►
From YouTube: Spec 3.0 meeting (February 1, 2023)
Description
B
C
B
C
Yeah
and
it's
90
degrees,
but
yeah,
it's
a
it's
easy.
It's
10
minutes
and
it's
gonna
be
released
right.
B
D
So
you
have
approval
and
approvals
from
Lucas
myself
said
here
and
nail,
but
you
got
four
code
owners
to
approve
your
repo
and
you've
been
waiting
for
Vladimir
to
prove
it
since,
when
again
two
weeks,
slash
RPM.
B
D
I
mean
you
got,
you
got
more
than
three
I
said.
The
minimum
is
three
right:
three
approvals
in
the
reaper
right.
Well,
in
this
case,
it's
it's
not
minimum
three
I
think
it's
not
set
up,
but
but
yeah
we
honor
it
like
it's
three
so
or
we
try
to
honor.
So
so.
You've
got
more
than
three
approvals
and
you
left
some
decent
amount
of
time
like
two
weeks.
I
think
it's
good
enough
amount
of
time
to
for
someone
to
just
jump
into
it.
You
pinged
him
how
many
times
I.
D
Well,
I
mean
yeah,
someone
thinking.
So
if
he's
not
responsive
for
this
specific
E2,
that's
completely
fine.
Let's
go
ahead
and
merge
it.
D
I
think
not
in
this
Reaper
but
I
think
on
the
spec
one.
We
have
something
written
down
there
that
it's.
You
should
get
at
least
we
approvals
or
something
like
that.
It
says
something
about
it
and.
C
I
think
I
think
there's
no
there's.
C
If
we
would
set
limit
to
three,
then
the
problem
is
with
the
bot,
because
bot
cannot
merge
other
PRS,
because
there
would
be
right,
yeah
and
I
quickly
thought
about
adding
two.
D
So
yeah
so
and
then
how
much
time
do
you
let
it
pass
until
you
hit
the
merge
I
think
that's
gonna
depend
a
lot
in
the
type
of
pull
requests
if
it's
a
big
one
important
one
or
if
it's
just
a
stupid
thing.
If
it's
a
stupid
thing,
and
this
guy
had
emerged
one
week
one
week
after
right,
if
you've
got
enough
approvals,
just
go
ahead
and
merge
it,
and
and
if
it's
a
big
one,
that
it's
going
to
master-
and
you
know
it's
important
one
and
so
on.
D
B
D
But
but
that's
the
thing
right
like
we're
humans
in
the
end
and
every
everything
is
relative
right
so
and
there's
no
easy
way
to
actually
put
it
there
clearly
for
each
case.
So
you
really
have
to
leverage
your
soft
skills
here
and
being
trying
to
be
polite
and
respectful
with
the
community
so
and
you've
been
respectful,
come
on
two
weeks
after
last
time
you
pinned
people
or
in
this
case
look
at
Spring
people
I.
D
Think
that's
more
than
enough,
and
on
top
of
that
you
got
four
co-downers
out
of
five
or
better,
because
I
think
we're
five
four
of
our
five
who
approved
it,
and
on
top
of
that,
it's
going
to
release
a
branch,
not
too
massive.
So
there
yeah,
not
everything,
is
or
or
I
I.
Don't
think
not
everything
should
be
coded
into
you
know
into
the
rules
and
everything
like
it's.
B
Hard
for
it's
hard
for
a
contributor-
that's
that
isn't
a
code
owner
to
know
when
when
something
is
that
fluent
right
so
but
yeah
it
might
just
be,
it
might
just
be
that
one
of
you
code
owners
or
a
code
owner
explicitly
State.
Well,
we
do
this
and
then
we
merge
it.
So
it's
up
to
the
code
owners
to
set
the
tone
of
when
something
is
okay
to
merge.
D
It's
part
of
the
game.
Man
like
it
might
be
hard
for
people
to
figure
it
out,
but
yeah
you're
part
of
a
community
to
chat
with
the
community
and
say
yes,
I,
don't
know
if
I
should
imagine
this
or
not,
and
then
someone
yeah
yeah,
you
know
like
it's
part
of
it
like
we're
humans,
we
can
communicate.
So
if
you,
if
you're,
not
right
or
not,
I,
know
I
know,
but
I
should
be
able
to
communicate.
D
Yes,
so
yeah,
okay,
like
we
don't
have
to
code
up
every
single
room,
but.
