[PATCH 1/2] clean: avoid looking for nested repository when appropriate

[PATCH 1/2] clean: avoid looking for nested repository when appropriate

[Patrick Marlier]

avoiding the unnecessary checks for is_nonbare_repository_dir() via setting DIR_NO_GITLINKS
---
 builtin/clean.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/builtin/clean.c b/builtin/clean.c
index 3ff02bbbff..18b37e3fd9 100644
--- a/builtin/clean.c
+++ b/builtin/clean.c
@@ -955,9 +955,10 @@ int cmd_clean(int argc, const char **argv, const char *prefix)
 				  " refusing to clean"));
 	}
 
-	if (force > 1)
+	if (force > 1) {
 		rm_flags = 0;
-	else
+		dir.flags |= DIR_NO_GITLINKS;
+	} else
 		dir.flags |= DIR_SKIP_NESTED_GIT;
 
 	dir.flags |= DIR_SHOW_OTHER_DIRECTORIES;
-- 
2.35.1

[PATCH 2/2] clean: avoid to differentiate untracked and ignored when appropriate

[Patrick Marlier]

---
 builtin/clean.c  |  4 +++-
 t/t7300-clean.sh | 24 ++++++++++++++++++++++++
 2 files changed, 27 insertions(+), 1 deletion(-)

diff --git a/builtin/clean.c b/builtin/clean.c
index 18b37e3fd9..1b1454d052 100644
--- a/builtin/clean.c
+++ b/builtin/clean.c
@@ -978,7 +978,9 @@ int cmd_clean(int argc, const char **argv, const char *prefix)
 		remove_directories = 1;
 	}
 
-	if (remove_directories && !ignored_only) {
+	if (remove_directories && ignored && !exclude_list.nr && force > 1)
+		; /* No need to recurse to look for ignored files */
+	else if (remove_directories && !ignored_only) {
 		/*
 		 * We need to know about ignored files too:
 		 *
diff --git a/t/t7300-clean.sh b/t/t7300-clean.sh
index 0399701e62..684eba914b 100755
--- a/t/t7300-clean.sh
+++ b/t/t7300-clean.sh
@@ -788,4 +788,28 @@ test_expect_success 'traverse into directories that may have ignored entries' '
 	)
 '
 
+test_expect_success 'avoid traversing into untracked directories' '
+	test_when_finished rm -f output error trace.* &&
+	test_create_repo avoid-traversing-untracked-hierarchy &&
+	(
+		cd avoid-traversing-untracked-hierarchy &&
+
+		mkdir -p untracked/subdir/with/b &&
+		mkdir -p untracked/subdir/with/a &&
+		>untracked/subdir/with/a/random-file.txt &&
+
+		GIT_TRACE2_PERF="$TRASH_DIRECTORY/trace.output" \
+		git clean -ffdx
+	) &&
+
+	# Make sure we only visited into the top-level directory, and did
+	# not traverse into the "untracked" subdirectory since it was excluded
+	grep data.*read_directo.*directories-visited trace.output |
+		cut -d "|" -f 9 >trace.relevant &&
+	cat >trace.expect <<-EOF &&
+	 ..directories-visited:1
+	EOF
+	test_cmp trace.expect trace.relevant
+'
+
 test_done
-- 
2.35.1

Re: [PATCH 1/2] clean: avoid looking for nested repository when appropriate

[Elijah Newren]

Hi,

On Tue, Feb 15, 2022 at 2:16 PM Patrick Marlier
<patrick.marlier@gmail.com> wrote:
>
> avoiding the unnecessary checks for is_nonbare_repository_dir() via setting DIR_NO_GITLINKS

Looks great, but a few details about commit messages that we like to see:

  * Please wrap commit messages at 72 characters
  * Describe your changes in imperative mood (i.e. "Avoid the
unnecessary" rather than "avoiding the unnecessary")
  * Use complete sentences for everything other than the subject.

