►
From YouTube: Spec 3.0 meeting (November 9, 2022)
Description
A
B
Okay,
so
I'm
Pedro
I'm
new
in
the
team
well
I
was
collaborating
with
we're
testing
kpi
since
in
January
and
basically,
basically
in
the
CLI
repository
so
far
and
well
now,
I'm
I'm
during
the
last
Model
I
joined
the
postman
team
and
in
1st
of
October
so
well
when
I'm
doing
I
have
right
now
in
my
mentor
is
Jonas.
C
B
And
this
is
the
first
meeting
I
joined
about
the
spec
3.0,
so
well,
I've
been
I've
been
reading
them
the
list
of
well
the
GitHub
issue.
You
know
about
the
all
the
contents
and
well
I,
know
more
or
less
how
to
spec
but
I.
Don't
I,
don't
I
I'm,
really
curious
about
how
he's
running
well.
B
D
Yeah
I
just
wanted
to
say,
I'm
happy
that
the
conference
is
over
and
I
can
fully
focus
on
on
work.
That's
all
I
wanted
to
say,
and
it's
sad
that
it's
already
dark
and
we
have
the
same
hour
and
Pedro
has
lights
outside.
It's.
B
D
E
A
Here
all
right,
let's
get
right
into
it.
Ico
do
you
want
to
start
everything
off.
F
Yeah
sure
I
can
yeah
I'm
going
all
through
all
those
request,
supply
issues
and
from
join
nice.
We
are
complete
and
yeah.
Let's
have
a
look
at
us.
You
are
wanted
me
to
adopt
more
samples,
but
also
samples
are
actually
in
as
in
API
to
zero.
What
we
want.
D
Yeah
yeah
I.
Did
you
see
my
answer
I
provided
it
a
few
hours
ago.
D
So
yeah
yeah
yeah
when
I
wrote
my
comment.
I
didn't
realize
that
we
did
not
yet
migrate
the
examples,
so
it
makes
sense
and
I
actually
wrote
their
that,
like,
since
the
conference
is
over
I
plan
to
actually
take
your
take
your
proposal
and
already
what
what
is
merged
for
3-0
by
by
front
and
I,
want
to
take
this
adeo
audio
example
that
I
mentioned.
Also
in
my
comment
and.
C
F
D
But
I
also
want
to
do
a
request
reply
to
because
they
they
their
perfect
use
case
because
they
are
using
request.
Reply.
C
D
Can't
use
it
so
they
just
have
descriptions
yep,
so
yeah,
okay,.
F
Okay
jump,
oh,
let's
jump
over
to
the
next
issue,
and
then
we
will
see
it.
We
have
some
blocker
here
or
not
yet
decided.
Maybe
we
get
it
done
now.
We
have
a
screen
here.
I
want
to
share
the
screen
here
here.
I
have
had
some
discussions
with
throne
and
yeah,
but
what
is
it?
Yeah
not
takes
a
second
really
much
discussion
done
so
far
so
yeah.
F
The
one
here
is
about.
You
have
two
types
of
request
reply.
One
is
a
static
one,
that's
pretty
easy
case
here.
You
say
you
have
a
request,
Channel
and
a
reply
Channel.
But
from
time
to
time
you
have
things
like
inboxes
for
the
reply,
so
every
process
creates
an
inbox
and
say
hey.
This
is
my
request
and
the
response
goes.
Please
do
my
inbox,
so
the
address
of
the
inbox
need
to
be
part
of
the
message
in
the
dynamic
case,
and
this
is
where
all
problems
are
starting.
F
So
you
can
say
here:
let's
go
to
the
dynamic
where's,
the
dynamic
one.
Here
we
have
a
request,
Channel
and
reply
Channel
and
say
hey
on
the
reply.
Dynamic
I
have
a
message
for
this
yeah.
Okay,
if
you
say
okay,
my
replay
location
is
defined
in
a
message.
This
is
what
we
have
discussed
about,
because
in
a
message
we
also
already
defined
the
correlation
ID.
