►
From YouTube: Kubernetes SIG Federation 20170613
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
We'll
be
posted
later
so
I
don't
see
well,
I!
Guess
we
just
don't
have
an
agenda
really
well-formed,
but
Quentin
comes
at
email.
The
list
talking
about
Hawaii
planning,
even
though
actually
here
yet
but
assuming
there
is
going
to
be
time
to
discuss
loose
from
a
preliminary
fashion
where
we're
gonna
be
targeting
for
aligning
what
people
wanna
target.
B
A
A
On
like
I
was
hoping
you
get
actually
separated
to
use
cases
out
in
the
fact
that
they're
still
kind
of
conjoined
means
I,
find
it
very
confusing.
To
read.
Some
stuff
is
relevant
for
the
local
staple
set
to
use
case
and
some
for
the
global,
and
we
discussed
sort
of
separating
those
two
just
never
saw
at
all,
and
one
can
certainly
be
done
and
separately
from
the
other.
I
means
your
intent
to
try
to
target
both
of
those
with
cyclists.
D
Yeah,
so
what
I
understood
from
the
earlier
comment
that
you
gave
in
the
last
meeting
was
that
in
the
in
the
use
cases
I
ensure
that
ban
use
cases
the
base
use
case,
which
is
when
you
create
stateful
sets
they
are
the
same
stateful
local
settle
sets
created
into
clusters
and
also
Banali.
We
build
the
other
use
case,
which
is
on
top
of
this.
You
can
have
global
identities
for
the
state
process
or
where
state
folks
at
replicas
can
communicate
across.
That
is
how
I
have
tried
to
be
it
in
the
document.
I
guess.
A
My
hope
would
be
is
so
there
is
one
have
one
document
for
each
with
the
intent
of
finishing
them
separately.
I
finished
the
simpler
case
and
we
review
it
and
we,
you
know,
merges
and
then
assuming
there's
time.
Then
you
work
on
the
expanded
like
global
use
case
and
we
merge
it
separately
like
having
that
one
day.
A
It's
hard
to
reason
about
because
it
staple
sets
are
complicated
as
it
is,
and
once
you
have
people
sort
of
up
to
speed
on
you
know
the
simple
use
case.
It's
likely
to
get
Monday
deterrent
back
on
the
more
complicated
one
if
they're
mixed
and
then
people
just
have
a
lot
of
complexity
to
try
to
master.
I
understanding
you
understand
that
complexity,
but
I
mean
take
pity
on.
D
I'm
jealous
actually
the
first
motion
of
or
the
first
phase
of
implementation,
that
I
am
proposing.
Doesn't
the
doesn't
propose
itself
to
be
that
complex
one,
and
second,
is
that
the
main
point
that
I
am
proposing
in
this
design
is
having
multiple
identities
for
each
replica
of
the
state
posit.
That
is
the
main
point
of
the
design
that
is
additionally,
we
need
to
implement.
D
A
Point
of
having
use
use
case
the
separable,
or
so
that
people
can
reason
about
it
without
having
to
have
all
the
state
in
their
head
I
guess
it
uses
the
designer
implementer
you
understand
what's
going
on,
but
for
other
people
we
have
a
lot
of
other
responsibilities
having
to
like
get
to
where
your
head
space
is
is
challenging.
If
there's
too
much
work
involved,
that's
my
sort
of
perspective
I'm
happy
to
hear
contrary
and
views.
If
anybody
else
wants
to
speak
up,
you.
D
A
It
clearer
and
more
sort
of
digestible
if
the
use
cases
were
separated,
so
you
can
focus
on
the
simple
case.
You
could
lay
out
what's
required
and
then
people
could
provide
their
feedback
on
just
that
and
then
separately.
You
could
ask
for
feedback
on
like
how
to
extend
that
design.
So
it's
like
it's.
You
know
the
foundation
and
then
you
build
on
top
of
it.
But
if
you
just
try
to
build
the
whole
house
at
once,
it's
hard
for
people
to
really
understand
or
interest
about
so
much
you're
trying
to
do
what
you're
proposing
I.
