►
From YouTube: 2022-10-13-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
API,
the
node.js
node
API
team
meeting
for
October
14
2022
will
follow
our
standard
approach
of
looking
at
the
issues
which
were
tagged
for
Milestone
11
I
will
just
bring
that
up
while
I
do
that
anybody
have
any
announcements
they'd
like
to
share.
A
No
I
guess
related
is
there
is
a
major
release
for
node
next
week,
node.js19
and
I
think
I've
seen
Kevin
was
suggesting.
Maybe
we
should
plan
for
a
release
for
node
API
as
well,
so
that
that
makes
a
lot
of
sense
to
me.
Okay,
let
me
share
screen
where
we
share.
Okay,
the
right
button.
A
B
C
Ahead,
yeah
UMass
is
not
really
yet
sorry,
but
I
feel
that
in
this
we
can't
I
can
complete
the
the
work
here.
Okay,.
C
I
I
missed
the
last
meeting,
but
I
was
thinking
in
the
the
same
yeah.
C
In
the
same
page,
we
could
have
two
sections
one
dedicated
for
the
yeah
different
binding
for
node.js
and
another
section
for
the
runtime
that
actually
Implement
Implement
node
DPI.
So
we.
A
One
section
but
yeah
that
makes
sense:
okay,
sounds
good.
Better
coverage
in
YouTube.
C
C
What
do
we
we
made
was
in
2020
I,
think
that
was
a
talk
that
Gabriel
and
Kevin
did
so
essentially
the
concept
yeah.
Our
are
good
for
me.
A
Yeah
and
I
think
they
need
new
ones
you
create
if
they
can
be
sort
of
Timeless,
because
the
apis
don't
change
that
much
right
so
like
once,
if
there's
a
hello
world,
one
just
focused
on
that
you'd
think
that
that
should
then
be
good
for
quite
a
long
time.
Right.
B
A
Okay,
lots
of
good
ideas,
and-
and
so
yes,
it
sounds
like
you're
working
on
it.
If
anybody
else
has
any
ideas
or
time
to
work
on
some
videos,
too
sure
there's
lots
to
do
there.
So
full
coverage
I
know
that
there's
been
some
work
there,
just
I,
don't
know
if
it
was
discussed
last
week.
Are
there
any
like
PRS
or
whatever
that
that.
C
B
A
Okay:
okay,
I'm,
just
looking
back
to
yeah.
A
See
there's
a
number
of
new
ones.
So
let's
just
look.
C
I,
don't
know,
if
could
be
a
good
thing,
have
a
discussion,
no
yeah,
because
yeah
I
won't
that
yeah
I
I
would
like
that
no
DPI
become
a
standard
right.
Yeah
write
native
code
for
for
JavaScript.
So
it's
a
I.
C
We
can't
maybe
share
ideas
with
all
other
people
that
now
are
implementing
and
no
DPI
for
a
different
runtime
yeah
like
Vladimir.
You
know
that
he
implemented
no
DPI
for.
A
C
D
So
it's
my
main
goal
is
to
have
react
native
support,
node
API.
Yes,
so.
C
Yeah,
it
could
be
good
yeah,
share
ideas
or
maybe
problems
that
people
have
in
a
different
runtime,
yeah
and
United
run
time.
So
right,
some.
C
A
Because
node
never
turned
on
the
the
thought
that
feature.
D
So
look
up,
you
guys
understand.
What's
the
issue
here,
like
sorry,
I
didn't
look
into
deep
into
this.
Are
they
talking
like
our
implementation
of
array,
buffers,
not
external
array,
buffers
not
working
correctly,
maybe
with
something
we
can
do
inside
of
our
implementation
to
address
it.
A
Yeah,
although
that's
like,
without
not
wanting
to
copy
data,
is,
is
certainly
a
use
case.
