►
From YouTube: Node.js N-API Team meeting 2020-02-10
Description
A
A
Okay,
all
right,
and
so
let's
go
through
the
milestone.
Do
we
need
some
additional
async
or
covariance
I?
Don't
think
anything
has
changed,
but
since
last
week,
but
please
chime
in
if
it
has
I,
haven't,
had
a
chance
to
look
at
a
synchro
rescue
worker
I
just
started
reviewing
it
before
the
meeting.
But
aside
from
that
I'm
not
aware
of
any
other
changes,
then
for
the
Mac
issue,
Jim
how's
it
been
going.
Oh.
B
A
B
Can
you
hear
me
now,
oh
okay,
so
what
I
just
shared
is
that
the
the
Windows
builds
have
been
failing
completely
and
this
week,
I've
determined.
Why
that's
happening
and
I'm
working
on
a
solution
to
at
least
get
the
Windows
builds
done
and
I've
got
a
little
package
of
build
issues
that
I'll
raise
with
Michael
when
he
gets
back
from
vacation.
A
A
This
one
is
also
kind
of
a
introductory
thing,
and
the
people
discovering
that
it's
not
so
easy
to
map
every
property
of
every
structure
to
to
a
JavaScript
property,
but
thanks
Kevin
for
chiming
in
there
yeah.
Now
this
one
I
don't
know.
If
there's
been
any
progress
on
this
okay,
I
guess
Kevin
started
working
on
it.
Hopefully,
hopefully
we'll
have
we'll
have
an
update
from
him
when
you
get
the
chance.
A
A
Lucas
you
got,
you
got
anything
new
on
this.
You
want
having
I've
been
working
on
this
I.
Think
I,
don't
know,
updates.
Okay,
all
right,
no
worries
all
right,
yeah,
that's
right
because
it
is,
it
is
thread-safe
function
really
if
it
so,
we
might
as
well
have
a
consolidated
effort
on
that.
Okay.
So
yes,
this
continues
to
be
case.
This
is
pretty
much
like
a
heads
up
and
yeah.
So
then,
let's
see.
A
Oh
we've
had
this
question
before
right:
we
had
like
a
steam
or
was
it
a
sink
promise
worker,
but
yeah
I
mean.
Maybe
we
should
do
an
example
of
this,
because
we
could
pass
the
problem
and
we
could
press
a
promise
or
a
deferred
into
a
an
ASIC
worker
or
an
async
RS
worker
as
a
reference,
I
guess
and
then
just
use
it
on
the
main
thread.
This
might
be
worth
an
example
settle
this
once
and
for
all,
because
it
does
come
up
every
once
in
a
while.
Okay
I.
A
And
to
just
just
make
an
example
for
it,
because
I
think
it
would
be
great
to
just
be
able
to
link
to
that
example
and
just
say
all
right,
all
right,
just
a
while
this
callback
is
not
called
immediately
I,
don't
know.
If
we've
we
reached
a
conclusion
with
this,
or
is
this
just
it's
just
somebody
realizing
that
these
things
are
asynchronous,
because
that's
what
it
looks
like
to
me.
A
A
Okay,
something
we
should
definitely
look
into,
but
I
think
this
is
just
a
matter
of
it
being
asynchronous:
okay,
I,
don't
I,
don't
know
if
they're,
if
we've
had
any
and
the
newer
ones.
Besides
that
something
that
has
cropped
up
in
the
meantime,
if
I'm
missing
something,
then
please
by
all
means
chime
in
but
I'm
inclined
to
just
move
on,
because
we
do
have
many
issues.
A
A
Right,
okay,
so
the
the
only
thing
as
far
as
I
know
that
has
changed
is
that
over
the
weekend,
I
tried
to
rewrite
the
reference
tracking
using
using
ojs
data
structures,
but
unfortunately
it
looks
like
it's
adding
a
performance
penalty.
So
so
that's
where
we're
at
with
that.
It
would
be
good
because
it
it
unifies
the
code
so
like
it
removes
all
this
stuff,
like
you
know
the
the
linked
list
sort
of
juggling
because
we're
already
doing
the
juggling
in
one
place.
A
Why
should
we
do
it
in
two
different
places,
but
unfortunately
it
looks
like
it
has
a
performance
penalty
which
I
still
haven't
figured
out.
Why
it
happens.
So
you
can
see
these
things
they
they
have
like
significant
declines
here
things,
and
especially
especially,
does
the
function
call
the
function
call
is
at
the
thirteen
percent.
Sixty
percent
incline,
so
yeah
I've
got
to
figure
out.
