►
From YouTube: 2022-01-14-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
Okay,
so
welcome
to
the
node
api
team
meeting
node.js
node
api
team
meeting
for
january
14
2022..
We
will
follow
our
standard
approach
of
looking.
A
I
think
this
is
the
right
one
with
the
10
things
tagged
for
the
agenda
and
we
can
go
from
there,
so
the
first
one
is
drive
towards
full
coverage.
I
know
jax
gave
us
a
heads
up
that
he
couldn't
make
today,
but
the
blog
post
did
go
out
and
hopefully
that'll
raise
some
awareness
of
some
of
our
good
first
issues,
and
I
don't
know
if
anybody's
seen
any
new
contributions
or
anything
in
the
repo
yet,
but
hopefully
we'll
see
something
there.
C
There
was
one
people
that
asked
to
write
the
test.
I
don't
remember
for
for
what
but
yeah
I
I
answered
yeah
yeah,
you
can
so
yeah.
B
C
I
I
think
that
the
blog
post
had
a
good
impact
about.
I
hope
that
more
people
come
to
to
give
the
the
contribution.
A
Yep
yeah,
so,
if
anybody's
watching
feel
free,
feel
free
to
join
the
meetings-
and
I
guess
we'll
keep
an
eye
on
that
and
and
see
how
that
goes.
A
The
next
issue
is
stale
issues
and
no
doubt
on
api
to
discuss.
So,
let's
just
go
down
to
the
one
at
the
bottom.
Okay,
so
I
don't
see
anything
new
new
this
one.
A
Okay,
so
this
one
is
just
you
know:
we've
got
something
a
workaround,
but
we're
not
you
know
leaving
it
on,
because
we
think
there's
some
additional
investigation
that
should
take
place
right
unless
anybody
has
any
updates
on
that
one
I'll
move
on
to
the
next
one.
A
Right
we
had
a
good
discussion
of
this
one
again.
This
is
in
terms
of
you
know
what
what
does
const
really
mean
and
how
should
we
use
it
in
general,
and
you
know
the
next
auction
next
action
was
vladimir,
was
actually
going
to
try
and
find
the
text
from
the
spec,
because
the
the
goal
from
that
is
to
like
really
understand.
A
If,
if
we
take
a
particular
position
on
how
we
use
it,
is
that
going
to
break
anything
that
this
back
is
depending
on
so,
for
example,
if
you
know,
if
you
say
constant,
the
compiler
can
assume
something.
Are
we
staying
consistent
with
that,
but
I
don't
see
a
a
an
update
there,
so
I
don't
think
we
have
anything
more
to
discuss
on
that
one.
Until
that
happens,
anybody
else
have
any
thoughts.
A
Right
so
I
mean
the
last
comment
is:
is
a
comment
from
gabriel
with
a
suggestion.
A
And
I
think
we
might
have
discussed
last
time
that
you
know
if
there
isn't
any
interaction
sort
of
soon.
We
might
consider
that
you
know,
or
we
can't,
we
can't
necessarily
move
it
forward
without
the
involvement
of
the
original
originating
poster.
A
D
So
for
this
one,
do
we
want
to
take
the
implementation
that
the
poster
made
in
this
pr,
or
are
we
waiting
for
him
to
respond
to
the
suggestion
and
say?
Oh,
this
is
a
good
suggestion,
let's
close
the
pr
or
no,
I
don't
like
the
suggestion.
I
I
will
go
with
this
pr
anyways.
Basically
like
is
this
suggestion
going
to
be
the
new
way
that
needs
to
be
done
in
order
for
this
pr
to
be
merged,
or
can
we
take
what
he
has
currently.
A
E
Yeah
yeah
this
was
based
on
what
vladimir
was
saying
that
we
should.
We
should
try,
try
to
avoid.
You
know,
passing
just
an
array,
because
you
know
you
you,
you
have
all
kinds
of
like
bounds
issues
and
that
kind
of
thing
yes-
and
no
I
mean
you
know
we,
the
the
the
final
solution
here,
would
be
to
create
the
class.
That's
that's
similar
to
to
what
what
span
will
be
in
in
c,
plus,
plus
2020.
E
That
would
that
was
basically
his
suggestion,
because
it's
it's
safer
now,
given
that
we
aim
to
interoperate
with
just
the
plain
c
api,
I'm
kind
of
ambivalent
about
this,
because
on
the
other
hand,
you
know
having
having
a
pointer
to
an
napi
value
which
could
be
you
know,
an
array
could
be
just
a
pointer,
and
we
don't
know
is:
is
this
part
of
working
with
the
c
api
so
yeah
like
I'm,
I'm
not
I'm
not
that
set
on
waiting
for
the
answer.
