►
From YouTube: CNCF Serverless WG - 2018-06-21
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
Alright,
hey
I,
think
I
got
everybody,
let's
go
and
get
started
first
off.
Thank
you
to
everybody
for
the
flurry
of
activity
relative
to
you
are
related
to
all
the
AIS.
We
had
see
from
all
the
ones
that
are
crossed
off.
We
went
through
quite
a
few
there's
still
some
that
are
left
open,
but
if
the
list
is
shrinking
fast,
so
thank
you
guys
for
working
on
that
appreciate.
It
first
up
is
a
new.
Yes,.
B
A
A
I,
okay,
then
Joey.
Let
me
move
it
up
here
and
cross
it
up.
Then.
Thank
you,
sir.
If
you
give
me,
if
you
get
a
chance,
can
you
put
a
link
here
to
the
PR
just
hope?
You'll
have.
A
You
very
much
all
right
next
community
time
so
for
those
of
you
who
were
not
at
the
face-to-face
or
and
they
dad
have
had
a
chance
to
read
the
meeting
minutes,
we
decided
that
we
wanted
to
do
a
little
bit
more
of
an
outreach
to
the
community,
so
people
who
are
not
regular
to
this
phone
call
but
who
do
want,
who
want
to
provide
feedback
or
input
in
terms
of
their
use,
cases
or
use
of
cloud
events,
and
so
this
is
an
opportunity
for
people
who
don't
know
due
to
our
phone
calls
or
basically
were
comfortable.
A
You
know
brought
in
the
broader
community
to
bring
up
any
issues.
Comments
concerns
whatever
to
the
group
at
large.
Hopefully
we
won't
take
up
too
much
time,
definitely
won't
schedule.
Ten
minutes.
If
it
turns
out,
we
have
a
lot
of
feedback
from
a
lot
of
people,
which
is
actually
a
good
thing.
There
too
may
end
up
having
a
dedicated
phone
call,
but
for
right
now
we
wanted
to
start
with
just
ten
minutes
on
the
way.
We
call
to
see
what
kind
of
feedback
we
get.
A
A
E
That
was
kind
of
putting
people
on
the
spot
that
maybe
there
might
be
something
at
the
end.
Yeah,
that's
fine!
You
could
circle
back
around
at
the
end
yep.
If
anyone
has
anything
to
talk
about
as
a
user
of
cloud
events,
not
necessarily
just
a
spec
writer
but
someone
he
may
have
the
use
case
or
may
have
tried
it
always
working
have
been
implementation.
E
A
All
right
and
if
I
forget
please
remind
me
to
go
back
around
to
the
end
of
the
call
all
right.
Next
up,
I
had
an
action
item
to
ask
about
stickers
and
t-shirts.
So
I
did
reach
out
to
Dan
Khan,
and
he
told
me
that
next
week
we
should
be
giving
me
a
coupon
code
in
order
for
us
to
order
stickers,
I,
don't
know
whether
that's
an
unlimited,
every
stickers
or
not,
but
when
I
get
the
coupon
code
I,
let
you
guys
know
so.
We
should
be
able
to
get
some
stickers
fairly
soon
t-shirts.
A
Unfortunately,
the
ciencia
will
not
pay
for
those,
because
we
are
just
a
sandbox
project.
We
have
to
be
a
little
bit
higher
on
the
totem
pole
in
order
to
get
them
to
pay
for
t-shirts.
For
us,
however,
he
did
point
me
at
a
particular
company
at
which
I
assume
he's
using
the
past
because
he
recommended
them.
So
what
we
should
probably
do
it
at
some
point
discuss
whether
we
wanted
to
get
together
as
a
group
to
order
Qi
search
because
I
assume
the
more
you
order,
the
cheaper
it
actually
is
so
I'm
not
gonna.
A
Put
anybody
on
the
spot
right
now
in
terms
of
whether
they
actually
want
to
fork
up
or
put
up
money
to
actually
buy
some
t-shirts
so
think
about
it,
and
now
next
week's
call
I'll
do
a
straw
poll
or
some
kind
of
poll
to
see
you
know
who's
interested
actually
buying
t-shirts
to
see
what
kind
of
order
we
can
put
together.
