►
From YouTube: CNCF Serverless WG - 2018-05-10
Description
Join us for KubeCon + CloudNativeCon in Barcelona May 20 - 23, Shanghai June 24 - 26, and San Diego November 18 - 21! Learn more at https://kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy and all of the other CNCF-hosted projects.
Join us for KubeCon + CloudNativeCon in San Diego November 18 - 21. Learn more at https://bit.ly/2XTN3ho. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy and all of the other CNCF-hosted projects.
B
A
A
C
A
A
A
A
H
B
B
A
H
F
A
E
A
A
It
I
got
it
make
it
easy
on
me.
Appreciate
that
Charlie
Pitkin
I,
don't
have
you
Gary
got
your
nevermind
sorry
ginger
collision,
either
its.
A
All
right,
let's
see
who
else
went
by
the
list:
Shawn
Smith,
hi
I'm,
with
Oracle
Oracle
got
it.
Thank
you
Steve.
Oh
you,
there
I
mean
all
right.
Thank
you.
A
A
And
here's
the
agenda
duck
again,
if
you
guys
want
to
add
your
name
to
the
list,
just
don't
put
an
asterisk
there,
we'll
get
you
our
roll
call
later
all
right,
Tay
what
it's
for,
after
when
I
go
and
get
started.
So
what
I
was
thinking?
I'm,
sorry,
so
I'm
gonna,
say
something
okay,
so
today,
I
would
I
want
to
do.
Is
focus
a
little
bit
on
process
for
lack
of
a
better
phrase.
A
We
are
part
of
the
servlets
working
group,
but
because
we're
actually
producing
a
specification,
it's
a
little
bit
fuzzy
in
terms
of
you
know
what
people
actually
call
us
and
what
our
role
is,
or
classification
is
within
the
scenes.
Yes,
so
we've
been
asked
to
make
it
a
little
more
formal
and
we
decided
to
try
to
join
as
a
sandbox
project,
I
believe
the
the
biggest
stumbling
block
for
us
to
go.
A
The
next
level
up,
which
is
incubator,
would
be
at
least
three
different
uses
in
production,
which
I
don't
think
we
have
there
yet
I
think
Microsoft
might
be
only
one
who
can
claim
that
as
it
today
as
an
official
supported
thing,
but
anyway,
we'll
start
with
sandbox,
but
so
I'm.
Mentioning
this
to
you
guys
for
two
people
for
two
purposes:
one
so
you're
aware
of
it,
but
two
to
please
look
over
the
status
and
proposal
itself,
mainly
to
make
sure
that
one
we
don't
have
anything
incorrect
in
there.
A
Any
additional
text,
you
think,
might
be
useful
in
there
just
go
ahead
and
you
know
either
ping
me
or
add
a
note
to
the
PR
and
I'll
get
that
added
in
there,
but
anyway,
with
that
I
just
want
to
bring
it
up
to
you
guys.
You
know
present
FYI
kind
of
a
thing
and
you
can
ask
he
goes
through
with
you.
When
did
you
get
a
chance?
Are
there
any
questions
or
comments
about
that.
A
All
right
case
next
excuse
me,
as
I,
was
going
through
the
list
of
issues.
I
came
across
these
four
that
I
thought
could
technically
be
closed
because
of
mainly
because
I
thought
we
covered
it
with
other
work
that
we've
done
now.
I
have
made
comments,
I
believe
that
all
these
saying
that
I
think
they're
ready
to
be
closed
and
no
one's
spoken
up
and
objected,
and
in
some
cases
some
people
have.
A
You
know
given
a
thumbs
up
to
do
it,
but
I
felt
a
little
uncomfortable
just
going
off
and
closing
them
without
making
you
guys
aware
that
I'm
going
to
be
doing
that
at
the
end
of
today.
So
please
look
at
these
four
issues.
If
you
believe
they
should
remain
open,
just
add
a
comment
and
I
won't
close
it
even
if,
for
some
reason
you
don't
get
a
chance
to
do
it,
we
can
always
reopen
them
later,
but
I
just
want
you
guys
to
be
aware
of.
M
A
Thank
you,
criminals,
okay,
so
anyway,
everybody
else.
Please
take
a
look
at
that.
If
you
think,
for
any
reason,
they
should
remain
open.
Just
add
a
comment
and
I'll.
Keep
it
open,
but
at
end
of
today
is
the
deadline
for
keeping
those
or
for
me
to
close
those
just
wanted
you
guys
to
be
aware
of
that.
A
Any
questions
or
comments
on
that
alright
cool.
