►
From YouTube: SIG Architecture 20181115
Description
http://bit.ly/sig-architecture
tl;dw
- Charter merged
- Windows GA deferral discussion (not going GA in 1.13, need to refine this process)
- Matt Farina added as co-chair
- Contributor summit update
- No meeting next week
A
Later
Aaron,
as
well
as
one
of
the
cig
architecture,
members
Brendon
still
on
planes
back
from
China.
So
if
there's
discussion
that
they
need
to
be
involved
in,
it
might
make
sense
to
take
some
of
this
over
to
the
release
meeting
tomorrow,
just
because
I
haven't
heard
from
either
of
them
on
this.
On
this
topic.
Okay,.
B
B
C
B
D
Right,
hi,
everybody
so
me
and
Patrick.
The
other
co-chair
for
sick
windows
are
here
so
we've
had
a
lively
discussion
yesterday
and
I
see
brands
here
as
well
around
the
sip
windows
and
eligibility
for
going
with
the
stable
designation,
1.13
there's
like
two
or
three
different
things
that
were
mentioned
in
in
the
thread
and
I
want
to
concentrate
very
very
specifically
into
what's
the
availability
bar
that
you
can
have
for
kubernetes
for
adding
a
new
feature
in
this
case.
D
It's
an
optional
feature,
and
but
it
is
a
major
feature
of
kubernetes,
so
I'm
gonna
start
from
at
least
the
Brian
started
right.
So
there
has
to
be
adequate
non
flaky
tests
enabled
as
part
of
the
release
blocking
suite
the
second
one
was.
Reliability
needs
to
be
high
and
I.
Think
that
the
way
we
measure
ability
to
a
certain
extent
in
the
kubernetes
community
is
using
those
tests
need
to
be
passing
and
have
the
repeatability
around
being
them
them
being
clean.
D
There
has
to
be
adequate
end
user
and
admin
documentation
and
then
the
last
one
is
compatibility
can
be
broken
for
existing
users
clusters,
both
for
today,
as
well
as
moving
forward
when
we
kind
of
look
at
that
list,
I'm
going
to
start
from
the
bottom,
because
that's
the
easiest
to
address
so
so
we're
not
breaking
compatibility.
You
know,
there's
an
optional
feature:
existing
apps
continue
to
work.
D
Moving
forward,
continue,
maintaining
our
support
for
Windows
there's,
no
real
big
API
changes
here,
so
we
don't
have
to
be
as
concerned
as
other
features
that
there
are
primarily
API,
API
driven
or
they
introduced
significant
changes
to
the
API
and
on
the
topic
of
documentation.
We
are
actually
that's
not
to
do
this.
We
are
gonna.
Do
it
we're
gonna,
have
good
documentation
when
we
exit
in
1
to
13
and
shown
that
in
the
past,
both
are
better
relation
or
alpha
that
we
actually
pulled
through
for
the
documentation.
D
We
know
we're
behind
now,
so
there's
no
visible
docs
that
someone
can
read,
but
the
dogs
deadline
was
the
19th
and
I'm
not
mistaken,
and
our
goal
was
to
actually
have
something
out
there
for
for
reviewing
and
start
approvals
by
then
so
so
that
kind
of
takes
care
of
the
last
couple
of
bullet
points
and
I
want
to
come
down
to
the
test.
Any
questions
on
those
before
we
we
get
down
to
the
test.
I.
C
Would
say
that
part
of
that
list
is
not
complete.
I
think
you,
the
current
needs
to
reproducible
ways
to
deploy
these
these
this
type
of
environment
right
because,
right
now
there
is
no
there's
no
documented
means
by
which
that
I'm,
aware
of
that
actually
outlines
for
an
independent
consumer
to
be
able
to
reproduce
an
environment
or
support
any
testing
that
is
also
missing
bred
so
that
that
piece
needs
to
be
there
as
well.
Yes,.
A
There
are
two
different
documents
I'm
available
on
that
I'm
actually
is
more
than
two,
but
the
ones
that
were
using
for
regular
testing
right
now.
One
of
them
is
from
Peter
hore
neck,
who
has
updated
the
scripts
that
are
already
in
tree
to
be
able
to
build
a
cluster
on
GCE
and
then
the
other
placeholder
that
we
have
and
the
documentation
links
out
to.
Of
course,
a
similar
project
on
ACS
called
ACS
engine
that
creates
a
cluster
on
Ana's.
