►
From YouTube: 2022-01-7-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
C
Pretty
much
a
request
to
community
to
help
with
full
code
coverage
for
node
api,
close
plus
classes.
A
C
A
Well,
they're,
all
generally,
a
combination
of
both
where
there's
like
a
c
c,
plus
plus
file
that
uses
the
node
add-on
api
like
each
of
the.
If
you
look
at
what
I'm
sharing
there,
each
of
them
says,
like
you
know,
we
have
the
m
class
and
there's
some
methods
on
that
class
that
we
don't
have
any
tests
cover.
C
I
just
at
some
point
want
to
resume
this
work
like
what
we
discussed
a
few
months
ago,
like
about
putting
some
code
test,
but
pretty
much.
I'm
mostly
concerned
about
this
c-based
apis
in
the
first
place.
A
Yeah,
these
are
c
plus
based
apis,
the
c
apis.
We
believe
we
have
pretty
good
code
coverage
in
the
core.
Now
I
know
there's
a
separate
discussion
of
like
well.
How
do
we
get
those
tests
separated
so
that
we
could
run
them
somewhere
else,
but
yeah?
We
talked
about
a
path
forward
on
that
and
I
think
we
have
next
steps
stuff.
This
is
for
the
the
add-on.
It's
the
the
node
add-on.
Api
wrapper
cool.
Thank
you.
A
D
I
am
it's
cool
for
me.
I
already
did
a
look
at
this
blog
post,
so
yeah.
A
D
Cleaner
than
these,
but
yeah.
D
It's
only
a
question
of
preferences,
so
it's
not
a
problem
for
me.
I
I'm
only
thinking
about
this.
Okay,
the
the
yeah,
the
image
that
we
are
using
here.
So
I
don't
know
if
the
light
lighted
version
is
is
better.
So,
yes,
it's
dark,
yeah,
that's
it.
So
this
is
only
the
the.
A
I
don't
have
any
strong,
strong
feeling,
one
way
or
the
other
what
yeah
yeah
it's
okay!
For
me,
okay,.
A
Okay,
so
I
think
at
this
point
I'll
just
forward
it
on
and
jack
I'll,
see
you
and
we'll
work
with
them
to
get
that
published
under
your
byline.
A
Okay
sounds
good,
so
stale
issues,
let's
take
a
look
there.
C
C
Go
sorry
go
ahead.
There
are
two
different
line
of
thoughts
like
what
this
constant
should
be
like
one
of
them
like
we
should.
We
should
actually
follow
like
with
normal
c
plus
plus
pattern.
Another
one
is
actually
like
what
I'm
proposing
for
more
practical
way
like
if
we
add
these
constants
to
all
methods,
it
means
like
our
classes,
can
be
used
in
much
more
different
contexts,
because
we
would
consider
all
these
different,
napi,
ic
plus
plus
classes,
to
be
rather
wrappers
around
javascript,
and
we
don't
use
cons.
C
We
pretty
much
just
put
kind
of
unnecessary
roadblocks
for
for
developers,
because
I
have
to
go
like,
for
example,
if
you
capture
in
lambda
some
c
plus
plus
class.
In
this
case
you
have
to
mark
with
lambda
mutual
mutual
mutable,
sorry,
and
and
so
on,
like
it's
a
lot
of
people
requiring
their
code
to
use
a
lot
of
cost
costs
unnecessarily
right.
A
With
like
and
there's
no
like,
it's
never
gonna,
adding
cons
in
those
places
is
never
gonna
cause
the
compiler
to
do
the
wrong
thing.
C
Technically,
not
because,
in
the
end
of
this
story
we
call
c
functions
and
they
think
we
don't
care
about
all
these
different
kind
of.
A
All
these
different
constants
right,
but
it's
more
it's
more
like
I
thought
I
thought
part
of
the
advantage
of
cost-
is
that
the
compiler
can
make
some
optimizations
so
like.
If
I
know
nothing
is
going
to
change,
I
can
actually
get
rid
of
code
because
I
could
like
admit
a
second
call.
I
can
reorder
things
and
if
things
are
actually
going
to
change
that
you
could
get
the
wrong
behavior
if
you
put
a
const
on
something
that
isn't,
but
maybe
I.