So, perhaps:

"""
clean: avoid looking for nested repositories when unnecessary

With `git clean --ff` we will be deleting nested untracked repositories,
so there is no need to differentiate them from other untracked files.
Use the DIR_NO_GITLINKS flag in dir.flags to signify this and avoid the
is_nonbare_repository_dir() checks.
"""

> ---
>  builtin/clean.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/builtin/clean.c b/builtin/clean.c
> index 3ff02bbbff..18b37e3fd9 100644
> --- a/builtin/clean.c
> +++ b/builtin/clean.c
> @@ -955,9 +955,10 @@ int cmd_clean(int argc, const char **argv, const char *prefix)
>                                   " refusing to clean"));
>         }
>
> -       if (force > 1)
> +       if (force > 1) {
>                 rm_flags = 0;
> -       else
> +               dir.flags |= DIR_NO_GITLINKS;
> +       } else
>                 dir.flags |= DIR_SKIP_NESTED_GIT;
>
>         dir.flags |= DIR_SHOW_OTHER_DIRECTORIES;
> --
> 2.35.1

Re: [PATCH 2/2] clean: avoid to differentiate untracked and ignored when appropriate

[Elijah Newren]

On Tue, Feb 15, 2022 at 2:16 PM Patrick Marlier
<patrick.marlier@gmail.com> wrote:

Could I suggest a bit longer commit message?  Perhaps:

"""
clean: avoid traversing into untracked dirs when unnecessary

When deleting all untracked and ignored files and any nested
repositories (such as with `git clean -ffdx`), we do not need to recurse
into an untracked directory to see if any of the entries under it are
ignored or a nested repository.  Special case this condition to avoid
unnecessary recursion.
"""

> ---
>  builtin/clean.c  |  4 +++-
>  t/t7300-clean.sh | 24 ++++++++++++++++++++++++
>  2 files changed, 27 insertions(+), 1 deletion(-)
>
> diff --git a/builtin/clean.c b/builtin/clean.c
> index 18b37e3fd9..1b1454d052 100644
> --- a/builtin/clean.c
> +++ b/builtin/clean.c
> @@ -978,7 +978,9 @@ int cmd_clean(int argc, const char **argv, const char *prefix)
>                 remove_directories = 1;
>         }
>
> -       if (remove_directories && !ignored_only) {
> +       if (remove_directories && ignored && !exclude_list.nr && force > 1)
> +               ; /* No need to recurse to look for ignored files */
> +       else if (remove_directories && !ignored_only) {
>                 /*
>                  * We need to know about ignored files too:
>                  *
> diff --git a/t/t7300-clean.sh b/t/t7300-clean.sh
> index 0399701e62..684eba914b 100755
> --- a/t/t7300-clean.sh
> +++ b/t/t7300-clean.sh
> @@ -788,4 +788,28 @@ test_expect_success 'traverse into directories that may have ignored entries' '
>         )
>  '
>
> +test_expect_success 'avoid traversing into untracked directories' '
> +       test_when_finished rm -f output error trace.* &&
> +       test_create_repo avoid-traversing-untracked-hierarchy &&

I think you may have been copying from an example earlier in the file
that I like wrote.  As someone else recently pointed out to me,
t/test-lib-functions.sh points out that test_create_repo is deprecated
and we should just use "git init <dirname>" instead.

> +       (
> +               cd avoid-traversing-untracked-hierarchy &&
> +
> +               mkdir -p untracked/subdir/with/b &&
> +               mkdir -p untracked/subdir/with/a &&
> +               >untracked/subdir/with/a/random-file.txt &&
> +
> +               GIT_TRACE2_PERF="$TRASH_DIRECTORY/trace.output" \
> +               git clean -ffdx
> +       ) &&
> +
> +       # Make sure we only visited into the top-level directory, and did
> +       # not traverse into the "untracked" subdirectory since it was excluded
> +       grep data.*read_directo.*directories-visited trace.output |
> +               cut -d "|" -f 9 >trace.relevant &&
> +       cat >trace.expect <<-EOF &&
> +        ..directories-visited:1
> +       EOF
> +       test_cmp trace.expect trace.relevant
> +'
> +
>  test_done
> --
> 2.35.1

Looks good, other than those minor items.  Thanks for working on this!