►
From YouTube: AsyncAPI SIG meeting 2 (Feb 19, 2019)
Description
This is the recording for the AsyncAPI Special Interest Group (SIG) meeting #2.
Attendees:
* Emmelyn Wang
* Enrico Hofmann
* Fran Méndez - Moderation
* Mike Ralphson
* Raisel Melian
Agenda:
* AsyncAPI 2.0.0 status update
* Roadmap for tooling and proposals on how to build tools
* Q&A
A
A
A
So
for
the
status
update,
I'm,
just
gonna
use
the
blog
post,
which
I
think
summarizes
it
well
so
I
like
to
do
to
mention
for
those
of
you
who
are
not
yet
aware,
we
did
some
changes
to
the
to
the
specification
that
are
not
yet
launched.
They
are
merged,
but
they
are
not.
If
lunch,
they
will
be
part
of
version
2.
A
A
The
events
section
in
favour
of
the
root
channel
so
instead
of
having
events
and
topics
and
or
topics
or
you
could
have
channels
and
slash,
and
this
last
channel
will
will
will
act
as
a
global
one
like
like
the
one
for
events,
so
events
is
not
needed
anymore.
Eventually,
we
will
get
rid
of
the
de
stream
one
as
well.
Whenever
we
have
the
protocol
matters
which
we'll
talk
about
it
later,
as
I
said
some
organizational
changes,
so
Mike
and
Reza
joined
us
as
maintainer
x'.
A
We
now
run
this
meeting
every
two
weeks
before
it
was
every
week
and
it
was
we
found
out
that
it
was
too
much.
Basically,
we
didn't
have
enough
news
to
share
and
yeah
I
refused
so
it,
but
we
did
some
major
changes
to
the
rift,
and
now
you
can
find
lots
of
information,
don't
check
it
out.
Whenever
you
have
time-
and
this
week
myself
at
least,
is
I'm
working
on
adding
support
for
multiple
schemas
and
when
I
mean
schemas,
because
this
is
schemas
is
a
prostituted
term.
A
Right
now
we
were
using
it
for
many
different
for
many
different
minions,
so
schema
as
schemas
as
a
way
to
define
the
content
of
a
message.
Right
now
we
were
using
JSON
schema
flavor,
let's
say
which
is
inherited
from
open,
API
and
and
we
will
start
supporting
different
ones.
Like
JSON
schema
draft
four
to
seven
proto
buffers
Avro,
you
name
it
that
the
spec
would
leave
it
open
and
it
will
be
responsibility
of
the
tool
in
to
parse
these
formats
to
be
able
to
actually
validate
the
messages
and
and
so
on.
A
Improving
connectivity
is
a
name.
It's
a
very
fancy
name
that
I
created
for
the
for
this
milestone
for
grouping
a
lot
of
tasks
that
we're
related
to
servers
and
to
the
server
section
of
the
of
the
of
the
spec
and
the
schemes
and
and
and
and
and
security
requirements,
separate
variables.
You
know
all
that,
so
that
shouldn't
be
that
shouldn't
take
too
long
because
there
are,
even
though
there
are
like
six
tasks,
they
are
all
very
small
like
some
of
them
are
just
5
min
changes,
so
it
shouldn't
take
too
long
and
that's
my
plan.
A
A
A
So
we
have
a
donate
link
here
and
we
also
have
one
here
which
is
become
a
become
a
sponsor
and
they
all
take
you
to
open
collective
table
so
open
collective
is
a
platform
from
for,
let's
say,
crowdfunding,
for
open
source
projects
or
for
open
initiatives.
If
you
want
from
public
and
transparent
initiatives,
okay,
so
we
are
hosted
by
an
open
source
by
a
non-profit
foundation.
A
The
open
source,
collective
5:01
see
six.
This
is
a
type
of
foundation.
That
means
it's,
not
it's
non-profit,
so
donations
can
some
people
think
actually
get
taxes
back
or
something
like
that.
I,
don't
remember.
To
be
honest.