C
Mean
I
described
english
speaking
like
people
talk
about
so
imagine
some
environments,
it's
possible
to
optimize
your
program
in
a
way
that
if
you
say
that
certain
class,
a
certain
object
is
constant,
so
potentially
a
kind
of
compiler
and
whatever,
like
loader
of
this
program,
can
put
this
up
code
inside
of
your
known.
U
in
immutable
memory,
if
you
like,
like
in
rome,
right
romable
memory,
so
it
means
that
this
object
is
there.
C
We
can
call
these
methods
and
I
think,
we'll
be
fine
in
what
would
we
in
our
case,
think
about
what
exactly
our
data
structure
will
put
there?
It's
effectively
a
an
api
underscore
value
right,
it's
what
we
put
there.
I
think
none
of
our
methods
will
change
this
pointer,
which
we
put
in
this
removable
memory.
A
G
A
If
the
stuff
in
between
is
all
marked
cons,
I'm
thinking
the
compiler
could
say,
oh
well,
I
don't
actually
need
to
call
get
again,
because
I
know
everything
that
was
done
in
between
was
a
const
method,
so
it
couldn't
have
changed
anything.
So
I
don't
actually
need
to
get
that
value
again.
So
I'm
just
going
to
optimize
that
away.
C
It
would
probably
be
true
like
in
a
very
kind
of
ancient
world,
then
we
don't
have
didn't,
have
multithreading,
but
now,
like
many
compilers
need
to
account.
That
even
say
you
have
a
in
this
couple
methods
which
just
constant
but
modern,
compilers.
All
assume
that
there
is
another
thread
which
might
change
this
value
anyway.
C
So
they'll
usually
not
alight
any
kind
of
additional.
A
Do
they
I
mean,
I
know,
I
know
in
a
lot
of
places,
certainly
when
I
not
so
much
in
the
java
world,
because
it's
less
multi-threaded
by
default,
but
like
we,
we
did
a
lot
of
work
on
the
java
side
to
emit
fences
like
proper
fences.
For
you
know,
the
different
platforms
power
in
particular
was
pretty
tricky
so
that
when
you
were
doing
things,
you
would
emit
the
right
fences
to
make
the
memory
coherent
and
it
so
it
wasn't
that
it
just
says.
A
So
I
don't
know
I
haven't
looked
at,
I'm
not
like
I'm,
not
an
expert
on
compilers
or
anything,
but
I've
seen
them
do
things
that
do
did
require
us
to
actually
say
no.
Here's
a
point
of
coherency
you've
got
to
make
sure
that
you
flush
caches.
You
do
whatever,
so
I
I
wouldn't
be
surprised
if
some
of
the
compilers
do
things
like,
oh
well,
yeah,
you've,
told
you've,
told
me
it's
caught,
so
I'm
going
to
use
that
information
to
make
your
code
faster
and
it
that's
why
lying
to
it
makes
me
uncomfortable.
C
I
just
not
aware
about
if
c
plus
specification
has
says
anything
like
this
against
the
methods.
I
don't
think
like
a
c
plus
plus
model
talks
about
this
stuff.
They
they
do
talk
about
like
data
members
and
and
so
we
have
all
these
different
atomics
and
kind
of
different
kind
of
memory
model.
But
it's
mostly
concerned
about
data,
not
not
methods.
As
far
as
I
know,.
C
Maybe
some
like
in
many
many
years,
people
added
to
specification,
but
I
guess
like
to
keep
a
backward
compatibility
with
existing
c
plus
plus
concepts
on
methods
does
not
provide
any
kind
of
guarantees
really
well.
But
but
these
things
like.
A
It's
they're
always
like
it's.
It's
not
specified
that
you
will
do
these
things,
but
in
my
experience
these
things,
you
add,
like
you
say,
give
us
this
extra
information
and
it
gives
the
compilers
the
flexibility
to
do
that.
So,
like
you
know,
if
you
tell
me
that
it's
cost
it
doesn't
mean
I
have
to
do
anything.
I
can
just
do
exactly
what
I
did
before,
but
the
whole
goal
is
that
it
then
lets
me
do
whatever
unspecified
behavior.
