threads / patch / 19017

patchtests: remove xargs in favor of --stdin where possible

Subject: [PATCH] tests: remove xargs in favor of --stdin where possible

## tl;dr

6 messages between Apr 23, 2009 and Apr 23, 2009. Diffs are folded; open one to read it.

replies: 5people: 3as markdown or json

Nguyễn Thái Ngọc Duy· Apr 23, 2009, 06:31 UTC · lore

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 Patch "Convert to use quiet option when available" reminds me I had a similar
 patch to remove (mostly) xargs usage from tests.

 t/t0000-basic.sh            |    2 +-
 t/t3100-ls-tree-restrict.sh |    2 +-
 t/t3101-ls-tree-dirname.sh  |    2 +-
 t/t4112-apply-renames.sh    |    2 +-
 t/t5000-tar-tree.sh         |    4 ++--
 5 files changed, 6 insertions(+), 6 deletions(-)

Show changes to 5 files +6 −6

t/t0000-basic.sh, t/t3100-ls-tree-restrict.sh, t/t3101-ls-tree-dirname.sh, t/t4112-apply-renames.sh, t/t5000-tar-tree.sh

diff --git a/t/t0000-basic.sh b/t/t0000-basic.sh
index f4ca4fc..5c07645 100755
--- a/t/t0000-basic.sh
+++ b/t/t0000-basic.sh
@@ -142,7 +142,7 @@ fi
 
 test_expect_success \
     'adding various types of objects with git update-index --add.' \
-    'find path* ! -type d -print | xargs git update-index --add'
+    'find path* ! -type d -print | git update-index --add --stdin'
 
 # Show them and see that matches what we expect.
 test_expect_success \
diff --git a/t/t3100-ls-tree-restrict.sh b/t/t3100-ls-tree-restrict.sh
index ee60d03..6e4faf8 100755
--- a/t/t3100-ls-tree-restrict.sh
+++ b/t/t3100-ls-tree-restrict.sh
@@ -39,7 +39,7 @@ test_expect_success \
      echo Lo >path2/foo &&
      echo Mi >path2/baz/b &&
      find path? \( -type f -o -type l \) -print |
-     xargs git update-index --add &&
+     git update-index --add --stdin &&
      tree=`git write-tree` &&
      echo $tree'
 
diff --git a/t/t3101-ls-tree-dirname.sh b/t/t3101-ls-tree-dirname.sh
index 51cb4a3..e0bf312 100755
--- a/t/t3101-ls-tree-dirname.sh
+++ b/t/t3101-ls-tree-dirname.sh
@@ -35,7 +35,7 @@ test_expect_success \
      echo 111 >path3/1.txt &&
      echo 222 >path3/2.txt &&
      find *.txt path* \( -type f -o -type l \) -print |
-     xargs git update-index --add &&
+     git update-index --add --stdin &&
      tree=`git write-tree` &&
      echo $tree'
 
diff --git a/t/t4112-apply-renames.sh b/t/t4112-apply-renames.sh
index f9ad183..6e56fa6 100755
--- a/t/t4112-apply-renames.sh
+++ b/t/t4112-apply-renames.sh
@@ -135,7 +135,7 @@ copy to klibc/arch/README
 +add a few lines at the end of it.
 EOF
 
-find klibc -type f -print | xargs git update-index --add --
+find klibc -type f -print | git update-index --add --stdin
 
 test_expect_success 'check rename/copy patch' 'git apply --check patch'
 
diff --git a/t/t5000-tar-tree.sh b/t/t5000-tar-tree.sh
index abb41b0..f036988 100755
--- a/t/t5000-tar-tree.sh
+++ b/t/t5000-tar-tree.sh
@@ -54,8 +54,8 @@ test_expect_success \
 
 test_expect_success \
     'add files to repository' \
-    'find a -type f | xargs git update-index --add &&
-     find a -type l | xargs git update-index --add &&
+    'find a -type f | git update-index --add --stdin &&
+     find a -type l | git update-index --add --stdin &&
      treeid=`git write-tree` &&
      echo $treeid >treeid &&
      git update-ref HEAD $(TZ=GMT GIT_COMMITTER_DATE="2005-05-27 22:00:00" \

-- 
1.6.2.2.693.g5a1be

Junio C Hamano· Apr 23, 2009, 08:36 UTC · re: Nguyễn Thái Ngọc Duy · lore

Re: [PATCH] tests: remove xargs in favor of --stdin where possible

Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:

> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
> ---
>  Patch "Convert to use quiet option when available" reminds me I had a similar
>  patch to remove (mostly) xargs usage from tests.

