Bazaar focus for 2.1 and 2.2

[Elliot Murphy]

On 12/17/2009 04:00 PM, James Westby wrote:
> On Thu, 17 Dec 2009 14:19:32 +1100, Martin Pool <mbp at canonical.com> wrote:
>> My hypothesis is that we (Canonical's Bazaar team) will get to grips
>> with UDD better if there is a tighter medium-term focus.
>>
>> The key question is whether vcs-imports is that thing, or not.
>>
>> If we do make imports that one important thing then I'll be asking
>> people not to work on looms, UDD merge support, or nested trees until
>> they're done.
> 
> What would you consider "done." Do you think that with 4 months effort
> you could get to 0 outstanding failures, or would there be some other
> stop point?
> 

I am definitely not a primary stakeholder in this thread, so take my
input with a bucket of salt. I am absolutely fascinated by the work
being done here, and very excited as the manager of an upstream team who
also maintains packages in Ubuntu to be able to use UDD.

Trying to decide between two important bits of work in a large connected
system is always tough. Thinking about this last night and today,
imports seem like a long tail kind of problem, where it's hard to
declare "done" as James points out.

It seems like "doing UDD" requires several pieces, one of which is an
import, and other pieces which are merge support, nested trees, looms or
pipelines. I don't know how many of the imports are working, so I'm
going to pretend that it's 80%.

Maybe one way of asking this question is: is it better to first fully
enable UDD for 80% of the packages (the ones with working imports), or
more important to partially enable UDD for 100% of the packages by
driving imports to 100%? I'm probably biased because as an upstream with
most projects in bzr and happy with the current git imports I'm not one
of the folks with a broken import, but I'd rather see the work on nested
trees, UDD merge, looms, etc. even knowing that it would postpone having
working imports for some percentage of packages.

My reasoning is that imports are a very well understood task, and it's
not likely that in the course of driving imports to 100% that we would
learn anything new. However, UDD merge, looms, nested trees are features
that would definitely influence the UDD workflow and tools and tutorials
that grow around that workflow, and postponing would perhaps make the
UDD experience suboptimal for all packages. When working on a large
system and choosing between working on a well-understood problem or
hooking up some new parts of the system that might hold some
yet-undiscovered problem or change the flow of data through the system,
my bias definitely goes towards discovering the new problems and hooking
up the new parts.

Whichever you end up choosing, I'm so thrilled to see progress on UDD.
This is going to absolutely rock.

-- 
Elliot Murphy | https://launchpad.net/~statik/

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 197 bytes
Desc: OpenPGP digital signature
Url : https://lists.ubuntu.com/archives/ubuntu-distributed-devel/attachments/20091217/4a31ed3c/attachment.pgp