►
From YouTube: Kubernetes SIG Service Catalog 20161205
Description
SIG service catalog discusses various issues around project governance, code reviews, milestones, and how close to follow the current state of the kubernetes repo
Agenda - https://docs.google.com/document/d/10VsJjstYfnqeQKCgXGgI43kQWnWFSx8JTH7wFh8CmPA/edit
A
A
B
I
think
that
the
meeting
is
the
call
the
weekly
call
is
going
really
well,
but
I
think
you
know
it's
obviously
an
expensive
call
and
I
think
you
know
getting
its
process
as
good
as
possible
will
help
us
to
to
use
it
to
focus
and
manage
the
work
we're
doing
outside
the
call.
So
so
that's
just
the
foundation
for
for
the
discussion
points
I'm
going
to
make.
So
I
think
it's
it's
been
really
good
for
us
to
have
the
agenda
doc.
B
B
So
that's
just
point
one,
the
the
other.
The
other
point
I
wanted
to
bring
up
is
that
it
seems
like
there
are
times
when
we
we
spend
a
lot
of
time
discussing
things
in
the
call
that
aren't
necessarily
beneficial
to
discuss
in
a
call
I
think
I
think
having
a
call
is
very
useful
for
some
things
and
also
not
a
very
good
forum
for
some
other
things,
and
so
it
would
be
helpful
to
me
to
get
really
good
as
a
group
at
identifying
conversations
we're
having
that.
B
A
B
I'll
give
two
examples:
I
think
11
example
I
had
that
I
was
a
little
frustrated
with
was
that
we
spent
almost
40
minutes
talking
about
how
many
lgtm
s2
to
have
for
for
a
merge
to
take
place.
I'm
not
sure,
I'm,
not
sure
what
the
appropriate
thing
would
have
been,
but
but
I
feel
like
that
time
would
have
been
much
better
spent.
If
we
had
identified
the
that
there
was
a
problem
and
and
found
some
way
to
to
have.
B
You
know
found
somebody
who
could
make
a
proposal
that
we
could
then
discuss
in
a
more
appropriate
forum
or
I.
Don't
know
what,
but
I
think
I
tends
to
be
a
lot
more
items
on
the
agenda
than
this.
Then
we
can
handle
in
the
hour
and
I
want
to
make
sure
that
we
are
it's
it's
great,
that
we
schedule
more
calls
throughout
the
week
as
necessary,
but
I
don't
want
us
to
be
doing
that,
simply
because
we're
wasting
time
and
in
previous
calls
or
we're
not
using
I
most
different
on
the
other
gentleman.
A
B
We
want
on
this
call
so
that
so
that
things
don't
devolved
into
long-winded
discussions
that
don't
have
final
outcomes
I
the
suggestions,
I'm
making
I
guess
my
goal
with
them
is
that
you
know
we
have
40
hour
work
weeks,
and
so
this
call
is
a
very
small
part
of
paedon's
as
much
as
we
can
use
this
call
to
focus
and
manage
the
work
we
are
doing
outside
of
the
call
so
that
that's
much
more
collaborative
and
and
focused
I
think
is-
is
beneficial
to
me.
Yeah.
A
A
Unfortunately,
there's
no
such
thing,
and
it's
like
a
honey
pot
right
like
it's
easy
to
get
drawn
into
very
pedantic
discussions
about
governance
and
stuff
like
that.
So
if
we
can
be
really
focused
in
precise
about
one,
what
matter
is
being
discussed
to
what
the
desired
outcome
is
three
be
mindful
that
there
is
no
perfect
solution
to
these
things.
I
think
that
would
be
great
uf
from
me
at
this
moment.
B
Great
so
I
think
the
only
other
example
I
have
and
then
I'll
close
the
floor
because
I
feel
like
maybe
this
conversation
is
going
to
take
up
too
much
time.
I
would
say
that
the
the
other
the
other
concern
I
have
is
that
we
spend
a
lot
of
time.
If,
if
we
get
to
the
point
where
we
are
spending
a
lot
of
time
on
this
call
working
through
designs
of
things
where
perhaps
a
better
mode
of
collaboration
is
that
and
for
that
type
of
thing
is
to
of
an
owner.
B
A
A
C
D
The
idea
of
there's
a
better
word
than
what
I
wrote
in
the
dock,
but
identifying
a
topic
or
discussion.