I
want
as
long
as
I
don't.
C
Right
in
some
sense,
like
this
constance,
would
be
pretty
much
share
like
what.
If,
for
example,
imagine
we
have
this
kind
of
effectively
wrapper
class
around
an
api
value
right
right,
we
have
a
method
which
says
like
a
set
property
for
this
object
and
and
we
mark,
if
a
market
is
constant,
we
pretty
much
say
like
you
know
what
we
will
never
change
with
pointer,
which
inside
of
this
class
right,
we
don't
break
anything
any
expectations
from
compiler.
A
A
C
A
A
Like
the
the
things
that
it
would
do
like
because
it
doesn't
know
what
you're,
when
you
say,
get
me
blah,
it
doesn't
know
what
you're
using
internally
or
not
right.
So
it
doesn't
know
whether
you're
using
a
method
like
you
could
like
you
could
just
be
doing
math
right
it
doesn't.
It
doesn't
really
relate
to
the
fact
that
it
uses
a
a
method
like
it
uses
data
within
the
the
the
class
itself,
or
it
does
something
else.
It's
just
telling
you
that
when
you
call
this
method,
this
object
will
not
change.
A
I
guess
the
only
way
that
like
it
would
that
I
can
I
kind
of
go
along.
Your
line
of
thought
is
if
it's,
if
that
was
just
tied
to
nothing
related
to
the
memory
in
the
like
in
the
cloud
like
the
memory
in
my
in-memory
thing
so
like,
but
I
was
allowed
to
change
the
answer
of
get
blah
after
you
know,
maybe
that
that
you
know,
maybe
if
it
was
limited
that
way
like
in
terms
of
like
the
const,
the
only
thing
it
means
so
the
constitution
can
still
return.
A
C
So
imagine
you
have
a
function
which
returns
time
current
time
right.
Every
time
this
function
is
being
called,
it
will
return
different
time,
but
this
function
can
be
const
because
it
may
not
change
anything
in
this
class
with
against
its
code
upon,
but
this
function
internally
uses
some
system
timer
to
retrieve
time
right
so
that
that
actually.
A
A
C
And
in
kind
of
example,
you
can
find
in
standard
c,
plus
plus
library,
if
you
look
on,
for
example,
stood
shared
putter
like
share
ptr
and
share
ptr
has,
for
example,
error
operator,
and
this
arrow
operator
is
actually
mark
itself
as
const,
but
it
always
returns
a
t
star
so
effectively.
It's
always
remove
removes
constants
from
the
staff.
It
returns
right.
C
Oh,
it's
it's
it's!
It's
you
don't
it's
kind
of!
You
have
to
pay
for
it,
but
the
only
way
for
you
to
look
at
that.
You
really
need
to
look
on
some
some
draft,
which
was
like
they
usually
have
published
draft
before
the
publish
final
one
gotta
pay
for
the
stat.
That's.
G
A
A
A
You
know
and
it,
but
then
I'd
be
comfortable.
Saying
yeah!
Okay
well,
based
on
this
definition
of
const
we're
not
violating
any
promise
and
therefore
we're
not
going
to
get
into
trouble,
because
you're
right,
like
that.
That
example
of
like
get
time
clearly
means
that
calling
a
const
method
doesn't
guarantee
that
you're
going
to
get
the
same
answer
the
next
time.
C
D
I
I
think
that
the
legendary
cast
said
that
at
one
point
you
can
get
the
nappy
underscore
value
pointer.
No
so
and
then
you
you
could
use
the
cpi
to
change
what
is
in
and
that
nappy
underscore
value.
So
this
is
the
point
I
think
that
or
or
obvious
in
the
cast
about
the
use
of
guns,
because
at
one
point
you
can
get
the
c
pointer
no
of
that
value
right.
So
you
think
he's
saying
that.
D
But
but
I
I
think
that
depends
on
what
the
cost
simplest
plus
is
no
so
yeah,
because
you
can
because
it
works
for
for
the
rapper
for
me.
But
if
you
can
able
to
get
to
the
row
pointer,
that,
for
us,
is
the
c
pointer
that
we
are
wrapping.
