►
From YouTube: 2021-06-11-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
Sounds
like
not
so,
let's
move
on,
I
will
share
my
screen
here
with
that
list,
so
the
first
one
is
to
look
at
any
new
stale
issues.
You
know
kevin
you'd
added
this
one
it
looks
like
did
we.
B
Yeah
we
talked
about
that
one
last
week.
I
think
legend.
A
Okay,
so
I
guess
that
one
we'll
just
leave
on
we've
and
we'll
see
you
next
week,
where
we're
at
with
it.
A
Okay,
tracking
issues
for
a
module
is
no
to
supported.
I
don't
think
I've
seen
anything
new
on
that
front.
Anything
anybody
wants
to
bring
up
on
that.
One.
A
Debug
testing-
I
know
I
haven't
done
anything
on
and
so
we're
done
to
issues.
So
are
there
any
issues
that
people
want
to
call
out
and
have
us
discuss
this
week.
C
This
one
right,
we
can
stability,
testing
the
garbage
collection
crash
with
this
test,
and
we
can
verify
that
we
it's
your
recently.
Your
recent
pr
have
been
have
fixed
the
issue
with
the
tests.
D
Right
just
where
yeah
right
here.
C
It's
very
tricky
to
get
the
engine
running
like
what
you
expect,
especially
on
the
gc
port,
so
I
just
like
say
directly
calling
into
the
big
callback
since
the
api
contracts
of
the
week.
Callback
is
that
the
embedder
has
the
responsibility
to
reset
the
proceed
persistent
handle
so
that
that
so
this
test
will
work.
Fine,
as
we
expected.
C
E
A
Right,
okay,
so
yeah,
that's
interesting!
The
the
question.
The
one
question
I
had
was
around
the
use
of
the
like:
it
adds
a
new
like
a
new
test
framework
right.
C
Yeah
we
no
the
test
framework
is
already
there,
but
we
are
adding
a
new
production
headers
for
gtask.
It's
a
very
simple
line
of
friends,
a
friend
declaration,
it's
in
the
bottom
of
the
file.
Okay,
so
just.
C
E
C
Test
cc
test:
oh
cc,
test;
okay
right,
sorry,
all
right
test
test
test
and
the
cc
test
right
cc.
A
D
A
C
C
So,
with
the
g-test
underscore
production,
we
are
going
to
distribute
the
file
content
with
our
binary,
so
I've
just
added
a
license,
but
if
you
that
that
might
be
not
maybe
not
mandatory,
I
think.
C
A
A
A
C
C
C
C
A
C
A
A
A
A
C
C
A
E
So
this
this,
the
the
the
declaration,
may
be
better
placed
into
js
native
api
v8
internal
dot,
h
because
v8
dot
h
is
kind
of
it's
kind
of
stuff
that
can
be
used
outside
of
node
right
and
we
don't
want
to
include.
We
don't
want
to
force
everybody
who's
using
v8
based
node
api
outside
of
node
to
also
have
g
test,
so
the
hash
include
may
be
better
placed
in
in
gs,
js
native
api,
v8,
internal
dot,.
E
E
A
A
D
C
C
A
E
Okay,
so
does
it
the
fact
that
it's
not
in
an
anonymous
namespace?
Does
this
add
symbols
to
the
node
binary
which
are
like
visible
using
like
nm?
You
know.
E
Yeah,
I
I
don't
remember
if
we
build
with
visibility,
it
was
hidden
or
not,
because
we
do
have
these
macros
like
a
node
or
an
api
external.
I
think
which
explicitly
declare
a
visibility
of
default
so
that
it
can
be
used
from
other
other
shared
libraries.
A
C
I
checked
about
the
js
v8
js
native
api
v8,
dot
h,
there
is
like
say,
left
checkers
and
then
not
p.
L
and
many
others
are
already
in
the
viet
import
name
space
regarding
without.
C
A
C
E
E
A
A
C
I
I
remember
that
when
I
first
put
the
declaration
declarations
in
the
v8
internals,
there
are
two
key
problems
that
we
have.
The
include
order
of
our
header
file
is
that
we
first
include
via
the
internals
and
the
other
files,
so
it
puts
them
in
the
va.
