►
From YouTube: CNCF Serverless WG Meeting - 2018-09-27
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,
3
after
when
I
go
and
get
started,
let's
see
in
terms
of
a
eyes
the
only
thing
they
want
to
point
out.
Actually
they're
mine
I'll
point
that
out
later,
if
you
have
an
AI,
please
get
to
it.
When
you
get
a
chance
there's
the
list
is
shrinking,
though.
So,
that's
all
goodness
about
the
face
to
face
in
Shanghai
we
do
have
a
list
of
riposte
topics
and
a
Google
Doc.
A
So
when
you
get
a
chance,
if
even
if
you're
not
going,
but
you
think
there
would
be
a
topic
that
isn't
listed
there,
you
think
might
be
a
good
something.
Chris
talked
about,
please
feel
free
to
add
it.
I
am
planning
on
setting
up
a
call
next
week
for
the
people,
who
will
be
there
to
discuss
options
for
what
we're
going
to
discuss
and
who
wants
to
talk
to
which
topic?
A
If
you
do
want
to
participate
in
that,
please
just
add
your
name
to
the
mailing
list
or
to
the
Google
Doc
and
include
your
email
in
case
I.
Don't
have
it
there's
one
person
in
there
like
I,
think
in
everywhere
their
email
address.
Unfortunately,
now
we
I'm
planning
on
doing
a
phone
call.
This
Friday
at
3
p.m.
based
upon
the
little
poll
that
I
sent
out
for
bringing
squirming
around
the
next
possible
interrupt
type
of
event.
Do
a
Google
Doc
here
with
a
list
of
ideas.
Nothing
is
set
in
stone
at
all.
A
The
new
logos
that
Austin
put
together
for
us
are
finally
available
in
the
CN
CF
shop,
and
so
they
are
finally
in
stock
and
I
was
given
a
coupon
code,
so
I
was
able
to
order
200
of
them
this
morning.
Hopefully
they
will
arrive
relatively
soon
and
I'll
be
able
to
bring
those
to
the
koukin
events
in
shanghai,
as
well
as
seattle
and
hand
those
out
to
not
just
other
people
but
to
you
guys
as
well.
A
So
you
guys
should
be
able
to
share
that
with
your
family
and
friends,
because
that's
always
exciting,
so
just
fine
I'm
sure
that
yep,
it's
the
latest
status
on
that
I.
Suppose
if
you
guys
have
a
conference
that
you'd
like
to
give
those
out
to
before
those
events
drop
me
a
note
that
I'll
do
my
best,
maybe
ship
some
off
to
you.
It's
up
to
you
or
you
can
wait
to
the
conference.
So
I
will
bring
the
route,
though
the
full
stack
available
to
those
conferences,
all
right.
A
A
All
right
now,
hang
anybody
that's
coming
before
word
then
I
don't
see
Austin
on
the
call,
so
I'll
try
to
update
wood.
As
far
as
I
know
what
happened
with
the
SDK
group
I
believe
we
had
a
phone
call
last
week
and
really
for
the
most
part
we
talked
about
the
github,
repos
or
maybe
I
was
acting
me
before
anyway.
I
think
from
our
last
Thursday
phone
call.
A
We
agreed
to
go
ahead
and
create
the
repos
one
per
language,
so
I
created
a
Python,
c-sharp,
Java
and
JavaScript
repo
VMware
wants
us
to
just
migrate
the
resisting
go
line
repo
over
so
they're
working
on
whatever
I
think
internal
approval
process.
They
have
to
make
that
happen,
so
hopefully
that
happened
fairly
soon,
but
then
we
should
have.
What
is
that
five
different
repos
I'll
set
up
I
need
to
be
given
the
names
or
I'm.
A
Sorry
they
get
up
IDs
of
people
who
would
like
to
be
made
admins
on
those,
because
that's
right
now,
I,
don't
think
I've
made
anybody
an
admin
I'm,
definitely
not
doing
all
the
work
on
all
five
repos.
So
I
need
you
to
send
me
a
note
or
ping
me
on
slack
with
the
github
IDs
of
who
you
want
have
admin
access
on
there.
A
At
some
point
we
should
probably
talk
about
the
governance
of
those
repos,
but
for
right
now
believe
it
a
little
bit
loose
just
to
get
the
things
bootstrapped
a
little,
but
anyway
we
should
part
about
it
at
some
point,
any
questions
about
SDK
work
or
anybody
who
was
on
the
sta
call.
Can
they
think
of
something
that
I
forgot
to
mention
all
right,
not
hearing
anything
moving
forward
Cathy?