Then
you
cannot
promise
that
that
yeah
at
one
point,
the
the
what
you
are
wrapping
will
be
never
modified
right
yeah.
D
A
D
A
Yeah
his
point
was
like
it's
almost
like
we
no
method.
It
may
be
that
we
should
mark
nothing
as
costs,
so
there's
no
there's
no
guarantee,
but
I
think
it
comes
back
to
like.
I
think
our
next
step
is
good
in
terms
of
like,
let's
look
at
the
spec
and
if
we
can
actually
say
this
is
what
the
spec
is
promising,
we
can
then
interpret
it
in
terms
of
like
is
that
being
are
we?
Are
we
maintaining
that
or
not.
D
Okay,
yeah
yeah,
it's
I
agree
with
it:
okay,
yeah.
A
And
if
the
answer
is
like
anything
behind
the
scenes
could
break
that
guarantee,
then
we
shouldn't
add
const.
If
the
answer
is
it's
tightly
enough
to
find
that
yeah,
none
of
these
things
that
we
could
do
matter
in
terms
of
that
definition
like
if
it's,
if
it
strictly
says
well
anyway,
let's
see
what
let's
see
if
we
can
pull
that
out,
and
so
so
vladimir
will
grab
that
we'll
put
it
in
the
issue
and
that's
a
good
good
way
to
sort
of
frame.
The
next
discussion.
D
A
A
Yeah
this
one,
I
actually
watched
the
video.
The
video
was
just
a
generic
video
on
why
reinterpret
cast
can
cause
problems,
but
I
don't
I
I
didn't
clue
in
at
least
myself
from
watching
that
as
to
why
that
means
that
this
it's
an
issue
in
this
case
and
why
that
would
be
related
to
the
specific
problem.
A
I
don't
think
it
was
like
a
report
on
a
on
something
where
it
was
like.
Okay,
go
look
at
this
other
change
a
while
back,
and
I
guess
I'm
at
the
point
where
I'm
not
sure
we
have
a
specific
enough,
recreate
that
it
wouldn't
take
a
large
amount
of
time,
and
you
know
I
looked
at
the
video.
It
wasn't
like
obvious,
like
oh
yeah.
That
means
this
is
wrong
because,
like
the
code
is
doing
a
cast,
you
can
always.
C
You
mean
if
value
of
the
date
is
now.
C
A
C
Well,
thankfully,
speaking
even
c,
plus
plus
compiler
allows
you
to
take
address
of
a
variable
which
is
not
even
initialized.
A
Right
well
exactly
right,
yeah,
so
that
it
doesn't.
If
this
even
this
had
nothing
in
it
was
none
initialized.
That
is
fine.
So,
like
that's,
where
I
looked
at
this
and
said,
like
the
video
didn't
add
any
any
insight
into
what
might
be
wrong
here,
it
just
said:
don't
use
reinterpret
cast,
you
can
run
into
problems
right
and
it's
like
okay.
I
can
understand
that,
but
I
don't
see
how
it's
a
problem
here.
C
Yeah,
nothing
obvious
to
me
because,
thankfully
I
assume
data
itself
is
is
a
field,
so
thankfully
it's
not
it's
addressing
memory
of
data
is
well
defined
at
this
point.
Yes,
content
of
this
data
also
well
defined
here,
because
it's
now
a
ptr
unless
class
itself
probably
been
you
know.
Sometimes
we
see
the
situation.
The
class
itself
kind
of
already
considered
with
that,
and
already
people
start
to
put
some
garbage
there
and
still
some
code
users
said
it's
only
kind
of
situation,
but
this
is
this
constructor
right.
A
Right
well
and
if
that's
happening
like
yeah
you're
using
an
invalid
class
or
an
invalid
pointer,
well,
there's
a
bigger
problem
like
the
problem.
Isn't
the
fact
that
we're
using
the
reinterpret
cast
here?
It's
the
fact
that
something
else
is
completely
broken
in
the
in
the
original
code.
Right
like
it's
somehow
using
a
collected,
object
or
whatever
so
right.
C
Do
do
we
have
a
wrapper
here
because
think
definitely
I
don't
buy
argument
that
doing
reinterpret
cast
is
wrong
yeah,
but.
