►
From YouTube: 2021-08-27-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.js
node
api
team
meeting
for
august
27
2021
before
we
get
started,
does
anybody
have
any
announcements
they'd
like
to
share.
B
Yeah
I
have
one
we
released
a
new
version
for
another
tone:
api,
the
version
4.1.0.
B
Essentially,
in
this
new
version
we
added
a
new
type,
maybe
a
class
to
handle
pending
exception
when
cpas
plus
exception
disabled.
We
added
wraps
for
cleanup
books
and
we
made
the
other
minor
fixes
and,
as
usual,
improved
the
test
suite
and
and
documentation
so
yeah.
This
was
my
my
announcement.
A
A
A
C
Maybe
go
ahead,
I'm
thinking
what
was
it.
I
think
we
basically
concluded
that
we,
we
can
do
symbol
at
least
because
the
the
hermes
engine,
if
I
remember
correctly,
treats
like
literals
as
literals.
C
So
you
know,
there's
no,
there's
nothing
to
reference,
unlike
v8,
which
treats
everything
as
like
a
value
which
can
be
which
can
have
a
persistent
and
a
weak
reference.
So
we
could
potentially
extend
the
the
ability
to
create
persistent
and
weak
references
to
symbols
without
much
work
at
all.
C
Yeah-
and
I
think
we
have
it
if
I
remember
correctly-
symbols
are
treated
as
objects,
vladimir,
I
think
looked
at
that
and
and
he
he
concluded
that
symbols
are
treated
as
objects
in
her
needs,
and
so
we
have
what
we
need,
because
the
objects
can
have
weak
references.
C
D
A
A
A
I
guess
this
might
be
partially
because
of
new
releases
drive
some
new
new
uploads
or
whatever,
but
that
looks
good,
there's
nothing
to
discuss
there,
creating
the
matrix
of
tested
methods
jack.
I
know
you've
been
working
on
that.
Do
you
want
to
have
an
update.
E
Yeah,
so
I
I
think
I've
documented
all
the
classes,
apart
from
threats,
functions
and
type
traffic
functions,
and
I
opened
up
a
draft
pr
for,
like
the
like
the
work
like
we
had,
I
was
I
had
a
question
about
how
we
should
test
various
functional
overloads
for
constructors,
for,
like,
like
stuff,
like
start,
say,
function
where
it
has
like
a
bunch
of
overloads,
and
I
think
gabriel
suggested
to
have
like
some
sort
of
front
clasps,
and
I
have
it
like
a
pr
open
for
it.
E
E
E
E
So
I
I
and
so
basically
I
just
like
inside
of
the
actual
napi.h
I
placed
the
friend
class
there
inside
of
the
async
worker,
but
I
added
like
a
surrounded
with
like
a
macro
saying:
okay
for
our
testing,
so
basically
we're
running
the
test.
Suites,
that's
what
we're
gonna
sort
of
include.
Otherwise,
we
don't
and
yeah
yeah.
C
C
It
just
sucked
just
a
sec,
sorry
sure
yeah,
basically
yeah,
that
was
the
idea,
is
to
to
just
use
this
rent
class
so
that
you
can
just
plain
old
test,
just
the
state
without
having
to
go
through
the
whole
long
test.
C
E
A
C
C
A
C
F
E
Nothing
on
the
like
best
friends,
but
there's
something
like
a
random
questions,
about
threats,
functions
and
on
the
doc
on
the
dock.
For
it
it
says
if
an
add-on
creates
additional
threats,
we
require
like
energy,
even
v
value
and
ref,
not
to
be
called
I'm
just
wondering
from
like
those
threats
that
we
created
from
the
add-on.
Just
I'm
just
wondering
why
that
is.
C
F
E
C
A
A
Okay,
so
each
of
these
is
open
for
a
class
and
if
it's
all
covered
it'll
be
closed
right.
Yes,
okay!
So
that's!
I
was
just
trying
to
understand
this.
So
if
I
look
at
any
one
of
these
yeah
there's
always
some
gap,
okay,
perfect
and
then
the
idea
is
to
I'd,
add
something
for
the
ones
that
aren't
and
yeah
okay.