Okay,.
B
E
Have
thought
so
because
that
makes
the
t-shirts
a
bit
unnecessary
at
that
point,
wealthy.
E
Mean
it
depends
how
long
the
logos
gonna
take,
but
let's
say
we
had
it.
One
of
the
things
we
could
do
is
to
create
a
an
illustrator
file
and
put
that
in
the
github
repo
and
then,
depending
on
how
coordinated
we
can
be
before
during
I
know.
When
we've
ordered
t-shirts
at
VMware
for
an
event,
they
can
be
as
little
as
six
six
to
nine
dollars
per
per
piece.
It
may
mean
that
people
could
order
their
own
if
it's
easier,
yeah.
A
F
Would
love
to
take
a
shot
at
this
time
is
just
I'm
is
just
hard
to
find,
as
always
I
say,
give
me
a
week
to
submit
a
PR
and
if
I
can't
come
up
with
anything,
then
let's
just
move
forward
and
order
some
stuff,
and
if
I
get
the
PRN,
then
we
should
look
at
it
as
a
group,
and
you
know,
I'll
try
and
provide
a
few
options.
If
we
like
them,
you
don't
we'll
vote
on
them.
If
not,
then
you
know
we'll
just
move
on
with
what
we
have.
Okay.
A
F
Sure
thing
we're
doing
a
discussion
to
discuss
the
we're
doing
a
separate
conference
call
to
discuss
the
design
of
the
cloud
events
SDK
there's
a
lot
of
interest
in
this.
It
turns
out
there
a
lot
of
companies
and
individuals
already
working
on
this.
So
we've
put
up
a
doodle
poll
to
vote
on
when
the
best
time
would
be
right.
Now
it
looks
like
Monday
is
the
most
optimal
time,
so
I
think
we're
just
going
to
go
ahead
and
go
with
that.
F
Have
you
closed
out,
the
poll
yeah
I,
think
I
think
we'll
we'll
probably
close
it
out
right
now
and
just
stick
with
the
Monday.
So
I
guess
this
is
more
of
a
heads
up
that
we're
gonna
do
the
call
on
Monday
and
the
time
is
8
a.m.
Pacific
and
I'll
close
the
poll
right
now,
but
if
you're
interested
in
SDK
design
for
cloud
events,
there's
a
github
issue,
I
think
maybe
Doug
could
put
it
in
the
in
the
dock.
F
F
F
A
E
I
would
like
to
be
meeting
I,
didn't
see
the
doodle
bug
or
whatever
it's
called
and
May
I'm
gonna
be
traveling
on
Monday
and
gonna,
be
in
Eastern
Europe,
so
I
don't
know
what
time
that
is,
that
I
might
not
be
able
to
make
if
I
would
have
liked
to
have
been
there.
Hey.
F
Alex
we'd
love
to
have
you
and
I
think
this
is
gonna,
be
the
first
of
many
and
we're
gonna
do
our
best
to
surface
all
of
our
conclusions
and
discussions
right
after
the
meeting,
so
you'll
be
able
to
follow
along
and
dive
in
whenever
you
have
time
we're
gonna,
try
and
face
this
out.
You
know:
initial
version
should
look
like
this.
The
next
version
should
look
like
this,
so
we're
not
gonna
like
do
anything
just
yet
we're
just
gonna
decide
the
scope
of
it,
and
you
said:
4
p.m.
PDT
8
a.m.
Pacific
on
Monday.
E
F
F
A
G
F
H
B
And
just
to
be
clear,
the
the
first
meeting
was
only
to
discuss
what
is
the
scope
of
things
that
we
could
talk
about
and
what
do
we
think
we
should
limit
it
to
and
then,
after
after
we've
figured
that
out,
we'll
start
doing
work
to
provide
code
or
whatever
else
base
based
on
that
feedback.
So
I
don't
think
that
people
will
be
missing
a
whole
lot
on
their
first
oneness
we're
just
trying
to
scope
it
so.
F
G
G
F
G
G
A
A
G
About
how
about
end
of
tomorrow,
Pacific
Theater
end
of
tomorrow
sounds.
