►
From YouTube: Node.js N-API Team meeting - January 28 2019
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
So
basically
in
11
is
cleaver,
so
I
don't
think
I
need
any
special
backporting
there
and
I
have
pr's
in
place
for
10
and
8.
But
I
noticed
that
with
10
there's
a
bit
of
a
problem
because
in
11
we
stopped
supporting
FreeBSD
10.
However,
when
we
seem
to
have
upgraded
the
infrastructure,
so
us
to
drop
support
for
FreeBSD
10,
because
it's
no
longer
like
version
temple
X,
which
is
the
branch
against
which
I'm
submitting
the
P,
are
no
longer
compiles
on
FreeBSD
10.
A
B
Yeah,
like
I'm,
not
sure
what
to
do
right
like
if
we,
if
we
are
not
going
to
be
sending
any
any
greater
than
greater
than
or
equal
to
11
CI
jobs
to
FreeBSD
10
them.
Maybe
we
should
leave
the
infrastructure,
as
is
on
FreeBSD
10,
so
that
the
older
ones
will
work,
because
I
noticed
I
noticed
that,
for
for
for
greater
than
or
equal
to
11
the
the
the
the
job
for
FreeBSD
10
is
sort
of
grayed
out
so
I'm
assuming
there's
some
sort
of
flag.
That
says,
ok,
don't
don't
build
on
FreeBSD
10,
though.
B
A
B
Yeah
yeah
yeah.
That's
that's
the
other
possibility
because,
like
I
like
I,
checked
the
I
checked
the
line,
you
know
dot,
eight
or
note
file
dot,
which
is
causing
the
errors
and
and
that
that
I
checked
the
last
commit
that
touched
it
in
intend
and
I
tried
to
run
a
build
against
that
commit
and
that
commit
fails
so
that
if
I
were
to
do
a
bisect,
that
would
probably
be
the
commit
that
needs
to
be
removed.
But
I'm
not
sure
because
it's
there
that
commit
happens
to
be
a
refactor.
So.
B
B
A
A
C
A
B
A
B
A
B
A
B
A
A
A
Ten
no,
no
same
thing
so
and
I
did
one
without
my
pair
as
well
so
yeah,
it's
just
I'm,
pretty
sure
it's
broken
in
both
of
those,
and
so
that
means
it's
even
something
that
was
back
ported
or,
like
you
said,
maybe
some
infrastructure
or
whatever
but
I,
think
yeah
I
think
once
we
can
get
the
release
team
engaged
they'll
either
say
we
know
about
that.
All
that's
new
or
it's
a
CI
thing
and
we
can
figure
out
how
to
push
it
forward.
A
A
Okay,
I
think
that's
good,
we'll
get
some
hopefully
we'll
go.
Bethany
can
help
and
we'll
figure
out
what's
going
on
there.
Next
one
is
353
focus
on
async
worker.
A
B
B
A
A
A
A
B
A
B
A
A
D
B
A
C
Just
wrote
a
a
to
wait,
that
is
that
he
is
the
maintainer
and
he
published
the
target
version
today.
I
think
one
two
hours
ago
and
I
work
it
on
on
be
gripped.
Essentially,
I
reported
again
because
they
changed
a
lot.
All
the
backend
code
in
a
function
and
I
are
the
support
for
low
preach.
It
and
I
have
a
new
focus
on
decrypt
repo
and
all
is
green
on
the
CI
and
I'm
waiting.
C
A
C
Thank
you
to
to
to
ask
him
so
maybe
if
I
didn't
receive
news
from
McRib
Matias
making
I,
remember
you
if
they
see
more
people
that
you
ask,
maybe
they
just
give
us
an
answer
right
just
know:
yeah.
C
C
A
Okay,
so
looking
looking
at
the
list,
so
I
know
Glen
unfortunate,
Glen
and
Edgar
had
to
drop
off.
It
sounds
like
they're
very
busy
in
terms
of
what's
going
on
for
them
today,
but
I
know
that
Glen
has
reached
out
to
the
maintainer
of
serial
port
and
got
his
Arduino
I
think
it
was
either
a
ship
door
on
its
way
or
arrived,
but
he's
starting
to
look
at
doing
a
port
of
that
one
in
terms
of
refuge.
Us
talked
about
I
cough,
so
this
one.