F
So
if
one
client
have
multiple
requests
at
the
same
time,
he
still
needs
correlation
IDs
to
see
what
is
the
answer
to
my
questions
to
relate
answer
and
question
together,
we
do
real
ID
and
your
Elite
as
well
and
reply
channel
to
say,
hey
here
my
message,
header,
you
find
the
address
of
my
inbox,
make
this
sense
for
everyone.
F
So
far,
and
now
we
want
to
say,
hey
already
originally
designed
it
having
the
reply
Channel
and
the
correlation
ID
defined
here
in
the
message,
then
formation
is
maybe
not
that
easy
to
understand,
maybe
not
a
straight
property.
We
could
have
it
as
well.
In
it
operation,
object,
say
hey.
This
is
my
replay
Channel
and
the
replay
address
can
be
found
here,
but
in
this
case
we'll
be
also
move
the
correlation
ID
to
have
everything
together
or
not.
F
It's
a
correlationary
part
of
the
message,
description
and
the
replay
location
part
of
the
operation
that
yeah
those
are
a
little
related
to
each
other,
I
think
and
yeah.
So
this
is
here
an
openclass
if
you
put
it
to
the
operation
and
option.
C
is
having
a
bug
per
message
here.
Here
is
having
everything
with
this:
okay
yeah.
You
say
this
option
here
you
can
have
multiple
replay
channels,
but
the
replay
Channel
at
rest
location
is
Define
per
operation,
so
it
have
to
be
for
all
possible
channels.
F
Here
have
to
be
part
of
the
header
in
this
case.
Well,
you
say
hey
same
here
on
level
operation
but
say:
I
have
a
message
called
request
and
I
can
have
multiple
messages
and
can
have
different
locations
for
the
replayed
Channel
for
different
messages
so
possible,
but
you
should
not
design
it
this
way,
but
it's
possible
another
option
not
written
down
here.
At
the
moment
you
could
have
it
on
level
channel
to
say:
hey,
I,
Define,
a
channel
the
channel
defines
a
channel
address
and
yeah.
F
If
you'd
you
have
a
channel,
you
have
the
channel,
address
and
say:
hey,
you
can
have
here
correlation
ID.
So
let's
do
it
this
way
here,
let's
copy
it
together
now
live.
F
To
say
here,
I
have
here
my
messages
in
my
channel
and
say
here:
I
have
here
my
address,
location
and
I.
Have
my
relation
location.
F
Maybe
say
this:
one
yeah
any
votes,
what
you
like
most
or.
A
I
have
one
comment
in
terms
of
the
correlation
ID,
because
we
have
already
a
keyword
for
specifically
correlation
ID
right,
yeah.
F
F
Are
also
used
for
fan
out
situations
where
you
receive
a
message
processes
and
send
out
your
results,
it's
for
a
message
or
pass
tracking.
F
C
E
Yeah
so
I
I
left
another
another
suggestion
on
on
the
comments.
Yeah.
G
E
Yeah
my
my
suggestion
there
is
that,
in
the
case
of
the
dynamic
ping
pong
here,
for
instance,
we
can
have
a
channel
which
is
a
channel
root
right.
In
this
case,
we
have
messages
and
we
say,
like
we
only
accepted
the
Ping
message
and
this
in
this
operation.
E
As
you
can
see
on
the
on
the
root,
Channel
definition,
there
is
well,
if
you
scroll
above
in
the
root
Channel
definition,
yeah.
F
E
There
can
be
two
messages
right
so
for
this
specific
operation
we
say
now
you
can
only
use
pink
here
in
this
in
this
operation
right
and
the
reply
in
the
reply
we
say
for
the
pink
message.
G
G
E
This
one
is
this:
one
is
the
static
one,
the
one
that
you're
highlighting
is
a
static
one
right,
yeah,
sure,
yeah,
so
and,
and
the
dynamic
says
location
instead
right
instead
of
having
a
dollar
ref,
you
have
a
location,
keyword
and-
and
that
is
like
yeah
like
this-
is
where
you
can
find
it
right.
Another
way
to
to
do.
F
E
F
This
is
why
I
originally
something.
G
E
F
But
in
this
case
I
would
guess
you
should
have
something
like
this
say:
hey
it's.
The
response
is
possible.
This
message.
What
here
is
not
working
as
well?
Here
you
say
the
response
for
the
Ping.