I'm
I
mean
I.
E
If
I
remember
correctly,
I
did
approve
this
pr
right.
So
yeah,
that's
that's
where
I
am.
E
I
I
have
to
check
where
we
use
c,
like
just
plain
pointers,.
D
A
E
A
A
D
Yeah,
but
I
think
we
could
make
some
sort
of
standard
in
our
in
our
library
that
only
the
underlying
napi
types
can
be
in
a.
D
Because
those
are
c,
you
know
c
types,
whereas
if
we
want
to
do
something
with
c
plus
plus
type
things,
maybe
we
would
require
them
to
be
inside
of
a
class.
Absolutely.
A
E
E
A
F
E
E
Yes,
that's
what
I
think
it
means,
but
but
usually
oh,
I
see:
okay,
okay,
okay,
no,
no
they're,
not
getting
rid
of
it
after
after
c
plus,
plus
20,
whatever
whatever
a
sign
is,
becomes
constant
expression.
E
I
think
yeah
so
yeah,
so
until
c
c
plus.
Oh,
why
don't
I
just
paste
the
link
here?
We
can.
We
can
all
look
at
it,
I'm
just
faster
than
am
I
trying
to
convey
everything.
The
pages
is
saying:
where's
the
chat
where's
the
chat
chat
paste
there.
You
go.
E
D
E
D
D
E
And
and
I
and
and
yeah-
and
I
I
don't
know
that
it
buys
us
a
lot
of
safety-
we're
we're
basically
deferring
safety
to
to
to
to
the
user
of
the
or
to
the
developer
of
the
add-on.
I
mean
which,
which
we
already
are
right.
I
mean
if,
if
we
get
a
garbage
pointer,
where
we'll
do
garbage
with
it,
you
know
so
I
don't
know
that
it
gains
as
much
type
safety
to
to
just
have
a
standard
vector
that
that
was
created
with
a
sign.
E
Yeah
yep
yeah,
so
yeah
I
mean
if
if
they
have
to
build
up
anything
from
a
pointer
and
a
length,
then
that
could
always
be
wrong,
no
matter
how
we
try
to
to
encourage
them
to
use
well,
well-formed
and
well-defined
arrays.
Sometimes
you
just
have
to
deal
with
a
pointer
and
a
size.
E
So
yeah,
I
don't
know.
I
don't
know
that
that
there's
a
lot
of
harm
in
that
well,
actually,
okay
hold
on
hold
on
just
to
go
back
into
the
circle.
Once
more.
B
B
E
D
E
They
are
very,
very
lightweight
wrappers
right,
like
you,
don't
think
about
the
storage
that
they
use
up
because
they
should
they
should
like
the
compiler
should
boil
them
away
90
of
the
time
and
just
replace
them
with
their
contents
and
and
should
replace
what
looks
like
a
c
plus
plus
call
with
with
one
or
more
calls
to
an
api
core
in
line
right.
E
That's
the
original
goal
of
of
these
bindings,
and
so,
if
you,
if
you
explicitly
set
aside
storage
on
the
heap
for
for
for
these
values,
then
they
are
becoming
real
objects.
That
cannot
be
boiled
away
right
because
they
they
have
to
be
evaluated
at
runtime,
and
so
in
principle.
I
think
we
should
avoid
that.
You
know.
D
And
if
you
look
at
one
of
his
tests,
I
guess
this
one
here
on
line
80
or
this
one
here
on
101.
The
way
that
he
creates.
It
is
by
creating
a
vector
and
then
calling
it
by
size
and
data
to
get
the
pointer
yeah,
which
doesn't
seem
right.
E
E
E
A
E
D
E
Makes
sense
yeah
it
would
be
a
little
more
code
to
to
like
wrap
it
in
a
standard
vector,
but
but
I
think
it's
worth
it,
it's
worth
maintaining
this
this.
You
know
policy
of
you
know
c,
plus
plus,
u
c,
plus
plus,
if
you
already
start
with
c
plus.
E
E
A
second
vector
has
vector
has,
it
seems,
to
have
an
override
for,
like
the
vector
constructor
also
seems
to
have
an
override
for
start
point
or
end
pointer.
E
G
A
A
A
D
C
A
Sounds
good
okay,
so
looking
at
the
next
one,
the
the
rebel
issue
jack
had
done
done
some
triage
and
basically
figured
out
that
it
it
like
somewhere
between
16,
5
and
16
6.
