►
From YouTube: Spec 3.0 Meeting
Description
C
A
Whatever
we're
not
wait,
this
is
my
profile,
not
hdb
profile.
Yes,
on
twitter
is
correct.
A
So
my
name
is
general
genes
lagoni.
A
I
work
for
google
services,
whatever
that
means-
and
I
hate
jesse
manning
and
I'm
saying
this
because
just
meaning
he's
about
to
join
the
call,
and
so
afterwards,
when
he
see
the
recording,
he
will
see
this.
A
E
C
C
But
it's
not
that
it's
it's
not
that
straightforward
to
do,
which
is
why
we
are
gonna,
discuss
it
a
little
bit.
From
my
perspective,
I
wanted
to
be
moved
because
everything
json
schema
related
should,
in
my
opinion,
be
in
one
repository
instead
of
multiple,
because
I
don't
see
the
bindings.
C
As
being
I
see
the
bindings
as
being
part
of
the
specification,
there's
no
difference
between
bindings
and
specification,
except
bindings
is
for
protocol
specific
things,
and
that
means
that,
ultimately,
I
want
my
spec
document
to
be
validated
against
these
bindings
as
well,
at
the
same
time
that
I'm
validating
the
the
main
json
or
the
async
api
document,
because
for
me,
there's
no
difference
between
them
and
it
makes
it
a
lot
more
easy.
C
E
E
B
B
B
Only
you
know
the
the
metadata
fields
and
this
polluters
in
exactly
unify
that
metadata
for
all
four
core
objects,
message,
operation
channels
and
servers,
so
yeah
yeah.
E
E
Those
would
be
allowed
within
the
within
the
info
or
sorry.
The
metadata
object.
E
Know,
but
this
would
not
you
know
at
some
point
lucas
and
I
had
talked
about
there.
There
was
going
to
be
an
extension
registry.
These
would
this
would
still
kind
of
be
just
ad
hoc
extensions
applied
to
that's
right.
Okay,.
B
B
Yeah
yeah,
but
we
have
also
options
to
define,
not
defined
fields
in
the
metadata
object.
I
also
describe
it
in
this
blue
request
and
we
should
probably
discuss
it
because
the
problem
is.
B
B
What
like.
B
So,
for
example,
if
you
want
to
add
something
like
the
environment
to
the
server,
you
exactly
write,
metadata.server
and
something
yes,
some
value,
and
by
this
you
don't
need
to
define
it
by
the
extension
x,
environment
yeah.
But
the
problem
is
how
to
validate
that
and
maybe
how
to
make
some
registry
for
the
metadata.
C
B
C
The
extensions
you
recall
the
arguments
for
why
you
wanted
why
you
wanted
to
not
have
extensions
and
use.
B
B
Yeah,
but
it
was
idea
from
lukas
because
we
copied
that
idea
to
introduce
the
metadata
object
from
the
kubernetes
and
in
the
kubernetes
world.
If
you
have
the
resource,
something
like
the
deployment.
Yes,
you
describe
the
deployment,
you
can
also
add
the
metadata
and
you
can
add
every
everything
as
you
want
exactly
the
metadata
labels
here.
A
So
the
idea,
if
I
recall
correctly,
is
that
we
go
with
this,
so
these
fields
are
like.
We
have
a
few
fields
that
are
defined
yeah.
A
Description
summary
tags
and
external
dots,
but
then
the
rest
will
be
defined
as
per
convention,
let's
say
so
exactly
convention
over
time
there
will
be,
I
don't
know
like
environment
in
some
objects
like
in
some
metadata
objects,
for
instance,
inside
inside
server
metadata,
you
can
put
there
like
environment
and
then
environment.
A
If,
if
it
happens
frequently
enough,
then
convention
will
know
that
that
this,
what
this
means,
what
what
environment
means?
It's
not
on
the
spec,
it
will
be
only
it
will
be
purely
a
tooling
thing.
