Skip to contentSkip to conversation

mingw: make is_hidden tests in t0001/t5611 more robust

PATCH In review Johannes Schindelin via GitGitGadget 2 participants 15 messages Apr 8, 2020 – Apr 11, 2020

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 mingw: make is_hidden tests in t0001/t5611 more robust

From: Johannes Schindelin <[email protected]>

We should not actually expect the first attrib.exe in the PATH to be the one we are looking for. Or that it is in the PATH, for that matter.

---
Make the tests that test core.hideDotFiles more robust

We have this feature on Windows where the files starting with a dot can be marked hidden (whether a file is hidden by default or not is a matter of naming convention on Unix, but it is an explicit flag on Windows). This patch improves the regression tests of this feature, and it has been carried in Git for Windows for over three years.

Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-603%2Fdscho%2Frobustify-is-hidden-tests-v1
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-603/dscho/robustify-is-hidden-tests-v1
Pull-Request: https://github.com/gitgitgadget/git/pull/603

 t/t0001-init.sh         | 2 +-
 t/t5611-clone-config.sh | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
t/t0001-init.sh +1 -1
index 26f82063267..2456688b281 100755
@@ -395,7 +395,7 @@ test_expect_success SYMLINKS 're-init to move gitdir symlink' '
 # Tests for the hidden file attribute on windows
 is_hidden () {
 	# Use the output of `attrib`, ignore the absolute path
-	case "$(attrib "$1")" in *H*?:*) return 0;; esac
+	case "$("$SYSTEMROOT"/system32/attrib "$1")" in *H*?:*) return 0;; esac
 	return 1
 }
 
t/t5611-clone-config.sh +1 -1
index 60c1ba951b7..87b8073cd74 100755
@@ -95,7 +95,7 @@ test_expect_success 'clone -c remote.<remote>.fetch=<refspec> --origin=<name>' '
 # Tests for the hidden file attribute on windows
 is_hidden () {
 	# Use the output of `attrib`, ignore the absolute path
-	case "$(attrib "$1")" in *H*?:*) return 0;; esac
+	case "$("$SYSTEMROOT"/system32/attrib "$1")" in *H*?:*) return 0;; esac
 	return 1
 }
 

base-commit: 9fadedd637b312089337d73c3ed8447e9f0aa775

Signature · 1 line
gitgitgadget
#1 lore raw

"Johannes Schindelin via GitGitGadget" <[email protected]>

t/t0001-init.sh test_expect_success SYMLINKS 're-init to move gitdir symlink' ' · line 400

Quoted 12 lines · writes:
t/t0001-init.sh +1 -1
index 26f82063267..2456688b281 100755
@@ -395,7 +395,7 @@ test_expect_success SYMLINKS 're-init to move gitdir symlink' '
 # Tests for the hidden file attribute on windows
 is_hidden () {
 	# Use the output of `attrib`, ignore the absolute path
-	case "$(attrib "$1")" in *H*?:*) return 0;; esac
+	case "$("$SYSTEMROOT"/system32/attrib "$1")" in *H*?:*) return 0;; esac
 	return 1
 }

I wondered if this (and the other one) want to be in test-lib but I am on the fence. All three tests that call this helper in t0001 are protected with MINGW prerequisite, but until I realized it, the call to "attrib" (whether it is given as a full path or relies on $PATH lookup) looked like a portability nightmare waiting to happen. It would make it even worse if we moved the above as-is to test-lib, as it is harder to see what the callers are doing once we did so.

With a change like this, however

	is_hidden () {
		if ! test_have_prereq MINGW
		then
			BUG "use of is_hidden outside MINGW prerequisite"
		fi
		case "$("$SYSTEMROOT"/system32/attrib "$1")" in 
		*H*?:*)	return 0 ;;
		*)	return 1 ;;
		esac
	}

I think it is OK to consolidate these two copies into one in test-lib

Thanks.

