git@vger.kernel.org mailing list mirror (one of many)
 help / color / mirror / code / Atom feed
From: Junio C Hamano <gitster@pobox.com>
To: "\"\"徐沛文  via GitGitGadget" <gitgitgadget@gmail.com>
Cc: git@vger.kernel.org, "René Scharfe" <l.s.r@web.de>,
	"Phillip Wood" <phillip.wood123@gmail.com>,
	"Johannes Schindelin" <Johannes.Schindelin@gmx.de>,
	"Elijah Newren" <newren@gmail.com>,
	"Aleen 徐沛文" <pwxu@coremail.cn>, Aleen <aleen42@vip.qq.com>
Subject: Re: [PATCH v18 3/3] am: support --allow-empty to record specific empty patches
Date: Tue, 07 Dec 2021 22:22:21 -0800	[thread overview]
Message-ID: <xmqqmtlb4nvm.fsf@gitster.g> (raw)
In-Reply-To: <4c3077f938435508850727e05ad514035f09bebb.1638939946.git.gitgitgadget@gmail.com> ("\"\"徐沛文 (Aleen)\" via GitGitGadget\""'s message of "Wed, 08 Dec 2021 05:05:46 +0000")

""徐沛文 (Aleen)" via GitGitGadget"  <gitgitgadget@gmail.com>
writes:

> @@ -1152,6 +1152,12 @@ static void NORETURN die_user_resolve(const struct am_state *state)
>  
>  		printf_ln(_("When you have resolved this problem, run \"%s --continue\"."), cmdline);
>  		printf_ln(_("If you prefer to skip this patch, run \"%s --skip\" instead."), cmdline);
> +
> +		if (advice_enabled(ADVICE_AM_WORK_DIR) &&
> +		    is_empty_or_missing_file(am_path(state, "patch")) &&
> +		    !repo_index_has_changes(the_repository, NULL, NULL))
> +			printf_ln(_("To record the empty patch as an empty commit, run \"%s --allow-empty\"."), cmdline);
> +

OK, so the idea is when we give back control to the user, if the
reason we stopped is because we did not have a usable patch in the
message and we didn't have any change to the index, we can offer
"--allow-empty" as an extra choice.

I guess that makes sense.

> @@ -1900,19 +1906,31 @@ next:
>  /**
>   * Resume the current am session after patch application failure. The user did
>   * all the hard work, and we do not have to do any patch application. Just
> - * trust and commit what the user has in the index and working tree.
> + * trust and commit what the user has in the index and working tree. If `allow_empty`
> + * is true, commit as an empty commit when index has not changed and lacking a patch.
>   */
> -static void am_resolve(struct am_state *state)
> +static void am_resolve(struct am_state *state, int allow_empty)
>  {
> +	int index_changed;
> +
>  	validate_resume_state(state);
>  
>  	say(state, stdout, _("Applying: %.*s"), linelen(state->msg), state->msg);
>  
> -	if (!repo_index_has_changes(the_repository, NULL, NULL)) {
> -		printf_ln(_("No changes - did you forget to use 'git add'?\n"
> -			"If there is nothing left to stage, chances are that something else\n"
> -			"already introduced the same changes; you might want to skip this patch."));


> +	index_changed = repo_index_has_changes(the_repository, NULL, NULL);
> +	if (allow_empty &&
> +	    !(!index_changed && is_empty_or_missing_file(am_path(state, "patch"))))
>  		die_user_resolve(state);

Why do we need this?  

Intuitively, because "--allow-XYZ" is "we normally die when the
condition XYZ holds, but with the option, we do not die in such a
case", any new condition that leads to "die" that only kicks in
when "--allow-XYZ" is given smell like a BUG.

The condition reads:

    unless we saw an empty patch (i.e. "patch" file is missing or
    empty, and did not result in any change to the index), it is an
    error to give "--allow-empty" to the command.