So
if
the
message
is
Ping,
the
response
is
Pawn.
Yes,
dollar
wrap
will
not
work.
You
can
have
multiple
response,
like
my
request
is
well.
E
F
E
F
So
this
mapping
here
would
not
work
so
when
you
say
here
pong
and
say
Okay
to
don't
work
like
this,
and
here
you
say
here,
the
location
is
at
this
point,
but
you
have
to
say
the
location
also
need
here,
because
the
request
message
contains
the
location
of
the
not
the
response.
So.
F
And
now
you
are
somewhere
on
my.
E
A
Confusing
I
think
what
makes
it
a
bit
confusing
is
not
only
the
syntax
but
the
use
case
around
it,
because
when
we
discuss
these
kind
of
use
cases,
all
of
them
like
these
four
specific
use
cases
right
and
sometimes
we
try,
we
kind
of
mixed
the
definitions
between
them
together
and
the
examples
together.
So
now
you
need
something
here,
but
here
you
need
this
and
it
to
me
it
gets
mingled
really
fast
and
it's
really
hard
to
follow
along.
F
E
D
E
G
D
So,
when
you
say
when
you
say
static
or
dynamic,
so
you
basically
mean
static
in
the
way
like
it's
with
the
with
Kraken
right,
so
it's
defined
well
what
messages
you
can
send?
What
do
you
get
in
response
when
you
say
Dynamic,
you
mean
this
case
with
Kafka
and
this
pattern
of
request
reply.
F
Yeah
true,
okay,
yeah
Dynamic
is
you
have
a
well-defined
addresses
aesthetic
is
you
have
redefine
addresses
like
your
web
circuit
example?
You
have
only
one
address
or
you
can
have
it
the
static
way,
all
as
well
as
with
Kafka
or
with
host
mqtt
broker
named
I,
don't
know
anymore,
but
yeah.
A
F
C
F
What
now
mean
you
have,
let
me
get
his.
F
F
R,
we
have
it
this
way
here,
let's
say
ping:
now
you
have
here
the
we
play
31st.
F
D
F
Location
and
something
else
yeah
location,
description,
but
yeah.
You
have
here
the
correlation
ID
and
you
have
here
the
reply
topic
or
something
like
this.
This
is
why
I
call
this
channel
originally.
F
G
F
Let's
see
if
I
find
it,
the
number
is
I'm
unable
to
find
it.
So
far,
but
yeah
you
have
the
channel
Indian
Channel.
You
have
the
location,
so
I
call
the
channel
address,
but
I
changed
this
so
often
now
I'm
not
really
a
member
anymore.
A
Should
oh,
this
is
just
a
suggestion,
an
overall
suggestion.
Actually
would
it
make
sense
if
we
took
the
smallest
use
case
first
merged
that
made
a
new
PR
for
the
next
use
case,
no
adapted
the
old
one,
we're.
A
D
Actually,
what
I
think
it's
that
actually
the
example
that
we're
looking
at
now
it's
it
was
easy
and
we
tried
to
complicate
it.
So
the
problem
is
that
the
request
reply,
as
you
show
here
it's
there
are
two
two
use
cases
static
and
dynamic
and
they
are
for
different
protocols
and
so-
and
we
now
just
basically
took
the
example
from
websocket
and
because
I
like
I'm,
saying
it,
because
I
haven't
seen
him
websocket
a
dynamic
approach
but
like
so.
D
D
But
we
don't
have
real
real
example
of
real
application
that
behaves
that
way.
That's
what
I
mean
so
in,
but
in.
D
F
D
F
F
D
Never
work
around.
We
have.
We
have
two
real
life
examples
that
are
in
2-0.
They
don't
have,
of
course,
request
replybook
because
that's
not
supported,
but
we
know
that
they
work
around
it
in
a
way.
So
we
can
model
them
with
3-0
true,
but
that's
so
just
basically,
what
I'm
saying
is
that,
like
I,
don't
think
it's
a
good
way
to
have
one
simple
example,
because.
D
Use
cases
so
I
would
say
like
if
the
best
is
actually
to
have
two
separate
examples.
Yeah.
