Bringing ~charmers and ~core back together around the Juju go port

Jorge O. Castro jorge at
Wed Feb 27 15:48:39 UTC 2013

I had a call with Dave Cheney last night to catch up on things. After
some discussion it's clear to me that we could do a better job working
together during this awkward transition period. Now that things are
really coming along and gojuju is dropping new releases every two
weeks, I think it's time to do a Blues Brothers and "Get the band back
together" and get out of this uncertain rewrite funk.

Here's what I'd like to propose:

- If you're not using Juju in production, use the Go port. I switched
over to it and was able to deploy things in AWS.
- No more pyjuju at charm schools. This can no longer be optional, we
will rehearse earlier and give the go team a chance to make our demos
work. They release every 2 weeks, there's no reason we can't sort
this. In hindsight, we probably could have done SCaLE's charm school
at least partially with the go port.
- Report bugs in go Juju. The core team is _keen_ on finding out what
works and what doesn't work. The "I brought this up last year and was
ignored" excuse is no longer a valid.
- If a bug is keeping you from switching to go Juju please mark
affects me in the launchpad bug so the core team has an idea of what's
keeping people from switching.
- I am going to start asking someone from the core team to sit in in
charm meetings.
- I am going to start asking to sit in on core team meetings.
- Every week during the charm meeting we are going to pick one bug
that is keeping the most amount of people from using the go port, and
we'll tell the core team and go from there. If there's an issue on why
you're not using the go port and it's not reported, you're going to
report it on the spot.
- As we closer to 13.04, and especially as the core team is doing the
go releases to fix our issues, I'll be ratcheting up the tough love at
the ~charmers meetings, so please make an effort to start filing bugs
with your problems.

I would also like to use this mail to kick off a discussion about
bugs. We have bugs tagged "prod" in lp:juju. We also have features and
wishlist items. We need to ensure these are on the core team's radar.
What should we do with them, Also affects juju-core? We also need to
get a handle on the bugs to prioritize them, etc. It should be
_obvious_ to anyone from looking at the bug list as to what the
priorities for the fixes are.

Basically, we need to be more active in the bugs and telling the core
team what we need to get things working. I'd also like to know how we
classify bugs with gojuju that are charmer specific. So bugs that
would be important for people who are writing charms, things like
debug-hooks, things of that sort, so that we can ensure that the bugs
for the people who are writing charms are getting the love as well.

And lastly, I can't force anyone to use the go port, but I can tell
you the #1 reason I am switching to it asap. Upgradeable environments.
The go port lets you deploy stuff, and then later on when there's a
new release, you just get all the new features without throwing
anything away. That means that if you find out the bare minimum of
what you need to get by, you can switch, and then you have a solution
moving forward.

I know that many of the providers are not ready, but they are making
progress and we need to start leaning forward in the saddle to get
ready for this.

Jorge Castro
Canonical Ltd.

More information about the Juju mailing list