►
From YouTube: Node.js N-API Team meeting - May 17 2018
Description
A
A
Got
okay,
so
here's
our
milestones,
so
we
can
walk
through
so
block
to
consign
coincide
was
six
and
eight
back
port
maroonish
and
I
did
get
together
to
discuss
that.
We
have
us
a
plan
to
get
together
in
another
think
like
not
this
Monday,
but
the
Monday
after
and
I'm,
hoping
to
start
to
write
a
little
bit
of
that
today.
So
I
think
we're
on
track
for
that
anything
to
orange
died
on
that
one
I
know.
D
A
Okay,
I
guess.
The
next
issue,
then,
is
creating
the
workshop
for
notes,
but
I
know
Jim
you
you
have
a
first
cut.
Don't
let
you
talk
about
that.
D
E
So
yeah,
so
that
what
I'm
assuming
here
is
the
person
has
experience
with
C
and
C++,
but
maybe
not
with
with
node,
particularly
building
native
add-ons,
so
I
do
list
the
prerequisites
and
the
tools.
So,
if
you're,
starting
like
from
ground
zero,
you
should
you
should
be
able
to
work
through
the
tutorial.
E
A
F
A
E
So
it
has
you
download
the
yeoman
and
then
the
generator
module,
and
then
it
just
takes
you
through
step
by
step,
generating
the
project
and
then
once
the
project
is
generated
right
here.
It
shows
you,
it
shows
you
the
how
the
project
is
laid
out
when
it's
generated,
and
then
it
takes
you
through
each
of
the
files
that
are
generated
and
explains
what
what's,
the
purpose
of
the
file
is
and
what
the
contents
of
those
files
are.
E
G
Doing
I
just
noticed
one
thing
in
the
in
the
tutorial,
though:
the
migration
tutorial
you're
using
STD
:
:
to
string
that
may
not
work
for
people
using
OS
10
because
I
know.
That's
that's
an
issue
for
some
reason:
the
standard
template
library
over
there
is
deficient
somehow,
okay,
so
you
you
might
want
to
just
use
a
stray
or
STR
error.
That's
like
it's!
It
that's
in
the
standard
c
library
and
it
converts
the
urna
number
to
to
a
string
and
then
string
is
that.
E
B
E
This
one,
you
know,
I,
did
a
survey
of
modules
that
were
out
there,
and
this
one
called
microtime
seemed
to
be
the
just
the
right
size
for
what
we
wanted
to
do.
I
looked
at
other
modules
to
convert
and
that
you
get
I
was
getting
bogged
down
and
additional
complexities
that
were
beyond
the
the
Nappy
concerns.
E
So
this
is
a
really
good
one,
because
the
conversion
tool
runs
and
then
you
try
to
build
it
and
it
fails
and
it
fails
in
ways
that
are
very
common
to
what
people
are
gonna
run
into
when
they
try
to
convert
a
module
so
shows
so
what
I?
What
I
have
them
do
is
clone
the
project,
and
the
first
thing
I
have
them
do
is:
is
build
and
run
the
project
the
way
it
is
verify.
E
Okay,
this
is
how
it's
supposed
to
work
and
then
I
have
have
them,
run
the
conversion
tool
and
then
try
to
build
again
and
then
what
you
have
is
you
have
of
issues
that
need
to
be
corrected
and
so
I
list
each
issue
that
you're
gonna
run
into
describe?
Why
you're
having
it
and
then
what
you
need
to
do
to
fix
it
yeah.
E
E
Exactly
and
it
kind
of
gives
you
an
idea,
okay
of
what
the
process
is,
and
then
the
sort
of
issues
that
you're
gonna
run
into
and
then
down
at
the
bottom
I
link
to
our
documentation.
Saying
okay,
when
other
issues
come
up
here
are
some
really
good
resources.
You
can
use
to
help
understand
what
you
need
to
do
to
for
Europe,
for
your
particular
needs.
A
A
I
think
my
other
thinking
is,
like
you
know,
sort
of
the
the
migration
is
sort
of
like
a
separate
topic.
Almost
like
you
know,
here's
migration
in
terms
of
the
sort
of
port
you
know
creating
your
own
module
I
can
see
the
first
one
you
have
is
a
really
good.
Like
you
know,
I
would
almost
call
it.
Sometimes
it's
like
exercise
one
right
right
and
then
you
know
I
can
see,
exercise
and
being
the
async
one.