This
is
this
is
something
that
that's
finances
from
the
US
I,
don't
know
if
you
write
or
or
Emily
know
better,
but
when
you,
when
you
patron,
when
you
donate
to
a
non-profit,
you
can
ask
for
the
taxes
packaging.
A
So
you
see
here
we
have
a
description
of
the
project
here
and
different
that
description
up
on
the
left
and
on
the
right.
You
have
different
ears
like
let
me
go
back
and
you
have
buggers.
You
have
bronze
sponsors,
silver
sponsors,
gold
and
very
support
sponsors,
so
they
all
share.
One
thing:
backers
are
basically
from
individuals.
A
C
A
Said
so
much,
but
we
we
appreciate,
and
then
you
have
bronze
silver
and
gold
sponsors
which
all
of
them
are
will
get
you
a
logo
in
in
our
repo
run.
Sponsor
will
get
a
a
logo
on
on
the
Packers
on
the
Packers
markdown
file.
There's
no
Packers
markdown
file,
yet
because
we
don't
have
any
brand
sponsors,
but
we
will
trade
one
as
soon
as
we
have
one
and
then
silver
sponsors
and
gold
sponsors
will
get
their
logo
on
the
ring
and
on
the
website
as
well
on
both
places.
A
A
I
think
that's
it
mostly
for
the
status
update.
Well,
let
me
share
something
with
you:
there's
a
pull
request
going
on.
This
is
a
data
issue,
not
the
pull
request,
but
here
it's
better
explained.
So
this
is
exactly
what
I
was
saying
that
we're
gonna
support
multiple
schema
formats,
so
this
is
an
example.
A
This
is
an
example
of
message,
payload
that
it's
defined
using
protobuf
here,
it's
defined,
put
in
some
hard
code,
Prada
back
there,
which
is
not
nice,
because
it's
a
string,
but
you
can
always
refer
to
the
to
your
protobuf
file
and
you
can
you
can
keep
it
where
it
is
right,
so
you
don't
have
to
write
it
here
in
the
document.
This
is
just
an
example
using
protobuf
because
it's
very
different
to
json
schema,
but
you
can.
You
could
replace
this
with
Avro
and
with
any
other
format
that
you
wish
to
use.
C
C
A
On
that
case,
because
you
may
not
have
it
on
the
same
format
so
think
about,
for
instance,
you
may
have
things
define
in
protobuf
like
the
payload
might
be
defined
in
proto
bot,
but
you
don't
have
headers
defined
in
proto
bar
right.
So
you
might
want
to
use
this
schema
here
right.
If
so,
you
probably
what
you're
suggesting
is
that
we
coul
move
this
app
one-level
app,
the
the
application
expert
above
one
level
app,
and
that
will
be
the
same
for
both
for
Heather's
and
for
Palos
right.
A
C
A
And
if
they
want
to
have
it
like
really
quickly
without
having
to
bother
creating
another
another
file
for
for
take
headers,
they
could
do
it
here
using
JSON
schema
into
the
proto
buffers.
So
we
so
allow
some
flexibility
like
some.
Some
some
part
of
the
of
the
document
might
be
use.
Json
schema
and
some
others
will
use.
A
C
A
It's
a
so
headers
will
start
here.
You
know
what
I
mean
I
mean
if
you
have
a
better
idea.
I
can
show
you
what
I,
what
I've
been
working.
What
I
we
were
here,
I
discovered
this
illusion
results
precisely
for
that.
Like
I
have
this
solution
where
I
was
putting
schema
type
here
and
then
you
have
headers
and
payload,
but
that
was
that
was
enforcing
people
to
have
headers
and
payload
in
the
same
format
they
might,
they
might
not
wonder
they
might
not
have
it,
which.
A
A
Just
these
I
just
added
this
line,
I
just
updated
this
this
the
following
is
not
valid.
Okay
and
these
discarded
solutions,
I
have
another
one
here,
which
is
schema
tied
and
URIs.
Instead
of
using
dress
yeah,
then
I
found
out
because
in
the
beginning,
I
thought
if
it
was
reference
and
something
that
it's
not
Jason,
it
might
cause
some
problems
to
to
some
resolvers
when
it
fetches
the
product
and
it's
not
Jason
right,
but
but
then
I
realized.