D
No
I
I
completely
get
it.
What
you
are
saying,
but
from
the
point
of
view
of
what
you
are
saying,
is
that
we
build
the
simpler
case,
which
is
we
have
partitioned.
Stateful
sets
possible
as
local
set
full
sets
in
local
clusters,
and
then
on
top
of
that,
we
say
that
we
can
have
global
communication
across
also
Oh.
What.
A
E
Read
the
document.
One
thing
that
bothers
me
about
sticking
to
single
separate
documents
is
designing
for
a
simpler
case
and
we
all
agree.
We
agree
that
a
reason
that
that's
the
right
thing
to
do
and
later
we
find
that
that's
a
dead
end
and
that's
one
of
the
reasons
why
I
think
it
should
be
in
the
same
dog.
Maybe
we
need
not
implement
the
whole
thing
in
one
goal,
but
at
least
we
should
know
what
is
going
to
happen
after
the
implementation
simpler
case.
So.
E
D
E
D
One
more
point
to
add
to
that
is
that
there
are
even
more
complex
cases
as
model
where
you
are
pointing
it,
which
are
actually
of
which,
which
complexity
will
be
included
with
the
possibility
of
upgrade
feature
or
set
full
setup
grace.
That
is
being
done
right
now,
so
that
I
have
suggested
that
we
will
handle
in
a
separate
design
document,
because
eventually,
then
we
have
a
braid
in
local
state
so
said,
we
will
need
that
in
citation.
Also.
B
B
D
D
So
as
I
mentioned,
the
main
pointer
or
the
main
neat
of
the
design
is
that
we
suggest
that
a
given
state
full
replica
can
have
multiple
identities
right
now.
It
has
a
single
usable
identity
in
locally
in
a
cluster,
we
say
that
that
identity
could
be
more
than
one
or
more
than
one
DNS
identity
could
map
to
the
same
IP
address,
but
if
I
remove
that
out,
then
actually
there
is
nothing
remaining
in
the
design.
Artsy
well.
A
I
mean
I'd,
be
laying
the
groundwork
for
future
work.
I
think
my
concern
is
that
I'd
like
to
see
a
correlation
between
a
feature
proposal
and
what
actually
gets
implemented
and
difficult
to
sort
of
draw
those
links.
If
there's
just
a
big
huge
amorphous
talk
of
everything
you
could
do,
and
then
the
reviewer
has
to
sort
of
interpret
like
what
your
intent
is.
When
you
write
code,
it
has
to
sort
of
try
to
figure
out
okay.
A
If
I
don't
really
have
the
background,
if
I
don't
really
have
you
know
the
rationale
for
why
you're
doing
what
you're
doing
and
then
I
see
a
bunch
of
code,
it's
really
confusing
I
mean
it's
hard
enough
to
go
through
code
and
deal
with
that
level
of
complexity,
the
implementation
side
without
clear
indication
from
a
design
document.
As
to
you
know
the
rationale
for
why
you're
doing
what
you're
doing
so,
that's
my
concern
and
I
apologize
that
it's
I'm
kind
of
picking
on
you
a
little
bit.
This
is
sort
of
a
general
statement.
A
I
think
that
it's
important
that
we
start
focusing
more
on
doing
a
little
bit
of
upfront
design,
reducing
the
complexity
so
that
we
can
actually
read
an
about
the
code
we're
implementing,
rather
than
doing
all
that
work
on
the
coding
level,
where
we've
kind
of
trying
to
solidify
things.
But
the
foundation
like
the
conceptual
foundation
may
be
lacking
so
make
sense.
D
So,
as
I
said,
I
I
agree
to
what
you
are
saying
that
there
are
two
separate
additional
I
mean
that
separate
things
with
epic
use
cases
which
sort
of
build
together.
But
what
I
am
trying
to
say
is
that
the
actual
proposed
design
is
going
to
be
implemented
as
one
shot.