A
So
you
know
I
I've
other
than
spending
a
lot
of
time
trying
to
recreate
a
fairly
complex
issue
with
you
know,
turning
on
address
sanitizer
building
this
other
big
project,
you
know,
I
don't
know
what
a
useful
next
step
is
and
and
with
the
originator
not
being
as
not
like,
not
responding,
I'm
kind
of
tempted
to
say.
Well,
we
maybe
we
should
pose
this
until
somebody.
You
know
reports
something
similar
or
is
you
know
actively
working
with
us
to
investigate?
I
don't
know
what
people
think,
though,.
C
Yeah,
I
think
we
could
probably
close
it
until
or
a
regional
author
maybe
come
back
and
with
some
more
helpful
wrapper
case.
Okay,.
A
A
B
G
It
does
just
yeah
the
the
example
given
in
the
docs
did
not
work
for
the
user.
G
The
previous
one,
which
you
showed
you
know.
G
I
think
it
was
the
issue:
three
nine
five,
six
four.
It
was
a
dog's
your
example.
The
the
example
in
the
docs
did
not
work
and
your
pr
actually
fixes
it.
You
had
added
the
variable
the
function
in
the
global
context,
and
then
it
worked
right,
sorry,
which
one
was
that
that
was
that
was
the
previous
one.
39564
am
I
okay.
A
A
A
G
A
What's
interesting
is
this:
this
has
like
an
update
to
v8,
so
it's
quite
likely
that
one
of
the
version,
one
of
the
things
in
v8,
changed
that
behavior,
because
I
think
in
this
earlier
discussions
we
were
talking
about
like
hey,
why
you
know
we're
telling
v8
the
right
info?
Why
isn't
it
working
properly
right,
yeah.
B
The
only
thing
I
can
see
is
before
we
when
we
wanted
to
try
to
autocomplete.
It
passes
like
a
sort
of
incomplete
code
into
va,
and
maybe
some
air
gets
around
there
and
now,
with
the
upgrade
it
sort
of
absorbs
it,
and
instead
right.
A
So
the
wasm
ones
don't
seem
doesn't
seem
like.
A
A
So
what
was
that?
It
was
like
pre.
A
A
A
A
A
I
think
the
only
thing
we
could
try
and
like
the
next
step
would
be
to
try
and
figure
out.
You
know
which
of
those
commits
you
know.
Is
there
a
specific
commit
that
actually
fixes
the
issue
itself
and
wouldn't
make
any
sense
to
backport
that
in
core,
but
I
think
it
seems
like
it's
probably
a
core
issue.
A
I
think
that
was
all
the
list
in
there
new
modules,
I
don't
think
there's
anything
new
there
and
since
we
just
have
a
little
bit
of
time
left,
I'm
going
to
move
forward,
enable
debug
testing
for
add-ons
d-pack
anything
to
report
on
that.
One.
G
So
kevin
he
reported
an
error
right.
So
basically
the
when
the
test
were
error
ignored
it
was
not.
It
was
returning
as
zero.
Always
it
was
exiting
the
zero
code.
I
fixed
that,
but
the
test
which
was
failing
for
him
was
actually
successful
for
me.
So
I
was
okay
curious
to
know
why,
okay,
so
he
doesn't.
A
A
G
A
And
we
can
fix
things
as
long
as
the
tests
still
pass.
We
haven't
broken
anything.
We
can
fix
things
afterwards
as
well.
So
it's
not.
I
think
we,
I
think
we've
spent
a
lot
of
time
and
it's
we
should
get
it
landed.
So
do
that
okay,
add
unit
test
to
help
async
worker.
F
A
And
this
is
actually
okay,
so
I
didn't
mean
to
go
through
all
the
pr's
anyway.
I
just
okay,
so
looking
for
things
that
are
new.
A
I
don't
know
if
there's
a
call
out
if
there's
any
that
you
in
particular
think
we
need
to
look
at.
You
were
mentioning
one
of
them
right
deepak,
that
was.
Is
it
closed
or
still
open.
G
A
One
zero
two
okay
thread
say:
function
with
array,
buffer.
