►
From YouTube: Node.js N-API Team meeting - July 20 2020
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
Okay,
if
not
I'm
just
going
to
open
that
up
and
share
that
oh
stone
11
share.
Okay,
I
think
everybody
should
be
able
to
see
it
now:
yep,
yep,
okay.
So,
let's
start
it
transfer
the
napi
tutorials
to
the
open
into
the
opengs
foundation,
gym.
B
Yeah,
okay,
so
I
have
a
pr
in
progress
the
ci
was
working
before
and
now
the
ci
is
failing
for
reasons.
I'm
pretty
confident
are
not
related
to
the
website.
So,
okay,
I
just
wanted
to
make
sure
it'd
be
okay
to
land
that
pr
in
the
current
state
that
it's
in.
C
A
B
D
A
Is
this
potentially
like
this
is
failing
on
windows
right
yeah?
I
wonder
if
this
has
anything
to
do
with
the
fact
that,
like
we,
3.01
was
not
good
on
windows,
but
maybe.
A
A
B
Yeah
yeah,
the
travis
cia,
is
passing.
B
B
B
Okay,
so
I'll
land,
this
pr,
when
we're
done
with
this
with
this
meeting
now,
if
we
go
to
the
issue
here
that
we're
looking
at
on
milestone,
11,
the
the
final
piece
of
the
puzzle
is
to
create
a
github
pages
branch
right.
C
B
If
you
scroll
up
just
a
little
bit
michael,
I
I've
listed
the
steps
that
I
took
when
I
was
working
on
my
repositories
where
you
know
where
I
have
complete
access.
I
can
do
I
can
push,
but
I'm
not
clear
how
to
make
that
happen
like
in
an
open
source
project.
B
So
I've
listed
the
steps
here
right.
I
I
don't
know
if
that's
something
that
you
could
do
or.
A
Okay,
so
it's
just
a
like:
it's
not
a
there's,
no
question
on
how
it's
really
just
having
the
access
rights
to
be
able
to
do
it
right,
correct.
B
Yeah
the
I
list
here
the
steps
that
I
took
when
I
had
complete
access
to
the
repo
it's
unusual
in
the
sense
that
it
has
to
be
an
or
well
it's
recommended
that
it
be
an
orphan
branch.
A
D
B
And
then
the
other
step,
the
step
two
is
to
enable
the
get
github
pages
within
github,
using
the
get
the
g
h,
dash
pages
branch.
A
A
Sounds
good,
yeah,
okay,
anything
else
you
want
to
talk
about
on
that.
One.
A
Okay
sounds
good,
okay,
so
the
tracking
issue
issues
we've
noticed
are
ported.
Okay,
I
don't
think
there's
anything
new
from
that
perspective
since
last
week.
But
let's
just
look,
I
know
that.
D
A
Just
see
how
we're
doing
there
still
still
on
the
upward
trend,
which
is
good
yeah,
I
do
know
we
had
our
an
issue
from
sqlite
right
that
you'd
help
to
look
at.
B
Okay,
I
think
it's
a.
We
have
an
issue
going
on
note,
add-on
api.
A
Know
so
there's
right
so
this
one
I
mean,
I
I
think
the
issue
they
ran
into
is
like
a
a
known
limitation
of
any
add-on,
not
not
node
add-on,
node
add-ons
if
I've
understood
correctly.
Okay,
so
that,
like
basically
the
vm
running
context,
which
is
what
they
were
using
here
right,
was
that
or
was
it.
A
So
this
vm
running
context
basically
says
I
want
you
to
run
in
a
context
where
things
like
the
globals
have
their
own
copy.
Okay,
but
there's
no
way
for
like
it
takes
work
for
that
to
be
passed
down
into
add-ons
and
node.
Just
doesn't
do
that
yet
I
mean
I
mean-
and
I
guess
gabriel
correct
me
if
I'm
wrong
here-
but
this
is
this-
is
a
separate
dimension
from
these
context,
aware
add-ons
that
we've
talked
about
right.
C
As
far
as
I
know,
no,
it
should
be,
it
should
be
part
of
context.
