Re: [PATCH v4 1/1] push: make 'set-upstream' have dafault arguments

[Abhradeep Chakraborty <chakrabortyabhradeep79@gmail.com>]

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

> That means if the user says push.default==nothing, we should error
> out "git push -u" as before, but that is not what the change to
> setup_default_push_refspecs() function does, is it?

Yeah. You're right. The current change does not throw error for
push.default=nothing. Because I thought that for all values of
`push.default` (except matching), 'git push -u' should create a
new branch with the same name as the current local branch. Now, It
seems that I was wrong.

> So, I am not sure if many of the above changes are sensible.  The
> first one certainly does not sound like sensible.

Actually, I didn't think deeply while commiting the changes. Today,
I think about it deeply and I realized the following points.

* if push.default='simple' or unspecified then it should not create
  a new branch on the remote. So, my proposed change of 'git push -u'
  for push.default='simple' is badly affecting the reason why
  push.default='simple' was built for.

* if push.default='nothing', It should throw error if no <refspec> is
  provided. Again, my proposed change is hurting it.

* For push.default=upstream, If an upstream is already defined then
  'git push -u' should only set that branch as the upstream of the
  local branch. This already works in git. But if an upstream is not
  provied, it should throw error. So, I am not sure whether 'git push
  -u' (with no upstream information) should create a new branch with
  the same name or not. What do you think about that?

* For push.default=matching, 'git push -u' should set all the existing
  matching branches as upstream of their respective matching local
  branches. It also already works. Same for 'push.default'=current also.

So, to put all in a nutshell, I think that the current behaviour of
'git push -u' is okay. It also seems that he/she who built the
setup_default_push_refspecs() was aware of this.

Sorry for the patch request and thanks for reviewing.

> Doesn't $i show in the output as-is here?  Quote the test title in
> double-quotes, while using single-qoutes around the test body.

Yeah. I observed this while testing. But had no idea why this happend
( as I am very beginner in shell scripting). I was waiting for the review
comment for it.

Thanks again for reviewing my patch request.