►
From YouTube: Node.js N-API Team meeting - Sep 20 2018
Description
A
A
A
You
know,
I
think
I
think
we
got
the
she's
gotten
the
test
cases
the
pass.
But
what
was
noticed
is
that
if
tests
fail,
they
just
hang
in
one
case
or
if
we
have
C++
exceptions,
disabled
and
if
C++
exceptions
are
enabled,
it
actually
throws
the
C++
exception,
which
is
interesting
because
I
think
we'd
expect
that
to
have
been
caught
somewhere
and
turned
into
a
JavaScript
exception.
A
But
it's
it's
a
bit
of
a
challenge
to
debug,
because
the
tests
run
with
mocha
and
there's
kind
of
like
several
layers
to
be
able
to
do
things
like
get
stock,
traces
or
debug
in
and
so
pushing
pushing
ahead
on
that,
but
still
trying
to
figure
out
basically
the
flow
of
either
you
know.
Where
is
it
hanging
or
in
the
case
of
the
C++
exception,
enabled
case
you
know?
B
A
A
C
A
A
little
longer,
if
it
takes
more
than
a
few
weeks,
maybe
we
can
think
about
how
to
ping
or
whatever,
but
for
now
it
seems
totally
reasonable.
Okay,
good
next
one
is
documentation.
Push
on
NEPA
know
that
on
API
that
one
I
think
Nicola
I
know
that
the
maybe
I
should
check
that
off
that
these
guys
were
landed
and
so
I
think.
The
last
thing
is
you
have
a
PR
already
open
to
do.
This
sort
of
last
cleanup
is
that
right.
A
I,
like
I'd,
be
happy
to
lie
on
this,
and
do
it
as
a
separate
pass.
If
anything
I
think
that's
easier
because
there'll
be
you
know
those
those
those
new
additional
PRS
will
be
easier
to
review
and
I.
Don't
really
see
any
downside
in
landing.
What
you've
already
got,
because
it's
only
an
improvement.
In
my
opinion,.
A
D
A
D
A
A
A
A
A
Yep
done
thanks.
Okay,
so
it
sounds
like
we'll
go
ahead
and
land
that
one
and
that'll.
Let
us
basically
I
think
closed
out
the
documentation
push
you
know,
maybe
a
little
bit
more
work,
but
I.
Don't
think
that
necessarily
needs
to
have
a
separate
unless
anybody
disagrees.
I
don't
think
it
needs
like
a
separate
thing
in
the
milestone
anymore.
A
No,
okay,
great
so
I'm
gonna
flip
over
to
milestone
11
men.
A
So
I
moved
over
these
burn
down
the
list
of
issues
raised
by
module
owners
burned
down
a
list
of
issues
and
there's
no
repo
I
did
do
a
little
bit
of
a
push
this
week
to
try
and
land
a
bunch
of
the
node
add-on,
API
things
that
are
out
there
or
suggested
a
new
release
which
Nicola
has
volunteered
to
do.
I
think
we
have
about
four
four
issues:
I
suggest
that
we
land
and
then
we
we
go
ahead
and
do
that
are
there
any
other
ones
are
specific
issues.
D
One
of
the
issues
which
is
to
implement
like
I,
can
extends
thing
in
napi,
so
that
people
can
can
subclass
and
have
signatures
on
their
methods
and
have
the
of
the
type
check
pass
and
the
the
person
on
the
v8
side
who
who
seems
to
be
the
expert
on
this,
has
returned
so
I've
resumed
conversation
with
him
about
about
implementing
this,
and
maybe
maybe
maybe
we'll
get
some
support
from
v8
for
this,
because
right
now
the
thing
that's
holding
it
back
is.
Is
that
you
cannot?
D
You
cannot
subclass
something
that
you've
defined
using
that
we
define
class,
you
can't
subclass
it
from
from
from
an
API.
You
can
only
subclass
it
from
JavaScript
and
so
we're
starting
to
define
a
v8
API
that
we
would
need
for
for
this
to
work,
and
so
that's
pretty
good
I'm,
not
sure
if
it's
going
to
materialize,
but
he
seems
to
be
the
person
to
talk
to.