That somehow does not make any sense to me.  Whether failed patch
application and manual tweaking resulted in the same tree as HEAD,
or a tree different from HEAD, if the user wants to say "I allow
Git to create an empty commit as necessary, instead of stopping",
shouldn't the user be allowed to?  After all, the option is not
"create nothing but an empty commit, it is an error if there is
something to commit."  It merely allows creation of an empty one.

This series has been trying to add code to punish users who give
"--allow-empty" when the command would have worked the same way
without it at least since the last round, and I am not sure where
that wish to stop users comes from.  Please explain.  I am starting
to think I may be missing something and this extra rigidity may be
helping a user (but again, I do not see how).

> +	if (!index_changed) {
> +		if (allow_empty) {

The index_changed variable being false is "the result is an empty
change", and is-empty-or-missing on "patch" file is "the input is an
empty change".  Ideally we would want to create an empty commit only
when both input and result are empty, but in order to prevent mistakes,
we may want to react to the case where the result is empty but the
input is not empty.

Here is where we can use is-empty-or-missing on "patch" and give a
different message (i.e. "even though the patch file is not empty,
the result of your patch application did not change anything"), if
we wanted to.  Or you could choose to reject such an attempt, which
might be simpler.  I.e.

		if (allow_empty &&
		    is_empty_or_missing_file(am_path(state, "patch"))
			"No changes - recorded an empty commit."
		else
			... the original error for no changes ...


Hmm?

  reply	other threads:[~2021-12-08  6:22 UTC|newest]