A
A
C
So
my
biggest
concern
here
is
from
a
cyclist
or
life
cycle
perspective.
Is
there
are
many
tools
to
deploy
clusters
right
and
we
have
not
been
consulted
on
any
of
this
at
all.
So
when
people
come
to
us
wanting
to
deploy
Windows
clusters,
we
need
to
be
able
to
point
them
in
a
canonical
location.
It
says
these
are
the
tools
you
can
use.
These
are
the
things
that
are
supported.
These
are
the
things
that
are
not
supported.
A
Yeah
and
so
that's
why,
in
that
documentation
that
we're
working
on
we're
linking
to
this
stuff,
that's
available
from
the
cloud
providers
that
can
set
this
up,
may
weren't
right
for
not
asking
people
to
use
a
specific
tool,
I
cube
a
DM
or
one
of
the
existing
ones.
That's
there
we're
asking
people
to
to
work
with
their
dystrophin
tour
on
that,
and
we've
got
two
different
examples
already.
There
is.
A
C
That
means,
if
there's
a
problem,
this
is
what
people
will
do.
They
will
say
a
feature
is
GA.
They
will
use
all
the
tools
to
try
and
understand
to
deploy
that
feature
and
unless
the
standard
patterns
exist
across
these
tools
and
these
other
tools
have
bought
into
them
like
whether
or
not
you're
gonna
use
paints
and
Toleration
zyk
Ross
this
across
all
the
different
deployment
mechanisms.
People
will
try
it
and
they
will
come
to
us
asking
the
questions
and
you
send
them
back
to
us.
Send
them
back.
C
C
To
understand
that
people
will
use
a
set
of
tools
and
to
make
sure
that
documentation
is
also
clear
and
the
expectations
are
clear
for
the
deployment
patterns
that
will
exist.
So
that
way,
it's
clearly
understood
from
other
parties
who
are
basically
the
ones
who
get
all
the
filter
for
when
things
go
wrong
that
that
they
have
a
routing
plan
as
well
as
have
had
buy-in
or
say
and
some
of
the
pieces.
So
that
way
they
can
actually
make
changes
and
other
tools
that
they
need
to
write.
D
Understand
so
we
haven't
had
that
discussion
right
so
I
mean
you
guys
were
aware
that
were
working
on
Windows
right,
so
I
mean
we
went
better
last
year
and
you
know.
But
yes,
we
did.
We
didn't
come
to
see
cluster
lifecycle
and
have
that
discussion.
That's
correct,
but
so
far,
so
let's
so
that
that's
all
for
right,
so
we
didn't
do
it.
But
in
the
past
you
know
when
we
went
bad,
that
we
had
a
lot
of
users
trying
this
well
actually,
a
pretty
big
community
on
slag
and
I.
D
Don't
think
a
single
one
of
those
guys
came
to
see
cluster
lifecycle
asking
for
guidance
or
support
or
the
elf.
We've
actually
said
that
we
don't
support
other
genius
clusters
and
we've
said,
and
that
might
be
fine
and
you've
heard
that
but
I
don't
think
you
know
when
they.
These
guys
need
support.
They've
come
to
our
to
our
sig
and
to
our
slide
channel
and
they've
gotten
that
support.
D
So
yeah
I
understand
that
you
know
you
know
we'll
work
with
you
guys
to
support
hybrid
environments
and
and
and
kind
of
smooth
that
that's,
of
course
stemming
from
a
kubernetes
community
standpoint,
but
I'm
not
as
worried
about
customers
having
trouble
and
not
finding
the
right
place
to
get
support
or
or
how
do
you
figure
things
out,
they'll
come
to
us.
I
also
have
one
more
thing
to
add
to
what
patrick
said.
Yes,
we
have
something
for
gke
and
and
agile
for
how
to
deploy
kubernetes
cluster
to
the
Windows
nodes.
D
A
B
C
I
just
want
to
make
sure
that,
like
here's,
here's
my
major
concern
and
my
beef
with
all
of
this-
that
we
do
get
questions
we
do
have
to
route
them
and
we
say
it's
unsupported,
but
it's
a
g8
feature,
so
that
leads
itself
to
ambiguity
from
the
community's
perspective
right.
I,
I,
don't
mind
doing
these
types
of
things.
So
long
is
that
the
other
SIG's
who
will
be
affected
are
involved.
There's
other
question
of
like
artifacts
right,
usually
upstream
owns
the
artifacts.