Is
there
anything
you'd
like
to
mention
relative
to
the
work,
clue?
I'm,
sorry,
workflow,
subgroup.
A
Okay,
all
right,
in
that
case,
are
there
any
questions
for
Cathy
all
right,
not
hearing
any
moving
forward,
then
all
right
peer
reviews.
We
have
a
couple
on
here
which
should
hopefully
be
fairly
straightforward.
Let's
see
what
we
got
here.
First,
one
all
right,
so
this
one
I
wanted
to
just
this
is
just
a
minor
typo.
Everything
else
in
this
list
was
singular,
so
I
decided
make
admin
and
singular
as
well.
A
The
bulk
of
the
PR
is
I
wanted
to
update
our
owners
file
to
be
a
little
more
consistent
with
owners
file
that
I've
seen
in
other
projects.
In
particular.
This
just
basically
said
the
admins
and
I,
which
I
still
keep
so
I
still
have
the
admin
section,
but
I
wanted
to
do
was
add
an
approver
section
now,
even
though
we
don't
have
a
formal
list
of
approvers
per
the
normal
get
up
process.
A
What
I
did
do
is
how
to
come
in
here,
pointing
to
our
spreadsheet,
telling
people
to
look
at
the
voting
rights
column
in
there,
and
that
will
give
them
the
current
list
of,
in
essence,
approvers
meeting
people
who
have
voting
rights.
So
that's
the
closest
thing
we
have
to
that.
I.
Don't
think
this
is
a
change
in
our
process.
I
just
find
a
document
we're
actually
doing
yeah.
A
And
I
actually
I
did
that
this
morning
and
since
that
I
think
in
our
previous
phone
call,
we
talked
about
open
up
an
issue
to
make
sure
that
we
had
additional
documentation
to
cover
that
I
open
the
issue
this
morning.
I
have
not
had
a
chance
to
actually
execute
on
these
yet,
but
that
is
not
my
to-do
list.
So,
yes,
yep
all
right,
any
other
questions
or
comments
all
right.
Any
objection
to
adopt
them
as
per
request.
A
All
right,
then
hearing
any.
Thank
you
guys
all
right
next
one,
this
one's
also
mine
a
couple
of
things
here.
One
is
we're
not
really
a
new
effort
anymore,
so
I
just
wanted
to
clean
that
up
when
I
was
in
here,
making
the
other
real
changes,
which
is
we
are
I,
wanted
to
make
it
clear.
That's
I'm,
sorry
I
got
distracted.
A
There
I
want
to
make
it
clear
when
we
actually
came
a
sandbox
project
mainly
because-
and
this
is
a
strictly
selfish
reason
every
now
and
then
somebody
pings
me
about
our
status,
what
we're
doing
or
when
we
became
a
real
project
and
I
like
me,
have
the
hardest
time,
remembering
it
and
I
figure.
Other
people
may
have
made
me
the
information
about
when
we
became
a
sandbox
project.
A
So
I
want
to
include
a
link
to
the
Google
Doc
for
the
TOC
meeting,
in
which
we
they
actually
voted
to
make
us
a
real
sandbox
project
and
it
was
a
May
15th.
So
I
just
want
to
put
that
information
in
there.
That's
really
the
main
purpose
behind
this
PR,
so
that
other
people
can
reference
it
if
they
need
it.
E
You
only
yeah,
hey
Doug.
This
is
the
only
the
comment
in
it.
It's
kind
of
a
dumb
comment,
so
I
apologize
but
I
think
we're
gonna.
Try
to
change
sandbox
to
cloud
native
sandbox
will
kind
of
in
discussions
on
that
right
now,
but
since
you're
making
changes,
if
you
want
to
just
instead
of
same
sandbox,
a
cloud
native
sandbox
in
front
of
that,
and
that
should
cover
you
going
forward
with
the
CTF
sandbox.
Your
cloud
native
sandbox.
A
A
F
A
A
B
F
A
A
A
A
Okay,
here
we
go
one
of
the
little
bit
meteor.
So,
if
you
guys
remember
correctly,
we
wanted
to
define
a
bar
for
adding
new
extension
attributes
to
the
extension
document.
I
believe
last
time
we
talked
about
this
may
have
been
Thomas,
who
suggested
that
we
actually
have
in
essence,
sort
of
sponsors,
meaning
at
least
two
voting.
Members
of
the
group
say
that
they
are
willing
to
say.