A
Where
you
you
know
it's
probably
the
most
complex,
complicated,
in-between
I
could
see
a
few
which
are
and
that
I
don't
know
them
off
the
top
my
head,
but
if
we
looked
at
the
API
it
would
be
like
teaching
you
a
little
bit
more
about
this
particular
part
of
the
API
right,
like
you
basically
got
up
saying:
you've
got
the
hello
world
on
the
front
and
then
is
there
something
else.
That's
you
know,
maybe
not
as
obvious
when
you
look
at
the
the
different
classes,
like
you
know,
working
with
the
object
type
or
they.
E
A
Yeah
promises
are
a
good
one,
so
that's
just
my
suggestion,
but
maybe
look
through
the
the
api's
and
say
well,
which
other
ones
look
particularly
interesting,
and
you
know
okay,
two
or
two
to
three
other
ones
to
fit
in
there
would
give
you
sort
of
the
you
know.
Some
people
may
get.
May
you
know
may
take
the
whole
time
just
to
get
through
the
the
first
two
you
already
have.
Some
people
who
are
sort
of
more
advanced
would
then
have
some
others
to
go
through
and
yeah
good.
A
D
D
D
E
E
E
E
E
A
A
Yeah
I
mean
I,
don't
know
it.
We
can
figure
that
figure
out
the
timing,
but
certainly
once
we've
gone
through
it
once
at
the
conference,
I
think
we
should
be
ready
to
say:
we've
got
feedback
there,
but
even
maybe
before
that
we
can
jump
into
that
yeah,
good,
okay,
so
yeah.
No,
that's
that's
great.
To
see.
B
A
I
It
today,
but
I,
have
like
have
a
gist
that
has
basically
what
he
his
proposals.
His
rough
proposals
are
he's
not
ready.
He
was
he
wanted
to
get
some
feedback
from
this
forum
before
he
published
them
so
I
prepared.
Oh
sorry,
I
should
have
just
Gabriel,
because
I
I'm
learning
how
soon
works
here
we
go
yeah.
I
I
So
yeah,
like
like
I,
say
this
is
a
kind
of
a
rough
cut.
I,
don't
think.
There's
too
much
controversial
here,
like
the
top
bullet
point,
is
basically
defining
what
an
a/v
API
is.
It's
nice
like
that,
was
a
link.
That's
at
YouTube
presentation,
I
haven't
actually
reacted,
but
I
haven't
watched
it
myself,
but
I
linked
it
in
the
original
one.
I
A
A
I
I
A
G
I
Yeah
and
then
so
she
see
a
idea
then,
like,
ideally,
it
should
be
discussed
at
a
meeting
like
I.
Do
I
tend
to
have
a
I
tend
to
just
miss
Peters,
occasionally
so
like
it'd,
be
good
to
have
just
a
call
a
call
out
of
any
new
API,
so
they're
gonna
land
at
the
meetings,
especially
if
the
person
who's
working
on
it
can
jump
in
I
know
that
there's
the
promise
API
is
I,
don't
know
what
the
state
of
those
were.
Did
they
land
or
not,
don't
believe.
A
I
I
G
G
F
I
That's
a
good
and
I
think
that
was
one
of
the
ones
to
attach
wasn't
didn't,
really
know
how
we
would
in
for
like
manage
that
sort
of
thing.
I
think
his
concern
was
that,
like
we
don't
want
a
new
API
to
be
like
we
want
to.
We
want
some
amount
of
time
where
we
are
able
to
modify
the
API
before
it
becomes
set
in
stone
in
case
like
in
cases
like
where
we
forget
to
add
a
finalized
callback,
for
instance,
sure
if
we
could
find
that
quickly.
I
A
I
D
D
A
Right
but
I
guess
I'm
worried
that
you
know,
as
you
ramped
up
in
numbers,
most
people
won't
even
read
the
docs
or
notice
that
is
experimental,
they'll
just
use
it
and
if
we
break
them,
they'll
be
like
what
right
and
and
it's
so
for
the
the
most
of
the
people
like
so
there's
this
new
method
and
if
people
really
care
about
it,
they're
the
ones
who
might
be
willing
to
turn
it
on.
If
they
don't
care
enough
about
it,
and
maybe
they're
not
going
to
be
the
people
who
give
us
feedback
anyway,
right.
D
A
A
A
F
A
I
I
It
follows
the
pattern
of
the
nappy
version
thing.
It
would
be
as
simple
as
doing
pound.
Defying
an
able
nappy,
experimental
one
or
something
like
just
defining
nappy,
enable
experimental
would
just
allow
those
api's
to
be
visible,
like
node
would
just
have
them
always
in
the
symbols.
It's
just
the
header
file
would
hide
them
from
you
unless
you
asked
for
them.