Thank
you.
Alright.
So
let's
talk
about
next
milestones
and,
as
I
said,
we
started
talking
about
this
at
the
face-to-face.
For
you
know
what
why
do
41.0?
What
I'd
like
to
do
is
suggest
a
process
where
we
obviously
use
github.
The
way
it's
meant
to
be
used
so
we'd
like
to
ask
is
that
everybody
who
had
an
idea
at
the
face-to-face
for
what
they'd
like
to
see
included
in
one
point,
though,
to
please
open
up
an
issue
by
default.
A
That's
either
post
1.0
or
not
required
for
1.0,
whatever
the
proper
term
we
want
to
use
is
but
I
figured,
that's
the
safest
way
to
go
right
now
is
to
assume,
as
for
1.0,
until
we
have
a
discussion
and
assume
otherwise,
and
so
with
that
what
I'd
like
to
do
is
I.
Last
night,
I
did
go
through
in
tagged
everything
that
I
thought
was
fairly
obvious
as
a
1.0
topic.
A
Now
that
doesn't
necessarily
mean
it's
necessarily
going
to
be
done
for
1.0,
but
I
think
that
we
have
to
have
the
discussion
for
one
point,
though,
because
if
we
do
decide
to
do
what
was
being
proposed
in
there,
it's
a
semantic
change
or
a
breaking
change.
The
spec,
and
obviously
we
need
to
consider
those
before
1.0,
so
I
went
to
him
tagged
a
whole
bunch
of
1.0,
but
then
there
was
a
set
of
them.
A
Take
the
entire
phone
call.
Is
that
okay
process
with
you
guys
or
for
me,
okay,
okay
and
then
I
figured
at
some
point
that
in
a
future
call
we
can
then
try
to
narrow
this
list
down
to
see
which
ones
you
want
to
do.
You
know
sooner
than
one
point
go
for
now
for
a
sooner
milestone,
but
that's
that
will
come
later
so
collaborate
on
libraries
and
support
tools,
this
one
while
I
opened
it
I
believe
I
open
this
for
urine
now
you're
on
I'm,
sorry.
A
This
was
from
the
roadmap
so
on
row,
map
version
0.4.
We
say:
CLIA
are
buried
on
libraries
and
supporting
tools.
My
initial
reaction
on
this
one
was
that
if
we
do
this,
it's
not
necessary
hard
requirement
for
1.0.
It
could
come
after
one
point,
though,
but
I
felt
a
little
uncomfortable
tagging
it
as
such
without
a
discussion
with
you
guys
first,
what
do
you
guys
think
does
this
have
to
be
done
before
1.0
or
at
least
does
the
discussion
have
to
happen
before
1.0?
A
E
E
A
A
So
say
what
let's
do
this
I'd,
rather
err
on
the
side
of
having
too
much
in
1.0
right
now,
so
today,
what
let
me
mark
this
as
1.0
and
there
you're
on,
let's
go
to
your
SDK
one,
and
you
can
briefly
very
briefly
summarize
this
one
and
we'll
decide
whether
we
want
to
for
sure
to
talk
about
this
before
1.0.
So.
L
You
know,
let's
start
with,
for
example,
we
have
HTTP
which
we
serve
in
high
level
agreed
now
it
looks
like
so
now
we
can
create
an
SDK,
for
example,
you
have
go
or
Java
or
node
for
each
one
of
them.
You
can
have
a
class
where
you're
initializing,
let's
say
the
cloud
event
structure
and
you're
saying
you
know
published.
Essentially
you
need
like
three
methods.
L
You
know
to
initialize
something
to
destroy
it
and
and
to
submit
or
publish
the
the
event
message,
and
then
we
can
create
sort
of
a
pluggable
mechanism
for
transports
initially
starting
with
HTTP
and
then
as
we
starting
to
define
AMQP
or
Kafka
or
other
transports.
We
just
plug
them
underneath
so
we
can
divide
it
to.
You
know,
serve
the
five
major
languages.
L
M
I,
don't
so
I
think
this
is
a
great
effort
to
do
and
I'm
all
for
it
and
just
I
just
think
it's
orthogonal
to
the
snick
work,
yeah
I
agree.
I
F
Wanted
to
make
that
point,
because,
like
urine
and
I
had
a
great
discussion
in
one
of
the
threads,
where
we
basically
proposed
two
different
interfaces
that
had
totally
different
rate
offs,
one
is
zero
copy.