That's
what
I
understand
and
it's
like
correlated
otherwise
are.
E
D
A
D
A
D
A
D
A
F
D
Because
I
get
my
rules,
concern
is
that
he
would
basically
want
to
map
where
and
even
to
what
get
submitted
as
the
implementation.
But
my
point
is
that
the
one
document
might
be
sufficient
and
I
can
probably
put
in
some
pointers
or
add
on
data
which
might
say
that
in
first
phase
of
the
first
PR
that
we
can
yes.
F
A
Just
a
little
bit
more
conceptual
clarity,
whether
you
reformat
this,
but
it
like
to
me
like
when
I
was
when
I
was
reading
and
reviewing
and
I
mean
you
I
appreciate
that
you've
added
more
use
cases
I
still
find
there's
a
lot
of
sort
of
Occidental
complexity
and
how
its
explained
and
like
I
want
to
understand
it.
I
really
have
to
digest
it
more
than
I'd
like
to
and
then
like
I
said
I'd
be
curious.
If
I
was
the
only
one
who
had
that
feeling,
I
just
I'm
kind
of
slow,
I
guess
yeah.
D
A
All
right,
so
it
just
sounds
like
yeah.
More
review
is
required,
so
I'm
going
to
suggest
I
apologize,
but
maybe
we
can
just
push
this
to
next
week.
It
sounds
like
other
people
need
to
still
read
the
document
and
I
need
to
provide
more
substantive
comments
rather
than
I
to
complicated.
So
you
can
actually
respond
to
that.
For
you
back.
D
A
D
Yeah
that'll
be
great,
that's
actually
what
I
have
been
looking
for.
It
has
been
like
open,
the
design
has
been
put
there
being
there,
and
it
has
been
out
for
the
view
since
almost
two
three
months
now
so
I
I
just
need
a
little
more
focused
review
for
that
and
it'll
be
great.
If
you
can
do
it
outside
this
meeting,
it
will
save
time
for
most
of
us.
E
One,
given
that
we
are
actually
postponing
the
design
review,
I
had
a
question.
My
understanding,
so
many
pages
design
review
was
that
you
would
come
which
use
it
and
walk
through
your
design.
For
one
of
those
use
cases
I
tell
you
how
you
hired
a
bunch
of
use
cases,
but
I
did
not
see
a
walk.
I
would
still
like
to
see
part
because
okay.
E
I
don't
have
like
I,
don't
have
clarity,
basically
I,
don't
understand.
I
did
not
still
understand.
Maybe
just
me,
I
still
do
not
understand
how
they
design
match
to
running.
For
example,
REO
HPD
question
globally
or
running
a
real
gas
and
rock
buster.
So
taking
one
example
and
walking
through
the
design
would
be
great.
Okay,.
D
D
B
A
Like
most
of
them
are
gone,
but
let
me
just
pull
up.
We
just
have
to
mate.
We
have
to
make
sure
that
we
actually
have
people
tasked
with
merging
the
remaining
things
or
more
punt
them.
So
we
have
one
do
not
add
unique
label
to
Daniel
set
looks
like
the
tests
are
actually
pouncing.
I
do
I,
think
you've
been
reviewing.
That.
Are
you
able
to
make
sure
that
gets
merged
generally.
E
A
Lgt
end
it,
it
did
not
merge,
however,
because
the
release
time
is
needed,
so
he's
added
a
release
now,
and
you
may
want
to
review
that
it
doesn't
really
look
great
to
me
just
to
see
your
feedback
but
yeah,
so
I
think
those
are
the
only
outstanding
issues
for
one,
seven
and
I
think
there's
minimal
work
required
more
just
procedural.
This
point
I.
B
B
So
it's
a
workaround
for
the
user,
but
is
it
a
bug
that
can
be
addressed
yeah,
but
it's
a
design
issue
that
can
be
addressed.
Minions
Alex.
We
can
talk
about
it
afterwards.
I
don't
think
it's
worth
in
the
whole
thing
into
it,
because
it's
somewhat
of
a
minor
issue
with
how
the
how
do
those
getting
them
works.
