►
From YouTube: CNCF Serverless WG Meeting - 2018-08-30
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
Thank
you
all
right.
It's
three
up
to
the
hour,
we'll
finish
with
the
attendees
list.
Later,
let's
jump
right
into
it,
I'm
sorry,
who's
that
I
thought
I
heard
someone
speak
in
there.
Okay,
all
right,
let's
go
through
the
a
I,
see,
there's
anything
that
jumps
out.
Is
there
anybody
ie
who
want
to
mention
their
AI
status?
That's
worthy
of
note,
I!
Think
most
people
know
about
their
a
eyes.
I
just
need
to
find
time
to
do
it.
A
A
We
are,
we
may
do
a
beating.
It
will
not
be
an
official
meeting.
Cuz
I,
don't
believe
we
have
quorum
is
on
the
doodle
poll.
I.
Guess
that
actually
closes
today.
Double
check,
I'm,
pretty
sure.
Actually,
I
know
what
there
was
no
new
votes
actually
I'm.
Sorry.
This
is
probably
wrong.
This
probably
should
have
said
Seattle
anyway,
oh
I'm,
sorry,
but
like
fusion,
it's
been
a
busy
week,
I'll
clean
up
that
stuff,
but
I
dizzy.
A
B
A
A
But
I
think
I
think
I
just
messed
up
on
this
cuz
I
think
we
already
decided.
We
were
not
going
to
do
a
face-to-face
meeting,
an
official
face-to-face
meeting
at
Shanghai
cuz
we
didn't
get
quorum,
but
I
did
put
a
doodle
plug
there
for
Seattle
I
think
that
one's
still
going
so
I'll
fix
this.
We
have
time
for
that
stupid
poll
to
finish
up,
so
we
need
to
worry
about
it,
so
I
apologize
for
that.
Little
bit,
confusion.
You.
A
Community
time
this
is
a
little
time
for
people
who
don't
normally
join
the
call
for
just
part
of
the
community.
In
general,
they'd
like
to
bring
up
a
topic
for
discussion,
we
usually
spend
ten
minutes
or
so,
if
there's
a
topic,
is
there
anybody
on
the
call
who
maybe
night
it
is
a
regular?
Was
the
topic
they'd
like
to
bring
up.
A
D
Yeah
yeah
for
sure,
so
how
can
I
find
where
to
or
should.
A
Yeah,
you
could
send
II
know
to
the
group,
but
usually
the
best
way
is
to
go
to
the
github
repo
I,
think
it's
github.com,
slash
cloud
events,
slash
spec
and
just
open
up
an
issue
there
with
your
concerns,
and
then
you
can
get
some
offline
discussions
going
and
see
where
we
land,
okay,.
E
A
F
G
E
It's
for
me
Luciano
one
example:
dot
we
were
discussing
correlation
ID.
Maybe
I
have
a
payroll
of
thousands
of
employees
that
I
have
to
pay.
I
start
just
one
even
to
pay
all
the
all
the
employees,
but
I
have
thousands
of
events
in
their
accounts
of
each
one
and
in
the
future,
I
want
to
track
that
all
the
the
deposits
was
started
by
the
same.
Even
there
was
the
payroll
event.
A
F
So
maybe
I
think
you
know
maybe
one
way
to
how
about
we
have
you
know
if
you
can
present
your
use
case
in
next
meeting
and
then
we
can
probably
work
on
that.
Oh
I
can
work
with
you
on
the
Yosuke
and
the
use
case,
and
then
we
can
see
how
that
solved
and
then
present
it
back
to
the
team.
Does
that
work.
F
Actually,
I'm
thinking,
because
this
for
this
correlation
idea,
I
think
you
know
different
I-
feel
some
people
still
not
clear
about
what
are
the
use
cases.
So
what
is
the
problems?
Probably
it's
good.
You
know
in
the
next
meeting
you
know
we
present
some
use
cases.
I
can
present
some
use
cases
which
people
has
put
in
what
flows
back
I
can
go
through
those
or
and
there
you
know
I
think
Lucy
know.
You
can
also
present
your
use
case.
So
people
has
a
better
understanding
of
you
know
what
problems
were
trying
to
solve
and
they're.