That's
taking
way
too
long
for
this
call
and
then
coming
up
with
just
a
way,
not
a
process,
but
just
a
really
easy
way
for
someone
to
volunteer
to
take
it
on
asynchronously
and
also
after
they
take
it
on
in
some
kind
of
proposal
or
design
doc
or
whatever
they
need
to
do
then
bring
it
up
in
a
better
forum
could
be
another.
Video
call
could
be
something
else.
C
C
A
A
Is
is
that
we
will
work
to
ensure
that
all
items
are
handled
with
like
one
keep
group
discussions
concise,
especially
around
governance,
have
a
specific
goal
in
mind.
That
is
explicitly
stated
call
out
at
the
beginning
of
each
meeting.
What
the
agenda
is
and
we'll
have
a
will
say,
strange,
attractor,
it's
okay
to
interject,
with
the
words
strange
attractor.
This
signal
at
something
is
getting
too
long-winded
or
verbose
and
needs
to
be
handled
offline
or
in
another
VC.
A
F
D
Ok,
this
is
a
bullet
point
specific
to
seeing
in
progress
the
progress
of
in
progress
PRS
on
the
list
of
pull
requests
in
our
repository
so
before
I
get
started,
I
want
to
scope
this
to
10
minutes
and
no
more
than
the
first
three
sub
bullet
points
which
are,
as
I
said,
determining
the
lgtm
status
of
a
PR
from
the
list.
We
have
to
propose
solutions
and
for
me
the
most
important
thing
is
reviewing.
D
Prs
right
now
is
fairly
painful
because
it
requires
a
lot
of
clicking
and
a
lot
of
me
reading
through
long
ish,
pyaar
comment,
threads
and
I
understand
that
it
sounds
like
a
petty
issue
and
a
shallow
issue,
but
it's
error-prone
as
I've
seen
quite
a
few
times
in
my
almost
merging
pr's,
that
shouldn't
have
been
merged
yet
and
also
it
has
a
cognitive
drain
associated
to
it.
So
that
being
said,
I
would
propose
that
either
we
adopt
pull
approve
of
the
case.
D
I
think
there's
a
Kate's
bot
which
I
actually
didn't
write
in
the
proposals
or
the
lgtm
labels
thing
and
I
would
propose
implementing
one
of
those
things
sooner
rather
than
later
so.
I
would
say:
let's
start
the
10
minutes
here
and
be
done
by
129,
pacific
standard
or
29
minutes,
after
whatever
our
you're
on
so
I'd
love
to
hear
thoughts
on
this.
B
D
A
H
So
this
is
Doug.
I
have
used
pull
up
crew
before
and
it
does
a
fairly
nice
job
of
counting
lgtm
and
won't
let
the
PR
get
merge
until
you
get
the
right
number
from
that
perspective,
it
works
really
well.
It
does
not
support
the
idea
of
automatically
adding
a
lgtm
type
label
if
that's
really
looking
for
in
particular,
you
know
one
that
says
the
LGBT
em
one
two
or
three
will
not
do
that
as
right
now,
I
have
asked
for
that
to
be
added,
but
who
knows
whether
a
ladder
or
not?
H
H
H
D
H
That's
what
I
meant
to
imply
you're,
okay,
with
with
people
manually
having
the
lgtm
one,
two
three
labels
I
personally,
would
would
prefer
some
automated
way.
So
I
would
love
it
if
we
could
get
someone
in
there
to
modify
the
munch
github
bot
to
do
that,
counting
automatically
I
just
haven't
had
Tim.
Look
at
the
code
know
whether
it's
harder,
not
I,
can't
imagine
it'd
be
hard
for
it
to
count,
because
I
think
it
does
look
at
LG
teams
right
now.
H
A
H
H
D
H
H
B
D
D
I
think
we're
yarn
Erin.
Did
you
see
my
comment
in
chat
to
assign
labels?
You
need
admin
access
to
the
repo?
Does
that
present
a
hindrance
or
no
I,
don't
know
the
answer,
but
I
would
say
for
now
we
have
a
possibly
less
than
optimal
solution
that
solves
the
problem
of
seeing
the
review
status
on
the
slash
polls.