B
A
A
If
also
open
that
box,
then
like
anybody,
can
put
whatever
they
want
there
and
it
will
have
meaning,
then,
even
if
it's
from
the
tool
inside
then
it
will
become
a
mess
like
or
it
could
potentially
become
a
mess.
That
say,
for
instance,
that
I
don't
know,
I
don't
have
an
example,
but
imagine
that
some
people
put
environment
as.
B
A
A
A
A
B
A
I
are
sorry
and
then
it
could
stop
being
as
an
extension
right,
okay,
but
but
yeah
or
not
not
stop,
because
you
will
not
be
able
to
stop
it,
but
but
yeah.
B
C
Make
sense
I
had
one
we
talked
about
whether
you
are
adding
extra
fields
or
it
should
be
extended.
A
B
Yeah
or
on
the
plane
that
yeah
that
pluricas
introduced
the
extra
field,
if
you
go
to
the
files
change
it
exactly
to
the
changes
specification
in
this
pulley
request.
No,
no,
the
file
changes.
B
Yeah
for
servers
channels,
operations
and
messages
yeah,
so
you
can
look
at
it
in
some
way.
It's
a
good
idea.
Maybe,
but
if
you
look,
for
example,
for
examples
for
channels
where
exactly
you
have
channel
operation
and
message,
it
can
be
redundant
that
on
every
object
you
have
to
define.
But
of
course
the
metadata
will
be
the
optional
filters,
but
you
have
to
define
the
metadata.
B
B
You
know
not
by
the
extension
but
by
composition,
yes
to
the
server
operation,
channel
and
message,
so
it
means
that
this
course
object
has
that
filter
metadata
object.
Yes,
but
then
this
is
the
problem.
If
you
introduce
how
to
you
know
handle
the
extension
only
for
the
metadata.
Yes,
as
we
discussed
so
the
x
environment
should
be
inside
the
metadata,
it
will
be
better
place
or
better
will
be
to
add
x,
environment
to
the
server
not
to
the
server
metadata
yeah.
A
E
Yeah,
I
think,
organizationally,
it's
really
it's
very
appealing
and
having
a
reusable
object
such
that
if
we
come
along
like
later
and
say
like
oh
well,
it's
really
useful
to
have
a
version
here.
Oh
it's
really
used
to
have
useful
to
have
a
display
name.
We
don't
have
to
go
in
and
change
15
different
places.
We
add
it
to
the
single
object
and
that
gets
applied
to
every
every
level.
B
Yeah
exactly
it
can
be
the
the
the
the
pro
prongs
yes
or
cons,
because
my
english
is
not
very
well
yeah
exactly.
It
can
be
good
options
because,
for
example,
for
a
message
or
operation
you
have
the
trades
so
for
you
can
reuse
some
descriptions,
some
name
yeah,
but
at
the
moment
you
don't
have
the
trades
for
servers
on
channels
and
by
the
metadata
object.
You
can
reuse
some
some
title
yeah
for
the
given
objects.
Yeah
exactly
so.
Thank
you
very
much
for
that.
E
Yeah
I
mean
right
now
we're
you
know,
we
have
event
portal
and
so
we're
exporting
out
in
each
events
and
topics
which
we
equate
to
channels
and
schemas.
E
They
all
have
their
own
independent
life
cycles
and
versions
and
display
names
and
sort
of
they
can
be
semantically
versioned
or
they
can
be
sort
of
more
intuitively
versioned,
like
you
know,
instead
of
10.1.3,
we
call
it
big,
sur
or
whatever,
like
apple
does,
and
that
applies
at
every
level
of
the
hierarchy
and
so
having
something
that
was,
you
know
a
metadata
field,
that's
very
clearly
separated
and
allows
us
to
do
that.
I
think,
would
be
really
really
appealing.
B
So
yeah
some
some
group
of
people
probably
will
say
that
okay,
the
metadata
object
is
good
like
for
you,
you
see,
and
for
your
companion
but
yeah.
Some
people
can
also
say
that
why
I
need
to
define
us
the
metadata
so
yeah
I
mean
I
am
with
you
mostly
because
I
I
really
like
the
structure
that
exactly
I
I
have
some
nested
fields
to
define
some.