A
Alright,
any
questions
come
on
tonight
all
right
coming
forward,
then
at
the
face-to-face
I
hear
who
it
was,
but
someone
brought
up
the
fact
that
we
forgot
to
actually
do
sort
of
a
post-mortem
about
the
interruptive
vent.
So
a
turn
agenda
item
for
us
discuss
that
are
there
any
comments,
observations
or
anything
along
those
lines
that
people
want
to
bring
up
relative
to
the
inter
up
event.
Aside
from
Austin,
did
an
awesome
job,
putting
everything
together.
D
Is
Rachel
I
think
it
would
have
been
like
I
didn't
think
this
at
the
time.
I
didn't
understand
the
importance
of
this,
but
in
retrospect,
I
think
it
would
have
been
fun
to
have
more
image
providers
like
more
sources,
admitting
the
events
so
maybe
for
the
next
demo
that
we
do.
We
could
do
that.
Okay,
we're,
maybe
changing
them
together,
might
be
the
thing
to
do.
Just
like
show
more
of
the
process.
I
didn't.
E
Understand
what
an
inter
up
event
was,
but
now
I
I
think
this
was
a
demo
from
Keep
Calm.
Yes,
maybe
we
could
talk
about
this
in
in
plainer
language,
Keep,
Calm
event,
gateway
demo,
so
I
felt
like
we
missed
out
on
that
a
open
vast
community
really
would
have
liked
to
be
in
a
part
of
it.
Somehow
we
didn't
get
the
message
and
we,
you
know
we're
used
to
working
very
late
notice
with
a
few
hours.
We
can
turn
things
around
so
for
the
next
demo.
E
I
E
A
What
do
we
do?
What
what,
if
we
just
try
to
make
sure
that
going
forward
announcements
about
meetings
or
anything
I
think
they
like
this
is
put
out
on
the
mailing
lists.
I
know
some
people
like
slack
but
I,
feel
like
that
everybody's
actually
watches
slack
very
closely,
but
mailing
lists
should
show
up
in
everybody's
inbox.
What,
if
we
just
assume,
that's
the
preferred
mechanism
for
getting
to
notice
out
about
something,
is
that
okay
with
people
yeah.
A
B
E
K
I
L
A
Yeah
I'm
trying
to
find
the
intern,
who
might
have
some
free
time
to
put
together
basically
a
Facebook
I'm,
sorry
that
it's
but
a
Twitter
light
kind
of
thing,
basically
to
to
visualize
the
stream
at
some
point
so
that
we
can
want
to
rely
on
Twitter
but
gives
the
same
sort
of
net
effect.
But
we'll
see
how
that
goes.
Just
so,
you
guys
know.
B
B
B
E
It's
not
it's
not
easy,
as
it
was
manager,
even
when
you
know
how
so
that
I
think
that
just
speaks
alike.
The
preparation
for
the
next
for
the
next
relative
we're
talking
about
RSS,
which
sounds
you
know,
I,
remember
that
back
in
the
day
as
well.
What
about
something
like
our
our
group
slack
and
if
we
guess
get
a
token
from
the
administrator,
we
could
use
that
the.
I
Nice
thing
about
RSS
is
that
you
can
everybody
can
go
and
publish
to
a
feed
that
they
have
and
then
you
can
go
and
pull
it
all
together
in
an
aggregator,
which
means
you
can
really
make
it
make
it
distribute
it.
And
then
you
have
an
aggregator
which
basis
shows
it
all.
In
one
place
like
Google
Reader,.
M
I
A
C
F
F
So
that
was
that
was
pretty
neat
Oh
I
think
you
know
for
a
first
demo
for
just
coming
out
of
the
gate,
wanting
to
get
that
visibility
wanted
to
demonstrate,
what's
possible,
being
inclusive
of
everybody.
That's
involved
as
long
as
you
know
they
got
the
memo,
then
you
know
we
I
think
we
nailed
that
you
know
up
up
next,
we
got
to
make
it
real
for
people.
We
got
to
get
closer
to
those
real
world
use
cases.
We've
got
to
get
the
tools.