C
B
F
C
Federation,
so
we
had
four
four
instances
where
the
Federation
he
tough
sweet
was
turned
off
out
of
the
four
three
of
them
were
because
of
tested
current
changes.
So
there's
probably
some
discussions
we
had
when
we
finished
releasing
one
seven
to
at
least
bring
up
with,
seek
release
about
having
testing
for
make
changes
water
while
we're
trying
to
get
everything
stuff
in
this
cases
impacted
us.
So
we
look
good,
but
it
wasn't
necessarily
our
fault.
C
The
other
bug
was
an
unfortunate
case
of
a
demon
sent
regression
that
got
in
and
he
got
in
I'm,
not
sure
the
timing,
but
I
think
he
might
have
got
it
in
a
while
blocking
was
disabled
because
of
the
testing
stuff
excuse
so
hopefully
we'll
the
coming
days
will
be
to
better
better
outcome,
but
the
the
true
percentage
of
the
staff
that
we
keep
for
eg
test
suite
quality
are
actually
pretty
high.
I
would
say
higher
than
95
percent.
If
we
abstract
the
cases
that
were
broken
by
Justin
from
pose.
B
No
sugar
no
go
ahead.
We
also
there
was
an
issue
with
them
with
the
federated
types.
Ete
test,
mat,
mafioso
cell
that
had
to
do
with
go
the
way
that
go
and
ginkgo
interact
with
with
the
loop
variable
and
for
loops
and
closures
and
pen
scoping,
which
was
causing
us
gears
to
run
the
same
context
or
a
potentially
run
each,
although
in
parallel
run
federated
types
test
for
the
same
type
over
and
over
again,
rather
than
four
different
types,
so
that
is
now
fixed,
which
is
which
is
a
good
thing.
B
E
Okay,
I'm
heading
out:
okay,
yeah,
okay,.
B
B
F
F
G
A
Five
of
them
so
on
on
the
spreadsheet,
there's
a
second
tab
on
the
bottom
here
on
the
first
tab:
q,
17,
q2,
there's
the
second
one
is
C
17
q3
another's
driving,
but
you
fit
in
a
spreadsheet.
You
already
have
open.
You
don't
need
a
link
to
it.
You
need
to
go
down
to
the
bottom
and
pick
it
it
from
town.
B
Yeah,
okay,
I
think,
oh,
we
haven't
I,
don't
think
I
haven't
had
a
chance
to
go
in
and
add
anything
to
this
list.
I
will
add
getting
the
distinct
controller
stuff
in
for
replica
sets
and
deployments,
and
probably
in
doing
some
adding
some
tests
for
those
trying
to
figure
out
how
to
make
sure
that
the
test
that
we
have
that
they're
tested
in
the
right
way
doubles
avoid
issues
like
the
bug
that
I
was
looking
at
yesterday
for
deployments.
B
A
A
We've
already
started
on
some
of
them.
I
think
things
like
virgins
queue
I
mean,
so
we
have
certain
priorities
at
Red
Hat
to
try
to
productize.
This
version.
Skew
is
huge.
Documented
upgrade
procedure
is
huge.
These
are
kind
of
blockers
for
us.
We
can't
deliver
something
to
customers
unless
these
are
doable.
I
guess
like
for
anybody,
who's
providing
Federation's
of
service,
it's
less
important,
because
you
can
try
to
just
kind
of
do
it
internally,
but
like
we
have
to
provide
software
that
users
actually
have
to
manage
them.
A
A
We
have
some
customers
who
basically
want
to
create
kind
of
mirrors
in
terms
of
how
things
are
spread,
so
private
appointment
with
nine
replicas
I
actually
want
nine
replicas
in
each
cluster.
So
when
I
add
or
remove
clusters
I'm
not
having
to
rebalance
the
number
of
replicas
I
expect
to
match
the
number
of
clusters
and
the
number
of
replicas
I
want.
Any
truster
should
be
not
a
hard
problem.