B
A
B
A
Sorry,
I
I
got
the
connection
issue
and
I
couldn't
hear
what
you
were
saying
so
my
mic.
I
was
about
to
to
ask
a
question:
it's
like.
Do
you
have
a
specific
example
where
we
will
want
to
use
to
reuse
this
metadata
definition,
like
you
know,
like
the
title,
the
name,
the
description
to
me
that
those
sound
like
a
specific,
like
very
specific
thing,
that
I
will
not
want
to
reuse
them.
Maybe
one
part
of
the
metadata
object.
I
will
but
the
whole
metadata
object
in
which
cases
would
you
want
to
reuse
them?
E
Yeah,
I
I
don't
know
where
we
got
cut
off
fran,
but,
like
you
know,
we
the
solas
event
portal.
You
know,
there's
independent
life
cycles
for
events,
and
you
know
that's
essentially
at
the
at
the
channel
level
for
us
and
then
the
application
which
is
sort
of
at
the
root
level,
and
then
the
schema
each
one
of
those
things
has
their
own
independent
display
name,
their
own
independent
version,
their
own
independent
semantic
version,
their
own,
the
independent,
which
would
be
an
extension
for
us.
It's
right
yeah.
E
I
think
you
know
internal
id
so
that
we
can
correlate
things
back
in
so
that
you
know
each
is
the
way
we're
moving
with
async
api
is
each
of
those
objects
are
independent,
except
for
for
reuse
and
so
to
have.
The
ability
to
attach
metadata
at
every
level
seems
really
really
valuable,
really
really
valuable.
E
A
Maybe
what
I'm
hearing
here
is
it's
a
different
argument.
So
maybe
wouldn't
it
be
better
that
we
organize
extensions
under
an
extensions
key?
Maybe
and
then
all
the
extensions
are
under
a
specific
object
right
so,
like
the
same,
the
same
thing
that
we're
doing
with
metadata,
but
imagine
something
with
with
extensions.
E
But
I
don't
know
if
that
addresses
it,
though,
because
it's
a
mixture,
you
know,
there's
some
use
case
specific
stuff
that
we
have
to
use
extensions
for,
but
I
also
think
there's
a
large
shared,
pretty
constant
group
of
metadata
that
would
widely
widely
be
applicable
to
across
industries
and
use
cases
and
also
objects
in
the
hierarchy
of
async
api.
But.
A
It's
not,
but
it's
not
here,
like
the
all
the
the
the
only
fields
that
we
are
considering
here
is
id
name,
title
summary
description,
tags
and
external
docs.
So
that's
the
only
set
of
fields
that
you
will
be
able
to
reuse.
E
A
E
E
A
That
source
code
yeah,
so
that
so
then
so
that
my
proposal
last
time
we
chat
about
it
was
like
that's
composition,
that's
fine!
We
can.
We
can
treat
it
as
composition,
but
you
can
use
the
same
object.
Let's
say
the
same
definition
in
in
code
or
in
json
as
as
inheritance.
So
you
can.
You
can
inherit
this
list
of
fields
and
it
it
can
be
still
structured
to
some
other
object,
but
instead
of
doing
composition
under
metadata
field,
you
can
say
this.
One
is
inheriting
this
other
one.
So
you
automatically
inherit
these
fields.
A
A
A
So
name
title
whatever
all
this
stuff
from
an
implementation
point
of
view,
but
from
a
user
point
of
view
I
mean
it's
to
me:
it's
providing
no,
no
value,
it's
just
putting
everything
under
metadata
which
actually
to
me
to
me
and
to
my,
but
that's
a
just
an
opinion.
It's
even
worse.
It's
one
more
level
of
implementation
and-
and
if
you
think
about
it,
like
everything,
that's
defined
there
on
the
object.
It's
meta
data
everything
everything,
because
the
whole
spec
is
metadata.
