►
From YouTube: CNCF Serverless Working Group 2020-01-23
Description
CNCF Serverless Working Group 2020-01-23
B
A
A
A
A
A
A
A
Good,
pretty
good
Mike,
hey
there
I'm
here,
that's
right
and
Ryan.
Yes,
I
am
Balu
excellent,
hello,
we'll
just
get
started
at
3:00
after
just
get.
A
A
H
A
Lionel
all
right,
it's
three
after
one
I
go
and
get
started
cool.
Thank
you,
I
missed
you
all
right.
Let's
do
this
thing.
Excuse
me
in
terms
of
a
eyes,
I
did
reach
out
to
a
couple
different
people
in
the
sensitive
organization,
in
particular
Liz
Rice,
who
chairs
up
the
TOC,
and
she
said
it
was
okay
for
us
to
send
a
note
to
the
TOC
as
opposed
to
a
formal
presentation.
Just
let
that
update
them
on
her
status
so
mark
and
I
are
working
on
that
you're
really
really
interested.
A
You
could
look
at
the
tail-end
list
of
the
situation,
discovery
doc,
it's
the
rough
draft.
They
were
working
on,
but
I,
don't
think
is
anything
too
controversial
in
there.
So
once
mark
and
I
settle
on
the
exact
wording,
I'll
just
send
that
out
as
an
email
to
the
TOC
and
I'll
probably
be
CC
1
of
our
mailing
lists
here.
So
you
guys
can
see
that
when
it
goes
out
with
that
community
time
anything
from
the
community,
the
people's
like
to
bring
up.
A
All
right
moving
on
kook
on
you,
you
so
I
did
create
a
planning
doc
just
to
gather
all
of
our
thoughts
and
activities
there
so
that
I
bring
that
up
here.
So
a
couple
things,
first
of
all,
if
you
are
planning
on
going,
please
add
your
name
to
the
list
here,
just
so
that
we
can
keep
track
of
it
for
face
to
face
planning
purposes
in
terms
of
making
sure
we
have
a
big
enough
room
and
also
we
will
be
having
a
kiosk.
A
A
We
will
need
volunteers,
so
please,
okay,
so
for
the
35
minute
session,
I'm
still
trying
to
find
out
from
the
workflow
guys
who's
going
to
be
presenting
there
and
what
their
abstract
is
going
to
be
I,
think
they've
been
busy
or
sick
or
something
so
I
haven't
yell
to
sync
up
with
them,
but
the
current
status
is
a
five
minute,
quick
update
on
cloud
events
followed
by
the
bulk
of
its
on
on
the
workflow
stuff.
Okay,
now
for
our
intro
session,
we
agreed
to
a
quick
update
on
cloud
events
and
then
an
overview
of
description.
A
Api
I,
don't
mind
doing
the
cloud
event
stuff
unless
someone
else
really
wants
to
do
it,
but
if
you
do
either
want
to
do
that
or
the
subscription
API
spec
just
put
your
name
here.
Just
because
you
see
a
name
in
any
of
these
things,
don't
hesitate
to
put
your
name
if
you
really
want
to
do
it
as
well.
This
is
just
for
the
list
of
people
to
choose
from
and
then
one
who's
actually
gonna
do
it
later.
So
don't
be
shy
if
you
already
see
a
name
on
any
of
these
spots.
A
I'm
a
deep
dive
here
is
the
abstract
that
Scott
gave
me
that
I
sent
in
anyway
and
he's
volunteered
to
be
a
speaker
they're,
giving
it's
a
hands-on
that
we
probably
don't
need
too
many
speakers
per
se.
But
if
you
do
want
to
join
Scott
on
the
stage
to
talk
about
stuff,
that'd
be
great.
However,
we
will
probably
need
several
people
to
help
out
to
walk
around
the
lab
to
help.
You
know
the
actual
participants
do
stuff
if
they
get
stuck
so
we'll
need
as
many
people
down
here
as
possible.
A
So
please
have
to
think
about
how
your
name
here
is
a
lab
helper
and,
as
part
of
her,
probably
also
need
to
decide
which
SDKs
were
going
to
focus
on.
Is
it
going
to
be
just
one
to
keep
it
easy
for
us,
this
stuff
or
a
presentation
perspective,
or
do
you
want
to
potentially
have
a
couple
different
ones?
People
can
choose
from
no
decision
on
that
yet,
but
be
thinking
about
that
one.