Okay.
A
A
D
I
just
had
one
question
about
the
use
of
application:
X,
protobuf
I
know
it's
just
an
example
here,
but
I
had
a
quick
look
at
the
RFC
that
defines
application
protocol,
which
is
an
expired
RFC
and
it
doesn't.
It
doesn't
go
into
any
detail
about
how
to
resolve
the
fragment,
ID
and
fragment
ID
resolution
is
supposed
to
be
defined
by
the
media
type.
A
D
A
A
A
We
don't
have
to
stick
to
one
I
just
need
to
be
clear
on
which
ones
are
the
support.
That's
right,
but
yeah
I
agree
that
I
mean
this
was
kind
of
an
invention
of
mine
like
if
you
go,
it
makes
sense.
To
me
like
a
fragment
is
actually
for
that.
The
definition
of
a
fragment
is
exactly
that.
I
want
just
a
specific
part
of
the
of
the
Fetchit
document,
but
yeah
it.
D
A
C
A
A
A
A
A
Not
or
or
you
you,
you
don't
ll
to
go
messages,
because
what
I'm
saying
is
that,
for
instance,
the
examples
that
I've
seen
for
nesting
is
that
if
you
want
to
have
a
imagine
that
you
have
a
special
type
here,
like
user
and
continuing
inside
the
search
request,
continuing
inside
you
define
what
you
means,
but
you
will
not
want
to
target
this
user
object
right
under
the
top-level
right.
The
other
thing
that
people
do
is
that
they
they
don't
put
it
on
the
inside.
They
put
it
aside
like
here
and
there
and
it
it
works.
A
A
I'm
not
sure
if
you
can
probably
have
eight
innings
on
top
level,
maybe
in
any
case
I
think
we
so
target
always
messages
right.
That's
what
I
think
so
that's,
but
that's
Prada
buff
specific
we
can
we
can,
then
we
can
go
through
the
specific
of
this
one
one
while
we
are
defining
how
it
should
work
this
one
specifically
but
but
yeah.
Another
thing
is
the
way
it
should
be
resolved.
Might
change
for
every
schema
format,
because
different
schema
formats
might
have
different
needs
like
I'm
I.
A
Don't
have
a
use
case
right
now,
but,
for
instance,
thinking
about
binary
formats.
If
you
want
to
specify
something
here,
that's
a
binary
file.
You
may
need
some
sort
of
way
to
indicate
that
your
you
need
to
do.
You
need
to
refer
to
some
part
of
your
of
your
binary
file.
Like
imagine,
I
want
to
just
drive
from
by
3
to
write
10,
something
like
that
right,
I,
don't
know,
I'm
just
invent
in
the
newspaper,
so
don't
take
it
to
so
much
to
account,
but
but
probably
the
way
it
works.
A
Because
I'm
using
too
many
new
words,
so
this
is
the
the
document
for
the
for
the
tooling
roadmap,
the
strategy
and
and
also
the
technical
proposal
on
how
to
how
to
help
implement
it.
So
once
thing
before
we
go,
there
is
this
terminology
that
I
was
using
its
new
terminology.
I
want
to
make
it
clear
that
the
way
I
see
the
way
I
see
that
this
implementation
is
by
separating
things
in
at
least
that
I
have
think
of
it's
for
four
parts.
So
we
have
the
the
high-level
specification
partial
validator.
A
This
is
the
one
in
charge
of
parsing
the
whole
document,
the
the
whole
structure
and
checking
if
it's
actually
missing
KPI
documents
like
verifying
that
the
first
field
is
a
async
API
with
version
two
blah
blah.
It
has
channels,
it
has
servers,
they
have,
they
all
have
the
destructor
the
correct
structure
right,
but
it
doesn't
go.
A
It
does
not
go
into
details
as
it
says
here
on
how
to
parse.
So
here
here,
good
par
schemas
and
that's
very
important.
We
just
want
to
have
as
a
piece
of
software
that
we
check
for
this
or
this
structure.
