git.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Fabian Ruch <bafain@gmail.com>
To: Eric Sunshine <sunshine@sunshineco.com>
Cc: Git List <git@vger.kernel.org>,
	Michael Haggerty <mhagger@alum.mit.edu>,
	Thomas Rast <tr@thomasrast.ch>, Jeff King <peff@peff.net>,
	Peter Krefting <peter@softwolves.pp.se>,
	Phil Hord <phil.hord@gmail.com>
Subject: Re: [PATCH v2 02/23] rebase -i: allow squashing empty commits without complaints
Date: Mon, 11 Aug 2014 09:01:48 +0200	[thread overview]
Message-ID: <53E86A5C.2050202@gmail.com> (raw)
In-Reply-To: <CAPig+cSvLwcwZ_8R0nRYst9V9mfRqc-HMDhyrCoU6cxE254Agg@mail.gmail.com>

Hi Eric,

Eric Sunshine writes:
> On Wed, Aug 6, 2014 at 7:59 PM, Fabian Ruch <bafain@gmail.com> wrote:
>> The to-do list commands `squash` and `fixup` apply the changes
>> introduced by the named commit to the tree but instead of creating
>> a new commit on top of the current head it replaces the previous
>> commit with a new commit that records the updated tree. If the
>> result is an empty commit git-rebase stops with the error message
>>
>>    You asked to amend the most recent commit, but doing so would make
>>    it empty. You can repeat your command with --allow-empty, or you can
>>    remove the commit entirely with "git reset HEAD^".
>>
>> This message is not very helpful because neither does git-rebase
>> support an option `--allow-empty` nor does the messages say how to
>> resume the rebase. Firstly, change the error message to
>>
>>    The squash result is empty and --keep-empty was not specified.
>>
>>    You can remove the squash commit now with
>>
>>      git reset HEAD^
>>
>>    Once you are down, run
> 
> I guess you meant: s/down/done
> 
> Same issue with the actually message in the code (below).

Fixed.

>>      git rebase --continue
>>
>> If the user wishes to squash a sequence of commits into one
>> commit, f. i.
>>
>>    pick A
>>    squash Revert "A"
>>    squash A'
>>
>> , it does not matter for the end result that the first squash
>> result, or any sub-sequence in general, is going to be empty. The
>> squash message is not affected at all by which commits are created
>> and only the commit created by the last line in the sequence will
>> end up in the final history. Secondly, print the error message
>> only if the whole squash sequence produced an empty commit.
>>
>> Lastly, since an empty squash commit is not a failure to rewrite
>> the history as planned, issue the message above as a mere warning
>> and interrupt the rebase with the return value zero. The
>> interruption should be considered as a notification with the
>> chance to undo it on the spot. Specifying the `--keep-empty`
>> option tells git-rebase to keep empty squash commits in the
>> rebased history without notification.
>>
>> Add tests.
>>
>> Reported-by: Peter Krefting <peter@softwolves.pp.se>
>> Signed-off-by: Fabian Ruch <bafain@gmail.com>
>> ---
>> Hi,
>>
>> Peter Krefting is cc'd as the author of the bug report "Confusing
>> error message in rebase when commit becomes empty" discussed on the
>> mailing list in June. Phil Hord and Jeff King both participated in
>> the problem discussion which ended with two proposals by Jeff.
>>
>> Jeff King writes:
>>>   1. Always keep such empty commits. A user who is surprised by them
>>>      being empty can then revisit them. Or drop them by doing another
>>>      rebase without --keep-empty.
>>>
>>>   2. Notice ourselves that the end-result of the whole squash is an
>>>      empty commit, and stop to let the user deal with it.
>>
>> This patch chooses the second alternative. Either way seems OK. The
>> crucial consensus of the discussion was to silently throw away empty
>> interim commits.
>>
>>    Fabian
>>
>>  git-rebase--interactive.sh    | 20 +++++++++++---
>>  t/t3404-rebase-interactive.sh | 62 +++++++++++++++++++++++++++++++++++++++++++
>>  2 files changed, 79 insertions(+), 3 deletions(-)
>>
>> diff --git a/git-rebase--interactive.sh b/git-rebase--interactive.sh
>> index 3222bf6..8820eac 100644
>> --- a/git-rebase--interactive.sh
>> +++ b/git-rebase--interactive.sh
>> @@ -549,7 +549,7 @@ do_next () {
>>                 squash|s|fixup|f)
>>                         # This is an intermediate commit; its message will only be
>>                         # used in case of trouble.  So use the long version:
>> -                       do_with_author output git commit --allow-empty-message \
>> +                       do_with_author output git commit --allow-empty-message --allow-empty \
>>                                 --amend --no-verify -F "$squash_msg" \
>>                                 ${gpg_sign_opt:+"$gpg_sign_opt"} ||
>>                                 die_failed_squash $sha1 "$rest"
>> @@ -558,18 +558,32 @@ do_next () {
>>                         # This is the final command of this squash/fixup group
>>                         if test -f "$fixup_msg"
>>                         then
>> -                               do_with_author git commit --allow-empty-message \
>> +                               do_with_author git commit --allow-empty-message --allow-empty \
>>                                         --amend --no-verify -F "$fixup_msg" \
>>                                         ${gpg_sign_opt:+"$gpg_sign_opt"} ||
>>                                         die_failed_squash $sha1 "$rest"
>>                         else
>>                                 cp "$squash_msg" "$GIT_DIR"/SQUASH_MSG || exit
>>                                 rm -f "$GIT_DIR"/MERGE_MSG
>> -                               do_with_author git commit --amend --no-verify -F "$GIT_DIR"/SQUASH_MSG -e \
>> +                               do_with_author git commit --allow-empty --amend --no-verify -F "$GIT_DIR"/SQUASH_MSG -e \
>>                                         ${gpg_sign_opt:+"$gpg_sign_opt"} ||
>>                                         die_failed_squash $sha1 "$rest"
>>                         fi
>>                         rm -f "$squash_msg" "$fixup_msg"
>> +                       if test -z "$keep_empty" && is_empty_commit HEAD
>> +                       then
>> +                               echo "$sha1" >"$state_dir"/stopped-sha
>> +                               warn "The squash result is empty and --keep-empty was not specified."
>> +                               warn
>> +                               warn "You can remove the squash commit now with"
>> +                               warn
>> +                               warn "  git reset HEAD^"
>> +                               warn
>> +                               warn "Once you are down, run"
> 
> s/down/done/

