threads / patch / 61986

patchdocs: explain the order of output in the batched mode of git-cat-file(1)

Subject: [PATCH] docs: explain the order of output in the batched mode of git-cat-file(1)

## tl;dr

6 messages between Aug 21, 2024 and Aug 22, 2024. Diffs are folded; open one to read it.

replies: 5people: 3as markdown or json

ahmed akef via GitGitGadget· Aug 21, 2024, 09:28 UTC · lore

From: ahmed akef <aemed.akef.1@gmail.com>

The batched mode of git-cat-file(1) reads multiple objects from stdin and prints their respective contents to stdout. The order in which those objects are printed is not documented and may not be immediately obvious to the user. Document it.

Signed-off-by: ahmed akef <aemed.akef.1@gmail.com>
---
    docs: explain the order of output in The batched mode of git-cat-file(1)
    
    The batched mode of git-cat-file(1) reads multiple objects from stdin
    and prints their respective contents to stdout. The order in which those
    objects are printed is not documented and may not be immediately obvious
    to the user. Document it.

Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-git-1768%2Fahmedakef%2Fexplain-the-order-of-output-in-cat-file-batch-operations-v1
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-git-1768/ahmedakef/explain-the-order-of-output-in-cat-file-batch-operations-v1
Pull-Request: https://github.com/git/git/pull/1768

 Documentation/git-cat-file.txt | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Show changes to Documentation/git-cat-file.txt +2 −2

diff --git a/Documentation/git-cat-file.txt b/Documentation/git-cat-file.txt
index bd95a6c10a7..f1e0b4a7219 100644
--- a/Documentation/git-cat-file.txt
+++ b/Documentation/git-cat-file.txt
@@ -270,8 +270,8 @@ BATCH OUTPUT
 ------------
 
 If `--batch` or `--batch-check` is given, `cat-file` will read objects
-from stdin, one per line, and print information about them. By default,
-the whole line is considered as an object, as if it were fed to
+from stdin, one per line, and print information about them sequentially in the same order.
+By default, the whole line is considered as an object, as if it were fed to
 linkgit:git-rev-parse[1].
 
 When `--batch-command` is given, `cat-file` will read commands from stdin,

base-commit: 80ccd8a2602820fdf896a8e8894305225f86f61d

-- 
gitgitgadget

ahmed akef via GitGitGadget· Aug 21, 2024, 10:05 UTC · re: ahmed akef via GitGitGadget · lore

[PATCH v2] docs: explain the order of output in the batched mode of git-cat-file(1)

From: ahmed akef <aemed.akef.1@gmail.com>

Signed-off-by: ahmed akef <aemed.akef.1@gmail.com>
---
    docs: explain the order of output in The batched mode of git-cat-file(1)
    
    this is the same change as https://github.com/git/git/pull/1761 but dues
    to missteps, the PR got closed and I couldn't fix it, also applied the
    review comments from @pks-t
    cc: Patrick Steinhardt ps@pks.im

Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-git-1768%2Fahmedakef%2Fexplain-the-order-of-output-in-cat-file-batch-operations-v2
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-git-1768/ahmedakef/explain-the-order-of-output-in-cat-file-batch-operations-v2
Pull-Request: https://github.com/git/git/pull/1768

Range-diff vs v1:

 1:  86a982884ec ! 1:  3f742957aa1 docs: explain the order of output in the batched mode of git-cat-file(1)
     @@ Documentation/git-cat-file.txt: BATCH OUTPUT
       If `--batch` or `--batch-check` is given, `cat-file` will read objects
      -from stdin, one per line, and print information about them. By default,
      -the whole line is considered as an object, as if it were fed to
     -+from stdin, one per line, and print information about them sequentially in the same order.
     -+By default, the whole line is considered as an object, as if it were fed to
     - linkgit:git-rev-parse[1].
     +-linkgit:git-rev-parse[1].
     ++from stdin, one per line, and print information about them in the same
     ++order as they have been read from stdin. By default, the whole line is
     ++considered as an object, as if it were fed to linkgit:git-rev-parse[1].
       