A
I
think
the
idea
of
gathering
all
the
use
cases
that
people
want
to
address
with
this
I'm
trying
to
figure
out
whether
a
Google,
Doc
or
one
issue
that
people
add
comments
to
it
would
be
a
better
way
to
go.
You
wanted,
you
want
to
decide
Kathy,
which
way
you'd
like
to
go
with
that
since
I
think
this
is
a
topic.
That's
very
you
care.
What
about.
F
Yeah
I
think
you
know
so
so.
There's
I'm
gonna
put
a
link
that
floating
in
this
in
the
meeting
minutes
and
then
people
can
in
that
workflows
back
there
are
section
on
use
cases.
People
can
go
through
those
use
cases,
but
I
assume.
Even
you
know,
even
you
go
through
the
document,
those
use
case
section.
You
might
still
have
some
questions,
so
I
I
think
it's
a
and
then
you
know
I
can
probably
present
a
give
me
some
time.
F
Next
meeting
I
can
quickly
go
through
those
use
cases,
or
if
people
have
questions
we
can
another
way
is
you
know
people
have
questions
and
then
you
know
we
can
help
answer
the
questions.
If
all
those
things
it's
not
just
a
lot
of
different
people
putting
those
use
cases,
so
other
people
can
also
jump
in
to
help
answer
questions.
That's
a
one
way.
Another
way
is
I
just
present
those
use
cases,
so
they
should
be
well.
A
I
think
presenting
the
use
cases
would
definitely
be
good,
but
I
do
think.
It'd
also
be
useful
to
allow
a
place
for
people
to
put
their
use
cases
down
in
advance
of
next
week's
phone
call
and
I'm
trying
to
figure
out
whether
the
have
you
best
done
as
a
Google
Doc
or
as
an
issue
I'm,
leaving
a
little
more
towards
an
issue
that
way
it's
saved
within
our
repository
for
weakening.
So
we
can
go
back
and
reference
it
later
so
Kathy.
A
F
A
The
use
case
is
a
document
that
we
have
is
I,
think
slightly
different
I,
think
that's
a
use
case
sake
for
where
we
think
cloud
events
will
be
used
and
how
how
could
solve
those
particular
problems.
I
think
what
we're
doing
here,
though,
is
were
more
of
an
information
gathering
stage
where
we're
trying
to
figure
out
which
use
cases
we
want
to
support
and
so
I
think
that's
I
think
it
needs
to
be
a
little
bit
more
freeform.
A
F
That's
fun,
but
I
think
you
know
those
uses
the
document
in
the
workflow
I
already
know
people
are
putting
a
lot
of
effort
put
in
that
your
cases
and
also
it's
a
discussed
and
the
Greek
bank.
You
know
that
group
of
people
which
should
be
solved
you
know
for
relate
to
the
crowd
events,
so
I
think
that
part
should
also
be.
E
H
H
I'm
saying
every
time
new
people
come
to
the
cold
and
we're
trying
to
expand
again.
What's
the
difference
between
the
envelope
and
the
message
and
I
think
it
need
to
be
clarified
that
things
that
are
in
the
envelope
need
to
serve
things
that
observe
the
message.
We
don't
try
to
put
comments
in
the
intro
I.
C
With
that
I
also
agree,
I
think
it'd
be
useful,
even
beyond
sort
of
the
workflow
to
have
a
repository
of
use
cases
where
people
can
ask
questions
and
people
can
respond
to
it.
So
I
don't
care
whether
it's
an
issue
or
Google
Doc,
but
the
thing
I
think
there's
the
use
cases
that
we
officially
support
and
that
we
bring
into
the
standard
but
I
think
from
a
community
perspective.
A
D
A
I
think
we
may
have
tried
in
the
past
I
think
the
primer
does
try
to
touch
on
this
I.
Don't
think
it
does
a
good
enough
job
though,
but
I
think
that's.
The
attempt
is
that
we'll
use
the
primer
to
help
explain
topics
that
don't
necessarily
go
into
the
spec
itself,
but
our
help
explain
how
to
use
the
spec
and
some
of
the
decisions
we
made
behind
what
the
spec
says,
but
we
are
running
a
little
low
on
time
here
and
I.
A
Don't
want
to
rattle
too
much
on
this,
because
obviously,
there's
gonna
be
a
very
important
topic.