So
that
would
be
awesome,
because
then
we
can
have
proper,
proper
subclasses
defined
in
an
API,
and
we
could
have
a.
D
A
D
D
This
whole
thing
back
is:
is
the
inability
to
retrieve
the
function
template
from
Brima
from
the
function,
and
so
what
he
suggested
was
that
we
had
an
API
to
the
function,
object
which
allows
you
to
extend,
and
it's
good
it's
it's
it's
good
to
have
like,
but
even
if
not
in
an
API
and
I
mean
in
other
engines,
we
can
fall
back
to
like
prototype
assignment
because
that's
that's
functionally
equivalent
and
I.
Think
I
think
it's
only
v8
that
has
this.
It
has
this
exception.
A
Like
I'm,
not
saying,
don't
doesn't
make
sense
or
isn't
good,
like
I,
think
it's
it's
good
to
make
an
improvement
they're
just
that
would
be
the
one
thing
to
keep
in
mind
in
terms
of
yeah.
We
don't
want
to
add
something
that
then
you
know
the
other,
implement
a
ters
or
duct
tape
or
whatever
can't
implement.
A
D
A
Sounds
good
that
I
guess
related,
not
a
bad
time
to
mention
I've
seen
two
things
this
last
week,
one
was
like
somebody's
got,
an
experimental
jst
port
of
like
node
running
on
JSC,
and
then
there
was
also
the
I
think
the
iOS
port
from
neo
Gnaeus.
That
right
was
it
I
would
know
so.
The
iOS
was
the
Jessee
one,
but
there's
also
like
a
more
embedded
one
from
the
onus
which
was
no
GS
on
ducting.
D
D
They're,
actually
more
serious
about
it
than
I
am
because
I
I,
just
kind
of
wrote
it
as
an
experiment,
but
they
seem
to
want
to
support
it.
And
so
so
my
takeaway
from
the
issue
was
that
they
would
like
us
to
to
package
the
n
API
tests
as
an
NPM
package,
so
that
so
that
they
can
also
run
the
test
suite,
and
that's
that's
actually
something
we
might
want
to
talk
about
at
some
point
ya.
C
A
Do
you
think,
there's
any
chance
of
getting
somebody
from
shadow
note
to
join
this
team?
Well,
I
can
I
can
certainly
comment
on
the
issue
yeah.
That
would
be
worthwhile
because,
if
they're
interested
in
and
like
you
know
having
another,
if
they're
gonna
be
pushing
the
Jerry
script,
port
of
of
any
P
I
have
me
involved
and
what
would
be
good
yeah.
A
Other
two
that
I
mentioned
they
don't
have
any
actually
don't
have
any
P
I
ports
yet,
but
it
shows
more
of
a
drive
to
be
having
node
running
on
other
platforms
or
engines
and
I
think
long.
You
know
if
they
don't
need
it
in
the
short
term.
You
know
longer
term
it.
It's
really
likely
still
make
sense,
so
that
modules
would
run
on
those
platforms
without
having
to
be
specific,
exactly
yeah.
That's
that's
the.
A
F
F
A
A
D
A
D
There's
only
one
concern
that
I
that
I
remember
having
and
that's
that
if
we
do
factor
out
the
tests
into
into
a
separate
repo
than
then
we
would
have
to
sort
of
copy
that
repo
into
the
nodejs
a
code
base
as
like
Depp's
an
API
test
suite
and
then
we'd
have
to
integrate
that
with
what
you
call
that
the
the
the
the
test
running
framework
that
we
have
in
Cir
Jenkins
or
no,
no,
not
Jenkins.
It's
a
format.
It's
like
an
output
format
that
we
use
make
tapper.
A
D
A
D
E
A
A
A
F
A
F
Because,
certainly
there
there's,
you
know,
there's
a
lot
of
interest
in
this
I
mean
the
one
thing
is
to
how
to
link
this
to
an
API
based
module
growth
and
I
was
talking
to
Alexis
the
genera
systems
and
he
did
an
iOS
port
of
chakracore
and
built
a
a
node
system
for
that
and
has
a
few
apps
in
the
App
Store
which
using
this
and
I
was
like
well,
do
you
need
an
API
and
he
was
like
you
know,
I.
