►
From YouTube: Config Working Group 4 26 18
A
A
A
So
daddy,
as
far
as
I
know
the
CRT
sync,
it's
almost
done
so
I
would
say
90%,
but
I
will
give
all
the
final
call,
because
he
is
he's
not
going
today.
So
the
public
internal
format.
We
had
a
discussion,
it's
done,
and
so
you
can
look
into
the
Devan
talk.
I
can
go
over
view
of
that
design
later
in
the
meeting
and
the
tested
framework
I
think
also
still
experimenting.
It
may
be
in
total
typing
stage.
A
A
B
A
B
C
C
A
A
So
the
row
maps,
and
so
we
just
come
up
with
our
QT
roadmaps
and
even
if
it
has
almost
a
mouse
past
q2
so
I
think
we
had
a
few
modifications
to
what
we
intend
it
to.
We
actually
accelerating
the
gallery
alpha
and
that's
gonna
stuff
on
b1
there
soon
we
have
decided
the
and
the
intrud
external
form
and
that's
make
its
the
further
making.
The
alpha
support
will
be
easier
by
using
the
external
CRD,
a
peer
group
which
is
different
funding
kind
of
one
and
existing
one.
A
So
what
we
plan
to
do
is
by
the
end
of
q2,
we
try
to
getting
some
minimal
support
of
conflict
model
meteor
with
service
config
and
on
the
distribution
side.
We
having
galley
client
and
the
galley
start
to
distribute
configs
with
components
with
county
clients.
Jason
will
go
a
little
bit
detailed
to
to
that
because
he
has
a
design
and
to
show
today.
A
So
on
the
row
map
is
basically
we're:
gonna
finish:
basic
suicide,
comfort,
validation
to
beta
incrementing,
Gary,
Alpha,
Sinfonia,
v1
design,
prototyping
service
config
with
Stata
features
and
the
component
config
distribution
data
client.
We
can
start
a
design
consumer
config,
but
may
not
be
some
of
the
governance
process.
We're
gonna
start
it
and
we
were
still
putting
our
API
styling
her
and
I'm
part
of
the
equivalent
process,
but
we
may
not
enter
to
do
it.
A
So
this
is
only
for
supporting
galley
and
galley
clips
donne
0,
like
the
first
iteration
of
gallium
sinking,
to
set
soceity
above
almost
the
identical
contents,
so
so
some
of
the
guidelines,
because
it's
an
Indian
intermediate
solution
which
will
be
replaced
in
a
few
months
by
over
the
foot
load
like
solutions.
So
we
try
to
pick
some
solution
that
kind
of
allow
us
to
move
faster
than
move
faster
than
than
other
solutions.
A
So
we
kind
of
listed
are
kind
of
different
solutions,
alternatives
and
pros
and
cons,
and
so
we
end
up
like
choosing
solution
5,
which
is,
we
are
not
changing
the
existing
config
and
the
resources
and
the
series
we're
going
to
keep
using
config
Lisa
to
I/o,
so
the
external
CR
DS
will
will
be
in
the
writes,
Eclair
group
and
the
right
type
and
versioning,
and
those
theories
will
be
watched
by
Gary.
So
once
galley
watch
those
see
are
these
and
they
will
convert
to
the
internal
conflict.
A
I'll
use
do
I,
owe
the
goose
a
is
for
this
to
work,
the
components
that
I
need
to
any
do
any
change,
because
what
I
learn
from
training
pilots
is
changing
the
Eclair
group.
It's
actually
deeper
model
change
and
that's
kind
of
destructive,
and
if
we
want
to
support
a
galley
with
differently
other
solutions
like
using
external,
if
you're
opening
an
ikea
cook
may
end
up,
and
the
new
change
is
expected
on
Holland
mixer
to
switch
this
to
a
peer
groups
to
supporting
Gary
alpha.
A
So
it
was
solution
5.
We
don't
need
to
do
any
of
this,
and
when
get
is
a
placed,
it
has
it's
an
up.
