Re: [PATCH] submodule documentation: Reorder introductory paragraphs

[Stefan Beller <sbeller@google.com>]

On Fri, May 22, 2015 at 10:35 AM, Philip Oakley <philipoakley@iee.org> wrote:
> From: "Stefan Beller" <sbeller@google.com>
>>
>> On Fri, May 22, 2015 at 7:36 AM, Junio C Hamano <gitster@pobox.com> wrote:
>>>
>>> Stefan Beller <sbeller@google.com> writes:
>>>
>>>> On Thu, May 21, 2015 at 1:03 PM, Philip Oakley <philipoakley@iee.org>
>>>> wrote:
>>>>>>
>>>>>> +Submodules are not to be confused with remotes, which are meant
>>>>>> +mainly for branches of the same project;
>>>>>
>>>>>
>>>>> This use of 'branches' didn't work for me. "remotes are meant mainly
>>>>> for
>>>>> branches of the same project" ?
>>>
>>>
>>> The "branch" in the original is used in a much wider sense than
>>> usual branch (i.e. ref/heads/ thing you have locally); it refers to
>>> forks of the same project but with a bit of twist.  When you say
>>> repository A is a fork of the same project as my local repository,
>>> you would give an impression that A is not the authoritative copy of
>>> the project.  But you can say my repository and that repository A
>>> are branches of the same project, you give zero information as to
>>> A's authoritativeness.
>>
>>
>> While this is correct, I think it is also confusing, because 'branch'
>> is a command which deals with local branches only in my perception
>> To deal with remote branches you need to use the commands
>> {remote, fetch, pull}.
>>
>> So when someone mentions "branch" I need to think of local operations
>> in one repository and not on different distributed histories.
>>
>
> If we are having difficulties defining a "remote" here (its not defined in
> gitglossary.txt anyway),

Now that we have a discussion on what remotes are, I'll send a patch for that
as well.

> why not simply put a full stop (period) after the
> "Submodules are not to be confused with remotes.", and bypass the problem,
> avoiding digging the hole deeper.

I think we should dig deeper and point out the differences as it may
not be clear what
the differences are for new comers. Not digging deeper sounds to me like saying

    'git frotz' is not to be confused with 'git bar' FULL STOP AND I
WONT TELL YOU WHY!

which is not helpful. (Why is the documentation pointing out there is
a difference to
that other command/concept, but not saying what is different?)

>
>>>
>>>>     Submodules should not be confused with remote repositories, which
>>>> are
>>>>     meant to track the same repository, just at another location; ...
>>>
>>>
>>> I do not think this is a great improvement.  You now conflated
>>> "repository" to mean "project" in the latter half of the sentence,
>>> while you are trying to explain what a "remote repository" is.
>>
>>
>> That's true.
>>
>>>
>>> Your copy of git.git is not the same repository as mine; they have
>>> different histories.  Both repositories are used to work on the same
>>> project.  "submoules are not remotes, which are other repositories
>>> of the same project", perhaps?
>>
>>
>> That makes sense.
>>
>
> If maybe that the feature we should pick on is the common root of the
> development between the local and remote repository, and quite distinct for
> the submodule. This allows remotes to be on the same machine, as well as
> distant machines and server.

I don't think this is actually true for all remotes. Think of shallow clones
(they have no root or a different root) or even subtrees which are pulled
in via a remotes?

The main thing about remotes is "not being here" (as in "part of this
repository". As you point out it can be nearby in the local fs or even on
another machine, or in the cloud)

>
> It is I believe technically possible to have a submodule which is its own
> super project, with and without recursion, but would be very atypical, and
> would belong in the 'don't do that' category