Thanks for the thorough reading.

   Fabian

  reply	other threads:[~2014-08-11  7:01 UTC|newest]

Thread overview: 148+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-06-19  3:28 [RFC PATCH 0/7] rebase -i: Implement `reword` and `squash` in terms of `do_pick` Fabian Ruch
2014-07-02 17:47 ` [PATCH RFC v2 00/19] Enable options --signoff, --reset-author for pick, reword Fabian Ruch
2014-07-02 17:47   ` [PATCH RFC v2 01/19] rebase -i: Failed reword prints redundant error message Fabian Ruch
2014-07-08 20:31     ` Junio C Hamano
2014-07-10 14:30       ` Andrew Wong
2014-07-10 16:35         ` Fabian Ruch
2014-07-10 17:04           ` Andrew Wong
2014-07-02 17:47   ` [PATCH RFC v2 02/19] rebase -i: reword complains about empty commit despite --keep-empty Fabian Ruch
2014-07-08 20:37     ` Junio C Hamano
2014-07-09 18:02       ` Fabian Ruch
2014-07-02 17:47   ` [PATCH RFC v2 03/19] rebase -i: reword executes pre-commit hook on interim commit Fabian Ruch
2014-07-08 20:43     ` Junio C Hamano
2014-07-13 11:00       ` Fabian Ruch
2014-07-02 17:47   ` [PATCH RFC v2 04/19] rebase -i: Teach do_pick the option --edit Fabian Ruch
2014-07-02 17:47   ` [PATCH RFC v2 05/19] rebase -i: Implement reword in terms of do_pick Fabian Ruch
2014-08-04 15:16     ` Matthieu Moy
2014-08-04 15:45       ` Fabian Ruch
2014-07-02 17:47   ` [PATCH RFC v2 06/19] rebase -i: Stop on root commits with empty log messages Fabian Ruch
2014-07-08 22:26     ` Junio C Hamano
2014-07-10  9:29       ` Fabian Ruch
2014-07-10 16:57         ` Junio C Hamano
2014-07-10 17:33         ` Junio C Hamano
2014-07-02 17:47   ` [PATCH RFC v2 07/19] rebase -i: The replay of root commits is not shown with --verbose Fabian Ruch
2014-07-08 22:29     ` Junio C Hamano
2014-07-11 13:46     ` Fabian Ruch
2014-07-15  9:29       ` Chris Webb
2014-07-02 17:48   ` [PATCH RFC v2 08/19] rebase -i: Root commits are replayed with an unnecessary option Fabian Ruch
2014-07-08 22:32     ` Junio C Hamano
2014-07-18  9:16       ` Fabian Ruch
2014-07-18 16:52         ` Junio C Hamano
2014-07-19 18:14           ` Fabian Ruch
2014-07-02 17:48   ` [PATCH RFC v2 09/19] rebase -i: Commit only once when rewriting picks Fabian Ruch
2014-07-02 17:48   ` [PATCH RFC v2 10/19] rebase -i: Do not die in do_pick Fabian Ruch
2014-07-02 17:48   ` [PATCH RFC v2 11/19] rebase -i: Teach do_pick the option --amend Fabian Ruch
2014-07-02 17:48   ` [PATCH RFC v2 12/19] rebase -i: Teach do_pick the option --file Fabian Ruch
2014-07-02 17:48   ` [PATCH RFC v2 13/19] rebase -i: Prepare for squash in terms of do_pick --amend Fabian Ruch
2014-07-02 17:48   ` [PATCH RFC v2 14/19] rebase -i: Implement squash in terms of do_pick Fabian Ruch
2014-07-02 17:48   ` [PATCH RFC v2 15/19] rebase -i: Explicitly distinguish replay commands and exec tasks Fabian Ruch
2014-07-10 20:03     ` Junio C Hamano
2014-07-02 17:48   ` [PATCH RFC v2 16/19] rebase -i: Parse to-do list command line options Fabian Ruch
2014-07-02 17:48   ` [PATCH RFC v2 17/19] rebase -i: Teach do_pick the option --reset-author Fabian Ruch
