►
From YouTube: AsyncAPI SIG meeting 11 (July 9, 2019)
Description
This is the recording for the AsyncAPI Special Interest Group (SIG) meeting #11.
Attendees:
- Abel Cal
- Andrew Snodgrass
- Fran Méndez
- Łukasz Gornicki
- Marc DiPasquale
- Raisel Melián
- Rubén Hervás
Moderation:
- Fran Méndez
Agenda:
- Contributions from Kyma (Converter and REST service) — Łukasz Gornicki
- WebAssembly PoC — Łukasz Gornicki
- Issues review
- Q&A
B
Yeah
sure
I
will
sure
right.
That's
gonna
be
the
best
I
guess.
B
B
So,
the
all
the
details
are
in
this
issue,
but
I
think
the
most
important
is
what
marching
from
our
kima
project.
Did
he
created
a
really
simple
example:
repositories
also
linked
into
the
into
the
issue,
what
he
did.
He
simply
provided
really
simple
instructions:
how
to
set
up
an
engineer's
hosting
the
the
web
assembly
file,
and
then
you
use
it
to
showcase
that.
B
It's
really
working
that
you
can
really
parse
and
VPI
spec
in
the
UI,
so
just
to
show
you
quickly,
I
have
it
running
with
docker
and
the
the
main
outcome
of
the
POC
is
that
go
is
not
much
enough.
The
the
default
go.
P--
compiler
compiles
into
really
big
size
packages,
so
the
the
go
parser
for
us,
an
api,
the
size
of
the
of
the
compiled
file,
a
wasn't
vial
is
14
megabytes,
and
even
if
you
enable
like
in
this
showcase,
we
have
a
gzipping
enabled
on
the
server
side
for
nginx.
B
A
B
For
us,
it's
super
cool
because
we
can,
with
with
this
pop,
we
proofed
it
works.
So
we
can
start
working
on
the
on
the
react
component
without
waiting
for
a
for
other
parser,
so
it
unlocks
the
work
we
can
yeah.
We
can
do
it,
but
long
term
yeah
I
would
not
say
go
is
a
good
solution
for
now
for
using
my
blossomy
yeah.
A
So
just
so
you
know
I'm
hoping
you
have
the
parser
in
native
JavaScript
working
for
the
browser
and
from
Algiers
I'm
hoping
to
have
it
like
I,
don't
know
like
probably
end
of
this
month
or
something
like
that.
Cuz,
it's
really
people.
You
know
it's
well,
I
mean
we.
We
are
the
lead
with
version
one
where
they
do
it
like
this.
It's
just
changing
this
schema
to
version
two
release
candidate,
one
and
and
basically,
since
there
is
nothing
else,
I
mean
we
can
keep
improving
it.
A
But
since
it's
JavaScript
and
it
doesn't
have
types
so
parsing
is
basically
the
validator.
Nothing
else
is
just
and
given
it's
giving
you
the
whole
raw
JSON
object
right.
So
the
parsing
is
really
easy.
You
didn't
have
to
match
the
Jason
structures
to
data
types
in
Java,
Script
right
because
there
are
no
data
types.
A
So
it's
it's
really
easy,
so
I'm
hoping
to
have
it
ready
for
the
for
the
end
of
the
month
for
sure
after
the
conference
or
beginning
of
the
next
month,
because
left
next
month,
because
I'm
taking
holidays
right
after
a
day
there
San
Francisco,
which
is
next
week
and
so
yeah,
but
it
shouldn't
be
hard.
It
shouldn't
take
long
to
be
honest.
I
actually
have
a
proof
of
concept
of
on
my
machine
now
and
with
JavaScript,
and
it
works
perfectly
in
browser.
I
mean
OGS
I'm,
just
not
happy
with
the
code
quality,
but
it
works.
A
No,
no
just
the
basics
without
proto
power.
For
that
for
support.
Just
you
know
the
basic
suggestion
schema
and
that's
it,
which
is
enough
to
unblock
your
work
right
as
well
so
and
for
most
of
the
cases,
is
gonna,
be
enough.