Why
is
the
interesting
thing
is
what
the
only
thing
that
should
have
declined?
Is
this
rest?
Because
this
you
know
this
thing
adds
and
removes
to
the
linked
list?
A
A
A
A
Okay,
I,
I,
guess
Nikola
could
also
update
us
on
this.
If
he's
seen
any
more
movement,
we
should
probably
check
on
on
notes
SAS
again
because
they
have
been
working
on
unfort
into
an
API.
That
would
be
a
big
big
win
for
us
if
they
moved,
but
I
haven't
had
a
chance
to
look
so
any
of
you
have
updates.
Then
please
just
mention
them
in
the
issue
on
this
front.
I
think
we're
also
fairly
quiet,
and
here
some
activity,
yeah.
B
So
if
you
scroll
down
to
the
bottom
Gabriel,
so
the
first
step
for
me
is
to
get
all
of
the
examples
on
to
the
examples,
repo
and
so
I've
created
this
table.
There
are
there's
one
that
for
context,
awareness,
that's
already
there,
so
the
ones
that
are
under
my
control,
I'm
working
to
get
those
on
the
repo
and
then
Nikola
and
Kevin
also
contributed
examples
there
that
are
kind
of
built
into
the
markdown.
B
That
would
be
better
if
they
were
separate
examples,
so
we'll
need
to
follow
up
with
them
when
they're
back
on
the
call
again
to
see
how
they
want
to
do
that,
then
the
next
step
is
to
discuss
what
tool
that
we
want
to
use
for
publishing.
And
if
we
have
a
discussion
about,
if
we
have
a
decision
on
a
tool
that
we
want
to
pursue,
then
what
I
would
do
is
do
some
playing
around
with
it
and
do
an
example
to
show
everybody
what
that
might
look
like.
B
And
see
the
hexo
tool
we'll
do
that
and
so
Haley,
ideally
that
that's
why
I
want
to
get
the
examples
into
a
repo
is
because
then
the
tutorial,
the
code
in
the
tutorial
will
be
always
live
code,
and
if
we
get
CI,
you
know
if
we're
running
tests
on
that
every
night,
then
we
have
a
great
deal
of
confidence
that
the
code
examples
in
the
tutorials
are
actually
working.
You
know
yeah
so.
B
Part
of
that
is
coordinating
with
the
rest
of
the
node
community
and
just
figuring
out
how
we
want
to
proceed
notice
that
node
has
a
page.
If
you
scroll
back
a
bit,
let's
see
you
know,
look
back
a
little
bit
more
I
think
there's
a
link
to
yeah
that
link
there
were
exist,
guides
yeah,
so
this
has
been
one.
One
proposal
is
that
we
put
them
here.
B
A
A
A
A
A
A
Yeah
and
and
and
we
have-
we
have
at
least
like
we
have
the
list
that
that
Nicola
put
together
and
then
they
have
the
list
from
from
Vincent.
We
universe,
so
I
think
I
think
we
can
even
start
with
those,
because
the
thing
is,
if
I
remember
correctly,
the
packages
on
NPM
that
use
an
API
right
or
or
no
dad
on
API.
They
don't
necessarily
have
an
an
API
tag
right.
A
Usually
they
have
an
an
MPI
tag
if,
if
the
main
release
hasn't
yet
switched
to
an
API,
but
they
are
offering,
they
are
offering
an
alternative
release
that
uses
an
API
experimentally
and
then
they
have
an
API
tag
right
but
or
at
least
a
release
tag,
but
I
guess
I
guess
they
can
have
like
a
package
tag
on
the
on
the
site
for
identifying
so
yeah
I
guess
you're
right.
We
can
probably
go
with
those
and
I
just
come
up
with
a
list,
but
ultimately.
A
The
thing
is
like:
if
we
create
a
list
right-
and
we
show
the
list
here
sooner
or
later-
that
list
will
be
out
of
sync
right.
So
so,
but
there
is
still
value
in
having
the
list
so
then,
so
then
do
you
think
it
would
be
best
to
just
place
a
link
on
this
page
that
that
brings
you
to
two
like
a
pre-recorded
query
on
the
NPM
website.
A
C
C
A
I,
don't
know
API
right
right,
okay,
but
those
packages
would
still
be
tagged
as
an
API,
though
right
like
it
wouldn't
be
tracked
by
their
dependencies.
But
rather
they
would
be
tracked
by
by
the
tags
right
and
they
can
still
be
tagged
right
as
an
API
and
then
we
would
finding
them
and
then
we
would
list
them
and
then
the
package
would
be
exposed.
Am
I
not
seeing
that
correctly.