[PATCH 0 of 9] improvements for named branches

Michael Gebetsroither gebi at sbox.tugraz.at
Wed Mar 5 04:39:21 CST 2008


* Patrick Mézard <pmezard at gmail.com> wrote:

> Well, you can already have multiple heads *without* named branches,
> and there was no compelling reason to disallow them with named
> branches.

Why not just see heads as unnamed branches?
It would make the whole system much nicer.

I really don't see any difference between a head and a branch, both are
branches, heads just don't have a name.

Heads are nice if there are 2 or 3 locally, but they are hard to manage
in projects with many devs.
They are error prone und forbidden by policy in many projects, so why
allow them?

> Besides, you have to tolerate "named branches" sub-branches
> to allow several people to work on the same named branch, even if you
> may prevent them from pushing additional named branches heads in a
> reference repository.

No you dont. With this you open a can of worms.
Just prevent them to push if that would create a remote head. And force
them to either merge locally or push their changes as a new branch.

That is the burden of working in one single repo in _one_ branch with
multiple people.

The _REAL_ problem is, that a head looses it's meaning as soon as it is
pushed (no one except the author knows what it does, and even for the
author it's quite a bit of work to find his head again).

cu,
michael
-- 
It's already too late!



More information about the Mercurial-devel mailing list