A
C
To
merge
options
so
anyone
can
trigger
release
the
emerge,
but
it
will
not
work
if
there
are
some
blockers
so
like
in
theory,
like
imagine,
you're
a
contributor
that
you
you've
done
that
in
meetings
you're,
not
in
slack,
you
don't
know
us.
So
you
you,
like
you
shouldn't
wait
for
us.
You
should
just
trigger
RTM.
C
C
My
opinion
yes
and
then
it's
our
problem,
the
the
code
owners
like
how
to
prevent
that
you
do
it
before
before.
We
actually
want
you
to
do
it.
So
in
all
the
other
repos,
it's
not
a
problem,
because
this
RTM
is
really
depending
on
the
settings
in
the
GitHub.
The
only
problem
is
this
and
the
spec,
where
we
know
we
want
to
have
three
reviewers,
but
in
the
settings.
B
C
So
so
basically
it
should
be
like
you
should
just
merge
it.
Then
we
figure
out
how
them
how
to
prevent
it
in
the
future.
And
it's
then
it's
on
the
code
owners
I
think
the
like.
It
should
be
the
me
from
Vladimir
and
Dale
and
and
Sergio.
We
should
meet
and
agree.
Okay,
then
we
should
injection,
schema
and
spec.
We
should
always
use,
do
not
merge
label
and
whenever
we
don't
want
someone
to
use
RTM,
that's
why
do
not
merge
what's
also
invented
sure.
B
So
with
the
spec
3.0
release
should
how
do
we
handle
versioning
of
bindings?
When,
for
example,
let's
say
that
one
of
the
Kafka
bindings
explicitly
have
to
be
have
a
major
change,
so
you
have
to
do
something
that
breaks
backwards.
Compatibility.
B
Then
we
would
have
to
explicitly
state
that
one
binding
is
for
one
major
version
and
another
is
for
version
two,
for
example.
So,
up
until
now,
we
don't
have
any
consensus
on
what
to
do
about
bindings.
It's
just
there
and
it
hasn't
been
part
of
the
the
validation
rules
either.
So
it's
been
kind
of
like
just
there.
B
B
D
But
we
have
a
change
of
perspective,
for
instance,
the
public
subscribes
and
received.
So
this
might
be.
A
problem
can
be
a
problem.
I
don't
know,
I
have
I
would
have
to
check,
especially
on
the
HTTP.
One.
Probably
I
haven't
checked,
but
might
be
a
a
problem
to
look
at,
but
in
my
opinion
like
what
I
will
do
is
since
we
don't
maintain
multiple
versions
at
the
same
time,
so
we
will
major
versions
at
the
same
time.
So
we're
not
going
to
maintain
two
anymore.
D
We're
just
gonna,
leave
it
there,
but
we're
not
gonna
keep
developing
version.
Two
I
think
it's
as
easy
as
marking
a
line
on
the
on
the
binding
like
up
until
here
up
until
0.2
version
of
The
Binding,
it's
for
version,
two
of
the
spec
and
from
there
on
is
for
version
three
of
the
spec
right.
So
whatever
we
add
on
The
Binding
might
or
might
not
make
sense
on
version
two,
but
I
want
this
feature
on
version
two
I'm.
D
Sorry,
you
had
to
migrate
to
version
three,
that's
what
I
will
do,
because
we
don't
maintain
multiple
versions
of
the
major
versions.
At
the
same
time,
right
so
so
then
I
will
do
select
up
until
here
is
written
to
there
on
its
version.
3.
up
until
here
is
version
three
from
there
on
this
version.
Four,
you
know
like
I'll
continue
like
this,
and
we
can
make
it
clear
on
each
of
the
bindings,
for
instance,
for
in
each
of
the
binders
we
can
just
select.
D
This
is
the
last
version
for
version
two
and
for
version
three.
This
isn't
there's
another
one,
but
that
also
triggers
or
raises
the
question
that
we
don't
maintain
multiple
bindings
versions.
Right,
we
I
mean
we
I
think
we
do
on
the
on
decent
schema
definitions,
I
think,
or
we
were
planning
to
do
it.
I
recall
that
we.
B
D
Right
yeah
but
I
mean
it's
having
a
list
of
all
the
versions
of
a
specific
bindings
you
know
like
and
to
see
how
it
has
been
evolving,
for
instance,
or
I
want
to
go
back.
One
version
on
The,
Binding
I
think
this
is
human
definition,
since
we're
having
something
like
that,
if
I
recall
correctly,
but
yeah
I'll
have
to
check
for.