Quoted 15 lines
t/t5611-clone-config.sh +1 -1
index 60c1ba951b7..87b8073cd74 100755
@@ -95,7 +95,7 @@ test_expect_success 'clone -c remote.<remote>.fetch=<refspec> --origin=<name>' '
 # Tests for the hidden file attribute on windows
 is_hidden () {
 	# Use the output of `attrib`, ignore the absolute path
-	case "$(attrib "$1")" in *H*?:*) return 0;; esac
+	case "$("$SYSTEMROOT"/system32/attrib "$1")" in *H*?:*) return 0;; esac
 	return 1
 }
 

base-commit: 9fadedd637b312089337d73c3ed8447e9f0aa775

#2 lore raw

cover make "is_hidden" even more robust

Here are two patches on top of "mingw: make is_hidden tests in t0001/t5611 more robust" from yesterday to clean up the code by consolidating two copies and make it harder to misuse.

Junio C Hamano (2):
mingw: refactor test_path_is_hidden out to t/test-lib-functions.sh
t: protect against use of test_path_is_hidden outside MINGW

 t/t0001-init.sh         | 13 +++----------
 t/t5611-clone-config.sh | 13 +++----------
 t/test-lib-functions.sh | 14 ++++++++++++++
 3 files changed, 20 insertions(+), 20 deletions(-)
Signature · 1 line
2.26.0-106-g9fadedd637
#3 lore raw

PATCH 1/2 mingw: refactor test_path_is_hidden out to t/test-lib-functions.sh

Two scripts had a copy of the same helper function, which needed the same fix at the same time. Let's move it to a common place.

---
 t/t0001-init.sh         | 13 +++----------
 t/t5611-clone-config.sh | 13 +++----------
 t/test-lib-functions.sh |  9 +++++++++
 3 files changed, 15 insertions(+), 20 deletions(-)
t/t0001-init.sh +3 -10
index 2456688b28..6b2e2e3dc2 100755
@@ -392,13 +392,6 @@ test_expect_success SYMLINKS 're-init to move gitdir symlink' '
 	test_path_is_dir realgitdir/refs
 '
 
-# Tests for the hidden file attribute on windows
-is_hidden () {
-	# Use the output of `attrib`, ignore the absolute path
-	case "$("$SYSTEMROOT"/system32/attrib "$1")" in *H*?:*) return 0;; esac
-	return 1
-}
-
 test_expect_success MINGW '.git hidden' '
 	rm -rf newdir &&
 	(
@@ -406,7 +399,7 @@ test_expect_success MINGW '.git hidden' '
 		mkdir newdir &&
 		cd newdir &&
 		git init &&
-		is_hidden .git
+		test_path_is_hidden .git
 	) &&
 	check_config newdir/.git false unset
 '
@@ -419,7 +412,7 @@ test_expect_success MINGW 'bare git dir not hidden' '
 		cd newdir &&
 		git --bare init
 	) &&
-	! is_hidden newdir
+	! test_path_is_hidden newdir
 '
 
 test_expect_success 'remote init from does not use config from cwd' '
@@ -456,7 +449,7 @@ test_expect_success MINGW 'core.hidedotfiles = false' '
 		sane_unset GIT_DIR GIT_WORK_TREE GIT_CONFIG &&
 		git -C newdir init
 	) &&
-	! is_hidden newdir/.git
+	! test_path_is_hidden newdir/.git
 '
 
 test_expect_success MINGW 'redirect std handles' '
t/t5611-clone-config.sh +3 -10
index 87b8073cd7..8e0fd39823 100755
@@ -92,24 +92,17 @@ test_expect_success 'clone -c remote.<remote>.fetch=<refspec> --origin=<name>' '
 	test_cmp expect actual
 '
 
-# Tests for the hidden file attribute on windows
-is_hidden () {
-	# Use the output of `attrib`, ignore the absolute path
-	case "$("$SYSTEMROOT"/system32/attrib "$1")" in *H*?:*) return 0;; esac
-	return 1
-}
-
 test_expect_success MINGW 'clone -c core.hideDotFiles' '
 	test_commit attributes .gitattributes "" &&
 	rm -rf child &&
 	git clone -c core.hideDotFiles=false . child &&
-	! is_hidden child/.gitattributes &&
+	! test_path_is_hidden child/.gitattributes &&
 	rm -rf child &&
 	git clone -c core.hideDotFiles=dotGitOnly . child &&
-	! is_hidden child/.gitattributes &&
+	! test_path_is_hidden child/.gitattributes &&
 	rm -rf child &&
 	git clone -c core.hideDotFiles=true . child &&
-	is_hidden child/.gitattributes
+	test_path_is_hidden child/.gitattributes
 '
 
 test_done
t/test-lib-functions.sh +9 -0
index 352c213d52..39b478e731 100644
@@ -760,6 +760,15 @@ test_path_is_missing () {
 	fi
 }
 
