►
From YouTube: 2020-04-09-Node.js N-API Team meeting
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
and
but
before
we
start
that,
does
anybody
have
any
announcements
that
they'd
like
to
make.
A
If
not,
okay,
we
will
move
on
to.
I
will
open
share
the
screen
with
milestone
11
and
we
can
start
from
there.
So
the
first
tagged
issue
is
renaming
nap
guy
is
something
more
appropriate
and
descriptive.
I
know
jim
you've
been
pushing
through
a
number
of
the
tasks
there.
B
A
Perfect:
okay,
okay.
That
sounds
good.
If
there's
nothing
else
to
talk
about
on
that
one
we'll
move
on
to
the
next
one.
If
I
can
get
back
okay,
the
blog
post,
about
the
transition
to
node
api,
I
think
I
know
I
haven't
had
a
look.
We've
got
an
outline.
A
C
What's
changed
not
in
references,
but
it
does
it
yeah.
It
all
really
depends
on
if
it's
a
lot
to
to
work
on
relax
but
did
not
approach,
because
I
feel
like
if
one
person
were
to
take
the
like,
what's
changed
and
what
did
not.
I
feel
like
that's
one
unit
of
things
you
know,
and
then
maybe
one
unit
is
the
background
discussion
of
like
why
we're
changing
it,
maybe
linking
to
the
the
issue
that
started
everything-
and
you
know
paraphrase
that
or
something
like
that
and
I
can
take
I'll,
take
the
approach.
Maybe
okay,.
B
A
A
A
I
mean,
I
think
any
of
these
sections
are
particularly
long
so
great
okay,
so
why
don't
we
try
that
at
least
that
way,
when
we
we
get
back,
we
can
it'll
be
good.
I
know
that
I'll
have
a
reminder
to
if
nothing
else
will
be
a
reminder
next
time
that
I
should
spend
some
time
a
little
bit
of
time.
I
don't
think
any
one
of
these
should
take
a
huge
amount,
so
right,
okay,
that
sounds
good.
So,
let's
move
on
to
the
next
one.
A
A
A
Let's
go
back
to
what's
in
there.
I
know
I
haven't
reviewed
the
external
examples
and
I
don't
I
don't
think
gabriel
had
time
I
haven't
enabled
debunk
for
testing.
A
So
we
can
move
on
to
burning
down
the
issues
raised
by
module
owners
so
note
that
on
api
are
there
any
ones
in
particular
that
people
want
to
sort
of
call
out
things
that
we
should
talk
about
yeah?
I.
D
I
have
one
request
that
that's
quite
a
long
time
down
the
risky
watch
reviews
it's
on
the
not
a
core
ripple,
the
issue.
The
purple
number
is
365
and
10.
D
There
are
some
issues
raised
also
in
the
comments,
but
the
the
pr
is
quite
doesn't
have
any
changes
through
a
long
time,
so
it
will
be
great
to
have
people
reviewing
it.
A
D
Yes,
and
so
in
the
first
version-
and
let's
say
the
major
version,
the
option
is
opting
so
we
can
like
say
in
the
next
major
version
we
make
the
operating
by
default,
so
people
would
have
to
opt
out
in
the
class.
If
I
recall
correctly,
we
have
a
discussion
of
the
adoption
policy
on
the
future
past.
A
Right,
I
guess
we'd
want
to
see
like
how
breaking
it
was,
but
yeah
we
can
at
some
time
in
the
future.
We
could
flip
the
default
from
one
to
the
other.
If
we,
if
we're
kind
of
like
convinced
it's
not
gonna,
it's
not
going
to
cause
any
grief
to
people
right,
yeah,
okay,
so
that's
one
that
yeah
I'll
go
back
and
I
it
wasn't
on
my
radar
if
you've
done
all
the
updates
and
it's
opt-in,
I
think
that
I'll
go
back
and
do
another
review
thanks
for
bringing
that
up.
C
C
C
Right
and
so
last
week
we
also
discussed
this,
and
so
what
he,
what
the
poster
suggesting
here
is.
Basically,
we
would
really
need
it
to
go
on
core,
and
then
that's
what
gabriel
said
here,
because
the
the
node
add-on
api
end
is
just
a
wrapper
around
the
core
object.
C
I
didn't
really
process
the
second.
This
last
comment.
Yet
your
case
is
probably
best
added
one.
Let's
put
it.
C
A
At
least,
I
think
that's
still
where
the
disagreement
is,
I
think
you
know
previous
times
we
discussed
it.
We
kind
of
agreed
that,
like
that,
having
the
option
to
have
that
behavior
we
can
see
making
sense,
but
it
should
be
opt-in
because
there's
sort
of
a
trade-off
right
like
if
it
would
be
kind
of
a.
