►
From YouTube: CNCF Serverless WG Meeting - 2019-08-22
Description
Join us for Kubernetes Forums Seoul, Sydney, Bengaluru and Delhi - learn more at kubecon.io
Don't miss KubeCon + CloudNativeCon 2020 events in Amsterdam March 30 - April 2, Shanghai July 28-30 and Boston November 17-20! Learn more at kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects
A
B
A
Forward,
let's
see
nothing
exciting
on
the
a
is
actually
maybe
I
should
ask
the
question
so
Clemens
you
have
to
a
eyes
out
there.
Do
you
plan
on
working
on
those
and
then
your
future?
A
D
E
D
F
A
A
G
A
G
G
A
Cool
any
questions
for
this
Takei
crib,
okay,
moving
forward
I
did
ask
to
get
on
the
TOC
schedule,
so
I
could
put
to
a
good
show
or
I'm
sorry,
so
we
could
ask
to
go.
Do
to
incubate
er
status
and
Chris
Anna
check
put
us
on
for
the
September
17th
TOC
call
still
looking
for
more
end-users.
Another
a
couple
of
people
are
still
investigating
that.
A
Yet
I
am
planning
to
put
that
up
this
week
so
that
we
can
start
gathering
our
ideas
for
what
talk
about
there.
Even
though
we
have
a
you
know,
sort
of
rough
that
line
already,
but
the
old
men
allow
us
to,
then
you
know
get
people
to
sign
up
to.
That's.
Do
the
talking
so
look
for
that
later
this
week,
all
right
before
we
jump
into
PRS
and
stuff
any
other
topics.
A
Let's
see
here
all
right,
he
took
a
action
item
to
modify
the
definition
of
a
string
to
limit
it
or
to
get
a
better
definition
for
what
it
means
to
be
had
a
printable
character
set,
and
this
is
basically
way
to
came
up
with
I'll.
Give
you
guys
a
second
to
read
this
in
case
you
haven't
a
chance
to
read
it
yet.
A
G
Yeah
I
think
this
is
basically
just
trying
to
use
the
same
technique,
we're
using
with
content
type
for
binary
encoding
for
HTTP
for
the
payload
for
the
data
content
encoding.
So
content
encoding
is
already
have
a
valid
HTTP
header,
where
we're
trying
to
redefine
what
that
term
is,
but
for
binary
tree
mappings.
We
should
just
use
the
normal
HTTP
header.
A
A
C
That
is
exactly
that:
it's
exactly
what
it
is,
so
there's
a
I'm
in
parallel,
responding
to
the
other
issue,
that's
open
with
data
transferring
Cody
and
and
that's
a
misunderstanding
because
there's
a
Content
code.
So
what
we
mapped
to
is-
and
this
is
what
the
spec
literally
says-
we're
mapping
to
RFC
7231
we're
just
mapping
to
we're
mapping
a
cloud
event
to
an
HTTP
message.
We
don't
do
anything
else.
C
So
data
transferring
so
content
transfer
encoding
is
in
in
so
with
with
what
we
do
here,
we're
mapping
to
both
HTTP
2
and
we
will
back
to
HTTP
3
and
we
met
map
to
HTTP
1
1,
because
we
only
really
focus
on
mapping
to
the
message.
The
transfer
stuff
is
all
specific
to
RFC
30
to
30,
it's
72
30s,
all
specific
to
HP,
1
1
and
looks
different
in
in
hd2
and
hd3.
A
So
I
think
I
was
anything
for
a
slight
different
perspective
and
correct
me
if
I'm
wrong,
you're
gonna
probably
am,
but
if
data
content
encoding
for
the
structured
format
of
stuff
is
base64,
because
it's
binary
data
and
you
can't
put
binary
data
into
adjacent
string
right
when
you
then
convert
that
into
the
binary
format
for
cloud
events
correctly.
My
wrong,
but
wouldn't
you
just
put
the
raw
binary
data
into
the
a
shippi
body.
So
therefore
you
wouldn't
need
content
encoding.
A
H
A
C
So
what
you
can
do
with
gate,
so
what
you
can
do
with
content
encoding?
Is
we
we're
actually
leaving
that
open
to
the
spec?
We
say?
Well,
you
must
support
base64
right,
but
it's
also
possible
that
you
that
you
in
that
express
other
values
that
you
use
as
extensions.
So,
for
instance,
in
in
RFC
3272
31.
You
can
also
for
the
content,
encoding
use,
gzip.