So
tell
you
what
why
don't
I
take
the
X
action
items
create
an
issue
to
gather
people's
thoughts
requirements.
Use
case
is
very
want
to
call
it
around
correlation
and
and
see
where
that
discussion
leads
us
as
part
of
that
issue.
What
I'll
also
do
is
create
another
issue
to
make
sure
that
we
we
touch
on
this
topic
of
envelope
versus
data
worst
last
body,
whatever
you
want
to
call
it.
J
I
There's
not
that
many
examples,
I
know
because
the
SDKs
are
still
being
worked
on
as
soon
as
the
SDKs
are
in
I
would
say
better
state,
but
more
ready
people
are
gonna
start
building
stuff
me
included
and
more
examples
are
gonna
come
out,
but
right
now
it's
still
a
bit
early
due
to
the
lack
of
easier.
The
case.
A
I
F
Dog
gentleman
who
spoke
before
you
I
think
raised
a
good
point.
I
think
you
know
we
really
need
to
have
a
document
that
is
plan.
Well,
what
is
what
have
issues?
So
people
are
on
the
same
page,
otherwise,
I
think
this
topic
is
going
to
be
dragged
down.
You
know
a
long
time,
so
I
I
can
volunteer
to
write
up
something
on
that.
So
to
see
whether
it
know
people
and
then
people
can
add,
you
know
more
input
or
comments
on
that.
F
A
F
No,
it's
I'm
not
of
a
middle
for
the
payroll.
That's
a
very
easy,
oh
I'm,
talking
about
about
correlation
that
you
know
why
we!
So
what
why
we
need
it.
It's
not!
Okay,
I
think
I'm
not
going
to
talk
about
the
solution.
I'm
going
to
just
talk
about
the
problems
you
know.
Basically,
we
need
to
correlate
multiple
events.
F
A
K
Is
this
is
Gemma
I
have
a
question
I'm,
not
sure
if
it's
a
process,
question
or
just
a
group
question
when
we
look
at
propagating
other
contexts
around
so
I
I
was
looking
at,
for
instance,
open
tracing,
which
seemed
to
be
something
that
we
could
propagate
that
context
through
the
extensions
in
the
context.
Where
do
you
see
those
sort
of
specs
or
guides
living?
Would
you
see
that
as
a
an
open
tracing
best
practice
or
as
a
cloud
events
best
practice
so.
A
L
K
So
so,
when
we
look
at
something
like
open
tracing
yeah,
which
has
defined
mechanisms
for
propagating
tracing
contexts
across
either
different
transports,
we
are
HTTP
and
messaging
and
whatever
it
would
seem
that
you
could
map
those
similar
constructs
into
the
extensions
in
the
context
of
a
know
of
a
cloud
event
yeah
and
then
unify
that.
So
so
the
question
then
becomes.
Is
that
an
open
tracing
piece
of
work
or
is
that
a
cloud
events
piece
of
work
so.
L
To
have
a
particular
metadata
that
has
particular
types
and
has
particular
names
you
want
to
meet
that.
You
want
to
harmonize
that
across
multiple
transports.
Then
the
right
way
to
do
this
is
to
define
how
you
project
that
into
various
transports
or
envelopes,
as
you
would
do
this
with
with
any
firearm.
L
It's
some
XML
that's
to
find
an
open
trace,
and
you
would
basically
slot
that
into
a
soap
pattern
as
this
and
and
I
looked
at
this
here
in
a
very
similar
way,
is
that
you
define
an
extension
and
put
that
into
the
extension
catalog,
probably
external
or
here,
and
that
probably
also
depends
on
the
character
of
the
project
and
whether
we
would
take
that
here
into
the
our
repo
like
if
it's
open
runs
on
the
umbrella
office
foundation,
etc
and
then
define
it
as
the
extension,
and
you
would
use
those
things
as
this.
L
So
for
correlation
more
broad,
but
because
that's
a
correlation,
I'd
correlation.
More
broadly,
if,
let's
say
there
is
the
and
I'm
just
making
one
up
right
now:
kind
of
the
open
home
automation
alliance,
alright-
and
they
say
we
need
to
have
a.