F
The
other
pieces
that
you
know
are
just
kind
of
stable
stakes
for
using
this
stuff
more
easily,
like
the
SDKs
integrations
into
you
know
all
pieces
of
important
popular
pieces
of
infrastructure
in
the
ecosystem.
So
we
got
to
get
more
real
now
in
what
we
do
next,
so
that's
my
I
think
that's
my
biggest
feedback
and
but
over
I'd
say
you
know,
given
those
criteria
which
are
hard,
it's
really
hard
to
get
visibility,
get
people
excited,
be
inclusive
of
all
types
of
vendors
I
think
we
knocked
it
out
of
the
park
there.
F
Yes,
I've
heard
the
same
thing:
it's
one
thing
to
put
out
this
put
out
a
standard,
there's
another
thing
to
do
it
with
all
of
the
industry
influencers
coming
together
and
at
the
end
of
the
day,
that's
probably
going
to
be
the
thing
that
makes
the
most
difference
and
we
we've
done
such
a
great
job
with
that.
So
far,
so
exciting
stuff,
yeah.
B
M
A
A
Okay,
cool,
alright
moving
forward,
then
issue
maintenance,
I,
don't
what
I'm
trying
to
do
going
forward
is,
as
new
issues
are
opened
up
when
appropriate,
I'm
going
to
try
to
see
we
can
find
an
owner
for
them
just
so
we
have
someone
who
can
sort
of
Shepherd
it
through
the
process.
I
think
we
have
actually
had
more
than
one
open,
but
I
think
only
one
of
them
actually
make
sense
to
have
an
owner
at
this
particular
time
so
Thomas.
A
A
It's
it's
more
just
to
help
make
sure
that
something
happens
with
the
issue.
It
gets
ret
resolved
in
some
fashion.
So
whether
that's
I
mean
that
doesn't
mean
you
yourself
out,
spread
the
PR
but,
as
you
said,
get
the
red
people
involved
to
want
to
have
the
discussion
around
it
and
see
if
you
can
come
up
with
some
kind
of
PR
for
it.
If
you
want
sure
yeah
I.
A
Okay,
thank
you
all
right
now,
I
believe
during
the
face-to-face
meeting.
We
briefly
talked
about
these
two
issues
and
we
agreed
that
we
could
probably
close
them,
but
we
want
to
give
people
time
to
look
it
over.
So
I
did
put
in
notes
and
each
issue
saying
if
you
care
about
this
one
and
want
to
champion
it,
please
speak
up
and
that's
the
people
that
we
were
probably
gonna
into
closing
the
issue
fairly
soon.
If
no
one
speaks
up.
A
So
this
is
my
way
of
saying
if
no
one
speaks
up
by
next
week,
I'd
like
to
close
these
two
issues,
if
everybody's
okay
with
that,
so
this
is
just
a
heads
up
for
people-
is
there
any
concern
with
giving
one
more
week
and
then
closing
these
two?
If
there's
no
comments
or
somebody
willing
us
to
a
champion,
it.
A
Alright
cool,
thank
you
guys.
Next,
this
is
just
for
me.
Jason,
like
I
think
is
last
thing
is
Roper
I
believe
in
Australia,
so
he
probably
isn't
on
the
call,
but
he
opened
up
here
if
it's
any
short
PR
and
you
say
it's
a
pull
request
for
dad
and
then
key
feel
to
the
spec
during
the
face-to-face.
We
briefly
talked
about
this
and
the
notes
have
an
AI
for
cummins
myself
to
add
this
non
goal
of
defining
transport.
I
have
no
idea
what
that
means.
Oh
wait,
who
wrote
this
right
here
is
that
I
didn't.
A
I
I
We
have
a
scenario
that
actually
layers
on
an
existing
spec
where
we
have
another
name,
be
binding
and
we
need
partition
keys
for
event
hubs,
or
at
least
it's
not
seen
to
do
that
for
us,
which
uses
ANCA
P
and
we
I
would
have
to
go,
and
we
would
have
to
go
in
to
find
effectively
a
constraint
over
the
piece
that
that
says.
Well,
here's
how
you
generate
your
partition
key
got.