F
Sure,
as
soon
as
you
have
specs
three
zeroes,
including
workarounds
or
description
Fields
or
whatever
I
can
model
it.
F
But
at
the
moment
this
is
what
I
see
the
only
possible
solution.
What
is
not
very
nice
here
is
that
you
have
the
channel
address
location
on
the
operation
base
and
the
correlation
ID
on
a
message
base,
but
I
don't
see
here
in
solution,
because
the
correlation
ID
is
a
multi-purpose,
and
this
here
is
a
single
purpose,
so
you
cannot
move
them
to
the
same
level.
F
C
F
C
F
F
E
I
mean
I
have
an
opinion,
but
foreign.
F
E
My
but
my
opinion
is
it's
really
not
it's
not
good
at
all
like
it's
just
a
I,
don't
have
a
solution.
That's
what
I
mean,
like
my
opinion,
is
that
if
we
move
it
to
the
message,
it
might
be
a
little
bit
confusing
that
you
have
to
put
these
specific
things
on
the
message
and
not
on
the
reply.
Part
of
it
of
the
operation,
but
but
I
I,
understand
that
if
we
put
it
on
the
reply,
part
of
the
actually
let
them
reply
on
the
messages
part
of
the
of
the
operation,
Dave.
F
E
No,
so
what
I'm
saying
is
that
I
understand
that
this
is
a
problem
now?
What
I'm
thinking
is
that
maybe
so
now
I'm
thinking
about
a
different
thing
as
well
like
what,
if
we
don't
have
reply
on
the
operation
at
all
right.
So
what,
if
reply
the
section?
The
reply
section
is
on
the
message
right
and
it's
not
in
the
operation.
F
F
E
Yeah,
my
point
is
that,
if
what
if
we,
for
instance,
on
the
message
level,
there
will
be
a
reply
like
we
have
now,
but
you
cannot
have
channel
right.
E
You
can
only
have
what
you
what
you
added,
actually
the
the
location
right
like.
Where
is
this?
What.
C
F
You
say
having
it
here,
let's
say
here:
we
have
a
reply.
Channel,
maybe.
E
And
location
will
point
to
a
a
to
a
dynamic
case
right
where
the.
E
F
Case
you
would
we
had
a
message
to
find
the
location
exactly.
F
E
Operations,
what
what
are
the
chances
that
you
want
to
reuse
the
same
message
in
in
in
in
different,
like
with
different
with
different
responses
right,
you
usually
Define
a
message
and
it's
always
going
to
be
use
the
same
way
right.
You
don't
use
it
for
other
things.
Right,
if
you
mean
reusing
payload,
for
instance,
payload
payload
can
be
a
schema.
A
reusable
schema
headers
can
be
a
reusable
schema,
but
this.
F
D
C
You
can
use
this.
Every
message
can.
D
Be
reused
in
different,
like
when
I
send
a
subscription
event.
I
can
also
screw
it
up
and
instead
of
subscription
status,
I
can
receive
an
error
messages
as
well.
So
you
have
use
case
where
you
have
multiple
messages
in
response,
and
you
have
use
case
where
you
actually
have
the
same
message
that
you'd
like
to
reuse
in
different
services.
D
E
G
F
Yeah,
this
is
a
possible
solution
to
say,
hey.
My
channel
is
reply
to
Ping
and
reply
to
Ping
contains
all
valid
messages
for
so
campaigns.
Message.
Pong.
F
But
then
I
have
to
give
the
the
ref
in
both
cases,
because
I
need
to
define
the
channel
anyway.
This
is
this:
defines
the
Channel
with
messages
with
the
message
body
schema
and
everything,
and
here
in
the
location,
so
in
the
location,
I,
just
Define
the
location
but
I
need
the
message.
What's.
E
C
G
E
D
G
F
F
F
F
Yeah,
let's
let
me
start
over
here:
yeah,
it's
Source
life.
Hacking
is
something
sometimes
a
little
bit
complicated,
but
it
works
so
I
have
here
now
ping
and
I
have
porn
and
I
have
a
big
fly,
address,
location.
G
F
I
have
the
correlation
ID
at
this
level,
so
I
can
have
to
look
damn
I
have
two
locations
together
here
see
here.