A
E
That's
deep
man,
I
I
hear
what
you're
saying
I
guess
I
would
defer
to
magic
on
the
outside
of
it
like
it
seems
like
having
one
an
additional
indent
actually
makes
it
easier
to
read
and
easier
to
implement,
but
magic
is
much
closer
to
implementation
than
I
am
day-to-day.
So
I
think
I
would
defer
to
his
expertise.
A
A
A
new
class
or
something
like
that
right
and
then
you,
you
just
implement
an
interface
from
java
point
of
view.
The
same
thing
like
you,
don't
extend,
you
don't
inherit
another
object
or
you
don't
extend
another
object,
but
instead
you
implement
an
interface
and
that's
what
you
think
you
can
do
here,
but
that's
from
an
implementation
point
of
view,
but
and
that's
a
way
for
you
to
reuse
this,
like
hey.
I
want
this
object.
This
object
and
this
object.
A
That
doesn't
mean
that
if
they
have
to
be
inside
a
specific
field,
you're
already
reusing
that
implementation
level,
but
at
user
level,
let's
say
on
the
yaml
or
on
the
json,
it's
still
on
the
same
on
the
same
level.
At
the
same,
you
don't
need
another
one,
more
field.
You
know
what
I
meant
so
my
point
here
is
that
yeah
it
could
be
reusable
yeah.
But
when
are
you
going
to
reuse
an
object
like
this
because
it's
pretty
unique
like
a
server
with
a
title?
A
Are
you
going
to
reuse
the
title
of
the
server?
Then
it's
probably
the
same
server.
Why
are
you
going
to
duplicate
the
two
and
what,
if
you
already
have
a?
What,
if
re
you
imagine,
you're
reusing
this,
the
name,
the
title
and
the
description
right,
or
only
the
name
of
the
title
and
then
one
day
you
go
and
add
the
description,
sorry
and
that
description
matches
one
of
them
and
not
the
other.
So
it's
a
mess.
A
So
I
I
feel
that
this
metadata
is
so
pretty
unique
to
the
specific
object
that
it's
being
applied
to
that
you
will
not
want
to
reuse
it
right,
like
the
same
with
id
the
same
with
name,
the
same
title
descriptions
summary
tags
right
like:
why
would
I
describe
a
server
or
operation
or
a
channel
right,
two
different
of
those?
Why
would
I
describe
them
the
same
way
then.
B
A
B
The
other
fields,
you
must
do
you
know
to
copy
yes
or
write
in
the
trades
or
something
like
that.
A
Let's
introduce
trades
from
on
servers
and
channels
as
well
yeah,
but
if
that's
the
concern,
then
that
concern
could
be
addressed,
introducing
trades
on
servers
and
channels.
A
B
E
B
Am
back
with
you
know,
we've
described
my
idea,
so
it
means,
for
example,
if
we
introduce
that
six
fields.
Yes,
on
the
same
level
on
the
server.
B
Yes,
as
you
want,
then
we
can,
for
example,
the
traits
define
only
that
six
fields
and
it
will
exactly
behave
like
the
metadata
object.
D
A
Amount
of
fruits
yeah,
that's
my
point!
So
it's
to
me
it's
already
possible!
That's
why
I
would
say
like
it's
having
no
new
value
right,
it's
a
it's!
It's
already
possible!
Well,
it
could
be
adding
value
to
server
and
channel,
but
we
already
have
trades.
So
why
not
extending
trades
to
server
and
channel.
C
A
A
And
also
like,
why
does
why
this
list
of
fields
and
why
not,
for
instance,
why,
in
a
message,
for
instance,
why
the
payload
is
not
metadata.
A
Or
why
the
headers
is
not
metadata
in
a
message.
I
think
it's
metadata
the
same
as
id
yeah,
exactly
because
the
whole
object
is
metadata.
If
you
think
about
it,
it's
defining
it's
defining
a
real,
a
real
message
that
will
be
flowing
through
the
broker
at
some
point.
So
it's
the
whole
thing.
The
whole
scheme
is
metadata.