Yes,
we
should
adopt
this
as
a
well-known
extension
that
doesn't
mean
that
they're
necessarily
going
to
it
implement
it
themselves.
A
Yeah,
these
have
at
least
two
voting
members
and
the.
If
the.
If
the
author
of
the
pull
request
is
a
voting
member,
they
are
allowed
to
be
one
of
the
two
as
well
that
we
wouldn't
have
to
worry
about
people
playing
funny
Dean's
getting
it
like
other
people
submitting
PRS.
You
know
on
behalf
of
other
people
just
so
they
can
vote
anyway.
I
think
it's
been
out
there
for
at
least
five
days
or
something
like
that.
Any
questions
on
this
or
comments
on.
F
A
That
was
easy.
Thank
you
guys,
all
right,
Christophe
and
now
I
know
Christophe.
There
may
be
some
open
comments
on
this
one,
so
we
may
not
be
able
to
prove
it
today,
but
I
did
want
to
get
people's
general
sense
about
the
direction,
because
I
think
it
was
a
very
interesting
approach.
I
just
wanted
to
bring
it
up
for
the
group.
You
want
to
talk
to
this
one.
C
Yeah
I
apologize
in
advance.
My
two-year-old
is
here
so
he
may
talk
in
between
if
it's
too
annoying
just
stop
me.
So
this
came
from
a
couple
of
problems
specifically
for
HTTP.
A
problem
is
that
names
or
headers
are
not
case-sensitive,
so
we
have
a
problem
if
you
want
to
convert
a
case-sensitive
attribute
name
to
her
to
an
HTTP
header,
so
we
already
had
a
caller
ID
for
your
naming
convention,
so
I,
basically
in
this
PR
I
sharpened
it.
So
maybe
you
can
scroll
down.
C
C
C
Okay
should
be
available
for
portion
eight
integration
of
common
languages
and
I
think
like
if
you
limited
to
alphanumeric
characters,
we
actually
get
the
closer
to
this
goal.
So
it's
really
well.
It
takes
away
a
lot
of
freedom
like
okay,
but
then
we
were
kind
of
sure
that
we
can
use
this
name
in
all
programming
languages
and
we
also
open
up
the
possibility
to
convert
between
cavalcades
in
a
case
or
for
HTTP
query
to
on
case
sensitive
ways,
so
that
that
one
I
did
then
in
the
upper
file.
C
C
That
one,
so
it
follow
this
follow
along
this
rule,
strictly
means
that,
for
example,
ID
would
be
separated
by
two
dashes,
which
is
also
not
how
you
would
do
it
in
the
a
cheapy
convention.
So
we
could
think
about
making
exceptions,
for
example,
ID
and
Joelle,
and
so
on,
but
I
think
we
can
discuss
this
yeah.
A
H
It
I
think
about
the
special
casing
is
we
would
have
to
make
that
list
huge
upfront
because
adding
to
it
later
would
break
version
changes
later
so
the
earlier
SDKs
that
don't
know
about
the
special
named
ones
wouldn't
know
how
to
deal
with
those
later.
So
if
we're
gonna
go
that
approach,
we
should
find
every
special
case
that
we
could
possibly
want
up
front
to
avoid
breaking
the
the
SDKs
later
just.
H
I
Problem
yet
again,
now
with
in
another
context,
with
trying
to
get
MVP
and
HTTP
aligns
we'll
have
the
same
problem
and
I
think
ultimately,
from
a
solo.
This
mapping,
from
with
injecting
the
dashes
and
pluses
I,
think
that's
gonna,
get
ugly
very
fast
in
practical
use,
because
it's
not
entirely
clear
what
that
means,
and
then
you
may
may
have
nice
mixed
use
cases
where
so
I
don't
want
to
make
it
too
complicated.
I
So
if
we
can
say
because
case
folding
and
then
what
is
a
character
and
are
we
constraining
ourselves
to
ASCII
and
if
we're
not,
then
your
utf-8
their
territory,
then
you're
in
case
folding
scenario.
So
I
think
if
we're
more
constrictive
about
what
we
allow
and
say
you
can
only
use
lowercase
and
you
can
only
use
effectively
ASCII
characters.
J
This
is
Tim
my
that
I
think.
That's
probably
a
good
idea
anytime
case
folding
can
possibly
be
avoided.
It
should
be.
It
tends
to
be
a
performance
bottleneck
in
some
languages
because
they
try
and
be
Unicode
complete
and
locale,
sensitive
and
so
on.
We
did
that
in
AWS
cloud
watch
events,
everything's
lowercase,
and
it's
we've
not
regretted
it.