A
D
H
A
H
B
A
I
A
Okay,
now
question:
for
you
guys,
I
did
confirm
that
we
will
have
an
answer
bar
ask
me:
anything
kind
of
a
thing
was
just
a
booth
in
the
main
you
in
one
of
the
big
halls
there,
and
apparently
there
are
three
different
options
we
could
do.
We
can
do
dedicated
kiosk
all
to
ourselves,
but
it's
basically
means
someone
has
to
be
there
full-time
or
as
a
man,
full-time
I
should
say.
The
other
options
are
either
half
time
dedicated
a.m.
or
p.m.
or
that
they
shared
they
describe.
A
A
However
I'd
like
to
know
from
you
guys
how
you
feel
about
this,
do
you
want
to
try
to
staff
it
full-time?
Do
you
think
there's
gonna
be
enough
interest
in
people
who
are
from
the
audience
cuz
I.
Can
you
know
I
can
I
can
be
the
NAG
to
try
to
get
her
right
to
sign
up
it?
Won't.
Do
full
time
but
I
don't
want
to
pester
people
off
if
everybody
thinks
it's
too
much.
A
C
E
J
G
E
E
C
E
D
E
This,
whoever
the
staff
is
that
would
do
it.
I,
don't
know
if
it's
CNS
CF
SAP
or
if
it's
the
conference
staff,
but
they
would
come
out
and
switch
the
signage.
So
if
it's
cloud
event
for
the
morning,
then
it
would
have
a
cloud
event
sign
and
then,
if
it
was
somebody
else
in
the
afternoon
and
they
would
just
change
the
sign
project,
okay,.
E
B
Can
we
post
a
schedule
or
something
if
we
choose
the
second
of
the
third
option,
I
remember
in
San,
Diego
I
wanted
to
chat
with
somebody
at
the
booth,
and
there
was
nobody
there.
It
was
unclear
which
were
what
so
having
like
a
public
when
a
timetable
or
something
and
I
don't
mean
public
in
the
sense
that
in
the
working
group,
but
in
public,
like
on
the
CN
CF
websites,
Gabe,
whatever
will
help
a
lot.
I
can.
A
A
K
C
K
A
H
A
Big
a
deal
if
you
were
already
planning
on
going,
however,
if
your
travel
approval
travel
plans,
whatever
it
may
be,
are
contingent
upon
was
having
a
face-to-face.
You
may
not
find
out
until
February
14th,
which
I
think
is
within
that
right
around
that
four-week
window
for
our
governance,
talk
of
but
now
I'd
say
of
announcing
a
face
to
face
so
just
be
warned.
You
may
want
to
get
a
travel
approval
in
now,
I
mean
maybe
I'm
gonna
say
lie,
but
tell
your
manager
or
whoever
that
you
think
there
will
be
a
face-to-face
meeting.
A
A
Is
definitely
a
possibility
that
would
make
it
harder,
though,
because
if
I
beat
em,
if
I
had
to
be
more
after-hours,
then
which
makes
it
a
little
bit
more
difficult
to
get
our
way
together.
But
we
can
ever
look
at
that.
Yeah,
okay,
so
let's
hold
off
and
see
what
they
say.
I
I
think
it'd
be
very
odd
for
them
not
to
be
able
to
find
a
room
for
two
hours,
but
we'll
see.
B
A
Cool
okay,
I
think
we
already
talked
about
all
these
things.
That's
good
all
right,
right,
SDK!
We
did
not
have
mean
last
week,
so
nothing
to
talk
about
I
believe
we
will
have
a
meeting
today
right
after
this
one's
over.
So
anybody
on
this
case
I
had
a
house
be
prepared
to
join
Cathy
I,
believe
it's
sick.
Unfortunately,
so
she's
out
on
a
call.
Do
you
have
anybody
else
from
the
workflow
subgroup
on
the
call
I
want
to
give
status,
No?
A
Okay,
in
that
case
Shh
in
terms
of
issues
for
caught
events
itself.
These
are
still
being
worked
in
the
background.
I
assume.
However,
this
one
is
new.
I
don't
want
to
so
I
talk
about
it
too
much
here
other
than
to
poke
people
to
take
a
look
at
it,
because
it's
a
good
to
question
about
serialization
to
Jason
and
I
know
that
we
talked
about
this
in
the
past
and
I.
