RFC, poke at 'shallow' branches

[Martin Pool]

On 4 February 2010 15:55, John Arbash Meinel <john at arbash-meinel.com> wrote:
>
> I believe Collin Watson pointed out one of the unique features of the
> UDD workflow. Specifically, that rather than making 100s of branches of
> a given project, he makes 1-2 branches of 100s of projects. As such, he
> isn't helped a lot by a shared repository, since it is generally always
> going to be new data he has to download. So decreasing the data
> downloaded for "give me something to work with" is quite useful for
> improving his daily workflow.
>
> It is something that our 'shallow' branch design could help. Here I'm
> defining it as a stacked branch that has enough data locally to do work.
>
> For the implementation, I was thinking that it wouldn't be too hard to
> add a "get_shallow_stream()" to the StreamSource object, which would
> fetch a revision and its contents without filtering out the basis
> inventory. And at the UI level, it would be exposed as "bzr branch
> - --shallow".

It might work.  I might expect some amount of fallout, so it would be
good to do early in 2.2.

I was wondering if we should tag bugs as 'fallout' so we could later
count them, if they're a problem introduced by an earlier landing that
we could have caught with better review or testing.  In a sense all
bugs that are not feature requests are in this category, yet some seem
much more like this than others.

-- 
Martin <http://launchpad.net/~mbp/>