C
J
A
A
L
A
Fair
enough,
at
least
you're,
honest.
Thank
you,
okay.
Okay,
let
me
pick
up
somebody
else,
Lynne
just
for
fun,
because
I
know
the
background
and
some
of
these
protocols,
like
this
I'm
at
Rakowski,
and
your
experience
with
some
of
your
other
inventing
type
of
protocols
and
stuff.
Do
you
have
an
opinion
on
this?
One
I
mean.
M
Specifications
side
and
impose
things
and
the
authors
of
these
events,
because
if
people
bought
into
the
spec
and
they're
gonna,
they
think
you're
gonna
author,
the
spec,
why
suffer
any
issues
in
terms
of
case
folding
or
having
to
add
things
to
your
pipeline
to
to
deal
with
keynote
answer
in
case
unicode,
all
the
things?
Why
inject
other
tools
in
the
pipeline
and
things
you
don't
need
to
if
people
behind
the
specter
and
code
do
it
with
the
more
prescriptive
things
to
make
processing
more
efficient.
N
A
O
A
I
C
I
The
so
the
C
II
is
the
SI
prefix
was
meant
to
be
basically
just
a
name
spacing
where
you
everything
is
prefixed
with
C
II.
You
can
go
and
take
the
strain
that
follows
C,
E
and
basically
stash
that
back
into
a
dictionary
and
then
and
then
clients
will
go
and
consume
attributes
from
the
dictionary
and
the
CEO
be
stripped.
So
that's
just
the
purely
just
a
wire
construct.
I
Guess
that
a
there's
a
case
where
we
actually
have
properties
which
differ
only
by
by
name
by
a
case,
and
then
you
can
also
need
to
go
and
and
find
out
the
casing
without
any
help,
just
from
that
arriving,
HTTP
header
so
again,
I'm
leaving
towards
towards
that
sounds
all
very
complicated
and
basically
speaks
for
us.
Our.
M
I
just
wanted
to,
did
you
got
me,
my
brain
cells
firing
book
if
I
acknowledge
that
I
think
that
this
discussion
is
indicative
of
the
maturity
spec
and
the
current
state
of
working
or
at
with
it
in
terms
of
implementation
and
which
means
that
people
are
looking
at
what
time
they
need
to
use?
What
utterance
sorry
geek
out
every
last
bit
of
performance,
how
to
avoid
confusion,
implementations
and-
and
the
next
logical
step,
then,
is
to
ask
yourself
since
this
discussion
is
about
you
know,
how
do
you
do
word
separation?
M
The
question
is:
why
do
we
need
birth
separation
body?
You
know,
if
you
look
at
all
the
examples
here
above.
Why
is
why
do
we
have
to
have
event
and
every
name
it
seems
redundant
by
context
unless
you
have
multiple
times
or
multiple
IDs,
we
need
to
be
the
differentiator
now
we're
gonna
trigger
or
some
other
indicator,
but
unless
it's
needed,
why
wouldn't
you
the
event
format
and
on
the
keys
as
much
as
possible,
so
I'll
just
throw
that
out
there?
Okay.
A
Thank
you,
so
I
have
my
hand
raised.
I
just
want
to
point
out
one
thing:
I
put
this
as
a
comment
and
one
of
the
other
issues
out
there
you
scroll
up
in
the
chat
you
can
see
where
I
put
in
see
my
property
I
just
want
to
feel
to
remember
that
HB
headers
can't
have
in
essence
multiple
values
in
there.
So
you
could
a
semicolon
and
then
put
you
know:
name
equals
the
real,
proper
name
in
the
proper
case
saying
with
dashes
or
whatever
you
want,
and
that
could
be
completely
independent
from.
A
However,
it's
specified
on
the
left
hand,
side
of
the
colon
as
HB
header,
so
we
do
have
options
like
that
as
well.
I'm,
not
I,
say
advocating
that
it's
why
to
make
sure
people
remember
that
we
had
that
kind
of
option
so
going
forward,
though
I
hear
at
least
two
options
here,
I
suspect
people
we're
going
to
want
to
think
about
both
of
them
going
forward.
What
I
would
like
to
suggest
in
terms
of
a
next
step
here
is
maybe
Clemens.
Could
you
open
up
a
poor
request
with
your
approach
of
just
lower
casing
everything?
A
A
A
Okay,
obviously,
if
someone
has
a
brilliant
idea
and
for
a
third
alternative,
don't
hesitate
to
open
up
another
poor
request
because
I
think
having
them
in
their
complete
form
as
full-blown
spec
edits,
makes
it
really
easy
to
be
able
to
compare
and
contrast
them
going
forward.
