Can't push to a karmic branch because of a Lucid one?

[Ted Gould]

Trying to make packaging branch for my karmic branch. Got this error, not sure why. The branch is 2a. I'm not sure why it's using some Lucid branch that is outside my project.

$ bzr push lp:~indicator-applet-developers/ubuntu/karmic/indicator-applet/series0.3
Using default stacking branch /~ubuntu-branches/ubuntu/lucid/indicator-applet/lucid at lp-46121488:///~indicator-applet-developers/ubuntu/karmic/indicator-applet
bzr: ERROR: KnitPackRepository('lp-46121488:///~ubuntu-branches/ubuntu/lucid/indicator-applet/lucid/.bzr/repository')
is not compatible with
CHKInventoryRepository('lp-46121488:///~indicator-applet-developers/ubuntu/karmic/indicator-applet/series0.3/.bzr/repository')
different serializers

[Michael Hudson-Doyle (mwhudson)]

For some reason the official source package branch for indicator-applet is not in 2a format. Sounds like something for Mr. Westby to fix I think...

[Ted Gould (ted)]

I'd also add that it doesn't make sense to me that a karmic branch would be stacked on a lucid one.

[Jonathan Lange (jml)]

On Wed, Nov 18, 2009 at 2:18 PM, Ted Gould <email address hidden> wrote:
> I'd also add that it doesn't make sense to me that a karmic branch would
> be stacked on a lucid one.
>

Stacking just a storage and network optimization trick. Every time you
have to think about it (at least wrt Launchpad), it's a bug.

In this case, stacking makes sense because the karmic branch then
takes up hardly any space. By stacking karmic on lucid rather than the
other way around, we make network access for lucid branches faster.

jml

[John A Meinel (jameinel)]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Michael Hudson wrote:
> For some reason the official source package branch for indicator-applet
> is not in 2a format. Sounds like something for Mr. Westby to fix I
> think...
>
> ** This bug is no longer a duplicate of bug 393677
> pushing a new 1.9 branch stacked on a 2a branch fails leaving an empty branch behind.
>

I actually guessed that the official package branch *was* in 2a, and the
branch being pushed was not. If that is true, then I think it is bug
#393677. If not, then you're probably right.

John
=:->

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAksEYKMACgkQJdeBCYSNAAPt8wCdGfX8999uKRTD/3ACUlos/dvK
JCIAn3xJ6fYPRevA6SozaAyUWkBMOWs9
=FuSc
-----END PGP SIGNATURE-----

[Michael Hudson-Doyle (mwhudson)]

I did actually check before making that comment :-)

[Robert Collins (lifeless)]

On Wed, 2009-11-18 at 20:05 +0000, Michael Hudson wrote:
> For some reason the official source package branch for indicator-applet
> is not in 2a format. Sounds like something for Mr. Westby to fix I
> think...

James, did you migrate indicator-applet to an 'upstream derived
branch' ? If so, can you add to your requirements for that that the
upstream upgrades to 2a ? :)

-Rob

[Michael Hudson-Doyle (mwhudson)]

Robert Collins wrote:
> On Wed, 2009-11-18 at 20:05 +0000, Michael Hudson wrote:
>> For some reason the official source package branch for indicator-applet
>> is not in 2a format. Sounds like something for Mr. Westby to fix I
>> think...
>
> James, did you migrate indicator-applet to an 'upstream derived
> branch' ? If so, can you add to your requirements for that that the
> upstream upgrades to 2a ? :)

I didn't realize there were any 'upstream derived branches' yet. If
there are I am interested and wish to subscribe to your newsletter...

Cheers,
mwh

[Robert Collins (lifeless)]

On Thu, 2009-11-19 at 00:07 +0000, Michael Hudson wrote:
> Robert Collins wrote:
> > On Wed, 2009-11-18 at 20:05 +0000, Michael Hudson wrote:
> >> For some reason the official source package branch for indicator-applet
> >> is not in 2a format. Sounds like something for Mr. Westby to fix I
> >> think...
> >
> > James, did you migrate indicator-applet to an 'upstream derived
> > branch' ? If so, can you add to your requirements for that that the
> > upstream upgrades to 2a ? :)
>
> I didn't realize there were any 'upstream derived branches' yet. If
> there are I am interested and wish to subscribe to your newsletter...

See the u-d thread I linked from the udd list and mentioned last week.

-Rob

[James Westby (james-w)]

On Wed Nov 18 17:24:39 -0600 2009 Robert Collins wrote:
> James, did you migrate indicator-applet to an 'upstream derived
> branch' ? If so, can you add to your requirements for that that the
> upstream upgrades to 2a ? :)

No, I think this was one of a few imported in an older format, I forget
why now, but it was certainly my intention to upgrade them. Michael
was helping me with this the other day.

Thanks,

James

[Tim Penhey (thumper)]

My question here is what, if anything, are we going to do about this?

[Martin Pool (mbp)]

This is not ideal behaviour but I also don't think it's a specific bzr bug.

[Michael Hudson-Doyle (mwhudson)]

We should make sure all package branches are in 2a format -- there's a thread on the udd mailing list about this.

[Michael Hudson-Doyle (mwhudson)]

This is done now (well, apart from like 5 branches, none of which are the stacked-on branches).