One
is
strongly
typed
and
I'm
a
little
bit
cautious
about
saying
that
we
will
make
each
trade
off
is
like
the
standards
body
or
the
working
group
recommends
exactly
one
SDK
yeah.
A
B
E
L
A
L
A
A
N
D
D
It's
not
being
required
for
window,
though,
but
it's
similar
in
that
there's
really
something
of
a
conformance
tool
to
you
know
to
validate
a
given
providers,
implementation
against
the
spec,
and
so
the
client
SDKs
certainly
facilitate
adoption,
which
is
you
know,
which
who's
the
the
project,
as
we
you
know,
as
it
moves
into
something
like
a
sandbox
or
or
to
that
next
step.
I,
wonder
I!
D
L
A
L
A
A
Okay,
so
tell
you
what
we
were
gonna
get
to
it
eventually,
but
Lisa,
and
you
brought
it
up
here.
Let's
talk
about
this
one
I
originally
marked
this
last
night
does
not
refer
to
1.0,
because
I
didn't
think
it
was
required
for
the
spec
itself
to
reach
one
point,
though,
even
though
I
do
think
it's
obviously
a
very
important
piece
of
work.
A
E
A
E
Completely
different
thing
to
that's:
just
they
are
certified
right
right.
Yeah,
like
kubernetes,
has
a
certifying
it.
That's
completely
different
thing
to
what
this
proposing
it's
still
useful,
they're,
both
they're
both
useful
they
both
should
probably
be
done
over
time.
Otherwise
we
will
all
write
our
own
client,
SDKs.
I.
Think
that's
why
yahrens
raised
this.
It
just
needs
some
I
guess
we
need
some
direction
on
it.
So.
A
D
D
D
That
you're
getting
a
label
and
a
fixing
the
label
is
a
lot
more
involved
because
you
gotta
validate
the
results
and
the
integrity
of
those
results.
If
it's
just
a
tool
to
help
those
that
are
implementing
and
make
sure,
and
then
then
they
maybe
build
it
into
their
CI
systems
to
just
validate
that
they've
they're
continuing
to
your
adhere
to
the
spec
and
there's.
D
A
So
if
we
were
to
scope
this
down
to,
as
you
described
it,
a
tool
that
does
validates
that
you're
spitting
out
a
valid
crowd
event,
is
that
something
that
one
we
did
keep
under
this
github
repo
or
in
our
organization
somewhere?
None
aside
same
repo,
but
you
know
within
the
cloud
events
organization.
Yes,
and-
and
is
this
something
we
need
to
discuss
before
1.0.
D
And
say
yes
to
one
and
then
before
one
so
technically
you
can
one
o
the
spec
without
having
you
know
this
tool
and
the
same
thing
goes
for
client-side
SDKs
I.
Think
in
my
mind
the
client-side
is
the
case.
Help
with
adoption
in
general.
The
conformance
tool
like
this
yeah
I,
guess,
yeah
I,
don't
I,
won't
I
would
hold
opinion
because
I
think
it
you
could.
You
know,
push
it
away.
What.
A
Other
people
think
is
there
something
we
should
that
we
should
discuss
before
1.0
now,
keep
in
mind
just
because
we
tagged
something,
isn't
that
required
for
in
point.
Oh,
it
doesn't
mean
we
don't
get
to
it.
It
just
means
it's
not
necessarily
are
all
hard
requirement
for
us
to
do
so.
But
if
someone
takes
me,
it
makes
the
effort
to
do
it.
It
could
get
out
to
the
agenda.
Do.
E
M
We
have
a
in
mqp.
There
is
a
validator
that
base
just
goes
and
and
looks
at
the
bits
on
the
wire
without
building
a
fancy,
library
and
API
around
it.
Yeah
so
and
then
basically
just
goes
and
checks
that
I
think
that's
even
written
for
HP,
that's
written
in
Python,
and
so
because
it
doesn't
need
to
have
multi-language
support.
You
don't
need
any
of
those
things.
You
basically
just
need
to
see
whether
the
bits
come
fall
down
the
wire
in
the
correct
way,
yep
and
I.
M
B
A
So
at
least
one
person
as
I
mentioned
I,
know
yesterday,
if
at
least
one
person
says
they
wanted
for
one
point,
though,
we're
gonna
take
the
conservative
approach
and
tag
it
as
such,
so
okay.
So
this
one's
not
one
point,
though
any
strong
objections
to
that,
because
I'd
rather
play
it
safe
and
force.
The
discussion
then
skip
it.
So,
okay,
we
got
those
there.