We
need
to
have
a
standard
for
defining
where
device
is
hanging
in
a
home
and
there's
a
notion
of
a
rule,
geometry
and
there's
a
notion
of
a
floor
and
there's
a
notion
of
a
unit
and
there's
a
notion
of
a
building
per
se
and
there's
all
that's
all
correlation
data.
L
Then
that
would
also
probably
go
into
an
extension
that
then
everybody
who's.
You
know
building
systems
that
have
to
do
with
a
home
automation,
standard
uses,
but
they're
only
applicable
to
that
subset
of
consumers
and
publishers,
and
they
don't
affect
the
standard
per
se
like
everybody.
Who's
using
cause.
Events
must
be
using
and
that
home
automation
standard
that
everybody
who's
using
that
home
automation
standard
will
use
that
extension
to
lay
it
at
the
top
of
called
events.
A
N
Ahead,
sorry,
just
as
an
FYI,
this
doesn't
need
to
be
distantly
a
hypothetical
discussion,
so
we
wanted
open
tracing
to
work.
We
did
some
research
from
what
I
could
tell
it
seems
like
open
tracing
all
over
HTTP
is
based
on
the
strip
retracing
and
that's
why
there's
a
distributed
tracing
extension.
My
attitude
was
a
like
this.
Beck
was
about
one
yet
and
we
are
the
smaller
body
and
the
WT
Beck
is
stable.
There's
there
was
very
little
understanding.
A
Okay
with
that
I
think
we
got
the
call
time
because
it's
supposed
to
be
bounded
by
ten
minutes,
I
think
we've
got
a
little
bit
over.
However,
that's
not
to
indicate
that
the
topics
that
are
brought
up
here
aren't
worthy
of
further
discussion.
What
I
would
ask,
though,
is
please,
don't
hesitate
to
open
up
issues
in
our
repo
for
some
of
these
topics.
A
That
way
we
can
get
some
fly
discussions
going
if
we
don't
have
to
limit
ourselves
to
these
10-minute
time
slots
for
some
of
these
things,
because
some
of
these
things
are
rather
important
that
we
get
settled
and
I
think
discuss
those
two
issues
with
you
best
foot
forward
on
those.
So
thank
you
guys
very
much,
but
we
didn't
even
record
so
I,
don't
think,
there's
any
update
on
the
SDK
or
on
the
workflows
workgroup.
So
let's
go
ahead
and
jump
right
into
the
PRS.
A
A
I
A
F
I
read
Claire
I
change,
my
vote
and
then
later
I
think
dog
will
work
what
we
sneak
together.
Putting
on
your
PR
to
clarify
those
points
you
either
sports,
which
means
we
allow
quarterbacks
with
his
type
clearly
defined,
but
with
its
content
open
for
open.
I
A
B
A
F
A
A
J
A
M
A
Dan
Barker
is
gonna,
go
make
the
changes
to
our
website
to
pick
up
the
new
logos.
I
will
start
the
process
of
getting
the
stickers
updated
with
the
new
logos
and
then
I
do
have
a
coupon
code
I'll
order,
some
stickers
and
I
believe
we
should
probably
get
them
in
time
for
coop
con.
It
was
shanghaied
or
at
least
Seattle
I
would
assume
so
I'll
work
on
that.
In
the
background,
all
right
moving
forward,
we
have
to
PRS
that
were
open
for
additional
transfer
bindings
if
I'm
incorrect
Lee.
A
There
was
a
general
sense
on
both
of
these
that
they
may
not
have
met
the
new
bar
relative
to
acceptance
of
new
transports
and
bindings,
but
people
were
given
another
week.
In
fact,
it's
not
been
two
weeks
to
look
these
over.
How
does
anybody
change
their
opinion?
Is
there
any
reason
for
us
to
accept
either
of
those
transport
bindings.
M
I
submitted
it
early
yesterday
and
there
was
like
or
sometime
yesterday,
it
was
just
kind
of
in
response
to
the
whole
discussion
on
that
thread
and
I
reread
the
article,
the
the
qualifying
topics
and
I,
don't
think
it's
being
you
didn't
feel
like
it
was
reflect.
The
wording
was
reflective
of
the
conversation
right.
