►
From YouTube: 2022-09-16-Node.js Node-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
So
welcome
to
the
node.js
node
api
team
meeting
for
september
16
2022
we'll
follow
our
standard
approach
of
looking
at
the
issues
tagged
for
milestone
11.
before
we
do
that,
does
anybody
have
any
announcements
or
info
they'd
like
to
share.
A
I'll
just
add
my
my
regular
message
to
anybody
who
might
be
watching
the
videos
that
collaborator
summit
doublet
is
coming
up
is
coming
up
a
couple
days.
I
think
it's
the
first
and
second
before
the
the
node
conf
view
conference.
A
If
you
have
a
if
you're
in
the
area
and
can
make
it
that
should
hopefully
be
a
good
time
to
get
together
with
collaborators,
you
know
it's
it's
geared
towards
collaborators,
so
not
necessarily
like
an
intro,
but
if
you're
you
know
thinking
that
you
want
to
become
a
collaborator
and
sort
of
listening
in
and
learning,
it's
still
a
good
opportunity
so
and
check
out
there's
an
issue
in
the
openjs
foundation,
repo
under
a
summit.
A
A
A
We
discussed
this
last
time.
I
I
didn't
have
a
chance
to
take
a
look
at
it.
I
don't
see
any
comments,
so
I'm
guessing.
Nobody
else
has,
but
that's
a
good
one
for
us
to
go
in
and
and
provide
some
info
on,
checking
for
memory,
leaks
and
add-ons.
B
How
did
this
one
come
about
like
what
was
the
discussion
or
issue
or
whatever
that
created
this?
There.
A
Was
this
issue
where
somebody
reported
that
there
was
a
memory
leak
and
the
answer
really
is
yes,
it
will
be
reported
because
add-ons
don't
like
we
don't
actually
unload
add-ons,
not
necessarily
node
api
specific.
A
A
So
we
have-
and
we
do
have
this
like
this-
actually
maybe
I
should
have
put
it
there.
I
actually
think
we
were
thinking
of
putting
it
into
the
like
the
note
api
docs
itself,
so
the
developers
would
find
it
so
find
a
place
in
there
that
made
sense
and
and
add
in
some
maybe
some
pointers
there.
There
is
a
like
a
contributing
doc
that
has
some
info,
but
that's
more
for
like
node
contributors
versus
like
developers,
so
we
thought
it
would
be
best
in
the
node
to
know
docs
itself.
A
Okay,
adding
the
c
list
of
non-c
language
bindings,
I
think
nicholas,
been
working.
It
was
planned
to
work
on
that,
but
he's
he's
been
out,
so
I
don't
think
there's
any
updates
on
that
one:
better
coverage
of
node
api
and
youtube,
so
I
think
nick
also
volunteered
so
again
he's
been
out.
So
no
progress
on
that
one.
I
don't
think
there's
anything
to
talk
about
unless
anybody
else
is
thinking
of
doing
anything.
There
has
any
updates.
B
I
didn't
add
any
new
tests
this
week,
but
I
got
some
comments
from
chandra
and
I'll,
probably
I'll.
A
Make
those
changes
and
yeah?
Thank
you,
okay.
So
there's
a
few
comments
on
these
tests
or
yeah
yeah,
I
think,
was
a
second
one
yeah
just
one
yeah,
okay,
yeah!
Thank
you.
A
Okay
sounds
good,
so
if
we
can
just
review
and
help
get
those
those
ones
landed,
that'll
be
good,
I'll,
try
and
block
some
time
off
myself.
Other.
C
A
Gonna,
take
a
look
I'll
I'll
assume
you
take
a
look
and
if
I
do
get
time,
try
and
spend
it
more.
Looking
at
your
your
pr
in
core.
Thank
you.
A
C
A
Okay,
okay,
stale
issues,
so
looking
at
the
stale
issues,
I
know
I
added
one
new
one,
which
was
this
one.
A
Cover
but
it
seemed
like
we
still
needed
some
thought
on
how
we
would
integrate
it
with
no
data
on
api.