Those
two
support
work
chain
and
workflow
you're
on
I
think
this
was
by
you
the
other
day
as
well.
L
Yeah
I
think
we
had
a
lively
discussion
at
the
face-to-face
around
labels
and
workflows
and
and
all
that
and
I.
Don't
it's
not
necessarily
a
new
work
item
it
just
maybe
we
need
to
clarify
within
the
standard
what
happens
when
you
observe
the
router
implementation?
A
workflow
is
one
one
form
of
router
now
I'm
getting
an
event
and
I'm
passing
it
to
to
another
function,
and
maybe
daisy
chaining
several
functions
so,
for
example,
when
I'm
daisy
chaining
an
event,
you
know
from
one
function
to
another.
L
What
is
the
source
is
that
the
original
source,
or
am
I
creating
a
new
source,
or
am
I
embedding
my
source
in
the
original
source?
So
I
think
there
needs
to
be
some
guidance
into
how
we
create
workflows
because
I
think
cloud
events
to
be
a
an
awesome
tool
for
building
workflows,
because
essentially
you
could
start
labeling.
You
know
the
event
with
things
like
work,
workflow
identifier.
As
you
know,
it's.
L
E
Yes,
I
agree
with
that:
that's
it
that
is
a
interesting
limitation
of
the
current
spec.
Is
that
if
you,
if
you
receive
a
cloud
event
and
then
you
yourself
want
to
pass
it
on
to
another
cloud
event
handler,
then
how
do
you
decorate
that?
Because,
with
a
HTTP
proxy,
there
are
ways
of
doing
that?
For
instance,
I
know
that
we
may
be
familiar
with
the
X
forwarded
buying
things
like
that.
Yeah.
M
I
think
we
should.
We
should
figure
out
figure
out
some
scenarios
for
this
and
see
how
we're
going
to
solve
those
like
where
the
problems
where
the
problems
lie.
So
that
I
think
that's
something
that
we
should
go
and
address,
because
that's
that
might
have
breaking
I'm
worried
more
mostly
worried
about
any
breaking
change.
In
fact
that
we
might
have
like
like
real
hard
issues
and
that
may
be
it
candidates
for
one
okay,.
A
O
Have
a
comment
on
this
one
I
know
in
a
white
paper:
we
have
a
definition
for
workflow,
which
is
somebody's
workflow,
so
I
think
I
think
you
know
to
avoid
confusion.
I
think
this
is
more
like
a
chain
of
events,
not
the
real
surveys
workflow,
which
involves
you
know,
events
and
function
and
an
interaction
between
the
events
and
functions.
L
O
Think
this
is,
if
you
are
talking
to
this,
as
you
know
like
for
the
router
right
when
I
say
the
the
real
workflow
is
like
a
state
machine.
You
know
you
have
different
states
and
then
you
know
the
transition
between
the
states
involve
events,
and
then
you
know
the
action
could
be.
You
know
one
at
one
function
or
multiple
functions
executed
in
parallel
or
in
sequence,
or
could
be
branching.
I
think
that
concept
is
different
from
this
one.
I
just
don't
want
people
to
be
confused.
L
They
very
complement
each
other,
because,
even
if
you
look
at
the
step
function
from
from
Amazon,
then
you
have
the
ability
to
control
the
output
of
one
function
and
the
input
to
the
second
function.
So
this
is
in
the
context
of
sort
of
the
cloud
event.
There
is
an
addition
which
is
yours,
you're,
saying:
okay,
there's
a
state
machine.
Am
I
waiting
for
a
condition?
Am
I
doing
several
things
in
parallel,
etc?
That's
not
a
certain
part
of
the
cloud
event.
L
A
You're
on
I
think
Kathy
is
just
saying
that
we
may
need
to
pick
a
different
title,
because
the
wording
makes
it
sound
like
we're
going
to
be
defining
the
workflow
mechanism
as
opposed
to
just
how
do
we
possibly
represent
chaining
inside
of
an
event
or
you
know.
What
do
we
do
about
that?
I
think
it's
I
think
she
just
wanted
a
wording
change.
The
title
is
that
right,
Kathy,
that's.
E
A
F
Things
before
we
move
on
to
the
next
issue,
so
one
of
my
action
items
for
this
week
was
to
do
some
research
into
distribute,
tracing
or
open
tracing
and
whether
or
not
it
could
solve,
and
though
I
was
at
the
opinion
that
no,
it
is
not
the
penisy
we're
looking
for
there.
I
do
think.