C
D
D
The
latest
version
I
think
but
I
think
somewhere
into
us,
I
record
this
business
because
or
no
no,
it's
just
for
the
latest
yeah
yeah
yeah.
So
we.
What
we
had
is
that
at
Glee
we
had
a
need
which
is,
which
was
that
we.
D
D
Okay,
I'm
gonna
remove
my
face,
which
is
even
better
so
so
yeah
no
problem
with
video
anymore
I
hope
it.
It
helps
with
the
connection
so
yeah
on
on
me.
We
needed
to
have
the
version
of
The
Binding
on
the
distance
schema
for
that,
because
they
were
breaking
changes
from
one
version
to
another
and
yeah,
but
that's
it,
but
we
don't
have
we
don't
get
the
history
of
or
I
think
no,
we
we
needed
it
yeah
yeah!
No,
we
needed
it.
We
needed
it.
D
So
we
had
to
validate
The
Binding
information
that
we
get
I'm
Glee,
for
whatever
reason
I
can't
remember,
I
have
to
I
will
have
to
check
and
we
had
to
have
two
Jason
schemats
at
the
very
least
one
for
for
version
0.1,
another
150.2,
so
so
yeah
I'll
have
to
I'll
search
for
it
but
yeah
anyway.
D
The
way
it
is
right
now
to
me
is
as
easy
as
saying
that
question
two
is
about
here
and
version
three
years
from
there.
That's
it
and
I
will
not
complicate
lives,
especially
because
it's
still
on
zero
point
hunting
versions.
So
we
expect
this
to
be
continuously
broken,
so
yeah.
B
And
this
is
something
we
have
to,
at
least
in
the
current
setup.
It's
something
we
have
to
do
in
the
in
the
readme's,
because
this
spec
files
them
set
like
the
validation
rules
for
the
bindings,
has
no
information
about,
or
has
no
constraint
that
these
only
works
for
an
async,
API,
2.0
or
3.0.
It
doesn't
have
that
constraint
in
itself
as
like.
It's
only
it's
only
validating
the
bindings
themselves
and
it's
not
until
it's
actually
used
now.
I
can't
remember
what
it
is
somewhere
here.
B
D
No
I
mean
what
I
will
do
is
I
will
just
agreement
that
we
can
have.
This
is
getting
outside
the
meeting,
okay,
but
but
other
than
that.
The
only
thing
that
I
will
do
is
just
living
this
node
in
each
of
the
bindings
that
I'm
not
sure
if
I
will
do
it
in
each
of
the
bindings,
but
I
will
review
the
bindings
and
make
sure
they
make
sense
on
version
3..
D
If
they
make
sense,
I
will
not
have
anything
now,
because
bindings
are
an
experimental
stage.
Yet
so
I
will
not
add
anything
yet
and
and
for
and
if
we
get
something
like
I,
don't
know
the
HTTP
binding
it's
a
problem,
then
we
leave
a
message
in
the
on
the
HTTP
by
name
another
rest
and
we'll
leave
a
message
saying
what
I
just
said.
Like
version
two
is
happened
to
here,
question
three
is
from
there.
That's
it
and.
D
B
D
D
B
D
B
For
The
Binding
repository
to
review
all
of
the
bindings
for
version
three
and
then
I'm
gonna,
add
an
issue
about
adding
a
node
in
the
bindings
Repository.
For
specifically
stating
this
compatibility
thing.
C
No
I
didn't
get
it
in
the
second
sentence.
You
said
like
so
I
get
like.
So
definitely
we
need
an
issue
that
the
review
is
needed.
We
have.
C
C
Is
going
but
the
other
part?
What
did
you
mean.
A
C
End
like
it's
not
a
it's
we're
not
making
here
any
decision
and
it's
it's
a
kind
of
decision.
Oh.
C
Yeah
yeah
yeah
yeah.
We
need
an
issue,
but
maybe
in
the
same
like
in
during
the
because
in
the
in
this
issue
for
the
review,
you'll
ping,
all
the
code
owners,
so
they
can
already
share
an
opinion
what
they
think
because
I
on
my
own,
like
I'm,
struggling
I,
imagine
solution
without
actually
knowing
that
it's
needed
the
solution
itself
like.
C
Maybe
there
will
be
no
changes
and
that's,
but
we,
but
we
still,
of
course,
need
a
note
that
that
the
current
bonding
conversion
is
compatible
with
two
and
three
but
I'm,
not
convinced
because
I
I
just
look
now
on
the
Kafka
and
it's
zero
four
right
and
it's
at
the
moment
we
have
an
approach
with
Bindings
that
it
works
with
any
minor
release
right.