M
We
had
this
conversation
many
weeks
ago
and
were
like
it's
hard
to
describe
de
facto,
and
so
now
that
we're
looking
at
it
through
the
lens
of
another
test
case
I
I
was
like
well
I'm,
not
sure
that
de
facto
was
like
needs
to
be
as
narrow
as
it's
literally
interpreted.
If
you
read
that
paragraph
okay.
A
Did
okay,
I
apologize
them
for
missing?
I
was
traveling
all
day
yesterday,
so
I
missed
that.
So
tell
you
what
I
think
Rachel
I
think
you're
right,
then
why
don't
we
hold
off
on
this
decision
of
these
two?
Since
that
says,
we're
headed
towards
closing
them?
It
does
no
harm
to
keep
it
open
end
of
the
week
or
so
so
we
can
look
at
your
PR
Sarah
that
you
just
opened
up
and
we'll
discuss
that
next
week
to
see.
A
A
O
Yes,
so
what
it
is
I
look
or
issue
is
that
if
you
put
too
much
stuff
into
extension,
then
into
extensions
that
we
may
run
into
technical
stuff
like,
for
example,
the
HTTP
transport
we
put
this
in
the
headers
and
most
HCP
servers
will
at
some
point
reject
requested
the
header
data
gets
too
large,
so
I.
Actually
you
an
issue
where
describe
side
for
a
couple
of
things
like
nginx,
for
example,
as
limit
of
eight
through
a
bias
like
the
ADA
was
API
gateway,
has
ten
cool
bars
and
so
on
so
yeah.
O
So
this
is
like
sort
of
just
guidance
saying
you
shouldn't
put
too
much
extension
extensions
into
your
cloud
event,
because
if
that
happens
to
be
transported,
for
example
over
HTTP,
then
you
have
so
much
header
data
that
your
request
will
be
rejected
and
I
think
this
isn't
only
apply
to
HTTP.
It
will
also
apply
to
other
transport
bindings,
but
is
this
just
used
as
an
example?
All.
A
A
Okay,
on
a
previous
call,
and
even
two
weeks
coming
out
her
percent
sure
when
we're
talking
about
a
possible
face-to-face
meeting,
actually
I
think
actually
might
have
been
for
the
USS
summit
this
week
it
was
brought
up.
I.
Think
Clemens
may
have
mentioned
it
that
we
may
have
overlooked
the
fact
that
our
governance
stock
doesn't
talk
about
face
to
face.
Meetings
have
to
be
announced,
but
there's
our
in
time
period
in
advance
to
give
people
a
chance
to
get
travel
approval
and
set
up
travel
plans.
A
Stuff,
like
that,
so
I
took
the
action
item
to
create
a
PR
to
say
that
we
have
to
do
at
least
four
weeks
in
advance
to
give
people
plenty
of
warning.
So
here's
my
PR
to
add
that
to
the
governance
stock.
Hopefully
it's
fairly
straightforward,
just
says
meetings.
Might
these
face
meetings
will
be
announced
four
weeks
in
advance?
A
F
Can
you
know
I
use
that
for,
for
example,
for
correlation
purpose,
like
you
know,
for
for
I
present,
some
use
cases
before
and
like
for
travel
requests,
somebody's
application
and
that
late
identity
able
could
be
the
travel
request,
ID
or
the
employee
ID,
and
then
you
know
the
consumer
can
use
that
to
correlate
that
event
correlate.
You
know,
travel
requests,
event
with
a
travel
approval
event,
and
there
are
many
such
cases
like
I
present
another
use
cases
which
is
a
particularly
detection
system.
That
could
be
a
motion,
motion,
detection
event
and
the
window
opening
event.
F
And
then
you
know
people
can
use
the
identity
label
of
these
two
events
to
correlate
them
to
the
same
service
application
instance.
So
this
is
for
that
purpose.
I.
Let
me
see
here
yeah,
basically,
and
that's
about
it.
I
think
this
correlation
issue
has
been
brought
up
multiple
times
in
previous
meetings.
In
order
to
support
those
use
cases,
I
would
like
the
crowd.
I
think
we
neither
crowd
events
to
have
a
place
for
the
event
producer
to
put
in
this
information,
and
this
is
an
optional
field.
F
If
no
even
producer
doesn't
have
identity
labels,
then
there's
no
need
to
have
this
field
and
also
from
the
event,
consumers
point
of
view.