F
We
just
skip
developing
and
we're
still
flashing
this
queue,
and
I
would
like
this
flush
to
happen
so
that
we
can
get
into
you
know
just
regular
development
mode.
So
some
of
the
slowness
is
basically
as
soon
as
we
send
a
PR.
We
have
another
one
dependent
ready
to
go
and
the
queue
is
at
this
point
maybe
15
deep.
A
Yeah,
so
I
think
that
we
I
would
like
to
see
the
queue
go
down
to,
but
I
would
also
counseled
against
building
chains
of
dependent
changes,
because
I
many
times
done
that
in
the
past,
with
KU
benetti's
and
then
the
first
one
in
the
chain
is
like
I'm,
not
sure
about
this,
and
then
you've
done
all
this
work
and
you're
not
sure
if
you're
going
to
be
able
to
use
it
anymore.
So
that
might
not
be
the
best
time
management
strategy
to
have.
F
F
You
will
see
Ron
series
of
Michael's
changes
that
are
about
to
hit
the
repo
one
of
them
has
already
done,
but
some
of
them
just
doesn't
make
sense
to
try
to
paralyze
because
they
would
conflict
with
others
that
are
in
that
in
a
chain,
but
none
of
them
I
would
put
in
the
category
of
this-
is
going
to
cause
vetoes
or
long
discussions
or
resets
whatnot.
Okay,
well,.
A
I
mean
I
personally
think
the
pace
has
been
pretty
good
and
I.
I
do
think,
to
some
extent
that,
like
having
in
the
open
source
world
and
meeting
to
get
group
by
in
on
things,
it
just
goes
a
little
slower.
I
am
not
sure
what
we
can
do
to
increase
the
velocity,
but
I
will
say
that
I,
don't
think
many
polls
that
are
on
controversial
have
waited
for
more
than
a
day.
Do
you
do
you?
Disagree?
Martin
am
I
not
perceiving
it
correctly.
I.
F
Would
say
the
pace
has
been
maybe
day
and
some
Delta
around
it.
Some
have
gone
faster,
others
have
gone,
maybe
bit
slower,
but
I'm
generally
impatient
person.
So
you
know
when
there's
something
non-controversial
he's
got
one
lgtm
trilogy
team
and
that
is
really
low
probability
that
somebody's
gonna
object,
I
would
just
say:
let's
keep
merging
two
to
value
velocity
over
stalling
the
pipeline
on
small
non-controversial
changes.
But
that's
just
me,
you
know
I'm
totally
happy
once
we
flush
this
clue,
you
shouldn't
always.
A
F
H
So
my
point
of
view
on
this
is
I'm
very
nervous
about
decreasing
it
down
to
21,
mainly
because
I
think
having
three.
There
is
a
good
practice.
It
gets
set
up
and
I
don't
want
to
feel
like
we're
going
to
do
anything
special
just
because
we're
just
because
we
have
a
backlog
of
it.
She's
coming
I
would
actually
recommend
that
you
make
your
PR
is
a
little
bit
bigger
than
if
you
feel
like
some
of
them
getting
held
up,
and
that
really
are
is
inconsequential.
H
As
you
say,
I
personally
have
no
problem
with
one
PR
solving
multiple
nit
picky
little
things
all
at
once.
I,
don't
think
you
have
to
actually
split
them
up,
but
also
there's
aspect
to
making
it
be
three
that
I
think
maybe
glossing
over
and
that's
how
I
have
him
three
in
there
you
give
more
people
a
chance
to
actually
catch
up
on
what
you're
doing
if
a
PR
gets
merged
too
quickly
and
people
focus
on
just
the
open
PRS,
they
may
not
mr.
H
A
Okay,
I'm
sorry,
the
chainsaw
has
completely
blanked
my
internal
cash,
but
I
got
it
back
now,
I
you
can
yeah.
I
think
if
you
feel
like
things
are
non-controversial,
I
think
that
it
would
probably
make
things
go
a
little
faster
if
non-controversial,
pr's
were
a
little
bit
bigger.
I
I
do
think
that
the
ones
that
are
in
the
category
that
Martin
mentioned
have
been
fairly
fine
grain,
and
maybe
you
can
make
them
making
double
them
up
for
now
and
we'll
see.
A
If,
if
they
have
the
same
velocity,
you
can
probably
triple
them
up
and
we'll
stop.