2014-07-02 17:48   ` [PATCH RFC v2 18/19] rebase -i: Teach do_pick the option --signoff Fabian Ruch
2014-07-02 17:48   ` [PATCH RFC v2 19/19] rebase -i: Enable options --signoff, --reset-author for pick, reword Fabian Ruch
2014-07-03 10:33   ` [PATCH RFC v2 00/19] " Michael Haggerty
2014-07-08 20:45   ` Junio C Hamano
2014-07-09 16:08     ` Fabian Ruch
2014-07-18 12:10       ` Thomas Rast
2014-07-28 23:18 ` [PATCH v1 " Fabian Ruch
2014-07-28 23:18   ` [PATCH v1 01/19] rebase -i: failed reword prints redundant error message Fabian Ruch
2014-07-28 23:18   ` [PATCH v1 02/19] rebase -i: allow rewording an empty commit without complaints Fabian Ruch
2014-07-28 23:18   ` [PATCH v1 03/19] rebase -i: reword executes pre-commit hook on interim commit Fabian Ruch
2014-08-01 23:47     ` Jeff King
2014-08-04 18:51       ` Fabian Ruch
2014-08-06 21:46         ` Jeff King
2014-07-28 23:18   ` [PATCH v1 04/19] rebase -i: teach do_pick the option --edit Fabian Ruch
2014-07-28 23:18   ` [PATCH v1 05/19] rebase -i: implement reword in terms of do_pick Fabian Ruch
2014-07-28 23:18   ` [PATCH v1 06/19] rebase -i: allow replaying commits with empty log messages Fabian Ruch
2014-07-28 23:18   ` [PATCH v1 07/19] rebase -i: log the replay of root commits Fabian Ruch
2014-08-02  0:04     ` Jeff King
2014-08-04 21:21       ` Fabian Ruch
2014-08-06 22:01         ` Jeff King
2014-07-28 23:18   ` [PATCH v1 08/19] rebase -i: root commits are replayed with an unnecessary option Fabian Ruch
2014-08-02  0:13     ` Jeff King
2014-08-04 21:31       ` Fabian Ruch
2014-07-28 23:18   ` [PATCH v1 09/19] rebase -i: commit only once when rewriting picks Fabian Ruch
2014-08-02  0:22     ` Jeff King
2014-08-07  0:24       ` Fabian Ruch
2014-07-28 23:18   ` [PATCH v1 10/19] rebase -i: do not die in do_pick Fabian Ruch
2014-07-28 23:18   ` [PATCH v1 11/19] rebase -i: teach do_pick the option --amend Fabian Ruch
2014-07-28 23:18   ` [PATCH v1 12/19] rebase -i: teach do_pick the option --file Fabian Ruch
2014-07-28 23:18   ` [PATCH v1 13/19] rebase -i: prepare for squash in terms of do_pick --amend Fabian Ruch
2014-07-28 23:18   ` [PATCH v1 14/19] rebase -i: implement squash in terms of do_pick Fabian Ruch
2014-07-28 23:18   ` [PATCH v1 15/19] rebase -i: explicitly distinguish replay commands and exec tasks Fabian Ruch
2014-07-28 23:18   ` [PATCH v1 16/19] rebase -i: parse to-do list command line options Fabian Ruch
2014-07-28 23:18   ` [PATCH v1 17/19] rebase -i: teach do_pick the option --reset-author Fabian Ruch
2014-07-28 23:18   ` [PATCH v1 18/19] rebase -i: teach do_pick the option --signoff Fabian Ruch
2014-07-28 23:18   ` [PATCH v1 19/19] rebase -i: enable options --signoff, --reset-author for pick, reword Fabian Ruch
2014-08-02 13:52   ` [PATCH v1 00/19] Enable " Jeff King
2014-08-04  8:37     ` Fabian Ruch
2014-08-06 23:59 ` [PATCH v2 00/23] " Fabian Ruch
2014-08-06 23:59   ` [PATCH v2 01/23] rebase -i: allow replaying commits with empty log messages Fabian Ruch
2014-08-06 23:59   ` [PATCH v2 02/23] rebase -i: allow squashing empty commits without complaints Fabian Ruch