G
So
so
looks
like
the
user
had
reported
that
you
know
the
main
thread
and
the
main
thread
had
exited
the
j's
thread
had
exited,
but
the
background
thread
has
still
going
on,
so
I
gave
a
so
the
the
example
that
was
that
is
provided
by
the
user
is
not
compiling
for
me.
I
think
it
is
a
windows
based
thing
that
he
had
given
if
you're
a
good
repo
he
has
provided.
Also
internet
repo.
G
I
was
I
wanted
to
try
it
in
mac.
I
am
only
bringing
a
mac,
so
I
gave
an
example
a
simple
example
from
the
specific
exam,
so
I
thought
the
question.
A
couple
of
things
that
are
there
in
this
example
is
that
he
was
using
array
buffer
was
returning
back
an
array
buffer
and
then
also
using
a
wrapped
function
right.
So
the
thread
save
function.
He
was
returning
a
using
a
array
buffer
or
returned
in
the
js
callback,
and
also
he
was
using
a
wrapped
object
right.
I
I
tested
it.
G
The
example
that
I
gave
was
only
with
the
array
buffer
and
then
he
commented
that
he
wanted
to
use
a
wrapped
object.
I
used
a
wrapper
object
as
well.
I
have
the
example
in
my
local.
I
can
share
my
screen
and
show
you,
but
the
thing
is
that
either
way,
I'm
not
able
to
recreate
the
issue
that
he
has
asked
for
that.
The
js
thread
just
exits
and
the
you
know
the
workers.
Child
thread
is
still
going
on.
The
worker
thread
is
still
going
on.
F
G
A
A
G
Let's
catch
up
in
the
next
session
yeah
and
what
was
the
other
one?
You
wanted
to
talk
about
one
one,
zero
three!
I
guess
one,
one:
zero,
three,
one:
one:
zero:
three:
okay!
It
was
a
performance
thing.
G
So
I
created
a
function
like
with
ten
thousand
exports
yep.
Initially
it
didn't
seem
to
show
on
any
like
any
performance
this
one
right,
but
I
I
did
do
a
user
performance
and
found
out
the
very
first
call
in
the
very
first
call.
There
is
a
slight
difference
between
the
larger
module
and
the
smaller
model.
The
smaller
module
say
if
the
very
first
call
takes
about
two
milliseconds,
the
larger
one
takes
about
six
milliseconds,
but
that
is
the
only
thing
after
that
they
are
all
the
same.
G
I
even
have
the
example.
I
can
run
it
and
show,
but
there's
no
significant
difference
right,
but
if
it
matters
for
some
for
some
reason
for
a
graphics
or
you
know
for
some
kind
of
graphics
module
if
it
matters
the
very
first
call,
after
after
the
loading
of
the
module,
a
native
module,
the
very
first
call
to
a
larger
module
is
slightly
like
three
times
three
x
times
more
fast
slower
than
the
smaller
one.
D
A
G
So
yeah,
I
can
run
the
just
repeatedly
and
see
repeat,
repeat
the
test,
multiple
time
and
see
what
if
it
is
even
recreatable,
but
I
I
tested
a
couple
of
times
in
my
local
and
it
seems
consistent
to
me
the
you
know
the
only
the
very
first
call
takes
a
little
longer
and
after
that
it
is
always
in
less
than
a
millisecond.
It
takes
less
than
a
millisecond.
Both
calls
are
equal
right,
larger
and
smaller.
A
A
So
I
I
I
guess,
I'm
at
the
point
where,
from
from
what
you've
running
like,
if
you
want
to
run
it
a
few
more
times,
that's
good,
but
I
think
you
pretty
much
confirmed
that.
It's
not
it's
not
just
the
hey.
It
takes
longer,
there's
something
much
more
subtle
going
on
or
something
in
this
code
or
whatever.
Okay,.
G
A
Yeah
exactly
yeah
so
yeah,
I
would
say
you
know
you've
done
some
good.
It's!
You
know
it's
good
for
us
to
have
done
that
and
to
understand
and
I'd
kind
of
say.
Like
again,
you
made
some
good
comments
unless
there's
a
response-
and
you
know
another
month
or
something
we
should
feel
pretty
comfortable
in
saying
we've
done
whatever
investigation
is-
was
appropriate
in
this
case.