So
don't
don't
hesitate,
though,
create
another
one.
If
you
need
to.
C
All
right
one
point:
this
is
also
like
I
said
it
limits
it
to
alphanumeric
characters
and
it
means
you
cannot
use
utf-8
or
like,
like
Tim
said,
like
Turkish
names
wouldn't
be
possible,
so
I
would
just
like
to
go
out.
Is
that
cool
of
everybody?
Or
do
you
think
this
no
go,
because
that
would
also
change
with
the
direction
of
things.
I
Now
I
think
I
think
we
should.
We
should
really
go
and
constrain
constrain.
The
character
set
to
be
stay
within
ASCII,
because
you
know
everything
out,
so
you
have
two
choices
on
either
allowing
everything
else
in
utf-8
or
constraining
it
to
asking
it
seems
to
be
the
two
fair
choices
and
I
think
that's
something
that
we
can
reasonably
do
and
should
do.
J
A
A
A
A
A
A
C
Can
try
okay
yeah,
this
didn't
start
out
from
me,
but
I
think
like
from
yes
commercials.
The
company
I
work
for
this
is
like
one
thing.
We
generally
have
that
people
are
really
so
very
good
in
the
converse
paste
and
it's
for
a
lot
of
things,
important
to
understand
the
order
of
events
and
that
you
make
sure
that
you
process
them
in
the
right
order
yeah.
C
So
what
I
did
is
then
started.
It
I
think
and
then
I
added
some
stuff
to
it,
but
is
what
is
what
is
basically
does
is
two
things.
It
generally
says
there
is
an
attribute
called
sequence
and
there
is
some
value
in
it
and
we
don't
really
know
what
it
is.
But
if
you
know
how
to
interpret
it,
then
you
can
find
out
which
event
comes,
and
the
second
thing
it
does
is
it
defines
a
sequence
type.
C
And
if
that
is
known
to
you,
then
you
know
how
to
interpret
what
you
find
in
sequence
and
this
we
have
one
sequence:
I've
defined
called
integer
and
it's
basically
an
increasing
integer,
so
pretty
straightforward
starts
with
1
then
goes
to
2,
3,
4
and
so
on,
and
then
it
wraps
around
at
some
point
yep.
That's.
A
Now
keep
in
mind,
this
is
just
an
extension,
so
the
bar
is
lower.
Don't
they
have
to
be
perfect,
it
can
change
over
time.
The
question
here
for
the
group
is
whether
this
seems
appropriate.
You
know
thing
for
us
to
add.
As
an
extension,
is
it
in
scope
for
our
spec?
Does
it
seem
like
it's
at
least
well
thought-out,
not
necessarily
saying
you
agree
with
it
or
that
you
can
implement
it,
but
does
it
seem
like
it's
a
valid
thing
that
people
may
want
to
do
as
an
extension,
I.
H
Think
it
is,
and-
and
part
of
that
is
because
I
believe
somewhere
in
the
spec,
we
specifically
call
out
being
able
to
process
stuff
kind
of
in
an
order
and
that's
not
built
into
the
the
basics
back.
So
I
think
this
is
a
very
least
the
call
for
an
extension,
if
not
a
full-on
property,
because
I
can't
remember
exactly
where
it
is,
but
there
it's
called
out
as
a
like
a
use
case
somewhere
in
this
back
being
able
to
process
stuff
in
order.
A
N
N
There
was
an
argument
for
yo,
saying
it's
of
type
gapless
sequence
rather
than
integer
or
just
a
gapped
sequence
and
I'm
more
sort
of
coming
at
that
from
the
fact
that,
as
soon
as
you
say,
this
is
a
a
gapless
sequence,
it
means
your
publishers
become
stateful,
you
know
so
that
could
that
could
result
in
problems
in
some
situations,
but
that
was
more
of
a
I
think
a
semantic
argument,
anything
else.
Okay,.
A
It
sounds
like
also
the
kind
of
thing
that
might
be
worthy
of
a
follow-on
PR,
mainly
because
I
think
it
may
require
a
little
bit
of
back
and
forth
to
get
the
wording
and
the
needs
and
the
syntax
of
it.
Quite
you
know
a
little
bit
more
fleshed
out
right
and
so
I
think
maybe
have
a
follow-on.
Pr
might
be
more
appropriate
and
keep
in
mind.
These
are
just
extensions.
They
can
change
any
time,
they're
not
going
to
be
version
with
the
rest
of
the
specs.