C
It
seems
like
every
time
someone
new
comes
in
everyone
assumes
well,
of
course,
this
is
synthesis,
goes
over
a
message
queue.
It
must
be.
It
must
include
your
topic.
It
must
include
your
shark
key,
whatever
I
wonder
if
there's
a
good
place
to
put
in
like
our
spec,
what
non
goals
are
like
things
that
we
have
learned
should
not
be
here.
Yes,.
I
A
In
fact,
that
I
already
opened
up
the
PR
for
that
and
I
think
that
it
would
be
a
great
place
to
put
this
type
of
information
because
it's
not
necessarily
normative
text,
but
it
is
sort
of
background
history
guidance.
What
do
you
want
to
call
it
and
that's
usually
what
a
primers
for
so
yep
I
agree
with
you
Thomas
all
right
moving
forward?
Actually,
really
that's
what
you
just
said
there
Thomas
I,
don't
want
this
track
of
your
suggestion,
though
we
do
want
to
open
up
an
issue
just
so
we
don't
lose
track
with
that.
A
A
N
To
excuse
for
mounting
for
the
cloud
event,
basically,
we
are
trying
to
playing
around
with
the
cloud
event
fact:
that's
why
we
want
to
see
how
a
Picasso
can
integrate
with.
How
can
that
be
one
of
the
transport
funding
for
our
event
and
how
we
can
map
the
possible
masters
between
the
car
events
back
in
that
way,
cloud
event
can
be
use
of
some
event:
presentation
for
the
fossil
functions
to
process
the
event
in
the
stream.
I
That
pasty
just
says
here
comes
a
protobuf
message
message
and
then
the
the
protocol
basically
just
puts
photo
buff
on
on
the
wire.
So
it's
it's
effectively.
Tcp
preamble,
proto
bus
and
that
doesn't
meet
the
bar
for
me
for
something
that
is
a
protocol
in
a
bigger
sense.
Frankly,
and
so
I
find
it
and
it
will
probably
probably
evolve
a
new
work
and
an
incubation
project
actually
in
the
Apache
foundation
and
that
doesn't
I,
feel
I
feel
it's
wrong
to
go
and
elevate,
singular
efforts.
I
A
I
No,
it's
I
think
it's
one.
If
it's
one
project
so
I
can
I
could
go
so
if
we
I
could
go
and
make
us
make
a
project
in
the
Apache
foundation
or
in
Linux
Foundation
somewhere
else
just
by
their
rules,
make
up
any
proprietary
protocol.
That
I
want
and
then
come
into
cloud
events
and
propose
a
binding
which
is
then
kind
of
a
standards
body
there's
rescue
operating
and
then
it
kind
of
gets
blessed
by
that.
As
you
know,
official
thing
kind
of
by
by
indirectly
by
us
I
was
giving
it
official
blessing.
I.
I
J
C
I
And
I
agree
so
having
this
having
this
having
a
catalog
of
and
here
our
event,
your
implementations
for
particular
products
and
that
are
using
cloud
event
in
here.
So
you
met
those
and
and
pointer
to
their
project.
Documentation
I
think
that's
that's
great,
but
just
having
that,
as
you
know,
having
these
sorts
of
specs
didn't
side
by
side
with
mappings
for
ratified
standards
like
HTTP
and
Kapiti,
which
are
you
know,
foundations
for
little
broader
product
categories.
That
just
doesn't
seem
right
to
me
so.
A
Let
me
share
I
think
also
on
the
same
page
though,
because
I'm
not
high
percent
sure
you
are
I,
think
you're
on
the
same
page
and
that
there's
a
second
tier
of
protocol
mappings,
but
do
those
second
tier
documents
sits
within
our
repository
or
do
they
sit
in
the
other
person's
repository
with
just
a
pointers
from
ours?
What's
the
suggestion
there.
I
C
A
N
E
Not
streaming
standardless
that
has
a
spec
and
a
meat-eating
smtp
has
a
consortium
behind
it.
Kitty's
noisy
slander
and.
I
E
A
What
if
we
did
this,
what
if
we
do
this
in
two
step
process?
First
people
review
the
current
PR
to
see
if
the
text
of
the
protocol
definition
here
sounds
good
as
a
first
draft
and
then
to
maybe
Clemens.