Thread overview: 129+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-11-12  4:58 [PATCH 0/2] am: support --always option to am empty commits Aleen via GitGitGadget
2021-11-12  4:58 ` [PATCH 1/2] doc: git-format-patch: specify the option --always Aleen via GitGitGadget
2021-11-12  4:58 ` [PATCH 2/2] am: support --always option to am empty commits Aleen via GitGitGadget
2021-11-12  6:17 ` [PATCH 0/2] " René Scharfe
2021-11-12  6:42   ` Aleen
2021-11-12  6:47   ` Junio C Hamano
2021-11-12  7:10     ` Aleen 徐沛文
2021-11-12 15:28     ` René Scharfe
2021-11-12 16:08       ` Junio C Hamano
2021-11-12  6:53 ` [PATCH v2 0/4] am: support --allow-empty " Aleen via GitGitGadget
2021-11-12  6:53   ` [PATCH v2 1/4] doc: git-format-patch: specify the option --always Aleen via GitGitGadget
2021-11-12 22:17     ` Junio C Hamano
2021-11-12  6:53   ` [PATCH v2 2/4] am: support --always option to am empty commits Aleen via GitGitGadget
2021-11-12 22:23     ` Junio C Hamano
2021-11-12  6:53   ` [PATCH v2 3/4] test: am: add the case when not passing the --always option Aleen via GitGitGadget
2021-11-12  6:54   ` [PATCH v2 4/4] chore: am: rename the --always option to --allow-empty Aleen via GitGitGadget
2021-11-15 10:39   ` [PATCH v3 0/2] am: support --empty-commit=(die|skip|asis) option to am empty commits Aleen via GitGitGadget
2021-11-15 10:39     ` [PATCH v3 1/2] doc: git-format-patch: describe the option --always Aleen via GitGitGadget
2021-11-15 10:39     ` [PATCH v3 2/2] am: support --empty-commit option to handle empty patches Aleen via GitGitGadget
2021-11-15 11:13       ` Aleen 徐沛文
2021-11-16  5:18     ` [PATCH v4 0/2] am: support --empty-commit=(die|skip|asis) option to am empty commits Aleen via GitGitGadget
2021-11-16  5:18       ` [PATCH v4 1/2] doc: git-format-patch: describe the option --always Aleen via GitGitGadget
2021-11-16  5:18       ` [PATCH v4 2/2] am: support --empty-commit option to handle empty patches Aleen via GitGitGadget
2021-11-16 10:07         ` Phillip Wood
2021-11-16 10:31           ` Aleen 徐沛文
2021-11-17  8:39           ` Junio C Hamano
2021-11-17  9:33       ` [PATCH v5 0/2] am: support --empty-commit=(die|skip|asis) option to am empty commits Aleen via GitGitGadget
2021-11-17  9:33         ` [PATCH v5 1/2] doc: git-format-patch: describe the option --always Aleen via GitGitGadget
2021-11-17  9:33         ` [PATCH v5 2/2] am: support --empty option to handle empty patches Aleen via GitGitGadget
2021-11-18 10:50         ` [PATCH v6 0/3] am: support --empty=(die|drop|keep) " Aleen via GitGitGadget
2021-11-18 10:50           ` [PATCH v6 1/3] doc: git-format-patch: describe the option --always Aleen via GitGitGadget
2021-11-18 10:50           ` [PATCH v6 2/3] am: support --empty option to handle empty patches Aleen via GitGitGadget
2021-11-19  0:56             ` Junio C Hamano
2021-11-19 10:33             ` Bagas Sanjaya
2021-11-19 12:10               ` Ævar Arnfjörð Bjarmason
2021-11-19 12:20                 ` Eric Sunshine
2021-11-19 16:49                   ` Junio C Hamano
2021-11-19 16:46               ` Junio C Hamano
2021-11-18 10:50           ` [PATCH v6 3/3] am: throw an error when passing --empty option without value Aleen via GitGitGadget
2021-11-19  1:13             ` Junio C Hamano
2021-11-19  2:11               ` Aleen 徐沛文
2021-11-18 23:47           ` [PATCH v6 0/3] am: support --empty=(die|drop|keep) option to handle empty patches Junio C Hamano
2021-11-19  1:45             ` Aleen 徐沛文
2021-11-19  5:46               ` Junio C Hamano
2021-11-19  7:23                 ` Aleen 徐沛文
2021-11-19  7:25                   ` =?gb18030?B?QWxlZW4=?=
2021-11-19 16:54                   ` Junio C Hamano
2021-11-19 17:14                     ` Aleen 徐沛文
2021-11-19 19:25                       ` Junio C Hamano
2021-11-22 11:57                     ` Johannes Schindelin
2021-11-19  4:16             ` Aleen 徐沛文
2021-11-19  5:04           ` [PATCH v7 0/2] " Aleen via GitGitGadget
2021-11-19  5:04             ` [PATCH v7 1/2] doc: git-format-patch: describe the option --always Aleen via GitGitGadget
2021-11-19  5:04             ` [PATCH v7 2/2] am: support --empty=<option> to handle empty patches Aleen via GitGitGadget
2021-11-19 22:50               ` Junio C Hamano
2021-11-19 23:07               ` Junio C Hamano
2021-11-22  6:46             ` [PATCH v8 0/2] am: support --empty=(die|drop|keep) option " Aleen via GitGitGadget