That's
like
surveys
at
Keep,
Calm
and
any
vendor
who
did
any
sort
of
visualizations
eyes
lit
up
when
I
suggested
that
maybe
we
could
get
cloud
events
to
support
visualization
across
these.
D
A
F
M
F
There's
the
question
of
chaining
that
you
can
see
from
within
the
functions,
context
and
chaining
that
might
be
only
like
supportable,
a
lower
volume
for
an
actually
cooperator
to
see
an
official's
agent
or
know.
For
example,
if
you
use
some
type
of
tracing
framework,
you
might
need
an
eventually
consistent
database
that
has
a
very
low
trace,
raped
or
as
something
that
we
want
to
depend
on
is
an
annotation
that
happens
every
time,
so
they
are
kind
of
different.
So
what.
A
A
A
Well:
let's
let
it
that's
when
to
get
put
out
there
two
days
ago,
let's
wait
till
next
week
and
if
no
one
speaks
up,
then
we
can
do
an
auto
closed
or
closed
on
the
call.
But
let's
get
people
a
chance
to
to
review
your
thoughts.
That's
not
fair
sounds
great
and
the
meantime
will
tell
you
at
one
point:
oh
because
we
should
result
when,
where
the
other
before
that
and
the
objection
to
that,
alright,
next
alignment
without
ATF
security
event.
Token
we'll
think
about
this
one.
Is
this
something
we
need
to
discuss
before
1.0.
M
Should
we
should
I
haven't
looked
at
this
yet,
but
this
is
something
we
should
look
at
I'm
for
the
HTTP
web
book.
Spec
I
mentioned
that
on
the
slack
channel
yesterday
we
need
to
figure
out.
What's
the
right
model
is
for
the
abuse,
protection
feature
and
I
have
a
proposal
in
there,
but
I'm
happy
to
not
invent
anything
and
take
something
that
exists
so
I,
don't
know
what
that
Alliance
will
have
to
go
into.
Okay,.
A
C
M
As
far
as
routing
is
concerned,
there
are
400
different
ways
of
doing
these
things,
including
existing
crypto
frameworks,
which
can
go
and
take
the
cloud
event
that's
expressed
in
in
and
Jason
and
and
wrap
that
so
there's
like
Jason
web
web
encryption
and
Jason
web,
the
web
signature
and
all
those
things
and
then
there's
a
few
other.
So
what
I'm
saying
is
and
I
there's
a
quote:
I
want
to
quote
myself
just
scroll
a
little
bit
up
a
little
bit
more.
M
Buddies
so
I'm
I'm
worried
I'm,
really
worried
about
the
W
security
rat
hole,
and
this
can
easily
go
into
that
so
I
would
I
would
literally
want
to
go
and
scope
this
out
of
v1,
and
then
we
can
go
and
think
about
message
level.
Security
in
the
next
version,
but
I
would
rather
not
try
to
go
down
that
route,
because
that
gets
very
complicated,
very
fast.
Okay,.
B
Was
also
discussed
about
when
discussing
labels,
annotations
and
bag
of
property
properties
thing,
because
an
event
might
want
to
sign
itself
over
encrypt
itself
with
its
labels,
and
that
would
require
integrity
checking.
So
this
might
be
required
for
one,
because
we
do
want
labels
and
annotations
for
events
and
defense
might
want
to
be
sign
we're
encrypted
or
whatever.
This
was
brought
up
at
the
face
to
face
thing.
Yeah.
F
And
I
think
that
be
to
some
extent
it's
questionable
whether
this
is
even
an
on
goal
beyond
things
like
TLS,
because
as
I
noted
in
the
one
of
the
labels,
issues
or
pull
requests,
sometimes
it
got
a
little
creepy
and
it
might
actually
be
inspect
that
you
could
have
a
router
that
were
Dax
PII
or
something
like
that
and
I.
Think
that
is
an
open
question
or
valid
question
whether
or
not
not
editing
the
event
as
they
should
or
a
must
sort
of
behavior
and
like
only
a
must,
should
be
cryptographic,
decide.
Okay,.
A
A
Okay,
so
we're
gonna
have
the
discussion
later.
We
don't
need
ever
right
now,
but
it
sounds
like
everybody
is
okay,
but
please
keep
it
at
1.0
for
right
now.
So
next,
what
I'd
like
to
do
is
look
at
all
the
issues
that
I
tagged
is
not
required
for
one
point:
oh
yeah
now
Clemens
ears
are
on
here.
First,
the.
Why
reason
I
marked
those
is
not
required
to
because
I
didn't
think
it
was
a
hard
requirement
1.0,
even
though
we
may
get
to
it
beforehand.