So
well
for
I.
Forget
thanks
for
this
proof
of
concept
with
WebP
assembly,
because
that
proves
that
I
mean
it's
cool.
It's
nice
and
it
may
be
there
in
the
future.
A
But
it's
not
there
yet
I!
Guess
the
problem
ISM
or
the
hard
work
is
always
on
the
side
on
the
go
compiler
right,
the
faucet,
your
compiler
might
have
to
work
harder
into
building
smaller,
smaller
web
assembly
files
and
C++
files
as
well
so
yeah.
So
rice
l
suggested
to
why?
Don't
we?
Why
don't
we
use
frost
right?
C
A
C
C
A
A
You
know
and
put
it
into
the
into
the
spec
and
go
it's
being
already
a
problem,
like
he's
been
doing
a
proof
of
concept,
but
it's
really
three
key
and
in
JavaScript
you
already
have
it
you
have
you
really
have
libraries
that
test
that
right,
but
and
then
for
instance,
you
have
libraries
that
may
convert
from
Avro
I
if
I
remember
correctly
from
Avro
to
JSON
schema
or
getaway
around
I.
Don't
remember
so
there's
this
there's
something
in
go
please
or
something
right,
but
I
don't
know
about
the
rest.
A
I
mean
that
the
thing
whenever
I
try
to
find
something
we
dressed
like
support
for
proto
buffers
or
for
a
draw
or
for
JSON
schema.
For
instance,
the
support
is
really
you
know
it's
in
early
days
or
in
existence.
Basically,
so
it's
hard
to
is
happen.
It's
not!
For
that
reason
we
cool
I
mean
anyone
could
argue
that
if
we
created
for
this
project,
then
we
contributed
to
communities
right.
So
we
contribute
to
have
it
basically,
but
that's
like
a
huge
word
around.
That
will
be
amazing
over
the
over
time.
A
That
will
be
amazing,
but
speaking
the
other
day
with
don't
remember
who
what
I
was
speaking,
someone
think
about
the
parser
and
and
I
realized
this
another
problem.
We
haven't
think
I
think
it
was
with
you
a
bit
with
the
dis
another
pro
with
the
person
and
it's
like
if
we're
gonna
make
it
pluggable.
So
we
want
to
plug
a
parcel
for
protobuf
and
for
extensions
and
all
this
stuff.
A
If
you
cried
this
tension,
for
instance,
in
JavaScript
and
the
parser
is
in
go,
how
do
you
use
a
JavaScript
extension
with
a
cold
person
or
I
mean
one
cool
cool?
Say
it
yeah,
but
you
use
the
JavaScript,
but
you
have
a
script
partially,
not
to
go
perfect,
but
you
may
be
using
the
person
in
your
go
code,
so
you
can
do
that.
You
cannot
do
it
so
that
takes
us
to
a
point
that
may
be.
A
If
we
want
to
make
things
pluggable,
we
need
to
have
eight
parts
that
are
only
on
each
language
right,
so
you'll
get
what
I'm
saying
like
we
understand
or
am
I
just
throwing
ideas
and
doesn't
make
sense.
If
someone
doesn't
understand
what
I'm
talking
about
just
please
raise
your
hand
I'm
happy
to
okay,
so
I
would
say
so
before
I
continue
look
at.
What
is
that
that
you
don't
understand?
I
was.
B
A
So
how
do
you
plug
it
to
a
another
type
of
language?
You
know
what
I
mean
so
imagine
if
you
have
to
passing
it
all
in
a
call
to
the
parse
method
you
want
to
pass
at
the
at
the
end
of
the
of
the
method
you
want
to
pass.
An
array
of
parsers
of
extension,
bar
tips,
right
or
extension,
validator
city
wants.
Not
part
is.
A
Yeah,
but
because
the
thing
is
the
the
support
for
proto
buffers
Avro,
you
know
another
and
another
ski
mask
schema
languages
like
xst
for
xml.
The
idea
is
to
make
all
of
them
pluggable
because
there
are
other.