It's
just
something
I,
don't
think
that
was
really
considered
up
to
this
point.
A
Another
thing
we
talked
about
previously
mikheil
was
leading
the
effort
around
like
Federation
preferences,
moving
away
from
annotations
Quinton
talked
few
weeks
ago,
I
think
about
really
like
getting
people
together
to
talk
about
this
and
try
to
nail
it
down.
I
think
that's
super
important
because
it
seems
like
it's
kind
of
foundational
and
all
the
work
we
do
sort
of
depends
on
being
able
to
store
preferences,
that
our
federation
specific
and
whether
we
decide
to
use
a
separate
object
like
per
federated
object
or
whether
we
decide
to
have
some
other
way.
A
What
do
we
override
the
types
or
the
thing
to
do
as
as
proposed
and
have
you
know
customized
kubernetes
type
and
supports
federation
fields,
but
that's
another
thing,
I
think
there's
a
lot
of
complexity
involved
in
the
sooner
we
get
started,
the
sooner
we
have
a
chance
to.
Actually,
let
me
make
one
eight
a
couple
things
around
off,
which
is
also
very
important
for
red
hat
I.
A
Think
Christian,
you've
asked
about
you
know
what
we
do
around
like
supporting
LDAP
and
openshift
goddesses
I've
gotten
renewed
calls
for
supporting
delegated
and
sanitation,
which
I
think
is
what
you're
talking
about,
and
there
are
I
think
there's
this
prior
are
in
the
form
of
Service
Catalog
and
how
they
integrate
with
this
journal
off
solutions.
So
I
think
we're
maybe
follow
their
example.
A
Having
not
necessarily
I,
don't
necessarily
think
will
be
a
huge
amount
of
work.
I
think
it
will
mainly
be
documenting
like
how
to
enable
the
integration
that
I
think
the
underlying
kubernetes
api
server
supports
and
will
just
need
to
make
sure
that
that
support
is
carried
over
to
federation.
We
have
documentation
around
using
will
to
set
it
up.
This
roulette
hose
robably,
I
mean
I
maybe
I
can
provide
the
Service
Catalog
Docs
I'll
put
it
in
the.
A
And
the
other
thing
was
supporting
our
back
in
the
Federation
control
plane
directly
I
was
discussing
this
with
Derek's.
The
concern
was
that,
like
the
Federation
control,
plane
doesn't
really
support
our
back
like
because
there's
only
a
subset
of
the
cube
API
that
it
exposes,
and
if
you
want
to
be
able
to
support
our
back,
you
actually
need
to
be
able
to
enable
like
creates,
like
mana,
creating
updating
deleting
whatever
are
back
objects
in
the
Federation
control
plane,
which
kind
of
requires
both
enabling
those
types
in
the
Federation
API
and
also
controlling
their
propagation
I'm.
A
D
A
Like
be
I
think
this
is,
this
is
a
very
important
thing
as
well.
We
have
customers
saying
you
know
we
want
to
be
able
to
use
Federation,
we
don't
want
to
give
full
access
to
everybody
and
it's
kind
of
a
separate
issue.
It's
related
but
separate
to
how
Federation
control
playing
interacts
with
underlying
like
the
number
clusters,
because
fine,
you
control
access
and
you
give
limited
permissions
but
currently
like
if
you
have
access
to
Federation
control
playing,
you
have
Spock.
A
A
Anyway,
but
yeah
so
I
mean
selective
permission,
not
just
at
the
member
cluster
level,
but
also
at
the
Federation
control,
plane
level,
and
so
this
is
obviously
there's
a
lot
of
work
items
that
I'm
putting
on
a
table
here.
I,
don't
expect
that
this
is
something
that
everybody
else
is
interested
in
and
I'm
not
asking
to
be
able
to
do
it.
We
do
have
another
Federation
resource,
hopefully
starting
from
Red
Hat
side
on
26.
In
a
couple
weeks,
we
have
a
couple
more
people
who
are
requisitioned.