O
O
A
Q
This
time,
I'm
not
sure
if
it's
a
strong
objection
but
I
think
we're
adding
redundant
data.
We're
generally,
the
transport
protocols
already
have
sequence
numbers
and
there's
a
large
complexity
where
this
all
only
works.
If
there's
one
producer
of
the
event,
because
now,
if
I
have
let's
say
five
producers
to
the
event
going
to
the
same
namespace,
which,
as
a
transport
could
be
a
topic
or
a
queue,
the
ordering
is
only
for
that
stream.
M
This
is
the
same
place
earlier,
I
hesitated
because
I
didn't
want
to
put
you
off
agenda,
but
you
know
we're
gonna
see
a
lot
of
these
things
coming
up
where
people
have
one
added
value
and
it
always
comes
in
a
pair.
You
have
to
give
it
a
type.
You
always
have
a
type
value,
so
you
know
one
thing
that
we
we
did
in
past
approaches
is
we
went
to
a
tagging
methodology.
We
have
it
just
an
area
where
you
just
put
tags
the
tags
themselves.
M
You
use
a
URI
and
your
I
by
means
of
having
a
protocol
and
domain
what
me--let
protocol
it
indicates.
The
class
and
the
value
itself
can
be
appended
to
the
path
or
other
other.
You
are
means
to
the
end
as
a
as
a
value
or
a
query
string
or
whatever.
So
if
you,
if
you
might
want
to
think
about,
if
you
have
a
lot
of
these
things,
come
up
having
a
single
line
way
of
doing
it,
instead
of
having
to
all
cases,
have
pairs
or
sometimes
tuples,.
A
N
A
A
This
is
strictly
me
this,
because
my
representives
in
the
group
I
tend
to
have
a
fairly
low
bar
for
extensions,
because
the
whole
point
of
them
is
to
see
whether
they're,
useful
or
not,
and
in
some
cases
it
may
be
duplicate
information
and
that's
part
of
the
things
that's
going
to
get
flushed
out
when
people
play
with
it
and
if
so
people
may
decide.
Ok,
we're
gonna,
kill
the
extension,
but
I
know
that
there
are
necessarily
none
to
say.
A
Oh
I'm,
sorry
notice,
they're,
all
protocols
have
some
sort
of
ordering
aspect
to
them
that
that
this
would
fit
nicely
into.
So
that's
why
my
bar
is
fairly
low
for
what
should
be
an
extension
I.
Look
at
this
as
more
of
a
playpen
type
area
and
that's
why
I'm
very
resistant
to
to
having
a
high
bar.
That's
my
take
on
it
anyway.
Anyway,
Jesse
I
think
your
hands
up
so.
G
You
know
that
you
you've
kind
of
gone
in
that
direction.
I
that's
what
I
was
gonna
say
earlier,
as
I
I
feel
like
it's
an
extension,
so
I'd
be
inclined
to
allow
something
that
isn't
explicitly
airtight,
but
this
this
sort
of
thing
around
kind
of
a
sequential
protocol,
does
open
up
a
few
things
that
I
wonder
about
uhm
one
I
believe
it
was
Matt
that
brought
it
up
where
or
actually
meant,
replied
to
someone
who
brought
it
up.
It
might
have
been
Jem
just
recently
brought
up.
G
The
idea
of
you
know:
multiple
producers,
so
multiple
producers,
multiple
consumers
and
possibly
something
where
the
sequence
matters
from
the
beginning
to
an
end
and
there's
n
number
of
steps.
So
that's
that's
not
really
something
that's
informed
here
and
also
I
think
this
sort
of
opens
a
it
opens
a
door
to
advising
in
the
case
of
you
know
what
happens
if
if
you
receive
things
out
of
order-
and
that's
equate-
you
know
one
one
or
more
members
of
the
sequence
are
missing.
G
What's
you
know
sort
of
that's,
that's,
obviously,
something
that
is
to
find
out
size.
The
scope
of
this
back
in
this
extension
but
I
think
calling
that
out
sort
of
goes
down
the
road
a
little
bit
so
I
think
yeah
I
mean
I'm,
not
I.
Don't
have
any
specific
advice
around
this
and
I
wouldn't
block
the
acceptance
of
this
particular
PR,
but
I
think
I
just
wanted
to
voice
that.
That's
where
my
head
starts.
Going:
okay,.
G
A
N
N
Maybe
that
could
be
addressed
by
saying
that
sequences
are
within
a
type
and
a
source
or
something
like
that,
because
I
would
assume
you
know
you
would
generally
have
one
event
source
given
one
publisher
per
source.