So
these
are
all
good.
A
A
Okay,
thanks:
let's
go
back
to
here
crete
semester,
so
yeah!
Unless
you
have
any
other
questions,
we
can
move
on
on
that.
One.
D
Yeah,
hey
michael
yeah,
thanks
yeah,
I
have
two
pull
requests:
a
new
particular
request
that
you
know
we
can
go
over.
It's
very
simple.
D
All
right,
the
first
two
one
one
is.
A
D
Yeah,
so
that
is
just
for
allowing
to
give
a
build
path
in
the
you
know
the
test,
so
that
we
can
even
pick
it
up
from
a
relative
path.
It
does
not
it
the
test
pass
all
the
tests
pass
in
this.
If
you
are
okay
with
so
basically
what
I'm
doing
is
I
in
the
in
the
one
of
the
function
parameters.
I
am
passing
in
a
relative,
build
path
and
also
defaulting
it
to
a
you
know:
a
relatable
path.
Maybe
the
naming
can
be
better
there.
D
D
The
story
that
I'm
working
on
I'll
be
able
to
point
at
the
generated
binding
dot
cc
right
that
that
I'm
so
the
intention
of
that
pr
was
to
generate
the
custom,
build
into
a
different
location
in
inside
the
unit
test.
Folder
right.
So
this
way
I'll
be
able
to
ask
the
test
to
run
on
top
of
that.
A
Right,
okay,
the
the
only
question
I
wonder
here
and
I'll
type
it
is
this
is
like,
including
this
slash,
makes
sense.
D
Well,
if
you
want
to
do
any
other
path
outside
of
the
build
directory
right
and
outside
the
source
directory
outside
of
the
git
repo,
then
I
think
even
that
can
be
done
like
we
can
step
out
of
the
report,
a
repository
and
build
it
somewhere
else.
But
I
think
mostly,
we
are
generating
the.
D
Yeah,
it
could
be
made
better
yeah.
Yes,
that's
true
that
that
could
be
better
ways
of
doing
this.
Yes,
I
agree.
Yeah.
F
D
F
A
C
D
The
other
pull
request
that
I
have
is
for
the
unit
test
names,
the
first
one,
so
I
just
attempted
to
change
the
unit
the
file
names,
because
most
of
the
file
names
are
not
standardized
like
we
have
so
the
challenges
that
when
we
want
to
filter
out
and
do
the
binding.js
and
you
know,
create
the
export
object
from
the
each
one
of
those
tests
are
almost
similar
to
the
file
names
right
right.
It
will
be
easier
if
it
is
standardized
this
way
so
that
I
can
change
from.
D
I
think
this
is
the
this
is
not
the
camel
case.
I
can
change
it
to
camel
case,
so
I'm
just
looking
for
suggestions,
but
if
it
isn't
say
if
it
is,
this
is
in
a
camel
case
right
where
we
have
the
underscore
so
then
I
will
be
able
to
change
it
to
you
know:
standardize
capital
letters
remove
on
the
all
the
underscores
removed,
and
then
that
would
be
the
export
object
and
that
would
be
easy
to
do
the
generate
the
binding.cc
in
our
unit
test
filtering
pr.
D
So
I
just
intended
to
change
the
names,
but
when
I
tried
to
push
the
code
the
lint
complained
about
you
know
I
had
to
do
a
lint
fix.
If
there's
a
way
to
bypass
the
lint
fix,
then
that
would
be
much
better,
because
I
all
I
intended
to
do
is
just
a
file
name.
Is
that
a
way
to
bypass
linting.
C
A
A
D
Making
that
having
the
dashes
or
the
underscore
helps
me
to
generate
the
export
object
from
within
right,
because,
if
there's
a
continuous
running
string,
I
I
wouldn't
know
the
the
export
objects
are
all
capitalized
right,
async
with
an
a
caps
and
the
context
to
the
c
caps.
It's
a
it's.
D
So
it
is
easier
for
me
this
way
to
just
change
from
one
case
from
a
camel
case
to
a
capitalized
format
to
generate
that
was
just
to
generate
the
export
object.