Awareness
is.
E
C
Understanding
but
we
we
we're
not
really
testing
this
with
with
with
like
with
this
vm
module
like
built-in
modules,
so
maybe
maybe
we
should
add
some
tests
to
make
sure
that
that
that
the
initializer
gets
called
again.
Well,
I
think
the
answer
is
it
doesn't.
A
C
A
Because
we
weren't
using
the
context
that
was
related,
you
know
we
would
end
up,
throwing
an
error
that
was
created
in
a
different
context
from
where
the
check
was
being
made
and
they
weren't
matching.
A
C
C
E
C
Idea
that
that
you
know
we're
basically
tied
to
a
global,
then
I
guess
you
could
say
right,
because
you
know
each
each
thread
gets
its
own
global
right,
but
so
does
the
context,
and
so
we
need
to.
We
need
to
basically
tie
ourselves
to
the
global,
and
then
that
means
that
means
having
separate
instances
of
the
add-on
for
for
context
as
well
as
threads.
A
C
C
Yes,
well
actually,
the
the
the
end
yeah,
the
m,
if
I
remember
correctly,
doesn't
get
stored
anywhere.
C
C
C
It
all
starts
with
with
nappy
module
register
by
symbol
and
there
we
create
a
new
end.
So
if,
if
nobody
calls
nappy
module
register
by
symbol,
then
then
yeah
then
that's
not,
then
we're
going
to
be
doing
things
wrong
right.
C
C
E
C
You
don't
have
to
today,
okay,
so
so,
okay,
the
the
question
is
when
you
when,
when
you
when
you,
when
you
run
in
in
when
you
run
in
or
you
do
the
vm
running
in
context
or
whatever
it's
called
the
the
scope
of
the
variables
right
that
that
you
have
on
the
javascript
side,
you
know,
can
you
use
those
variables?
Because
if
you
can
then
then
yeah,
then
it
makes
sense
and
then
they
are
somehow
connected
right
and
it's
not
a
separate
environment.
A
C
A
C
C
A
A
Have
so
at
least
that's
you
know
not
it's
I,
but
I
think
it
is
something
that
is
is
like
you
said
again,
it's
unexpected
and
like
we're
seeing
problems
with
jess,
there
was
another
problem
issue
where
I
you
know
investigated
found
it
to
be
the
same.
I
was
wondering
if
it
was
yes.
C
Sorry
go
ahead,
I
was
just
going
to
say
so
this
this
seems
to
be
the
root
cause
is
is,
is,
is
this
distinction
between
native
and
javascript
add-ons
in
in
the
in,
in
the
context
of
this
vm
right
module.
A
A
Caching,
something
and
but
it
it
seems
like
anna
said-
and
of
course
it
was
correct
that
get
current
context
returns
the
same
value
anyway.
A
All
right,
I
think
those
places
where
we
use-
and
vice
li
get
current
context
they're,
just
like
historical.
We
didn't,
you
know
and
context
should
be
doing
that
internally
if
it
needs
to
be
anyway,
so
I'll
probably
submit
a
pr
to
fix
that
up.
But
fundamentally
it's
that
you
know:
we've
hooked
up
the
context,
awareness
first
for
for
threads,
but
not
for
this
particular
case,
which
is
another
case
where
you
can
have
it.
C
But
this
is
actually
this.
This
is
actually
the
exact
opposite
of
what
you
want
for
threads
here.
You
do
not
want
a
new
instance.
In
fact,
you
want.
You
want
the
context
to
to
to
to
work
with
with
existing
values
right,
whereas,
whereas
when
you're
on
a
new
thread
you
want
to
initially,
you
want
to
create
a
new
instance
of
the
add-on
on
purpose
right.
A
Right,
it's
kind
of
like
I
guess
my
understanding
is
that
there
might
be
like
a
stack
of
contexts
yeah,
because
we
have
the
isolate
in
the
context,
so
it
could
like
it
could
be
like
for
a
given
isolate.
You
can
have
multiple
contexts.