So
maybe
that
is
a
clarifying
point,
but
other
than
that
yeah
I.
It
was
just
an
opinion.
I'm
not
dead,
set
on
anything.
Okay.
A
So
there
are
a
couple
of
comments
made
about
potential
changes
or
follow-on
pieces
of
work.
That
could
happen
here
of
those
ones
that
were
mentioned
are
any
of
those
things
that
people
feel
strongly
enough.
That
they'd
want
to
block
the
current
PR
from
going
in,
as
opposed
to
doing
it
and
follow-on
PRS
cuz
I'm,
trying
to
figure
out
whether
we
should
move
forward
with,
in
essence,
doing
a
vote
now
or
does
anybody
feel
strongly
that
their
suggestion
should
go
in
before
we
even
adopt
this
one.
A
A
Alright,
thank
you
guys
that
was
cool.
I
think
this
process
will
get
easier
as
we
start,
adding
more
and
more
alright
next
issue,
this
one
I'm,
not
sure
we
only
resolved
this
today,
because
people
may
need
more
time
to
look
at
it,
but
the
open
messaging,
spec
or
protocol
I
transported
Clery
call.
It
has
been
out
there
now.
A
There
was
some
questions
about
whether
this
spec
met
the
minimum
bar.
So
without
talking
for
the
moment
about
the
content
of
the
spec
itself,
I
was
wondering
people
have
had
a
chance
to
look
at
this
particular
comments,
which
I
think
was
edited
last
night
about
why
they
think
it
meets
the
minimum
bar
I'll.
Give
you
guys
a
minute
just
to
read
this
in
case
you
haven't,
read
it
yet.
A
Okay,
so
I'm
not
necessarily
going
to
ask
everybody
to
definitively
say
for
sure
whether
they
think
is
meth
bar
or
not,
because
I
think
they
may
need
more
time
to
to
think
about
it.
However,
I
ain't
gonna
ask
the
question
of
its:
does
somebody
have
an
initial
reaction
and
Ike
or
a
gut
feeling
as
to
whether
this
vacation,
based
upon
the
information,
this
comment
is
gonna
meet
the
minimum
bar.
I
That
depends
that
depends
on
what's
actually
so
if
there
has
been
changed
to
the
submission,
because
my
understanding
is
that
open
messaging
is
three
things:
it's
an
abstraction
and
then
it
also
has
a
wire
format,
and
then
it's
a
benchmarking
collaboration,
project
and
I.
Think
the
only
thing
that
really
matters
to
us
here
since
we're
talking
about
intro
will
be
a
specific
eight.
I
I
Organizations
so
that
clears
that
bar,
but
I
would
say,
if
you
look
at
at
the
contributors
list,
for
both
projects
for
rocket
and
queue,
as
well
as
for
open
messaging
you'll,
find
that
the
community
is
rather
in
lateral
on
both
accepted
open
messaging
in
the
benchmarking
project,
where
there's
a
lot
of
activity,
but
that's
not
related
to
this
submission
I
would
I
would
so
personally.
I
would
not
be
a
big
fan
of
of
this
finding,
but
if
it's
a
wire
specification
that
I'm,
probably
cool
with
it,
okay.
J
A
I
Spec
is
really
abstract
and
really
what
it
is
about
is
that
this,
it's
the
it's
the
rocket
MQ
wire
format,
which
is,
interestingly,
so
the
rocket
MQ
wire
format.
Apparently
that's
my
understanding
of
it
has
been
elevated
into
the
open
messaging
specification
without
the
actual
specific,
a
wire
specification,
even
making
it
yet
on
the
open
messaging
side,
which
is
the
thing
that
kind
of
rubs
me
in
the
wrong
way
and
is
a
little
weird.
Is
it's
it's
patchy
rocket
MQ
something
has
been
put
into
Apache
by
Alibaba.
I
N
The
when
I
learned
to
open
messaging
a
while
ago,
I
haven't
looked
I,
have
to
say
for
quite
a
while.
My
understanding
was
more
of
a
programming
model,
so
that
would
to
me
that's
akin
to
somebody
wanting
to
do
a
mapping
document
from
cloud
events
to
JMS.
It
was
more
of
that
sort
of
level.
I
I
never
took
it
as
a
wire
level
thing.
I
took
it
as
a
abstraction
of
a
messaging
infrastructure.
K
This
is
a
timing
nitpick,
but
in
the
comment
there
are
no
links
to
you.
