked -- "foo b*" &&
>      ++	test_path_is_missing "foo bar" &&
>      ++	test_path_is_file foo &&
>      ++	test_path_is_file bar &&
>      ++	git stash pop --no-index &&
>      ++	test_path_is_file "foo bar" &&
>      ++	test_path_is_file foo &&
>      ++	test_path_is_file bar
>      ++'
>      ++
>      + test_expect_success 'stash previously ignored file' '
>      + 	cat >.gitignore <<-EOF &&
>      + 	ignored
>   7:  4b92d47e16 <  -:  ---------- t3904: adjust stash -p test to account for index states with breaking changes
>   8:  af14dee1be <  -:  ---------- t3905: adjust stash -u tests for breaking changes
> 
> base-commit: 1ee85f0e215f22b0878d0ad4b2445d12bbb63887

>
> Hi Ben
>
> On 16/09/2025 01:37, D. Ben Knoble wrote:
> >
> > PS I've left some new t3903 tests as copy-pasta for now to get feedback
> > on the rest of the series; there are bits of that file that could use an
> > update to the modern style (e.g., not using "test 1 = $(cat file)").
> > Since some new tests are substantially similar to old tests that use
> > this style, such cleanup is /probably/ warranted but will delay eyeballs
> > on the core of this series.
>
> In situations like this even if we don't convert the old tests, I think
> it is worth using a modern style for the new ones. Some people argue
> that the style within a file should be consistent but in practice that
> means if someone decides to clean them up in the future they have even
> more work to do and in the meantime we have a harder time debugging test
> failures.

> With stash.index=true, git-stash(1) command now tries to reinstate the
> index by default in the "apply" and "pop" modes. Not doing so creates a
> common trap: "git stash apply" is not the reverse of "git stash push"
> because carefully staged indices are lost and have to be manually
> recreated. OTOH, this mode is not always desirable and may create more
> conflicts when applying stashes. Use "--no-index" to disable this behavior.

> With stash.index=true, git-stash(1) command now tries to reinstate the
> index by default in the "apply" and "pop" modes. Not doing so creates a
> common trap [1], [2]: "git stash apply" is not the reverse of "git stash
> push" because carefully staged indices are lost and have to be manually
> recreated. OTOH, this mode is not always desirable and may create more
> conflicts when applying stashes. As usual, "--no-index" will disable
> this behavior if you set "stash.index".