He
didn't
run
into
any
native
modules
in
the
mobile
space.
So
I
was
like.
D
D
B
A
And
that
is
an
area
I'm
still
trying
to,
unfortunately
hung
who
was
I
was
was
gonna
row
into
champion.
This
sort
of
embedding
push
has
moved
on
to
something
else,
so
I'm
trying
to
see
if
I
can
get
somebody
else
to
reinvigorate
you
know
we
had
the
one
meeting
to
reinvigorate
that
because
I
think
it's
it's
not
strictly
an
API
related,
but
you're
right.
It
does
have
some
synergy
on
in
that
respect.
A
D
A
A
D
D
But
the
fact
that
he
is
a
collaborator
tells
me
that
he
he
should
have
maybe
made
a
few
PR,
the
the
things
that
he
has
seen
right,
yeah
so
I
mean
he
does
maintain
a
fork
which
is
no
js'
mobile.
So
he
certainly
has
it
in
there
and
given
that,
given
that
this
whole
thing
is
monolithically
built
all
the
time
because
they
are
apps
running
on
mobile,
then
then
building
them
off
the
fork
makes
perfect
sense,
but
I'm
not
sure
if
we
could
backport
some
of
this
to
to
upstream.
You
know,
yeah.
D
Because
the
other
thing,
the
other
thing,
that
it
involves
and
I,
think
that
this
is
probably
where,
where
it
gets
very
difficult,
is
that
you
know
if
we
backboard
these
things,
then
we
would
want
to
dedicate
at
least
a
job
to
building
on
mobile.
You
know
that
we
can
actually
test
that,
but
at
that
point
it
becomes
a
very
big
project,
because
we
need
to
get.
A
D
Exactly
yeah
yeah
cuz
you'd
have
to
get
a
job
up
with
an
emulator
and
that
kind
of
stuff
I
mean
we'd
almost
start
duplicating
what
browserstack
does
at
that
point,
so
so
yeah
I'm,
you
know
it's
kind
of
a
chicken
and
egg
problem.
It's
it's
it's
a
it's.
It's
very.
It
seems
to
be
in
high
demand,
but
on
the
other
hand,
there's
a
there's,
a
pretty
big
barrier
to
entry.
You
know
right
so
I'm
not
sure
how
we
could
how
we
could
do
this
in
in
small
increments.
A
A
Think
about
so
I
think
we
went
on
a
little
bit
of
a
tangent
there,
but
I
guess
we're
just
back.
We
just
went
out.
We
were
on
that
tangent
we're
about
to
move
on
to
the
next
issue.
Yeah.
F
A
A
A
A
B
A
A
Mean
what
I,
what
I
was
thinking
too
is
that
to
sort
of
minimize
you
know
time
people
have
to
spend
for
prep.
You
know
most
of
this.
We
would
do
relatively
informally
unless
somebody
wants
to
go
into
volunteer
to
put
a
set
of
slides
and
stuff
like
that,
like
I
think
we
can
start
out
with.
If
we
see
that
in
the
room
we
get
a
number
of
people
who
are
new
to
NTP
I,
we
can
sort
of
just
you
know
verbally
give
an
overview,
pull
up.
A
Some
of
the
repos
walk
people
through
it
then
I
thought
we
could
go
through
like
here.
You
know
I
think
we
want
to
have
a
base
agenda,
but
if
there's
a
bunch
of
people
there's
ten
people
there
who
want
to
talk
about
some
specific
and
APRI
thing
that
we
don't
have
an
agenda,
we
should
do
that
right.
So
we
should
then
adjust
the
agenda
based
on
who's
there
and
what
they
want
to
talk
about.
A
Then
we
have
the
hands-on.
So
again,
I
just
thought,
like
you
know,
whoever
feels
most
comfortable
among
us
can
talk,
can
walk,
helped
to
walk
through
the
files,
we'll
just
bring
up
node
API
and
node
API
under
CC
and
walk
through
the
content
there
to
help.