It
only
works
when
people
start
writing
external
facing
configs
and
it's
easy
to
end
it
on.
The
major
count
of
this
is
that
we
are
not
using
the
scopes
provided
by
the
API
group
so
end
up,
but
we
still
need
to
require
our
kind
of
name
to
be
unique
and
that's
also
impose
requirements
on
our
photo
message.
Design,
/,
API
style,
guide
this
to
be
unique.
A
A
So
there's
a
then
we
need
to
do
some
row
backs
and
to
the
CLE
changes
to
pilot
that
is
probably
okay
and
and
there's
some
other
count
house.
It's
the
one.
A
major
one
is
the
something
somehow
it's
not
fully
friendly
with
your
photo,
because
when
they
resolve
the
groups,
it's
kind
of
implicit,
if
you
have
to
kind
name
and
the
same
kind
name
and
different,
if
you
book
it
will
end
up
giving
you
a
default
resource
from
certain
groups.
So
your
end
up
using
fully
qualified
domain
name.
A
So
there
is
some
alternatives,
so
we
can
resolve
the
kind
uniqueness
but
being
the
complications.
So
we
expect
we
are
not
going
to
do
it.
So
this
is
summarize
the
decision
and
the
why
we
made
a
decision
and
the
main
reason
is
allows
to
move
faster
and
to
the
service
config
mode
and
and
the
component-level
any
restrictions
we
have
right
now
will
be
removed
once
to
beginning
the
distribution
GRCC
stuff
on
kissing.
A
So
any
questions,
no
more
questions,
maybe
Jason.
You
want
a
first
and
so
Jason
has
a
design
proposal
about
component
config
distribution,
and
so
we
plan
to
do
a
design
review
and
after
Kubica
and
then
the
next
working
group
meetings
so
but
Jason.
Maybe
you
want
to
do
an
overview
of
what
to
see
maybe
share
the
dock,
so
people
can't
stand
up
handed
yeah.
C
C
C
C
A
C
Conversion
that
each
component
needs
to
do
which
could
be
the
nulls.
Are
we
don't
want
to
miss
the
a
server?
That's
not
still
there
from
the
center
for
configuration.
This
is
just
about
the
components.
Api
symptom
point
is
like
highlighters
Feldman
didn't
need
to
talk
to
the
or
anyway
for
service,
renders
your
information
yeah.
So.
C
Do
you
see
Ernie's
engineer,
PC
series
for
persistence,
bidding
observability
all
the
things
that
we
reason
people
like
using
kubernetes?
We
want
to
keep
that.
We
need
to
keep
that
because
we
need
to
migrate
from
where
we
are
now
to
whoever
we're
going
to
be.
But
the
component
we
want
to
propose
another.
The
subscription
model
where
the
component
has
subscribed
to
pick
updates
from
a
fixer
Gallery,
and
then
we
have
a
galley,
would
there's
a
proxy
config
from
the
CRV
to
the
SUBSCRIBE
component
and
two
starters
will
adopt
the
ongoing
STS
protocol.
C
A
C
With
what
what
they
design
and
of
all
vision
to
our
this
arrow
means
the
nice
thing
about.
This
is
they're
both
watch
based
and
they
both
carry
its
data
differently.
So
we're
gonna.
The
names
are
gonna
change,
so
it's
here
de
carries
a
resource
where
they
component
would
subscribe
to
configuration,
have
no
be
delivered
in
pieces
of
config
enlist
pieces.
If
we
want
to
call
fragments
differentiate,
lettuce,
CRD,
what's
in
a
C
or
D,
what
a
component
gets
for
starters
will
be
the
same
thing.
C
But
then
anyone
we
can,
we
can
map
it
to
the
current
kettle
and,
as
we
diverged
over
time,
it
needed
every
River
time.
The
resources
are
clearly
see.
Our
piece
completes
the
fragment
than
they
necessarily
the
same
thing.
I
couldn't
know
what
detail
about
how
they
we
track
that
with
snapshot
and
take
to
the
track
in
criminal
rollout.
But
that's
the
basic
idea.