The
claims
I
again.
This
is
a
myth
big,
but
if
Doug,
if
you
could
move
the
comment
there
saying
that
several
companies,
the
voice
I,
would
really
really
like
to
see
links
for
this-
and
this
is
not
just
for
this
Pierre
birth,
but
for
most
peers
who
claim
such
stuff
are.
K
Nothing
I
didn't
believe
the
comment
but
I'm
seeing
their
links
for
open
messaging,
the
Pesci
rocketing,
cue
and
a
vegetable
sir,
but
no
links
for
claims
that
companies
would
support
or
they're
getting
involved
or
have
another,
but
its
participation.
If
we're.
If
we
have
a
board
that
bar
should
at
least
come
with
not
just
claiming
they
have
participated,
they
definitely
if
they
renounce
their
participation,
they
definitely
have
a
blog
post
or
a
tweet
or
something
that's
official
and
that's
good
building
can
I.
Q
O
I
was
just
looking
at
their
github.
Repo
seems
to
have
a
have
under
specification,
or
it's
a
little
bit
spread
out,
but
you
can
get
in
to
see
what
they're
expecting
in
message
headers
and
their
subscriptions
about
types
expectations.
So
it
is
a
bit
more
prescriptive.
I
think
it's
it's
less
flexible.
It
was
like
there's
concerns
that
seem
to
be
outside
of
the
messaging
space
to
me.
There's
an
information
there,
people
on
a
reference
that
okay.
A
So
I
all
the
points,
everybody
I
think
the
speaker
key
is
empty.
Now,
so
all
the
points
you
guys
brought
up
to
you,
like
very
valid
things,
can
I
ask
you
all
you
guys
who
spoke
up
to
please
add
comments
to
the
pull
request
itself,
in
particular,
if
you
needed
additional
information
like
glad,
you
asked
for
pointers
and
stuff,
like
that.
Please
put
those
as
comments
into
the
pro
requests
so
that
the
author
can
respond
to
those
and
then
please
take
a
look
at
the
PR
itself.
A
In
particular,
the
the
the
actual
file
changes
themselves
to
see
whether
that
whether
it
meets
your
guys
needs
whether
you
need
more
information,
whether
it's
too
abstract
and
whatever
comments
you
feel
appropriate,
because
this
PR
has
actually
been
out
there
for
quite
a
while
and
they've
been
kind
of
blocked
by
that,
a
bar
that
we
defined
and
so
I
feel
kind
of
bad.
It's
taken
this
long
for
us
to
give
them
a
definitive
answer.
A
So,
if
possible,
I'd
like
the
on
next
week's
call
come
back
with
a
definite
yes
or
no
to
them,
and
that's
string
them
along
any
further.
So
please
put
your
comments
out
there
sooner
rather
than
later,
so
they
have
a
chance
to
respond
to
them
and
with
that
I
think
pretty
much
out
of
time,
but
there's
one
thing
that
I
forgot
to
do
that.
If
I
need
to
go
back
and
do
on
the
previous
PR
about
the
sequence
attribute
now
I
know
Christophe,
since
it's
basically
you
know
they
helped
write.
A
The
PR
and
I
know
he's
expressed
interest
in
supporting
it.
I'm
sorry
he's
claiming
he's
one
of
the
voters
who
says
yes,
I
want
to
you
know:
I
want
this
thing
in
there,
because
we
have
a
minimum
bar
of
two
voters
voting
entities,
approving
extensions
who
would
like
to
be
the
second
voting
entity
approving
that
extension.
A
Who's
that
opposes
Jesse.
Thank
you.
I
was
gonna
pick
on
Vlad
since
he's
president
just
before,
but
I'm
glad
someone
spoke
up
voluntarily.
Thank
you
all
right,
cool
with
that
I
believe
at
the
end
of
the
agenda,
or
at
least
for
the
time
for
today.
Are
there
any
last-minute
questions
or
comments
before
I
go
back
and
do
a
final
roll
call
all
right,
not
hearing
any.
Let's
go
back
up
here,
I
think
I
heard
most
people
mad
I
heard
Rohit.
A
R
A
Attendees
I
appreciate
that.
Thank
you,
you're
welcome!
It's!
Alright!
Anybody
else,
all
right
in
that
case,
since
we
have
a
whole
two
minutes
last
chance
any
other
topic.
We
wonder
when
it's
the
bring
I'm
really
really
quick,
all
right,
and
that
case
we're
done.
Thank
you
guys
very
much
you
better
through
quite
a
few
today
I
appreciate.