Explain
it
same
with
no
down
it
and
no
doubt
on
API
and
really
just
that
that
be
driven
by
what
people
think
of
mentioning
questions
that
come
up
stuff
like
that.
A
We
can
then,
similarly
look
at
you
know
the
current
areas
in
our
milestones,
as
you
know,
starting
points
with
the
good
areas
as
well.
As
you
know,
if
there's
look
at
the
open
issues-
and
you
know
see
if
there's
some-
that
we
think
are
good
for
people
to
jump
into
I,
think
one
one
for
sure
is
like
no
doubt
on
API.
We
have
a
list
of
those
things
that
aren't
tested
yet.
So
that's
a
good
way.
Everyone.
A
B
A
F
How
long
is
this
it's
an
hour
and
a
half
now
explain
discuss
you
know
based
on
the
current
agenda,
yeah.
A
A
C
A
F
A
C
A
Okay,
so
comment
in
there,
but
I
guess,
like
I'm
gonna,
not
worry
about
this
too
much
more.
If
we
go
in
with
this
I'm
happy-
and
you
know,
if
you
think
we
should
tweak
it
before
then
otherwise,
let
us
know-
and
still
we
can
tweak
it
either
way,
but
make
sense
so
back
to
the
milestone,
create
matrix
of
tested
methods.
A
Okay,
take
that,
but
we
have
this
initial
list
of
ones
which
I
believe
are
the
ones
that
aren't
tested
and
andrew
is
also
working
on,
like
a
make
coverage
target
for
no
doubt
on
API,
so
we'd
be
able
to
run
that
and
they'll
actually
give
us
just
like
that.
If
you
looked
at
coverage
dot,
nodejs
org,
we
see
coverage
numbers
for
all
of
node,
core
C
and
C
and
JavaScript
this.
A
This
one
would
just
be
seen,
but
it
would
give
us
a
clear
view
of
where
were
what
we're
testing
and
what
we're
not
testing
who's
Andrew,
but
Andrew
he's
an
intern
here
in
in
IBM,
oh
okay,
so
anisha's!
So
you
know
we
we
we
have
rotating
interns,
I!
Think
they're,
like
an
18-month
internship
Oh
Misha
was,
is
now
back
at
school.
So
we
have
a
new
intern.
I,
don't
have
all
of
their
time,
but
you
know
I
try
and
get
a
better
time
to
help
out
on
these
kinds
of
things.
Yes,.
B
A
D
C
D
A
D
D
D
F
D
C
D
Believe
so
because
we
control
Doc's
of
nodejs
org,
so
you
know
we
could
filter
de
logs
for
for
how
many
times
those
SVG
files
were
downloaded
I'm
fairly,
certain
I
mean
that
that's
something
for
or
the
infrastructure
team
to
answer,
but
I
think
I.
Think
we
can
do
that.
Like
I
mean
we
can
get
stats
on
every
URL
that
that
is
served
by
one
of
our
servers,
I
think
so
the.
A
A
So
I
guess
the
you
know
it
is
something.
Do
we
think
the
core
Doc's
in
particular
give
us
more
visibility,
better
visibility
than
the
foundation
website,
the
I
mean,
certainly
it's
a
higher
visibility
than
the
node
add-on
API
repo,
but
it
seems
like
where
there's
more
convincing,
that
this
is
a
good
idea.
D
D
Why
not
just
write
it
out?
Yeah
yeah
yeah,
it's
yeah,
I
I!
Guess
it's
a
tough
call,
but
but
I
mean
at
least
in
my
opinion.
Putting
this
putting
this
at
the
top
of
the
of
the
napi
documentation
is
where
it
makes
the
most
sense,
because
that's
the
documentation
that
people
will
be
reading
when
they're
porting
their
module
and
so
right.
So
this
kind
of
this
kind
of
gives
them
a
goal.
Yeah.