This
is
just
a
JSON
schema
partner.
It's
nothing
else,
it's
just
what
we
have
right
now,
but
ignoring
the
the
part
of
where
we
define
a
message.
That's
it
and
then
once
this
is
one.
Let's,
let's
forget
about
that.
A
Then
we
have
another
another
piece
which
is
a
schema
part,
and
this
is
the
one
or
these
are
the
ones
that
will
be
in
charge
of
parsing
every
every
type
of
schema.
If
we
have
a
decent
schema,
Edition
schema
definition
for
free
message.
We
will
use
the
JSON
schema
parser
for
this
message.
If
it's
a
Prada
path,
we
will
use
the
protobuf
for
that
for
this
message,
definition
and
so
on
right
and
we
will
have
to
support
open
API
scheme
as
well.
I,
don't
know
Mike.
A
A
Yeah
yeah
yeah,
all
of
them
I
mean
all
of
the
B
version
is
like,
for
instance,
brother
bath,
I,
think
it
has
a
mainly
version,
two
and
three
so
say
yeah
and
then-
and
this
part
is
specifically
the
protocol.
Mapper
is
specifically
for
for
SDKs.
If
you
want,
so
this
one
is
in
charge
of
telling
the
the
SDK
how
it
will
encode
and
decode
or
how
its
utilized
or
this
utilize
the
message.
A
So,
for
instance,
if
we
want
to
send
the
message
over
mqtt,
we
will
like
to
have
these
headers
like
quality
of
service
header,
and
they
can
just
be
one
or
two
for
those
who
don't
know
I'm
critical
if
service
can
also
be
zero,
but
imagine
for
ass
for
our
specific
use
case.
We
want
just
to
have
one
or
two:
we
don't
allow
zero,
okay
and
any
other
protocol
specific
configuration.
A
So
the
SDK
will
read
this
part
of
the
file
and
will
validate
the
message
or
will
construct
the
message
depending
on
what's
defined
here.
So
imagine
that
we
are
trying
to
send
a
message
via
mqtt
with
quality
of
service
zero.
It
will
fail
because
we
know
that
it
can
only
get
one
or
two
right.
So
that's
that's!
That's
this
part.
A
If
you
notice,
this
is
a
specification
extension,
that's
why
it
starts
with
X
I'm,
doing
it
as
specification
extensions,
because
I
think
it's
the
easiest
the
easiest
way
now
we
might
think
about
standardizing
these
things
in
a
different
way
in
the
future
in
the
spec,
but
right
now,
I
think
it's
it's
fine
like
that,
and
and
that's
why
I
said
this,
they
are
like
special
type
of
specification
extension
right.
A
Yet
it
makes
sense,
yeah,
okay,
full
and
now
specification
extensions.
So
this
is.
This
is
very
important
because
I
know
I
know.
I
know
it
sounds
sometimes
like
it's
it's
it's
crazy.
How
are
you
going
to
parse
specification
extensions?
No
we're
not
been
a
parse
it,
but
we're
not
going
to
build
the
parsers
right.
So
what
we're
gonna
do
is,
for
instance,
mike
shared
with
me
an
example
of
an
open,
EDX
specification
extension,
which
is
ex
Twitter
and
in
the
info
object.
A
Imagine
that,
along
with
name
description
out
there,
we
want
to
have
the
tweeter
contact,
that's
cool,
but
it
needs
to
be
validated
right,
like
you
cannot
put
here
whatever
you
want
in
writing
that
we
have
X,
Twitter
and
file.
We
put
a
five
the
number
of
five
there.
It
doesn't
make
sense
right,
so
we
need
to
have
a
piece
of
software
that
will
validate
that.
A
A
What
I'm
waiting
for
here
is
that
at
some
point
we
will
have
a
directory
of
extensions
where
people
can
go
and
find
and
they
have
to
define
where
they've
been
headed,
that
it's
actually
allowed
to
have
the
extension
like
here,
for
instance,
the
the
only
place
you
can
have.
This
attention
is
no
object
and,
and
also
they
sue.
This
will
provide
a
decent
schema
document.
