On Tue, Mar 2, 2010 at 5:10 PM, Scott Olmsted <[email protected]> wrote:

> I want to handle merging their changes, so I've asked them to create
> branches 'task1', 'task2', etc.

Never do that. The root principle of Git is it makes branching and
merging very easy, so you can easily avoid them at all costs.

Use a command line front-end, such as rake, and create miniature rake
tasks that pull, test, and push your code.

Push and pull from the same repository. Use TDD and "continuous
integration", and integrate each time the code grows even a tiny bit
better.

Doing it like this completely flattens the odds of "integration hell".
Just because the integration is automated doesn't make it any less
hell.

Only branch and merge if u have a customer-support emergency, and
don't want to put metadata barriers around new features that an old
customer is not ready for. Branch the labelled version of the code
they got, make their change, and merge & update as soon as possible.

-- 
  Phlip
  http://c2.com/cgi/wiki?ZeekLand

-- 
SD Ruby mailing list
[email protected]
http://groups.google.com/group/sdruby

Reply via email to