►
From YouTube: 2021-10-15-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
B
B
I
think
that
yeah
we
need
some
other
times
but
yeah
all
it's
going
on
well
for
okay!
Well,.
A
Okay,
let's
look
at
our
stale
issues.
I
think
we
are
down
to
one,
and
this
is
the
one
I
know
jack.
You
spent
some
time
looking
at
it.
A
I
don't
know
if
you
have
anything
you
want
to
add
to
that.
I
haven't
had
a
chance
to
really
read
in
detail.
I
don't
know
if
anybody
else
has.
D
Could
you
go
down
just
a
little
bit
in
this
last,
the
second
to
last
paragraph.
It
ends
in
which
the
program
invoked
who's
cons,
which
eventually
gets
rejected
by
the
eight,
with
a
raw
string
exception.
D
F
A
And
thanks
for
everybody's
who's
been
involved
in
looking
at
that,
let's
go
back,
so
that's
all!
We
have
under
stale
issues
so
not
too
much
there
tracking
issues
for
modules
that
are
reported.
I
guess
nothing
new
to
report
this
week,
but
we'll
just.
F
A
A
A
Right,
okay,
maybe
it's
here
create
a
matrix
of
tested
methods,
so
I
guess
that's
usually
jack
where
we
you
know
see.
If
there's
anything
you
want
to
discuss
or
things
we
should
look
at
on
the
test.
C
Yeah,
I
just
have
a
I
just
fixed
up
that
my
test
for
function,
references
based
on
some
feedbacks
and
yeah.
I
think
it's
ready
for
review
now.
A
A
All
right,
okay,
here
we
go,
I
haven't
enabled
debug
testing
so
before
we
go
on,
so
I
guess
deepak
do
you
want
to
have
anything
you
want
to
talk
about
on
your
your
front
of
improving
the
ability
to
like
run
individual
tests.
G
I
think
last
time
when
I
demoed
kevin
and
others
were
not
here,
gabriel
was
not
here.
Do
I
can
do
the
demo
again?
You
know
just
a
quick
demo
because
because
there
was
only
small
changes
from
last
time,
let
me
do
it
sure
I.
G
G
Start
then,
it
would
run
everything
within
the
thread
safe,
folder
right,
so
it
picked
up
everything
from
the
thread
subfolder
because
I
gave
a
star
at
the
end.
It's
a
wild
card.
A
A
G
And
I
can
also
give
like
multiple
options.
I
can
quote
it
and
then
deal
with
deliver
the
my
filter
conditions
with
space
npm
run.
H
So
I'm
wondering
since
we're
on
the
since
we're
in
the
process
of.
H
You're
breaking
up
gabriel
of
scripting
this
up
to
oh
you're
back
now:
okay,
yeah,
okay,
so.
G
Okay,
so
here,
if
you
see
I
I
I
added
a
star
in
the
front,
so
it
picked
everything
that
ended
with
the
function.
All
the
folders
there
is
a.
There
are
a
couple
of
folders
here
that
I
ended
with
a
threadship
function.
G
It
it
picked
that,
and
from
last
time
there
was
a
there
was
a
limitation
from
our
last
time
when
we
did
the
when
I
did
the
presentation,
if
there's
some
folder,
that
is,
that
matches
the
wild
card,
a
file
that
matches
the
wild
card
was
not
picked
up
right.
The
folders
were
given
more
priority.
I
have
fixed
that
as
well,
so
the
file
is
also
picked
up
right.
G
This
just
matched
the
object,
wrap
file.
It's
a
file.
Yeah
object
is
just
one
single
file
yeah,
so
it
ran
all
the
tests
right.
It
also
go.
It
compiled
all
the
selected
conditions
that
match
the
filter,
filter
options
and
then
it
also
went
ahead
and
ran
the
tests
on
them.
G
A
G
Yes,
yeah,
I
think
they
are
valid
comments.
Minor
comments
like
why
do
we
need?
I
can
answer
them?
The
condition
between.
I
can
answer
that
in
the
pr
here.
The
reason
why
we
are
having
this
is
a.
Why
did
I
add
it,
I'm
just
thinking,
maybe
I
have
to
think
on
how
to
play
back
in
the
pr
I'll.
Let
me
reply.
D
I
just
took
a
look
at
this
a
little
before
the
meeting
just
wanted.
H
G
D
None
of
them
are
like
blocking
right
for
us
to
merge
this
in,
though,
because
you
know
like
this
pr
is
getting
quite
large.
So
if
we
wanted
to
merge
this
and
then
do
incremental
updates
to
it
that
that
would
be
fine
too.
H
A
G
Sure,
yes
I'll,
let
me
address
the
comments
right
away
and
then
probably
like
in
a
couple
of
hours
I'll
thank
kevin,
so
we
should
be
ready
to
merge
this.
I
think
kev
gabriel
has
put
in
his
comments,
so
it
would
be
easy
to
run
with
make
flags.