Saying
hey
this
is
this
piece
of
JSON
schema
is
actually
validating
what's
in
here,
so
you
cannot
put
whatever
you
want
there.
A
We
will
use
this
piece
of
JSON
schema
in
our
parser
at
the
end
of
the
whole
cycle
to
validate
that
the
specification
extension
is
there,
we
will
use
it
if
the
if
the
user
provides
us
this
piece
of
decent
schema.
If
not
it's
totally
fine
by
default
extensions
will
not
fail.
We
don't
care.
These
are
free
right.
You
can
do
whatever
you
want
so,
and
this
is
the
data
the
graphic
the
chart
for
for
the
parser.
A
A
We
don't
use
protocol
buffers.
Protocol
mappers
are
not
using
on
the
parser.
So
again,
this
first
one
is
the
one
in
charge
of
parsing
the
ml
or
Jason
the
input,
the
elongation.
The
second
one
is
a
schema
parser
or
if
you
want
this,
this
piece
here,
we'll
just
we'll
just
make
one
of
the
other
persons
below
like
Justin,
schema
proto
before
abre
to
take
care
of
the
message
person
of
the
schema
person.
A
Then
we
will
have
the
specification,
Stenson's
parser
and
then
once
we
have
the
whole
document
validated
and
built,
we
will
validate
it
with
the
de
Jason
schema
validator
right.
So
we
provide
a
body
JSON
schema
document,
and
this
JSON
schema
document
will
be
what
we
actually
will
provide
to
the
rest
of
the
tooling
and
the
rest
of
the
tooling
will
not
have
to
worry
about
prototype
abra
or
anything
they
will
just
use
JSON
schema.
So
all
this
JSON
schema
wrote
above
average.
A
They
will
need
to
convert
positions
here
to
a
specific
version
of
Jesus
scheme
right.
So
we
we
unify
everything
there.
It's
not
hard,
I've
been
looking
into
that,
and
protobuf
already
have
a
way
to
come
to
convert
libraries
to
convert
to
distant
schema.
I've
read
the
same.
It's
been
a
little
bit
harder
to
find
it
for
to
forego,
but
they
are
there
and
they're
not
complex.
If
we
have
to
read
it
and
go,
it's
gonna
be
fine.
A
If
we
don't
do
it,
this
is
the
parser.
So
imagine
that
we
have
imagine
that
we
have
a
code
generator.
The
code
will
have
to
breathe
the
whole
document
right
that
the
whole
async
API
document
and
from
there
it
will
have
to
generate
some
code
right.
That's
simple,
as
that,
if
we
don't
convert
to
the
schema
all
the
J's
or
the
the
code,
generators
all
the
other
tooling
as
well-
that
immediacy
generators
we
need
to
understand
protobuf,
we
need
to
understand
Avro.
We
need
to
understand
JSON
schema
the
open,
API
schemas.
A
A
A
But
all
of
them
have
in
common
that
they
need
the
partial
first
they
need
to
use
the
puzzle
first.
So
let's
get
rid
of
this
dependency
end
up
on
the
person
right.
We
like
code
generator,
documentation,
generator
SDKs.
Everything
we'll
have
to
use
the
person
first,
just
obvious.
Like
okay,
give
me
the
document
and
I
will
read
it
right.
A
Then,
let's,
let's
get
rid
of
this
difficulties
for
the
rest
of
the
people
on
the
partner
right,
then
we
provide
a
full
valid,
fully
valid
JSON
schema
document,
and
if
we
have
to
annotate
with
our
Alan's
extensions
that
the
JSON
schema
document
Scalf
rate,
it
means
fine.
If
we
imagine
that
there
are
some
something
some
things
in
the
impro
Toba
for
anavra
that
we
cannot
do
with
just
in
schema.
That's
going
to
happen.
If
that's
that's
the
case,
we
can.
A
But
in
that
case
what
happens
is
it
is
it?
This
is
like
a
funnel
right.
