►
From YouTube: 2021-12-03-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
A
A
A
A
A
And
you
can
see,
like
kind
of
I
think
the
the
take
we
took
was
that
the
logical
state,
you
know
the
things
like
the
javascript
object
would
be
considered.
Part
of
the
logical
state,
logical
state.
The
original
poster
doesn't
seem
to
completely
agree
with
that.
A
You
know
I,
I
kind
of
think
it
makes
sense
to
me
in
that
it's
it's
still
like,
even
if
you
thought
of
like
the
pure
c-plus
plus
state,
if
it
was
something
that
was
part
of
the
state
of
the
object,
then
I
I
still
kind
of
you
know.
I
think
I
forget
who
it
was
who
pointed
out
the
last
time,
if
you
call
a
method
on
the
object,
call
a
constant
method
and
then
call
that
same
method.
Again,
you
know
you
and
nothing
else
is
run.
A
B
A
Right,
I
think
this.
This
is
where
I
think
I
tried
to
explain
that
one
right
like
so
basically,
but
instead
that
in
turn
result
in
a
different
answer
when
you
call
another
method
on
the
c
plus
object.
A
It
was
like
you
were
saying:
have
we
mentioned
the
the
aspect
of
like
calling
a
method,
getting
an
answer,
then
calling
a
const
method
and
calling
that
original
method
again
and
expecting
to
get
the
same
thing,
and
I
I
think
I'd
covered
it
in
that
particular
like
I've
highlighted
the
the
section
there.
But
if
it's,
if
it's
not
communicating
that,
maybe
you
could
add
another
comment
in
the
issue
to
call
that
out
or
yeah.
B
I
think
I
can
try
to
comment
and
state
my
point
of
view.
Okay,.
B
A
A
A
A
A
D
Yeah,
I
just
did
more
digging
into
this
issue.
Okay,
my
changes
merged
in
this
add-on
will
now
throw
an
error
saying
an
api
defined
property
fails
and,
after
looking
more
into
it,
looks
like
when
we
wrap
an
error
thrown
from
v8.
Using
that
error
wrapper
code
that
air
the
value
is
actually
a
type
not
b
null.
D
So
when
we
try
to
define
a
property
with
value,
not
be
not
on
an
object,
the
program
crashes,
but
if
we
add
a
check
in
the
error
wrapper
code,
basically
ignoring
areas
of
type
nappy,
no,
this
bug
will
disappear
and
like
it
will
work
inside
the
comment
inside
the
rebel.
So.
A
So
the
yeah!
So
if
I
go
to
no
api,
the
inl.
D
Yeah,
I
know
and
will
be
the
air
constructor.
Okay.
D
Yeah
this
one
so
pretty
much
that
value.
If
we
add
a
check
saying
okay,
this,
if
this
value
is
a
nappy,
no,
we
don't
actually
do
anything
to
it.
Then
just
then
we
sort
of
swallow
this
error
from
from
v8
and
then
repo
using
that
command
line
thing
will
work,
but
I'm
not
quite
sure
if
that's
the
intended
behavior.
B
D
Oh
yes,
but
like
this
check,
is
to
make
sure
that
val
isn't
a
null
pointer
like
a
c
plus
less
than
a
pointer
but
right
well,
the
value
like
it
could
be.
I
have
a
value,
but
it's
a
type
sort
of
nappy
node
inside
of
the
v8
yeah,
like
that,
okay.
D
D
I
believe
so
I
think
it's
I
think
it's
I
think
it's
from
after
we
call
the
function
from
a
function,
call.
B
Then
it's
possible,
it's
possible
to
run
a
normal
javascript
program
to
throw
a
null
value
as
the
exception
right.
So
we
can't
do
anything
when
the
code
exception
value
is
a
not
now.
A
B
A
D
Yeah
the
reference
works,
but
it
fails
when
we
try
to
attach
that,
like
an
apple
property,
descriptor
using
nappy,
define
properties
like
right
as
as
of
now.
D
When
I
like,
it
looks
like
it's
like
the
status
that
returned
when
we
try
to
attach
this
flag
to
the
descriptor.
Is
that
it's
it's
an
invalid
argument?
So
it's
probably
within
the
c
plus
that's
not
a
c
api.
That's
giving
us
a
narrow
status.
D
Right,
yeah,
you
like
nap
modern
properties,.
C
Why
do
we
need
the
value
from
here?
Because
we
already
have
an
api
value?
So
why
don't
we
just
say
value
here?
Instead
of
value
from
and
the
value.
C
Like
we
have
a,
we
have
a
we
have
a
c
structure
right,
which
is
which
is
then
getting
passed
to
the
core
api
right.
If
I,
if
I
remember
correctly
at
that
field,
the
c
structure
expects
an
napi
value
right.
We
already
have
an
api
value,
then
why
do
we
have
value
call
and
column
from
and
the
value?
Why
not
just
value
here.
D
Yeah,
I
think
yeah
that
might
be
unnecessary.
The
conversion.
C
A
A
C
B
If
I
understand
correctly
in
the
repo,
then
we're
previewing
values
and
the
defined
property
will
fail
for
the
this
larger
script.