C
E
Can
I
was
just
gonna
ask
very
quickly
if
Tim
is
there
a
list
of
things
that
we're
asking
for
here,
cuz
I
heard
you
know
there
are
some
things
we
don't
have
we'd
like.
Can
we
can
it
maybe
form
that
into
a
list
and
see
if
it's
the
things
that
we
can
go
get
checked
off,
so
we
can
move
along
with
GA?
Is
it
would
that
be
a
good
way
to.
F
Yeah
I
I
agree
that
having
a
checklist
or
the
GA
requirements
would
be
helpful
and,
looking
you
know,
I'm
also
trying
to
make
sure
that
we
have
the
right
experts
from
cluster
lifecycle
or
node
or
whatever,
who
are
looking
at
this.
But
you
know,
as
I've
been
hooking
into
this
the
past
few
weeks,
putting
quite
a
bit
of
time
into
it.
F
You
know,
and
the
beta
release
a
year
ago
was
not
even
alpha
quality,
in
my
opinion.
So,
given
how
short
this
release
cycle
is
and
how
many
people
are
going
to
be
away
for
Q,
Khan
and
holidays
and
other
things,
the
risk
of
this
feature
is
super
high,
but
it
just
does
not
meet
the
bar
right
now
and
my
confidence
that
will
meet
the
bar
without
putting
the
risk.
The
whole
release
at
risk
is
extremely
low,
so
I
think
right
now.
F
D
D
Future,
but
we
had
documentation
for
a
better
feature
without
actually
multiple
people
that
were
tiny
doubt,
and
you
know
that
actually
deployed
Windows
clusters
and
within
the
kubernetes
environment-
and
you
know,
we've
been
talking
to
these
folks
for
the
last
one
year
so
that
that
existed
so
I
understand
that
you
saw
that
the
piada
do
have
four
dogs.
Were
we
basically
eliminated
everything,
and
it's
just
basically
outlined
right
now,
like
the
sections
our
dogs
existed
before,
people
were
using
around
simple
scripts.
D
F
Documentation
I'm
talking
about
is
not
the
documentation
of
how
to
connect
Windows
node
to
a
cluster,
but
it's
the
documentation
for
any
users
that
want
to
understand
what
should
work
and
what
should
not
work.
And
what
are
the
differences
and
from
what
I
hear
you
know,
there
were
a
lot
of
things,
especially.
F
D
F
D
Can
talk
about
the
test,
a
squirrel
right
so
Patrick
us
some
feedback
on
that,
but
you
know
I,
don't
know
Matt
had
raised
his
hand,
so
I'll
defer
to
him
in
a
second.
But
you
know
when
we
get
feedback
like
that,
it'd
be
great
if
he
actually
gets
routed
to
the
C.
So
if
you
were
getting
feedback
that
certain
things
were
not
working,
I
would
love
to
actually
hear
that
both
just.
F
E
D
D
D
B
You
know
I'm
super
rooting
for
you
guys
in
this
work,
I
think
it's
amazing
what
you've
done
and
and
it's
frustrating,
because
when
I
think
about
the
release
cadence
and
how
important
it
is
to
maintain
stability
before
the
end
of
the
release,
we
have
to
make
some
really
heartbreaking
decisions
about
what
goes
in
and
what
doesn't
and
the
impact
on
the
release.
Team
of
the
overall
release,
quality
and.
D
Be
aware,
this
is
an
optional
feature
right,
so
if
he
doesn't
get
turned
on
it
doesn't
impact
anybody,
and
from
the
moment
the
hybrid
clusters
are
not
in
the
mix
you're
actually
guaranteeing.
Now
that
the
only
ones
are
going
to
try
say
windows
are
the
guys
are
gonna,
have
a
a
master
components
in
Linux
with
Windows
computer
notes.
No.
B
G
C
G
Sure
I
would
like
to
see
some
documentation
on
if
you
turn
on
Windows
clusters,
and
you
want
to
run
the
conformance.
What
do
you
need
to
switch
things?
What
do
you
need
to
do
to
switch
off
so
that
the
conformance
tests
will
pass,
for
example,
right?
That
was
one
of
the
things
that
I
was
looking
for
as
well,
even
before
yeah.
D
And
if
you
remember
that's
what
Patrick
is
working
on
right,
they
were
trying
to
isolated
the
set
of
test
cases
that
are
basically
would
like.
