[RFC] New name for 'repositories' - 'baskets'

[Martin Pool]

On 2006-03-02 Thu, at 10:36 AM, Jan Hudec wrote:

> On Wed, Mar 01, 2006 at 09:42:47 +0100, Erik Bågfors wrote:
>>>
>>> I agree, that the object should have a name. But from users  
>>> perspective,
>>> the Repository(is_shared=true) is manipulated by user explicitly,  
>>> while
>>> the Repository(is_shared=false) is created behind the scenes and the
>>> user does not need to think about it. However when explaining the
>>> concept, you need a name for Repository() as well.
>>
>> What am I missing here?
>>
>> -------
>> A repository is a "place/thing/whatever" that stores your revision
>> data. There are two kinds of repositories in bzr.
>>
>> A shared repository can contain multiple branches, this is the  
>> kind of
>> repository that you find in for example subversion.
>>
>> A "none shared/internal/branch specific/private/whatever" repository
>> is contained in a stand alone branch and is automatically created
>> whenever you create a stand alone branch.  This type of repository  
>> can
>> not contain multiple branches
>> -------
>>
>> Is there anything not clear about that description?
>
> This description is perfectly clear. The point is whether at least  
> all the
> core developers can agree on it and use it consistently.

Yes, I agree.

I feel comfortable with the qualifiers "shared" and "standalone" --  
perhaps just because that's the term we've used here so far.  One  
problem with the word "shared" is that it might be read as indicating  
the repository is shared over the network, or shared between users,  
or something.

Deciding how to document/explain this well really depends on which  
use model we intend to teach as primary.  I think it's good to pick  
one to explain first, and then get to the others later.  Standalone  
branches are the most well-established model so perhaps what we  
should recommend for the time being.  This suggests text like this:

   bzr stores the history of the branch within the .bzr directory.   
(Later you can learn about how to store the history of several  
branches within a common shared repository in a way similar to CVS.)

-- 
Martin Pool