+# Tests for the hidden file attribute on windows
+test_path_is_hidden () {
+	# Use the output of `attrib`, ignore the absolute path
+	case "$("$SYSTEMROOT"/system32/attrib "$1")" in
+	*H*?:*)		return 0;;
+	esac
+	return 1
+}
+
 # test_line_count checks that a file has the number of lines it
 # ought to. For example:
 #
Signature · 1 line
2.26.0-106-g9fadedd637
#4 lore raw

PATCH 2/2 t: protect against use of test_path_is_hidden outside MINGW

In reply to Junio C Hamano

This test helper function is meant to test if the "hidden" atttribute is set (or unset) as expected on Windows, and use of it outside MINGW prerequisite is an error. Ensure we have the prereq and trigger a BUG otherwise.

It is tempting to instead replace its implementation with something like

if test_have_prereq MINGW
then
... current Windows specific code ...
else
# ls without -a/-A hides paths that begin with a dot case "$(basename "$1")" in
.*) return 0 ;;
esac
fi
return 1

but one test in t0001 is designed specifically that with an option "repository/.git" of a newly created repository is *not* hidden on Windows, which means that it is impossible to make that feature work on POSIX systems and the above won't test what we want to test.

---
 t/test-lib-functions.sh | 5 +++++
 1 file changed, 5 insertions(+)
t/test-lib-functions.sh +5 -0
index 39b478e731..ad54863166 100644
@@ -762,6 +762,11 @@ test_path_is_missing () {
 
 # Tests for the hidden file attribute on windows
 test_path_is_hidden () {
+	if ! test_have_prereq MINGW
+	then
+		BUG "use of test_path_is_hidden without MINGW prerequisite"
+	fi
+
 	# Use the output of `attrib`, ignore the absolute path
 	case "$("$SYSTEMROOT"/system32/attrib "$1")" in
 	*H*?:*)		return 0;;
Signature · 1 line
2.26.0-106-g9fadedd637
#5 lore raw

v2 cover Make the tests that test core.hideDotFiles more robust

In reply to Johannes Schindelin via GitGitGadget

We have this feature on Windows where the files starting with a dot can be marked hidden (whether a file is hidden by default or not is a matter of naming convention on Unix, but it is an explicit flag on Windows). This patch improves the regression tests of this feature, and it has been carried in Git for Windows for over three years.

Junio, I'm sorry that I did not get to send v2 yesterday, and that you spent time on the two add-on patches even after I finalized this second iteration (but I was first waiting for the PR build to pass and while that happened, I got stuck in meeting after meeting).

Changes since v1:

* A preparatory patch now moves and renames is_hidden to test-lib-functions.sh
* A typo in the function's comment is fixed, while at it.

Johannes Schindelin (3):
t: consolidate the is_hidden functions
mingw: make test_path_is_hidden tests in t0001/t5611 more robust
t: fix casing of the operating system Windows

 t/t0001-init.sh         |  9 +--------
 t/t5611-clone-config.sh | 13 +++----------
 t/test-lib-functions.sh | 10 ++++++++++
 3 files changed, 14 insertions(+), 18 deletions(-)

base-commit: 9fadedd637b312089337d73c3ed8447e9f0aa775
Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-603%2Fdscho%2Frobustify-is-hidden-tests-v2
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-603/dscho/robustify-is-hidden-tests-v2
Pull-Request: https://github.com/gitgitgadget/git/pull/603

Range-diff vs v1:

 -:  ----------- > 1:  dd81ea68d6b t: consolidate the `is_hidden` functions
 1:  889f7c21333 ! 2:  fc4202cb548 mingw: make is_hidden tests in t0001/t5611 more robust

@@ Metadata
Author: Johannes Schindelin <[email protected]>

## Commit message ##
- mingw: make is_hidden tests in t0001/t5611 more robust
+ mingw: make test_path_is_hidden tests in t0001/t5611 more robust
+
+ This function uses Windows' system tool attrib to determine the state
+ of the hidden flag of a file or directory.

