Re: [PATCH] doc: clarify difference between `push.default` `simple` and `current`

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



"Dan Fabulich via GitGitGadget" <gitgitgadget@xxxxxxxxx> writes:

> From: Dan Fabulich <dan@xxxxxxxxxxxx>
>
> The documentation made `simple` and `current` sound identical. The
> difference is that `simple` strictly checks that the upstream tracking
> branch's name matches the current branch's name.

All of the above are correct, and a patch that sticks to fixing that
would have given us a great improvement.

Thanks for working on this documentation update, but it seems some
unrelated changes are mixed in.

>  	Different values are well-suited for
>  	specific workflows; for instance, in a purely central workflow
>  	(i.e. the fetch source is equal to the push destination),
> -	`upstream` is probably what you want.  Possible values are:
> +	`simple` is probably what you want.  Possible values are:

This change is not explained/justified at all why it was part of the
patch in the proposed log message.

And I do not think this is a good change.  `upstream` is recommended
for most people when they employ a purely central workflow.  You can
start working from the common 'master', even on multiple topics in
parallel at the same time, and perform "git push" with push.default
set to 'upstream'.  With 'simple' you cannot.

    $ git checkout -t -b theme1 origin/master
    ... work work work ...
    $ git checkout -t -b theme2 origin/master
    ... work work work ...
    ... changes for theme2 become complete first ...
    $ git push

Here, if your push.default is set to 'upstream', your theme2 updates
their master, which is exactly what you want.  Then

    $ git fetch origin
    $ git rebase origin/master theme1
    ... rebased on updated 'master' at theirs --- at least it should
    ... contain what we did on our theme2 topic, but possibly
    ... changes from other people.
    ... more work ...
    $ git push

Again, your theme1 updates their master, which is exactly what you
want.

> @@ -23,8 +23,8 @@ push.default::
>    given. This is primarily meant for people who want to
>    avoid mistakes by always being explicit.
>  
> -* `current` - push the current branch to update a branch with the same
> -  name on the receiving end.  Works in both central and non-central
> +* `current` - push the current branch to update the branch with the same
> +  name on the remote.  Works in both central and non-central
>    workflows.

Again, a change that is not explained/justified.  "a" -> "the" I can
understand (i.e. a branch with the same name is unique over there,
so "the" is more appropriate), but not the other one.

>  * `tracking` - This is a deprecated synonym for `upstream`.
>  
> -* `simple` - push the current branch with the same name on the remote.
> +* `simple` - push the current branch to its upstream tracking branch,
> +  but only if the upstream tracking branch has the same name as the
> +  current branch. (`simple` will fail with an error if the upstream
> +  tracking branch's name doesn't match the current branch's name.)

That is correct.  The additional text may be somewhat helpful for
somebody who just got an error message and wants to understand where
the error comes from.

But stepping back a bit, is understanding why it failed the primary
thing our documentation should aim for?  I'd rather see our
documentation help the user achieve what they wanted to do in the
first place.  I.e., Be able to push without an error to publish
their work.  And for that "this will fail when X" is less helpful
than "this is appropriate if you work this way."

    simple - this is like `upstream` but with additional restriction
    that the local branch must be named the same as its upstream
    branch.  Suitable with a very simple centralized workflow, where
    you fork off of their 'master' branch to create your own
    'master', work there, and push the branch back.

>  +
> -If you are working on a centralized workflow (pushing to the same repository you
> -pull from, which is typically `origin`), then you need to configure an upstream
> -branch with the same name.

I do not think this removal is explained/justified, either.  Those
who set push.default to 'simple' while using the centralized
workflow must use one-to-one correspondence, so this advice is very
relevant.  What makes it a good idea to remove it?

Thanks.




[Index of Archives]     [Linux Kernel Development]     [Gcc Help]     [IETF Annouce]     [DCCP]     [Netdev]     [Networking]     [Security]     [V4L]     [Bugtraq]     [Yosemite]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Linux SCSI]     [Fedora Users]

  Powered by Linux