Re: [PATCH RFC v1] stash: implement '--staged' option for 'push' and 'save'

From: Sergey Organov <sorganov@gmail.com>
To: Junio C Hamano <gitster@pobox.com>
Cc: "Ævar Arnfjörð Bjarmason" <avarab@gmail.com>,
	"Eric Sunshine" <sunshine@sunshineco.com>,
	"Git List" <git@vger.kernel.org>,
	"Thomas Rast" <tr@thomasrast.ch>,
	"Denton Liu" <liu.denton@gmail.com>
Subject: Re: [PATCH RFC v1] stash: implement '--staged' option for 'push' and 'save'
Date: Tue, 12 Oct 2021 21:25:52 +0300	[thread overview]
Message-ID: <87lf2yqevj.fsf@osv.gnss.ru> (raw)
In-Reply-To: <xmqqzgre40ff.fsf@gitster.g> (Junio C. Hamano's message of "Tue, 12 Oct 2021 10:28:52 -0700")

Junio C Hamano <gitster@pobox.com> writes:

> Sergey Organov <sorganov@gmail.com> writes:
>
>> Junio C Hamano <gitster@pobox.com> writes:
>>
>>> Junio C Hamano <gitster@pobox.com> writes:
>>>
>>>> More importantly...
>>>>
>>>> Whenever I think about a new "feature", I try to come up with a
>>>> story in which the feature effectively improves the end-user's life,
>>>> how it fits in the larger picture, and enables something that is
>>>> hard to do by combining other tools.
>>>>
>>>> The kind of "story" I would aim for is like this.  Suppose we were
>>>> selling not "git stash -S" but "git stash -k". ...
>>>
>>
>> [...]
>>
>>> So in short, I do not think I am strongly opposed to "git stash -S"
>>> existing, since I did find one use case story that it could be used,
>>> but I do think it is redundant and unnecessary.
>>
>> Redundant? Yes. Unnecessary? Yes. Useful? Yes. ;-)
>>
>> I took the steps to propose the new feature after yet another round of
>> "how do I quickly store this tiny bit of changes I just figured I need
>> for later, out of bunch of VIWIP changes?"
>>
>>   git stash --staged
>>
>> is exactly the (currently missing) answer for me, as I have pretty
>> interactive tool to stage diff chunks always handy.
>>
>> What's your answer, I wonder?
>
> I am the one who questions the usefulness of "stash --staged" and
> thinks "add -p", "stash -k", test, "commit" is a much better way to
> solve the "we have a messy working tree and we want to create a
> clean multi-step end result out of it" problem

I don't want to create a multi-step result out of it, if it means a
series of commits. The question is about a change that is *unrelated* to
the series I'm supposedly doing.

>
> I consider "stash --staged" as a solution in search of a problem, so
> you'd need to ask somebody else for a problem that "stash --staged"
> is suitable for.

I didn't ask you what --staged is suitable for, sorry. I asked how do
you solve the problem of saving an *entirely unrelated* subset of
changes for future use?

If the answer is "I don't have such problem", it's OK with me, but my
point is that I, and at least a few others, seem to have such a problem
frequently enough to justify introduction of the --staged option.

>
> And "I want to stash away this tiny bit" is better solved by *not*
> doing "git add" it to the index and then stashing.  Rather, I'd just
> do "commit" so that I can "rebase -i" to reorganize these bits
> later.  Of course, to test the "tiny bit" standalone, I may use
> "stash -k" first, but do not see such a senario shows the merit of
> using "stash --staged" over other tools.

That is a good solution for *different* problem. The changes I want to
stash-out supposedly don't belong to the series of changes currently
being worked on *at all*, and I don't want to test them right now as I'm
working on entirely unrelated set of problems and don't want to get
side-tracked.

So, the analog here is not using "stage -k"->test->commit cycle, it's
rather temporary switching to another branch and committing there, like
this:

 <hack, hack, hack...>
 <notice unrelated problem, give it a quick fix and stage it>
 $ git checkout -b tmp-fix-bla-bla
 $ git commit -m "Will have to look at bla-bla later"
 $ git checkout -
 <hack continues, probably using stash -k and rebasing as needed>
 <... time passes... >
 $ git switch some-branch
 $ git cherry-pick -n tmp-fix-bla-bla
 <... continue to work on the bla-bla fix ...>

See? But now, we already have such a wonderful place for temporary
states called "stash". Why should it be so hard to "commit" right to the
stash instead of stomping around and then house-keeping of these
temporary non-branches? That's what "stash --staged" is suitable for,
not for creating clean sequence of commits out of a mess, where "stash
-k" indeed shines.

>
>> That said, I'm also curious what story, if any, do you have for 'git
>> stash --patch', as exactly the same story should be applicable to
>> proposed 'git stash --staged', as far as I can see.
>
> "stash --patch" is also "Meh" from my point of view.  I do not
> strongly object to its existence, it may be a OK tool for a small
> scale use, but I suspect it would be more frustrating than helpful
> to users when applied in a larger workflow story, just like I view
> "git stash --staged".

I see, thank you for clarification.

Thanks,
-- Sergey Organov