Execution
scope
when
the
field
is
detecting
side
effects.
Just
like
why
we
got
the
the.
Why
we
run
into
this
error
is
because
we
cannot
run
any
javascript
javascript
script
in
the
in
the
scope.
B
C
Yeah,
that's
that
yeah
yeah
yeah.
What
lucas
is
saying
is
definitely
correct,
but
I
think
also
if,
if,
if
the,
if
the
value
is
null,
then
an
api
defined
properties
may
also
fail,
and
I
think
we're
on
the
right
path
with
checking
the
implementation,
because
I
think
at
least
one
of
those
things
has
to
be
non-null,
because
otherwise
you,
you
cannot
define
a
property
where
everything
is
known.
There's
no
method,
there's
no
accessors,
there's
no
value!
There's
there's
nothing!
C
A
C
C
You
know
if,
if
javascript
execution
is
not
allowed,
then
I'm
not
sure
that
we
can
do
much
honestly
right
because,
like
yes,
you
can
take
that
exception
and
report
it.
How
javascript
execution
is
not
allowed,
so
you
cannot
return
to
javascript
and
and
have
the
exception,
be
thrown
and
processed
and
stuff.
B
B
E
B
Okay,
I
just
checked
that
when
we
creating
the
native
method,
we
have
a
flag
that
tells
invade
our
method
has
said
effects.
I'm
not
sure
why
we're
still
calling
the
native
mess
negative
methods
for
previewing,
that's
what
I'm
not
sure
and
it.
If
there
is
a
problem
with
it,
we
have
to
fix
it
in
the
core.
C
B
I
was
mentioning
that
when
we
creating
the
np
function,
we
should
create
the
corresponding
with
function
right
when
we
creating
bit
function.
There
is
a
parameter
to
set
if
this
function
has
side
effects,
and
we
have
already
set
the
flag
to
be
to
tell
viet
that
that
this
function
has
set
the
effects
and
va
should
not
call
this
function
when
previewing
in
the
wrapper.
C
B
A
A
A
B
A
B
Next
step
is
sure,
but
it's
worth
investigating
ours.
Maybe
I
can
investigate
it.
Okay,
with
my
best.
C
Well,
one
thought
that
comes
to
mind.
Is
you
know
what,
if
the
rebel,
doesn't
heed
the
the
fact
that
we
said
that
the
it
doesn't
heat
the
flag
it
caused
the
method
anyway,
even
though
we
said
it
has
side
effects
right
like
the
ripple
implementation
is
on
top
of
v8.
Does
it
consult
v8
saying?
Is
it
okay
to
call
this
method
for
preview
purposes
if
it
doesn't
consult,
v8
and
unconditionally
cause
the
method,
then
then
that
could
be
a
problem
with
the
with
the
v8
implementation.
B
A
A
You
just
take
a
quick
look
to
see
the
general
trend,
so
it
looks
like
we
went
up
and
then
we
went
back
down
so
not
quite
sure.
The
result
of
that,
but
still
5
million
downloads
a
week
is
pretty
high
number.
Let's
see,
let's
look
at
the
next
one
is
the
drive
towards
full
coverage
so
jack.
Any
updates
on
that
front.
D
Think
is
it
it's
a
word
doc,
but
I
can
put
in
like
a
google
docs,
if
possible,.
A
D
D
A
A
A
A
A
A
B
A
Which
I
think
was
like
when
I
opened
the
dock,
because
I
didn't
have
the
word
perfect
word
installed.
That
created
me
a
doc
so
anyway
shared
share
that
with
everybody,
and
I
made
a
few
songs
like
I
made
a
few
suggestions
and
probably
other
people
should
be
able
to
get
in
there
and
and
make
suggestions
as
well
and
then
in
the
end
we
can
end
up
with
a
doc.
We
can
ask
to
be
published
thanks.
A
Okay,
so
that's
on
that
front.
I
have
not
done
that
d-pack
anything
you
want
to
discuss
on
the
the
work
in
terms
of
being
able
to
more
easily
run
the
tests.
E
Actually
kevin
had
a
inference.
He
said
that
the
test
should
be
exiting
with
the
error
code.
If
the
unit
tests
are
failing,
he
was
saying
that
the
exit
code
right
now
is
zero
and
it
should
be
changed
to
one
so
definitely
yeah
yeah
so
was.
E
Well,
that's
the
cia
part
right,
that's
the
actual
unit
test
when
you
run,
but
the
when
running
single
test
right.
A
A
Okay,
okay,
anything
else
you
want
to
talk
about
on
that
front.
E
No
nothing
else.
Okay,.
A
B
A
In
the
wrong
place,
actually
yeah
there's
one
more,
but
it
would
be
good
actually
to
come
and
look
there
for
a
second
so
that
in
a
second
though,
no
don't
be
api.
C
Yeah
this
I
this
may
be
the
case
that
they
are
both
correct,
because
I
remember
adding
overrides
where
you
can
pass
the
the
getter
and
the
setter
as
a
template
parameter
to
to
reduce
stack
size.