A
No
one
will
get
them
and
we
have
other
resources
that
may
be
able
to
contribute.
But
my
goal
and
1-8
is
to
try
to
have
more
sort
of
work
items
that
are
accessible
to
people
who
have
the
time
to
do
them
and
that'll
require
kind
of
laying
out,
what's
required
in
a
little
bit
more
detail,
rather
than
you.
A
A
Yeah,
oh
so
one
more
thing:
we
have
had
calls
for
aggregated
status,
I've
been
sort
of
I.
Did
a
cursory
read
over
of
reading
cube
resources
from
the
API
sir
Federation
API
server
I'm,
not
entirely
sure
that
document
captures
the
need
for
aggregated
status.
It
seems
to
be
focused
around
enabling
access
to
the
member
clusters
via
the
Federation
API,
but
without
really
a
lot
of
detail
or
vision
for
how
users
would
consume
that
and
I
guess
from
my
perspective,
like
a
user
should
be
able
to
say
for
say
a
replica
set.
A
Okay
show
me
this
federated
replica
set
and
I
would
see
the
cube
view,
but
I
would
also
see
its
Federation
status.
You
should
indicate
you
know
how
many
replicas
are
in
which
clusters
and
find
it.
You
know
we
can
get
that
Vav
API,
but
how
does
how
does
user
see
that
it
needs
to
be
easily
accessible,
I?
Think
and
so
I'd
be
happy
to
have
that
folded
into
the
the
reading
cube
resources
proposal.
A
I
just
wanted
to
make
sure
that
that's
captured
I
think
the
the
lack
of
status
or
visibility
of
how
Federation,
like
has
moved
objects
around
between
clusters.
I
think,
that's
I,
think
that's
almost
a
bigger
problem
than
being
able
to
just
access
resources
directly
from
the
member
requesters
or
at
least
a
big
enough
as
big
an
issue.
So
that's
it
I'll
stop
talking.
I
can.
H
Respond
to
the
last
one
there
I
agree:
I
mean
that
is
technically
you're,
not
the
only
one
who's
expressed
that
requirement.
The
current
proposal
is
essentially
that
it
should
be
possible
in
the
initial
version
of
the
API.
What
you'd
be
able
to
do
is
you'd
be
able
to
list
all
of
the
clusters.
Sorry,
all
of
the
let's
say,
replicas
set
as
an
example
that
all
of
the
replicas
sets
of
a
given
name
vistors,
including
the
Federation,
and
so
that
that
is,
you
know
the
simplest
way
that
one
could
get
a
view
of.
H
H
What
we
could
consider
doing
in
perhaps
a
future
revision
is
to
fold
all
of
that
up
into
the
status
of
the
federated
object,
but
that's
not
in
the
current
proposal,
so
I
guess.
A
question
to
this
group
would
be
with
the
initial
that
I
just
outlined.
Would
that
be
useful,
and
and
perhaps
would
that
fully
satisfy
the
requirement.
H
Want
to
not
provide
that
as
an
initial
version
and
go
straight
to
the
more
complex
version,
I
guess,
which
is
creating
a
new
API.
It
would
involve
different
status
for
the
replica
set
object
which
would
incorporate
all
of
the
underlying
stuff
and
I.
Don't
think
that
the
latter
would
be
consistent
with
the
sense
that
there
you
know
there
are
other
cases
similar
cases
in
the
kubernetes
use
case.
Where
you
know
you
don't
the
status
of
of
a
replica
so,
for
example,
does
not
include
all
of
the
pods
that
comprise
that
replica
set
at
the
moment.
H
Okay,
so
so
there
are
numerous
objects
in
kubernetes,
forgetting
about
Federation.
For
the
moment
where
there
is
an
object
in
the
API,
for
example,
a
replica
set
which
has
a
bunch
of
underlying
objects
in
the
case
of
replica
set,
those
are
pods
which
are
not
visible
in
the
status
of
the
replica
set
object.
You
have.