There
are
other
internal
and
external
as
well.
Languages
are
schema
formats
that
we're
not
going
to
support
natively
in
each
language
right,
because
that
will
be
insane
so
imagine
that,
for
instance,
I
use
the
example
of
flatbuffers,
for
instance.
Flatbuffers
is
it's
really
it's
highly
adopted
in
in
the
game
industry
right?
A
So
what
happens
is
if
we
don't
add
support
for
flatbuffers?
They
have
to
create
a
flatbuffers
message.
Parser,
like
the
idea,
the
person
was
there
like.
Ok,
you
create
a
flat
buffer
message,
parser
and
you
plug
it
into
our
pattern.
But
if
someone
wrote
it
on
JavaScript
the
flatbuffers
parser,
you
cannot
plug
it
into
the
go
parser
or
into
the
java
parser
into
the
Python
person.
You
can
only
plug
it
into
the
JavaScript
parser
and
you
may
not
be
running
your
code
in
JavaScript.
You
may
be
running
it
in
Java
or
Python
or
whatever.
A
A
To
be
honest,
it's
a
hard
decision,
because
in
any
case,
what
I'm
seeing
here
is
that
we
can
probably
not
make
the
parser
tribal,
maybe
in
extensible
so,
which
is
I,
mean
it's
fine,
but
it
loses
a
lot
of
power.
Right,
like
flex,
is
the
the
first.
The
first
idea
of
the
of
the
parser
was
to
make
it
playable,
so
it's
easily,
or
maybe
we
can
help
other
people.
I,
don't
know,
write
their
extensions
in
in
a
language
that
can
be
compiled
down
to
other
languages
like
like
over
interests
or
something
else.
Oh.
B
Yeah,
maybe
I
just
simply
didn't
get
it
at
first,
because
for
me,
it's
I
would
not
consider
this
a
concern
right
because
at
the
end,
if
you
just
have
let's
say,
go
parser
and
JavaScript
parser
and
they're
all
prepared
in
the
way
that
you
can
plug
in
your
custom
code
into
it.
And
then
somebody
is
having
everything
in
Java
and
they
have
some
parser
for
some
specific
protocol
written
in
Java,
then
I
mean
what
else
can
you
do?
You
simply
have
to
tell
them
write
it
in
the.
A
Yeah,
no,
no,
no
I'm,
just
trying
to
I'm
putting
all
the
examples.
You
know
all
the
situations
in
areas,
so
we
pick
one
you
know
and
we
make
sure
that
we
pick
one
wisely
before
continuing
work
on
that's
right,
so
yeah
I,
guess
maybe
in
that
case
I'm
just
thinking
about,
maybe
just
in
this
case
you
have
to
just
yeah.
So
the
part
that
is
pluggable,
but
you
only
can
plug
code
in
your
own
language.
A
So
obviously
right
and
not
all
the
extensions
are
gonna
be
available
in
all
languages
like
like
the
parse
it
right,
okay,
okay!
So
in
any
case,
coming
back
to
the
to
the
topic
of
using
breast
yeah,
you
know
problem
is
it
was
slow
down
development,
starting
because
starting
from
that,
the
fact
that
none
of
us
know
rest-
but
some
of
us
do
know
ego
so
so
yeah
I'll
stick
to
go
and
just
create
a
separate
parcel
for
JavaScript,
which
is
a
like
a
special
case
because
of
the
browser
right
over
time.
B
Is
already
a
tiny
got
any
goal
project
which
you
say,
yeah
I,
said
a
different
compiler
that
community
works
on
to
make
go
embeddable
in
really
miss
tiny
devices?
Yes,
but
it's
yeah
and
we
tried
it
with
this:
go
parser
for
passing
API,
so
the
generated
by
bus
and
if
I
did
not
fully
work
work,
because
not
everything
is
yet
fully
supported
there,
but
maybe
even
in
few
months
and
the
story
will
be
totally
different
with
what
I
see
photo.