> [1]: https://lore.kernel.org/git/CAPx1GvcxyDDQmCssMjEnt6JoV6qPc5ZUpgPLX3mpUC_4PNYA1w@mail.gmail.com/
> [2]: https://lore.kernel.org/git/c5a811ac-8cd3-c389-ac6d-29020a648c87@gmail.com/
> 
> Signed-off-by: D. Ben Knoble <ben.knoble+github@gmail.com>
> ---
>   Documentation/config/stash.adoc |  5 +++++
>   builtin/stash.c                 |  9 ++++++--
>   t/t3903-stash.sh                | 37 +++++++++++++++++++++++++++++++++
>   3 files changed, 49 insertions(+), 2 deletions(-)
> 
> diff --git a/Documentation/config/stash.adoc b/Documentation/config/stash.adoc
> index ec1edaeba6..e556105a15 100644
> --- a/Documentation/config/stash.adoc
> +++ b/Documentation/config/stash.adoc
> @@ -1,3 +1,8 @@
> +stash.index::
> +	If this is set to true, `git stash apply` and `git stash pop` will
> +	behave as if `--index` was supplied. Defaults to false. See the
> +	descriptions in linkgit:git-stash[1].
> +
>   stash.showIncludeUntracked::
>   	If this is set to true, the `git stash show` command will show
>   	the untracked files of a stash entry.  Defaults to false. See
> diff --git a/builtin/stash.c b/builtin/stash.c
> index d9b478d1d1..8a0eef3c70 100644
> --- a/builtin/stash.c
> +++ b/builtin/stash.c
> @@ -130,6 +130,7 @@ static struct strbuf stash_index_path = STRBUF_INIT;
>   static int show_stat = 1;
>   static int show_patch;
>   static int show_include_untracked;
> +static int use_index;
>   
>   /*
>    * w_commit is set to the commit containing the working tree
> @@ -662,7 +663,7 @@ static int apply_stash(int argc, const char **argv, const char *prefix,
>   {
>   	int ret = -1;
>   	int quiet = 0;
> -	int index = 0;
> +	int index = use_index;
>   	struct stash_info info = STASH_INFO_INIT;
>   	struct option options[] = {
>   		OPT__QUIET(&quiet, N_("be quiet, only report errors")),
> @@ -759,7 +760,7 @@ static int pop_stash(int argc, const char **argv, const char *prefix,
>   		     struct repository *repo UNUSED)
>   {
>   	int ret = -1;
> -	int index = 0;
> +	int index = use_index;
>   	int quiet = 0;
>   	struct stash_info info = STASH_INFO_INIT;
>   	struct option options[] = {
> @@ -864,6 +865,10 @@ static int git_stash_config(const char *var, const char *value,
>   		show_include_untracked = git_config_bool(var, value);
>   		return 0;
>   	}
> +	if (!strcmp(var, "stash.index")) {
> +		use_index = git_config_bool(var, value);
> +		return 0;
> +	}
>   	return git_diff_basic_config(var, value, ctx, cb);
>   }
>   
> diff --git a/t/t3903-stash.sh b/t/t3903-stash.sh
> index b8936a653b..d6127173b1 100755
> --- a/t/t3903-stash.sh
> +++ b/t/t3903-stash.sh
> @@ -1595,4 +1595,41 @@ setup_stash()
>   	)
>   '
>   
> +test_expect_success 'stash.index=true implies --index' '
> +	# setup for a few related tests
> +	test_commit file base &&
> +	echo index >file &&
> +	git add file &&
> +	echo working >file &&
> +	git stash &&
> +
> +	test_when_finished "git reset --hard" &&
> +	git -c stash.index=true stash apply &&
> +	echo index >expect &&
> +	git show :0:file >actual &&
> +	test_cmp expect actual &&
> +	echo working >expect &&
> +	test_cmp expect file
> +'
> +
> +test_expect_success 'stash.index=true overridden by --no-index' '
> +	test_when_finished "git reset --hard" &&
> +	git -c stash.index=true stash apply --no-index &&
> +	echo base >expect &&
> +	git show :0:file >actual &&
> +	test_cmp expect actual &&
> +	echo working >expect &&
> +	test_cmp expect file
> +'
> +
> +test_expect_success 'stash.index=false overridden by --index' '
> +	test_when_finished "git reset --hard" &&
> +	git -c stash.index=false stash apply --index &&
> +	echo index >expect &&
> +	git show :0:file >actual &&
> +	test_cmp expect actual &&
> +	echo working >expect &&
> +	test_cmp expect file
> +'
> +
>   test_done

>
> Hi Ben
>
> On 22/09/2025 02:39, D. Ben Knoble wrote:
> > With stash.index=true, git-stash(1) command now tries to reinstate the
> > index by default in the "apply" and "pop" modes. Not doing so creates a
> > common trap [1], [2]: "git stash apply" is not the reverse of "git stash
> > push" because carefully staged indices are lost and have to be manually
> > recreated. OTOH, this mode is not always desirable and may create more
> > conflicts when applying stashes. As usual, "--no-index" will disable
> > this behavior if you set "stash.index".
>
> Thanks for updating the tests, they look good. As I said before I don't
> have a strong opinion about this change but I certainly don't object to
> it. I think this change will also affect the behavior of "git
> merge/pull/rebase --autostash" which we should maybe call out in the
> commit message. I don't think that change in behavior is a problem as it
> is probably what the user would expect when they set this config.

