Re: udd at uds-n

[Martin Pool]

On 1 December 2010 09:27, Barry Warsaw  wrote:
> Hi Martin, thanks for posting this update, and apologies for taking so long to
> get around to reading it.
>
> On Nov 17, 2010, at 08:06 PM, Martin Pool wrote:
>
>>At the end of that discussion we picked two specific items for the bzr team:
>> * speed
>> * loom support, on lp and within bzr, and connecting them to packaging 
>> patches
>>
>>and for Launchpad
>> * build from branch into the main archive
>> * actually execute a merge from a merge proposal
>> * through launchpad. merge from a debian branch into an ubuntu udd branch
>
> I think we also have to address the package import failure issues.  I see two
> parts to that.  First, we need to make sure that if someone branches
> lp:ubuntu/foo (or ubuntu:foo ) on a branch that has had import failures,
> that some very prominent warning is displayed.  Perhaps the branch fails
> unless a --force flag or something is given.  I'm not sure exactly, but I'm
> fairly confident that silently producing a branch that's out-of-date is *not*
> a good thing. :)
>
> Second would be to address the issues actually causing the failures, but
> that's a longer term project.

jml asked a similar question but I just realized it was off the list.
That was, how do I reconcile what was discussed at UDS with the
feedback we got in the UDD survey results

which did emphasize import reliability.

The short story is that for now, we're going to work on holistic
network performance to/from Launchpad, on getting the package importer
working better, and on general bugs/reactive work.

-- 
Martin

-- 
ubuntu-distributed-devel mailing list
ubuntu-distributed-devel@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-distributed-devel

Re: udd at uds-n

[Barry Warsaw]

Hi Martin, thanks for posting this update, and apologies for taking so long to
get around to reading it.

On Nov 17, 2010, at 08:06 PM, Martin Pool wrote:

>At the end of that discussion we picked two specific items for the bzr team:
> * speed
> * loom support, on lp and within bzr, and connecting them to packaging patches
>
>and for Launchpad
> * build from branch into the main archive
> * actually execute a merge from a merge proposal
> * through launchpad. merge from a debian branch into an ubuntu udd branch

I think we also have to address the package import failure issues.  I see two
parts to that.  First, we need to make sure that if someone branches
lp:ubuntu/foo (or ubuntu:foo ) on a branch that has had import failures,
that some very prominent warning is displayed.  Perhaps the branch fails
unless a --force flag or something is given.  I'm not sure exactly, but I'm
fairly confident that silently producing a branch that's out-of-date is *not*
a good thing. :)

Second would be to address the issues actually causing the failures, but
that's a longer term project.

-Barry


signature.asc
Description: PGP signature
-- 
ubuntu-distributed-devel mailing list
ubuntu-distributed-devel@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-distributed-devel

udd at uds-n

[Martin Pool]

Just a few brief notes about UDD discussions at UDS-N.  A lot of this
is captured to various degrees in gobby documents and IRC logs at
, but they're pretty long:


* There was a lot of interest in adopting something like the bzr patch
pilot process  in Ubuntu; it's worked well for us and I'm pleased Rick
is willing to give it a go in the different situation of Ubuntu.  I
hope it goes well.


The main discussion about UDD is here, starting about
2010-10-27T16:08, and there are a lot of good things for us to work
on, including:

 * speed of getting source through UDD, considered holistically
(including outright speed, needing to fetch from the UK rather than a
closer mirror, shallow checkouts, proactive mirroring)
 * various merge cases that could be better
 * problems with packages with complex history already in bzr,
creating multiple histories

Some other observations:
 * currently used by very much early-adopters
 * build from recipe quite successful, though still quite young
 * focus on eliminating tedium
 * check out grab-udd-merge

At the end of that discussion we picked two specific items for the bzr team:
 * speed
 * loom support, on lp and within bzr, and connecting them to packaging patches

and for Launchpad
 * build from branch into the main archive
 * actually execute a merge from a merge proposal
 * through launchpad. merge from a debian branch into an ubuntu udd branch

We did not get around to doing interactive user testing; I'm trying to
reschedule that with mrevell.

We did get survey responses, and I will send a summary of them soon.

-- 
Martin

-- 
ubuntu-distributed-devel mailing list
ubuntu-distributed-devel@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-distributed-devel