From: Phillip Wood <phillip.wood123@gmail.com>
Date: Wed, 17 Apr 2019 15:04:19 GMT
Subject: Re: [PATCH v2 1/2] commit/reset: try to clean up sequencer state
Message-ID: <f1371b89-8f8b-d0b9-7bda-23f8cb5d0541@gmail.com>
In-Reply-To: <nycvar.QRO.7.76.6.1904171423370.46@tvgsbejvaqbjf.bet>

Hi Dscho

On 17/04/2019 13:26, Johannes Schindelin wrote:
> Hi,
> 
> On Wed, 17 Apr 2019, Junio C Hamano wrote:
> 
>> Phillip Wood <phillip.wood123@gmail.com> writes:
>>
>>> Avoid this potential problem by removing the sequencer state if we're
>>> committing or resetting the final pick in a sequence.
>>
>> The use-case story before this conclusion only mentioned "commit"
>> that concluded the multi-step cherry-pick/revert, and never talked
>> about "reset", which made my eyebrows to rise.
>>
>> As a part of "reset", we have already been removing CHERRY_PICK_HEAD
>> and REVERT_HEAD, so "git reset" during a conflicted "cherry-pick"
>> for example is already destructive and the user can no longer get
>> back to continuing the cherry-pick anyway after running it, even
>> without this patch.  So from that point of view, it does make sense
>> to remove the other sequencer states at the same time.
> 
> Do you mean to say that a `git reset` during `git cherry-pick <range>`
> aborts it?

No I mean it removes CHERRY_PICK_HEAD/REVERT_HEAD and so cancels the 
conflicting pick/revert, it does not abort the operation as a whole. If 
the conflicting pick/revert was the last in a range then we want it to 
remove .git/sequencer as well as the ..._HEAD file as it is easy to 
forget to run --continue in that case.

Best Wishes

Phillip

> In my experience, this is not the case. The advice printed out after a
> conflict even recommends to run `git reset` (followed by `git cherry-pick
> --continue`, in lieu of the `git cherry-pick --skip` we have yet to
> implement).
> 
> So I don't think it is correct to say that `git reset` does not let the
> user get back to continuing a cherry-pick...
> 
> Ciao,
> Dscho
>