Just
could
not
remember
the
definitive
answer,
so
I
was
hoping,
maybe
like
Scott
since
you're,
obviously
heavy
involved
in
the
goal-line
SDK.
A
A
A
And
I
know
we
went
back
and
forth
that
thing.
I
met
a
couple
of
times:
okay,
anything
else
relative
to
the
cloud
event
spec
itself
before
we
move
on
to
the
to
the
new
spec,
all
right,
moving
forward,
then,
okay,
since
Clements,
not
on
a
call
Mike.
Would
you
like
to
update
us
on
the
status
of
things
I.
D
A
H
Any
scope,
level
hey-
this
is
Ron
yeah
I
can
just
confirm
that,
unfortunately,
for
the
subscriptions
part,
majority
of
us
had
a
combination
of
heavy
travel
schedules
and
being
sick
over
the
last
two
weeks.
So
we
didn't
meet
this
week,
but
we
do
have
a
follow-up
on
Tuesday,
so
we'll
hopefully
have
a
more
meaningful
update.
Next
week.
Okay.
H
F
We
have
yeah
if
we
have
time
the
last
regular
core.
We
also
talked
about
the
subscriptions
on
sort
of
the
back
propagation
of
a
subscription
up
to
the
original
publisher
and
what
clements
not
on
a
call,
what
he
said.
It's
a
difficult
topic
and
he
kind
of
wants
to
push
his
back,
not
fall
a
bit
about
it.
But
from
my
perspective,
it's
not
so
difficult,
so
I
don't
know
if
it's
appropriate.
F
If
not
stop
me,
if
we
discuss
in
this
detail
now,
but
my
father's,
it's
actually
not
so
difficult
because
you
have
the
filters
you
just
take
all
your
filters.
You
propagate
them
up
once
that
and
then
the
other
guys
can
do
what
they
want
with
the
shooters.
If
you
apply
filters
on
your
events
that
don't
apply
to
you
and
it's
fine,
so
I
don't
see
where
he
thinks
it's
so
difficult.
A
J
A
J
It
depends
really
I
mean
I
brought
up
this
abstract
model
of
eventing
domains,
because
I
mean
there
can
be
different
messaging
technologies
behind
I,
don't
know
Christophe
what
what
scenario
you
have
in
mind
exactly
so,
if
it's
also
about
using
different
protocols.
So
then
you
might
have
something:
that's
bridging
between
different
eventing
technologies.
J
F
A
A
C
G
J
G
Yeah
yeah
I
wasn't
insinuating
that
this
would
need
to
be
a
new
broker
now
I
I
get
that
I
was
just
sort
of
stating
that,
with
all
the
protocols
listed
and
I
alluded
to
the
duty
to
this
in
a
comment
as
well,
it
looks
like
we've
got
a
real
discrepancy
between
the
behavior
of
HTTP
protocol
and
the
others,
which
are
our
broker
based
and
well.
It's
just
saying
is
in
this.
G
Eventing
domain
is
very,
very
easy,
with
a
broker
based
broker
based
protocol
right,
so
I'm
qtt,
Nats
whatever,
but
then,
when
you
get
into
HTV
Peele,
and
this
becomes
actually
very
difficult
and
I
see
the
problem
you're
trying
to
solve
I'm
just
sort
of
stating
that
there
may
be
more
here
than
this.
This
could
get
into
a
very
deep
pit.
J
Okay,
yeah
see
so
yeah
for
HTTP,
as
it
is
usually
point-to-point.
I
agree,
that's
more
difficult,
so
a
extreme
example
of
an
event
in
the
main
would
be
just
a
single
I
mean
a
point-to-point
connection,
so
you
wouldn't
then
have
a
lot
of
producers
and
consumers,
but
just
a
single
one
and
combine
them
then
buy
a
link
basically,
and
then
it
wouldn't
really
be
really
a
dedicated
to
Maine
or
something
would
be
just
hidden
inside
the
the
producer
and
the
consumer
could
also
think
of
it.
H
J
G
A
J
Klaus,
yes,
it's
not
really
about
the
subscriptions.
It
was
actually
the
issue
I
opened
more
than
a
year
ago,
I
think
about
the
nested
events.
So
I've
still
got
that
the
action
items
I'm
honestly
a
bit
clueless
after
our
last
discussion,
so
originally
I
brought
it
up
in
the
SDK
call
two
weeks
ago,
because
I
thought
it
was
just
more
or
less
an
implementation
detail
in
the
SDKs.