A
C
B
B
A
B
A
A
A
It's
it's
been
a
little
well,
so
welcome
back,
but
I
think
when
we
last
discussed
this,
we
thought
the
next
step
was
since
you'd
worked
on
level
down
to
figure
out.
If
you
could
go
back
and
talk
to
the
original
maintainer
to
see
what
the
path
might
be
to
getting
the
N
API
version
landed
and
and
the
default
okay.
E
A
Well,
I,
guess
it's
the
it's
the
bring
it
up
to
date
and
then
see
about
the
next
like
it's
whatever
the
next
steps
already
get.
It
landed
because
it's
one
of
these
ones
that
it's
gonna
break
in
twelve
I
think
well
we're
impacted
by
the
API
change,
I
guess
this
ones,
which
were
at
risk
and
that
our
incision,
so
it's
good
to
get
them
moved
over
to
any
fee.
I
yep
make
sense.
It's
a
separate
list
of
ones
that
are
actually
gonna
break
in
twelve
I.
Think
that's
somewhere
else.
A
We
should
dig
out,
but
I
think
part
of
its
overlap.
Okay,
so
you're
you'll
take
dashing.
Do
that
sounds
good
micro
time,
so
Nick
has
already
got
that
under
the
way
time.
Glenn
has
a
PR
and
is
engaged
in
getting
that
landed,
SQL
lights.
Jim
has
taking
the
leadership
on
that.
One
I
saw
you
reached
out
to
the
maintainer
who
had
a
question
which
I
tried
to
answer.
I'm,
not
sure
I
did
a
great
a
good
enough
job,
but
we'll
see
I
think
on
that
front
rate
right.
A
Node
reports
has
landed
as
emergent
212,
and
you
know
we
said
it
wasn't
a
good
candidate,
but
it's
it's
going
for
12
and
beyond.
It
won't
be
an
issue
anyway
and
0
mq
and
it
sounds
like
there
was
already
a
version
to
be
merged
back
to
master.
So
that
is
just
a
matter
of
we'll
hop
and
it
sounded
like
so
I
think
unless
anybody
has
sign
to
add
we're
good
on
that
front.
B
B
After
after
talking
to
to
to
ally
some
more
in
in
the
in
the
comments
on
the
on
the
electron
PR,
it's
becoming
clear
that
that
the
the
solution
that
he
proposed
seems
to
be
the
simplest
one,
but
that
puts
that
puts
an
API
in
an
interesting
position,
because,
because
the
the
whole
node
module
version
checking
does
not
apply
to
an
API
modules
right.
They
they
already.
A
A
B
Snow,
so
so
the
so
this
module,
this
ABI
declaration
may
still
apply,
but
may
may
its
its
usage
may
be
restricted
to
just
to
just
napi
modules,
as.
B
A
A
A
B
A
A
Well,
you
won't
have
to
change
your
code
for
d8
stuff,
like
if
you're
using
open
SSL
it's
a
different
version
of
open,
SSL
or
whatever
yeah
but
you're.
Making
that
choice
to
say:
okay,
wait,
a
sec
I,
don't
get
that
I,
don't
get
the
free
and
therefore
now
the
only
thing
I
wonder
is:
do
we
need
to
yeah?
A
B
Yeah
I'm
yeah:
how
do
we
like?
How
do
we
tell
the
napi
module
loader
that
to
create
a
an
old
module
structure
with
a
version
rather
than
a
negative
one,
yeah
and
yeah?
There
are
some
tricks
because
then,
then
the
entry
point
is
different,
so
ready
we
may
have
to
reach
into
like
the
the
the
DL
open
implementation
that
we
have
in
a
node
binding
dot
right
CC.
So
there
might
be
a
bit
of
an
interaction
there,
but
yeah
yeah
right
off
the
bat.
It
doesn't
look
impossible.
A
Yeah
I
think
today
it
would
be,
like
you
know,
know
if
you're
reaching
into
those
other
things
like
libuv
or
anything
else.
You
shouldn't
be
declaring
yourself
as
any
PI,
but
then
we're
restricting.
Who
can
use
it,
which
is
exactly
unreasonably
like
we
should.
We
should
basically
make
it
so
that
anybody
can
write
using
an
API,
but
I
still
have
the
safety
net.
So
yeah,
okay,.