A
B
So
we
it's
about
the
converter
and
they,
sir,
so
just
I
wanted
to
align
how
we
should
proceed
with
that.
So
we
for
our
use
case
for
kima.
We
need
not
only
we're
not
only
building
front
end,
but
also
we
lay
back
end
for
that
will
validate
this
pack
and
convert
it
for
the
customers
from
old
version
into
the
new
one
and
the
whole
stack
for
us.
We,
it
always
is
go
length,
so
we
need
to
provide
a.
B
A
sink
API
react
component
into
the
a
sink
FG
organization,
but
it's
up
to
you
to
decide
if
you
prefer
it
that
we
contribute
to
your
organization
or
we
keep
adding
kima,
because
if
we
will
keep
it
in
kima,
it
means
it's
gonna
end
up
in
our
huge
mono,
repo
somewhere
hidden
between
the
components.
So
nobody
know.
A
E
F
B
A
D
Well,
my
bad
I
mean
from
a
date
from
Spain
and
well
I
met
with
GU
band
and
he
contact
with
travel
and
it's
this
project.
I
spent
the
last
few
years
from
defending
the
sea
cappy's
api's,
because
I
was
working
for
a
media
company
and
the
only
way
that
I
phone
to
solve
some
issues.
So
big
issues
was
the
in
the
same
way
and
really
excited
about
this
display,
because
I
think
it
has
a
very
long
want
to
vary.
It
will
be
very
interested
way
to
solve
most
of
the
most
issues.
D
D
Deployer
will
implemented
with
the
async
ABA
specs
and
I'm
just
for
now,
I'm,
just
playing
with
the
the
edit
of
the
line
editor
and
try
and
I'm
using
an
example
about
quarantine
and
I'm.
Trying
to
to
make
it
work
with
the
front
on
the
front
that
I'm
doing
with
angular
is
the
most
the
way.
That
is
the
language
that
I
really
know
so.
D
A
A
D
A
D
A
Yeah,
so
for
those
of
you
who
don't
know
I
started
another
meeting,
that's
why
there
was
a
lot
of
confusion
and
in
the
beginning,
which
is
for
people
who
want
to
regularly
contribute
to
a
sink
API,
because
this
one
is
public
meeting
for
people
who
just
want
to
attend,
maybe
ask
some
questions
and
because
they
have
some,
they
have
interest
and
I
created
another
one
for
contributors.
So
we
keep
at
we
first
of
all,
we
know
each
other
and
we
can
share
the
progress
or
questions
about
the
development
of
certain
features
right.
A
This
one
is
gonna,
be
about
development.
It's
not
gonna,
be
about
like
this
one,
which
is
usually
to
expose
to
the
public
or
to
answer
questions
to
people
that
might
join
and
don't
join
anymore
because
they
don't
really
care
so
much
about
the
development
of
the
project.
They're,
just
curious
right,
so
so
yeah
just
so
you
know,
and
I
was
thinking
I'm
doing
something
easy
to
review,
but.
A
A
A
G
D
A
B
A
B
One
ticket
and
the
other
one
I
had
was
the
failing
in
the
in
the
generator
repo.
There
was
a
issue
files
that
you
cannot
generate
Park
down.
If
you
have
references
to
files
and
I
only
made
it
made
it
possible
to
kind
of
set
up
an
environment
where
I
can
reproduce
it.
Okay,
but
didn't
worry
yet
work
on
the
fix,
because
I
first
wanted
to
set
up
a
goals
for
this
converter
and
II
and
the
service.
That's
what
I
continued
working
on
it.
A
F
E
A
A
A
A
Yeah,
so
maybe
if
you
want
to
share
just
push
this
go
to
a
branch,
so
alright,
it
start
I
mean
helping,
maybe
not
just
reviewing,
but
maybe
we
can
just
synchronize
and
say:
hey
I
have
some
fixes
for
this
case
this
case
in
this
application,
and
we
can.
We
can
all
contribute
right
or
maybe
just
create
a
draft
pull
request.
Now
that
think
it
happens.