C
C
Yeah
we
have,
we
have
an
override
where
that,
where
the
the
getter
and
the
setter
are
passed
as
template
parameters
and
the
setter's
default
is
null
pointer,
and
so
it
is
possible
to
write
code
where
the
getter
is
getting
passed
as
the
sole
template
parameter.
C
C
A
C
We
have
them
documented
it's
in
it's
in
it's
in
instant,
it's
in
instancewrap
md,
okay,
so
let
me
find
that.
C
C
That's
true
yeah
there
should
be
at
the
top.
There
should
be
like
a
like
a
class
hierarchy
type
thing.
A
C
B
C
A
C
Yeah
so,
okay,
I
mean
the
type
exists,
yes,
but
the
type
is
produced
using
methods
that
that
are
class
specific,
like
instance,
accessor
returns.
One
of
these
yes,
but
you
do
not
create
these
directly.
You
create
them
using
instance,
accessor
and
instance
value
and
and
static
access
or
static
value.
And
that's
what
creates
these
things.
So
it's
I
don't
know
that
it's
possible
to
use
them
directly
so
like.
C
If
you
look
at
the
call
to
define
class,
yes,
you
could
have
like
a
a
property
descriptor
defined
at
the
top
and
then
pass
it
to
define
class.
But
everybody
just
does
it
in
line
so
you
you
know,
you
could
store
a
value.
You
could
store
the
return
value
of
instance,
accessor
and
then
pass
that
return
value
into
defined
class.
If
you
really
really
wanted
to.
A
C
Yeah,
I'm
not
sure,
I'm
not
sure
that
now
that
I
read
this
more
in
detail,
I'm
not
sure
that
it
makes
sense
to
mention
that
it
is
a
subclass,
because
it's
not
a
subclass
of
anything.
It's
it's
a
it's
kind
of
a
helper
class,
that's
defined
in
object,
wrap
and
and
and
it's
templated
to
the
subclass
so
that
they
don't.
C
They
have
no
common
types
for
different
object
wraps,
so
that
so.
C
C
C
A
C
Yeah,
okay,
I
think
the
way
we
could
open
this
this
page,
this
class
property
and
descriptor
or
this
class
property
descriptor-
would
be
to
say
that
this
is
a
helper
class
produced
by
instance,
accessor
and
then
linked
to
instance,
accessor,
and
you
know
that
would
basically
keep
this
documentation
valid,
but
it
would
sort
of
strongly
hint
that
people
should
be
using
instance
accessor
to
produce
these
things
rather
than
produce
them
by
hand.
C
Should
say
to
open
the
section
by
saying
that
this
is
a
helper
class
produced
by
the
the
object
or
the
instance
wrap
methods
such
as
instance,
accessor
instance,
value
and
so
forth.
C
A
Yeah,
would
you
possibly
be
willing
to.
A
Sorry,
let's
I
just
want
to
just
click
on
that
one
nope.
A
C
Yeah,
I'm
basically
asking
them
to
just
handle
their
own
exceptions,
because
we
don't
know
how
to
translate
any
old
standard
exception
into
an
api
error
that
you
know
we
I
was
thinking
like.
We
could
write
a
handler
for
it
and
just
use
the
what
method
to
produce
a
string
and
then
turn
that
into
an
api
error.
But
if
somebody's
not
happy
with
that,
now
we're
going
to
have
to
give
them
a
way
to
override
that
and
produce
their
own.
So
they
might
as
well
just
produce
their
own
to
start
with.
A
A
Okay,
so
this
is
like
nappy
apis
should
handle
out
of
memory
better,
so
this
was
quite
old.
C
So
for
for
for
this
situation,
this
pervasive
use
of
new,
we
actually
had
a
leak
related
to
this
like
two
years
ago,
where
we
were
creating
the
the
the
wrappers
for
for
moving
or
for
promoting
an
api
call.
We
were
receiving
into
a
node
add
on
api
call
to
the
user
and
those
things
needed
to
be
free.
They
needed
to
get
attached,
I
think,
to
the
function
so
that
they
get
freed
along
with
the
function,
and
I
think
that
that
addresses
some
of
the
leaks.
A
Right
or
you
know,
handle
it
and
handle
it
at
ad
hoc
is
kind
of
the
we
haven't
done
anything
for
three
years
going
on
four
years,
so
I'm
just
wondering
like.
Is
it
useful
having
this
still
be
open.
C
Yeah
yeah
handling
out
of
memory
is,
is
pretty
much
ad
hoc,
no
matter
what,
because
you
might
run
out
of
memory
just
trying
to
handle
out
of
memory
since
you're
out
of
memory.
You
know
it's
one
of
those
situations
right
so
I
mean
I
don't
know
that
we
can
handle
this
in
place.
If
we
can
handle
this
in
place,
then
maybe
but
it's
it's
a
it's
a
long-winded
investigation.
I
suspect
right.
C
A
C
C
C
C
A
C
C
C
C
B
B
C
A
And
we
are
at
time
actually
we're
overtime,
so
thanks
everybody
for
sticking
around
a
little
bit
longer,
but
maybe
we
should
close
out
this
week
and
we
will
maybe
continue
to
go
through
the
list
next
time.