But
the
interesting
enough.
The
discussion
brought
up
that
it's
different,
so.
E
J
It,
as
you
can
see,
I
brought
this
up
in
February
2018,
so
it
was
even
before
we
had
0.1
of
cloud
events
and
I.
Think
for
me
at
least
this
nested
event
case
is
not
really
relevant
right
now,
but
still
as
I
brought
it
up
and
was
interesting
discussion
I'm
so
years
or
we
had
the
idea
I.
Think
in
the
discussion
further
down
doc,
you
propose.
I
could
just
write
an
example
where
one
event
like
structured,
a
coded
event,
would
be
nested
inside
a
binary
and
I.
J
When
I
tried
doing
this,
I
ran
into
the
problem
that
we
determined
by
the
content
type
that
what
the
event
in
that
message
is
so
doing
this.
What
is
shown
here
this
this
example
is
not
possible
and
is
also
not
not
really,
and
that
came
out
in
the
discussion
two
weeks
ago
really
against
this.
Back,
as
we
say
that
no.
J
That's
what
I
said:
yeah
sorry
if
it,
if
I
somehow
didn't
express
that
clearly.
So
exactly
so,
the
idea
of
this
back
and
that
I
wasn't
aware
of
that
too,
and
at
that
time,
is
that
even
if
you
do
structured
encoding,
all
the
attributes
from
that
structured
encoded
event
can
also
be
replicated
to
the
header.
So
this
case
that
you
have
one
event
in
the
header
and
another
one
in
the
payload
is
not
really
possible,
as
least
if
you
use
the
normal
content
time
of
the
cloud
events
plus
JSON.
J
I
D
J
A
J
Yes,
I
said
I
created
that
issue
in
the
beginning,
when
I
still
was
was
trying
to
make
my
mind
about
all
this
eventing
and
by
now
on
our
end,
I
would
say
we
definitely
want
to
avoid
this
nesting.
But
one
example
where
I
saw
something
like
this
or
word
might
occur,
is
actually
in
Canada
eventing
with
some
of
the
sources
they
I
think
like
the
K
native,
the
the
Kafka
sauce
I
think
they
create
their
own
event.
I
look
like
Kafka
event.
J
J
A
A
Think
part
of
the
reason
was
because
these
properties
here
did
not
match
the
properties
here
and,
and
if
these
appear
here
they're
supposed
to
basically
be
a
duplicate
of
what
you
see
inside
the
body
memory
correctly,
yes,
okay,
the
one
thing
I'm
always
kind
of
bothered
me
about
that
is
I
know
that
the
spec
says
you
can
copy
things
up,
but
does
it
mandate
that
if,
for
example,
C
type
appears
here
it
has
to
match?
What's
in
the
body.
A
I
H
J
I
J
I
F
That
was
a
bit
my
question
why
we
have
to
have
it
like
in
the
spec
you
have
to
like
the
data
and
your
data
you
can
have
whatever
you
want,
including
another
nested
event.
So
that's
up
to
you.
If
you
really
want
to
do
it,
but
then
you
don't
get
all
the
transformation
from
one
protocol
video
drop
this
week,
but
that's
it
up
to
you,
so
you
can
do
it.
If
you
absolutely
have
to
or
I
don't
see
how.
Well
you
are
adding
the
complexity
of
all
this
nesting
into
the
spec.
F
A
A
J
A
J
A
J
A
D
I
Closest
is
correct
because
I
don't
think
it
would
round-trip
converting
a
rabid
message
between
binary
and
structured
if
it
wraps
a
structured
object,
because
I
think
it
promotes
the
content
type
and
in
that
case,
you're
getting
at
the
same
situation
of
the
above
example
so
seems
like
primer
texts
is
probably
in
order.
If
you're
gonna
wrap
cloud
events.
A
C
A
B
A
There
might
be
something
worthy
of
mentioning
some
place
in
our
primer
or
something
that
just
so
people
understand
our
versioning
scheme.
I'll,
double-check
I.
Have
this
vague
recollection.
We
may
actually
say
something
about
that's
some
place,
but
I'll
check
it.
If
not
maybe
I'll
open
issue
just
so,
we
can
think
about
it.
A
Okay
did
I
miss
anybody
for
attendance,
I,
think
I
got
all
the
late
how
the
people
showed
up
at
the
end,
all
right.