We should not actually expect the first attrib.exe in the PATH to be the one we are looking for. Or that it is in the PATH, for that matter.

+ Let's use the full path to the tool instead.
+

- ## t/t0001-init.sh ##
-@@ t/t0001-init.sh: test_expect_success SYMLINKS 're-init to move gitdir symlink' '
- # Tests for the hidden file attribute on windows
- is_hidden () {
- # Use the output of attrib, ignore the absolute path
-- case "$(attrib "$1")" in *H*?:*) return 0;; esac
-+ case "$("$SYSTEMROOT"/system32/attrib "$1")" in *H*?:*) return 0;; esac
- return 1
- }
+ ## t/test-lib-functions.sh ##
+@@ t/test-lib-functions.sh: test_path_is_hidden () {
+ BUG "test_path_is_hidden can only be used on Windows"

-
- ## t/t5611-clone-config.sh ##
-@@ t/t5611-clone-config.sh: test_expect_success 'clone -c remote.<remote>.fetch=<refspec> --origin=<name>' '
- # Tests for the hidden file attribute on windows
- is_hidden () {
# Use the output of attrib, ignore the absolute path
- case "$(attrib "$1")" in *H*?:*) return 0;; esac
+ case "$("$SYSTEMROOT"/system32/attrib "$1")" in *H*?:*) return 0;; esac
return 1
}
-
-: ----------- > 3: d0c0767a2aa t: fix casing of the operating system Windows

Signature · 1 line
gitgitgadget
#6 lore raw

PATCH v2 1/3 t: consolidate the `is_hidden` functions

From: Johannes Schindelin <[email protected]>

The is_hidden function can be used (only on Windows) to determine whether a directory or file have their hidden flag set.

This function is duplicated between two test scripts. It is better to move it into test-lib-functions.sh so that it is reused.

To make it safer to use, we specifically test for the MINGW prereq now, so that it is not used on a non-Windows platform by mistake.

---
 t/t0001-init.sh         |  9 +--------
 t/t5611-clone-config.sh | 13 +++----------
 t/test-lib-functions.sh | 10 ++++++++++
 3 files changed, 14 insertions(+), 18 deletions(-)
t/t0001-init.sh +1 -8
index 26f82063267..1edd5aeb8f0 100755
@@ -392,13 +392,6 @@ test_expect_success SYMLINKS 're-init to move gitdir symlink' '
 	test_path_is_dir realgitdir/refs
 '
 
-# Tests for the hidden file attribute on windows
-is_hidden () {
-	# Use the output of `attrib`, ignore the absolute path
-	case "$(attrib "$1")" in *H*?:*) return 0;; esac
-	return 1
-}
-
 test_expect_success MINGW '.git hidden' '
 	rm -rf newdir &&
 	(
@@ -406,7 +399,7 @@ test_expect_success MINGW '.git hidden' '
 		mkdir newdir &&
 		cd newdir &&
 		git init &&
-		is_hidden .git
+		test_path_is_hidden .git
 	) &&
 	check_config newdir/.git false unset
 '
t/t5611-clone-config.sh +3 -10
index 60c1ba951b7..8e0fd398236 100755
@@ -92,24 +92,17 @@ test_expect_success 'clone -c remote.<remote>.fetch=<refspec> --origin=<name>' '
 	test_cmp expect actual
 '
 