You
think
think
of
this,
as
like
kind
of
three
streams
right,
Stream
number
one
is
you
only
have
a
Linux
environment,
you
run
all
the
test
cases
everything
passes
string.
Number
two
is
basically
you
subset
the
test
cases
to
only
the
ones
that
work
on
window
guaranteed
to
work
on
windows,
and
you
run
them
on
a
Windows,
only
cluster
and
those
paths.
D
A
Yeah
and
so
the
work
on
that
third
stream,
we
were
asked
to
move
that
out
to
14
pending
the
discussion
with
suit
conformance,
and
so
the
workflow
that
we've
been
using
is
the
second
one
for
the
windows-only
cluster.
They
had
when
we're
running
the
subset
of
the
tests
and
that's
because
all
the
rest
of
the
changes
that
we
needed
for
the
other
scenarios
were
asked
to
move
out.
So
just
clarifying
that.
F
E
E
A
pad
on
and
I
see
things
like
daemon
set,
rolling
updates,
doesn't
work
is
failing
tests,
and
so
I'd
really
want
to
see
some
of
these
apps
things
that
people
expect
to
be
green
or
a
good
reason.
If
you
know
they're,
not
gonna
be
they're
gonna
be
I,
mean
daemon
set
not
having
a
working
rolling
update.
I,
don't
know
that
we'll
have
this
stuff
fixed
by
the
end
of
the
day
tomorrow,
and
doesn't
that
mean
it
needs
to
get
pushed
to
the
next
release
cycle.
D
I
think
you
know
what
Patrick's
opinion
is
bad.
Let's
say
we
actually
do
get
this
test.
You
know
and
we
get
all
these
things
fixed
all
right,
so
it
would
not
give
me
we
have
potential
tests
are
important
and
impactful.
They
are
not
done.
Then
I
agree
with
you
guys
weren't
already,
but
what?
If
we
actually
do,
if
we
get
the
tests
to
pass
and
green
they're,
the
ones
are
that
the
shoe
past?
Are
we
good
then.
H
H
That's
not
a
better
quality,
so
so
so
reason
we
work
together
with
the
single
windows,
so
we
at
the
beginning,
I
think
we
we
have
that
the
requirement
I
did
for
the
windows
container
support
not
as
the
actual
feature
I
did
to
the
kubernetes
community
and
if
you
cross
the
architecture
and
the
signal
and
the
Signet
work
and
a
sick
storage
is
set
up.
So
that's
how
the
I
think
this
is
also
proposed,
be
name
is
Patrick
also
went
to
the
cig
architecture,
and
this
is
a
kind
of
women
to
exact
time.
H
So
so
we
we
are
working
on
a
beanie
sigel
when
those
support
is
try
to
add
to
this
windows
kind
of
form,
extended
kubernetes
platform
and
to
include
windows
support
in
the
one
got
the
14,
but
in
the
one
that
asserting
they
only
call
the
windows
container.
Support
da
I
want
to
make
that
more
clear,
I'm
here,
so
so
so
only
if
you
talk
about
the
ga,
of
course
we
do
capture
those
testing
and
what
kind
of
feature
support
that
is
a
separator
I'm,
not
to
say
I'm
signing
up
for
the
year
yet.
H
D
I
mean
so
so
we
are.
We
are
basically
getting
to
the
point
now
where
we're
saying
that
every
signal,
six
cluster
lifecycle,
Sagarika
diction,
has
to
give
us
a
for
us
to
go
to
GA
and
what
I'm
hearing
across
the
board
is.
You
know
everybody
is
coming
up
with
their
own
idea
of
the
quality
metrics
that
we
want
to
build
and
the
quality
metrics
that
you're
trying
to
go
to
GA
and
I.
D
Think
what
you
guys
are
saying
is
that
you're
not
approving
and
if
we're
gonna
make
that
as
the
quality
gate
for
growing
GA
is
that
we
get
approval
from
all
those
six
and
you
guys
say
no,
then
there's
no
reason
for
us
to
find
it.
You're
basically
would
just
say
scratch.
It
I
mean.
Am
I
hearing
this
differently.
B
I
Thing
was
like
I
think,
maybe
looking
at
the
documentation
for
all
the
workload
stuff
for
the
policy
stuff,
it's
a
huge
amount
of
documentation
to
update,
including
the
examples
knowing
which
ones
work
trying
to
communicate
to
the
end
users.