Every
piece
of
software
will
have
to
go
through
the
parser,
and
then
it
will
go
through
the
their
own
business
logic,
and
it's
not
because
we
want
I
mean
if
you
think
about
it,
but
the
code
generator
that
we
have
the
NORTHCOM
gentleman,
not
another
generator
package.
Note
again
has
a
partial,
built-in
kind
of
partial
documentation.
Generator
also
has
same
thing:
it's
actually.
A
This
copy
basil,
the
participation,
is
so
it's
there.
It's
it's
a
building
person
and
that's
why
I
created
the
generator
package
like
this
is
insane
I
have
the
the
parser
in
both
places
and
those
two
packages
are
doing
the
same
one
for
generating
code
and
the
other
one
for
generating
documentation,
but
in
the
end,
what
they
are
generating
our
files
independently.
What's
inside
the
file,
if
it's
cold
or
if
it's
HTML
or
Magda,
it's
the
same
right.
That's
why
I
unified
both,
but
then
I
realize
crap.
A
A
This
is
little
things
we're
making
we're,
making
me
repeat
code
across
all
the
to
me
like
this,
for
instance,
this
one
just
just
this
one
like
having
to
do
to
to
make
to
join,
to
join
that
at
the
base
topic
and
the
topic
right
and
and
if
you
have
to
do
things
like
validating
certain
things
like
like
authentication
or
authorization
methods,
same
thing,
you
don't
want
to
go
through
this
right.
This
is
this
is
a
tool
for
generating
documentation,
not
for
parsing
right.
A
So,
yes,
that's.
Why
I
like
to
extract
the
parser
and
not
just
for
documentation,
generators
and
for
code
generators,
but
for
every
tool
and
because
it
has
to
be
for
every
tool
we
have
to
make
it
compatible
with
everything
right
and
we
have
the
opportunity
to
simplify
the
lives
of
the
developers
on
on
the
tool.
A
It
doesn't
mean
if
you
are
in
the
mood
of
creating
your
own
parser
go
for
it,
we
may
we
I
mean
we
may
offer
as
well
a
way
to
hook
into
the
person.
That
will
be
also
amazing,
but
that's
a
version
two
right
there
on
three
four,
even
for
like
once
we
have
the
person
we
may
have
the
ability
or
we
can
may
offer
the
ability
to
book
and
whenever
the
parser
is
going
to
validate
I,
don't
know
security
schemes
or
something
like
that.
Then
it's
like
Nana.
This
is
my
repo
responsibility.
A
They
will
take
care
of
it
things
like
that.
That's
fine,
that's
fine
or
it's
possible.
So
the
that's.
Why
I
want
to
put
a
lot
of
effort
in
the
parson
because
I
I
realized?
This
is
common
to
all
the
tooling,
obviously,
because
all
the
tooling
will
have
to
parse
the
document
and
to
understand
what's
inside
right
doesn't
make
sense,
cool,
so
yeah
I'm
not
going
to
I'm
going
to
go
through
all
these
examples.
A
A
We
would
want
to
go
through
creating
a
generating
framework,
so
this
is
most
probably
I
was
no
I
was
going
to
say
this
is
probably
going
to
be
like
swagger,
coach
gene
or
something
like
that.
Price
Hill
has
been
looking
into,
but
we
don't
know
yet.
So
that's
why
I
I
didn't
want
to
and
took
right
there,
that
we're
gonna
use,
swagger,
coach
and
or
sell
or
tweak
so
I
got
caught
Jen,
but
we
may
grab
some
some
approaches.
They
did
instead
of
codes
approaches.
A
A
How
does
it
call
price
I'll
help
me
with
that?
I?
Don't
remember
the
name
of
Java
this
spring
/
right,
yeah
spring
12
strings.
If
you're
used
to
spring
cloud
streams,
whatever
right
then
you're
in
and
you
love
it,
you
know
how
it
works.
You
may
want
to
build
your
code
using
spring
cloud
strings
and
not
our
SDK
right.
It
makes
total
sense,
but,
as
the
case
building
an
SDK
is
actually
for
a
different
purpose.
Is
the
purpose
is
to
to
e
e.
A
He
is
the
principal.