If
you
know
it
doesn't
need
this
information,
each
can
just
ignore
the
whole
that
this
bag
doesn't
need
to
go
into
despite
to
decode
every
label.
Yeah,
that's
that's
about
it.
Okay,.
A
F
B
F
P
P
P
So
for
most
of
the
cases,
these
information
possibly
be
good
enough
for
the
service
function,
but
for
some
other
cases
the
service
function,
my
needed
to
know
who
uploaded
the
image
to
pocket
one
or
bucket
two
or
package
free
and
possibly
need
some
other
extra
information
about.
You
know
some
sort
of
identification
context
of
the
event.
So
I
think
this
is
one
possibly
a
possible
use
cases
I
believe
there
are
some
other
use
cases.
I
think
this
is
one
thing
I
want
to
comment.
P
The
other
thing
is
that
this
attribute
well
I,
think
it
doesn't
matter
you
call
these
attribute
a
bag
or
not
to
me
least,
attribute,
is
defined
to
be
a
map
of
map
of
key
key
value
pair
and
the
value
of
these.
The
value
of
the
key
is,
is
also
Stringer's
defined
by
Cassie.
So
it
doesn't
really
matter
it's
a
map
attribute
or
a
string
attribute.
We
can
also
make
it
a
string
attribute
if
we
want,
by
making
it
a
string
of
comma-separated
string,
a
comma-separated
key
value
pair.
P
Another
thing
I
want
to
say
that
the
identity
context
for
each
event
is
very
different
from
other
events,
so
some
event
can
I
was
just
a
zero
identity
context,
but
some
others
may
have
can
or
some
others
may
have
more
contact.
So
the
information
related
to
light
and
fake
is
quite
different,
so
this
kind
of
dynamic,
it's
kind
of
harder
to
put
such
kind
of
information
into
the
top
level.
So
it's
I
think
is
spydra.
It's
a
good
idea
to
put
all
identity
related
information
into
the
into
least
attribute.
P
A
I
B
Think
I
said:
let
me
ask:
if
Scott
did
you
say?
Would
you
want
to
say
already
so
my
main
point
about
this
is
I
think
that
identity
labeling
is
one
use
case
and
the
general
case
of
labeling
events
is
actually
interesting,
and
so
my
suggestion
is,
is
that
we
not
use
a
map,
use
a
list
of
labels
and
you
can
optionally,
give
a
label
a
type
and
that
type
could
be
identity.
B
So
it's
it's
a
little
more
generic,
it's
a
little
more
extendable
and
it
applies
to
every
use
case.
That
applies
to
allowing
an
event
to
be
labeled,
not
just
identity.
So
if
you
allow
this
to
be
added
to
the
top
level,
and
then
you
also
want
to
label
events
in
the
future
now
you
have
two
places
where
generic
labels
get
applied
to
your
event
and
why?
A
I
So
there
were
two
topic
topics
discussed,
first
of
all,
Chatty
Cathy
labels
and
then
identity
in
AWS,
which
might
be
cognitive,
identity
or
access
key
secret
key
or
whatever
these.
These
really
have
to
be
part
of
the
another.
Payload
I
brought
up
this
issue
in
the
github
issue
for
event
versus
penalty.
There
are
limits
on
what
we
can
put
as
payload
and
I'm,
not
sure
that
this
is
the
best
place
to
be
for
them.
But
then
again,
this
would
be
heavily
impacted
by
the
security
discussion,
especially
by
signing
an
encryption.
A
F
Didn't
quite
get
that
question,
so
your
so
the
reason
you
know
we
put
a
bag
here
is
because
if
we
do
not
have
this
back,
there
will
be
a
lot
of
information.
As
we
know,
the
fact
is,
there
are
a
lot
of
information.
I
mean
for
different
events.
The
identity
could
be
different
right,
for
example,
for,
like
you
say,
for
travel
request'
event,
the
identity
will
be
like
the
travel
request.
F
Id
for
our
loan
application
event
is
where
a
loan
application
ID
right
for
a
motion
moving
event,
a
motion
detection
event:
the
the
label
could
be
the
house
address,
so
they
are.