What's
gonna
work
on
Windows
and,
what's
not,
is
probably
like
I
think
it's
a
rather
extensive
documentation
update
it's
not
straight
forward
and
small.
D
No
but
I
did
everything.
What's
not
working
right,
we're
gonna
tell
them
what's
gonna
work
like
we're
gonna
give
people
the
the
blueprint
for
how
they
can
actually
start
running,
Windows,
Server
containers
and
kubernetes,
and
the
things
that
we
know
work
very
well
and
have
been
tested
and
that's
what
we're
gonna
start
with
it's
a
v1
feature
right.
A
going
stable
doesn't
mean
that
you're
fully
functional
and
you're,
not
gonna,
iterate,
that
on
that
later
on,
it
is
an
MVP.
D
If
certain
things
who
work
they'll
be
able
to
deploy
their
pods,
will
they
be
able
to
run
stateful
sets
no
we're
gonna
explicitly
say
we
don't
support
that.
Will
they
be
able
to
excuse
some
other
features,
know
what
game
gonna
say:
there's
certain
things
that
just
won't
work
like
you
know
you
don't
have
calico
Falana
hostgator
the
exelon
doesn't
work.
If
there's
a
big
list
of
things
that
we're
gonna
say
it's
not
supported,
but
you're
gonna
give
them
the
blueprint
for
the
things
that
they
do
work
so
that
can
actually
get
started.
I
Okay,
I
mean
I,
just
been
communicating
that
clearly
and
concisely
so
that
we
don't
get
a
lot
of
requests
right
into
the
Sega
Blatt
Channel
we
don't
get.
A
lot
of
repressed
runs
hey.
How
does
this
not
work?
We
don't
get
a
lot
of
issues
and
about
like,
basically,
all
my
work
loose
and
work
just
if
we
know
that
we
have
a
lot
of
gaps
for
this
feature.
Why
are
we
rushing
it
in
like?
Why?
Don't
we
just
wait
a
quarter
and
address
all
the
gaps?
What's
the
benefit
to
rushing
it
in
right
now?
A
There,
the
thing
I
want
to
ask
you
is
I,
mean
I,
appreciate
all
the
feedback
on
all
this.
It
looks
like
there's
a
lot
more
people
that
want
to
give
feedback
in
this
process.
I
think
what
I'd
like
to
ask
from
the
release
team
specifically,
is:
we
need
to
get
a
specific
list
and
have
a
date
where
that
list
is
closed.
No
more
new
work,
adjectives
that,
after
that,
because
to
be
brutally
honest
here,
I
feel
like
every
time.
I
come
to
this.
A
There
are
more
people
creating
work
than
there
are
people
that
can
accept
the
work,
which
means
we
can
never
succeed
if
the
target
always
moves
and
I
feel
like
that
is
a
project
management
issue
that
we
need
to
get
on
top
of,
and
so
I'd
really
like
some
collaboration
to
get
that
list
crystal
clear.
So
we
can
so
we
can
get
this
done
and
everybody's
comfortable,
because
these
discussions
could
go
on
forever.
H
Earlier
I
didn't
see,
but
I
also
didn't
see,
I
think
sign-off
actually
there's
the
discussing
at
the
signal
and
the
way
to
agree,
and
we
need
a
clear
name
to
say
what
limited
front
of
signal
the
perspect
we
do
agree
and
a
lot
of
other
know
the
perspect
and
it
is.
We
are
okay
and
we
want
the
communication
and
clear
tasks.
So
this
basically
repeated
so
many
times
and
but
I
have
to
say
those
data.
H
Yet
so
so
I
cannot
just
sang
off
at
this
moment,
but
I
didn't
say:
oh
I
won't
sign
off,
so
I
just
want
to
clarify,
because
we
do
agree
at
the
beginning
of
the
planning.
Half
year
ago
we
said
we're
okay
with
limitation,
about
which
continent
time
and
an
imitation
about
the
which
she
windows,
the
server
and
the
limitation
of
the
certain
feature.
We
all
agree
on
those
kind
of
things,
but
I
didn't
say,
look
those
I
didn't
say
past,
but
I
didn't
see
those
documented
all
right.
D
There's
no
documentation
right,
I
mean,
like
I,
think
done
right
before
you
join
I,
explained
that
being
so
underwater
trying
to
get
all
our
tests
and
migrating
everything
into
a
screen
that
we
haven't
had
time
to
sit
down
to
write
the
dogs.