My
channel
is
ping.
My
message
is
ping.
My
replay
message
is
pong,
and
here
as
well,
I
say:
okay,
this
one
is
ping.
My
message
is
Ping
I
have
a
reply.
Channel
and
I
say
here's.
Those
are
my
messages
that
are
possible.
F
G
F
This
part
here
you're
getting
rid
of
this
part
here.
So
it's
easy,
but
the
best
from
here
is
you
have
really
the
address
and.
G
B
E
This
is
really
com.
What
does
it
mean
reply
address
there
like
it's?
It's
really.
It
doesn't
make
sense
to.
F
F
E
C
F
Yeah,
but
we
need
to
get
it
some
decision
so
otherwise.
C
E
F
D
E
Eventually,
we
have
to
eventually
yes
yeah,
we
are.
We
are
four
code
owners
on
the
spec
and
we're
only
two
of
us
here:
five
we're
still
missing,
Sergio
and
Vladimir.
Here,
sir
okay
yeah.
F
F
E
F
F
E
You
need
to
make
a
decision
soon,
I
mean
we
don't
have
to
take
it.
So
man,
that's
we
are,
we
are
we're.
We
are
like,
or
at
least
in
my
mind,
is
that
we
release
version
three
on
April,
so
we
still
have
time.
Don't
worry.
E
F
D
F
D
Suggestion
updates
pull
request,
so
we
have
but
I
also
suggest,
let's
clean
it
up,
because
there
are
some
unresolved
conversations
and
it's
hard
to
see.
G
D
One
we
follow
and
not
so
resolve
the
conversations
update.
What's
the
current
stage,
I
mean
the
current
proposals
and
I
my
commitment
that
I
will
take
a
year
example
and
for
and
we'll
rework
it
for
3-0,
with
all
the
proposals
yep.
C
F
Yeah
I
have
one
more
left
issue:
I,
don't
know
if
you
want
to
pick
up
another
thing
or
not
it's
about
the
message
thing
here.
So
if
we
want
to
limit
messages,
we
say
here
at
the
channel.
We
have
here
ping
and
root
and
the
link
to
the
message
object
and
now
we
are
going
to
operation
and
want
to
limit
ping.
So
I
want
my
ping.
E
I
will,
just
even
to
be
honest,
like
I,
will
just
make
it
an
array
of
of
you
know,
ping
and
and
names
the
names
of
the
mess.
The
idea
of
the
messages
in
the
channel
yeah.
C
F
E
E
A
map
I'm
saying
yeah
yeah.
You
can
then
do
that
what
I
mean
is
that
yeah
we
we
we
do,
leave
it
as
a
dollar
ref,
as
it
is
right
now
on
the
on
the
re
on
the
messages
inside
the
operation.
E
We
leave
the
dollar
rev
there,
because
that
that's
what
we
agreed
in
another
in
another
issue,
but
I
was
even
thinking
like
hey.
Should
we
make
an
exception
here,
because
it's
hard
to
read
like
it's
make
it's
actually
making
it
hard
to
read,
but
I,
don't
think
we
should,
because
otherwise
we
we
start
having
too
many.
We
have.
We
start
having
exceptions
right
and
dollar.
Refs
will
be
an.
E
E
F
E
G
E
E
So
I
will
limit
this
dollar
ref
to
0.2
the
channel,
not
to
the
to
the
not
to
the
components.
E
This,
so
that's
why
you
should
be
able
to
have
a
name
instead
of
because
otherwise
it's
an
ID,
it's
it's.
E
F
G
H
H
E
What
do
you
mean
by
registry
like
a
confluent
registry,
or
something
like
that?
Yeah.
E
That
so,
if
you
want
to,
if
you
want
to
make
it
to
make
it
like,
it's
used,
you
still
need
to
point
to
have
channels
in
your
file
that
will
reference
this
other
file.
Where
you
have
these
only
components,
you
know
what
I
mean.
E
E
The
idea
of
the
message,
for
instance,
in
this
case
like
I,
will
just
because
all
it
will
always
be
hashtag,
slash
channel
the
name
of
the
channel
messages
and
then
and
the
ID
of
the
message.
So
by
looking
at
this
string,
I
can
quickly
get
the
message.