There
was
a
fix,
it
looked
like
it
might
be
in
v8,
but
we
need
somebody
to
spend
some
more
time
to
sort
of
understand
or
figure
that
out.
A
You
know,
because
if
there
is
a
fix
in
v8,
maybe
that
should
be
pulled
into
14..
I
know
I
know
lucinda
cass
had
looked
at
it
and
sort
of
said
it's
strange
that,
even
though
we're
telling
the
console
that
it
is
a
mutable
method,
it
still,
it
still
goes
ahead
and
sort
of
executes
it.
Maybe
that
actually
got
fixed
in
a
later
version,
so.
A
E
A
No
yeah,
like
that's,
not
yeah
like
basically
what
I
was
trying
to
say.
Maybe
not
well,
is
like.
We
know
something
in
here:
changed
it
yeah.
Can
we
figure
out
what
it
is
it's
like?
Is
it
one
of
these
v8
you
know
commits
I
mean
I
think
one
of
them,
though,
was
like
a
whole,
as
is
not
necessarily
a
small,
commit
like
anyway
right
right.
A
It's
basically
going
down
to
say,
okay,
what
what
commits
in
here
or
you
know,
and
if
it's
a
move
v8
to
the
next
version,
it
might
even
be
like
figure
out
the
subset
of
that.
That
actually
needs
to
be
fixed.
E
I
ideally
ideally
would
perform
this
experiment
and
and
and
give
the
the
release
folks,
maybe
even
a
pr
saying,
with
with
the
with
the
back
port.
A
E
A
The
next
step
really
is
for
somebody
to
have
enough
time
to
look
at
it
and
actually
figure
out
which
of
these
in
here.
It's
great
it's
great,
that
it's
narrow
that
jack
narrowed
it
down
to
one
of
these,
but
now
we've
got
to
figure
out
which
one
of
these
you
know
actually
cost,
because
I
think
I
looked
at
it
and
it's
not
like
there's
a
tiny
number
of
things
right
like.
A
B
A
Back,
I
think,
that's
all
the
stale
ones
yeah.
So
that's
all
the
ones
on
the
stale
list
tracking
the
issue
for
modules
that
are
reported.
I
don't
think,
there's
anything
new.
There.
A
So
nothing
really
new
on
that
front
to
note
enable
debug
testing
for
add-ons
so
that
one,
I
think.
A
A
I
know
that.
Definitely
the
fact,
the
issue
that
deback
has
I'll
plan
to
try
and
land
that
this
afternoon
I
think
it's
been
approved.
It's
all
ready
to
go
just
I
meant
to
do
that
last
week
and
do
it
are
there
any
other
ones?
In
particular,
people
suggest
we
should
talk
about.
D
I
think
there
was
an
issue
that
I
was
working
on,
but
I
haven't
worked
on
it
since
my
vacation
and
I
may
need
some
help
on
it.
It
was
the
signal
was
not
being
triggered,
so
I
don't
think
it's
this
one,
because
maybe
that's
a
removal
of
cons.
It
was
under
issues.
I
think,
okay,
let
me
get
back
to
the
issues.
A
G
D
And
legend
has
mentioned
that
since
signal
is
user
initiated,
but
there
is
no
progress
data,
and
this
is
one
that
I
think
I
would
like
to
work
on.
But
I'm
not
exactly
sure
the
best
way
to
provide
a
fix
because
we
have
to
ping.
We
have
to
make
a
a
call
to
the
thread
safe
function
in
order
to
to
trigger
it.
But
we
need
to
be
able
to
differentiate
between
a
call
of
the
the
progress
and
a
call
first
for
signal
and
I'm
not
sure
the
best
way
to
handle
that.
D
E
E
Where
you
know
you
know,
if
it's
of
this
type,
then
this
is
a
signal
call
if
it's
of
that
type,
then
it's
a
it's
a
progress,
call.
A
A
It's
like
I'm
just
trying
to
wonder
like
does
it
make
sense
that
there's
actually
like
an
on
signal
method
and
if
you
called
signal
that
should
get
called
instead
of
on
work
progress
or
does
it
make
sense?
That,
like
signal,
is
an
exceptional
circumstance
and
you
know
you
want
that
in
the
on
work,
progress,
and
but
you
know
that
it's
that
that's
the
case
somehow.
G
In
my
understanding,
signal
is
just
a
specialized:
send
send
progress
as
it
doesn't
understand
any
data,
but
it
just
activates.
The
javascript
just
give
the
javascript
thread
a
chance
to
wrong
arbitrated
arbitrary
code,
even
if
the
loop
is
idle,
and
so
that's
the
very
the
I
to
my
understanding,
the
very
use
case
for
the
signal,
but
it
can
be.
