Use of named branches

Guido Ostkamp hg at ostkamp.fastmail.fm
Sat May 5 22:36:03 UTC 2007


On Sat, 5 May 2007, Mark A. Flacy wrote:
> The folks using named branches believe that "named branches" provides 
> the exact semantics and results that you would get from having cloned 
> repositories as branches.

Of course with cloned repositories you need to use pull and push, so 
semantics are not 100% the same. But the possibilities you get by using 
branches should be identical whether you are on an internal named branch 
or on one that's represented by a clone. Otherwise internal branches would 
make no sense to me.

Even the Mercurial book says "If you're more in the power users category 
... there is an alternative way of handling branches ... Mercurial ... can 
also work with multiple 'big picture' branches. The key to working this 
way is that Mercurial lets you assign a persistant name to a branch ...".

> That's a reasonable assumption to make, but I don't believe the current 
> Mercurial implementation will handle it.  Since the "merge" command is 
> currently designed (AFAICT) to reduce the number of heads in the 
> repository to 1, I don't see how the named branch model *would* be able 
> to work without a lot of care from the end user.

The current version obviously has a bug or is not mature yet. The 'git' 
implementation shows, how it should work.

Let's wait whether Matt has some news on this topic.

Regards

Guido



More information about the Mercurial mailing list