Why did "hg push", push my local bookmark to remote?

[Sean Farley]

Jaikiran Pai writes:

> On Friday 06 February 2015 12:41 PM, Sean Farley wrote:
>>> I however wish that it was much simpler than this and maybe that
>>> proposal in that mercurial-dev thread from 2012 could be revisited to
>>> improve the situation?
>> That is moot as far as I understand. What would be considered now is a
>> proposal to improve the 'paths' option to only push certain branches to
>> specific remote servers.
>>
>> I don't understand what you want to be easier. You are saying that
>> adding *one* command-line option of '-s' to your secret branch is too
>> hard? I find that very explicit and refreshing that Mercurial is doing
>> what I told it to do: mark this secret. Commits on top of secret commits
>> are always secret.
>
> Sean, the thing I wish was easier was local/lightweight branches in 
> mercurial. As a developer working on various different features, bug 
> fixes locally I wish it's easier to just do a "hg up 
> local-branch/bookmark/or-whatever-other-term", then do some commits on 
> it and don't have to worry that those commits will end up remotely 
> unless I explicitly push them. The "paths" proposal sounds like it could 
> take care of this. From what I understand of that proposal, it sounds 
> like I would be able to do in hg what I currently can do with local 
> lightweight branches in git like:
>
> git checkout local-branch-foo
> ... // do changes
> git commit -m "WIP commit locally"
> git push upstream master // i.e. *don't* push local-branch-foo but only 
> push the branch named master to upstream remote repo
> ... // more changes to local branch and now ready to have it included in 
> master branch
> git rebase upstream/master // rebase my local branch commits on top of 
> upstream master branch
> git push upstream master // now that the local commits are ready to be 
> pushed to upstream and are rebase on (local) master branch, push it to 
> the remote master branch

This is probably behavior we could experiment with in the remotenames
extension. Thoughts, Ryan?