G
M
A
A
F
A
The
reason
I
think
this
one
is
lingering
is
because
the
person
who
signed
up
for
it
has
been
distracted
on
other
things.
There's
someone
would
like
to
take
take
on
the
work.
I
think
that's
fine.
The
reason
I
marked
that
is
not
required
was
mainly
because
I
thought,
given
all
the
stuff
we
put
in
the
spec
about
what
it
is,
it
kind
of
implied
what
it's
not
can
particular
things
like
they,
it's
not
going
to
coud
that
the
desk
the
destination
address
in
there,
for
example,
things
like
that
yeah.
F
A
A
This
is
not
set
in
stone,
so
if
you
feel
like
things
are
mislabeled
to
speak
up
later,
is
that
okay,
all
right
moving
on
then,
let's
see
so
what
I'd
like
to
then
do
is,
as
new
issues
are
opened
up,
I
would
like
to
go
through
a
process
of
assuming
we
don't
get
a
boatload
of
issues,
every
call
maybe
take
the
first
minute
or
so
on.
Each
call
to
classify
each
issue
as
it
comes
in
whether
it's
1.0
or
or
not
required
to
one
point,
though,
hopefully,
I
won't
take
too
much
time.
A
We
get
a
lot,
then.
Maybe
we
only
do
a
couple
unless,
on
the
call
I
do
my
goal
is
to
make
sure
that
every
issue
is
tagged
appropriately,
so
that
would
help
or
that
helps
define
what
we're
looking
at
in
terms
of
goals
for
the
next
milestones
or
for
the
1.0
milestone.
Now,
along
those
lines,
we
have
I
believe
on
my
issues.
We
have
total.
A
We
have
41
issues
total
and
of
those
I
guess
if
we
have
what
six
or
seven
on
here
so
somewhere
in
the
mid
30
range
of
total
issues
for
1.0
I
would
like
it.
If
people
looked
at
the
list
and
volunteered
to
own
some
of
them,
that
does
not
mean
that
you're
responsible
responsible
for
necessarily
advocating
a
position
one
way
or
the
other,
it's
more
about
poking
on
people
to
come
to
a
resolution
and
help
drive
those
discussions,
because
without
owners
for
these
things
they
make
as
linger.
A
So
please
take
a
look
at
the
open
issues
and
volunteer
for
the
ones
that
you'd
like
to
at
least
help
drive.
I
may
take
a
few
minutes
on
each
call
to
pick
out
what
I
could
consider
the
heavy
hitters
or
the
the
more
important
ones
and
try
to
strong-arm
people
to
volunteer
to
take
them
just
so
we
get
some
forward
progress
on
them,
but
it
oh,
it's
always
best
to
volunteer.
If
you
can
I
appreciate
that,
so
I
don't
want
to
take
time
up
now.
A
Just
give
you
a
heads
up
that
I
may
start
nagging
people
to
volunteer
alright,
so
in
terms
of
next
milestones
and
I
did
want
to
spend
all
these
out,
calling
it
but
I
do
think
a
future
call.
We
should
start
talking
about
intermediary,
mind
some
milestones
before
1.0
and
what
on
that
list
of
issues
you
want
to
include.
So
please
be
thinking
about
that
and
if
you
have
any
suggestions
for
how
to
reduce
that
list
down
for
the
next
milestone
dates.
A
A
All
right
so
in
terms
of
when
1.0
is
ready.
In
my
mind,
this
is
what
I
sort
of
came
up
with
is
obviously
when
we
think
we're
all
done
with
breaking
changes
or
breaking
issues
when
those
are
resolved.
Obviously
typos
type
stuff.
You
know
we
donate
say
need
we're
not
too
much,
but
then
there
are
other
issues
that
came
up.
I
can't
remember,
someone
may
have
actually
open
up
an
issue
related
to
this
I
can't
remember
for
sure,
but
they
were
questioning
I've.
A
Actually
Thomas
Nathan,
you
you're
saying
you
know
how
we're
going
to
determine
criteria
41.0
in
terms
of
implications,
use
in
production
and
stuff
like
that.
So
I
include
that
in
the
list
here,
one
of
the
things
I
thought
might
be
interesting.
Is
we've
gone
a
certain
number
of
months
without
any
major
changes
to
the
spec,
but
I
wanted
to
sort
of
brainstorm
a
little
with
you
guys
in
terms
of
what
other
criteria
you
could
think
of
that
we
should
consider
for
when
we
feel
like
we're
getting
close
to.