Yeah
after
this
yeah
clear
that
this
is
a
draft
that
this
is.
This
is
just
a
key
code.
Maybe
I
know.
E
E
F
A
A
A
D
E
E
F
Know
if
you
remember
that
we
talked
about
the
reference
to
a
local
file,
but
the
file
was
a
yeah
man
file,
not
a
gem
file
yeah,
because
that
was
that's
one
issue.
Maybe
yeah.
A
A
A
A
The
thing
is
that
when
you
have
a
dollar
red
and
then
pointing
to
a
local
file
or
a
remote
file
wherever
and
it's
a
general
one,
the
one
that's
in
the
draft,
the
question
is:
will
it
work
or
not
because
it
might
seem
if
your
documents
in
general
and
reference
is
another
llamo?
The
feeling
is
like
it
will
work
right.
It's
referencing
another
piece
of
yellow
that
wood
will
be
converted
to
Jason,
and
then
you
know
and
evaluate
it
at
so
as
we
do
with
the
documents
but
they're
thinking
about
it.
A
This,
like
I,
don't
know
because
we
have
these
schema
formats
in
in
person
too,
and
with
respecting.
We
are
expecting
JSON
schema,
JSON
schema,
stuff,
Jana
is
Jason,
so
I
don't
know.
Maybe
we
can.
We
sue
allow
this,
but
only
if
you
specify
in
the
schema
format
that
you
suspect
in
JSON
schema
plus
yeah
Mel.
You
know
in
the
in
the
in
the
Mima
type
that
you
can
specify
with
a
plus.
You
can
specify
that
the
content
type
or
something
like
this
I
don't
know,
doesn't
make
sense.
What
I?
A
B
A
That's
a
work
even
even
though,
if
you
used
y'know
internally.
What
happens
is
that
is
that
we
used
a
Sarang
this
peg.
It
says
that
you
can
use
Gemmell,
but
only
the
subset
that
can
be
converted
to
Jason.
So
it's
like
a
nice.
It's
just
like
it's
sugar.
You
know
syntax
sugar
for
for
async
API,
but
the
real,
let's
say
official
format
of
facing
API
is
not
the
a
negotiation
just
use
yellow
for
ease
of
reading
for
people
right
for
humans.
A
B
D
D
A
E
A
A
A
If
we,
there
are
two
cases,
we
either
make
the
the
parser,
detect
the
format
and
converted
to
Jason
to
matically,
or
we
make
people
have
to
specify
this.
Otherwise
it
won't.
It
will
not
work.
So
this
one
is
like
more
I,
don't
know
how
to
say
it's
a
it's
more
like
strict
right,
unless
you
specify
it
correctly
right,
they
do
not
work.
Okay,
so.
A
D
Well,
I
think
that
people
expect
that
the
shaman
is
a
superset
of
Jesus.
So
if
it
works
with
the
past
works
with
Jamal
its
we
work
with
Jason.
So
if
we
have
the
capability
to
with
Jana
I,
think
Jason
is
already
done
so
I
see
that
the
property
of
scheme
format
is
not
needed.
If
you
are
going
to
refer,
make.
A
Sense,
okay,
I'm,
just
curious
I
mean
I,
have
my
own
opinion
on
that
and
I
think
I.
Think
the
saying
I
think
that
we
should
make
it
automatic
instead
of
explicit,
but
then
on
the
other
side,
I
have
the
you
know,
feeling
that
this
is
Peck.
Everything
should
be
specified
right.
Every
case
will
be
specified
and
I
mean
we
can
put
this
specification
as
well.
If
your,
if
your
reference
of
the
file
is
a
yellow
parsers
will
convert
it
to
Jason
whatever
we
have
it
in
the
spec
and
that's
it.
D
A
So
in
this
case
you
have
you
know
you
have
to
specify
this.
Otherwise,
this
format
can
be
anything
right.