2021-11-22  6:46               ` [PATCH v8 1/2] doc: git-format-patch: describe the option --always Aleen via GitGitGadget
2021-11-22  6:46               ` [PATCH v8 2/2] am: support --empty=<option> to handle empty patches Aleen via GitGitGadget
2021-11-22  7:06                 ` Junio C Hamano
2021-11-22  7:19                   ` Aleen 徐沛文
2021-11-22  7:02               ` [PATCH v9 0/2] am: support --empty=(die|drop|keep) option " Aleen via GitGitGadget
2021-11-22  7:02                 ` [PATCH v9 1/2] doc: git-format-patch: describe the option --always Aleen 徐沛文 via GitGitGadget
2021-11-22  7:02                 ` [PATCH v9 2/2] am: support --empty=<option> to handle empty patches Aleen 徐沛文 via GitGitGadget
2021-11-22  7:04                   ` Aleen 徐沛文
2021-11-22  7:51                 ` [PATCH v10 0/2] am: support --empty=(die|drop|keep) option " Aleen via GitGitGadget
2021-11-22  7:51                   ` [PATCH v10 1/2] doc: git-format-patch: describe the option --always Aleen via GitGitGadget
2021-11-22 12:00                     ` Johannes Schindelin
2021-11-23  1:25                       ` Aleen 徐沛文
2021-11-23 12:30                         ` Johannes Schindelin
2021-11-22  7:51                   ` [PATCH v10 2/2] am: support --empty=<option> to handle empty patches Aleen via GitGitGadget
2021-11-23 15:26                   ` [PATCH v11 0/2] am: support --empty=(die|drop|keep) option " Aleen via GitGitGadget
2021-11-23 15:26                     ` [PATCH v11 1/2] doc: git-format-patch: describe the option --always 徐沛文 (Aleen) via GitGitGadget
2021-11-23 16:12                       ` Johannes Schindelin
2021-11-23 22:02                         ` Junio C Hamano
2021-11-23 15:26                     ` [PATCH v11 2/2] am: support --empty=<option> to handle empty patches 徐沛文 (Aleen) via GitGitGadget
2021-11-26 20:14                       ` Elijah Newren
2021-11-29  9:19                         ` Aleen 徐沛文
2021-11-29 10:00                           ` Aleen 徐沛文
2021-11-29 17:10                             ` Elijah Newren
2021-11-30  5:45                               ` [PATCH v12 3/3] am: support --allow-empty to record specific " Aleen 徐沛文
2021-11-29 18:17                         ` [PATCH v11 2/2] am: support --empty=<option> to handle " Junio C Hamano
2021-11-29 18:57                           ` Elijah Newren
2021-11-30  5:37                     ` [PATCH v12 0/3] am: support --empty=(die|drop|keep) option and --allow-empty option " Aleen via GitGitGadget
2021-11-30  5:37                       ` [PATCH v12 1/3] doc: git-format-patch: describe the option --always 徐沛文 (Aleen) via GitGitGadget
2021-11-30  5:37                       ` [PATCH v12 2/3] am: support --empty=<option> to handle empty patches 徐沛文 (Aleen) via GitGitGadget
2021-11-30  5:37                       ` [PATCH v12 3/3] am: support --allow-empty to record specific " 徐沛文 (Aleen) via GitGitGadget
2021-11-30  7:21                         ` Junio C Hamano
2021-11-30  9:55                       ` [PATCH v13 0/3] am: support --empty=(die|drop|keep) option and --allow-empty option to handle " Aleen via GitGitGadget
2021-11-30  9:55                         ` [PATCH v13 1/3] doc: git-format-patch: describe the option --always 徐沛文 (Aleen) via GitGitGadget
2021-11-30  9:55                         ` [PATCH v13 2/3] am: support --empty=<option> to handle empty patches 徐沛文 (Aleen) via GitGitGadget
2021-11-30  9:55                         ` [PATCH v13 3/3] am: support --allow-empty to record specific " 徐沛文 (Aleen) via GitGitGadget
2021-12-01  3:37                         ` [PATCH v14 0/3] am: support --empty=(die|drop|keep) option and --allow-empty option to handle " Aleen via GitGitGadget
2021-12-01  3:37                           ` [PATCH v14 1/3] doc: git-format-patch: describe the option --always 徐沛文 (Aleen) via GitGitGadget
2021-12-01  3:37                           ` [PATCH v14 2/3] am: support --empty=<option> to handle empty patches 徐沛文 (Aleen) via GitGitGadget
2021-12-03 22:30                             ` Johannes Schindelin
2021-12-01  3:37                           ` [PATCH v14 3/3] am: support --allow-empty to record specific " 徐沛文 (Aleen) via GitGitGadget
2021-12-02  0:58                             ` Junio C Hamano
2021-12-06  1:35                               ` Aleen 徐沛文
