Shared repo setup

[Joshua Judson Rosen]

On 2014-07-02 14:05, Lucio Crusca wrote:
> Thanks for the replies. Now another question arises:
>
> Do I really need "bzr init-repo" on the server? Can't I just "bzr init" as
> if it were a local repo and forget about creating the branch on the
> server? What do I miss if I do so, apart from the warning about working
> tree not being updated?

The warning is actually orthogonal; you can either use "bzr init-repo --no-trees"
or initialise your individual branch(es) with "bzr init --no-tree".
Either one will prevent that warning.

There are a few reasons I'd advocate using a shared repository to hold
your branches, though:

	- it makes creating new branches for the same project on your server faster;
	  e.g.: when you create a new release branch, or push up a new experimental
	  branch, only the new data needs to be copied to the server because the
	  old/shared data is already there.

	- setting defaults (like "--no-trees") on the shared repo means that
	  you don't need to remember to do it for each new branch.

	- allocating a directory to hold branches for your project at the beginning
	  is generally a good idea; e.g.: if you start out with your main branch being
	  "/srv/bzr/myproject/trunk" rather than just "/srv/bzr/myproject",
	  you won't have to reorganise everything when you realise that you
	  actually do need more than one branch.


That last point is not really about shared repos per se, but if you're going
to make a project-directory to hold branches anyway, you might as well make
it into a shared repository to get the other benefits; and I've noticed
that, often, when people talk about `not bothering to create a shared repository',
they actually mean that they're not bothering to create the intermediate
project-directory at all.


-- 
"'tis an ill wind that blows no minds."