I
know
what
I'm
doing
or
you
know.
A
C
A
A
A
Doing
that,
but
then
also
adding
in
a
guard
which
would
say
you
know
you
can
opt
in
to
basically
having
it
just
eat
the
exception.
A
C
A
A
I
think
that
would
make
sense
like
it
needs
to
be.
Does
it
need
to
be
in
there?
Well
like
it's
basically
like.
C
C
A
Right,
it
could
be
something
as
simple
as
putting
into
your
code
like
the
thing
that
does
the
include
before
that
you
put
like
a
hash
define
you
know,
suppress
blah
blah
blah
in
this,
we'll
check
that
right.
C
Yeah
there's
multiple
ways
to
get
a
decline
right,
okay,
that
makes
sense.
So
I
can
take
this
then
okay
and
we
wanted
to
try
to
get
this
in.
A
A
A
C
E
C
C
I
had
a
suggestion
randomly
that's
like
what,
if
we
flipped
the
order
of
the
tests
to
run
the
typed
thread,
say
function
ones
first,
because
the
implementation
for
tsfn
and
typed
is
a
little
different,
but
the
tests
are
the
same,
I'm
wondering
if
we
switched
them
and
we
also
see
errors,
would
that
would
that
help
any.
C
Like
if
it's
in
the
tech,
if
it's
in
the
tests
or
or
not,
I
don't
know
right,
because
I
guess
it's
hard,
because
it's
intermittent
so
it
could
just
all
of
the
typed
ones,
will
always
pass
and
fail.
But,
like
I'm,
not
really
sure
how
else
to
to
troubleshoot,
because
I
haven't
had
any
I've
tried,
also
tried
running
locally,
and
I
haven't
because
you
said.
A
A
C
C
C
C
C
A
C
C
A
A
A
A
Else
we
could
consider
doing
releases
in
a
different
order,
like
we
could
agree
to
do
a
release
that
just
drops
support
for
node
10
and
then
agree
to
do
one
more
release
for
the
3.x
line.
Once
we
figure
out
like
once
we're
ready,
it
was
just
sort
of
the
idea
was
like.
Let's,
let's
get,
let's
get
one
last
three
release
that
has
you
know
most
of
the
up-to-date
stuff
because
it's
been
a
while
and
then
do
a
four
to
drop
support
for
10
around
the
time.
A
Maybe
we
should
see
if
we
can
try
to,
I
don't
know
like.
Maybe
we
should
try
hard,
try
some
more
if
we
more
like.
If
we
a
couple
of
us,
try
and
run
this
in
a
loop
for
a
long
time
and
see
if
we
can
recreate
it
and
then
next
week
we
can
figure
out
because
the
node
node
doesn't
go
like
node.
16
doesn't
come
out
until
the
week
after
next
right.
A
So,
okay,
so
yeah,
why
don't
we?
Maybe
let's
do
that
for
this
week,
is
let's
let's
try
and
see
if
we
can
reduce
it,
because
that's
the
only
one
that
I
think
if
we
look
at,
let's
look:
are
there
any
other
ones?
We
think
we
absolutely
want
to
get
in.
A
I
think
kevin
the
other
one,
like
the
throws
javascript
exception.
If
we're,
if
that's
done
in
time,
that
would
be
good.
I
think
the
rest
of
them
are
all
you
know.
Let
me
look
at
the
iterator
for
object
is
that
I
think
that
one
there's
still
some
discussion
going
on
there.
Wasn't
there.
A
A
A
A
Yep,
let's
it
should
be
just
if
you
do
an
npm,
install
and
then
npm
test.
A
I
don't
know
if,
if
kevin
on
mac,
is
there
anything
else
that
needs
to
be
done?
There.
A
Okay,
I
mean
I
did
try
in
some
cases
to
run
subsets
of
the
tests
because
they
run
a
bit
faster
and
and
to
do
that
in
the.
A
C
A
C
B
A
B
A
Which
gets
slightly
challenging
sometimes
because,
like
we're
launching
new
processes
and
stuff,
I
think
in,
like
the
I
think,
I'd
also
modified
the
index.js
to
not
spawn
a
new
environment
because
otherwise
your
debugger,
would
you
know
you
would
you
would
run
the
debugger
on
the
on
the
the
outer
process
and
not
the
process.
That's
actually
crashing.
A
A
A
A
Just
I
have
a
couple:
pull
requests
to
add.
Like
the
the
project
uses
the
dco,
so
we
should
have.
We
should
present
the
dco.