The reason being...?

 (1) xargs is not universally available, or portable enough?

 (2) people may learn from tests, and we should demostrate our ability?

 (3) something else?

If the reason is (1) I would understand it, but if it is (2) then I am a bit reluctant.

The tests may not break with your change because none of them may use problematic characters (especially "\n" and '"'), but update-index --stdin without -z is not suitable for reading from output from "find" without -0 option (on the other hand, "update-index -z --stdin" is good for reading output from "find -0"; but for portability we avoid GNUism "find -0").

Johannes Sixt· Apr 23, 2009, 09:22 UTC · re: Junio C Hamano · lore

Re: [PATCH] tests: remove xargs in favor of --stdin where possible

Junio C Hamano schrieb:

> The tests may not break with your change because none of them may use
> problematic characters (especially "\n" and '"'), but update-index --stdin
> without -z is not suitable for reading from output from "find" without -0
> option

Why should this be so? I thought I need -0/-z only if there are "\n" in file names (and I specifially talk about a find | update-index pipeline).

> (on the other hand, "update-index -z --stdin" is good for reading
> output from "find -0"; but for portability we avoid GNUism "find -0").

Sure, -0/-z is safer and to be used in scripts. But for an ad-hoc command that I type on the command line, why should I use -0/-z if I know that I don't have any "\n" in my file names?

-- Hannes

Nguyen Thai Ngoc Duy· Apr 23, 2009, 10:07 UTC · re: Junio C Hamano · lore

Re: [PATCH] tests: remove xargs in favor of --stdin where possible

2009/4/23 Junio C Hamano <gitster@pobox.com>:

Show 14 quoted lines

> Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:
>
>> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
>> ---
>>  Patch "Convert to use quiet option when available" reminds me I had a similar
>>  patch to remove (mostly) xargs usage from tests.
>
> The reason being...?
>
>  (1) xargs is not universally available, or portable enough?
>
>  (2) people may learn from tests, and we should demostrate our ability?
>
>  (3) something else?

I worked on busybox win32 port, less commands meant less work for me. So (1).

Show 5 quoted lines

> The tests may not break with your change because none of them may use
> problematic characters (especially "\n" and '"'), but update-index --stdin
> without -z is not suitable for reading from output from "find" without -0
> option (on the other hand, "update-index -z --stdin" is good for reading
> output from "find -0"; but for portability we avoid GNUism "find -0").

It can't be as safe as find -0| update-index -z, but it would be equivalent to find|xargs, isn't it? Both separate arguments by \n.

-- 
Duy

Junio C Hamano· Apr 23, 2009, 10:25 UTC · re: Nguyen Thai Ngoc Duy · lore

Re: [PATCH] tests: remove xargs in favor of --stdin where possible

Nguyen Thai Ngoc Duy <pclouds@gmail.com> writes:

Show 8 quoted lines

>> The tests may not break with your change because none of them may use
>> problematic characters (especially "\n" and '"'), but update-index --stdin
>> without -z is not suitable for reading from output from "find" without -0
>> option (on the other hand, "update-index -z --stdin" is good for reading
>> output from "find -0"; but for portability we avoid GNUism "find -0").
>
> It can't be as safe as find -0| update-index -z, but it would be
> equivalent to find|xargs, isn't it? Both separate arguments by \n.

Have you worked with pathnames with high-bit set and without core.quotepath set to false?

Nguyen Thai Ngoc Duy· Apr 23, 2009, 23:22 UTC · re: Junio C Hamano · lore

Re: [PATCH] tests: remove xargs in favor of --stdin where possible

On Thu, Apr 23, 2009 at 8:25 PM, Junio C Hamano <gitster@pobox.com> wrote:

Show 13 quoted lines

> Nguyen Thai Ngoc Duy <pclouds@gmail.com> writes:
>
>>> The tests may not break with your change because none of them may use
>>> problematic characters (especially "\n" and '"'), but update-index --stdin
>>> without -z is not suitable for reading from output from "find" without -0
>>> option (on the other hand, "update-index -z --stdin" is good for reading
>>> output from "find -0"; but for portability we avoid GNUism "find -0").
>>
>> It can't be as safe as find -0| update-index -z, but it would be
>> equivalent to find|xargs, isn't it? Both separate arguments by \n.
>
> Have you worked with pathnames with high-bit set and without
> core.quotepath set to false?

OK I see. There are some fancy unquoting going on inside. I'd take the patch back.

-- 
Duy