A
A
B
B
An
alternative
will
be
that
they
will,
they
can
allocate
a
they
can
allocated
everybody
in
the
bathroom
from
the
Via
from
the
engine,
and
then
they
can
use
that
that
block
of
memory
to
educated
their
own
data
so
either
they
will
copy
into
that
into
this,
and
everybody
from
the
engine
or
they'll
use
this
battery
to
the
their
external
usages
So.
Eventually,
they
will
have
to
avoid
it
using
the
external
Rebecca
when
they
would
like
their
apps
to
be
compatible
with
the
memory
cage.
D
D
A
B
B
I'm
doubting
that
if
that
would
be
possible,
since
let's
say
we,
we
can
compare
known
API
modules
ahead
of
the
time.
Well,
we
don't
know
if
it
can
be
run
if
we,
if
it
is
running
on
the
node.js
or
the
election.
So
when
we
compare
the,
we
don't
know
about
how
user
use
it,
if,
if
user
will
use
it
in
node.js
or
use
it
in
networking
or
even
use
it
in
node.js,
with
the
memory
cage
enabled
so
I
I,
don't
think
it's
the
information
it
is
available
at
the
time
when
building
that.
B
A
So,
but
if
this
is
already
an
existing
method,
nappy
create
buffer
copy
yeah
right,
that's
right!
Then
you
can
do
it
like
I
guess!
That's
you
know
it's
that
if
you
want
to
write
your
app
so
that
you
create
a
copy
instead
of
using
an
external
resource
directly,
you
can
do
that.
Although
it's
you
know
that
the
reason
external
resource
is
there
because
it's
handy
not
to
have
to
copy
the
data
right.
A
B
I
believe
I
believe
the
program
is
not
specific
to
electron.
When,
let's
say
when
people
would
like
to
use
node.js
as
a
when
they
would
elect
to
you
know
just
as
a
standard
box,
they
don't
want
the
the
user
code
to
access
the
memory
outside
the
engine
hip.
Then
they
can
enable
the
memory
cage
for
node.js2.
You
know,
since
it's
a
security
mechanism.
A
I
I
I
know
what
you're
saying
is
it's
not
like.
We
wouldn't
necessarily
call
this
option.
You
know
limit
to
electron
apis,
but
it
could
be
something
like
limit
to
I.
Don't
know
whatever
the
right
name
would
be
right
so
that,
if
you
because
I
I
don't
know,
if
there's
anything,
we
can
do
other
than
let
authors
opt-in
and
give
them
some
feedback
to
say.
A
I
was
just
thinking,
that's
something
concrete.
We
could
do
you
know
add
another
Rift
def
just
like.
If
you
say,
hey,
I
only
want
to
use
node
API
version
1.0,
not
one
of
the
later
ones,
because
I
want
to
support
these
versions
of
node
right.
This
would
be
a
similar
okay,
I'm
going
to
Define
this,
and
it
basically
tells
the
the
build
process
that
I
want
to
support
a
subset
of
the
environments.
A
B
D
So
support
support
sorry,
guys,
like
a
little
bit
slow
I
feel
like
the
support.
I,
don't
understand,
okay,
you're,
absolutely
right!
We
do
have
today
mechanisms
to
create
buffer.
Then
we
have
mechanisms
to
retrieve
its
buffer
infant
people
can
copy
any
data
inside.
But
I
don't
understand
why,
in
this
Blog
the
proposing
to
do
to
kind
of
allocate
data
using
this
different
method
to
create
external
resource
and
then
introduce
a
new
method,
an
API
create
buffer
copy,
which
we
don't
have.
D
D
A
So
that
I
think
I
think
what
they're
saying
is
like
here's.
Your
work
around
right
use
this
other
API,
which
will
I,
mean
it's.
It's
not
really
a
workaround
but
like
because
there's
some
use
cases
where,
like
you
know,
passing
in
a
buffer
pointer
to
existing
memory.