The
the
message
ID
right
inside
the
channel
list.
H
F
E
C
E
Write
some
examples
which
actually
takes
me
to
another
question
here
is
like
then:
do
we
really
need
channels
under
components,
yeah.
F
E
E
F
F
Not
but
yeah
I
do
know
some
crazy
Architects,
so
yeah.
F
E
E
Can
also
have
multiple
you
can
have
like.
Instead
of
root,
you
can
have
the
pink
Channel
and
the
pong
Channel
and
both
will
have
the
address.
Slash
you
can
you
can
compartmentize
I,
don't
know
like
you
can
divide
you
can
categorize
by
by
message
right,
even
though
they're
the
same
channel,
they
understand,
they
have
the
same
topic
in
the
broker
or
in
the
OR
in
websocket.
E
E
F
A
F
Them
so
then,
at
this
point
here.
G
E
F
Would
leave
it
this
way
here
because
it's
the
most
clears
and
even
for
the
for
the
easy,
easy
case,
it's
pretty
easy
to
read.
You
have
a
channel
and
you
have
a
messages
to
the
channel
and
yeah
for
easy
cases.
It's
nice,
but
you
mentioned,
is
easier
to
read
for
complex
cases.
F
E
E
C
F
Yeah
somebody
designed
it
yeah
I
guess
not.
Oh
yeah,.
E
F
So
you
may
get
rid
of
those
messages
seeing
here.
Yes,
this
idea
to
get
rid
of
those
messages.
F
The
last
question:
so,
if
soon
as
this
is
decided,
I
guess
those
messages
thing
is
not
really
part
of
the
request.
Reply
discussion.
He
said
another.
F
E
F
So
I
would
be
fine.
Yeah
I
will
create
those
proposals.
So
then
we
have
one
proposal.
This
message
is
here
and
one
was
out
and
then
need
to
vote.
Somehow
I
need
to
be
one
be
chosen.
F
E
We
on
the
pull
request,
we
have
the
discussion
and
and
we,
if
we
don't,
if
we
can't
come
to
an
agreement
all
of
us,
then
we
just
ping
the
code
owners
to
vote
which
I
don't
think
ever
happened
like
we
always
try
to
come.
C
E
Yeah
but
with
Lucas
said,
like
I,
wouldn't
worry
about
that
to
be
honest,
like
we
just
need
to
explore
and
make
it
clear
the
appetites
and
downsides
of
all
of
them
right
and
and
yeah
I'm,
trying
to
make
it
clear
for
everyone
to
to
grab
their
hands
around
and-
and
so
we
can
have
a
better
discussion
right
because
right
now,
I
agree
like
the
current
pull
request,
I'm
familiar
with
it,
but
whoever
lands
on
the
current
pull
request.
It
will
be
a
mess
all
right.
E
E
You
can
probably
add
this.
These
examples
that
we're
working
on
here
as
different
files
and
then
we
can
command
on
those
files
and
then
those
files
will
be
removed
right,
but
it's
just
to
to
improve
the
conversation
on
the
pull
request,
but
that's
pull
request
level
right
because
on
comments,
it's
it's
harder
to.
You
know
like
replying
to
your
comment
like
yeah
I,
like
your
idea,
but
I
will
change
this
line
or
this
other
line,
it's
repeating
the
same
code
over
and
over
and
over.
A
E
Just
want
to
say
that
I
mean
thanks
for
the
patience
man
you're
doing
a
great
job
here,
exploring
lots
of
possibilities
there,
so
yeah
we're
gonna,
make
it
just.
Let's
make
sure
that
we
we're
all
in
the
same
page,
so
yeah.
G
D
Before
we
started
live
stream,
we
talked
about
like
how
late
is
this
meeting?
So
just
remember
hey
call
that
like
before
we
did
not
discuss
it
on
3.0
meeting,
but
we
had
dedicated
meeting
at
a
hour
that
is
best
for
all
us
all
of
us,
so
we
can
sync
on
slack
if
we
want
to
have
a
meeting
on
different
hour.
Oh,
but
everyone
agree
with,
we
can
easily
set
it
up,
as
as
we
did
before,
yeah.