       When `--batch-command` is given, `cat-file` will read commands from stdin,
     + one per line, and print information based on the command given. With

 Documentation/git-cat-file.txt | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

Show changes to Documentation/git-cat-file.txt +3 −3

diff --git a/Documentation/git-cat-file.txt b/Documentation/git-cat-file.txt
index bd95a6c10a7..c39074b9ee6 100644
--- a/Documentation/git-cat-file.txt
+++ b/Documentation/git-cat-file.txt
@@ -270,9 +270,9 @@ BATCH OUTPUT
 ------------
 
 If `--batch` or `--batch-check` is given, `cat-file` will read objects
-from stdin, one per line, and print information about them. By default,
-the whole line is considered as an object, as if it were fed to
-linkgit:git-rev-parse[1].
+from stdin, one per line, and print information about them in the same
+order as they have been read from stdin. By default, the whole line is
+considered as an object, as if it were fed to linkgit:git-rev-parse[1].
 
 When `--batch-command` is given, `cat-file` will read commands from stdin,
 one per line, and print information based on the command given. With

base-commit: 80ccd8a2602820fdf896a8e8894305225f86f61d

-- 
gitgitgadget

Junio C Hamano· Aug 21, 2024, 17:19 UTC · re: ahmed akef via GitGitGadget · lore

Re: [PATCH v2] docs: explain the order of output in the batched mode of git-cat-file(1)

"ahmed akef via GitGitGadget" <gitgitgadget@gmail.com> writes:

Show 7 quoted lines

>  If `--batch` or `--batch-check` is given, `cat-file` will read objects
> -from stdin, one per line, and print information about them. By default,
> -the whole line is considered as an object, as if it were fed to
> -linkgit:git-rev-parse[1].
> +from stdin, one per line, and print information about them in the same
> +order as they have been read from stdin. By default, the whole line is
> +considered as an object, as if it were fed to linkgit:git-rev-parse[1].

A few "Huh?" I had while reading the above.

 * "as they have been read from stdin"; drop "from stdin" here, as
   we already know we are talking about the mode that reads object
   names from the standard input and there is no need to repeat it.

 * "considered as an object" -> "considered to be an object name" or
   "used as an object name" [*].  This primarily comes from my
   spinal reflex against "consider as", plus my desire to be more
   precise in terminology.

Thanks.

Nothing mentioned below should be part of this patch, but as I noticed it while studying the current documentation to prepare this review, I'll record them as #leftoverbits.

The description of how lines read from the standard input should look like needs more work. Documentation on "--batch" says "the input lines must specify the path, separated by whitespace", but is it clear that it expects "<object name>", followed by a whitespace (not necessarily a single SP), followed by "<path>"? Without prior knowledge, I wouldn't be surprised if somebody read the text as asking for paths separated by whitespace, e.g.

    README.txt COPYING Makefile

that gives three paths. The text needs to be tightened to say something like "must give the path after the object name, separated by whitespace. The path is used to find the textconv and smudge filter".

The section also says "See the section BATCH OUTPUT below for details." but the section it refers to does not talk anything about this whitespace thing. It also is unclear what would happen if the input line says:

    :COPYING Makefile

Would it apply the textconv/filters meant for Makefile to the blob stored at COPYING in the index? If we say

    :README.txt
    
would the command be smart enough to know that the blob came from
the path README.txt and apply the textconv/filters meant for that
path, without the input repeating the same information twice like:

    :README.txt README.txt

or something silly like that?

[Reference]

* https://www.britannica.com/dictionary/eb/qa/consider-and-consider-as

ahmed akef via GitGitGadget· Aug 22, 2024, 19:50 UTC · re: ahmed akef via GitGitGadget · lore

[PATCH v3] docs: explain the order of output in the batched mode of git-cat-file(1)

From: ahmed akef <aemed.akef.1@gmail.com>

Signed-off-by: ahmed akef <aemed.akef.1@gmail.com>
---
    docs: explain the order of output in The batched mode of git-cat-file(1)
    
    this is the same change as https://github.com/git/git/pull/1761 but due
    to missteps, the PR got closed and I couldn't fix it, also applied the
    review comments from @pks-t
    cc: Patrick Steinhardt ps@pks.im

Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-git-1768%2Fahmedakef%2Fexplain-the-order-of-output-in-cat-file-batch-operations-v3
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-git-1768/ahmedakef/explain-the-order-of-output-in-cat-file-batch-operations-v3
Pull-Request: https://github.com/git/git/pull/1768

Range-diff vs v2:

 1:  3f742957aa1 ! 1:  6adeb2c7bb3 docs: explain the order of output in the batched mode of git-cat-file(1)
     @@ Documentation/git-cat-file.txt: BATCH OUTPUT
      -the whole line is considered as an object, as if it were fed to
      -linkgit:git-rev-parse[1].
      +from stdin, one per line, and print information about them in the same
     -+order as they have been read from stdin. By default, the whole line is
     ++order as they have been read. By default, the whole line is
      +considered as an object, as if it were fed to linkgit:git-rev-parse[1].
       