D
F
A
D
Yeah,
so
we
have
been
talking
about
this
once
earlier,
so
I've
almost
completed
the
source
changes
and
the
left
one
is
making
existing
tests
to
be
compatible
with
maybe
or
maybe
not
enabled
so
the
most
of
the
api
are
compatible
between
the
maybe
enabled
and
disabled,
but
some
of
them
like
say
we
are
accessing
the
object
properties
with
the
with
the
bracket.
D
D
Impossible
to
let
users
let
the
developers
aware
that
the
bracket
accessing
is
throwing
so
in
these
changes.
I've
made
that
bracket
so
in
bracket
accessing
the
error
being
a
fatal
error,
if
the
maybe
is
enabled
and
to
make
it
a.
E
D
Since
then,
they
are
unable
to
ergonomic
ergonomically
to
see
that
if
there
is
anything
throwing
in
javascript
since
they
have
don't
have
any
means
to
let
compiler
checks,
the
value
is
empty
or
not,
since
the
value
wrapper
with
the
boolean
or
number
or
object
is
the
same
kind
with
the
value
in
value
wrapped
or
nothing
interrupt.
So
they
may
be
kind
is
just
maybe
type
just
make
sure
the
compiler
checks
the
return
value,
and
it's
if
you
got
the
maybe
or
value
is
in
empty,
maybe
is
an
empty
map.
D
D
So
this
p
appear
in
major
for
the
majorly
for
the
add-ons
that,
when
the
c
plus
plus
exception
is
not
enabled.
D
D
There,
in
no
problem
in
the
test,
I
have
to
work
with
existing
tests
to
be
compatible
with.
Maybe
so
we
don't
have
to
write
test
duplicated
for
the
maybe
enabled.
F
D
D
C
E
C
Okay,
right,
okay
and
then
so
because
you
said,
the
problem
is
with
the
bracket
operator,
because
we
have
the
the
ability,
the
the
shorthand
to
say,
like
object.
Bracket
string
equals
something
and
that
is
assigned
something
and
that
doesn't
work
with
the
maybe.
A
So,
but
I
guess
it's
like
I'm
wondering
now
like,
so
what
do
we
expect
if
we
can't
get
our
tests
to
work?
Is
it
a
is
it
that,
like
we'd,
have
to
change
all
the
tests
to
make
it
work
or
that
like
because
that
that
then
that
means
it's
like
a
it's
a
big
breaking
change
to
anybody
who
enables
it
right
versus
the
or
is
it
you
know
how?
How
would
like
somebody
who's
upgrading
to
say
I
want
to
enable
maybe
handle
this
case.
D
A
D
Nothing
yes,.
D
That
can
work,
but
the
I
may
explore
the
semantics.
If
it
will
trend
like
say,
make
it
harder
for
people
to
understand.
What's
going
on
here.
A
In
the
tests
and
stuff
sure
yeah
that
that
makes
sense,
but
that
might
be
a
way
where
you
know
we
can
update
all
the
tests
so
that
we
do
get
the
proper
testing
and
it
can
run
both
ways
and
it
the
other
advantage
is
like
if
it.
If
it's
not
confusing,
it
would
also
be
a
way
where
we
can
show
people
what
they
should
be
doing
right.
A
E
A
D
Oh
well
we're
taking
a
look
into
the
second
pass
callback.
I've
just
find
noticed
that
there
are
lexi
said
immediately
said
immediately
in
the
core
on
the
finalization
of
non-api
finalizers
yep.
So,
while
I'm
testing
the
second
pass
callback,
I've
noticed
that
this
setting
immediately
have
captured
the
environment
reference.
So
then,
in
the
environment
tell
down
the
immediate
is
not
cancelled
and
the
present
reference
is
still
there
and
that
caused
a
crash.
But
I
didn't
I
didn't
replay
it.
D
The
course
is
very
similar
that
you
say
we
are
running
users
penalizers
in
the
environment,
environment
tell
them
so
we
didn't,
but
the
fundamentals
are
very
tricky.
Like
say,
we
have
to
defer
the
finalizers
to
be
run
so
that
the
gc
is
properly
handled.
E
D
D
But
that's
require
another
exploration
to
see
if
that
change
is
in
value.
Okay,.
D
A
D
E
E
A
C
Sense,
though,
because
I
remember
when
we
were
doing
like
the
note
add-on
examples,
you
know
was
adding
a
thread
say
function
test
I
had
to
put
in
the
package
json
engine
greater
than
or
equal
to,
and
then
I
had
to
look
up
when
was
node
api
4
released,
and
what
version
was
that?
Because
that's
when
it
was
introduced?
So
I
had
to
look.
D
D
C
A
C
A
C
C
A
D
E
C
A
I
think,
if
not
thanks
for
everybody's
time
and
we'll
see
everybody
next
week,
thanks.