B
Yeah,
so
there
was
that
core
pr
was
there,
something
that
we
needed
to
do
in
node
add-on
api.
In
order
for
this
to
because
I
remember
yeah,
the
gendercast
was
working
on
that
one.
A
A
A
A
If
this
is,
we
should
just
turn
it
on
like
it
like.
You
are
right
there
like
there's,
still
bug
fixes
which
can
affect
code,
but
it's
is
it.
Is
it
likely
to
affect
things
where,
like
it
runs
on
one
version
and
does
the
right
thing,
but
doesn't
run
on
the
other
version
and
do
the
same?
The
right
thing.
A
C
Anyway,
I
think
I
need
to
take
a
look
into
this
to
see
how
the
flags
can
resolve
the
problem,
how
the
flag
can
elevate
the
problem
here
right
if,
if
the
default
value
switching,
can
be
the
solution
here.
If
so,
we
can
decide
what
we
can
do
next
as
the
next
step
or
if
we
need
to
do
anything
in
the
node,
I
don't
apply
to
right
or
even
if,
like.
C
A
Yeah,
because
this
seems
like
another
good
thing
where
you
could
just
say
like
yeah
in
your
code,
you
can
say
turn
it
on
and
you
know
maybe
it'll
even
it'll
tell
you
whether
or
not
the
versions
you're
targeting
support
it,
and
if
it
does,
then
you
you
could
and
we
wouldn't.
We
wouldn't
necessarily
affect
somebody's.
C
A
Yeah
yeah,
along
with
note
version
and
and
it
it
makes
the
story
kind
of
all
night-
all
hung
together
there.
So.
B
Yes,
I
I
was
also
sort
of
trying
to
catch
up
a
little
bit
on
this
one
as
well,
and
this
is
because
we
do
a
static
cast
of
that
negative
one
for
an
unknown
type,
and
there
was,
I
think,
there's
a
pr
that
was
opened
to
resolve
it.
Yeah.
If
you
go
to
the
pull
request.
B
B
He
just
linked
to
the
issue
again,
but
if
you
just
go
to
the
list
of
the
pull
requests
on
no
doubt
on
api,
okay,
sorry
go
here
right,
yeah,
yeah
and
it's,
I
think,
the
last
one
there
right
yep
and
then
I
think
there
was
some
discussion
here
talking
about
in
removing
the
concept
of
a
neo
type
and
to
me
that
actually
makes
the
most
sense,
because
in
the
underlying
node
api
there
is
no
concept
of
an
unknown
type.
B
A
A
So
our
comment
there,
like
this
comment
that
I
put
in
there,
was
like
we
discussed
that
we
basically
agreed
that
yeah.
We
should
just
remove
it
and
refactor
the
code
to
do
that.
I
think
we
were
just
you
know,
seeing
if
if
zach
was
maybe
gonna,
you
know
incorporate
that
into
his
pr.
A
B
And
just
another
thing
for
my
clarification:
once
you
create
a
typed
array
with
a
specific
length,
can
that
length
change?
My
thought
is
no.
B
Because
I
think
in
like
the
the
the
default
new
constructor
with
no
no
arguments,
just
cree
uses
a
length
of
zero,
so
you
in
my
like,
I
can't
imagine
what
you
would
be
able
to
do
with
a
typed
array
of
length
zero
and
if
that's
the
case,
then
like
the
type
really
shouldn't
even
matter
right.
So
then
that's
why
I
think
it'd
be
okay
to
use
just
some
some
default
type,
because
you
really
wouldn't
be
able
to
interact
with
the
typed
array.
Anyways.
Does
that
make
sense.
A
Yeah,
I
I
think
that
was,
I
can't
remember
the
details
now,
but
it
was
along
the
lines
of
like
how
could
you
you
be
usefully
using
this
right
right
and
it's
I
mean,
maybe
maybe,
when
we
look
at
it
more
carefully,
it
might
be
related
to
like
doing
a
you
know
like
copies
or
something
like
that,
but
but
yeah
I
know
I
I
it
just
seemed
like
somehow
we
created
it
that
way,
but
just
didn't
make
any
sense,
doesn't
make
any
sense.