Internals
cause
a
lot
of
ordering
issue
in
the
header
declarations.
C
C
C
A
C
C
C
A
Yeah,
no,
I
yeah.
If
these
were
like,
because
I
know
we
don't
do
as
much
of
a
like
within
node.
We
don't
necessarily
hide
things
as
much,
but
I
think
it's
better
not
to
have
it
exposed
yeah,
okay,
just
understanding
that
that
that
all
makes
sense
to
me,
I
guess
the
one
thing
you're
going
to
look
at
was
making
that
move.
Otherwise,
there's
any
other
questions
for
people.
B
Yes,
so
on
node
add-on
api
there
was
the
most
recent
issue
about
the
gcc
11
with
c
plus
plus
20.,
and
I
had
to
get
a
bug
account
created
for
for
the
gcc
bugzilla.
So
I
have
that
account
created
and
now
I'm
going
to
paste
create
that
bug.
So
my
thought,
though,
is
that
we
would
just
update
our
set
up
help
set
up
markdown.
B
A
A
B
A
Any
other
ones
that
people
want
to
call
out
to
discuss.
C
And
I
found
that
michael
you
have
raised
some
concerns
regarding
the
existing
change
yeah,
which
number
was
that
that
was
the
number
is
9
27,
nine,
twenty
seven,
okay,
yeah.
D
D
C
We
have
just
to
remove
the
black
bracket
right
right
here,
so
I've
just
added
a
wrap
up
document
in
test
readme
dot
mv
is
to
describe
how
we
writing
tests
for
different,
build
flavors
to
help
them
work
in
with
a
sim
same
source
code,
and
I
believe
there
is
nothing
nothing
special
to
do,
but
to
ensure
that
we
have
to
how
to
make
sure
each
code
to
each
code
link
in
I'm
sorry.
C
That's
the
core.
Okay,.
A
C
A
C
Was
it
we
don't
have
to
wrap
values,
but
we
are
just
going
to
unwrap
in
our
test
code
wrapping
it
down
in
the
in
the
source,
let's
say:
nappy,
dot,
h
and
then
bl
dot
h
in
the
test.
We
only
do
unwrapping.
A
C
The
problem
here
is
that
we
we
can
make
note.
I
know
api,
accept
user
functions
that
may
be
maybe
of
value
so
that
you,
we
don't
have
to
unwrapping
ourselves
in
our
code
I
can
make,
but
I
believe
that
will
not
affect
api
stability.
So
it's
a
very
new
construct,
signature
or
another
api
functions
so
that
we
can
declare
another
api
from
not
functions
with
returning,
maybe
value,
and
we
can
unwrapping
maybe
to
back
into
the
javascript.
A
C
The
coolest
thing
is
that
when
we
call
into
javascript,
then
it's
my
throwing
so
we
have
to
handle
that
handle
it's
that
written
value
properly
anything
other
functions
like
say
we
get
the
integer
values
of
a
number
type
that
didn't
call
into
javascript.
Then
we
don't
have
to
handling
those
bunch
of
things.
E
Basically,
basically
anything
that's
napi
preamble
on
the
core
side.
It
needs
this,
this
sort
of
handling.
So
then
you
you
can
you
want
to
if
you
want
to
make
sure
everything
that
is
covered?
If,
if
you
want
to
make
sure
everything
is
covered,
then
you
can
just
look
at
all
the
napi
preamble
apis
that
we
have
in
core
see
where
they
get
called
from
here
and
then
make
sure
that
those
calls
use,
use
the
structure
when,
when
exceptions
are
off,
so
that's
how
you
check
for
for
completeness.
A
C
Yeah
right
since
we
don't
have
maybe
type
in
like
say,
maybe
not
enabled
when
cpc
plus
plus
exception
is
ignored.
So
we
have,
we
don't
have
to
access
the
member
functions
or
maybe
so.
We
have
to
declare
a
global
function
for
that
and
it
doesn't
do
nothing
when
maybe
it's
not
declared
and
maybe
not
enabled.