So
any
one
to
two
to
three
to
four
to
five.
We.
C
In
my
opinion,
it's
not
like,
like
a
line
for
Aldi
for
all
the
findings
like
so
I'm,
not
sure
if
we
should
apply
the
rule
by
default
to
all
the
bindings
or
only
the
Bindings,
that
needed
to
be
fixed
and
then,
if
binding
owner
reviews
and
says,
okay,
there's
no
changes
then
like
we
just
had
a
a
note
that
the
binding
should
be
compatible
with
minor
versions
of
two
point
and
three
point.
B
Yeah
I
guess
we'll
see
right
how
how
well
it
plays
together
should
I
add
those
issues
or
that
issue
before
or
after
we
are
done
or
before
after
the
freeze,
because
we
kind
of
have
to
know
which
changes
are
there
in
order
to
give
feedback
to
The
Binding
owners
about
which
changes
might
affect
the
bindings
right.
C
But
the
request
reply,
I,
don't
think
it
was
in
any
binding.
Nobody
tried
to
do
like.
We
always
tried
to
block
people
from
work,
arounding
spec
in
The
Binding,
so.
C
D
B
D
If
you
want
to
do
it
right
now
or
right
after
the
call
just
do
it
now,
but
what
I?
What
I
mean
is
exactly
that
is
as
soon
as
possible
right
after
the
call,
if
you
want.
B
No,
no,
of
course
not,
but
my
follow-up
question
is
my
follow-up
question:
do
we
have
enough
information
now
to
let
the
code
owners
of
the
bindings
like
explicitly
State
what
changes
might
affect
their
bindings
and
which
changes
they
should
look
out
for
because.
B
Are
are
some
of
the
changes?
What's
it
called
like
are
finalized
it's
hard
to
explicitly
state
that
you
should
look
into
this,
but
later
down
the
line
when
it's
done
I
mean
maybe
just
that's.
D
My
point
you
created
today,
but
we
don't
have
to
address
it
today,
so
you
greet
it
today,
so
we
don't
forget
about
it.
So
we
take
it
into
account
right
and
but
we
can
start
addressing
it
after
we
freeze
the
aircon
battery
and
and
but
in
any
case
there
are
things
that
we
know
that
they're
gonna
make
it
to
version
three
right.
So
most
of
it
we
know
already
it's
a
it's
stable,
it's
I
mean
stable,
it's
it's,
it
makes
sense
or
it
looks
like
it
makes
sense.
D
So
so
we
know
that
we're
gonna
have
a
change
in
the
perspective.
We
know
that
we're
gonna
have
request
reply.
Somehow
we
don't
know
yet
exactly
the
details,
but
you
can
start
thinking.
Oh
in
the
case
that
we
have
request
reply,
is
it
going
to
have
conflict
somehow
cool
so
then
I
might
want
to
as
a
code
owner
then
I
might
want
to
put
a
note
or
reminded
for
me
here
that
after
the
freeze
after
request,
reply
is
merged
and
everything
is
frozen
already,
I'll
review
it.
D
B
C
Yeah
no
I
haven't
schema
bindings.
C
I
just
had
a
so
when
you
show
the
schema.
I
realized
that
I
think
we
have
a
problem
there
right
because
we
and
it's
something
that
Fran
also
mentions
like
in
Glee.
He
had
to
have
both
schemas
of
both
versions
of
probably
Kafka
binding.
B
We
have
all
the
all
the
all
the
versions
that
are
compatible
with
that
major
version
of
async
API,
and
then
one
of
them
is
of
course
marked
as
the
as
the
latest
one.
That's
the
newest
one.
So
that's
the
default
value
right,
but
all
the
others
are
explicitly
stated
that
they
are
only
matched
against
whenever
The
Binding
version
property
is
specifically
set
for
themselves.
C
C
B
When
did
that
change
the
capture?
Five
six
I
could
have
swore
that
I
have
no
idea
which
bindings
I
was
looking
at
them.
Okay,
I'm
gonna.
Take
a
look
at
that
as
well.
D
Okay,
so
other
than
that,
so
now
my
question
is:
do
we
have
anything
on
the
pipeline?
That's
still
missing
or
version
three,
and
that
we
want
to
make
it
part
of
a
person
tree,
because.
C
I,
don't
I
wanted
to
ask
this
yeah,
but
first
of
all
like
because
I
was
out
a
bit,
there's
no
objection
from
the
community.