A
A
I
understand
you
may,
but
just
because
you
need
it
for
structured,
there's
not
as
I
guarantee
that
you
need
it
for
binary.
No,
it
yes,
but
this
mapping
is
correct
right
and
I
guess
what
I'm
wondering
is
whether
we
need
additional
sentence
here.
That
says
just
because
you
need
this,
for
the
Jason
serialization
does
not
necessarily
mean
you
have
to
then
base64
encode,
all
HTP
messages,
in
particular
in
the
binary
format.
Yes,
yeah.
C
A
C
It's
okay
to
say:
Jesus.
We
have
not
specified
that
yet
should
not
get
worked
up.
It
worked
up
about
it,
but
that's
where
you
would
place.
That's
where
you
would
place
agreements
about
compression
right
now.
We've
been
we've
been
using
basics
before,
as
the
one
thing
we
think
is
necessary,
but
compression
all
those
things
would
go
here,
understood.
A
F
A
C
C
F
C
Well,
the
rule
is
pretty
clear
in
the
HTTP,
and
that
is,
you
can't
have
anything
but
7-bit
ascii,
which
means
we'll
have
to
have
some
kind
of
encoding
and
it
percent-encoding
is
the
closest
thing.
But
then
then
the
question
is,
you
know:
does
that
actually
catch
the
does
that
actually
catch
and
the
Unicode
said
like.
So
how
do
you?
How
do
you
encode
a?
C
C
A
G
D
A
A
Okay,
in
that
case,
we'll
have
to
delay
this
one
till
next
week
to
on
I,
actually
think
that's
it
in
terms
of
open,
PRS
or
version
one.
So
that's
good
I
only
have
one
right
now,
however,
we
do
have
a
couple
of
issues
that
I
wanted
to
get
people's
opinions
on,
because
I
thought,
if
they
actually
required
to
change
to
spec,
they
need
to
be
in
there
before
1.0.
A
A
Because
I
I
didn't
but
I
look
at
this
schema
for
the
header.
It
didn't
show
a
list
there,
but
I
know
in
general.
You
can
have
this
thing.
Hcp
header
appear
more
than
once,
but
I
couldn't
find
any
text.
That
said,
the
second
one
overrides
the
first
or
whether
it
appended
to
some
sort
of
list.
I
did
yeah.
F
F
C
Says
if
one
or
more
it
so
htp
says
and
because
I
just
have
that
right
here,
if
one
or
more
encodings
have
been
applied
to
the
right
to
a
representation
I'm,
so
that's
payloads,
that's
the
age
feeling
go
for
payload.
The
sender
that
applied
the
encoding
must
generate
the
content,
encoding
header
field
that
lists
the
content
coatings
in
the
order
in
which
they
were
applied
right.
C
C
A
H
D
C
H
C
C
No,
no,
no!
No!
You
look.
Look
at
that
sentence
if
one
or
more
encoding
is
applied
bond
so
of
one
or
more
in
courts
have
applied
to
a
representation
of
the
sender
or
must
generate
a
company
Coney,
head
or
field
that
lists
the
content
coatings
in
the
order
in
which
they
were
applied.
Okay,
so
that
has
a
list.
C
C
And
what
he
does
they
have
so
for
the
case
that
we
have
binary
this
is
this
is
now
getting
a
little
weird
I
mean
I
may
need
some
more
time
to
think
about
the
details.
So
in
the
case
when
I
have
binary,
I
want
to
be
able
to
do
exactly
what's
set
here.
That
which
means
I
want
to
be
able
to
declare.
This
is
gzipped
and
go
and
use
gzip.
A
C
Okay,
great
so
HP
does
not
use
the
content,
transfer
encoding
field
of
mine,
so
the
thing
that
we
refer
to
this
year
right
is,
is
not
engaged
in
HTTP,
but
there
is
a
transfer
encoding
field
and
the
transfer
encoding
field
in
in
30
to
70,
okay,
now,
okay,
now
I'm
back
on,
but
now
I'm
back
on
track.
So
sorry
for
the
confusion,
so
this
yield.
C
C
No,
this
really
is
correct.
Okay,
because
that's
what
we
mean
we
so
so
smtp
where
this
comes
from
right
has
can
only
do
text
and
you
have
to
trick
it
into
carrying
and
old
versus
of
well,
it
can
only
do
text,
and
so
you
have
to
trick
it
into
into
carrying
something
that
is
more
than
7-bit
ascii,
so
you're
effectively
telling
it
a
look
at
this
look
at
this
encoding
field.