A
One
point,
though,
and
I
did
put
a
proposal
here
for
what
I
thought
might
be
a
good
thing
to
shoot
for
which
was
three
uses
in
production
and
perhaps
three
months
of
a
time
when
the
spec
doesn't
actually
go
through
any
breaking
changes,
meaning
it
looks
fairly
stable
at
being
used
in
production.
But
I
wanted
to
open
this
up,
for,
as
I
said,
a
brainstorm.
Brainstorming
discussion
with
you
guys
see
what
you
guys
thought
and
how
you
like
to
address
this.
F
I,
because
I
like
being
able
to
trust
that
I,
can
rely
on
the
1.0
label.
It
really
sucks
when
you
have
2.0
come
out
right
after
at
one
point
now.
One
thing
I
thought
was
really
interesting.
Is
the
number
of
implementations
might
be
a
really
good
test,
just
to
make
sure
that
we
don't
have
some
blind
spot
and
they
use
days
for
not
handling
the
challenges.
You
can
measure
that
in
a
couple
of
different
ways,
you
can
say
that
we
have
two
implementations
due
to
you
know
Microsoft
and
circles
framework.
F
Supporting
this
though
I
don't
know
it
might
be
better
to
say
that
we
have
a
certain
number
of
event
providers
or
a
certain
number
of
customers
using
this
in
production
for
their
actual
application.
The
last
one
I
think
is
the
most
important,
but
maybe
the
most
business
sensitive
to
the
various
people
involved
all
right.
A
B
A
In
my
opinion,
I
think
it's
too
soon
to
tell
I'm
assuming
that
we're
gonna
follows.
You
know
the
semantic
versioning
rules
where,
as
long
as
we're
making
non-breaking
changes,
we're
not
going
to
go
up
to
2.0
and
everything's
backwards
compatible
and
whether
we
want
to
introduce
a
breaking
change
would
be
completely
up
to
the
group
to
decide.
A
A
F
E
This
is
Alex
again
what
why
is
there
such
a
focus
on
hearing
1.0,
which
is
actually
the
point
at
which
you
cannot
make
breaking
changes
and,
and
we
want
to
make
breaking
changes
on
their
very
low
cadence?
Do
we
not
need
a
lot
more
validation
from
people
using
this
and
adopting
it
before
we
go
to
1.0
is
what's
pushing
that
I.
E
A
I'm,
okay,
with
saying
this
is
a
premature
discussion.
I
just
I
think
maybe
Thomas's
issue
is
the
one
that
sparked
the
the
idea
for
having
a
discussion
on
today's
call
and
if
people
want
to
wait
and
see
how
things
go
before
we
even
have
the
discussion
further.
That's
fine
and
if
someone
was
to
ask
us,
for
example,
in
the
TOC
call
next
week,
you
know
what
is
our
criteria
for
1.0?
How
do
we
know
what
we're
there
I
could
I
could
dance
around
that
and
say
we
don't
have
a
firm
line
in
the
sand.
N
E
N
N
A
So
I
think
what
I'm
hearing
so
far
is.
This
may
be
a
little
premature,
but
it
may
be
a
good
thing
for
us
to
start
thinking
about
the
back
of
our
muffins
and
maybe
what
I
should
do
is
wait
a
couple
of
months
or
a
couple
more
milestones,
and
then
we
raise
the
question
to
see
if,
if
we
have
any
more
of
a
solid
answer,
well,.
A
N
So
don't
do
to
the
previous
discussion
about
0.2
3.4
that
was
in
the
milestones,
I
think
I.
Think
if
everyone
could
take
a
look
at
those
and
come
next
week.
Thinking
about
you
know
what
are
the
next
major
milestone?
Something
would
like
to
see
some
of
that
could
be
adoption
based
in
terms
of
providers,
consumers
open
source,
seven.
A
A
A
Okay,
next,
with
only
about
seven
minutes
left,
so
we
have
a
whole
bunch
of
topics
that
people
wanted
to
discuss
for
server,
additional
workflows
and
you're
on
I
believe
earlier
in
the
call
you
said,
we
don't
need
to
sterilize
everything
and
I
definitely
grew
at
that.
So
then
the
question
then
becomes
at
what
point
in
time
do
we
want
to
have
a
discussion
around
kicking
off
when
they
did
one
of
these
additional
work
streams,
because
I
do
think
there
will
be
a
little
bit
of
process
involved
there.
A
So,
for
example,
if
nothing
else
I
do
think
we
need
to
get
agreement
from
the
TOC
I'm
starting
another
work
stream.