So
we
do
the
freeze
right.
A
D
A
C
E
Yeah
I
think
that
may
be
a
mistaken
comment
because
it
looks
like
there's
another
proposal
that
deals
with
you
know
we're
very
interested
in
multi-format,
multi,
schema
format,
support
and
so
that
one
doesn't
look
like
it's
look
like
it
moves
forward,
but
it
looks
like
another
proposal.
That's
very
similar
for
multi
format.
Support
did
move,
move
forward
just
reading
through
it.
E
B
So
the
status
with
that
is,
it
will
not
be
included
as
part
of
version
3,
but
it
is
just
a
feature:
change,
okay,
so
going
forward.
It's
just
a
change
of
at
least
that's
the
idea,
because
we
don't
have
the
underlying
standard
for
it
regarding
references
so
yeah
and
we
don't
have
the
tools
for
it,
which
is
another
problem.
B
So
yeah
that's
going
to
be
a
longer
discussion
for
a
feature
in
the
future.
Hopefully
I'm
not
going
to
say
a
timeline
because
I
always
that
up.
E
Let
me
pull
up,
let
me
pull
up
the
what
I
was
looking
at.
B
E
One
there's
something
because
there
was
something:
let
me
let
me
find
it
first
and
if
I
find
it
then
I'll
I'll
share
my
screen,
so
it
looks
proposal
to
allow
defining
schema
format
other
than
default.
One
622.
E
A
E
But
it's
possible
that
this
would
be
included
in
B3
right
because
it's
aren't,
it
would
meet
the
the
code,
freeze,
requirements
and
everything
and
it
would
address
you
know.
Potentially,
you
could
use
this
to
put
an
XML
schema
in
there,
because
most
of
the
examples
are
Avro.
But
it
looks
like
you
can
reference
out
to
a
different.
B
So
the
problem
with
his
issue
or
pull
request
is
that
he
bundles
two
features
together.
E
E
Agree
with
sort
of
you
know
taking
over
a
very
standard
way
of
doing
things,
it
doesn't
make
sense,
and
you
know
I
agree
with
that.
I.
B
Mean
we
would
love
to
because
if
we
can,
if
we
could
just
say
we
use
this
standard
for
referencing,
schemas,
perfect
use
it,
the
problem,
is
it
doesn't?
It
doesn't
exist,
or
at
least
I
can't
find
it
so
yeah
be
wrong,
but
I
can't
find
it.
E
Totally
I've
been
down
the
same
road,
so
I
yeah
I,
agree
that
there
isn't
something-
and
it
certainly
seems
like
open
API
struggles
with
this
as
well
I
mean
I.
It
seems
like
we
go
round
and
round
and
round,
and
the
answer
maybe
is
just
have
this-
be
our
own
named
field
rather
than
you
know,
dollar
sign,
ref,
yeah,.
B
We
definitely
have
conversations
about
this.
Also
in
a
Json
schema.
I
can
link
you
something
accurate
two
seconds
so,
even
though
that
they
don't
have
this
problem
of
referencing
non-json
standards,
it's
still
a.
B
E
Yeah
yeah
I
totally
get
it
it's
tough
because
it's
a
pretty
pretty
crucial
functionality.
And
but
it's
helpful
to
me
for
me
to
understand
what's
in
what's
out
in
in
terms
of
of
that,
and
you
guys
do
think
and
I
think
that's
right-
that
it's
not
like
necessarily
a
breaking
change,
because
at
this
point.
C
I
actually
I
have
totally
different
opinion.
I
think
we're
mixing
two
different
topics
here:
okay,
so
so
so
there
are
two
things
brought
above
XML
and
stuff,
and
that's
about
referencing
and
and
the
custom
schema
it's
different
custom
schemites
that
you
just
want
to
say
what
schema
like.
You
can
say
that
you
have
XML
it's
just
not
going
to
work,
but
I
mean
you
can
Define
any
any
schema
format
you
want,
and
now
it's
on
the
message
level
and
but
you
just
can't
have
a
reference
to
some
nested
part
of
the
XML.
C
For
example,
data
point
or
whatever
is
the
XML
naming
and
that's
one
topic
and
that's
I,
don't
think
we
enabling
XML
would
be
a
breaking
change.
I,
don't
think
it
would
be
because
it's
as
we
also
mentioned
here,
that
we
can
just
come
up
with
alternative
dollar
reference
field
like
dollar,
whatever
or
whatever
name
of
the
field
that
would
be
custom
just
for
XML
and
that
I
don't
think
it
will
be
a
breaking
change.