In
that
case,
we
were
done.
Thank
you
guys
and
if
you
were
interested
in
the
SDK
call
go
ahead
and
stay
on
the
line.
We'll
start
that
one
up
in
a
about
two
or
three
minutes
just
that
people
leave
thanks.
Everybody
thank.
D
D
A
K
K
K
A
Hey
while
we're
waiting
there
was
that
face-to-face
meeting
between
K
native
and
kata.
What's
the
a
step
in
that
I
thought
personally,
a
follow-on
meeting
has
that
been.
K
Scheduled
yet
I
don't
think
it
has
been
scheduled.
Yet
there
is
a
kata
call
at
I
think
it's
in,
like
whatever
20
minutes
so
ten
o'clock
my
time.
So
the
next
meeting,
slot
and
I
think
there's
some
Chad
ease
about
that
I
know
some
folks
played
with
the
interaction
with
the
cave
native
on
kata
and
seeing
who
could
kind
of
fit
in
and
this
morning
in
the
channels
working
group.
K
We
also
chatted
about
making
sure
that
we
start
that
there's
the
whole
channel
spec
on
sort
of
kind
of
what
the
channels
look
like,
and
there
was
some
talk
about
making
sure
that
in
the
spec
we
do
not
assume
too
much
of
a
push
based,
but
we
also
allow
for
the
venting
based
sorry,
pull
based
models,
and
so
I
basically
broke
some
folks
volunteered
them
to
go
ahead
and
take
a
deeper
look
on.
What
does
that
mean?
Okay,.
A
A
I
J
I
So
at
the
moment
the
plan
is
we're.
Gonna
do
a
one,
oh
release,
that's
gonna
capture
the
original
method
and
then
the
because
we're
gonna
do
SEM
for
everything
else
in
the
repo
is
going
to
start
having
to
become
API
compatible
for
changes.
There's
going
to
be
a
oh
release
that
drops
the
original
transport
implementations
in
favour
of
the
bindings.
J
I
I
I
And
we've
been
coding
this
strategically
so
there's
pieces
that
are
leveraged
in
the
2o
implementation
that
get
migrated
out
or
copied
out
of
that,
the
one
o
virgin
so
we're
aware
of
the
Scratchy
and
then
the
other
thing
that's
going
to
come
is
the
we're
gonna
document,
the
2o
architecture,
so
that
it's
it's
a
little
more
clear.
What
exactly
we're
doing
and
having
add
a
new
binding.
I
Cool
the
goal
is
to
get
one.
Oh,
you
know
walk-down
released
for
coop
con
and
for
integration.
So
this
comes
because
there's
there's
been
some
people
that
would
like
to
integrate
with
the
SDK,
but
are
unwilling
to
do
this
like
in
the
and
the
readme.
It's
like
by
the
way
like
you're,
gonna,
break
you
willy
nilly
and
there's
customers
that
do
not
want
to
to
take
on
a
dependency
that
doesn't
follow
similar.
So,
even
though
it's
maybe
not
production-ready
exactly
there's
still
some
stuff
to
figure
out
we're
gonna
lock
the
API
down,
make.
B
You
scroll
down
a
bit.
Are
you
joined
this
call
because
I
was
interested
in
Kim's
questions
from
the
week
so
go
along
as
the
Carrie
super
stable,
very
much
hoping
of
the
spring
world
and
sitting
here,
C
sharp,
SDK,
Clemens,
obviously
uses
it
for
Asia,
Asia
and
stuff
is
maintained
so
on
and
so
forth.
Do
you
have
a
like
a
list
of
maintainer
for
the
others?
We
do
that
I'm
certain
interest
in
events,
but
a
lot
of
people
are
a
bit
off
by
the
state
of
the
SDK
yeah.
A
Unfortunately,
I
didn't
do
my
a
I'm
supposed
to
poke
the
bottom
for
people
in
terms
of
who
the
owners
are
I,
can't
say
who
they
are
offhand
other
than
I
think
Fabio
might
on
the
top
of
one
I
was
going
to
just
take
a
look
at
the
commits
and
see
who
did
all
the
commits
and
poke
those
people
to
find
out
the
status
of
these
things.
Well,
maybe
we
need
to
have
official
leads
that
are
kind
of
accountable
for
these
things.
That
would
be
a
good
thing
documents
at
some
place.
Yes,
in.