-# Tests for the hidden file attribute on windows
-is_hidden () {
-	# Use the output of `attrib`, ignore the absolute path
-	case "$(attrib "$1")" in *H*?:*) return 0;; esac
-	return 1
-}
-
 test_expect_success MINGW 'clone -c core.hideDotFiles' '
 	test_commit attributes .gitattributes "" &&
 	rm -rf child &&
 	git clone -c core.hideDotFiles=false . child &&
-	! is_hidden child/.gitattributes &&
+	! test_path_is_hidden child/.gitattributes &&
 	rm -rf child &&
 	git clone -c core.hideDotFiles=dotGitOnly . child &&
-	! is_hidden child/.gitattributes &&
+	! test_path_is_hidden child/.gitattributes &&
 	rm -rf child &&
 	git clone -c core.hideDotFiles=true . child &&
-	is_hidden child/.gitattributes
+	test_path_is_hidden child/.gitattributes
 '
 
 test_done
t/test-lib-functions.sh +10 -0
index 352c213d52e..09a2479fd38 100644
@@ -1543,3 +1543,13 @@ test_bitmap_traversal () {
 	test_cmp "$1.normalized" "$2.normalized" &&
 	rm -f "$1.normalized" "$2.normalized"
 }
+
+# Tests for the hidden file attribute on windows
+test_path_is_hidden () {
+	test_have_prereq MINGW ||
+	BUG "test_path_is_hidden can only be used on Windows"
+
+	# Use the output of `attrib`, ignore the absolute path
+	case "$(attrib "$1")" in *H*?:*) return 0;; esac
+	return 1
+}
Signature · 1 line
gitgitgadget
#7 lore raw

PATCH v2 2/3 mingw: make test_path_is_hidden tests in t0001/t5611 more robust

In reply to Johannes Schindelin via GitGitGadget

From: Johannes Schindelin <[email protected]>

This function uses Windows' system tool attrib to determine the state of the hidden flag of a file or directory.

We should not actually expect the first attrib.exe in the PATH to be the one we are looking for. Or that it is in the PATH, for that matter.

Let's use the full path to the tool instead.

---
 t/test-lib-functions.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
t/test-lib-functions.sh +1 -1
index 09a2479fd38..09e2cebb456 100644
@@ -1550,6 +1550,6 @@ test_path_is_hidden () {
 	BUG "test_path_is_hidden can only be used on Windows"
 
 	# Use the output of `attrib`, ignore the absolute path
-	case "$(attrib "$1")" in *H*?:*) return 0;; esac
+	case "$("$SYSTEMROOT"/system32/attrib "$1")" in *H*?:*) return 0;; esac
 	return 1
 }
Signature · 1 line
gitgitgadget
#8 lore raw

PATCH v2 3/3 t: fix casing of the operating system `Windows`

In reply to Johannes Schindelin via GitGitGadget

From: Johannes Schindelin <[email protected]>

The lower-case windows refers to something different.

---
 t/test-lib-functions.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
t/test-lib-functions.sh +1 -1
index 09e2cebb456..139647a6341 100644
@@ -1544,7 +1544,7 @@ test_bitmap_traversal () {
 	rm -f "$1.normalized" "$2.normalized"
 }
 
-# Tests for the hidden file attribute on windows
+# Tests for the hidden file attribute on Windows
 test_path_is_hidden () {
 	test_have_prereq MINGW ||
 	BUG "test_path_is_hidden can only be used on Windows"
Signature · 1 line
gitgitgadget
#9 lore raw

Re: [PATCH v2 0/3] Make the tests that test core.hideDotFiles more robust

In reply to Johannes Schindelin via GitGitGadget

"Johannes Schindelin via GitGitGadget" <[email protected]>

Quoted 4 lines · writes:

Junio, I'm sorry that I did not get to send v2 yesterday, and that you spent time on the two add-on patches even after I finalized this second iteration (but I was first waiting for the PR build to pass and while that happened, I got stuck in meeting after meeting).

Heh, mails cross all the time, so there is not much to be sorry, and it is not a race, and it's not much of an issue who sent an equivalent patch to the list first ;-)

A few issues I noticed that are not worth pointing out inline with the patches are

[1/3] The "consolidate" is a bit uneven. It keeps the 'windows'
(lowercase) to pretend to be a pure code movement, while adding the "prereq or die" that makes it an impure code movement. If I were doing this, this step would be pure code movement, and [3/3] would have the "prereq or die" as its main theme, i.e. make sure it is hard to misuse. [3/3] would also do "s/windows/Windows/" as "while at it".