Maybe
you
could
write
up
a
PR
with
how
to
structure
this
second-tier
type
of
system
that
we're
gonna
set
up
you
and
then
we
could
see
if
people
like
that
second
tier
system
or
not
and
and
we'll
hold
off,
actually
accepting
this
PR
until
we
decide
whether
we
want
a
second
tier
and
we'd.
I
I
I
E
I
We
have
a
in
the
calf
car
case.
There
is
I
would
say
an
out
that
is
relatively
new,
because
there
is
now
a
second
clean
Kafka
protocol
implementation
that
we
built
and
that
kind
of
makes
it
a
de
facto
multi-party
standard
and
we're
going
to
engage
in
the
Kafka
community
specific
on
the
protocol
pieces.
Yeah.
L
I'm
trying
to
understand,
why
are
we
trying
to
be
so
strict?
You
know
you
go
to
Sparky
or
a
bunch
of
plugins
for
any
database
in
the
world.
Why
still
do
it
make
it
so
complicated?
You
know
we
should
have
a
repo
where
everyone
who
feels
like
you
know
there
is
a
standards
and
specifications,
but
the
implementations
should.
I
Since
for
defining
effectively
a
standard
here,
I'm
just
trying
to
be
in
that
standard,
the
goal
of
that
standard
is
to
promote
interoperability.
It
seems
odd
to
me
that
we
would
be
happy
for
proliferation
of
project
proprietary
protocol
at
the
bottom
and
then
simply
go
and
layer
stuff
on
top
of
at
the
bottom
of
it.
I
think.
L
I
I
I
The
thing
right
right,
you
look
at
you,
look
at
four
standards
that
exists
and
instead
of
picking
one
of
the
four
because
it
doesn't
quite
exactly
know
what
you
do,
you
create
a
fifth
and
that's
how
you
get
twenty
and
twenty
five
and
that's
how
we
destroy
interoperability
and
I.
Think
if
we're,
if
we're
here,
to
go
and
promote
interoperability,
we
should
also
do
our
part
in
stopping
proliferation
of
proprietary
protocols
and
not
not
try
to
set
you
up
right.
L
J
L
But
who
are
you
to
decide
which
one
you
know?
Maybe
tomorrow
there'll
be
another
project
which
is
in
its
inception.
It's
still
not
widely
adopted,
but
people
want
to
use
it.
So
you
know
if
you're,
making
an
exception
for
Kafka,
which
is
not
an
idea
for
I,
Triple
E
or
whatever
is
standard,
then
I
think
we
should
open
up
for
one.
You
shouldn't
be
so
strict,
so.
A
Clemons
is
it
true
that
I
think
in
your
mind,
you
actually
have
a
distinction
between
a
protocol
binding
for
a
completely
new
type
of
binding
versus
a
protocol
binding
for
something
like
say,
HTTP
right.
If
someone
was
to
come
along
with
a
brand
new
HTTP
binding
that
it's
just
different
than
one,
we
have
I
think
you're
really
concerned
about
a
proliferation
of
HTTP
bindings
right.
You
want
only
one
HP
binding.
I
That's
a
part
of
the
that
doesn't
align
with
with
the
goals
that
we
that
I
have
in
in
you
know,
making
making
sure
that
we
have
more
interoperability
because
actually
reach
less
interoperability,
where
everybody
can
now
could
put
the
same
payload
on
a
less
interoperable
ecosystem,
and
that
doesn't
make
sense
to
me.
So
so,
instead
of
doing
that,
I
would
go
and
say
if
there
is
and
I
think
the
base
principle
is.
I
If
you
have
a
protocol,
that
protocol
must
be
implemented
in
at
least
two
projects
have
a
proper
protocol
spec
and
it
needs
to
be
implemented
in
at
least
two
projects
independent
of
vendors.
So
that's
similar
to
the
rules
that
you're
familiar
with
in
the
license,
even
though
they
say
three
and
and
basically
promote
interoperability
and
not
for
every
project
to
just
come
along,
make
their
own
thing
and
then
make
out
against
binding
and
then
declare
that
as
an
click
and
being
able
to
claim
that
that's
a
standard
thing
is
it's
not
right.