A
A
Right
and
that's
that's
effectively,
what
we're
seeing
is
that
the
global
and
the
copies
of
things
on
it
are
they
just
don't
match
like
in
the
case
of
the
one
that
nicola
and
I
were
looking
at
earlier,
like
it
was
telling
us
this
is
not
a
oh,
it's
type
of
exciting.
It's
like
it
was
a
particular
type
of
exception.
A
C
Yeah
so
basically
the
built-ins,
the
javascript
built-ins,
if,
if
they
are,
if
they
are
reinstantiated
for
each
context,.
A
C
Then
you
would
get
you
would
get
this
problem
even
even
without
native
add-ons
right,
so
so
like,
for
example,
let
me
think
about
this
like
if
you
have,
if
you
have,
let's
say:
let's
say
you,
you
create
a
new
error
object
and
then
you
you
and
you,
then
you
evaluate
you,
know
vm
running
new
context
that
error
object,
which
is
in
your
scope,
because
we
know
right
scope
gets,
gets
sent
into
the
context
that
error
object.
Instance
of
error.
Would
you
get
false
for
that
right
because.
A
Right,
but
I
think
the
problem
is,
is
that
what's
happening
is
even
within
that
vm
running
context,
the
javascript
code
actually
respects
the
new
globals
so
like.
If
you
create,
if
you
say,
is
this,
if
you
said
new
exception
and
then
you
said,
is
this
an
instance
of
that
exception?
It'll
say
yes,
because
it's
all
using
the
the
new
contact,
the
globals.
For
that
new
context,
however,
when
you
call
a
native
method,
it
actually
uses
the
outer
one,
because
that's
the
one
that
was
created
when
you
required
it
right
and
therefore,
if.
C
You
have
the
next
context
right
because
we
store
the
context,
and
so
okay,
I
see
so
when
you're
calling
instance
of
you
should
be
calling
it
on
the
inner
context.
So
you
shouldn't
be
retrieving
the
context
from
from
from
whatever
you
stored,
but
then,
but
then,
and
visceral
at
get
current
context
is
the
correct
way
to
do
things
rather
than
end
context.
A
But
that
doesn't
seem
to
that.
That
was
my
question.
That's
what
I
was
wondering,
but
anis
and
it
was
like
no,
it
doesn't
actually
do
that.
C
A
I
I
so
it's
something
like
it
might
be.
It
might
be
something
like
node
needs
to
do
something
to
support
that
or
yeah.
That
was
my
question,
but
yeah
for
that
other
one.
I
tried
actually
like
I'm
like
really
that
that
sounds
like
it
should
be
doing
it,
but
when
you
I,
unless
I
did
it
wrong,
which
is
always
possible,
I
hacked
the
code
so
that
it
it
says,
did
m.
Isolate,
get
current
context.
Give
me
exactly
the
same
thing
as
m
of
context,
and
the
answer
was
yes.
C
C
Let
me
paste
this
in
for
everybody.
Where
is
the
chat?
Here?
Is
the
chat
there
you
go
so
this
is
what
I
ran
just
now
in
the
in
the
like
the
rebel.
C
C
C
C
C
C
A
A
A
C
Okay,
okay
and
then
in
there
inside
the
context,
they
also
create
a
new
date.
A
C
A
C
C
That's
right,
that's
right!
Yes,
and
you
can
pass
it
into
the
new
context.
Okay,
I
see
so
okay.
So
what
this
means
is
our
our
function,
callback,
where,
where
we
translate
the
the
the
way
v8
calls
into
native
to
the
way
we
call
into
native
that's.
C
All
right,
let
me
let
me
let
me
quickly
check
the
code.
This
should
be.
This
should
be
in
js
native
api,
an
edit
note.
Oh
the
srcjs.
A
A
Yeah
so
yeah
it's
worth
it's
worth
thinking
about
because
it's
like,
I
guess
you
know
it's
not
that
it's
specific
to
no
to
to
any
pi
modules,
but
it
is
a
interesting
thing
in
that
we,
you
know
this.
Other
error,
you
know
basically
seems
to
be
related
as
well.
In
that
you
know,
just
was
was
basically
failing,
because
it's
trying
to
compare
errors
and.