C
A
C
Yeah
and
and
this
one
actually,
this
proposal
from
Machi
so
there
are,
there
are
many
different
proposals
involved
in
one
proposal,
but
that's
what
you're
showing
it's
actually
the
the
problem
is,
and
that
was
his
proposal.
So
the
current
problem
is
that,
if
you
don't
like
can
I
share
my
screen,
maybe
yeah.
C
Oh
yeah,
so
currently,
when
you,
for
example,
have
Avro
right,
you
can
have
reference
to
some
external
avrofile
and
it's
gonna
work
right
would
be
the
reference.
C
Wait,
oh
yeah
here
in
the
schedule
you
had
this
issue.
You
mentioned.
A
C
So
here
in
this
issue
that
you
linked
we're
discussing
like
this
XML
referencing
Etc,
now
there's
this
other
one.
Can
you
share
me
send
me
the
link
of
the
one
that
you
were
showing
Jesse
at
the
yeah
I.
Don't
remember
what
was
the.
C
Basically,
the
idea
of
and
that's
why
this
pull
request
from
Machi,
it's
about
extending
objects,
think
schema
objects
because
schema
object
is
in
components
right,
and
the
thing
about
schema
object
is
that
it's
purely
just
a
schema
right.
So
there's,
let
me
so
the
in
the
in
schema.
You
can
just
put
the
name
of
the
schema
and
the
schema
and
his
proposal
was
that,
like
to
overcome
this
issue
that
you
cannot
use
Avro
in
the
in
the
components
was
to
have
a
kind
of
custom
schema.
C
So
so
inside
payload
you
don't
Define
the
schema
directly,
but
you
have
an
object
where
you
can
specify
the
format
and
the
schema,
so
the
format
is
no
longer
defined
on
the
message
level
but
inside
the
payload.
So
that's
at
least
to
my
understanding.
This
is
the
pull
request
about.
This
is
the
solution
and
it
was
like
mainly
driven
by
one
of
the
like
one
of
the
issues
from
the
community
where
actually
from
Michael
from
Solas-
and
it
was
mainly
like
it
was
mainly
related
to
this.
C
This
problem
that,
in
the
message
you
have
you
say,
Avro,
but
you
don't
have
to
you-
don't
want
to
have
your
schema
inlined
in
the
payload.
You
don't
have
want
to
have
it
in
the
separate
file.
You
want
to
have
it
in
the
components,
and
this
basically
doesn't
work.
C
But
when
I
was
writing,
this
message
I
actually
thought
that
maybe
it's
just
a
bug
in
the
parser,
not
really
technical
problem,
but
that's
different
different
thing,
and
this
is
if
this
is
not
a
problem,
if
that's
not
just
a
bug
with
the
parser,
but
actually
something
more
complicated
like
like,
for
example,
what
was
proposed
by
mache,
then
this
is
a
brain.
This
will
introduce
a
breaking
change
because
it's
it's
a
change
in
the
object,
but
of
course
we
can
always
have
a
tuple
right,
so
we
can
have.
C
We
can
say
that
schema
can
be
just
a
schema
or
schema
objects
can
be.
We
can
contain
also
an
object
that
contains
information
about
schema
format.
So,
in
the
end,
of
course,
it
all
depends
how
we
approach
it,
but
that
would
that
I
think
are
totally
two
different,
separate
things
like
a
problem
with
that.
C
Currently
in
the
spec,
you
can
use
only
Json
schema
in
components,
schema
object
and
the
other
one
is
XML
protobuf
XML
protobuf,
not
breaking
change
in
my
opinion
can
be
done
in
3.1
3.2,
but
this
one,
depending
on
the
reason
why
it
doesn't
work,
might
be
a
breaking
change
and
yeah
I
agree.
We
should
maybe
try
to
put
more
attention
to
it
and
drive
this
discussion
in
the
other
issue
to
see.
What's
the
what's
the
reason
and
what's
the
possible
solution.
E
D
Jesse,
so
what
how.
D
D
Mean,
like
schema,
sorry
I'll,
ask
you
for
a
second
no.
D
Said
I
was
saying
that
for
Jesse,
so
the
question
was
for
Jesse
like
how
how
many
schema
formats
do
you
know
that
your
customers
are
using
customers
of
soil.
As
you
know,
it's
just
a
schema
Avro
XML,
maybe
xsd
instead
of
XML.