A
So,
let's
do
this
because
I
don't
want
to
take
up
the
rest
of
time
with
this,
because
I
think
this
is
a
little
bit
contentious.
So
it's
Clements!
You
took
the
AI
to
write
up
a
PR
to
describe
the
process,
we're
going
to
use
to
decide
what,
whether
we
want
to
have
a
second
to
your
set
of
specs
or
not,
and
how
we're
going
to
side
what
falls
into
that
category.
But
it's
human
potations,
where
you
want
to
call
it
so
and
that
will
once
you
once
you
have
the
PR
people
can
review
that.
A
A
But
in
the
time,
I
would
like
people
to
go
and
look
at
this
particular
PR
to
see
if
they're,
okay,
with
the
actual
text
in
there
as
a
starting
point
for
a
draft,
and
then
we
can
just
figure
out
based
the
Cummins
PR,
whether
we're
gonna
accept
it
and
how
we're
gonna
accept
it.
That's
not
fair
and
what
we
won't.
We
won't
both
on
that
today.
Obviously,
a
way
for
Cummins
PR
get
results,
I'm.
A
A
C
G
A
It
depends
what
you're
talking
about
if
you're
talking
about
a
new
attribute
from
the
spec
perspective,
it's
gonna
be
just
like
all
the
others.
If
you're
talking
about
an
extension
attributes,
that's
gonna
be
done
or
actually
handled
at
the
transports.
Spec
level
and
Clemens
has
a
P
as
I
can
modify
all
of
our
transport
specs
to
say
how
extensions
will
appear
in
each
of
those
transports.
G
I
So
the
what
we
read
to
is
that
we
don't
are
not
going
to
make
we're
not
gonna
make
in
a
distinction
here
in
for
in
the
main
spec
for
extensions,
we're,
basically
just
allowing
that
extensions,
can
go
and
extend
the
the
attributes
that
instead
of
the
cloud
events
envelope
and
then
it's
really
up
to
the
extension
spec
per
se,
to
define
how
that
projects,
how
its
attributes
project
into
the
respective
transport.
So
the
extends
has
become
a
layer
on
top
of
the
existing
specification
set
and
and
then
and
then
there
will
be
a
generic.
I
So
in
each
transport,
spec
will
have
a
generic
way
to
to
map
extensions
if
the
extension
itself
doesn't
have
an
opinion.
So
if
the
extension
itself
says
the
attribute,
X
must
be
mapped
to
a
particular
HTTP
header,
because,
like
in
in
the
case
of
I,
think
open
tracing
it
was.
There
is
a
set
of
very
specific
HTTP
headers
that
are,
it
must
be
used.
I
Then
the
extension
spec
is
effectively
overriding
anything
else
and
says:
if
you
have
this,
if
you
have
to
find
this,
this
actually
then
I
think
that
must
produce
the
following:
HTTP
HTTP
headers
in
in
DHCP
mapping,
and
then
there
was
going
to
look
for
all
other
cases
where
the
extension
is
not
specific
about
this.
There's
going
to
be
a
default
mapping
that
the
that
the
transport
specialist
does
that
answer.
G
I
Engine
per
extension,
so
you
write
a
spec
in
the
extension
and
the
extension
defines
where
the
fields
are,
and
it
then
also
defines
special
handling
for
special
transports.
So
it's
all
self-contained
extension
user
statement
it
sends
inspect
needs
to.
As
you
write
a
new
one
needs
to
take
care
that
you
don't
have
clashes
with
the
base
specification
and
or
and
then
also
avoid
clashes
with
other
extensions
and
that's
mostly
about
convention,
and
with
that
we
actually
avoid
having
you
know,
prefixes
or
anything
in
special.
I
So
if
you're
writing
something
that's
for
open
tracing,
you
would
probably
go
to
be
safe
to
go
and
name
your
attributes
open
tracing,
something
so
that
you
avoid
clashes.
But
I.