A
H
Not
necessarily
the
underlying
clusters,
you
could
query
the
Federation
for
the
set
of
replicas
sets
in
the
underlying
clusters,
which
underpin
the
Federated
replica
set
in
the
same
way
that
in
the
kubernetes
api,
you
can
query
the
kubernetes
api
to
get
the
list
of
pods
that
underpin
a
given
replica
set.
But
you
don't
get
that
in
the
same
API
response,
you
don't
say
tell
me
the
status
of
this
replica
set
and
part
of
its
status
is.
These
are
the
pods,
and
this
is
their
you
know:
liveness,
probe
status,
etc.
H
E
One
one
of
the
differential
signals
died
in
case
of
kubernetes.
For
example,
replica
said,
the
replica
set
controller
doesn't
actively
track
defaults,
as
in
the
pod
spec
other
than
actually
looking
at
the
number
of
parts
per
second,
whether
the
part
is
dead
or
not,
which
is
one
of
the
reasons
why
it
doesn't
show
the
status
ending
ethical
said,
which
is
which
is
in
contrast
to
what
we
do
in
Federation.
H
E
That's
it
doesn't
actually
track
the
de
spec
of
the
object
or
anything
as
in
if
somebody
goes
and
modifies
the
board
spec
for
one
of
the
parts
created
by
a
replica
said,
the
controller
actual
doesn't
do
anything
as
opposed
to
what
happens
in
Federation,
where,
if
you
go
to
one
of
the
underline
just
doesn't
change
something
in
the
underlying
replica.
Third
sedation
actually
can
fight
back.
E
E
H
E
H
E
H
Guess
I
understand
what
you're
saying:
yeah
yeah:
there
is
a
difference
there
and
the
question
is
whether
we
surface
that
in
the
in
the
status
yeah,
that's
a
good
point
and
I
would
I
think
we
should.
You
know,
probably
have
all
of
these
conversations
in
that
design
document
review
and
rather
than
necessarily
dig
into
them
too
much
here,
yeah,
okay,.
E
B
So
it
sounds
like
there's
a
decent
amount
of
stuff
that
people
that
is
a
1.8
at
this
point.
We
certainly
don't
have
enough
time
to
really
I
think,
go
through
them
in
detail,
any
sort
of
prioritization
or
assignment
so
maybe
next
week
we
can
go
through
in
more
detail
and
plan
to
spend
quite
a
bit
of
time.
C
Yeah
and
we
think
well
too
much
on
the
side
of
the
back
backlog,
but
it
has
gotten
much
larger,
it's
1:7,
so
we
had
item.
We
have
a
lot
of
PR
that
it
should
kind
of
fall
into
place
early
in
the
cycle
and
we
kind
of
at
least
agreed
on
there,
like
almost
up
GTM.
So
my
expectation
is
a
background
should
reduce,
but
it's
generally
not
a
big
trend.
I
know,
there's
a
lot
more
stuff
in
now,
so
I
think
we'll
want
the
reprioritize
across
now
and
want
that
alone.
H
H
H
B
H
The
proposal
was
to
prioritize
buddy
by
the
next
meeting,
essentially
or
and/or
yeah
so
a
week
to
put
stuff
in
there
a
week
to
prioritize,
and
then
you
know
allocate
stuff,
so
we
can
get
get
going
towards
energy.
Is
it?
Is
there
any?
You
know,
organization
that
is
significantly
interested
in
having
items
on
this
list
that
has
not
captured
at
least
the
first
part
of
what
we'd
like
to
work
on
in
188
or
are
we
mostly
covered?
It
looks
like
Red,
Hat
Google
quarrei
have
their
stuff
in
there.
Is
that
true.
C
H
H
H
Maybe
if
there
is
such
a
publicly
visible
taken
by
Google,
maybe
you
could
just
and
where
it
fits
in
here,
because
we
should
put
a
I
guess,
a
clear
message
out
to
the
community
that
they're
quite
a
lot
of
companies
I
mean
great
hat-
is
an
example
of
you
know:
putting
a
lot
of
effort
into
the
Federation
project
for
their
for
their
products
and
I
guess.