E
Yeah
I
mean
like
xsd,
is
like
the
major
one,
but
I
mean
I
just
feel
like.
We
need
to
have
some
sort
of
generic
solution
for
it,
but
I
would
say
that
90
of
what
we're,
what
we
see
right
now
is
demand
for
XML.
You
know,
events
coming
out
of
sap
are
still
formatted
in
in
XML,
so
that's
pretty
important
for
what
we
deal
with
but
again,
I.
Don't
think
that
we
should
solve
it
only
for
XML,
but
more
more
generally,
because
I
do
think
that
those
those
cases
are
not
not
rare.
D
Because
my
so
my
thought
here
is
that
maybe
we
don't
have
to
find
the
generic
solution
for
all
the
cases
to
work.
So,
even
though
it
might
look
like
the
ideal
solution,
but
given
that
it's
a
complex
situation,
maybe
what
we
can
have
other
components
is
is
something
okay,
I'll,
stop
sure.
A
D
I'm
gonna
stop
I'm
gonna,
stop
sharing
sharing
my
video
today.
This
is
my
general
connection,
is
not
good,
so
I
was
saying
that
maybe
we
can
have
schemas
under
components
just
for
decent
schemas
for
the
default
ones.
If
you
want
and
other
components,
we
can
have
access
to
schemas
and
protobub
schemas
as
different
keys,
and
then
it
would
be
just
a
matter
of
the
parser
check-in
on
the
payload.
D
When
we
reference
something
from
the
components
we
will
check
if
it's
a
if
it's
referencing
a
decent
schema
or
or
a
product
above
schema
or
a
Nexus
schema,
and
this
way
we
can
make
differentiation
between
the
three
and
we
don't
have
to
support
more
because
I
never
hear
anything
about
the
any
other
format.
Avro
is
the
only
one
but
average
Json
so
average
is
not
a
problem.
Apples
will
be
easily
doable.
We
can
use
dollar
wrap
inside
our
as
well.
So
that's
completely
fine.
So
so
average
is
not
it's
not
really
a
problem.
D
We
can
still
have
another
section,
average
schemat
in
inside
inside
components
and
and
that's
it.
We
list
all
the
possible
schema
formats
that
we
support,
which
are
four
in
this
case,
and
that's
it
maybe
in
the
next
major
version,
we
find
a
cool
solution
that
works
for
all
of
them
without
having
to
separate
them
into
different
keys,
but
yeah
like
just
an
idea
that
we
probably
don't
have
to
yeah.
E
I
mean,
but
then
you
start
getting
into
like
protobuf,
but
to
me
I
mean
I,
guess
I,
don't
you
know
you
say
you
have
all
these
different
key
for
XML
different
key
for
you
know,
yeah,
probably
coming
on
the
second
like
but
like
it
seems
like
if
you
could
just
have
like
one
key
for
the
one
key
for
the
you
know.
D
B
E
Yeah
I
mean
that's
tough
right,
you
know.
Maybe
you
have
a
link,
then
you
get
into
sort
of
like
you
know.
You
have
three
free
Fields
like
you,
have
a
schema
format
field
and
then
you've
got
like
an
outside
reference
field.
That's
not
dollar
sign
ref,
but
something
that
we'd
come
up
with
on
our
own
and
then
you've
got
like
the
inline
plain
text.
Schema
of
you
know
whether
that's
an
xsd
or
a
protobuf
definition,
or
something
like
that.
D
D
I
know
I
know,
but
the
thing
is
that
in
any,
but
anyway
we
should
be
yeah.
We
will
be
supporting
in
lining,
because,
even
though
you
probably
you're
probably
using
another
file
where
you
have
the
accessd
file
or
the
product
at
some
point,
you
might
want
to
bundle
things
right.
E
D
Yeah
yeah,
so
you
might
want
to
bundle
things
into
one
file
and
then
fit
it
into
some
of
the
tools
or
services
that
you're
using,
and
in
this
case
you
need
to
be
in
line
and
not
referencing
any
other
file.
So
but
but
yeah
like
in
lighting
is
not
really
a
problem,
because
the
only
problem
is
that
yeah,
it's
a
pain
to
to
edit
manually
but
yeah.
D
We
just
I
mean
the
tools
to
be
clear
about
it
like
right,
like
it's,
you
can
link
it
and
don't
edit
files
inside
this
field.
Here
you
don't
have
to
you.
You
can
lock
yourself,
but
but
if
you
want
to
still
export
it,
you
can
export
it
as
a
single
file
and
that's
preferable.