H
Yeah
so
yeah,
basically
I
I
noticed,
I
don't
know
if
you
can
hear
me
properly,
but
I
I
noticed
that
when
you
ran
this,
it
ran
only
one
job
at
a
time
right.
So
so,
if
you
run
no
jip
internally,
then
you
know
you
can
easily
get
like
require
os
cpu's
length
from
from
node
right
right
and.
F
H
A
H
To
minus
j
and
then
that
number
returned
by
node
that
will
that
will
run
your
build
in
parallel,
so
so
that
it
runs
as
many
jobs
as
you
have
cpu.
G
H
H
Yeah
I
mean
that
the
test
runner
will
run.
However,
it
runs
that
this
is
not
going.
H
H
H
Calling
it
right-
and
this
would
automatically
adjust
itself
to
to
the
user's
system
right
right,
okay,
if
not,
then
if
it's
not
easy,
if
we
don't
call
nodejib,
if
we
rely
on
like
you
know,
npm
to
call
no
jib
for
us,
then
then
we
can
just
document
that
that.
H
A
B
Now,
in
the
documentation
about
the
tests,
we
have
only
a
paragraph
on
the
readme
file,
so
my
opinion
is
that
we
should
move
the
content
of
that
paragraph
in
a
file
in
our
documentation
and
explain
all
what
we
added
with
these
new
new
game.
Sure.
G
B
Yeah
in
readme,
you
can
move
the
section
test
on
a
separate
file
and
in
the
header
you
can
link
the
file
the
the
the
new
file.
So
we
have
in
this
way
my
opinion.
We
could
have
better
organized
documentation
about
the
the
the
tests
and
how
it's
possible
to
execute
or
add
a
new
test.
Okay,.
B
B
Merge
all
your
work
that
you
did
until
now
and
then
you
can
concentrate
on
right
this
documentation.
B
So
with
all
what
you
you
you,
you
explain
what
to
you
did
and
that
you
can
start
to
improve.
So
my
my
idea,
I
I
don't
know
if
you
agree,
yeah.
A
Okay,
so
that's
something
yeah,
that's
looking
pretty
good
and
it'll
be
nice
to
get
the
first
piece
and
the
good
first,
a
few
good
follow-ons.
After
that.
Let
me
go
back
to
the
okay.
If
I
can
find
there,
we
go
okay,
I'm
gonna
share
again.
A
So
I
think
the
next
things
we
have
on
is
to
look
at
issues
in
the
node
and
node
on
repos.
I
don't
know
if
there's
any
ones
that
people
want
to
call
it
to
start
with.
D
A
D
D
But
I'm
not
sure,
like
we
discussed
how
the
the
replications
are
for
breaking
changes
and
whatnot
when
you
change
it
from
non-from
const
to
non-const.
D
There's
that
and
then
there's
also
some
oddity
happening,
where
signal
isn't
being
triggered
for
the
async
progress
q
worker,
but
it
is
being
triggered
for
the
async
progress
worker
or
async
worker.
Excuse
me,
so
that's
like
a
secondary
issue,
and
that
is.
D
This
was
reported
in
his
car.
The
first
comment:
he
edited
it
to
add
the
fact
that
this
was
failing,
because
he
added
his
test
to
show
his
correction
right.
A
D
Vr
and
I
guess
perhaps,
signal
wasn't
being
tested
for
this
method,
I'm
not
sure.
D
So
now
he's
saying
that
he
tried
to
add
the
signal
test
for
his
pr
to
test
it
and
now
it's
failing
and
that
he
says
that
it's
only
for
one
of
these
specializations
of
progress
worker.
H
D
We
can
look
into
so
I
guess
we
can
discuss
this
sort
of
one
at
a
time,
the
first
being
the
const
removal,
I'm
not
sure
the
breaking
change-ness
of
it
was
the
thing.
A
A
D
H
D
This
method,
this
this
method,
doesn't
modify
the
met
its
own
properties,
which
it
doesn't.
D
A
D
A
D
H
A
Right,
I
guess
we're
like
we,
don't
actually
think
it's
const
in
the
end,
but
maybe
maybe
we
should
just
maybe
it
should
be
that
we
have
like
a
new
signal
and
deprecate
the
old
one.
A
H
Make
you
go
ahead,
I'm
just
saying
we
have
a
few
const
casts
in
in
the
code
like
I'm
just
looking
at
the
implementation
of
of
of
node
api
like
in
core,
and
we
only
have
two
of
them
in
the
entire
code
right.
A
A
A
The
name
will,
maybe
perhaps
be
a
not
as
nice,
but
I
don't
know
what
we'd
call
it
right.
A
D
A
A
D
D
D
So
we
really
would
only
be
making
a
duplicate
of
use
of
user-facing
apis.
I
think
only
the
the
execution
progress
signal.