A
B
B
C
B
Because
a
warning,
then
you
know
it
wouldn't
break
any
existing
code,
at
least
from
compiling.
So
all
right,
this
seems
like
a
simple
enough
thing
that
I
could
be
able
to
to
work
on
then.
B
Releasing
the
thread
safe
function
and
if
you
have
an
object
that
holds
a
tsfn
and
you
want
to
release
at
destruction
time
or
or
something
like
that.
If
tfsn
is
held
longer
in
time
than
objects
which
contain
the
tss
are
created
and
destroyed
on
the
fly,
there
must
be
a
way
to
properly
release
a
function.
That
is
a
member
of
an
object
inside
the
object's
destructor.
B
A
At
one
point
so
we're
like
well
we'll
defer
till
then
so
you
get
back
and
then
this.
C
A
The
this
one-
and
there
was
an
example
I
think
that
he's
got
in
where
was
it.
A
I
think
it's
probably
landed
but
he'd
written
an
example
to
show
show
that
as
I
remember,
but
he
was
going
to
add
we're
leaving
this
open
because
he
was
also
going
to
add.
Like
another
example
like
I
can't
remember
if
he
did
the
note
add-on
example
or
the
raw
note
api
one,
but
he
was
going
to
do
the
other
one
as
well.
So
so
I
don't
think,
there's
necessarily
anything
to
talk
about
on
that.
One.
A
B
A
C
A
B
B
Segmentation
faults,
inconsistent
errors.
It
has
a
little
blue
line,
yeah
this
one
yeah.
So
this
is
the
one
that
I
was
sort
of
working,
trying
to
figure
out
what
was
going
on
and
the
original
poster
actually
merged
his
pr
into
master.
B
So
I
don't
know
if
they
were
able
to
identify
what
the
problem
was
or,
if
or
if
they're,
just
accepting
the
the
limitation
that
this
this
whatever
this
issue
was,
so
I'm
just,
I
guess
I'm
just
sort
of
waiting
to
hear
back
from
them
on
that.
A
A
C
C
C
A
C
Yeah
right,
I'm
posting
the
first
example
of
how
to
construct
a
henderscope
here,
yeah.
So
this
I
I
just
find
that
maybe
our
documentation
might
somewhat
be
misleading
to
newcomers,
especially
for
those
are
new
to
c
plus,
plus
that
they
cannot
distinguish
the
the
function.
Declaration
within
the
actual
example
usage
examples.
C
So
right,
maybe
we
should
add
a
examples
for
each
of
the
each
of
the
function
or
we
should
clarify
that
clarify
that
the
the
the
code
snippet
in
the
documentation
is
just
the
declaration
of
the
function,
which
is
not,
which
is
not
how
people
should
use
it
directly
right.
So
yeah
right.
So
that's
my
ideas.
B
I
think
it
would
be
worthwhile
just
to
add
a
simple
example,
but
what
I'm
sort
of
thinking
is
what
would
be
a
good
example
to
show
when
someone
needs
to
explicitly
create
a
new
handle
scope?
B
You
know
because,
like
if
you
think
about
you
know
just
say
you
have
a
native
function
that
you
put
on
to
your
module
exports
or
whatever
you
know
it
would.
It
would
be
really
simple
just
to
add
that
handle
scope
creation
in
your
function,
but
I
don't
think
that
would
actually
signify
like
a
good
use
case
of
right
when
the
handle
scope
needs
to
be
made.
B
B
A
B
Okay,
I
think
the
the
convention
that
we
have
is
that
we
put
the
examples
at
the
end,
but
I
think
it
would
be
good
to
explain.
B
Maybe
in
the
beginning,
why
I
mean
we
say
these
handles:
keep
an
object,
alive,
object,
making.
Okay,
oh
well!
There's
an
example
right
there.
B
B
B
B
Yeah
I
mean
maybe
we
can
just
put
that
bot
that
last
example
also
inside
the
handle
scope,
doc
and
then
maybe
that
would
be
enough.