> On Mon, Sep 22, 2025 at 10:11 AM Phillip Wood <phillip.wood123@gmail.com> wrote:
>>
>> Hi Ben
>>
>> On 22/09/2025 02:39, D. Ben Knoble wrote:
>>> With stash.index=true, git-stash(1) command now tries to reinstate the
>>> index by default in the "apply" and "pop" modes. Not doing so creates a
>>> common trap [1], [2]: "git stash apply" is not the reverse of "git stash
>>> push" because carefully staged indices are lost and have to be manually
>>> recreated. OTOH, this mode is not always desirable and may create more
>>> conflicts when applying stashes. As usual, "--no-index" will disable
>>> this behavior if you set "stash.index".
>>
>> Thanks for updating the tests, they look good. As I said before I don't
>> have a strong opinion about this change but I certainly don't object to
>> it. I think this change will also affect the behavior of "git
>> merge/pull/rebase --autostash" which we should maybe call out in the
>> commit message. I don't think that change in behavior is a problem as it
>> is probably what the user would expect when they set this config.
> 
> Agreed, I hadn't considered that here. Should we also update the docs,
> do you think?

> we did not advertise the connection between the new config
> option stash.index and the implicit use of git-stash via --autostash
> (which may also be configured). Do so.
>
> Signed-off-by: D. Ben Knoble <ben.knoble+github@gmail.com>
> ---
>
> This builds on dk/stash-apply-index from gitster/git and is published at
> https://github.com/benknoble/git/tree/stash-apply-index-doc
>
>  Documentation/config/stash.adoc | 4 ++++
>  1 file changed, 4 insertions(+)
>
> diff --git a/Documentation/config/stash.adoc b/Documentation/config/stash.adoc
> index e556105a15..fcb9a4a7a0 100644
> --- a/Documentation/config/stash.adoc
> +++ b/Documentation/config/stash.adoc
> @@ -2,6 +2,10 @@ stash.index::
>  	If this is set to true, `git stash apply` and `git stash pop` will
>  	behave as if `--index` was supplied. Defaults to false. See the
>  	descriptions in linkgit:git-stash[1].
> ++
> +This also affects invocations of linkgit:git-stash[1] via `--autostash` from
> +commands like linkgit:git-merge[1], linkgit:git-rebase[1], and
> +linkgit:git-pull[1].

>
>  stash.showIncludeUntracked::
>  	If this is set to true, the `git stash show` command will show
>
> base-commit: 9842c0c7492d2858d64ef81128f7b1f0b38e326b
> --
> 2.48.1

>
> This follow-up patch makes sense.
>
> • It reads like a logical continuation of the previous commit 9842c0c749
> • The log message is clear (and with no spelling mistakes)
> • The markup is correct (list continuation, links)
> • `make lint-docs` passes
> • `./ci/check-whitespace.sh @^` passes
>
> On Mon, Oct 6, 2025, at 14:59, D. Ben Knoble wrote:
> > With 9842c0c749 (stash: honor stash.index in apply, pop modes,
> > 2025-09-21)
>
> Curiously, since this is also the base commit, referring to “the
> previous commit” would also work if this patch is indeed applied on top
> of that one. But maybe that contextual reference is a bad idea?

> > merged in a5d4779e6e (Merge branch 'dk/stash-apply-index',
> > 2025-09-29),
>
> This is over-specified IMO. Like mentioned this patch could be applied
> on top of commit 9842c0c749. Then that merge commit will not be
> reachable from this resulting commit.
>
> I also don’t see the point of mentioning when things were merged in in
> the commit message.

> > diff --git a/Documentation/config/stash.adoc b/Documentation/config/stash.adoc
> > index e556105a15..fcb9a4a7a0 100644
> > --- a/Documentation/config/stash.adoc
> > +++ b/Documentation/config/stash.adoc
> > @@ -2,6 +2,10 @@ stash.index::
> >       If this is set to true, `git stash apply` and `git stash pop` will
> >       behave as if `--index` was supplied. Defaults to false. See the
> >       descriptions in linkgit:git-stash[1].
> > ++
> > +This also affects invocations of linkgit:git-stash[1] via `--autostash` from
> > +commands like linkgit:git-merge[1], linkgit:git-rebase[1], and
> > +linkgit:git-pull[1].
>
> According to these
>
> • `git grep -- --autostash`
> • `git grep merge-options.adoc`
>
> This text exhaustively covers all commands which have this option.
>
> ... which might mean that “like” is an unneeded hedge? (it’s probably
> not intended to be a hedge)