D
D
D
I'm
just
not
sure
of
I
guess
I
can
try
taking
his
his
tests
in
a
different
versions
of
an
api
header
files
and
seeing
if
it
just
compiles
or
not.
A
D
Yeah:
okay
I'll
tell
you
that.
D
D
Yeah,
I
think
now
I
can't
like
one
is
just
like
a
q
worker
is
backed
by
a
threat,
say,
function
and
the
norm.
The
async
worker
is
not,
and
that's
why
the
async
works
like
there's
different.
D
A
Sounds
good
like
it's,
maybe
so
behind
them
they
have
a
different
the
q
yeah
anyway
yeah.
If
you
can
take
a
look,
and
it
may
be
again
like
signal
didn't
work
at
all,
you
could
never
have
called
it.
So
that
would
if
that
is
the
case
that
gives
us
the
flexibility
to
say
well,
like
you
know,
should
async
progress,
q
work
or
have
a
signal,
and
so
forth.
A
A
E
A
Okay,
so
yeah,
so
one
was
like
this
was
some
discussion
on
one
of
the
issues
jack
opened
and
it's
it's
interesting
like
we
only
have
a
single,
extended
error,
info
per
m
for
performance
reasons,
and
it
even
our
examples,
were
basically
doing
the
wrong
thing
and
that
they
were
using
that
extended
error.
They
were
getting
it
but
then
making
another
call
and
then
using
it
which,
because
it's
a
shared
structure,
it
doesn't
work.
A
H
A
H
I
I
don't
know,
I
think
I
think
that's
pretty
safe
yeah,
okay,
yeah
yeah,
basically,
okay,
yeah.
A
D
A
H
So
I
don't
know
that
performance
matters
that
much
unless
it's,
unless
it's
a
an
error
that
that
implementation
or
that
the
application
knows
how
to
handle-
and
it
happens
in
a
tight
loop,
I
mean
that's,
that's
a
pretty
long
list
of
ifs.
A
H
Exactly
because,
and
because
in
in
this
in
this
crazy
situation,
you
are
in
a
tight
loop,
then
you
are
very
likely
or
very
easily
end
up
with.
You
can
very
easily
end
up
with
like
a
memory
leak
right
because
you
forgot
to
if
you
forgot
to
free
it
now
now
it
explodes,
because
you
have
so
many
things
copied.
A
Right
so
this
is
right
correct,
so
I
think
he
then
pointed
out
that,
like
we're
actually
not
clearing
the
error,
which
is
correct,
so
jack,
I
don't
know
I
commented
on
the
other
issue.
If
you
want
to
take
a
look
at
a
pr
for
that.
A
The
cleanup
of
what
we
call
that
error
message.
B
D
A
I
don't
think,
there's
anything
else
new,
then,
okay,
so
any
other
things
we
should
talk
about
this
week
before
we
close
out.
C
Sorry
just
have
a
question
for
the
error
and
exceptions
is
that's
okay,
yep
yep?
So
what
do
we
do
about
the
like
the
how
to
like
handle
it
within
the
air
new
implementation?
Because,
right
now,
like
we
they're
sure
the
structure
gets
mutated
so,
like
we
sort
of
don't
handle
some
of
the
cases
right,
which.
C
A
Right
so
this
code
right
yep,
so
I
think
what
you
need
to
do
is
once
we
get
this
last
in
error
info.
You
need
to
copy
this.
So
basically
do
this
up
here.
Okay,
right
after
you
know,
because
basically
this
call
will
mutate
what
you've
got
back
in
info.
So
you
need
to
copy
out
what
you
need
from
there.
C
Yeah-
and
I
think
the
thing
is
like
if
we
keep
the
error
code,
I
think
the
code
as
it
is
now
we
might
end
up
with
a
null
pointer
at
the
end
for
air,
because
the
switch
statement
doesn't
handle
like
certain
like
cases.
If
we,
if,
for
example,
if
it's
nappy,
if
we
expect
an
object,
it
won't
handle
it.
So
we
won't
actually
create
an
error
and
subsequently
we
won't
pass
anything
to
the
sort
of
the
constructor
and
get
a
reference.
C
Yeah
the
air
code
like
yeah
after
yeah
after
you
call
not
the
exception
pending,
so
that
will
always
be
okay,
so
we
always
go
to
the
default
case.
No
matter.
A
You
know
and
then
two
two,
like
you
know
in
the
the
other
one
I
just
had
like
you
know,
a
const
car
store
error
message
equals
this
right
yeah
and
then
you
can
copy
the
as
well
the
error
code
out
and
then
use
those
instead
of
whatever
was
returned
in
info:
okay,
okay,
because
yeah
you're,
no
gay.
Basically,
once
you
make
this
call,
the
contents
of
that
info
is
no
longer
valid.
The
doc
actually
says
that,
but
it's
very
easy
to
not
like
our
code.
Does
it
doesn't
doesn't
do
the
right
thing.