[2/3] The test numbers on the title is no longer relevant, and I
would suggest retitling this step. No matter which test uses it currently, or any new uses added in the future to other tests, with this patch the helper is more robust, and it was the point of [1/3].

Thanks.

#10 lore raw

v3 cover Make the tests that test core.hideDotFiles more robust

In reply to Johannes Schindelin via GitGitGadget

We have this feature on Windows where the files starting with a dot can be marked hidden (whether a file is hidden by default or not is a matter of naming convention on Unix, but it is an explicit flag on Windows). This patch improves the regression tests of this feature, and it has been carried in Git for Windows for over three years.

Junio, I followed your advice, and did one more thing: the function is now renamed also only in 3/3.

Changes since v2:

* The first patch is now a true code move, no edits are snuck in.
* The typo fix, the function rename, and the prereq check are now all done as a "cleanup" patch (the third one).

Changes since v1:

* A preparatory patch now moves and renames is_hidden to test-lib-functions.sh
* A typo in the function's comment is fixed, while at it.

Johannes Schindelin (3):
t: consolidate the is_hidden functions
mingw: make test_path_is_hidden more robust
t: restrict is_hidden to be called only on Windows

 t/t0001-init.sh         |  9 +--------
 t/t5611-clone-config.sh | 13 +++----------
 t/test-lib-functions.sh | 10 ++++++++++
 3 files changed, 14 insertions(+), 18 deletions(-)

base-commit: 9fadedd637b312089337d73c3ed8447e9f0aa775
Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-603%2Fdscho%2Frobustify-is-hidden-tests-v3
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-603/dscho/robustify-is-hidden-tests-v3
Pull-Request: https://github.com/gitgitgadget/git/pull/603

Range-diff vs v2:

1: dd81ea68d6b ! 1: f13f9f78cda t: consolidate the is_hidden functions
@@ Commit message
This function is duplicated between two test scripts. It is better to move it into test-lib-functions.sh so that it is reused.

- To make it safer to use, we specifically test for the MINGW prereq
- now, so that it is not used on a non-Windows platform by mistake.
+ This patch is best viewed with --color-moved.

@@ t/t0001-init.sh: test_expect_success SYMLINKS 're-init to move gitdir symlink' '
test_expect_success MINGW '.git hidden' '
rm -rf newdir &&
(
-@@ t/t0001-init.sh: test_expect_success MINGW '.git hidden' '
- mkdir newdir &&
- cd newdir &&
- git init &&
-- is_hidden .git
-+ test_path_is_hidden .git
- ) &&
- check_config newdir/.git false unset
- '

## t/t5611-clone-config.sh ##
@@ t/t5611-clone-config.sh: test_expect_success 'clone -c remote.<remote>.fetch=<refspec> --origin=<name>' '
@@ t/t5611-clone-config.sh: test_expect_success 'clone -c remote.<remote>.fetch=<re
test_expect_success MINGW 'clone -c core.hideDotFiles' '
test_commit attributes .gitattributes "" &&
rm -rf child &&
- git clone -c core.hideDotFiles=false . child &&
-- ! is_hidden child/.gitattributes &&
-+ ! test_path_is_hidden child/.gitattributes &&
- rm -rf child &&
- git clone -c core.hideDotFiles=dotGitOnly . child &&
-- ! is_hidden child/.gitattributes &&
-+ ! test_path_is_hidden child/.gitattributes &&
- rm -rf child &&
- git clone -c core.hideDotFiles=true . child &&
-- is_hidden child/.gitattributes
-+ test_path_is_hidden child/.gitattributes
- '
-
- test_done

## t/test-lib-functions.sh ##
@@ t/test-lib-functions.sh: test_bitmap_traversal () {
@@ t/test-lib-functions.sh: test_bitmap_traversal () {
}
+
+# Tests for the hidden file attribute on windows
-+test_path_is_hidden () {
-+ test_have_prereq MINGW ||
-+ BUG "test_path_is_hidden can only be used on Windows"
-+
++is_hidden () {
+ # Use the output of attrib, ignore the absolute path
+ case "$(attrib "$1")" in *H*?:*) return 0;; esac
+ return 1
2: fc4202cb548 ! 2: 8650936b8d4 mingw: make test_path_is_hidden tests in t0001/t5611 more robust
@@ Metadata
Author: Johannes Schindelin <[email protected]>