B
Yeah,
but
I
always
think
about
the
description
for
title
for
the
given
object
as
the
metadata
for
for
the
user
that
exactly
user
can
see
and
okay,
okay,
I
can
read
the
description
yes,
but
the
payload
or
the
message
id
is
the
metadata
from
you
know
the
code,
the
runtime
in
this
way.
Of
course
the
human
can
also
read
that
and
understand
that,
but
you
usually
use
that
field
from
their
anti
from
the
code
perspective,
yes
from
application
perspective,
but
but
the
description
is
mainly
for
the
human.
E
B
B
A
B
A
Then
you
can,
you
can
reuse
it
on
adjacent
schema
level
and
and
also
at
programming
language
level.
Yeah.
A
B
A
E
E
C
B
C
B
Okay,
so
I
will
change
every.
I
will
change
the
metadata
object
to
the
you
know
to
iota
inheritance,
but
probably
we
need
maybe
some
you
know
some
another
feedback
from
community.
B
A
So
so
yeah
just
leave
the
feedback
there
and
and
yeah,
and
I
will
I
can
leave
my
opinion
there.
If
you
want
okay
and
maybe
this
pull
request
can
be
reconverted
to
extending
trades
to
server
in
general.
A
I
mean
that's
only
in
case,
that's
only
in
case
it.
We
think
it
makes
sense
like
it
makes
sense
to
reuse,
title
description
and
so
on.
Maybe,
like
you
said,
maybe
in
the
case
of
virtualin,
we're
going
to
reuse
some
of
these
fields,
not
all
of
them.
B
A
B
B
Yeah
the
next
one
is
the
plural
quest,
as
I
remember
the
front
already
accepted
it.
I
think
that
the
root
tax
and
external
documentation
is
exactly
the
metadata,
but
the
metadata
for
the
user
yeah.
No,
not
everything
is
metadata
yeah.
So
I
moved
the
external
documentation
tax
from
the
root
to
the
info
object.
The
main
info
object
is
yeah
and.
B
E
C
I
don't
know
if
it
makes
the
the
spec
more
clean
to
look
at
if
it's
in
line,
then
those
can
like,
because,
as
everything
is
in
the
room,
I
I
mean
it's
the
same
print.
It's
the
same
discussion.
We
just
had
it's
just
whether
we
are
removing
an
info
object
that
has
been
there
since
just
gonna,
say
beginning,
but
I
don't
know.
B
B
Or
contact
or
maybe
but
but
extension
yeah,
it
will
behave
the
same
yes,
but
you.
E
C
B
A
E
A
A
A
In
the
implementation,
but
then
in
spec
it
will
all
be
on
the
same
level.
So
I'm
actually
thinking
that
yeah
actually
info
doesn't
make
sense,
doesn't
make
any
sense
there.
What
now
that
I'm
thinking
about,
I
just
got
so
used
to
this
from
open
api
that
yeah
that
maybe
like,
when
are
you
going
to
want
to
use
info
objects?
B
E
B
B
A
A
B
A
A
A
What's
the
meaning
of
it
so
think
about
it,
like,
like
extensions
applied
to
the
root,
are
extensions
of
the
whole
application
like
it's?
The
the
root
is
the
application
level
right,
so
extensions
on
the
root
are
extensions
for
the
application
level,
but
extensions
in
the
info
object
are
also
extensions
at
application
level.
Right.
A
So,
let's
put
it
on
the
root
level,
and
in
which
case
will
you
say
no
not
to
not
on
the
root
level?
Let's
put
it
on
the
info
object,
because
it
makes
more
sense
in
which
cases.
B
A
I'm
not
I'm
not
saying
that
I
mean
I
am
the
culprit
here.
Of
course
everything
everything
that's
done
there,
or
almost
everything
is
my
fault,
so
yeah.
Now
I'm
just
challenging
myself
and
everyone
like
do.
We
need
an
info
object
at
all.
Why
can't
these
fields
be
on
the
road
right
like
I'm,
creating
a
new,
a
new
asynchpa
file
and
it
has
a
title:
it
has
a
version.
It
has
a
description,
blah
blah
blah
blah
blah
blah
blah.