A
E
A
A
B
Sometimes
the
only
reason
we
have
a
different
node
module
version
is
we
have
a
different
version
of
the
yay
open,
SSL
hasn't
changed,
libuv
hasn't
changed,
crawled
and
and
an
N
API
module
will
never
be
using
v8
right.
So
now
we're
with
time
we're
unaware
unnecessarily.
You
could
argue
that
we
are
unnecessarily
tying
an
N
API
module
to
changes
in
v8
which
it
doesn't
care
about
anyway.
So
so
you
know
it
seems
like
it
seems
like
not
enough
granularity
to
me
so.
A
B
B
A
I'm
I'm,
like
from
my
perspective,
the
better
we
can
make
it
the
more
flexible,
but
if
we
have
trouble
getting
that
far
I
still
think
there
would
be
value
in
this
first
step
of
okay,
you
can
take.
You
can
take
advantage
of
any
P
I.
You
can
write
your
code
so
that
it
uses
that
and
you
it's
it's
no
worse
than
for
other
modules
and
that
the
node
API
version.
This
is
the
only
thing
you
have
and
Italy's.
B
That's
that's
what
I'm
thinking
like
if
we
go.
If
we
go
with
this
with
this
tie
to
the
node
module
version
first,
then
then,
then
you
know
people
will
will
adopt
that
people
who
use
libuv
and
openness
to
self
right,
but
then,
but
then
you
know,
they'd
have
to
make
another
code
change.
If
if
we
then
go
ahead
and
give
them
more
granularity
later,
you
know
yeah.
A
If
we
give
more
granular
layers
later
anybody
yeah
so
I,
you
know
I'm
I
think
we
should
think
about
it.
Come
up
with
the
best
thing
we
think
we
could
get
landed
now
yeah
and
if
we
can
do
you
know
if
we
can
get
the
whole
shebang
in
there,
that's
great,
but
I,
don't
think
we
should
say
we'll
do
nothing.
If
that
we
can't
because
I
think
what
we
just
talked
about
would
still
be
at
least
a
stop
right.
B
B
I
will
drop
tracking
of
the
of
the
engine
version,
because
the
whole
point
of
any
API
is
that
you
don't
have
to
track
the
engine
version
so
so
we'd
have
we'd,
have
fewer
we'd
have
fewer
defined
ABI
component.
We
basically
just
have
open
SSL
libuv
and
the
Lib
I
see
you
I
think,
but
but
since
that's
just
a
proof
of
concept,
we'll
also
have
to
agree
on
which,
which
parts
of
the
ABI
we
wish
to
track
independently.
You
know
right.
A
B
Well,
only
only
insofar
as
having
a
place
where
you
declared
these
things,
where
you
formalize
the
fact
that
you
know
just
as
just
as
when
you
update
the
the
v8
minor
version,
you
you
update,
node
module
version.
So
wouldn't
you
update
the
OpenSSL
setting
no
to
be
ABI
incompatible.
You
update,
like
the
the
node
module,
API
OpenSSL
version,
and
then
you
increment,
that
and
and
so
forth.
So
so,
basically.
A
A
E
B
Well
it
has
to
be
a
different
discussion
right
because
in
the
in
the
classical
node
modules,
discussion,
v8
and
the
version
of
v8
plays
a
role
true,
so
so
so
so
so
the
vendor
prefix
is
not
usable
there
right
there
and
and
all
these
arguing
that
that
that
that
the
the
v8
was
OpenSSL
plus
libuv
plus
everything
else
constitutes
a
vendor
prefix
right,
but
but
well
you
it
just
so
you
go
ahead.
I.
A
Used
to
say
you
can
make
the
same
argument
in
terms
of
it's.
It's
correct
that
you've
got
this
one
thing,
but
just
because
so
in
the
case
you
know
an
API
has
removed
v8
from
the
max
yeah,
but
in
in
the
case
of
you
know,
if
you
still
have
two
or
three
different
things
that
you
want
to
declare
separately
so
that
if
they
don't
change
in
an
ABI
braking
way,
you
don't
have
to
rebuild
yeah.
B
A
B
B
To
build
to
build
like
Ali,
said:
130
different
builds
right,
however,
those
that
use
an
API
they
wish,
presumably
to
have
as
few
build
as
possible
right
and
so
so
so
to
give
them
builds
for
exactly
those
environments
that
a
change
and
none
other,
that's
I
would
say
a
good
thing
right,
like
don't
unnecessarily
build
your
module
like
ever.