Like
your
your
that's
a
way
to
actually
communicate
like
your,
whereas
in
this
one
in
the
copy,
it's
like
I,
don't
know
whether
you
can
actually
start
to,
then
you
know
modify
that
copy
data
or
at
the
very
worst.
You
have
this.
C
D
Right
but
I
just
trying
to
understand
like
what
does
it
achieve
like
what
what
it
gives,
so,
why
not
just
use
normal
API
create
buffer.
A
D
But
my
understanding
work
out
what
makes
it
then
very
incompatible
that
they're,
adding
this
new
method
to
allocate
resource
right.
A
D
So
what
what
was
an
incompatibility
becomes
just
because
what
they're
saying
if
somebody
creates
a
module
they're
using
Create
external
buffer
and
this
approach
does
not
work?
Is
it
what's
happening.
A
Exactly
like
I
think
fundamentally,
the
problem
is
that,
if
you're
in
a,
if
you're
in
a
64-bit
process
that
pointer
can
point
to
anywhere,
whereas
when
you
turn
on
this
pointer,
cage
and
compress
pointers,
the
V8
can
only
refer
to
something
within
a
smaller
subset
of
the
address
range,
and
so
now
you've
got
a
pointer
to
something
which
is
potentially
outside
that
range.
So
it
just
doesn't
know
how
to
deal
with
it.
D
But,
fundamentally,
what
we
can
do
we
can
do
exactly
what
they're
shown
here.
We
can
do
it
inside
of
our
create
external
buffer,
so
thinkly,
but
again
I,
I've,
I'm
working
on
this
creating
features
and
I
try
to
apply
these
features
everywhere.
So
what
is
here
this
feature
would
say
like
whenever
you
use
a
create
external
buffer
in
this
case,
if,
with
this
feature
for
cage
occasion
available,
what
will
do
for
you
internally,
we
will
do
this
copy
right,
like
we
will
in
turn,.
A
B
A
A
There
there
are
methods
to
get
like
the
underlying
pointer
and
stuff
like
that,
but
but
like
if,
for
example,
I'd
memory
mapped
a
file,
you
know
the
the
old
code
would
would
let
JavaScript
see
whatever
updates
were
made
to
that
file
without
having
to
do
anything
right.
Yes,
in
the
second
code,
we've
created
a
copy,
so
you
won't
see
any
changes
to
that
underlying
memory.
B
A
Yeah
and
it's
you
know
so
it's
it's
it's
down
to
like.
What
can
we
usefully
do?
We
could
potentially,
like
you
said,
have
a
feature
that
says
like
hey
I'm
going
to
opt
in
to
not
be
create
external
buffer
is
actually
immutable
or
something
so
you're
giving
us
something
that's
immutable.
Therefore,
we
could
create
a
copy
right
if
you're
already
going
to
set
the
flag.
A
Does
that
really
you're
gonna
have
to
do
something
in
your
application
right.
So
does
that
really
help
more
than
just
saying,
well,
you're
going
to
have
to
refactor
to
call
this
create
buffer
copy
I'm,
not
so
sure,
but
the
simple
thing
of
like
pretty
pretty
simple
for
us
would
be
to
add
in
that
extra
Define,
and
that
would
help
people
at
least
identify
when
they're
doing
something
that
might
run
into
the
problem.
D
A
D
Want
to
have
some
kind
of
function
which
would
allocate
memory
inside
of
V8
everything,
because
we
want
to
do
like
a
disable
kind
of
exclude
from
API
said
this
function.
So,
let's
just
record
it
as
we've
kind
of
discussed
it
and
yeah.
D
It
was
interesting
because
I
was
I
was
planning
to
introduce
something
like
external
strings.
Do
they
play.
A
D
Yeah
the
question
like
what,
if
we,
you
want
to
use
node
in
some
way
kind
of
node
API
in
some
specific
environment
where
you
allow
to
have
this
external
buffers,
like
you
said,
imagine
you
have
some
kind
of
resource
inside
of
your
application.