Why
grouping
them
inside?
A
If
info
grouping,
usually
the
to
me,
you
group
things
when
maybe
you
want
to
make
them
more
readable,
that's
that
would
be
an
argument
and-
and
that's
and
I
think
that's
the
argument
of
the
object,
probably
so
they're
all
together,
so
they're
not
spread
around
the
root
object
in
some
places.
A
B
A
A
So
maybe
the
only
argument
here
would
be
to
put
everything
together.
It's
a
good
place
in
a
single
place,
everything
grouped
together.
So
so
all
these
fields
are
visible
at
once
right
and
because,
if
if
we
move
them
to
the
root
object,
then
it
will
be
async
api,
title
channels,
operations,
description,
then
components,
then
terms
of
service.
A
So
there
will
be
all
like
spread
around
the
document
and
which
is
also
not
a
problem,
if
you
think
about
it,
but
but
I
mean
as
as
long
as
it
makes
sense
for
you
but
yeah.
The
only
argument
that
I
see
here
is
that
we
don't
want
to
to
keep
them
grouped
somehow.
A
A
If,
if
you
have
contact
the
contact
object,
has
email
has
a
url
so,
but
you
want
to
have
twitter,
so
you
want
to
put
twitter
inside
the
contact,
object.
C
A
B
Make
some
voting
for
community
or
something
like
that,
because
to
be
honest,
I
don't
see
any.
You
know
the
comments
in
the
open
api
community
that
I
don't
like
the
info
object.
Yes,
something
like
that.
No.
A
No,
no
me
neither
I
mean
I'm
just
I'm
just
challenging
thinking
here.
So
so
it's
just
like,
let's
not,
let's
not
give
anything
for
or
take
anything
for
granted
right
like
it
has
to
be
an
info
object
because
yeah,
how
not
right,
maybe
not
like,
for
instance,
I
don't
know
like
graphql,
doesn't
have
an
info
object.
Jrpc
doesn't
have
an
info
object
right,
so.
E
A
C
A
B
A
C
A
B
A
B
B
You
accept
the
dot
pr
which
one
I
put
the
root
tax
and
external
documentation.
You
can
see
that
yeah.
A
B
A
D
C
B
Maybe
on
the
next
meeting
is.
B
C
B
Yeah,
it's
very
easy
because
it's
only
create
another
reasonable
part.
The
reasonable
object
inside
the
component,
so
operation
object
at
the
moment.
In
the
second
version
we
don't
have,
but
maybe
in
the
next
major
version
we
want
to
introduce
it
because
you
know
we
want
to
split
the
operation
from
the
channel.
Definition
is
to
separate
field.
So
by
this
I
propose
to
also
add
the
components
operation
when
you
want
to.
You
know,
decrease
the
create
a
reasonable
operation
object
and
the
external
documentation,
and
also
the
tax.
C
C
B
Yeah
we
have
this
proposal
from
the
front.
Yes
about
this,
removing
confusion
in
the
public.
C
B
E
B
You
can
create
some
some
fields
for
the
given
operation,
maybe
description,
I
don't
know
title
etc
and
then
by
the
trades
you
only
change
something
like
the
type.
Yes,
that
the
request
or
sent
and
receive
something
like
that.
So
yeah.
We
should
think
about
that
yeah,
but.
B
A
A
C
B
The
same
way,
so
I
can
answer
by
the
questions.
What
is
the
purpose
for
the
servers
and
channels.
A
A
A
It's
reusable
stuff
in
the
same
company
in
the
same
document,
but
if
someone
same
for
say,
for
instance,
like
servers,
maybe
we
don't
need
servers
and
components
as
well
like
like,
like,
like
servers,
is,
might
be
reused
across
multiple
documents
cool,
so
you
define
the
ones
somewhere
in
one
of
the
documents
and
the
other
ones
are
going
to
be
pointing
to
to
the
to
the
first
one
directly
to
the
servers.
Why?
Why
is
it
better
to
point
to
components
that
slash
server
than
to
point
to
directly
to
servers
from
other
documents.