Yeah.
G
G
We
with
your
PR
Kyle,
it
could
be
as
simple
as
you
know,
like
defining
nappy
version
as
nine
nine,
nine,
nine,
nine
and
then
and
then
you
know
five
thousand
years
from
now.
Then
you
know,
then
that
will
break
yeah.
We
have
a
define
which
is
not
be
experimental.
It's
that
highest
number
exactly
yeah
yeah
yeah.
We
could
go
to
like
max
safe
integer
or
whatever
yeah.
G
A
G
Well,
actually,
actually,
if
we
turn
this
argument
around,
then
we
can
say
that
that
our
decisions
as
to
which
which
which
new
API
is
we
take
out
of
experimental,
will,
will
be
heavily
influenced
by
whether
we
can
back
port
them
or
not,
because
you
know
we
want.
We
want
the
highest
possible
and
API
version
to
be
available
as
widely
as
possible
right.
But
if
that's
not
possible,
because
we
see
v8
simply
had
lacks
certain
features
in
version
six
or
whatever
the
the
oldest
version
becomes,
then
you
know
we
we
have
to.
G
G
Insects,
it's
true,
and
you
know
what
that
that.
Also
that
also
kind
of
kind
of
it.
It
kind
of
impedes
us,
because
what?
If?
What?
If?
After
that,
we
run
on
which
we
could
backward.
But
now
we're
stuck
because
because,
in
order
to
back
board
the
one
which
we
could
we'd
have
to
back
port,
the
previous
one,
which
came
which
we
can't,
and
so
therefore
we
can
no
longer
back
port
anything
ever
again.
Well,
I.
A
Guess
my
thought
on
that
was:
we
could
still
back
port
functions,
but
it
it's
just
that
you
can't
use
the
versioning
potentially
to
know
that
right
like
you
have
to,
then
you
then
have
to
have
enough
knowledge
of
saying,
okay.
Well,
I
know
this
one
was
actually
back
ported
to
six
and
I
can
use
it
as
opposed
to
being
able
to
check
the
version
number
right,
I,
don't
I,
don't
see
an
easy
way
around
that
done
for.
G
A
G
A
A
We
have
a
bit,
you
know,
I
think
we
have
a
little
bit
more
of
a
flexibility
for
any
P
I,
because
we
are,
you
know,
there's
a
recognition
that
making
sure
that
people
can
go
back
to
six
is
important
right,
but
you
know
realistically
we're
now
down
to
ten
months
of
six
lifetime
right,
yeah.
Exactly
so.
C
A
A
I,
like
yeah
I,
think
I
think
that's
a
good
suggestion,
because,
just
in
terms
of
version
management
that
would
help
significantly
as
well
and
and
it
does
allow
us
to
land
things
and
for
people
to
use
things.
So
we're
not
like
being
a
blocker
in
the
way
in
in
the
way
completely.
But
it
does
give
us
a
little
bit
more
time
to
be
able
to
then
review
those
before
they
get
pulled
out
of
experimental
right.
I
G
A
A
I
A
A
I
A
A
A
Well,
I,
guess
you
know
it's
like
get
it
and
get
it
documented
in
that
in
that
issue,
and
then,
basically
you
know
at
CC
the
TSCA
well
at
CC,
all
the
collaborators
to
say
this
is
our
proposed.
You
know
it.
This
is
our
proposal
for
how
we
want
to
manage
the
new
new
new
API
is
going
into
any
pie.
Okay,.
D
A
A
I
D
D
A
I
A
I
D
I
I
A
I,
don't
think
we
can
I
I
think
maybe
some
wording
in
there
about
you
know
at
the
front
that
says
you
know
we
really
want
to
keep
the
the
set
of
api's
to
the
smallest
required
set,
and-
and
so
you
know
not
I,
don't
think
we
can
require
production
usage,
but
I
think
we
could,
you
know
require
we
could
basically
try
and
capture
in
words
like
this
really
needs
to
be
there,
not
just
something
that
seems
nice.
Yes,.
G
E
A
D
G
G
We
report
node,
we
tested
it
on
like
I,
don't
know
one
like
a
0.5%
of
our
capability
and
and
and
it
worked
you
know
and
that
that
is
that's
a
production
usage,
but
it's
kind
of
experimental
you
know
and
then
and
then
you
know
we
can
like
just
tip
our
hat
and
say,
of
course
you
know,
but
you
know
the
that's.
That's
perhaps
an
extreme
example,
but
even
even
something
less
than
that,
where
you
know
they
just
put
it
through
the
CI
for
their
own
shop
and
and
and
they
showed
that
it
made
things
possible.