The
same
way,
we
did
on
getting
agreement
to
start
cloud
events
at
all,
because
I,
don't
think
it'd
be
appropriate
for
us
to
do
something
out
of
out
of
socrata
phrase
charter
or
bounds
of
what
we
wrote
the
TOC
agreed
for
us
to
look
at.
A
So
how
do
you
guys
want
to
approach
this?
Do
you
want
to
wait
until
it
feels
like
the
cloud
event
thing
spec
is:
is
sort
of
calming
down
and
we're
not
going
through
so
much
churn.
Otherwise,
we
may
get
our
time
split
up
too
much
or
you
want
to
start
it.
You
know
basically
right
now
and
start
picking
out
the
next
work
item.
How
do
you
guys
want
to
approach
this
so.
L
A
Agree
if
I,
if
I
phrased
it
backwards,
I
apologize
I,
definitely
think
we
should
come
up
with
the
idea,
but
before
we
start
doing
it
like
actually
writing
down
a
spec
I
think
it
would
be
best
to
get
a
agreement
from
the
TOC,
because
the
TOC
comes
back
and
says:
no,
then
that
may
change
how
we
process
it
going
forward.
It.
L
L
O
I
would
like
to
propose
that
we
know
we
stopped
working
on
the
server
dysfunction.
Workflows,
which
is
you
know
about.
You
know
how
to
what
kind
of
language
primitives
we
can
use
to
define
the
to
specify
the
or
to
compose
the
function
workflow,
and
also
the
interaction
between
the
event
and
functions.
I'm,
seeing
you
know
some
companies
or
they
started
doing
this,
for
example,
Amazon
step
function
platform
is
doing
it.
Huawei
is
in
about
an
IBM
I.
Think
he's
doing
that
so
I
think
we
need
to.
O
A
O
I
support
that
I
think
you
know
when
we
discuss
the
location
things
or
what
information
we
should
add
to
the
events,
how
we
should
add
it.
That's
all
related
related
to
you
know
the
workflow,
how
we
should
do
that
so
I
think
you
know
working
on
the
workflow.
Well,
you
know
we
can
come
back
to
see
what's
missing.
In
the
event
current
events
back,
okay,.
A
A
A
So
then,
what
would
do
okay,
so
what
would
do
was
we'll
take
as
much
time
as
we
needed
on
next
week's
call
to
figure
out
which
of
this
list
or
others.
If
you'll
want
to
add
to
the
list,
we
want
to
tackle
next
as
a
work
stream.
Okay,
so
please
be
thinking
about
it.
If
you
have
additional
items,
you'd
like
to
add
to
the
list,
please
get
them
in
before
the
phone
call,
so
people
have
a
chance
to
think
about
it
in
advance.
A
A
So
if
you
want
to
add
things
to
this
list,
go
ahead
and
do
a
PR,
okay
and
I'll,
don't
you
don't
have
to
appear
for
anything
in
this
list
here,
I'll
make
sure
that
everything
is
list
is
in
the
documents,
so
you
can
think
of
other
things.
So
do
you
want
to
add
them
sure,
and
then
we
could
discuss
in
the
next
week.
Oh.
O
A
A
A
O
A
Just
an
indication
that
yeah
that
you've
actually
been
heard
on
the
call
we
keep
attendance
for
voting
rights.
I
guess
I
should
say
that,
for
those
of
you
are
new
to
the
call,
because
we
don't
have
formal
maintained
errs
like
a
in
a
traditional
code,
github
project.
What
we
do
is
if
we
ever
get
to
the
point
where
there's
a
contentious
decision
to
be
made-
and
we
just
have
to
take
a
vote
because
we
can't
decide
to
consensus
which
way
to
go.
A
We
actually
people
earn
voting
rights
by
attending
three
out
of
the
last
four
meetings
and
that's
why
I
make
sure
that
I
actually
hear
you
on
the
call
and
yeah?
And
you
didn't
just
add
your
name
without
actually
during
the
call
and
then
I
keep
track
of
everybody's
attendance
here
and
then
everybody
with
a
green
Fox
has
voting
rights
and
it
is
based
upon
company
and
that's
why
I
was
asking
for
the
company
affiliation
they're
really
going
to
call
that
help
Alex.
It.
E
E
A
E
A
Let's
talk
about
that
offline
I.
Think
company
may
need
to
decide
who
you're
representing
in
this
workgroup
and
that's
up
to
you
to
decide
it's
my
opinion.
So
what
okay.