C
Think
you
usually
steeper,
because
you
also
have
you
also
have.
I
think
the
issue
is
deeper
than
that,
because,
for
example,
if
you
bundle,
if
you
want,
if
you
use
or
have
heavy
views
of
references,
and
you
want
to
bundle
documents
together,
for
example,
if
you
reference
servers
across,
let's
say
you
have
three
application:
three
async
api
documents
that
reference
an
external
server
file
with
a
with
the
bundling
behavior.
C
It
has
to
bundle
them
either
in
line
in
the
document,
or
it
has
to
have
some
kind
of
reusability
in,
for
example,
components,
right,
yeah
and
there
it
makes
sense
to
have
like.
Even
if
you
don't,
even
if
you
don't
reference,
multiple
servers
within
the
same
or
the
same
server
within
the
same
document.
A
B
C
But
it's
a
it's
a
it's
actually,
a
good
thought
experience
or
not
thought
experience,
experiment
but
like
question,
because
you
can
also
just
have
an
extension
that
has
all
of
these
components.
It
doesn't
matter
where
exactly
you
place
those
references
like,
for
example,
the
the
example
with
the
bundle
you
can
have
it
anywhere
in
the
document.
As
long
as
you
do
the
extensions
and
then
local
references
and
then
your
time.
Yes,
yes,
I
know
search,
that's
true.
A
With
version
3,
we
need
to
rethink
everything
before
it's
too
late
right.
A
So
that's
why
I'm
questioning
everything
like
what
the
hell
is
components
then,
because
I
think
like
when
we
started
to
put
so
many
things
in
components
I
started
feeling
like
this
is
like
this
is
becoming
a
black
box
where
you
can
throw
whatever
you
want
there
and-
and
my
gut
feeling
was
saying
like
I
don't
know,
what's
happening,
but
I
have
the
feeling
that
we're
abusing
components
and
but
I
don't
know
why
and
I'm-
and
I
was
thinking
that
maybe
yeah,
maybe
it
will
be
so
components
if
I'm
not
mistaken.
A
I
think
the
purpose
of
components
originally
was
to
have
reusable
stuff
inside
the
same
document.
Right,
I'm
not
saying
on
amazing
kpi
here,
I'm
saying
an
open
api,
but
but
yeah
check
yeah
for
this
particular
specification.
A
A
By
reference,
what
I'm
saying
is
that
it
needs
to
be
reusable
by
reference.
I
know
that
you
can
point
somewhere
else,
of
course,
but
but
it's
a
place
where
you
can
put
it
like
this,
is
my
security
yeah.
A
Whatever
not
pointing
to
specific
security
that
you
define
first
and
then
you
reason
the
next
ones,
so
so
I
think
it
makes
sense
whenever
I
think,
whenever
it's
only
for
for
reusability
inside
the
document,
I
think
components
make
a
lot
of
sense.
B
Yeah
but
okay,
but
with
security,
schemas
yeah,
the
security
schemas
inside
the
component.
Yes,
we
had
this
problem
that
we
make
the
reference
in
the
security
folds
yeah
by
the
name
of
the
given
field
in
the
components.
B
Problem
because
no,
it
says
something
like
that,
if
you
have
the
channels
and
the
servers,
it
exactly
includes
your
application
yeah,
but.
B
Is
only
you
know
the
reasonable
part,
so
it
means
that
you
can,
for
example,
create
redundant
messages
and
you
don't
you
don't
need
to
reuse
that
in
your
specification,
yes,
yeah
in
the
same
way
the
security
schemas?
Yes.
So
this
is
the
problem
and,
as
I
remember
in
open
api,
the
core
team
created
discussion
about
the
next
major
version.
4.00
and
one
person
say
why
we
need
the
security
schemas
inside
the
components,
because
it's
break
the
functionality
of
the
components
in
this
meaning
yeah.
So.
B
C
C
C
It
might
also
be
a
bit
too
late
to
answer
it
now,
but
let's
or
maybe.