You
know,
yeah.
A
But
III
think
it's
I,
don't
I,
don't
know
if
it's
that
you
know
people
choose
to
use
v8
and
say:
okay
I'm
willing
to
build
138
things.
It's
like
oh
I'm,
using
open,
SSL
I'm
using
Libby
V.
That's
my
only
choice,
I,
don't
know
I,
guess
I'm
like
I,
don't
think
necessarily
that
they
care
less
about
it.
Yeah.
B
A
B
Yeah
but
but
you
know,
if
you're,
if
you're
starting
out
with
a
node
application,
then-
and
you
still
want
to
use
native
threads
because
a
sync
worker
doesn't
work
for
you,
then
there
you
might
be
using
the
bubi
and
I
mean
there
are
others
that
use
that
use
exactly
two
things
v8
and
leave.
You
be
right
sure.
A
B
B
A
A
B
B
A
E
B
B
B
A
A
B
A
B
A
B
A
A
A
B
Yes,
exactly
because
the
version
numbers
are
not
expressive
enough,
they
they're
just
saying
the
ABI
of
nodejs
has
changed.
That's
it
right,
yeah,
so
and
and
and
of
course,
nodejs
assumes
the
worst
which,
which
is
a
which
is
a
good
conservative,
good
conservative
assumption
yep.
But
but
it's
it's
it's
overkill
for
napi
to
be
so
conservative,
because
the
whole
point
of
an
API
is
that
you
need
not
be
so.
Conservative
yeah.
B
A
A
A
I
think
it
might
be
worse,
I'm
gonna
say
actually
no,
let's
do
that.
Okay,
so
yeah,
okay,
I
think
there's
also
one
and
no
doubt
on
API
as
well.
Correct.
Okay,
sorry-
and
it
was
good
that
we
did
that,
but
wasn't
enough
SQL
late
part.
We
already
talked
about
burned
down
issues
raised
by
module
owners.
I
don't
have
anything
particular
to
highlight
this
week.
A
D
Hey
guys,
yeah
I
actually
have
missed
a
whole
bunch
catching
up.
So
what
is
the
kind
of
overall
status
of
adoption
that
we
have
seen
so
far
like?
Are
people
generally
kind
of
finding
a
lot
of
issues
that
are
hurting
adoption
of
people
kind
of
not
into
it
like?
What
do
we
have
a
general
sense
of
what's
going
on.
B
My
overall
picture
and
correct
me
if
I'm
wrong
is
that
is
that
we're
sort
of
we've
got
we've
got
like
a
healthy
pace
going
on.
It's
not
easy.
It
could
be
faster,
but
but
just
just
judging
from
from
the
fact
that
we're
having
so
much
so
much
conversation
of
know
that
on
know,
then
on
AAPI,
and
so
many
proposals
and
people
coming
in
and
submitting
patches,
I
think
I
think
we're
we're
not
exactly
sidelined.
Like.
A
A
A
C
A
The
other
thing
we'll
see
is
that
FS
events
is
one
of
the
ones
I
think
it
supposedly
gets
like
a
million
downloads
a
month
or
something
and
they're
pushing
you
know
they.
They
basically
said
they
really
wanted
it
back
ported
to
eight
okay,
because
it
would
allow
people
to
adopt
it,
and
so
that
actually
happens
we
might
even
see
a
bigger
bump.
Although
we
don't
because
because
you
pre
build
your
binaries,
you
may
not.
A
D
A
Basically
said,
that's
dependency
that
we've
included
in
their
package
that
Jason
right
on
so
yeah
I
think
I.
Think,
like
Gabriel
said
we
could
do
more
evangelization.
We
could
push
it
further,
but
it
still
seems
like
we
have
a
sort
of
a
slow
and
steady
growth
and
certainly
that
I
did
notice,
like
Nick
did
that
you
know
around
the
November
and
when
was
when
was
some
note
cards
not.
A
C
A
D
A
D
B
We
really
need
to
go
over
them
very
closely
because
you
know
where
we're
coming
we're
committing
ourselves
to
a
C+
Enterprise
for
this
at
that
point
yeah,
so
so
so
yeah
that
that's
sort
of
the
wider
discussion,
but
but
honestly
the
the
whole
threading
thing
is
is
part
of
like
a
bigger
trend,
I
think
of
just
getting
multi-threaded
Java
scripts
going
I
mean
what-what.