       When `--batch-command` is given, `cat-file` will read commands from stdin,

 Documentation/git-cat-file.txt | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

Show changes to Documentation/git-cat-file.txt +3 −3

diff --git a/Documentation/git-cat-file.txt b/Documentation/git-cat-file.txt
index bd95a6c10a7..d5890ae3686 100644
--- a/Documentation/git-cat-file.txt
+++ b/Documentation/git-cat-file.txt
@@ -270,9 +270,9 @@ BATCH OUTPUT
 ------------
 
 If `--batch` or `--batch-check` is given, `cat-file` will read objects
-from stdin, one per line, and print information about them. By default,
-the whole line is considered as an object, as if it were fed to
-linkgit:git-rev-parse[1].
+from stdin, one per line, and print information about them in the same
+order as they have been read. By default, the whole line is
+considered as an object, as if it were fed to linkgit:git-rev-parse[1].
 
 When `--batch-command` is given, `cat-file` will read commands from stdin,
 one per line, and print information based on the command given. With

base-commit: 80ccd8a2602820fdf896a8e8894305225f86f61d

-- 
gitgitgadget

Ahmed Akef· Aug 22, 2024, 07:31 UTC · lore

Re: [PATCH v2] docs: explain the order of output in the batched mode of git-cat-file(1)

> * "as they have been read from stdin"; drop "from stdin" here, as
>   we already know we are talking about the mode that reads object
>   names from the standard input and there is no need to repeat it.

it is needed to explain that git will not do any optimization to the order of paths before printing the output. I entered a discussion with someone who was worried that git may optimize the paths order because it is not stated clearly that output follows the same order as input.

> * "considered as an object" -> "considered to be an object name" or
>   "used as an object name" [*].  This primarily comes from my
>   spinal reflex against "consider as", plus my desire to be more
>   precise in terminology.

this is not related to my changes, I just moved this line but didn't change it.

Thanks, ahmed akef

On Thu, Aug 22, 2024 at 9:20 AM Ahmed Akef <aemed.akef.1@gmail.com> wrote:

Show 113 quoted lines

>
> > * "as they have been read from stdin"; drop "from stdin" here, as
> >   we already know we are talking about the mode that reads object
> >   names from the standard input and there is no need to repeat it.
>
> it is needed to explain that git will not do any optimization to the order of paths
> before printing the output. I entered a discussion with someone who was worried
> that git may optimize the paths order because it is not stated clearly that output follows
> the same order as input.
>
> > * "considered as an object" -> "considered to be an object name" or
> >   "used as an object name" [*].  This primarily comes from my
> >   spinal reflex against "consider as", plus my desire to be more
> >   precise in terminology.
>
> this is not related to my changes, I just moved this line but didn't changed it.
>
> Thanks,
> ahmed akef
>
>
> On Thu, Aug 22, 2024 at 9:18 AM Ahmed Akef <aemed.akef.1@gmail.com> wrote:
>>
>> > * "as they have been read from stdin"; drop "from stdin" here, as
>> >   we already know we are talking about the mode that reads object
>> >   names from the standard input and there is no need to repeat it.
>>
>> it is needed to explain that git will not do any optimization to the order of paths
>> before printing the output. I entered a discussion with someone who was worried
>> that git may optimize the paths order because it is not stated clearly that output follows
>> the same order as input.
>>
>> > * "considered as an object" -> "considered to be an object name" or
>> >   "used as an object name" [*].  This primarily comes from my
>> >   spinal reflex against "consider as", plus my desire to be more
>> >   precise in terminology.
>>
>> this is not related to my changes, I just moved this line but didn't changed it.
>>
>> Thanks,
>> ahmed akef
>>
>>
>> On Wed, Aug 21, 2024 at 7:19 PM Junio C Hamano <gitster@pobox.com> wrote:
>>>
>>> "ahmed akef via GitGitGadget" <gitgitgadget@gmail.com> writes:
>>>
>>> >  If `--batch` or `--batch-check` is given, `cat-file` will read objects
>>> > -from stdin, one per line, and print information about them. By default,
>>> > -the whole line is considered as an object, as if it were fed to
>>> > -linkgit:git-rev-parse[1].
>>> > +from stdin, one per line, and print information about them in the same
>>> > +order as they have been read from stdin. By default, the whole line is
>>> > +considered as an object, as if it were fed to linkgit:git-rev-parse[1].
>>>
>>> A few "Huh?" I had while reading the above.
>>>
>>>  * "as they have been read from stdin"; drop "from stdin" here, as
>>>    we already know we are talking about the mode that reads object
>>>    names from the standard input and there is no need to repeat it.
>>>
>>>  * "considered as an object" -> "considered to be an object name" or
>>>    "used as an object name" [*].  This primarily comes from my
>>>    spinal reflex against "consider as", plus my desire to be more
>>>    precise in terminology.
>>>
>>> Thanks.
>>>
>>> Nothing mentioned below should be part of this patch, but as I
>>> noticed it while studying the current documentation to prepare this
>>> review, I'll record them as #leftoverbits.
>>>
>>> The description of how lines read from the standard input should
>>> look like needs more work.  Documentation on "--batch" says "the
>>> input lines must specify the path, separated by whitespace", but is
>>> it clear that it expects "<object name>", followed by a whitespace
>>> (not necessarily a single SP), followed by "<path>"?  Without prior
>>> knowledge, I wouldn't be surprised if somebody read the text as
>>> asking for paths separated by whitespace, e.g.
>>>
>>>     README.txt COPYING Makefile
>>>
>>> that gives three paths.  The text needs to be tightened to say
>>> something like "must give the path after the object name, separated
>>> by whitespace.  The path is used to find the textconv and smudge
>>> filter".
>>>
>>> The section also says "See the section BATCH OUTPUT below for
>>> details." but the section it refers to does not talk anything about
>>> this whitespace thing.  It also is unclear what would happen if the
>>> input line says:
>>>
>>>     :COPYING Makefile
>>>
>>> Would it apply the textconv/filters meant for Makefile to the blob
>>> stored at COPYING in the index?  If we say
>>>
>>>     :README.txt
>>>
>>> would the command be smart enough to know that the blob came from
>>> the path README.txt and apply the textconv/filters meant for that
>>> path, without the input repeating the same information twice like:
>>>
>>>     :README.txt README.txt
>>>
>>> or something silly like that?
>>>
>>>
>>> [Reference]
>>>
>>> * https://www.britannica.com/dictionary/eb/qa/consider-and-consider-as
>>>
>>>

Junio C Hamano· Aug 22, 2024, 15:07 UTC · re: Ahmed Akef · lore

Re: [PATCH v2] docs: explain the order of output in the batched mode of git-cat-file(1)

Ahmed Akef <aemed.akef.1@gmail.com> writes:

Administrivia: please do not break the discussion thread by dropping
In-Reply-To: and / or References: headers.  For those who are
following from sidelines, this is a reponse to

    https://lore.kernel.org/git/xmqqa5h5ztd9.fsf@gitster.g/

Show 7 quoted lines

>> * "as they have been read from stdin"; drop "from stdin" here, as
>>   we already know we are talking about the mode that reads object
>>   names from the standard input and there is no need to repeat it.
>
> it is needed to explain that git will not do any optimization to the
> order of paths
> before printing the output.

I do not think "from stdin" is necessary for that. The sentence begins ...

>  If `--batch` or `--batch-check` is given, `cat-file` will read objects
> +from stdin, one per line, and print information about them in the same
> +order as they have been read from stdin. By default, the whole line is
> +considered as an object, as if it were fed to linkgit:git-rev-parse[1].

... by explaining that the command reads from the standard input, and does something to each in the same order as they were read. If you already said you are reading from the standard input, the order you read them is the order youread them from the standard input.

Drop the "from stdin" from "as they have been read from stdin".