A
Like
maybe
it
was
like
here's
an
example
of
using
it
and
the
example
is
explained
more
in
more
detail
in
this
link
right.
This
yeah.
B
B
C
B
A
Nothing.
Okay!
So
I
just
flipped
over
here:
let's
look
at
the
labels
yeah,
so
this
has
got
everything
too.
C
A
Well,
on
its
way
to
landing,
and
otherwise
I
don't
think
I
see
anything
particularly
new
to
pull
out
this
week.
A
This
one
all
mentioned
just
because
I
was
doing
some
experimentation
if,
like
for
for
containers,
there's
there's
some
experimentation
with
letting
have
like
wasn't
bare
metal
containers,
and
I'm
always
like
well
hey.
Why
don't
you
just
burn
your
wasm
unknown,
and
so
they
were
like
well,
hey
show
us
how
you
do
that
so
interesting,
the
the
proof
of
concept
that
was
doing
that
with
wasn't
time,
uses
c
run,
which
is
pure
c
code.
So
I
tried
to
start
to
integrate
things.
I'm
like!
A
So
I
actually
have
applied
this
patch
and
I'm
using
that
to
like
experiment
with
you
know:
loading
loading
node
as
a
shared
library
and
just
running
just
enough
script
code
to
to
load
some
wasm.
A
So
just
thought
I'd
mention
that,
as
like
I'm
actually
playing
around
a
little
bit
with
that,
and
it's
just
like
a
concrete
example
of
where,
like
the
c,
the
c
interfaces
really
make
things
a
lot
easier
in
some
cases,
because
you
can,
you
can
sort
of
deal,
load
them
and
load
the
symbols
versus
having
to
link
against
a
whole
bunch
of
things
and
absorb
the
build
chain
of
the
that
you
get
along
with
them.
A
C
Yeah,
I
mean
that's
an
interesting
use
case,
but
I'm
I
was
still
thinking
about.
How
do
we
keep
api
stability
for
platform
integration,
since
that's
most
of
the
engine,
setups
are
not
guaranteed
by
node.js
as
it's
maintained
in
the
v8.
So
that's
my
thinking.
Oh,
how
do
we,
if
we
land
this?
How
do
we
maintain
that
stability.
A
C
A
C
A
And
like
I
had
to
do
some
other
things
like
is
to
get
the
wasm,
I
had
to
read
the
wasm
in,
and
I
just
basically
poked
that
in
as
an
object
on
the
global,
the
global
object
and
then
the
script
that
I
run
pulls
it
off
and
has
has
enough
stuff
to
actually
run
the
the
the
javascript.
It
has
enough
to
run
the
wasm
through
the
javascript
that
runs,
but
yeah.
So
is
it?
C
That's
correct
about
the
execution
argument,
which
is
the
second
third
and
fourth
argument,
also
note
about
it:
it's
notable
that,
when
knowledge
is
encountered
on
record
unrecognizable
flags,
not
just
will
just
exit
with
an
error
and
saying
that
these
flags
are
not
recognizable
and
I'm
I'm
done
with.
I
don't
know
how
to
deal
with
it,
and
so
I
just
actually
know
the
record
so
yeah,
that's
a
another
and
things
that
are
just
notable
to
exposing
these
flags
to
a
api,
stable
api.
B
Yeah
the
api
would
be
stable
because
all
it
is
is
just
accepting.
You
know
an
array
of
strings,
so
it
it
like.
I
don't
think
this
method
would
change.
What
may
change
would
be
the
engine's
underlying
implementation
of
what
it
does
with
those
arguments,
and
I
think
that
was
something
that
was
brought
up
somewhere
in
in
the
the
vast
messages
of
the
conversation
in
the
pr,
and
I
think
that
was
sort
of
the
the
original
pull
request
or
his
his
design
implementation.
B
C
Yeah
that
that's
a
fair
point,
but
I
was
thinking
about
the
actual
use
case
out
here
since
actually,
since
we
declared
this
as
a
bi,
stable
and
people
might
just
pre-compile
the
binary
and
link
it
against
a
let's
say,
new
version
of
node,
since
it
may
just
the
node
version,
the
node
should
library
might
and
provide
a
higher
version
of
knowledge
version
which
which
we,
which
it
should
be
provide
a
similar
behavior
or
at
least
to
say,
compatible
behavior.