2014-08-07  7:16     ` Peter Krefting
2014-08-07 22:03     ` Eric Sunshine
2014-08-11  7:01       ` Fabian Ruch [this message]
2014-08-13 19:24     ` Phil Hord
2014-08-06 23:59   ` [PATCH v2 03/23] rebase -i: allow rewording " Fabian Ruch
2014-08-06 23:59   ` [PATCH v2 04/23] rebase -i: hide interactive command messages in verbose mode Fabian Ruch
2014-08-08 19:09     ` Thomas Rast
2014-08-11  8:26       ` Fabian Ruch
2014-08-11 18:22         ` Thomas Rast
2014-08-06 23:59   ` [PATCH v2 05/23] rebase -i: failed reword prints redundant error message Fabian Ruch
2014-08-06 23:59   ` [PATCH v2 06/23] commit: allow disabling pre-commit and commit-msg separately Fabian Ruch
2014-08-06 23:59   ` [PATCH v2 07/23] rebase -i: squash skips commit-msg hook Fabian Ruch
2014-08-06 23:59   ` [PATCH v2 08/23] rebase -i: reword executes pre-commit hook on interim commit Fabian Ruch
2014-08-08 19:09     ` Thomas Rast
2014-08-11  8:45       ` Fabian Ruch
2014-08-11 18:22         ` Thomas Rast
2014-08-06 23:59   ` [PATCH v2 09/23] rebase -i: teach do_pick the option --edit Fabian Ruch
2014-08-06 23:59   ` [PATCH v2 10/23] rebase -i: implement reword in terms of do_pick Fabian Ruch
2014-08-06 23:59   ` [PATCH v2 11/23] rebase -i: log the replay of root commits Fabian Ruch
2014-08-06 23:59   ` [PATCH v2 12/23] rebase -i: root commits are replayed with an unnecessary option Fabian Ruch
2014-08-06 23:59   ` [PATCH v2 13/23] rebase -i: commit only once when rewriting picks Fabian Ruch
2014-08-06 23:59   ` [PATCH v2 14/23] rebase -i: do not die in do_pick Fabian Ruch
2014-08-06 23:59   ` [PATCH v2 15/23] rebase -i: teach do_pick the option --amend Fabian Ruch
2014-08-06 23:59   ` [PATCH v2 16/23] rebase -i: teach do_pick the option --file Fabian Ruch
2014-08-06 23:59   ` [PATCH v2 17/23] rebase -i: prepare for squash in terms of do_pick --amend Fabian Ruch
2014-08-06 23:59   ` [PATCH v2 18/23] rebase -i: implement squash in terms of do_pick Fabian Ruch
2014-08-06 23:59   ` [PATCH v2 19/23] rebase -i: explicitly distinguish replay commands and exec tasks Fabian Ruch
2014-08-06 23:59   ` [PATCH v2 20/23] rebase -i: parse to-do list command line options Fabian Ruch
2014-08-08 19:10     ` Thomas Rast
2014-08-11 20:56       ` Fabian Ruch
2014-08-06 23:59   ` [PATCH v2 21/23] rebase -i: teach do_pick the option --reset-author Fabian Ruch
2014-08-06 23:59   ` [PATCH v2 22/23] rebase -i: teach do_pick the option --signoff Fabian Ruch
2014-08-06 23:59   ` [PATCH v2 23/23] rebase -i: enable options --signoff, --reset-author for pick, reword Fabian Ruch
2014-08-08 19:10     ` Thomas Rast
2014-08-12 21:04       ` Fabian Ruch
2014-08-13 12:47     ` Michael Haggerty
2014-08-14 17:24       ` Fabian Ruch
2014-09-21 16:59       ` Fabian Ruch
2014-08-18 21:22 ` [PATCH v3 00/27] Enable options --signoff, --reset-author for pick, reword, edit Fabian Ruch
2014-08-18 21:22   ` [PATCH v3 01/27] rebase -i: allow replaying commits with empty log messages Fabian Ruch
2014-08-18 21:22   ` [PATCH v3 02/27] rebase -i: allow squashing empty commits without complaints Fabian Ruch