## Commit message ##
- mingw: make test_path_is_hidden tests in t0001/t5611 more robust
+ mingw: make test_path_is_hidden more robust

This function uses Windows' system tool attrib to determine the state of the hidden flag of a file or directory.
@@ Commit message

## t/test-lib-functions.sh ##
-@@ t/test-lib-functions.sh: test_path_is_hidden () {
- BUG "test_path_is_hidden can only be used on Windows"
-
+@@ t/test-lib-functions.sh: test_bitmap_traversal () {
+ # Tests for the hidden file attribute on windows
+ is_hidden () {
# Use the output of attrib, ignore the absolute path
- case "$(attrib "$1")" in *H*?:*) return 0;; esac
+ case "$("$SYSTEMROOT"/system32/attrib "$1")" in *H*?:*) return 0;; esac

 3:  d0c0767a2aa < -:  ----------- t: fix casing of the operating system `Windows`
 -:  ----------- > 3:  b2604459917 t: restrict `is_hidden` to be called only on Windows
Signature · 1 line
gitgitgadget
#11 lore raw

PATCH v3 1/3 t: consolidate the `is_hidden` functions

From: Johannes Schindelin <[email protected]>

The is_hidden function can be used (only on Windows) to determine whether a directory or file have their hidden flag set.

This function is duplicated between two test scripts. It is better to move it into test-lib-functions.sh so that it is reused.

This patch is best viewed with --color-moved.

---
 t/t0001-init.sh         | 7 -------
 t/t5611-clone-config.sh | 7 -------
 t/test-lib-functions.sh | 7 +++++++
 3 files changed, 7 insertions(+), 14 deletions(-)
t/t0001-init.sh +0 -7
index 26f82063267..9cc919d8d1a 100755
@@ -392,13 +392,6 @@ test_expect_success SYMLINKS 're-init to move gitdir symlink' '
 	test_path_is_dir realgitdir/refs
 '
 
-# Tests for the hidden file attribute on windows
-is_hidden () {
-	# Use the output of `attrib`, ignore the absolute path
-	case "$(attrib "$1")" in *H*?:*) return 0;; esac
-	return 1
-}
-
 test_expect_success MINGW '.git hidden' '
 	rm -rf newdir &&
 	(
t/t5611-clone-config.sh +0 -7
index 60c1ba951b7..c861e12ea44 100755
@@ -92,13 +92,6 @@ test_expect_success 'clone -c remote.<remote>.fetch=<refspec> --origin=<name>' '
 	test_cmp expect actual
 '
 
-# Tests for the hidden file attribute on windows
-is_hidden () {
-	# Use the output of `attrib`, ignore the absolute path
-	case "$(attrib "$1")" in *H*?:*) return 0;; esac
-	return 1
-}
-
 test_expect_success MINGW 'clone -c core.hideDotFiles' '
 	test_commit attributes .gitattributes "" &&
 	rm -rf child &&
t/test-lib-functions.sh +7 -0
index 352c213d52e..7a7e7a38312 100644
@@ -1543,3 +1543,10 @@ test_bitmap_traversal () {
 	test_cmp "$1.normalized" "$2.normalized" &&
 	rm -f "$1.normalized" "$2.normalized"
 }
+
+# Tests for the hidden file attribute on windows
+is_hidden () {
+	# Use the output of `attrib`, ignore the absolute path
+	case "$(attrib "$1")" in *H*?:*) return 0;; esac
+	return 1
+}
Signature · 1 line
gitgitgadget
#12 lore raw

PATCH v3 2/3 mingw: make test_path_is_hidden more robust

In reply to Johannes Schindelin via GitGitGadget

From: Johannes Schindelin <[email protected]>

This function uses Windows' system tool attrib to determine the state of the hidden flag of a file or directory.

We should not actually expect the first attrib.exe in the PATH to be the one we are looking for. Or that it is in the PATH, for that matter.

Let's use the full path to the tool instead.