You
can
stop
once
we
start
saying
us
too
much,
but
I
think
what
has
gone
in
so
far
has
been
like
I
can
see
twice
as
big
being
totally
manageable
at
the
same
pace,
which
gives
you
twice
as
much
throughput
change.
Wise
sounds.
D
F
A
So
the
next
agenda
item
is
I
want
to
see
what
the
appetite
is
in
the
group
for
another
face-to-face
meeting
in
the
remainder
2016,
with
the
fought
already
in
mind
that
we
all
probably
have
end
of
your
holiday
schedules
to
do,
but
I
could
probably
squeeze
in
another
one.
If
there's,
if
there's
interest
in
availability
for
one
in
2016
alternately,
I
think
that
we
should
try
to
do
something
in
january
of
2017.
G
A
A
A
All
right
things
are
going
pretty
fast
in
this
meeting
now
that
we've
all
agreed
to
keep
it
brief.
Okay,
so
then
the
next
milestone
that
is
on
here
or
I'm.
Sorry,
next,
next
agenda
item,
my
name
next
to
it
is
I,
wanted
to
discuss
milestones
for
things
that
we
will
actually
try
to
tell
the
community
a
go
use
this.
This
represents
a
checkpoint
on
our
work
and
I.
A
A
Well,
let's
forget
about
timing
for
now,
let's
talk
about
the
second
milestone
and
then
back
out
to
the
first
one.
So
the
second
milestone
would
be
something
that,
for
me,
I
would
like
to
be
something
that
we
think
is
more
of
like
a
ga
release,
and
that
might
be
a
lot
for
a
second
milestone.
Maybe
we
can
have
three
of
them,
but
ultimately
I'm
shooting
for
something.
That's
a
ga
release
that
we
think
is
full-featured
enough
for
people
to
get
the
right
experience
when
they
use
it.
A
I
I,
wear
right,
experience
is
TBD,
but
I
don't
want
it
to
feel
genki.
I
would
want
the
ga
release
first,
GA
release
to
be
something
that
can
use
API
Federation,
so
that
implies
like
at
a
minimum
after
one
dot,
seven,
where
I
believe
API
Federation
will
become
the
norm
in
Coober
Nettie's
and
that
can
use
service.
Injection
policy,
which
has
already
been
discussed,
is
like
an
eventual
angle
and
also
has
like
a
nice
CLI
experience
that
kind
of
thing.
A
For
those
of
you
that
don't
already
know
the
cube,
CTL
binary
supports
things
like
raw
get
and
resource
discovery
from
a
thing
that
looks
like
a
cube,
API
server,
so
we
could
point
cube
CTL
at
like
a
rough
version
of
service
catalog
that
has
an
API
server,
that
I
can
introspect
and
do
and
get
a
raw
get
and
create
and
list
probably,
but
to
do
something,
that's
closer
to
what
you
have
for
cube
CTL
for
existing
API
resources.
We
probably
have
to
write
our
own
code,
which
is
okay
and
I.
A
Think
if
you
also
back
out
the
things
that
we
want
to
do
in
the
core,
you
take
that
service
injection
policy.
You
take
a
federation
and
I
know
like
the
dais
guys
have
mentioned
it
various
points
in
our
previous
discussions
that,
like
it,
is
important
for
them
to
have
a
usage
path
on
older
versions
of
cube.
I.
Think
that's
that's
good.
To
have
in
that
that
initial
milestone
is
one
where
you
can
use
it
on
an
older
version
of
cube,
because
we
don't
make
any
assumption
about
service
injection
policy.
F
Martin
yeah
I'll
start
speaking,
so
we
had
a
good
conversation,
I
think
with
Paul
and
Tim
Hawking
actually
jumped
in
as
well
on
the
issue
of
realigning
our
repo
with
co,
Brandis
repo,
which
is
a
good
goal
and
Tim,
contributed
some
observations
in
terms
of
what
did
Cooper
Nellie's
repo.
Do
that
doesn't
look
like
it
was
a
good
idea,
looking
back
and
some
of
this,
if
we
were
to
realign
with
what
Cabrera's
wished
they
had
done
as
opposed
to
what
they
did
do.
We
would
need
to
make
some
changes.
F
A
I
think
we
settled
PKG,
and
I
got
a
respectfully
disagree
with
tim
on
this
one.