C
D
D
D
D
A
D
A
Is
so
I
don't
know
it
might
make
sense
to
reach
out
to
her
and
kind
of
say
you
know
it's
like
it's.
It's
probably
the
best
pace
to
achieve
that
goal,
and
is
the
downside
like
if,
even
if,
it's
not
a
perfect
set
this
what's
the
negatives
that
don't
weigh
that
the
the
benefit
that
we
see
anyway
right
yeah.
F
A
D
A
D
Right,
yeah
yeah,
because
I
mean
most
people
know
how
to
publish
modules
and-
and
they
understand
that
an
API
is
a
different
interface
and
so
so
they'll
they'll
simply
start
going
through
their
module
and
and
start
changing
stuff
over
to
an
API
and
and
go
directly
to
the
to
the
API,
Docs
I.
Think
so
you
know
I,
don't
think
they
need
a
guide
on
how
to
publish
in
an
API
module.
You
know
I
mean
the
guide
concerns
itself,
mostly
with
with
you
know,
publishing
under
an
experimental
for
agonal
that,
but
that's
they're.
C
D
D
A
Lines
right
so
yeah
what
yeah
yeah.
F
D
D
D
D
A
D
F
So
how
about
we
do
this?
How
about
we
put
this
recommendation
on
a
bi-stable
note
as
well
and
while
we're
trying
to
while
we're
trying
to
get
this
in
the
APA
reference
as
well,
the
the
reason
why
this
thing
is
like
we
need
to
at
least
be
out
that
there's
a
badge
available
I
want
to
talk
to
mark
and
say:
well,
you
know
he
has
an
Appy,
enabled
module
deputy's
badge
and
hope
hoping
to
kind
of
get
a
few
badges
available
before
the
talk,
so
that
we
can,
you
know,
start
talking
about
it.
D
That's
that's.
That's.
Certainly,
that's
certainly
something
we
should
do,
but
when
it
comes
to
the
actual
badge
we
may
want
to
have
like
an
authoritative
URL
for
it.
You
know
so
that
people
don't
start
making
their
own
because
I
mean
they
can.
They
can
just
go
to
shields
I'll
type
in
three
three
words
and
they
have
a
badge.
Sorry.
A
A
D
F
A
Yes,
right
like
we
want
the
visible
the
reason
we
had
put
it.
In
course,
we
just
want
to
visible.
There
are
the
actual
badges.
If
we're
gonna
have
SPG's
of
our
own
could
be
easily
there,
and
so
the
initial
version
of
the
dog
could
just
be
in
our
API,
stable,
node
repo,
and
then
you
know
we
retain
those
links
for
anywhere
else
that
it's
documented
yeah.
F
A
A
D
Yeah,
in
order
to
do
that,
though,
we
kind
of
need
to
sort
of
get
up
to
speed,
get
the
repo
up
to
speed,
because
we
haven't
updated
the
actual
repo.
In
a
long
time.
We've
only
been
using
the
issues
right,
so
we
should
probably
pull
from
from
master
in
the
main
Rico
so
that
the
actual
napi
document
is
up-to-date
and
so
that
so
that
the.
F
F
E
A
Add
it
in
like
on
the
main
readme
I
would
just
add
it
in
on
the
main.
Read
we're
not
saying
make
it
in
a
fork
of
it.
It's
like
put
the
content
into
this
repo.
There
could
be
like
you
know.
We
now
have
data
meeting
notes.
There
could
be
a
separate
badges
subdirectory
where
the
actual
images
lives.
Oh
I,
see
okay,
yeah.
A
A
D
F
D
D
C
A
Document
limitations,
the
babyĆs
stability
bar
for
any
PRI
I,
know
Gabriel.
You
had
a
couple
peers.
D
I
think
I
I
think
those
are
close,
but
I
remembered
Nicola
and
I.
We,
we
discussed
another
thing
that
we
might
want
to
PR,
but
I
never
got
around
to
doing
it.
So
not
sure
if
we
should
close
this.
It's
close
the
bull
ish,
but
but
we
should
probably
get
rid
of
that
last
issue
that
we
discussed
in
one
of
the
comments
and
then.