2021-12-06  9:41                           ` [PATCH v15 0/3] am: support --empty=(die|drop|keep) option and --allow-empty option to handle " Aleen via GitGitGadget
2021-12-06  9:41                             ` [PATCH v15 1/3] doc: git-format-patch: describe the option --always 徐沛文 (Aleen) via GitGitGadget
2021-12-06  9:41                             ` [PATCH v15 2/3] am: support --empty=<option> to handle empty patches 徐沛文 (Aleen) via GitGitGadget
2021-12-06  9:41                             ` [PATCH v15 3/3] am: support --allow-empty to record specific " 徐沛文 (Aleen) via GitGitGadget
2021-12-07  5:01                             ` [PATCH v16 0/3] am: support --empty=(die|drop|keep) option and --allow-empty option to handle " Aleen via GitGitGadget
2021-12-07  5:01                               ` [PATCH v16 1/3] doc: git-format-patch: describe the option --always 徐沛文 (Aleen) via GitGitGadget
2021-12-07  5:01                               ` [PATCH v16 2/3] am: support --empty=<option> to handle empty patches 徐沛文 (Aleen) via GitGitGadget
2021-12-07  5:01                               ` [PATCH v16 3/3] am: support --allow-empty to record specific " 徐沛文 (Aleen) via GitGitGadget
2021-12-07  8:31                               ` [PATCH v17 0/3] am: support --empty=(die|drop|keep) option and --allow-empty option to handle " Aleen via GitGitGadget
2021-12-07  8:31                                 ` [PATCH v17 1/3] doc: git-format-patch: describe the option --always 徐沛文 (Aleen) via GitGitGadget
2021-12-07  8:31                                 ` [PATCH v17 2/3] am: support --empty=<option> to handle empty patches 徐沛文 (Aleen) via GitGitGadget
2021-12-07 18:12                                   ` Junio C Hamano
2021-12-07  8:31                                 ` [PATCH v17 3/3] am: support --allow-empty to record specific " 徐沛文 (Aleen) via GitGitGadget
2021-12-07 18:23                                   ` Junio C Hamano
2021-12-07 18:24                                 ` [PATCH v17 0/3] am: support --empty=(die|drop|keep) option and --allow-empty option to handle " Junio C Hamano
2021-12-08  5:05                                 ` [PATCH v18 " Aleen via GitGitGadget
2021-12-08  5:05                                   ` [PATCH v18 1/3] doc: git-format-patch: describe the option --always 徐沛文 (Aleen) via GitGitGadget
2021-12-08  5:05                                   ` [PATCH v18 2/3] am: support --empty=<option> to handle empty patches 徐沛文 (Aleen) via GitGitGadget
2021-12-08  5:05                                   ` [PATCH v18 3/3] am: support --allow-empty to record specific " 徐沛文 (Aleen) via GitGitGadget
2021-12-08  6:22                                     ` Junio C Hamano [this message]
2021-12-08  6:46                                       ` Aleen 徐沛文
2021-12-08 11:32                                         ` Junio C Hamano
2021-12-09  7:25                                   ` [PATCH v19 0/3] am: support --empty=(die|drop|keep) option and --allow-empty option to handle " Aleen via GitGitGadget
2021-12-09  7:25                                     ` [PATCH v19 1/3] doc: git-format-patch: describe the option --always 徐沛文 (Aleen) via GitGitGadget
2021-12-09  9:28                                       ` Bagas Sanjaya
2021-12-10  1:26                                         ` Aleen 徐沛文
2021-12-10  6:50                                           ` Bagas Sanjaya
2021-12-11  9:22                                             ` Junio C Hamano
2021-12-09  7:25                                     ` [PATCH v19 2/3] am: support --empty=<option> to handle empty patches 徐沛文 (Aleen) via GitGitGadget
2021-12-09  7:25                                     ` [PATCH v19 3/3] am: support --allow-empty to record specific " 徐沛文 (Aleen) via GitGitGadget

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: http://vger.kernel.org/majordomo-info.html

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=xmqqmtlb4nvm.fsf@gitster.g \
    --to=gitster@pobox.com \
    --cc=Johannes.Schindelin@gmx.de \
    --cc=aleen42@vip.qq.com \
    --cc=git@vger.kernel.org \
    --cc=gitgitgadget@gmail.com \
    --cc=l.s.r@web.de \
    --cc=newren@gmail.com \
    --cc=phillip.wood123@gmail.com \
    --cc=pwxu@coremail.cn \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
Code repositories for project(s) associated with this public inbox

	https://80x24.org/mirrors/git.git

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).