A
A
If
you
do
it
in
javascript
all
seems
okay,
whereas
if
you
actually
run
it
in
a
new
context,
then
you
get
the
same
error
that
they
were
seeing
before,
which
is
like
yeah.
This
doesn't
match,
which
makes
sense,
because
you're
again
in
you
know
in
the
the
exception,
was
thrown
in
one
context
and
created,
but
then
checked
in
a
different
one.
C
Yeah,
but
I
mean,
but
that's
okay,
I
mean
that's
actually,
okay,
because
if
it
was
thrown
in
one
context,
well,
you
know
in
so
so
you're
saying
this
gives
you
false
yeah
like
this,
gives
you
a
different,
but
why?
Because
because,
if
the,
if
the
native,
if
the
native
method
running
in
a
new
context
gives
you
gives
you
another
type
of
error,
that
means
it's
using
the
inside
context.
C
Because
right
yeah,
so
so,
if
you're
saying,
if
you're
saying
that
that
it
it
that
the
that
the
add-on
erroneously
stores
the
outside
context
and
uses
it
even
when
running
in
the
inside
context,
right
then
yeah,
then
the
console
log
here
should
give
you
true.
The
second
one.
F
C
A
A
This
is
this
is
yeah,
although,
like
the
actual
error
that
they
have
like
the
thing
that
they
had
actually
does
a
bunch
more
code
like,
I
suspect
that,
in
the
case
of
what
just
is
doing
that,
like
they're
running
a
whole
bunch
of
code
in
so
maybe
maybe
my
test
case
is
wrong.
A
C
That,
or
or
or
our
our
implementation
is
inconsistent,
and
sometimes
it
uses
the
right
scope.
Sometimes
it
uses
the
wrong
scope.
So,
for
example,
creating
an
error
happens
to
use
the
correct
scope,
but
but
other
parts
like
what
like
what
what
the
sqlite
is
doing
with
an
api,
it's
using
the
wrong
scope
or
context.
A
Yeah,
except
that,
like
the
problem
that
they
was
reported,
was
that
just
was
having
problems
right
right,
whereas
like
like,
when
under
just
we
can,
we
can
see
that
it
it's
trying
to
compare
something
which
is,
which
is
looks
like
a
type
error,
but
isn't
a
type
error.
So
it's
the
same
problem
as
we
saw
before
in
that
it
was
like
it's
returning
false
because
they're
not
the
same
thing
now,
maybe
I'm
maybe
I've
read
too
much
more.
Maybe
I've
made
a
mistake
in
terms
of
saying
it's
related
to
vm
context.
C
A
A
C
Actually,
actually,
the
way
to
change
this,
the
way
to
change
this
test
might
be
to
say
to
say,
vm,
running
new
context,
try
add-on
dot,
ror
catch,
something
and
then
something
instance
of
type
error,
so
put
all
that
in
the
double
quotes.
C
We
expect
this
to
be
true
right,
because
I
mean
if,
if
the
situation
is
what
we
think
it
is,
which
is
that
we're
using
the
context
given
at
at
instant
it
yeah
instantiation
time,
then
then
then,
comparing
even
though
the
the
error
was
created
inside
the
vm
context.
When
you
check
that
same
error
outside
the
vm
context,
it
should
be
true
right
because
well,
it
was
created
using
the
outside
context.
E
C
A
A
Okay,
so
that
one's
an
interesting
one
and
a
few
related,
I
think
I
I'm
guessing
these
are
still
related,
although
maybe
I'm
not
sure
anymore,
okay.
So
let's
go
back
to
the
list
we
were
looking
at.
Where
are
we.
A
Okay,
help
oops,
sorry,
okay,
so,
okay,
so
back
on
to
no
confu
workshop
proposal.
B
Yeah,
michael
I've
not
heard
back
from
the
conference.
Okay,
the
deadline
is
a
week
from
today.
B
Okay
and
once
again,
I'd
like
to
open
it
up
again
to
lucas
and
to
kevin.
If
you
would
like
to
be
part
of
the
team
that
works
one-on-one
with
people
having
questions
about
native
add-ons.