How
can
I,
how
can
explain
the
onboarding
of
people
into
into
inventories
and
systems
because
currently
is
very
hard
for
people
to
get
into
infantrymen
because
of
the
multiple
systems
out
there
like
mqtt
and
QB,
if
Nats,
for
instance,
different
languages,
different
frameworks,
different
ways
of
working
right,
while
HTTP
is
much
simpler
in
this
case,
because
you
have
a
lot
of
a
lot
of
unified
Julian
and
there
and
everything
is
like
unified
into
HTTP
protocol
area.
It's
fine!
A
And
the
cool
idea
will
be
to
create
an
SDK
that
allows
you
to
send
and
receive
messages
from
a
cell
from
a
server
or
broker,
and
you
don't
have
to
know
if
the
broker
is
Kafka,
if
it's
MQTT
Nats
and
could
be
whatever
so
your
code
will
remain
the
same
and
you
only
have
to
go
to
a
config
file
and
say
hey.
Actually,
this
broker
here
is
Kafka.
A
You
saw
you
get
into
partition
number
two,
whatever
I'm
just
inventing
the
details
right
and
if,
at
some
point
you
decide
to
change
Kafka
for
mqp
or
I,
think
you
just
change
the
config
file,
but
the
code
still
works
right.
That
will
be
like
to
me
that
will
be
like
a
Mason
like
ok,
so
the
coding
will
be
the
same
right.
Everything
is
obstructed
in
terms
of
code,
nothing
in
terms
of
configuration.
A
A
This
is
kind
of
a
jump
here
in
this
cell,
because
that's
and
that's
why
I've
been
putting
all
the
dependency
section
here
on
every
one
on
every
on
every
milestone,
so
to
make
to
make
it
explicit
that
this
one,
for
instance,
it's
not
blocked
by
by
the
previous
one.
They
don't
have
to
happen.
You
know
all
every
one
after
the
other
right.
A
A
Yeah,
most
probably
will
have
to
work
on
both,
so
you
can
generate
react,
HTML
and
markdown,
and
they
also
so
behave
more
or
less
the
same
right
with
their
own
peculiarities
and
in
an
ideal
world,
as
I
was
saying,
we
probably
did
the
current
editor
in
favor
of
the
react,
playground
that
that
Lucas
and
the
chemo
team
has
been
developing
and
that
it
needs
its
to
need.
Some
improvements
like
it
needs
to
support,
for
instance,
choosing
between
react,
HTML
and
mardow,
and
some
other
things.
We
will
go
into
details
on
that.
C
D
A
D
Kind
of
layer
and
then
you've
got
layered
approach
with
multiple
parts
of
the
parser
I.
Think
it's
really
good,
and
it
should
avoid
the
kind
of
duplication
and
the
fragmentation
you
get
in
tools
around
open,
API
yeah,
because
I
think,
even
if
somebody
reimplemented
the
the
entire
parser
they're,
probably
going
to
implement
it
along
the
same
lines
of
the
roadmap
yeah.
So
we're
kind
of
guiding
the
way
tools
should
be
built
as
well
as
well
as
well
as
building
a
reference
implementation.
D
A
C
A
C
A
D
It's
a
slightly
long
story,
but
yeah,
it's
kind
of
it
was
erased
from
existence
yeah.
My
other
quick
question
was:
when
you
talk
about
beautifying
the
past
version
of
respect.
Are
we
going
to
standardize
of
what
the
beautified
version
should
look
like,
so
would
all
pass
as
beautiful
a
respect
in
the
same
way
at.
A
Least,
I
mean
yes,
so
at
least
at
least
our
parser,
so
beautifying
in
a
specific
way
and
by
beautifying
thinking
a
lot,
because
it's
it's
always
a
mess
about
making
join
between
base
channel
and
the
channel.
This
is
always
a
mess.
You
know
we
always
have
to
run
into
this,
and
but
we
might
also
want
to
have
on
the
root
level.
We
might
also
want
to
have
a
list
of
messages.