A
C
A
Sounds
good,
I
think
I
mean
there's
lots
of
changes,
but
in
a
lot
of
cases
the
changes
are
just
like
this
right,
which
is
pretty
straightforward,
yeah,
it's
basically
wrapping
this,
which
will,
instead
of
being,
I
guess
some
of
those
may
be
a
little
bit
more
complicated,
but
some
of
them
just
add
this.
Maybe
your
value,
in
which
case
it's
pretty
obvious.
What's.
C
A
C
C
C
A
A
C
It
added
a
new
parameter
to
declare
the
written
type
of
the,
maybe
right,
okay,
so
that
is
when
the
maybe
is
not
enabled
we
can
return.
The
tab
default
default,
constructor
is
to
say
it's
written,
an
empty
of
the
bullying
right.
A
C
A
A
Okay,
any
other
issues
that
people
want
to
bring
up.
A
C
A
D
C
So
we
previously
declared
the
object,
get
property
names
as
a
const
function,
but
it
does
changes.
It
does
have
sad
effects
in
especially
in
the
javascript
side.
So
I'm
not
sure
the
original
intention
of
the
cons
qualifier
so
and
there
I
mean
they
are
not
const
in
in
the
meaning
of
javascript
style.
So
I'm
not
sure
people
are.
A
A
E
Right,
I
think
the
idea
here
is
that
the
the
the
object
whose
property
names
you're
getting
is
not
changed.
So,
yes,
there
may
be
side
effects,
but
those
side
effects
need
to
be
checked
using
by
different
means.
Right,
like
you
need
to
check
if
an
exception
was
thrown
so.
C
Right
I
mean
in
javascript
you
can
do
anything
in
the
getter,
so
the
object
itself
can
be
changed
using
getters.
E
Well,
okay,
but
I
think
I
think
we
we
should
restrict
this
to
to
just
looking
at
the
c
plus
side,
because
yes
you're
right,
but
but
I
mean
I
don't
think
we
should
try
to
attempt
to
capture
like
javascript
concepts,
because
for
the
purposes
of
this
function
call
right.
The
object
is
not
changed,
it's
the
same
object
right,
and
so
if,
if
there
are
changes
on
the
javascript
side,
then
then
you
know,
then
then
that's
a
that's
a
javascript
speciality
right.
E
A
A
Const
should
mean
that
it
wouldn't
change
that
pointer
to
the
other
object
right
in
the
c,
but
would
it
imply
that
that
other
object
would
not
have
any
changes
in
you
know
if
I
did
something
like
I
dereferenced
that
that
pointer
and
I
did
a
plus
plus
because
it
had
an
increment
function,
would
that
consider
would
would
that
be
considered?
Breaking
the
consciousness.
A
A
So,
even
though
it's
like
a
it's
a
pointer
as
opposed
to
like
because
you're
not
changing
the
object
itself,
you're
changing
something
that
it
points
to.
E
Yeah
yeah,
I
know
what
you
mean.
I
I
think
the
the
compiler
might
be
smart
enough
to
to
track
you
down,
but
yeah.
I
think
where
we
break
the
chain
is
with
the
core
node
api
call
right,
because
the
the
the
the
core
node
api
call
does
not
accept
the
const
and
api
value.
It
accepts
a
plain
old
and
api
value
right.
A
A
E
A
E
Basically,
if
we
wanted
to
be
absolutely
pedantic
about
this
and
we
wanted
to
enforce
this
const-ness
on
the
javascript
side,
what
we
would
do,
I
suspect,
is
you
know
we
would,
we
would
freeze
the
object
and
then
we
would
unfreeze
the
object
before
we
return.
That
would
be
enforcing
constantness,
but
I
mean
that's:
that's
huge.
A
E
Yeah
yeah
yeah
yeah
yeah
yeah,
if
yeah,
so
no,
that
would
be
a
huge
thing,
but
that
that
that's
that's
that's
that's
what
we
would
be
doing
if
we
wanted
to
really
truly
enforce
this
concept
of
const-ness.
Even
you
know
extending
it
to
the
javascript
side,
but
I
don't
think
we
want
to
do
that.
So
I
I
don't
know
that
we
want
to
do
this
either.