That
I,
like
wishing
you
had
done
something
is,
is
totally
understandable
and
good,
but
I
think
the
practicalities
of
like
we
need
to
keep
in
mind.
I
think
a
very
high
priority
thing
is
to
ease
the
the
cognitive
like
commit
charge
of
doing
a
context,
switch
from
cooper
Nettie's
to
service
catalog,
and
if
we,
if
we
start
departing
from
what
Cooper
Nettie's
does
like,
we
will
have
people
that
will
be
confused.
A
They
will
and
remember
folks
confuse
people
generate
questions
and
it
takes
time
to
answer
them.
So
I,
though
things
might
be
imperfect,
this
is
just
my
opinion
here.
I'm
not
being
dictatorial,
but
I
want
to
make
an
argument
on
the
merits
is
that
it's,
it's
probably
has
a
higher
cost
than
we
can
imagine
right
now
to
do
something
that
is
different
m
and
make
something
organized
differently.
A
I
totally
understand
the
impulse
to
do
what
you
think
is
right,
but
I
think
that
what
we
in
a
perfect
world,
probably
things,
do
look
different
from
Cooper
Nettie's
and
there's
a
cost
associated
with
each
Delta.
So,
like
pack
PKG,
I
feel
really
really
strongly
about
I
feel
less
strongly
about
hack,
but
I
also
feel
like
we
should
like.
A
We
should
not
that
perfection
be
the
enemy
of
progress.
We
just
moved
things
into
hack,
I,
really
don't
want
to
move
them
right
back
unless
there's
a
strong
reason
to
do
so,
and
I
will
also
say
that
the
tooling
that
I
expect
that
we're
going
to
want
to
use
or
on
code
generation
for
crew
benetti's
is
very
opinionated
about
where
things
go,
and
I
think
the
path
of
least
resistance
is
very
much
one
where
we
probably
go
with
the
conventions,
and
we
can,
we
can
evolve
things
at
the
same
rate
as
kuber
netezza
wolves.
I
I
think
I
basically
said
everything
in
the
chats,
but
my
thinking
is
that
we
also
establishing
a
pattern
from
the
incubator
projects
and
hence
it's
going
to
be
basically
a
goal
by
for
everything
coming
after
us,
at
least
that's
my
understanding.
So
when
we
make
some
decisions,
it
would
be
good
to
understand
if
there's
a
reason
for
it.
That
is
my
personal
thing.
A
A
I
All
I
was
going
to
say
is
that
I?
My
understanding
is
that
we
are
simply
going
to
be
establishing
a
pattern
for
all
the
other
incubator
projects
moving
forward,
at
least
that's
my
understanding
where
we
are
so.
Hence
any
project
coming
after
us
will
be
probably
asking
the
same
questions
as
we
are
asking
ourselves
right
now,
which
is
like.
I
Why
did
they
do
things
this
way,
so
at
least
the
people
to
understand
why
some
decisions
are
made
and
and
so
forth
and
part
of
it
is
just
for
my
own
edification,
which
is
to
say
why
on
earth?
Would
anybody
do
this,
but
the
other
part
also
is
then
there's
a
cost
for
any
project
coming
after
us
that
will
be
coming
from
a
you
know,
standard
project,
if
you
will
their
standard
is
uber
navy's
is
one
standard
at
then
there
might
be
another
standard
that
is
the
rest
of
the
world.
I
A
And
I
have
not
been
trying
to
make
an
argument
to
authority
that
it
is
right
to
do
it
because
Cooper
Nettie's
does
it.
My
concern
is
like
Arab,
basically
around
easing
people
into
it
and
avoiding
a
situation
where
we
hit
a
roadblock.
That
takes
like
a
lot
of
time
to
clear,
because
we
have
to
change
some
upstream
tooling
thing
that
is
used
upstream
and
in
service
catalog
and
I
have
like
I,
basically
have
worried
that
it
we
will
have
to
fight
like
really
hard
strange,
attractor
latent
battles
with
other
groups
in
the
communities
community.
A
That
might
just
be
extremely
counterproductive
that
we
could
avoid
if
we
just
shape
things
the
same
way
and
we
work
to
change
people's
minds
instead
of
changing
code.