You
just
want
to
memory
map
to
this
resource
like
and
it
can
be
kind
of
big
inexpensive
resource
like
so.
Maybe
it's
an
image
for
him
or
something
would.
A
Be
right
well
exactly
right
and
I
guess.
The
other
question,
though,
is
over
time.
Are
we
going
to
like
a
V8?
Is
V8
ever
going
to
remove
it,
in
which
case
we're
kind
of
out
of
luck
right,
but
if
they're
actually
going
to
like
completely
so
that's
worth
thinking
a
little
bit
about
as
well
is
like.
Is
this
something
that
you
know
in
fact
may
maybe
not
sustainable
long
long
term?
Right?
You
know
what
I
mean
okay.
D
I
would
suspected
Securities
they
want
to
have
efficient
books
V8
as
much
as
possible.
A
A
Anyway,
so
that's
yeah,
I'll
read
some
more.
It's
worth
us
reading,
some
more
and
thinking
about
that,
but
I
I
honestly,
don't
think
we're
gonna
change
their
their
mind
on
that.
So
it's
it's
kind
of
a
us
thinking
like
what
does
this
mean
longer
term,
and
what
can
we
do
to?
What
can
we
do
to
sort
of
help,
which
is
that
the
one
idea
is
like
yeah
I'd
like
to
Define
that
you
know?
Maybe
it's
like
support,
sandbox
sandbox
run
times
or
I.
Don't
know
what
right.
C
Yeah
I
don't
know
if
it's
greater
or
over
web
boss
Maybe
these
Club.
A
D
I
think
we,
it
is
we're
resurrected
last
time,
because
we
found
that
the
issue
is
closed,
but
technically
we
never
fix
this
finalizers
to
be
run
as
a
part
of
GC
juicy
run.
It's
something
called
Chang
Zonk
is
working
on
right
now
and
okay
I
had
also
some
PR
in
this
area,
but
we
found
that
this
issue
was
automatically
closed,
so
we
reopened
it.
C
Maybe
we
could
on
set
to
be
not
closed
because
we
are
working
on
and
yeah,
maybe
I
think
yeah.
We
could
say.
A
B
C
Yeah,
like
I,
mean
it's
the
the
pr
about
unknown
array
type,
so
yeah
right.
A
A
Okay,
that
was
this
one
yeah
and
then
the
last
one
here
right,
yeah.
A
Okay
right,
so
we
looked
at
stale
issues
and
in
terms
of
modules
that
are
ported.
C
Yeah
yeah
just
one
week
last
week,.
A
A
Haven't
I
haven't
done
anything
on
that
one,
but
still
doesn't
hurt
to
have
a
list
burn
down
the
list
of
issues
in
the
repo
okay.
So
let's
go
to.
A
So
we
don't
have
too
many
issues,
PR's
open,
so
I
think
this
one's
new
right,
ish.
C
And
refer
to
add
these
these
these
paracas
before
to
to
to
create
the
release.
Okay,.
C
Yeah,
essentially,
he
he
needed
to
parse
a
big
Json
file
and
you
know
node.js
has
a
limit.
512
megabytes.
So
you
you
you,
you
get
error.
If
you
try
to
open
a
file
bigger
than
yeah
these
sites
right.
D
A
C
In
my
opinion,
you
could
use
a
sorry,
some
module
like
Json
stream,
and
you
solve
it.
Maybe
your
problem
right.
A
C
This
is
the
the
problem.
Yeah.
C
A
reasonable
suggestion
for
me
I
answer
later:
okay,.
C
Makes
sense
I
can
make
there
it
is
when
these
PR
will
be
emerged.
For
me,
thanks
yeah
sounds.
A
A
Okay,
so
basically,
that's
already
been
discussed.
I
think
that
probably
takes
us
back
to
like
nothing
new
that
hasn't
been
talked
about
in
previous
ones,
any
other
ones
in
this
repo.