But
you
know
dogs
is
like
a
two
three
day
effort
to
me
and
Patrick
or
or
we
can
actually
knock
that
out
and
have
really
good
documentation
about
what
works
step
by
step.
What
do
they
have
dogs
on
how
to
deploy
these
on
on
GCE
ANSI
ball
with
oviya?
D
No,
yes,
as
well
as
I
sure.
So,
yes,
we
are
I,
mean
I,
don't
know
if
dogs
is
a
gating
factor,
then
we
are
actually
I'm
I'm
a
little
sad
about
our
process
and
this
discussion
today,
but
I
do
agree
that
the
tests
and
then
passing
or
not
passing,
is
I.
Consider
the
gating
factor
and
I
agree
with
most
of
the
comments
that
people
said:
I
don't
want
the
dogs
to
be
that
we
can
fix
that
and
we
have
a
month
a
lot
of
months.
We
have
well
have
at
least
like
in
a
few
days.
D
D
I
mean
we
don't
want
to
keep
rehashing
this
over
and
over
again
seems
that
we
have
critical
concern
from
a
lot
of
different
folks
on
the
state
of
the
test.
Our
goal
was,
you
know,
working
as
many
hours
as
possible
to
actually
get
all
those
to
be
green.
Yes,
you're
right
or
not,
anticipating
code
changes,
but
on
the
other
hand
you
know
we're
trying
to
decipher
those
tests
and
flip
the
bit
on
the
ones
that
should
be
passing
what
the
ones
that
shouldn't
are.
D
If
someone
asked
earlier,
what's
the
goal
for
trying
to
rush
this
we're
not
trying
to
rush
this
right,
we
are
you
know
it's
we're,
always
trying
to
target
GA
around
this
timeframe
and
being
clear
from
that.
From
from
last
year,
we've
announced
add
on
the
kubernetes
community
meetings
of
announced
at
and
we
talked
about
it.
One
went
to
bed
the
last
year,
so
this
is
not
rushing
it.
The
rationale
behind
this
is
that
we're
actually
trying
to
get
users
to
use
it.
D
When
you
have
the
bed
a
designation
or
an
alpha,
there's
a
certain
class
of
people
that
just
you
know,
don't
trust
it
and
now
Barry
I,
don't
want
to
push
this
to
for
the
sake
of
just
going
stable
without
actually
having
the
stability
and
the
reliability
that
people
should
expect.
So
as
Dalek
part
of
it
right
me
and
Patrick
will
never
be
comfortable,
saying
we'll
go
nga
without
that,
really
passing
all
the
tests
and
asking
comfortable,
but
having
that
designation
allows
more
people
to
use
it.
D
F
D
A
A
F
A
That's
you
know
three,
probably
and
other
three
versions
from
now
and
Coupland,
and
so
we're
still
waiting
on
the
new
Windows
Server
2000
19,
to
be
available
on
on
Azure
and
GCE
to
support
the
test
infrastructure.
It's
not
there.
Today,
it's
not
going
to
be
there
by
code
freeze,
which
is
why
I've
said
we're
going
to
commit
to
that
for
version
14.
A
We
don't
have
any
reason
to
believe
that
it
would
actually
require
any
work,
but
that's
just
the
way
the
timing
lined
up,
because
you
know
our
expectation
was
that
version
of
Windows
Server
was
supposed
to
be
out
by
the
end
of
October,
which
was
what
was
told
to
customers
back
in
September
and
that
turned
out
to
not
not
line
up.
Okay.
F
D
D
D
So
I
long
story,
short
I,
don't
want
to
eat
up
the
whole
cigar
kotecha
time.
Are
you
guys
all
ok
with
going
down
the
path
of
saying?
Should
we
get
all
the
tests
to
grunt
Green,
there
ones,
the
sherbet
passing
by
end
of
day
tomorrow
we
can
continue
proceeding
with
a
ga
plan,
provided
that
you
know
we
complete
the
dogs
in
a
short
period
of
time
frame
after
that
other.
F
Than
other
than
an
announcement,
basically,
what
does
the
G
a
designation
affect
about
what
you
plan
to
do
right?
Presumably,
we
need
to
get
tests
working
to
get
the
ending
readings
blocking
sweet.
We
need
to
get
the
documentation
in
place.
All
that
work
seems
independent
of
whether
we
officially
declare
it
here
or
not.
Okay,
so.
D
D
In
a
mech
there
too,
so
we
are
basically
having
folks
are
looking
to
the
test
tweaking
whatever
needs
to
be
tweaked
and
marking
the
test.