F
I
haven't
actually
had
any
time
to
think
about
it.
Okay,
all
right
now!
Let
me
actually
think
about
it
and
get
back
to
you.
B
F
B
F
B
A
week
from
today,
so
just
let
me
know
I'll
drop
you
an
email.
B
B
So
I'll
I'll
get
that
in
this
week,
I'll
get
that
in
before
the
deadline.
A
Okay
sounds
good,
okay,
enable
debug
testing
for
add-ons.
I
don't
think
anything.
We've
done
anything
on
that
burn
down
list
of
issues
raised
by
module
owners.
So
I
think
this
one.
There
is
an
interesting
one
to
talk
about
as
well,
which
is
this
issue.
This
pull
request.
A
A
I
don't,
I
think,
lavell's
certainly
been
working
on
things
like
he
was.
He
was
asking
about
our
app
bayer
a
test,
so
it
sounds
like
he's
working
on
it,
but
my
take
is.
We
should
probably
do
a
release
like
a
302
that
backs
out
the
original
chain.
The
original
fix,
if
we
don't
have
it
by
the
end
of
the
week,
does
that
make
sense
to
everybody.
A
A
G
Yeah,
it
makes
sense.
I
I
I
was
thinking
if
we
can
only
set
the
last
latest
version
on
npm
23.00.
A
Right,
that's
a
great
idea.
I
was
wondering
if
we
could
do
something
like
that
as
well.
Is
that
I
think
we
should
do
that
immediately
if
we
can.
G
Yeah,
so
the
lotus
is
the
3.0.0
and
when
we
are
ready
we
can
prove
this
new
new,
okay.
A
And
so
somebody,
the
only
way
somebody
would
get
301
is
if
they
specifically
put
in
301
right,
yeah,
okay,
yeah.
I
think
that
that
let's
do
that
right
away
because
we
don't
want
we've
already
had
a
few
people
meant.
You
know,
report
issues
and
even
like
our
testing,
I
wouldn't
be
surprised.
If,
if
I
guess
the
question
is,
why
isn't
our
internal
testing
failing?
A
C
A
G
Is
michael
could
could
you
share
the.
A
A
A
G
Yeah
last
monday
all
worked
good
and
yeah.
I
pushed
the
right
the
release.
A
A
C
Yes,
we
do
okay,
it's
it's
in
or
is
it
common
jipi,
okay,
so
in
a
root
directory
common
gp.
C
A
G
The
okay!
What
I
understood
is
that
if
you
now,
the
the
the
patch
on
that
we
made
on
3.0.1,
if
you
are
on
on
windows,
using
the
add
on
include
underscore
beers.
G
Include
this
is
a
rendered,
is
it
will
not
be
rendered
as
a
list,
but
as
a
an
item,
and
it
will
become
to
to
work
and
the
problem
with
the
space
on
the
folder
name
is
related
only
on
unix
system,
because
it
it
worked
good
on
windows,
even
if,
before
this
right
last
patch.
A
G
Okay,
so
I
will
wait
one
two
days
and
if
I
don't
I,
I
will
not
see
any
activity.
I
will
push
some
some
tests.
A
B
A
Okay,
thanks
yep,
thanks
for
all
your
help
on
that
one,
let's
see,
I
think
that
was
the
main
one
that
was
sticking
out
in
my
mind.
I
don't
know
if
people
have
other
ones
that
they're
have
in.
A
A
A
B
A
B
B
A
A
A
C
C
Yeah,
I
just
I
think,
I'm
I
I
think
I'm
I'm
stuck
on
on
that
on
that
problem
with
that
you
just
opened
with
the
with
the
benchmark
things,
because
it's
breaking
it's
breaking
the
ci
for
for
for
this
pr
and
yeah
I
I
thought
I
had
done
something,
but
I
don't
think
I
did
because
this
is.
This
is
something
else,
but
anyway
yeah.
So
yeah,
it's
moving
forward.