If
someone
wants
to
avoid
these
recommendations
and
inline
things
themselves
and
I
think
things
there
that's
their
problem,
but
but
yeah
and
the
reason
I
was
saying.
D
This
is
because
in
each
of
the
components
section
problem
above
schemas
other
schemas,
and
all
of
that
we
can
use
different
referencing
models
as
well
right
for
each
of
the
schema
for
each
of
the
schema
formats.
So
for
inside
protoba
schemas,
we
could
be
using
something
custom
that
will
be
give
me
the
URL
of
the
schema.
D
What
is
it
it's
a
file
on
my
system,
it's
somewhere
else
and
in
another
format
that
will
be
allowing
you
to
reference
a
specific
message
inside
inside
the
product
file
and
messages
can
be
nested
so
inside
specific
messages
in
a
specific
field
or
whatever,
whatever
it's
loaded
there
and
that's
for
product
and
for
xsd,
will
be
completely
different,
probably
Maybe,
not
maybe
not
so
different,
but
instead
of
referencing
with
dollar
ref
or
with
our
own
way
of
doing
it.
We
can
use
XPath,
for
instance,
which
is
standard
for
that.
D
So
so
we
can
use
expat
or
XML
preferencing
and
that's
my
point
like
we
don't
have
to
use
a
single
solution
for
all
and
at
least
not
in
the
beginning,
because
it's
not
easy.
It's
not
going
to
be
easy
right,
but
if
we
do
it
like
this,
we're
already
providing
value
right,
we're
already
providing
value
to
users
and
that
will
be
able
to
use
product,
buff,
xsd
and
everything
else
inside
the
async
API,
maybe
not
in
a
elegant
way.
But
who
cares
about
that?
Being
elegant,
yeah.
E
I
mean
I,
don't
care
about
Elegance,
but
I
do
care
about
tech
debt,
like
you
know,
if
we,
if
we
do
it
this,
if
we
do
it
this
way,
then
you
know
when
you
when
it
comes
to
when
we
do
it
for
real
and
think
about
it,
then
people
have
to
radically
change
both.
You
know
their
tooling,
both
Import
and
Export
sides.
So
yeah.
D
That's
true,
but
that's
only
one,
that's
assuming
that
we
find
one
assuming
that
we
find
a
way
to
to
you
know
to
unify
all
the
all
the
formats
which
is
honestly
something
is
going
to
happen
ever,
but
but.
B
D
E
Sorry
everything
that
you've
talked
about
bran,
you
know
I
I,
totally
agree
that
there's
always
gonna
be
about
quirks,
but
it
sounds
like
and
I
think
we
actually
have
an
issue
open
with
us
that
I
just
didn't
pursue
long
like
what
we
want
is
to
me
is
like
a
place
where
you
can
specify
what
this
thing
is,
what
the
format
is
a
place
where
you
can
have
a
reference
to
an
external
file.
E
If
you
want
a
place
where
you
can
have
the
entire
schema
inlined,
if
you
want,
then
also
some
sort
of
pointer,
whether
that
be
x-rf
or
you
know,
and
protobuf
pointer
to
you,
know,
sort
of
a
sub
section
of
that
particular
of
that
schema,
and
it
seems
like
that
would
cover
80
percent
of
of
what
we're
asking
for
here
and
provide
a
lot
of
a
lot
of
value
without
having
the
sort
of
proliferation
of
specialized
fields
for
every
schema
that
comes
along.
B
I
think
I'm
gonna
afford
the
because
we're
running
out
of
time,
I
think
I'm
gonna
create
a
specific
call
just
for
that
discussion,
maybe
not
in
the
next
following
weeks,
but
just
to
move
that
issue
along
or
proposal
along
because
it's
going
to
be
a
very,
very
complex
discussion.
I
think
many
different
and
what's
it
called
like
many
different
passes:
branches
yeah.
Exactly
that's.
B
I
think
I'm
gonna
do
that
and
for
Magic's
proposal,
but
allowing
schemas
within
components,
I
think
I'm,
gonna
ping
him
again
to
get
his
opinion
on
his
or
these
status
and
what's
needed,
and
we
can
take
it
from
there.
E
E
What
techniques
you
have
for
for
extracting
your
your
imposing
your
will
on
people.
E
Perfect
I,
like
it
I
like
it
all
right,
guys,
I
gotta
drop,
but
it's
been
fun
and
I'm.
Looking
forward
to
continuing
later
thanks
man.