C
But
the
fact
is
not
just
is
going
to
reject
that
that
argument
and
just
complaining
that
this
argument
is
not
recognizable
since
it's
being
removed
in
v8,
so
the
binary
compiled
against
a
lower
version
of
another
another
api
version
will
not
be
able
to
run
with
a
higher
api,
not
not
just
shared
library,
with
declared
with
a
higher
node
api
version.
A
If
you
like
yeah,
I
mean
like
today.
If
I
I
could
write
an
add-on,
I
think
so
so,
just
like
I'm
just
like
the
the
the
you
know
the
the
example
code
that
I'm
running
now.
I
could
take
that
code
and
I
could
stick
it
into
an
add-on
today
and
then
that
add-on
would
run
or
not
run
depending
on
whether
node
had
been
started
with
the
right
arguments
to
allow
wasn't
it.
A
Yeah,
so
it's
you
know,
despite
using
the
despite
using
the
the
stable
avi,
there
still
are
like
how
you
run.
Node
itself
can
still
actually
have
an
effect
on
to
whether
you
know
may
basically
make
a
requirement
on
which
versions
of
node
you
may
or
may
not
run
with,
which
is
an
interesting
it.
A
A
And
so
yeah,
it's
a
it's
a,
I
think,
long
term.
It
would
be
really
good
for
us
to
have
some
sort
of
c
api,
whether
whether
it
makes
sense
as
part
of
the
the
node
ap
or
not
having
some
c
api
makes
it
much
more
consumable.
C
Also,
I
just
like
to
mention
that
this
will
be
part
of
the
embedding
embedding
api,
so
it
would
be
great
to
have
discussion
with
folks
that
are
working
on
them
yeah
and
have
experience
on
the
embedding
api.
Since,
like
say,
embedding
is
a
different
story
with
the
add-on
authoring.
C
Let's
say
electron
may
may
election
many
experience
on
the
embedding
and
they
have
a
lot
of
use
cases
on
embedding
and
they
might
have
and
thoughts
on
how
to
shape
and
easier
to
use
api
to
embed
node.js
in
their
application.
C
So
I
would
like
to
I
would.
I
would
like
to
express
that
how
likely
an
embedding
another
jazz
would
break
would
would
without
how
their
requirement
might
how
to
like,
say
it?
C
What
and
how
the
possibility
is
what
the
possibility
is
to
like
say
they
will
need
apis
other
than
know
they
don't
appear
to
in
in
order
to
embed
not
just
into
their
application.
A
Yeah,
I
know
I
agree.
We
should
be
good
to
get
the
the
simple
this
this
seems
to
like
what
was
interesting.
Is
it
seemed
to
work
for
the
simple
case,
which
is
all
I
needed
in
this,
and
what
I
was
experimenting
with.
It
was
just
like
run
a
very
simple
script
and
like,
in
fact
I
can
kind
of.
I
can
show
you
the
yeah.
C
As
a
first
step,
I
think
it's
very
interesting,
but
if
the
embedding
application
grows
day
by
day,
then
if
they
need
apis
other
than
know
the
adam
api,
then
things
will
change,
though
they
will
have
to
either
have
a
approach
to
convert
another
api
values
to
the
row.
V8
object,
or
rather
they
will
and
they
will
have
to
convert
all
their
existing
code.
That,
based
on
the
api,
know
the
api
embedding
api
to
the
to
the
to
the
original,
not
just
c,
plus
plus,
f
and
binary
yeah.
C
Buttonsome
to
this
embedding
authoring,
since
you
have
either
have
to
break
the
api
stable
guarantees
or
you
have
to
convert
your
code
base
to
the
to
the
more
powerful
to
the
powerful
api.
Instead,
yeah.
A
C
A
I
just
want
it
was
just
mentioning
because
it's
it
was
interesting
in
the
and
it
at
least
at
least
ran
the
simple
use
case.