Would
worker
threads
now
being
available
without
a
flag
and
everything
it's
the
we
are
just.
We
are
just
the
native
side
of
the
whole
multi-threaded
JavaScript
discussion,
I.
Think.
A
E
A
B
A
E
A
B
Examples
I
think
that's
that's
another
thing
like
they
just
I
had
of
an
example
going
with
with
how
to
how
to
asynchronously
fill
in
an
array
buffer
and
pass
it
back
to
JavaScript.
B
B
It's
a
fundamental
contact,
yeah
yeah,
so
no
matter
what
framework
you
use,
you
will
not
be
able
to
get
away
from
from
having
to
keep
persistent
references
and-
and
it's
it's
it's
a
fundamental
way
of
adjusting
yourself
to
thinking
about
about
how
to
bridge
a
garbage
collected
environment
to
one
that
that
has
explicit
allocate
and
free
and
so
so
yeah.
So
those
questions
will
keep
on
coming
and
and
since
we
are
now
those
who
provide
the
environment
as
opposed
to
an
end,
we
will
be
the
one
who's
leaving
those
questions.
A
The
other,
the
other
thing
so
we're
just
catching
up
if
we
had
unlimited
resources,
the
other
thing
we'd
probably
push
on
to
is
there
was
some
discussion
of
building
a
layer
that
would
implement
man
on
top
of
an
API
yeah,
so
I
mean
I,
know
Gabriel,
you
know,
did
some
investigation
on
that.
So
it's
not
impossible.
It's
just
a
fairly
big
effort.
Yeah.
B
B
Basically,
I
it's
better,
so
so
what
one
of
the
important
benefits
of
having
such
a
shim
would
be
that
you
could
you
could
do
what
looks
like
mixing
v8
code
with
an
API
code
and
the
reason
why
it
would
be
possible
possible
is
because
the
v8
code,
this
being
a
shame,
would
would
just
really
be
know
that
on
API
calls
in
the
back
right.
And
so,
unlike
the
genuine
v8
implementation,
our
shim
would
would
then
have.
B
We
would
then
have
methods
for
for
constructing
what
looks
like
a
v8
object
using
an
API
or
know
that
on
API
objects
and
and
what
looks
like
know
that
on
the
API
objects,
using
v8
objects
and,
and
so
so
for
add-on
maintainer,
the
huge
benefit
would
be
that
that
they
could.
They
could
gradually
port
over
their
code.
So
as
soon
as
they
depend
on
the
shim
there,
their
entire
code
base,
with
all
of
a
sudden
shift
from
calling
v8
to
calling
to
calling
an
API
ultimately.
A
B
Would
be,
it
would
be
mostly
mostly
inline
like
like
most
of
it
would
be
inline,
and
so
so
it
wouldn't
introduce
perf
issues
except
except
the
function.
Signature
again,
we'd
have
to
do
yet
another
layer
of
thunking
because
the
the
function
signature
simply
don't
line
up,
and
so
so
you'd
be
going
from
a
function.
Signature
that
looks
like
v8
and
really
is
v8
inside
the
N
API
implementation
to
C,
n
API
function,
signature
to
I
know
that
on
API
function,
signature
and
then
back
to
a
function.
B
Signature
that
looks
like
v8,
so
so,
okay,
you
can
cut
out
one
step
there,
which
is,
which
is
to
skip
to
know
that
on
API
function,
signature
and
and
just
and
and
supply
translates
the
the
sort
of
full
v8
signatures
directly
to
to
see
an
API.
Because,
because
you
can,
you
can
interchange,
know
that
on
API
and
CN
API
very
easily
by
just
constructing
an
environment
constructing
on
a
p-value
from
from
from
the
C
stress,
but.
A
Kind
of
like
a
chakra
course
shim,
so
I
don't
know
if
a
Rona
shinja
test,
you
haven't
think
a
bit
about
more
about
that
and
maybe
next
meeting
you
know
we
can
discuss
it
some
more
because
Isis
you
know,
some
of
the
same
challenges
might
have
been
with
what
you
saw
with
chakra
core,
or
you
know,
the
chakra
shim
and
how
effective
it
was
for
native
modules.
So
yeah.