That
should
that
should
never
run
on
Windows,
as
in
like
they
very
compatible
or
the
test
is
testing
for
things
that
will
never
work
on
windows
like
stateful
sets
like
one
of
the
things
I
mentioned
earlier,
or
something
that
requires
privileged
containers
or
something
else,
then,
once
you
actually
finish
that
list
to
produce
that
and
what
we
seek
relates
to
tell
them.
D
This
is
what
should
be
passing
on
Windows
and
look
at
this.
Everything
is
green.
This
is
what
should
not
be
passing
on
Windows
and,
as
expected,
it's
red.
If
we
were
to
produce
that
by
end
of
day
tomorrow,
would
you
still
be
eligible
to
proceed
on
a
path
of
koenji
a
provided
that,
once
that's
done
with
the
testing,
we
will
complete
the
documentation.
That's
one
of
the
easiest
things
for
us
to
do.
We
just
have
no
time
to
start
on
it.
Yeah.
G
Right,
if
you
tell
us
a
lovely
tomorrow,
then
we
won't
have
any
time
to
do
those
kind
of
reviews
to
see
which
tests
we
actually
need,
but
are
not
passing
right.
And
if
you
have
to
fix
that,
then,
like
I
said
we'll
end
up
in
eating
into
a
code.
Freeze,
yeah
I
know
my
last
week,
then
it
would
have
been
better,
but
I
know
that
you
guys
are
working
non-stop
and
you
know
night
Sonny,
so
yeah
part.
D
Doing
that
all
right,
so
you
know
we're
gonna
argue
just
for
the
sake
of
argument.
We'll
say
that
Windows
is
not
gonna.
Go
Jane
1
to
13
will
proceed
in
getting
all
our
tests
passing
in
test
grade
and
working
with
sig
release
to
identify
that
who's
gonna.
Take
the
lead
on
producing
the
set
of
GA
criteria
for
any
future,
not
just
for
sig
windows,
so
we
can
actually
start
checking
those
off
for
114
I.
Don't
want
us
to
two
months
from
now
to
get
into
a
situation.
We
actually
start
having
the
same
discussion.
A
C
A
G
What
I
would
recommend
there
is
Patrick
and
you
Michael,
you
had
a
lot
of
feedback
for
the
over
the
last
week
right
if
you
can
turn
that
feedback
into
a
cap
and
say
okay,
this
is
what
we
know.
This
is
what
you
guys
told
us
is
this
enough.
That
would
be
the
kind
of
kept
that
we
all
of
us
should
sign
off
on
like
signal
signal,
ease
and
architecture
will
find
off
on,
and
so
you
will
know
exactly
what
to
do
and.
D
Know
and
one
of
the
problems
that
we've
had
and
Patrick
alluded
to,
that
in
the
in
the
slack
Channel
sorry
in
the
chart
for
zoom
is
that
you
know
we.
We
actually
gonna,
come
up
with
the
release
criteria
and
when
a
document
that
and
we're
gonna
give
every
seed
an
opportunity
to
basically
greenlight
this
yay
or
nay
and
we're
gonna
give
people
a
deadline
because
in
the
past
we've
been
ignored
until
now,
when
we're
actually
getting
blocked.
So
we're
gonna
give
everybody
an
opportunity
to
read.
G
D
You
guys
are
unfortunate
deadlines
on
us
exactly
right
now,
right,
that's
exactly
what's
happening.
Why
can
we
know
them
force
a
deadline
for
providing
feedback
to
the
ciock?
If
you
guys
are
approving
us
moving
forward?
We
talked
about
114
right,
so
if
in
the
next
month
and
a
half,
we
produced
all
the
release
criteria
for
us
when
stable
and
we
ask
the
usual
six
to
either
approve
or
reject
that.
Why
not
be
able
to
give
them
a
time
frame
for
giving
us
an
answer
so.
G
F
B
I
wish
it
did
because
a
lot
of
things
just
languish
as
PRS
forever
and
don't
ever
get
merged
because
nobody
reviews
them-
or
you
know
like
this
happens
all
the
time
and
unfortunately
the
onus
is
on
the
person,
who's
driving
the
technical
work
to
get
consensus
on
it,
and
it
is
I'm
miserable
and
impossible,
but
it
has
to
be
done
because
there's
no
there's
no
way.