I
have
to
test
some
more
things,
but
I'll,
probably
chime
into
the
issue.
Just
saying
like
hey,
I
tried
it
out
and-
and
I
think
I
think,
you're
right
like
trying
to
understand
what
beyond
like
this
is
a
nice,
simple
interface
and
it's
like
okay,
so
what's
missing
or
like
maybe
the
more
complicated
use
cases
would
be
a
good
thing
to
understand.
B
A
I
think
we
probably
want
to
reach
out
to
somebody
on
the
electron
team,
and
you
know
maybe
that's
like
you
know,
because
they're
mentioned
like
see
if
we
can
find
somebody
to
just
you
know,
talk
to
them,
maybe
invite
them
to
this
meeting
or
you
know,
have
one-on-one
or
whatever
to
see.
If
we
can
get
some,
they
may
be
able
to
say
well
yeah,
but
you
can't
do
blah
blah
and
blah
right.
A
B
A
A
Point
is
often
they.
They
will
also
need
much
more
control
over
things
like
v8,
but
then,
in
my
mind,
like
once
you're
once
you've
gone
that
far,
maybe
you
should
just
be
using
the
c
plus
one
right.
It's
there's
sort
of
the
line
between
there's.
A
We've
always
done
the
line
of
like
if
you
need
access
to
all
sorts
of
internals
and
v8
knowledge
in
particular
right
like
you're,
then
you
know
we're
not
going
to
expose
all
that
through
the
apis
necessarily,
but
if
you
have
the
the
80s
in
the
in
the
case
of
add-ons,
it
was
the
80
use
case
right
so
it.
But
I
think
what
the
discussion
is
here
is.
If
this
api
is
a
10
use
case,
then
it's
not
necessarily
as
valuable.
C
C
The
story
for
the
adam
authors
is
that
if
the
the
choices
are
made
to
decide
if
they
are
going
to
use
no
node
api
or
the
v8
api
that
if
they
are
going
to
have
any
hooks
into
via
internals,
if
not,
then
they
actually,
if
they
just
use
bridging
the
and
bridging
a
external
library,
to
the
no
not
just
then
great
they.
Maybe
they
just
need
to
create
a
javascript
values.
C
Then
no
data
is
the
perfect
choice
for
them,
since
they
are
not
likely
to
cross
the
line
to
to
to
to
touch
the
v8
raw
vps.
But
how
would
embarrass
what?
What
is
the
possibilities
for
embarrassed
across
the
line
like
say,
if
embodies,
have
have
a
great
possibility
to
cause
the
land
to
control?
However,
it
behaves.
A
Yeah,
I
think
it
kind
of
comes
down
to
like
if
all
you
want
to
do
is
run
some
scripts,
that's
different
than
possibly
some
of
the
other
ones
where
you're
you
know
you,
but
that's
where
I
don't
have
off
the
top.
My
head,
like
in
the
what?
What
beyond
running
a
script,
is
that
you're
going
to
want
to
do,
isn't
better
right,
yeah
and
getting
that
list
would
help
us
understand.
A
A
Running
the
script
there
there
is,
you
know
some
things
where
you
could
say.
This
doesn't
quite
feel
like
just
go,
run
my
regular
script
right
that
I
can
run
say
as
a
as
a
something
I
passed
a
note
itself,
but
that
that
might
be
a
little
that
I
mean,
I
think,
if
I
I
could
probably
get
around
that
like
if
we
can
get
the
eval
function
through
the
globals
or
whatever,
and
then
you
can
just
develop
something.
A
Agreed,
okay,
so
yeah,
I
thought
I'd
mention
it
I'll
chime
in
on
that
issue.
Just
saying
hey
like
I
was
doing
this
experimentation
and
I
needed
to
see
rapper,
so
I
tried
it
out
and
then
can
sort
of
maybe
after
that
I'll
we
can
see
if
there's
like
people
in
the
issue
or
whatever
that
we
could
sort
of
start
to
build
that
list,
because
I
think
that
would
be
interesting.
A
Okay,
anything
else
we
should
talk
about
this
week
before
we
close
out.