C
It's
going
to
have
some
implications
for
know
that
on
api,
because
the
the
way
we
the
way
we
test
for
the
effects
of
of
of
gc,
will
have
to
change
we're
going
to
have
to
use
the
same
approach
as
core
which
is
basically
to
to
to
to
keep
gc
in
an
idle
loop.
Until
you
see
the
the
desired
change
in
state
and
if
you
don't,
then
the
test
fails,
if
you
don't
see
it
after,
like
10
iterations
or
whatever
right.
C
True
yeah,
I
mean
there's
a
lot
of
things
broken
right
now,
so
so
it's
kind
of
hard
to
to
figure
out
which
fix
to
land
first,
because
we
in
core,
we
got
the
benchmark
thing
and
then
in
in
you
know
that
on
api,
we're
still
waiting
on
that
on
that.
What
is
that
called
the
handle?
Scope
thing
right,
your
your
fix
for
the
handle
scope
has
to
be
released
in
order
for
us
to
be
able
to
to
to
test
test
some
of
the
pr's.
C
You
know
that
on
api,
and
then
this
is
another
change,
that's
sort
of
coming,
and
I
don't
I'm
trying
to
figure
out
if
we
update
the
the
g
seeing
now
if
well
yeah.
That
should
be
okay.
That.
C
C
Yes,
true,
true,
okay,
so
so
then
all
right!
So
then
I
guess
I
guess
I
will.
I
will
take
the
opportunity
that
that
the
breaking
benchmarks
have
have
given
me.
Would
I'm
ready
actually
fix
the
or
or
update
the
the
the
stuff
and
know
that
on
api
yeah,
okay,
great
yeah,
I'll
I'll?
Do
that
idle
loop
thing?
You
know
the
thing
is
like
this.
This
comment
dot
this
this
common
stuff
that
we
use
in
in
core
tests.
A
lot
of
that
would
be
useful.
C
A
C
Yeah,
but
I
mean
if
we
wanted
to
share
with
with
cord,
and
that
would
be
a
pretty
big
implication,
because
we'd
have
to
have
another
one
of
those.
You
know,
like
depths
kind
of
sure,
includes
where
it
just
imports
the
whole
thing
wholesale
and
stores
it
in
the
repo,
and
then
you
know
we
we
update
it
occasionally
like
we
do
with
v8.
You
know
only.
C
Yeah,
it's
yeah
exactly
so
I'm
not
sure
if
that's
ever
going
to
happen,
but
but
anyway,
for
now
we
can
just
keep
copying
the
the
stuff
that
we
need
and
know
that
on
api.
A
Right:
okay,
I'm
just
going
to
flip
back
just
at
one
minute.
I
think.
C
C
Well,
it
was
well
I
mean
it
was
no,
it
was
pretty
consistent.
I
I
don't
think
it
was
it
was.
It
was
luck.
It
was
that
that's
just
how
the
gc
worked
before,
but
now
now
they've
they've
quote:
unquote,
unhinged
it,
and
so
so
you
know
and
if
it
can
run
without
without
a
stack,
that's
ready
to
execute
javascript.
If
now
there
is
such
a
possibility
and
occurs
frequently,
then
then
we
have
to.
C
We
have
to
defer
using
set
immediate
to
ensure
that
we
have,
you
know
a
stack
capable
of
running
javascript,
it's
as
simple
as
that
and
and
then
the
implication
of
that
is
that
all
our
tests
now
must
must
must
assume
that
gc
is
asynchronous
as
opposed
to
synchronous,
and
so
so
that's
that's
why
the
the
gc,
until
basically.
E
A
E
A
C
C
A
Okay,
so
let's
see
is
there
any
since
we're
over?
A
A
B
C
B
A
If
not,
everybody
have
a
great
week
and
we'll
talk
to
everybody,
except
for
jim
next
week
and
we'll
talk
to
you
the
week
after
I
think
yeah
yeah.
Actually,
although
I
might
be
on
vacation
so
who
knows,
I
think
I'm
trying
to
remember
it's.
It's
I'm
not
away
next
week,
yeah!
No,
I'm
on
a
vacation
the
week,
you're
back
but
anyway,
so
I'll
see
in
a
little
bit
all
right.
Well,.