We
should
make
sure
that
that
we
care
how
it
relates
to
any
alternative
to
their
because
we've
kind
of
taken.
C
What
would
the
other
one
federated
preferences-
and
these
are
all
discussions
we
have
as
well
with
other
customers
on
the
Cape
of
eBay-
nothing
rates,
to
my
mind
specifically,
but
at
the
I
guess
we
had
our
own
prioritization
of
stuff.
We
would
like
to
work
on.
None
of
these
I.
Don't
think,
there's
anything
to
surface
here
in
this
budget
at
this
time,
but
yeah
I
mean
we're
just
we're
fighting
the
air
next
week
to
help
prioritize
the
the
efforts
that
are
pending
on
especially.
H
I
know
it's
just
going
to
clarify
so
so
I
mean
it's
totally
fine
for
any
company
Google
included
to
be
you
know,
working
on
stuff
that
is
not
going
to
be
open
source,
so
that
might
be
part
of
a
proprietary
product,
ek
or
whatever.
But
I
guess
is
what
you're,
referring
to
other
things
that
Google
is
working
on
that
are
going
to
be
part
of
the
open
source
Federation
product.
Are
they
receipt
or
going
to
be
or
not
either.
H
H
H
C
Main
we
have
a
large
working
groups
across
big
working
groups,
tried
to
tackle
different
problems
that
are
related
between
multiplexer
and
between
that
were
doing
some,
some
of
them
overlay
with
other
GPT
efforts,
so
I
mean
there's
a
lot
going
on.
We
don't
that
we
have
to
draw
the
line
somewhere
and
if
there
is
some
customers,
it's
true,
don't
necessarily
see
where
that
line
is
and
echo.
These
concerns
I
can't
always
so
we
clarify
for
the
purposes
of
the
safe
where
that
line
is
drawn,
that
that
makes
it.
B
A
I
do
actually
have
a
request,
so
when
you
mention
that
eBay
is
actually
rolling
up
Federation
everywhere,
not
necessarily
right
now,
but
I
would
be
very
interested
in
hearing
more
about
how
they're,
using
it
kind
of
curious.
To
my
mind,
I,
wouldn't
necessarily
want
to
deploy
a
Federation
production,
so
I'm
curious
how
they're
doing
and
what
their
rationale
for
doing
it
is.
They
said
something
you
can
share.
H
Sure
I
can
actually
ask
them
to
attend
this
meeting
and
present
that
to
the
group,
I
wouldn't
want
to
speak
on
their
behalf.
My
understanding
and
I,
don't
think
would
mind
me
talking
about
this
here-
is
that
they've
done
considerable
work
internally,
which
they
haven't
necessarily
open
sourced
all
of
it
they
are.
They
have
at
least
philosophically
adopted
Federation
as
their
way
of
managing
the
reasonably
large
number
of
clubs,
because
that
they
have
precisely
what
they've
been
to
make
it
congrat.
H
This
point,
I,
don't
know,
but
they
they
actually
asked
to
have
a
meeting
with
me
and
a
bunch
of
the
engineers.
In
fact,
I
think
it
is
this
afternoon
where
they
want
to
become
more
actively
involved
in
the
open
source
part
of
this
project
as
well.
So
you
can
imagine
them
joining
us
shortly
and
yeah.
They
should
be
able
to
give
more
detail
themselves.
Cool
cool.
C
H
I
have
had
that
conversation
of
the
Meridian
they
apologize
for
not
having
been
acted
more
actively
involved
here.
I
think
it's
basically
being
driven
by
you
know
very,
very
tight
deadlines
on
their
side
and
they
just
haven't
had
the
cycles
to
be
involved
in
this
group.
They
just
had
to
get
their
own
stuff
out,
but
but
they
are
aware
that
this
is
not
an
ideal
situation
and
I
believe
they
are
trying
to
address
that
by
being
more
actively
involved
here
in
the
coming
weeks,
cool.