---
 t/test-lib-functions.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
t/test-lib-functions.sh +1 -1
index 7a7e7a38312..2f81463a240 100644
@@ -1547,6 +1547,6 @@ test_bitmap_traversal () {
 # Tests for the hidden file attribute on windows
 is_hidden () {
 	# Use the output of `attrib`, ignore the absolute path
-	case "$(attrib "$1")" in *H*?:*) return 0;; esac
+	case "$("$SYSTEMROOT"/system32/attrib "$1")" in *H*?:*) return 0;; esac
 	return 1
 }
Signature · 1 line
gitgitgadget
#13 lore raw

PATCH v3 3/3 t: restrict `is_hidden` to be called only on Windows

In reply to Johannes Schindelin via GitGitGadget

From: Johannes Schindelin <[email protected]>

The function won't work anywhere else, so let's mark it as an explicit bug if it is called on a non-Windows platform.

Let's also rename the function to avoid cluttering the global namespace with an overly-generic function name.

While at it, we also fix the code comment above that function: the lower-case windows refers to something different than Windows.

---
 t/t0001-init.sh         | 2 +-
 t/t5611-clone-config.sh | 6 +++---
 t/test-lib-functions.sh | 7 +++++--
 3 files changed, 9 insertions(+), 6 deletions(-)
t/t0001-init.sh +1 -1
index 9cc919d8d1a..1edd5aeb8f0 100755
@@ -399,7 +399,7 @@ test_expect_success MINGW '.git hidden' '
 		mkdir newdir &&
 		cd newdir &&
 		git init &&
-		is_hidden .git
+		test_path_is_hidden .git
 	) &&
 	check_config newdir/.git false unset
 '
t/t5611-clone-config.sh +3 -3
index c861e12ea44..8e0fd398236 100755
@@ -96,13 +96,13 @@ test_expect_success MINGW 'clone -c core.hideDotFiles' '
 	test_commit attributes .gitattributes "" &&
 	rm -rf child &&
 	git clone -c core.hideDotFiles=false . child &&
-	! is_hidden child/.gitattributes &&
+	! test_path_is_hidden child/.gitattributes &&
 	rm -rf child &&
 	git clone -c core.hideDotFiles=dotGitOnly . child &&
-	! is_hidden child/.gitattributes &&
+	! test_path_is_hidden child/.gitattributes &&
 	rm -rf child &&
 	git clone -c core.hideDotFiles=true . child &&
-	is_hidden child/.gitattributes
+	test_path_is_hidden child/.gitattributes
 '
 
 test_done
t/test-lib-functions.sh +5 -2
index 2f81463a240..139647a6341 100644
@@ -1544,8 +1544,11 @@ test_bitmap_traversal () {
 	rm -f "$1.normalized" "$2.normalized"
 }
 
-# Tests for the hidden file attribute on windows
-is_hidden () {
+# Tests for the hidden file attribute on Windows
+test_path_is_hidden () {
+	test_have_prereq MINGW ||
+	BUG "test_path_is_hidden can only be used on Windows"
+
 	# Use the output of `attrib`, ignore the absolute path
 	case "$("$SYSTEMROOT"/system32/attrib "$1")" in *H*?:*) return 0;; esac
 	return 1
Signature · 1 line
gitgitgadget
#14 lore raw

Re: [PATCH v3 0/3] Make the tests that test core.hideDotFiles more robust

In reply to Johannes Schindelin via GitGitGadget

"Johannes Schindelin via GitGitGadget" <[email protected]>

Quoted 8 lines · writes:

We have this feature on Windows where the files starting with a dot can be marked hidden (whether a file is hidden by default or not is a matter of naming convention on Unix, but it is an explicit flag on Windows). This patch improves the regression tests of this feature, and it has been carried in Git for Windows for over three years.

Junio, I followed your advice, and did one more thing: the function is now renamed also only in 3/3.

Looks good.

I would have renamed it when making it global, exactly for the reason you stated in [3/3], but it's not like we would merge only the first two without the fixup, so it does not matter in practice (as long as other people won't pick up a bad habit from the example, that is).

Thanks. Let's start merging a lot of things down to 'next'.