A
These
disposing
in
a
different
way
like
instead
of
channel
and
then
publish,
and
then
the
message
we
might
want
to
have
just
a
list
of
all
messages
with
some
information
about
the
topics
in
which
this
message,
these
messages
use
just
to
ease
the
the
the
the
the
the
development
on
the
other
side
like,
for
instance,
you
you
might
want
to
have
a
list
of
topics
or
sorry
of
messages.
We
are
topics
or
anything
on
your
documentation.
A
You
just
iterate
further
on
this
list
right
instead
of
going
on
every
topic,
then
on
every
operation
and
checking
if
it's,
if
it
has
a
message
on
publishing
and
subscribe,
you
know
so,
but
but
answering
your
question
is
I
think
we
should
at
least
offer
minimum
standardization
of
what
our
parser
offers.
Okay.
D
A
A
No
thinking
so
much
under
on
the
one
that
that
we
have
on
generator
right
now.
This
is
I
think
this
is
too
specific.
Just
a
minimum
thing
like
yeah
like
like
generating
a
collection
of
messages
or
color,
or
a
collection
of
topics
or
collection
or
of
wherever
things.
So
you
don't
have
to
traverse
the
whole
tree
every
time,
right,
just
basic
stuff
and
and
and
and
join
in
base
topic
and
the
topic.
A
But
we
may
we
may
want
to
define
there
whenever
we
start
working
on
that,
we
may
want
to
define
exactly
how
it
should
be
beautified
and
defining
what
should
be
or
what's
required
like.
If
you
create
your
own
parser,
it
should
follow
at
least
this
like
it
so
habit
with
this
beautiful
things
right,
fine.
A
This
is
funny
a
funny
thing
that
I
found
an
internet
but
I
completely
agree
with
this
bill.
This
way
of
building
products-
and
this
is
a
product.
So
this
is
exactly
what
I
want
to
achieve.
That's
why
I
want
to
put
a
lot
of
effort
in
building
a
nice
parser
and
then
starting
building
nice
things
useful
things,
but
slowly,
but
they
also
be
useful
right,
even
though
we
might
not
have
the
car
really
quickly,
but
in
the
meantime
we
will.
A
We
will
offer
bicycles
right
so
at
least
there's
something
and
we
keep
building
and
keep
building
and
iterating
on
that
instead
of
the
other
approach,
because
what
most
probably
will
happen
if
we've
followed?
The
first
approach
is
that
requirements
might
change
during
the
way
or-
and
we
find
that
this
is
we
find
that
the
requirements
we
had
in
the
beginning
or
not.
A
What
do
we
thought
and
we
have
to
change
things
and
suddenly
it's
math
people
are
under
on
the
project
and
the
code,
and
things
are
not
working
well,
and
this
is
and
we
we
can,
we
can
start
working
on
that
and
and
I
don't
know
and
have
a
year
of
work
without
having
anything
really
ready
for
any
one
to
two
years.
That's
not
decidable.
We
want
to
have
feedback
as
soon
as
possible.
Right
we
went
from
step
one.
We
want
to
start
having
feedback
and
saying
hey
guys.
This
is
very
this.
Is
this
parcel?
A
C
B
A
Already
and
and
one
thing
that
I
would
like
to
to
spread
the
word
for
you
to
spread.
The
word
is
we
people
do
to
start
creating
content
like
blog,
post
and
and
and
the
commentation
tutorials
know
that
I'll
be
working
on
that,
but
I'm,
not
a
technical
writer,
so,
and
so
for
you,
the
native
English
people,
if,
if
you've,
if
you
notice
something
that
something's
wrong
with
some
language
that
I'm
using
or
whatever
just
let
me
know
because
I
mean
don't
feel
bad
of
telling
me
hey,
this
is
this
is
not
correct.
A
A
A
Yeah,
so
this
is
something
that
that
we
definitely
need.
We
need
to
do
it
more
people
to
create
blockers
and
on
our
blog
and
also
in
another
vlogs
and
I'll,
be
happy
to
do
to
join
two
other
companies
or
other
blogs
to
to
write
a
joint
blog
post.
So
and
as
well
as
you
and
feel
free
to
write
about
it
and
don't
wait
for
me.
Well.