This
is
what
you
find
in
the
air
in
here:
there's
no
7-bit
ascii!
This
is
actually
something
else.
I
C
Yeah,
so,
basically
for
basically
for
is
nowhere,
but
here-
and
this
is
what
we
refer
to
that
spec
now,
the
the
the
transfer,
the
content
encoding
of
HTTP
is
a
different
thing
for
which
we
don't
have
a
field
for
yet
the
so
this
is
so
our
truck
data
content.
Encoding
is
specifically
only
for
this
for
the
case,
when
we
put
data
into
straight
in
structured
mode
into
a
data
it
what
we
put
the
string
into
data,
but
the
string
really
is
supposed
to
be
a
binary.
C
A
C
A
The
reason
I
say
that
is
because
this
implies
to
some
people:
it's
the
HTTP
header
content
encoding,
which
is
a
list,
and
your
grades
and
I
think
what
you're
saying
is:
no,
no.
We
don't
want
to
list.
This
is
strictly
only
for
the
binary
case
and
we're
stealing
data
transfer,
encoding
or
content
transfer
encoding
and
therefore,
to
avoid
any
possible
confusion.
We're
transferring
here.
You
avoid
that.
Yes,.
C
A
C
C
A
C
A
C
A
A
Okay,
so
we're
think
I
think
we're
talking
about
dropping
the
word
content
dropping
this
section
to
loosen
it
Peggy,
even
broader,
set
of
strings,
but
keep
the
reference
to
defining
base64.
So
we
have
a
little
bit
of
inter
outbuilding,
yeah
and
I
suppose
drop
this
so
he's
got
the
resident.
C
And
then
and
then,
and
we
still
need
to
string
what
we
still
need
to
have
a
have
a
sentence
that
says
you
know
that
basics
t4
is
the
string
that
is
like
this
is
not
clear.
Yet
it's
this
attribute
their
support.
The
basic
c4
in
colon
must
be
supported,
yeah
I
think
it's
that's.
That
I
think
is
sufficient.
Okay,.
A
A
G
A
G
It's
tricky
because
we
we
might
be
talking
about
the
RFC
for
transfer
encoding
if
it's
HTTP
binary,
but
we're
not
we're
not
necessarily
talking
about
that
as
as
part
of
the
spec
but
part
of
the
binding
so
like
in
in
structured
mode
data
content.
Encoding
does
not
mean
that
and
it
shouldn't
be
mapped
to
the
header,
because
we
have
a
string
inside
the
JSON
payload.
That
is
basic
support.
Encoded.
You
know
when
that
thing
gets
converted
to
a
binary,
HTTP
request.
G
We
should
pop
the
quotes
off
the
JSON.
Base64
string,
use,
content,
encoding
and
then
provide
it.
But
the
thing
is
like
you,
you
might
lose
the
intent
of
the
original
sender
if
they
want
that
date.
The
data
sent
as
base64
encoded
in
the
next
conversion.
We
don't.
We
don't
necessarily
know
that
the
original
request
was
supposed
to
be
base64
encoded,
so
yeah
structured
might
be
something
completely
different
and.
C
C
We
leave
the
choice
that
you.
If
you
want
to
write,
you
can
still
have
a
basic
c4
body,
and
then
this
field
says
well.
This
is
a
basic
c4
thing.
You
should
go
and
interpret
that
as
binary
or
you
leave
that
off
completely
this
field
and
then
the
encoder,
that's
binary
so
more
effectively,
now
giving
you
both
choices,
also
for
for
HTTP.
But
that
means
because
that
this
basics,
before
encoding
things
is
not
a
concept
at
HTTP
supports
in
its
content,
encoding
field
which
exists,
but
that's
really
just
for
gzip
we're
now
having
we're.
C
Now,
we
now
set
ourselves
up
with
a
core
the
collision,
just
by
choosing
that
name.
So
if
we
just
call
this
data
encoding,
then
it
gets
and
we
don't
map
it
to
http.
Then
it
becomes
in
HTTP
becomes
ze
no
data
encoding,
and
it
goes
from
from
one
side
to
the
next
and
then
there's
clarity
that
this
is
really
cloudy.
Events
concept,
yeah.
C
A
A
A
B
A
A
A
D
A
A
A
Okay,
let's
see
if
we
can
move
on
to
Evans
PR
about
tricky
cases
now
keep
in
mind.