Before
we
move
over
to
core.
A
A
Okay,
well,
it's
always
nice.
That's
part
of
the
advantage
of
multiple
different
platforms
and
combinations.
Eventually,
one
of
them
shows
the
the
the
result.
D
No,
yes,
yeah
a
return,
an
API
value,
so
if
yeah
return
now
in
there,
yes
you're
right.
Yes,
it's
only
changed
it's
it's
only
one.
It's
only
test.
It's
only
with
specific
method,
not
right.
It.
D
Because
we
have
an
API
callback,
these
specifications
as
we
need
to
return
NPI
value,
but
this
method
does
return
void
and
actually
look
for
other
tests.
I
did
this
kind
of
right,
quick,
better
expression,
a
search-
and
it
seems
like
this
was
the
only
case.
Okay,.
B
B
A
Makes
sense
because
I
think,
if
nothing
else,
A
reproduction
would
let
us
out
a
test
which
would
be
good
right,
so
yeah
makes
sense
yeah
this
one
I
think
I,
don't
know
if
you'd
had
a
chance
to
take
a
look.
B
B
A
D
Like
a
second,
oh
sorry,
last
one
I
made
some
progress.
Obviously
this
one
okay
I
submitted
new
implementation.
What
we
discussed
last
time
so
technically
I
updated
the
description,
change,
documentation,
so
I
think
it's
it's
ready
for
the
prime
time.
A
D
B
D
Extract
it
all
these
different,
embedding
apis
in
a
separate
file.
I'm
really
happy
about
this
change,
and
it's
only
probably
another
thing
I
would
like
them
to
do
like
is
actually
implementation.
Also
I
believe
should
be
extracted
in
separate
file,
because
I
I
kind
of
try
to
write
they
like
the
same
term
would
be
nice
to
not
to
put
anything,
not
engine,
not
anything.
A
D
Definitely
there
was
some
progress
and
I
think
you
commented
out
on
this.
You
probably
want
to
edit
your
your
comment
because
it
seems
like
it
was
repeating
itself.
It
would
be,
could
come.
D
But
but
it's
great
to
see
that
that
progress
section
made
in
this
on
this
PR
and
especially
I,
think
after
your
comment,
developer
made
a
few
iterations.
Okay.
A
A
D
D
I
want
to
understand
what
to
do
with
this
finalizer
stuff,
because
I
do
have
another
kind
of
PR
open
full-on
time.
I
I
want
to
have
this
finalizers
to
be
around
essentially,
as
a
part
of
this
garbage
collector
runs
change
on
I'm
going
to
address
this
area
or
I
I
should
kind
of
update
my
PR.
What
do
you
think.
D
Different
types
and
I
definitely
at
this
point,
I
try
to
use,
create
new
apis,
but
I
think
our
latest
discussions
was
not
to
have
separate
apis.
That
may
be
using
this
feature
instead,
but
you
do
also
have
changes
in
a
similar
area.
So
my
question
to
you:
do
you
want
me,
continue
this
PR
or
you
going
to
implement
it
yourself?
It's
I
guess.
My
question
is.
B
Next
and
I
have
to
say
that
my
PR
currently
didn't
Reserve.
This
is
the
problem
that
is
the
penalized
again
code
into
the
JavaScript,
so
I
believe
when
the
respecter
has
landed
in
the
main
branch.
Then,
if
you
would
like
to
do
it,
then
you
can
add
a
feature:
detection
that
the
add-on
can
disable
the
desert,
love
that
penalizes
to
according
to
JavaScript
and
they
can
get
the
their
finalizer
to
be
invoked
more
eagerly.
So
I
think
it's
a
separate
issue.
We
should
check.
B
D
A
Okay,
well,
if
not
we're
pretty
much
at
time
anyway,
thanks
for
everybody's
time
and
we'll
see
everybody
in
GitHub
and
next
week,.