> On Thu, Oct 9, 2025 at 6:55 PM Kristoffer Haugsbakk
> <code@khaugsbakk.name> wrote:
>>
>> This follow-up patch makes sense.
>>
>> • It reads like a logical continuation of the previous commit 9842c0c749
>> • The log message is clear (and with no spelling mistakes)
>> • The markup is correct (list continuation, links)
>> • `make lint-docs` passes
>> • `./ci/check-whitespace.sh @^` passes
>>
>> On Mon, Oct 6, 2025, at 14:59, D. Ben Knoble wrote:
>> > With 9842c0c749 (stash: honor stash.index in apply, pop modes,
>> > 2025-09-21)
>>
>> Curiously, since this is also the base commit, referring to “the
>> previous commit” would also work if this patch is indeed applied on top
>> of that one. But maybe that contextual reference is a bad idea?
> Generally I think so, but that's just my preference. Once commits have
> stable reference points, I'd rather use that. But I'm not attached to
> this one, so if we end up re-rolling, I can adjust either way.

>> This is over-specified IMO. Like mentioned this patch could be applied
>> on top of commit 9842c0c749. Then that merge commit will not be
>> reachable from this resulting commit.
>>
>> I also don’t see the point of mentioning when things were merged in in
>> the commit message.

>> >       If this is set to true, `git stash apply` and `git stash pop` will
>> >       behave as if `--index` was supplied. Defaults to false. See the
>> >       descriptions in linkgit:git-stash[1].
>> > ++
>> > +This also affects invocations of linkgit:git-stash[1] via `--autostash` from
>> > +commands like linkgit:git-merge[1], linkgit:git-rebase[1], and
>> > +linkgit:git-pull[1].
>>
>> According to these
>>
>> • `git grep -- --autostash`
>> • `git grep merge-options.adoc`
>>
>> This text exhaustively covers all commands which have this option.
>>
>> ... which might mean that “like” is an unneeded hedge? (it’s probably
>> not intended to be a hedge)

> 
> Le 11 oct. 2025 à 13:29, Junio C Hamano <gitster@pobox.com> a écrit :
> 
> "D. Ben Knoble" <ben.knoble+github@gmail.com> writes:
> 
>>> On Thu, Oct 9, 2025 at 6:55 PM Kristoffer Haugsbakk
>>> <code@khaugsbakk.name> wrote:
>>> 
>>> This is over-specified IMO. Like mentioned this patch could be applied
>>> on top of commit 9842c0c749. Then that merge commit will not be
>>> reachable from this resulting commit.
>>> 
>>> I also don’t see the point of mentioning when things were merged in in
>>> the commit message.
> 
> Yeah, that is less useful to me (there is a tool, given a commit
> object, to figure out at which merge it got merged to the mainline);
> I didn't think of a way the information can be useful to general
> readers.  If the mainline merge was a release or more ago, then it
> may make sense to say "commit X, which appeared in version Y, was
> broken in such and such way, and here is to fix its breakage"

>>>[snip]
>>
>> Yeah, that is less useful to me (there is a tool, given a commit
>> object, to figure out at which merge it got merged to the mainline);
>> I didn't think of a way the information can be useful to general
>> readers.  If the mainline merge was a release or more ago, then it
>> may make sense to say "commit X, which appeared in version Y, was
>> broken in such and such way, and here is to fix its breakage"
>
> It’s not clear to me if you’d like to see a v2 or not, but given you
> were already happy to merge in the latest what’s cooking, I don’t mind
> not sending one ;)

>
> On Sun, Oct 12, 2025, at 20:04, Ben Knoble wrote:
> >>>[snip]
> >>
> >> Yeah, that is less useful to me (there is a tool, given a commit
> >> object, to figure out at which merge it got merged to the mainline);
> >> I didn't think of a way the information can be useful to general
> >> readers.  If the mainline merge was a release or more ago, then it
> >> may make sense to say "commit X, which appeared in version Y, was
> >> broken in such and such way, and here is to fix its breakage"
> >
> > It’s not clear to me if you’d like to see a v2 or not, but given you
> > were already happy to merge in the latest what’s cooking, I don’t mind
> > not sending one ;)
>
> Sorry.  I misread a What’s Cooking email[1] as asking why a topic hadn’t
> been picked up. But you were asking about sending another patch on top
> of something which was already in `next`.