We
cannot
be
magical
out
of
detecting
because
there
there
may
be
the
case
where
two
formats
are
compatible
between
each
other
and
two
files
may
look
like
to
four
different
formats
and
you
don't
know
which
one
to
choose
like
I'm
thinking,
for
instance
a
maverick
Avro-
and
this
is
kima-
some
pieces
are
very
similar.
Both
are
Jason.
A
E
G
A
G
A
F
A
A
A
Do
you
look
at
that
header,
this
cool,
being
fair
from
the
in
the
case
of
HTTP,
the
school
being
fair
from
the
mime
type,
but
in
the
case
of
a
file,
for
instance,
the
local
file?
You
don't
have
this
right.
Imagine
that
you
have
my
system
wherever
my
stuff
and
then
you
have
I,
don't
know
file
damu
or
you
just
have
file,
and
this
file
is
a
yellow
file
to
work
in
a
way
right
by
sniffing
the
content
and
trying
to
parse
the
content
and
if
it
works
with
yellow.
We
assume
this
is
channel
right.
A
E
A
With
just
explode
on
exactly
so
currently
what
we're
doing
is
we
try
to
parse
the
the
content
destination
if
it
doesn't
work
within,
try
sgml?
If
it
doesn't
work,
it's
mediation
again,
and
so
we
fail.
We
would
rest
never
but
the
more
stuff
that
we
add
or
not
the
more
stuff,
because
this
is
only
for
the
case
of
the
default,
where
we
don't
have
the
schema
format
right.
We
only
consider
those
two
so
thinking
I,
don't
know.
Maybe.
A
A
So
this
is
the
idea.
The
idea
of
this
is
it
the
default,
which
is
like
this
without
schema
format
right,
the
default
will
will
only
support
JSON
and
Gamble,
and
if
you
want
to
support
something
else,
you
need
to
split
eclis
specify
the
schema
for
me.
Otherwise
it
won't
work,
but
we're
talking
now
about
the
default,
like
you
don't
specify
anything.
What
do
we
assume
is
this
Jason?
This
is
Yemen.
A
Is
it
we
don't
care,
you
know
it's.
We
tried
both
there's
a
by
the
way.
There
is
another
I'm
considering
another
change
in
spec
for
the
final
version,
which
is
something
like
this
right
instead
of
schema
format.
You'll
have
something
like
this,
so
this
becomes
mandatory
all
the
time
the
schema
for
one
becomes
mandatory.
All
the
time
you
cannot,
we
will
not
have
a
default
because
you
have
to
put
it.
Why
is
this?
Because
some
people
are
asking
for
multiple
formats
like
this
I
want
to
have
the
imagine
I
want
to
have
my
I
won't.
E
A
My
message
defined
in
different
schema
and
xst,
and
one
will
think
why
you
want
to
define
it
in
two
different
formats
right.
That's
what
I
thought
in
the
beginning
and
that's
why
I
chose
for
the
the
other
way
of
specifying
it.
But
then
someone
told
me
you
know
what's
what
happens
in
old
in
old
applications
with
legacy
code?
A
A
A
A
It
might
not
be
possible
because
the
channel
invariants
Kefka
topic-
you
only
have
one
Kafka
topic
right
and
you
you
may
receive
the
the
the
data
adjacent,
but
you
have
it
defined
in
orange,
acting
in
XML,
but
you
have
a
defining
both
XML
and
and
XS
d
and
is
a
schema.
So
this
is
again.
This
is
not
the
content
of
the
message.
This
is
the
definition
of
the
content,
so
you
can
define
JSON
using
x,
SD
and
you
can
define
XML
using
JSON
schema.
A
A
This
is
I,
think
we're
entering
like.
You
know,
dark
areas
where
we
may
be
complicating
things,
a
lot
for
methan
or
for
some
edge
cases
that
some
people
have
I
only
had
about
this
one's
from
someone
not
twice
so
maybe
it's
just
a
niche
case.
This
person
is
having
and
that's
it
and
then
there
are
workarounds
for
this
as
well.
Like
I
said,
you
can
have
two
files
to
a
sink
a
very
fast
while
you
are
migrating
legacy,
kill
to
to
new
code.