A
G
No,
no,
no
no
I
mean
it
should
be
credible.
Is
what
I'm
saying
right?
Okay
and
the
thing
is,
the
thing
is
like
we
are
helping
them
out
by
by
publishing
as
experimental
right
like.
If
we
publish
with
this
flag,
then
we'll
make
it
easy
for
them
to
use
their
existing
a
deployment
pipeline
for
using
a
published
version.
All
they
need
to
do
is
add
an
Appy
experimental
to
their
Flags,
you
know
and
then
and
then
and
then
yes,
that's,
so
that's
a
development
step
on
their
part,
but
they
want
this
right.
A
G
Yeah
yeah,
no,
no,
not
and
I,
mean
I'm,
not
suggesting
that
we
make
this
a
like
a
like
a
hard
requirement,
but
but
it
it
would
support
the
case
for
for
it,
having
seen
some
real-world
usage,
even
even
even
experimentally,
right
and,
and
so
so.
The
thing
is
like,
while
it's
experimental,
we
can
break
it
right.
But
after
that
we
cannot.
So
we
need
to
know
that
it
works
for
at
least
somebody.
You
know
yeah.
A
I
I
C
A
J
The
port
itself
is
working,
but
there's
two
failing
test
cases
and
the
reason
for
that
is.
We
don't
have
a
way
to
check
whether
a
javascript
object
is
a
date
or
a
regular
expression.
So
working
on
making
a
plural
request.
To
add
that
so,
similarly,
to
how
we
check
whether
some
JavaScript
object
doesn't
promise.
A
G
A
I
I
Cord
doesn't
really
we
do
intrinsic
types
and
then
we
kind
of
leave
it
to
you
to
figure
it
out
from
there.
We
actually
don't
even
have
a
good
instance
of,
but
that's
more
of
a
gap
than
a
like.
We
don't
have
a
good
native
API,
for
instance,
of
so
that's
a
gap
that
we
would
almost
certainly
like.
We
could
fix
that,
but
I,
don't
where
I
know
in
the
past
we've
an
apprehensive
about
like
adding
a
new
public
API
for
every
single,
like
you
know,
are
you
this
are
you
that?
G
A
I
A
I
A
I
Other
they
have
one
caveat
to
that.
To
call
out,
is
you
have
to
make
sure
that
you
get
it
off
the
global
for
the
right
context,
I'm
not
actually
sure
how
multi
context
nappy
can
even
be,
but
I
just
run
into
a
issue
recently
with
the
node
chakracore,
where
we
were
using
instance
of,
but
the
promise
came
from
another
context.
J
A
F
A
Just
didn't
know
how
that
it
was
that
easy
to
do
so.
That's
why
I
didn't
suggest
it,
but
I
think
that
we
should
do
that
to
start
with,
and
then
I
mean
we
have
a
similar
request
for
like
now.
Those
may
be
slightly
different
is
like
the
in
32
s:
oh
yeah
yeah
yeah.
So
it's
kind
of
the
same
thing
except
in
that
case
like
in
32,
isn't
standardized.
A
A
I
That's
definitely
a
trickier
one,
because
I
don't
think
any
engine
actually
truly
like
theirs.
They
always
their
story.
Numbers
I
mean
big.
Ants
are
a
different
thing.
I
know
that,
like
you
can
still
usually
if
you
can
get
the
double
and
then
interrogate
the
bevel
to
figure
out
whether
it
is
an
int
and
then
whether
it's
a
32
bits.
So
it's
one
of
those
things
for
like
like
a
node
add-on
API
could
maybe
do
like
a
an
implementation.
I
A
I
A
Okay,
so
yeah
an
issue
that
I
think
from
this
perspective.
Let's
see
if
we
can
just
implement
it
without
adding
anything.
Sorry
for
sending
you
on
that
sort
of
wild
goose
chase
for
a
little
while
and
then
that's
we
may
have
to
well
I
think
we're
pretty
much
out
of
time,
unfortunately
today,
but
we
should
schedule
some
time
to
talk
about
the
32
side
of
things.
Okay,.
A
I
A
A
The
one
last
thing
I'll
mention
is
I
won't
be
able
to
make
it
next
week
actually
or
the
week
after
so
I'd
say,
fall
back
to
this
to
the
Hangout
in
terms
of
doing
the
meeting
and
I'll
try
and
read
up
on
on
what
comes
out
out
of
those
two
but
I'm
at
those
like
that's
a
collaborator
summit,
two
weeks
from
now
and
I'm
at
a
conference
next
week.
So,
okay.