If
we
do
not
have
a
black,
there
will
be
a
prolific
variation
of
a
lot
of
email
and
labels.
So
I
think
it's
better.
We
have
a
back
of
course
for
each
specific
event.
F
I
don't
expect
the
identity
labels
to
be
a
knot,
because
for
each
specific
event
the
event
producer
will
just
put
you
know
whatever
it's
the
identity
and
a
few
I
dental
Abel's
in
innate,
so
I,
think,
and
also
this
is
a
good
for
consumer
point
of
view.
If
the
consumer
does
not
need
it,
the
consumers
can
just
pass
this
back
name
and
they
generate
that's.
It
do
not
need
to
go
into
to
pass
every
label
and
also
good
for
debugging
purpose.
F
So
this
is
my
Mike.
This
is
why
I
think
he
knows
better
to
do
this
way
and
also
we
do
find
that
you,
because
sometimes,
if
we
leave
the
type
open,
I,
think
some
transport
probably
will
have
problem.
So
that's
why
we
defined
what
is
a
content
type.
It
would
be
a
key
value
string
with
a
key
to
view
a
string
and
the
radio.
It's
also
a
stream
format.
A
Okay,
so
I
apologize,
I
could
stay
monk,
but
to
peeny
I.
Think
you
had
your
hand
up.
Did
you
want
to
say
something
yeah.
Q
They
still
have
this
free
from
definition
between
producers,
and
it
seems
that
issue
belong
in
the
payload,
but
on
the
other
hand,
it's
useful
or
outside
the
payload
when
you
want
to
drop
the
vent
or
something
and
I
think
it
should
be
clarified.
What
the
how
this
relates
to
that
discussion
since
it's
it's
doesn't
seem
that
the
purpose
is
very
clear
in
that
regard.
Q
M
F
Think
that
today,
I
think
if
we
make
it
too
broad,
if
we
say
okay,
we
take
the
identity
out,
then
they'll
be
up.
Then
the
producer
had
put
just
random
information
inside
it.
I
hear
other
comments
say
you
know
yeah
the
producer
put
random
information
inside.
So
that's
why
I
think
it's
better
to
scope
it
to
be
identity.
Otherwise
you
know
it's
going
to
be
a
blow
unless
it
upload
it
and
also
I,
want
to
suggest
the
point,
and
that
you
know
is
its.
F
A
I,
don't
think
anybody
has
their
hand
ups.
Let
me
ask
a
question
because
I
it
doesn't
seem
like
it's
very
clear
to
me
that
everybody
is
on
the
same
page
relative
to
the
purpose
behind
this
new
bag.
That
Kathy's
looking
to
add.
So,
for
example,
Kathy
I
know
you
talked
about
it
as
being
for
identity
purposes,
but
to
me
the
word:
identity
isn't
very
precise.
It
can
be
most
different
things
different
people
and
then,
when
I
take
a
look
at
what
Scott
and
thank
you
Scott
for
the
link.
What
Scott
posed
here
as
an
alternative?
A
This
seems
like
a
very
sort
of
generic
value,
pair
kind
of
tagging
mechanism
right
and
it's
not
necessarily
specific
to
identity,
Perseids
any
label.
You
want
to
add
to
there
and
I'm,
not
saying
which
one
is
necessary
right
or
wrong,
but
it
just
seems
to
me:
we
don't
necessarily
have
agreement
on
what
the
purpose
is
behind.
What
we're
trying
to
solve
here.
Is
it
just?
A
We
won't
have
a
mechanism
to
be
able
to
add
additional
properties
for
some
purpose
in
general,
or
is
it
only
for
one
very
slice,
thin
slice
of
semantic
meaning
behind
it,
and
if
it's
that
very
thin
slice
I'm
not
sure
that
the
word
identity
means
enough
to
me
to
understand
very
clearly
what
would
go
in
that
bag
versus
not
because
when
I
see
it
it
has
currently
defined.
I
would
sit
there
and
wonder
why
isn't
eventid
part
of
this
identity
bag
because
obviously
I
event
ID
is
the
identification
of
the
event
is
a
unique
identifier.