2014-08-18 21:22   ` [PATCH v3 03/27] rebase -i: allow rewording " Fabian Ruch
2014-08-18 21:22   ` [PATCH v3 04/27] fake_editor: leave standard output unchanged Fabian Ruch
2014-08-18 21:22   ` [PATCH v3 05/27] rebase -i: hide interactive command messages in verbose mode Fabian Ruch
2014-08-18 21:22   ` [PATCH v3 06/27] rebase -i: discard redundant message when rewording fails Fabian Ruch
2014-08-18 21:22   ` [PATCH v3 07/27] commit: allow disabling pre-commit and commit-msg separately Fabian Ruch
2014-08-18 21:22   ` [PATCH v3 08/27] rebase -i: verify squash messages using commit-msg Fabian Ruch
2014-08-18 21:22   ` [PATCH v3 09/27] rebase -i: do not verify reworded patches using pre-commit Fabian Ruch
2014-08-18 21:22   ` [PATCH v3 10/27] rebase -i: teach do_pick the option --edit Fabian Ruch
2014-08-18 21:22   ` [PATCH v3 11/27] rebase -i: implement reword in terms of do_pick Fabian Ruch
2014-08-18 21:22   ` [PATCH v3 12/27] rebase -i: log the replay of root commits Fabian Ruch
2014-08-18 21:22   ` [PATCH v3 13/27] rebase -i: do not use -C when --no-edit is sufficient Fabian Ruch
2014-08-18 21:22   ` [PATCH v3 14/27] rebase -i: commit only once when rewriting picks Fabian Ruch
2014-08-18 21:22   ` [PATCH v3 15/27] rebase -i: do not die in do_pick Fabian Ruch
2014-08-18 21:22   ` [PATCH v3 16/27] rebase -i: teach do_pick the option --amend Fabian Ruch
2014-08-18 21:23   ` [PATCH v3 17/27] rebase -i: teach do_pick the option --file Fabian Ruch
2014-08-18 21:23   ` [PATCH v3 18/27] rebase -i: remove no-op do_with_author git commit --amend Fabian Ruch
2014-08-18 21:23   ` [PATCH v3 19/27] rebase -i: prepare for squash in terms of do_pick --amend Fabian Ruch
2014-08-18 21:23   ` [PATCH v3 20/27] rebase -i: implement squash in terms of do_pick Fabian Ruch
2014-08-18 21:23   ` [PATCH v3 21/27] rebase -i: explicitly distinguish replay commands and exec tasks Fabian Ruch
2014-08-18 21:23   ` [PATCH v3 22/27] rebase -i: parse to-do list command line options Fabian Ruch
2014-08-18 21:23   ` [PATCH v3 23/27] rebase -i: teach do_pick the option --reset-author Fabian Ruch
2014-08-18 21:23   ` [PATCH v3 24/27] rebase -i: teach do_pick the option --signoff Fabian Ruch
2014-08-18 21:23   ` [PATCH v3 25/27] rebase -i: do not overwrite user author information Fabian Ruch
2014-08-18 21:23   ` [PATCH v3 26/27] rebase -i: refuse to commit when resuming with updated head Fabian Ruch
2014-08-18 21:23   ` [PATCH v3 27/27] rebase -i: enable --signoff, --reset-author for pick, reword, edit Fabian Ruch

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

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

  git send-email \
    --in-reply-to=53E86A5C.2050202@gmail.com \
    --to=bafain@gmail.com \
    --cc=git@vger.kernel.org \
    --cc=mhagger@alum.mit.edu \
    --cc=peff@peff.net \
    --cc=peter@softwolves.pp.se \
    --cc=phil.hord@gmail.com \
    --cc=sunshine@sunshineco.com \
    --cc=tr@thomasrast.ch \
    /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.
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 NNTP newsgroup(s).