We
can
just
tacitly
let
things
drift
into
into
GA
or
into
the
codebase
without
some
sort
of
thorough
review,
but
I
also
want
to
add
something.
B
I
was
going
to
say
this
earlier
and
I
this.
There
there's
a
whole
lot
of
failure
here.
This
the
fact
that
you
guys
are
having
so
much
trouble
with
getting
this
end
and
all
that
is
partly
indicating
that
we
need
to
improve
these
communication
lines
and
also
just
the
process
generally
so
Michael.
What
I
would
what
I
would
suggest
is
team
Patrick?
B
D
B
B
We
need
more
reinforcements
on
this
effort
just
generally
speaking,
and
we
need
reinforcements
on
a
lot
of
things,
but
this
is
specifically
something
where
we
need
to
get
more
critical
mess
around
this
to
vet
the
use
cases
and
to
help
you
at
documentation
and
do
those
things,
because
I
think
this
is
an
incredibly
important
effort
and
have
it
done
in
a
way
on
such
a
shoestring
is,
is
a
disservice
to
the
value
that
you
got
that
you
working
on.
This
is
incredibly
valuable,
so
we
need
to
do
it
right.
B
G
D
I
So
just
one
request:
when
you
do
the
cap,
can
you
also
define
the
features
that
you
intend
to
support
for
Windows
as
part
of
GA?
Not
just
when
you
do
the
cap,
can
you
also
define
the
features
you
intend
to
support
as
part
of
what
the
GA
product
looks
like
you,
as
in
like
there
are
requirements
like
test
passes
over
and
so
on,
but
we
will
support
deployment.
We
will
support
IB
the
support
staple
set.
We
will
not
support
security
context.
We
will
not
support
high
disruption,
button
or
policy
just
so.
I
D
B
Great
Michael
and
Patrick
I'm,
really
sorry
that
this
has
been
so
difficult
and
please
for
all
the
people
that
are
working
on
this
on
your
team.
Tell
them
that
we
we
really
do
want
to
make
this
better
in
this
process
more
smoother
than
the
future.
They
do
appreciate
the
work
we
just
need
to
be
cautious.
That's.
B
B
G
B
B
B
Fantastic
I
did
a
little
bit
of
housekeeping.
We
are
cancelling
next
week
because
the
US
Thanksgiving
holiday-
and
that
does
not
mean
that
the
work
stops.
However,
all
of
you
there
so
continue
looking
at
the
the
tagged,
secure,
architectures
things
in
the
repos,
if
you're
feeling
bored
and
want
to
do
some
extra
stuff,
for
the
sake
there's
a
lot
of
just
general
documentation.
That
needs
to
be
done,
and
obviously
this
work
with
Michael
and
Patrick
on
the
windows
and
getting
that
GA
check
lists
together.
B
That's
something
that
we
probably
should
should
think
about
anyways
and
also
getting
ready
for
a
cube
con.
If
anybody.
If
anybody
wants
to
help
with
that,
because
we
do
have
the
architecture
update
and
right
now,
I,
don't
think
we
have
a
list
of
things
that
we
want
to
talk
about
or
we
we
had
it
for
a
while,
but
we
didn't
actually
agree
on
anything.
So
yeah
I,
don't
know
Brian.
Do
you
have
anything
you
want
out
on
that,
since
we
kind
of
skated
over
that.
F
Yeah,
so
the
contributor
summit
is
that
the
update
you're
talking
about
correct
yes
yeah.
So
there
is
a
talk
at
a
contributor
summit
which
I
plan
to
give
about
the
technical
directions
for
the
soudanese
project
for
2019,
so
I'm,
starting
to
put
that
together,
I
asked
for
ideas
for
that
a
couple
of
times
in
this
meeting.
It's
my
current
thinking
is
an
evolution
of
the
architectural
layers
that
we
talked
about
last
year.
F
So
those
were
the
main
topics.
Other
folks
have
ideas.
I
suggest
starting
an
email
thread
on
the
mailing
list
or
reach
out
to
me
when
I
have
a
draft
I
will
send
it
out
the
suit,
assuming
that
I
have
it
early
enough,
but
the
that
the
the
gist
of
what
I'm,
currently
thinking
about
I,
just
want
to
say
thanks
again
to
our
new
chair
volunteer,
because
the
layers
diagrams
that
I
started
with
last
year
were
derived
from
Matt's
diagrams,
so
I'll
be
building
on
those
and
reshuffling
them
a
little
bit.
Brian.