A
A
And
that's
part
of
a
concern
that
I've
had
in
the
past,
which
is
this
feels
like
a
generic
bag
for
almost
anything,
even
though
it's
called
identity,
identic
identification
or
identity,
it's
gonna
get
used
for
just
about
anything
because
it's
so
loosely
defined,
which
means
we're
adding
yet
another
accessibility.
Point
and
people
aren't
gonna,
be
clear
where
to
add
things.
A
Q
F
So
I
think
I'm
here
in
two
different
perspective
on
this
one
saying
you
know
this
identity
level
is
too
restrictive
another.
You
say
you
know
it's
too
broad
because
you
say
I
think
doc.
Your
point
is,
you
know,
I
didn't
tell
you,
no
it's
a
and
you
can
put
people
can
put
anything
inside
it
and
but
I
see
right
here.
I
heard
another
comment
before
say:
you
know:
identity.
Just
you
call
it
label
because
identify
to
restrictive.
F
A
Q
F
Think
here
I
want
to
address
it.
One
of
your
point
is
event
ID,
because
for
you
you
have
event
ID
right.
You
know
from
event
source.
It
gave
an
event
ID
it
that
even
ID
has
you
know
and
there's
nothing
to
do
so
so
for
the
use
case
that
that
travel
request,
you're
gonna
have
the
event
ID,
but
then
how
we
correlate.
This
travel
request
event
to
the
travel,
prevent
those
two
IDs
has
nothing.
You
know
they
do
not
have
any.
You
know
on
relationship.
F
So
that's
why
we,
you
know
if
the
even
producer
can
put
like
at
ribose
basic
travel,
request,
ID
or
the
tribal
applications.
You
know
employee
ID,
something
like
that
that
will
help
doing
the
correlation.
So
that's
why
you
know
event.
Id
is
not
enough,
so
this
is
so.
This
is
a
place
for
the
producers
with
additional
identification
information
on
which
you
know
think
you
know
it.
It
has
that
okay.
A
A
S
N
Wanted
to
respond,
there's
a
suggestion
to
change
it
to
routing
labels.
I
actually
prefer
Kathy's
approach
of
labeling
of
naming
something
based
on
what
it
is,
not
how
you
expect
it
to
be
used.
I
think
that
it
shouldn't
look
weird
in
a
system,
but
we
find
any
use
for
a
property.
For
example,
I
could
use
this
for
filtering
as
well.
Okay,
cool.
A
P
A
Like
yeah
and
with
that
I
think
we're
almost
out
of
time,
so
Kathy
I'm,
trying
to
figure
out
the
best
way
to
move
forward
here.
I
say
two
potential
paths.
One
is
to
set
up
an
offline
meeting
to
have
people
who
are
interest
in
this
topic.
Discuss
it
but
I
know.
Sometimes
those
meetings
aren't
as
productive
as
they
could
be
because
of
limited
attendance.
The
other
option
is
to
just
bring
this
up
on
next
week's
phone
call
to
continue
the
discussion
as
long
as
you
don't
have
anything
else
pressing
to
work
on
it.
A
F
Think,
probably
you
know
in
the
next
meeting
after
we,
you
know
present
those
use
cases,
so
people
can
also
think
about
that
and
then,
if
this
identity
is
not
I,
this
is
a
name.
I
can
come
up
with
if
people
think
this
I
didn't
take
whatever
name.
It's
not
good.
People
can
suggest
another
name,
but
I
think
for
labels.
I
think
he's
a
little
bit
too
broad.
That's
my
take,
but
if
you
know
someone
to
come
up
with
a
better
name
so
so.
A
A
F
A
Okay,
so
let's
see
how
the
conversation
goes
next
week
after
we
have
the
correlation
discussion
and
see
where
maybe
something
will
pop
up
and
someone
will
have
a
brilliant
idea
and
how
to
move
forward
on
here.
Obviously
this
is
a
very
touchy
subject
or
a
complicated
one.
I
should
say:
okay
with
that,
we
have
three
minutes
left
and
I'm,
not
sure
we
have
a
whole
lot
of
time
to
do
anything
needy.
A
A
N
A
Got
you
Doug,
Doug
MIT
I
can
never
pronounce
your
last
name.
I
apologize.
I,
don't
see
him
still
on,
so
they
missed
him
again.
This
time.
Okay,
we
already
have
Luciano
I,
believe
we
heard
you
that.