This
is
all
non
normative
stuff,
but
we
do
need
to
make
sure
it's
right,
since
these
are
guiding
examples
come
on
there
we
go
so
we're
back
to
Clemons
under
Clemons
that
Evans
PR
did
people
get
a
chance
to
look
at
this.
One
I
think
he
made
some
very
very
minor
typographical
fixes
last
night,
I,
don't
think
of
any
semantic
changes.
A
A
G
C
C
G
A
C
C
A
A
Christopher's,
looking
at
this
one,
so
I
don't
think
we
can
discuss
it.
I
think!
That's
it.
Let's
actually
pause
for
the
webhook
spec,
which
I
think
is
an
AI
for
me
to
deal
with
anybody
else.
Yet,
okay,
I
think
that's
it
for
the
agenda.
Are
there
other
I'm?
Sorry
come
as
we
gonna
say
something
yeah.
The.
G
C
C
C
No,
of
course
no
it's
not.
This
is
not
the
only
way.
This
is
why
I
think
it's
a
separate
one
right
the
web,
so
expect
is
one
way
you
can
do
these
things,
and
this
is
what
so
so
the
HDP
spec
the
web
look.
Script
are
separate
specifically
because
I
wanted
to
make
sure
that
the
you
can
just
map
the
cloud
event
onto
an
HTTP
message
and
independent
of
the
direction.
Alright,
the
HP
messages
are
the
same
for
requests
and
responses
other
than
the
status
line
versus
the
crest
line
and
that
some
headers
are
allowed.
C
No,
but
the
rest
is
all
the
same.
So
that's
what
the
H
should
be
binding,
my
it
does
and
then
for
pushing
stuff
out.
That's
what
the
web
expect
covers,
and
we
don't
have
anything
for
solicitation
of
events.
That's
right,
but
that's
something
we
could
go
and
we
could
go
and
add,
but
but
for
delivering
events
via
push.
A
C
There
is
it
so
there's
a
there's,
a
spec
project.
Apparently
then
I
learn
about
inside
of
inside
of
the
Linux
Foundation,
which
is
you
know,
all
the
granite
if
the
grand
umbrella,
which
is
specific
existing
for
things
of
that
sort,
but
I
have
not
done
a
follow
up
on
the
investigation
to
see
whether
we
can
go
in
and
move
it
to
a
different
place.
C
Ultimately,
it's
just
a
question
of
whether
it
should
be
home
so
that
it
gets
more
audience,
but
we're
doing
inventing
here
and
that's
kind
of
the
primary
use
case
for
aspect
like
this.
So
that's
why,
even
though
it
doesn't
have
a
kind
of
time
into
the
payload,
that's
why
it
makes
sense
to
have
here.
If
we
remove
it,
then
there's
all
the
things
that
are
in
that
spec
are
all
about.
You
know
creating
a
profile
of
a
of
HTTP
that
helps
to
draw.
C
Then
all
those
constraints
for
a
drop
are
gone
and
we're
still
having
the
issue
of
like
how
do
how
do
you
do
off?
And
how
do
you
do?
Abuse
prevention,
like
all
the
things
that
are
specifically
addressed
here,
are
then
again
left
up.
Left
to
no
speculation
and
mutual
agreements,
and
that's
what
I
want
to
avoid
with
this
right.
A
So
anyway,
I
think
this
is
I.
Think
between
the
two
of
us,
the
ones
we
have
any
ID
figure
out
what
they
want
to
do,
whether
you
want
to
propose
to
keep
it
here,
opposed
to
move
it,
and
we
could
take
that
offline,
yeah,
okay,
okay,
okay,
any
other
topics
for
the
agenda
since
we're
almost
out
of
time.
A
A
J
A
A
All
right
cool,
just
since
we
have
a
second
here,
Kyle
you
still
on
York
Kyle.
If
you
hope
you
don't
mind,
you
feel
free
to
say
you
can't
comment
if
you
can't
comment,
but
are
you
guys
actually
using
cloud
events
or
are
you
still
just
sort
of
investigating
it?
Just.
A
The
reason
I'm
asking
is
because
I
I
get
the
sense
that
your
company
would
fall
into
the
category
of
an
end-user
company
and
we're
looking
for
end-users
to
say.
Yes,
we
we
use
cloud
events.
So
if,
at
some
point
you
you
feel
like
you
can
say
that
without
lying,
please
let
me
know
so
we
can
add
your
company
to
the
name
to
the
list
assuming
you're,
okay,
with
having
your
name
in
there,
really
all
right
cool.
Thank
you
all
right.