G
The
previously
fixed
a
issue
is
that
we,
since
the
thresher
function,
behave
differently
with
the
uv
async,
since
we
are
deferring
the
actual
directive
function,
call
with
a
uv,
immediate
immediate,
so
there
are
chances
that
the
actual
there's
a
function
might
get
deferred
and
with
and
in
the
activation
of
the
uv
async
and
between
the
uv,
immediate.
There
are
chances.
We
stand
using
reactivate
your
basic
again.
G
First,
if
the
evasion
will
trigger
again
because
we
are
deferred
by
the
uv,
immediate.
G
D
I
think
I
I
do
recall
I
I
do
unders
understand
what
you're
saying,
but
I'm
not
exactly
sure
how
we
can
differentiate
it
correctly.
If
that
makes
sense.
D
And
this
this
little
code
block
that
I
posted
here
solved
one
of
the
problems,
but
the
test,
didn't
it
like
solved
the
case
of
sending
null
pointer,
but
then
it
actually
broke
the
other
tests.
So
that's
why
I
said
the
test.
The
fix
is
premature
to.
D
G
It's
currently
swallowed
because
we
are
sending
our
pointer
and
we
are
not
able
to
distinguish
the
now
pointer
of
data
to
no
data
and
the
data
is
already
consumed.
So
we
cannot
distinguish
these
two
state.
G
It's
all.
We
are
now
distinct.
We
know
using
the
no
point
as
a
as
a
state
of
the
data.
The
data
is
already
consumed,
so
we
are
swallowing
the
signal.
E
B
B
E
Opportunity
to
just
to
just
create
a
data
structure
which
wraps
the
user's
data
and
which
has
the
hints
that
we
need.
You
know
we
can
have
an
enum.
That
says
this
is
a
signal
and
this
is
data
and
then
and
then
the
data
structure
contains
one
of
those
enum
values
and
it
contains
the
user's
data
right
and
then
on
the
on
the
flip
side.
We
just
say
you
know
if
this
is
a
signal,
then
I'll
do
whatever
signals
do.
If
this
is
data,
then
do
whatever
data
does.
D
E
D
E
B
D
G
E
G
D
Okay,
so
then
inside
the
user's
javascript
function
they
can
check.
If
the
argument
is
null,
then
it
was
a
prop,
then
it
was
a
signal
call
and
then
they
can
handle
it
that
way
inside
their
code.
Yeah,
okay,
I
think
that
makes
sense.
Then
all
right,
I
think,
from
both
that
and
then
gabe's
suggestion
of
that
you
know
type
with.
I
think
that
makes
sense.
I
think
I
can
work
on
this
now.
D
Yeah
great,
all
right.
A
Okay
sounds
good
thanks,
sorry
about
that,
were
there
any
other
ones
that
people
wanted
to
call
out
that
we
should
take
a
look
at.
D
I
see
there's
this
pinned
issue.
Do
we
plan
on
that?
You
have
at
the
top
here.
Are
we
planning
on
another
release
and
if
so,
are
we
waiting
for
issues
to
be
closed
before.
D
C
A
Sounds
good
okay,
if
nothing
else
in
this
repo.
Let's
just
take
a
very
quick
look
in
the
node
repo
for
any
issues
that
are
tagged.
A
Yeah,
so
this
one,
I
haven't,
had
a
chance
to
read
his
answers.
But
the
question
is
like
what
happens
because
the
recent
like
note,
17
plus,
is
open
a
cell
3.
E
D
E
And
and
yeah
and
that's
compile
time,
that's
a
compile
time
decision
right,
whereas
we
are
explicit,
we're
all
explicitly
about
runtime
decisions
so
that
we
can
maintain
api
stability.
E
A
A
A
It's
kind
of
like
because
we
we
even
change
the
default
node
api
version.
So
if
you
actually
build
with
a
newer,
a
newer
version
of
node
you
may
you
may
build
something
that
won't
run
by
default
on
older
versions.
Right
like
you
can
you
can
actually
set
the
node
api
version
to
avoid
that,
but
by
default
yeah
you
can't
go
backwards.
So
I
yeah
anyway,
we'll
we'll
see
what
he
says
in
that
answer,
and
that
will
maybe-
and
if
he's
here
at
the
next
meeting
we
can.
A
A
D
A
Guys
next
week,
yeah
thanks
yeah.
We
are
at
time.
So
I
think
at
this
point
we
should
probably
drop
off
unless
there's
anything
like
super
high
priority
people
think
we
should
look
at.