Don't
think
we
need
to
have
hard
and
fast
rules
here
to
go
and
do
prefixes
for
everything
that
are
kind
of
weird,
but
we
just
make
it
in
an
organic
fashion
and
just
put
it
on
me,
put
the
responsibility
on
the
extend
inspect
own
or
to
go
and
do
the
right
thing,
but.
J
A
I
No
they're,
not
you
write
an
extension
and
the
extension
is
to
the
cloud
events
envelope
and
then
you
and
which
effectively
adds
to
the
abstract
the
abstract
model.
Yesterday
the
model
that
we
have
and
then
su
serialize
that
as
you
serialize
and
map
that
cloud
events
document
down
to
the
transport,
then
you
also
apply
rules.
So
the
way
I
think
about
this
in
software
is
that
you
effectively
are
you
have
two
pieces?
You
have
the
parliament's
document.
Let's
say
that's
a
an
object
graph
and
that
object
graph.
The
spec
tells
you
hey.
I
You
can
add
the
following
three
properties
since
in
an
object
graph,
then
you
run
that
through
a
pipeline
in
the
pipeline,
there's
a
pipeline
stage
here
in
the
HTTP
transport
that
says,
oh
I
know
those
three
things
and
it
will
then
go
in
the
HTTP
pipeline
and
then
take
these
three
properties
and
not
them
appropriately
to
HTTP.
So
think
of
it
really
as
a
plug-in
model
and
the
extension
spec
is
defining
how
you
ought
to
write
your
plugin.
G
I
Correct
I'll
write
this
I
summarize
this
for
ducks
document
and
then
also
make
the
additions
and
I'm
gonna
and
I'm
gonna
make
that
I
mean
that's
the
the
one
work
item
I'm
going
to
go
into
along
with
the
the
principles
for
protocols
next
week
and
then
have
that
in
time
for
the
Tuesday
deadline.
Right.
A
A
So
there's
one
thing
I
want
to
bring
up
before
we
do
you
circle
back
around
things.
Where
is
it
so
I
did
oh
I
mentioned
I
open
up
a
PR
for
the
first
pass
at
a
primer.
It
is
by
no
means
a
complete
primer.
It's
just
sort
of
a
reorganization
of
existing
documentation
that
we
have
just
pull
it
along
together
into
a
single
document
and
I
did
add
a
at
least
I
think
at
least
one
placeholder
for
additional
text
to
come
later.
A
This
is
just
the
first
draft
of
the
document
just
to
get
something
out
there,
so
people
can
start
issuing
other
PRS
to
it
not
going
to
ask
for
a
vote
right
now.
Obviously,
but
please
just
take
a
look
at
it,
I'd
like
to
see
if
we
can
get
comments
on
that
and
potentially
address
the
comments
before
next
week's
call,
so
maybe
we
can
get
that
document
or
that
PR
emerged.
A
A
O
A
Sure,
if
you
were
here,
not
okay,
Matt
are
you
there
I'm
here
duck
excellent
Clemens
I
got
you
Louie
I
heard
Cathy
I
heard.
Is
there
anybody
who
is
not
on
the
agenda
or
a
list
of
attendees
with
a
star
I?
Think
I
might
have
everybody
all
right.
In
that
case
we
have
a
whole
two
minutes.
Is
there
any
other
topics
who
would
like
to
bring
up
that
we
can
quickly
hash
through
in
two
minutes.
P
I
just
wanted
to
ask
one
question
about
cloud
events.
So
are
the
cloud
events,
something
that
we're
expecting
cloud
providers
like
AWS
and
Azure
to
support,
or
is
this
something
where
we
kind
of
take
events
from
the
cloud
providers
and
then
convert
it
into
a
more
standard
format?
Just
to
help
me
understand
how
this
is
going
to
be
working
in
practice.
B
E
Are
we
seeing
any
thing
in
the
industry
other
than
from
I
sure
I,
someone
from
Azure
I
met
at
dr.
Cohen,
told
me
that
the
previous
back
for
event,
grits
was
very
similar
to
spec
for
cloud
events
being
a
JSON
body
with
some
of
the
same
elements.
What
about
were
we
have
a
much
more
varied
spec
like
the
AWS
suspect
for
events?
Are
we
seeing
any
any
interest
that
I.