First,
you
know
because,
like
I,
I
personally,
I
think
that,
like
a
point
that
went
into
the
a
point
that
Tim
made
is
that
the
the
build
the
interface
to
the
build
system
is
via
to
make
file
so
like
I,
don't
think
it
should
be
that
hard
to
change
KU
benetti's.
A
If
we
really
think
that
that
hack
is
wrong,
then
it
should
be
doable
to
change
communities
in
the
meantime.
I'd
prefer
to
not
have
to
explain
to
people
in
the
chat
once
this
repo
becomes
popular
that
oh,
the
build
scripts
are
here,
maybe
I'm,
maybe
I'm
blowing
this
risk
out
of
proportion.
It's
totally
possible,
I'm
just
some
dude
right
like
I
don't
I
don't
know
anything,
I'm
just
a
guy.
A
Let
me
let
me
read
Kent's
comment
in
here
ken
says,
dear
Marty,
if
my
calculations
are
correct,
you
should
have
received
this
shortly
after
you
saw
the
DeLorean
hit
by
lightning
know
that
I
am
alive
and
in
a
year
living
in
the
year
1885.
Oh,
wait!
I'm!
Sorry,
that's
not
the
that's!
Not
what
camera
he
says.
My
mic
is.
Apparently
my
mic
is
apparently
not
working.
Wait
now,
I
just
lost
it
all
right
hold
on
a
second
hold.
A
My
mics
not
working
so
I
type
I'm
typing
what
I'm
thinking
instead
I
think
we
need
to
deal
with
this
on
a
case-by-case
basis.
Hack
vs.
script,
for
instance,
sounds
like
something
simple
where
we
should
just
main
tera
T
maintain
parity
with
them
to
me.
I
feel
like
we
should
consider
partying
what
they
have
done
only
in
places
where
there's
a
more
functional
operational
difference
that
impacts
our
ability
to
get
things
done
smoothly
or
promote
community
understanding
of
our
project.
A
D
I
would
give
that
a
big
plus
one
the
best
argument.
I've
heard
to
do
things
like
uber
Nettie's
is
tooling,
and
I
agree
it
would
it's
going
to
be
very
nice
for
people
to
come
in
and
have
our
guy
look
just
like
our
guy
or
gal
look
just
like
the
kuru
Nettie's
image.
But
that
being
said,
as
tim
has
said,
the
coup
Bernie's
image
has
and
probably
will
change
a
lot
and
even
his
I
think
he
call
it
a
seed
repo
or
his
kind
of
his
best.
F
F
A
F
One
thing
would
be
nice
to
quantify,
I've
been
asking
for
that.
Last
time
is,
are
we
on
par
yet
or
what
is
left
to
do
and
we
should
be
deliberate
about
each
of
those
things
whether
we
want
to
exert
the
cost-
and
I
would
say
burden
should
be
on
the
person
trying
to
make
a
change
justify
that
to
change
worth
it
so
I.
F
That
sounds
good
to
me
that
sound
good
yeah
be
nice.
If
they
had
rationale
so
too,
please
valets
desire
as
well,
and
then
we
can.
We
can
weigh
the
cost
of
doing
it
now
versus
doing
it
in
the
future
when
the
tool
breaks
that
we
are
trying
to
use,
but
at
this
one
actually
have
no
idea
what
even
the
scope
is
like.
What
are
the
things
that
we
would
be
changing?
Okay,.
A
I
think
actually,
we've
come
a
fair
way
to
doing
it.
There's
maybe
one
or
two
things
that
I
want
to
change
right
now
and
then
I
can
shut
up
about
this,
and
we
can
all
just
keep
going
and
concentrate
on
solving
the
business
problems
in
the
repo
sounds
good.
Ok,
so
we're
we're
at
time
now
just
about
there's
four
minutes
left
and
so
I'm
inclined
to
inject
one
item,
which
is
not
on
the
agenda,
which
is
I
would
like
to
have
a
dis
design,
discussion
on
unbind
and
possibly
deep
provision
this
week.
A
I
think
those
are
very,
very
important
to
have
group
consensus
around
I
I
do
not
think
that
I
am
available
to
meet
tomorrow
at
this
time.
How
do
people
feel
about
a
completely
design
based
discussion
or
design
discussion?
One
hundred
percent
design
discussion
meeting
on
Wednesday
at
this
time,
I
see
a
thumbs
up
from
Aaron.