Names
to
be
placed
in
that
can
be
placed
in
the
binding
cc
right.
A
newly
generated,
binding
cc
custom
generated
binding
dot,
rc's.
C
A
D
D
And
if
you
go
to
the
bottom,
you
know
very
the
very
bottom
run
to
the
very
bottom.
E
D
The
export,
if
you
look
at
the
export
the
unit,
the
unit
that
is
exported
in
it
array
buffer.
If
you
look
at
the
array
buffer
right,
the
a
is
caps
and
b's
the
array
buffer,
as
a
word,
has
a
caps
and
b
caps
right.
So
what
I'm
doing
is
once
you
provide
a
filter,
filter,
name,
I'm
going
and
generating
the
binding
dots
you
see
at
that
time.
The
binding.cc
file
should
have
all
this
fact.
We
are
generating
custom,
binding.cc
yeah
if
you
want
to
filter
and
run.
D
In
that
case,
I
need
this
name.
The
init
arrival.
For
that.
That
is
a
you
know,
that's
the
comp
that
should
be
part
of
the
binary
compilation
of
you
know
the
binding
dot
newly
generated
binding.cc
so
that
it
can
it
can.
You
know,
gonna
get
all
this.
So
that's
the
reason.
D
F
C
A
D
It's
the
other
way
around
from
the
file
name,
I'm
trying
to
generate
the
unitary
buffer.
You
know,
then
the
name
so
if
it
is
a
dash
buffer.
A
D
A
Okay
yeah,
if
you
just
commented
like,
if
you
can,
if
it's
almost
like
in
the
commit
comment,
you
could
update
the
commit
to
say
you
know
the
that
they
know,
though,
changing
the
names.
But
then
you
know
to
for
the
other
generation.
But
then
you
know
also
fixing
white.
You
know
the
linter
comment
and
any
other
changes
other
than
file
name
changes
are
fixing
the
linter.
D
I'll
mention
all
that
in
the
comments
yeah
I'll
put
that
in
the
comments
and
then
I'll
add
a
get.
The
comment
comment
description
as
well.
D
No
once
this
is
done,
the
the
other
pr
should
be
done.
Pretty
quick
like
in
a
couple
of
days,
probably
by
the
next
meeting,
we
will
be
able
to
review.
Yeah
sounds
great.
A
Okay,
so
on
to
the
next
things
enable
debug
test
for
add-ons.
I
know
I
haven't
done
that
so
now,
looking
at
issues.
Actually,
this
is
the
one
I
should
probably
I
can
bring
up.
I
I
added
this
one.
There's
sort
of
an
ask:
somebody's
been
working
on
a
serial
port.
F
F
C
B
I
think
worker
instead
of
the
of
calling
the
c
api
to
do
the
same
thing
that
I
think
worker
do
and
then
and
then
they
yeah
they
don't
didn't
check
for
the
status
no
when
they
call
the
on
node
api,
so
yeah.
This
is
the
the
what
we
we
did
till
now.
A
C
I'm
also
looking
at
the
pr-
and
I
think
you
know
you
know-
moving
to
async
worker-
might
be
like
a
big
commitment
on
their
part,
because
it's
like
a
different
way
of
of
writing
the
code,
because
you
have
to
like
organize
something
as
a
subclass
of
async
worker
and
that
might
be
too
big
of
a
change
for
them.
So
what
I,
what
I
suggested
to
them
as
an
alternative
to
moving
to
async
workers
to
just
use
like
nappy,
throw
it
fail
so
as
to
process
the
return
values.
C
You
know
without
migrating
to
like
this
different
way
of
doing
async
work,
so
then
they
can,
they
can
still
mix.
I
mean,
though,
the
whole
point
of
know
that
on
api
is
that
you
can
mix
in
plain
old,
core
node
api
calls
right,
and
so,
if
they're
doing
that,
then
that's
great
and
it's
okay
and
it's
supported,
but
they
should
process
the
return
values
using
the
convenient
macros
that
we
provide
right,
which
is
this
an
api
throw
with
bail?