►
Description
B
B
A
B
B
G
A
I
J
Linux,
one
to
the
tests
in
the
sea
I
opened
a
related
issue.
Some
involvement
on
AIX
issues
working
with
en
on
the
a
bi-stable
API
and
getting
ready
for
the
API
working
group
meeting
this
week
and
I
involvement,
some
of
the
other
working
group
meetings
and
then
just
miscellaneous
reviews
and
lands
and
so
forth.
Next
is
Brian.
K
M
B
N
Got
the
release
candidate
for
v
four
point:
five
point:
zero
out
and
got
a
whole
bunch
of
people
to
help
with
back
porting
working
on
some
enhancements
to
canary
in
the
gold
mine
that
should
see
it
moving
a
lot
faster
and
also
being
parsed
quicker,
which
should
help
with
some
of
the
Jenkins
problems.
We've
been
having
and
then
also
helped
organize
a
meeting
with
the
DEA
team
to
improve
the
workflow
for
floating
patches
on
LTS
and
next
is
Trevor.
O
O
D
Yes,
I
I'm
trying
to
handle
a
flaky
tests
that
break
on
freebsd.
Although
a
turn
looks
like
it
might
be,
an
infra
related
problem,
that's
easily
solved,
so
that's
exciting
various
es
length
updates
improvements,
yay
and
banging
my
head
against
the
long-standing,
be
a
tick
processor
test.
That's
non-deterministic
that
fails
occasionally
so
and
we
didn't
miss
anybody.
It
doesn't
look
like
so
I
guess
we're
on
to
the
agenda
now.
B
Yeah,
so
we
got
a
few
items
today.
I
think
we
can
roll
through
miss
them
fairly
quickly,
pretty
quickly
on
the
previous
meetings
agenda.
If
anything
here
stands
out
but
worthy
of
comment,
please
do
so
otherwise
we'll
move
on.
So
we
talked
about
well,
I
was
in
here
so
it
was
talked
about.
Dropping
visual
studio,
2013
support
in
v7
and
ugly
little
you
dinner
again.
Real
path.
Changes
is
on
the
agenda
again
and
then
the
unambiguous
JavaScript
grandma
thing
for
ES
modules
was
accepted
and
merged.
B
M
Pick
this
one:
don't
the
decision
to
drop
support
for
us,
it
was
entertaining
v7
doesn't
seem
controversial.
We
need
to
prepare
some
repeats
machines
to
do
visit
with
2015
and
also
I
tests
to
the
matrix
to
test
modules.
Built
will
2015
with
not
build
2015,
as
always
currently
taking
care
of
that
the
work
is
it
under
control.
M
The
remaining
questions
were
were
whether
we
would
drop
support
for
v6
and
there
is
a
potential
breaking
change.
So
it
turns
out
that
the
issue
affecting
no
sass,
which
is
a
known,
ABI
and
compatibility
across
the
compiler
version,
only
affects
windows,
XP
topics,
we
don't
support,
XP,
that's
encouraging
and
the
only
one
that
we
were
well
so
I
had
a
chat
with
the
visual
c++
team
and
has
been
suggested.
M
M
Compiler
tries
to
maintain
compatibility,
although
there
are
some
current
cases
and
at
this
point
the
best
way
to
there's
no
way
to
be
sure
of
how
people
are
using
it
out
there
or
the
best
way
to
gain
some
confidence
is
do
some
testing,
so
I
was
wondering
if
I
could
get
some
input
from
miles
if
Canadian
the
gold
mine
would
be
would
provide
good
coverage
for
elective
modules
and
from
the
rest
of
the
group.
If
passing,
that
would
be
good
enough
I,
but
in
a
criteria
without
support
in
b6.
N
So
we'd
have
to
make
some
changes,
at
least
in
CI,
to
see
how
that
would
work
for
canary
in
the
gold
mine,
because
it's
installing
all
the
modules
each
run.
So
what
we
would
really
want
to
do
is
we'd
want
to
run
the
install
with
the
original
version
and
then
run
the
test
suite
with
the
new
version
and
see
you
know.
If
there's
any
breakages
I
could
look
into
what
it
would
take
to
get
something
like
that
working,
I'm,
actually
kind
of
curious.
B
It's
actually
exactly
the
same
stuff
that
I
wanted
to
do
on
that
front,
been
I,
think
Ben's
going
to
do
some
inspection
of
binaries,
but
the
actually
compatibility
testing
for
add-ons
yeah.
That's
exactly
the
same
thing.
So
if
that
worked
for
clearing
the
gold
mind,
then
that
would
cover
our
case
for
upgrading
ABI.
Well,
the
cover
the
particular
case
that
I
was
interested
in
loved
one
have
been
was
interested
so.
N
M
N
B
M
B
A
M
O
As
far
as
the
deep
circling
issue
you
had
us
solve
this
in
my
PR,
but
they're
still
a
windows
as
you
wear.
It
was
similarly
sim
looking
across
drives
or
network
drives.
That
causes
its
resolving
correctly.
That's
it
another
PR.
Basically,
just
re
implement
the
old
JavaScript
API
to
to
solve
that
I
feel
like
you
can
be
simplified.
B
M
Stolen
well,
I
think
the
concern
is
that
is
driving
back
most
of
the
jes
implementation.
Just
kind
of
this
windows
corner
cases
and
people
would
prefer
to
see
and
what
education
in
sneakers
black,
but
Butler
shall
be
working
on.
This
I
didn't
have
a
chance
to
review
in
detail,
placing
two
to
think
that
this
is
the
best
way
forward.
A
B
O
O
O
This
this
patch
honestly,
is
in
like
over
a
year
coming,
and
it's
not
a
little
speed
up.
It's
like
over
an
order
of
magnitude,
and
what's
your
name,
she
hears
broccoli.
Do
you
actually,
the
original
reason
why
that
you
can
came
up
is
because
she
was
having
to
set
a
huge
number
of
files
and
it's
causing
a
big
performance
gap
in
her
library,
and
so
this
freedom
was
put
in
to
help
with
that
case.
O
O
B
N
B
K
Yeah,
okay,
so
so
this
for
people
that
are,
you
know,
this
is
just
about
preventing
lowercased
header
names
from
clashing
with
properties
that
are
inherited
from
the
object
prototype
like
underscore
underscore
proto
underscore
underscore
in
particular.
Might
the
my
original
solution
or
the
current
solution
really
is
just
use
the
null
prototype
and
just
have
that
be
the
end
of
it?
This
is
the
solution
we're
already
using
for
query
string
parsing
since
feed
the
original
v6
release.
K
One
alternate
solution
to
this
was:
it
was
suggested
to
cherry
pick
at
least
one
of
the
properties
from
the
object
prototype,
and
to
stick
that,
like
specifically
like
to
string
for
example,
and
to
stick
that
on
the
object.
My
argument
against
that
was
that
I
personally
think
it
should
just
be
like
an
all-or-nothing
thing
to
prevent.
You
know
like
possible
issues
with
like
duck
typing
and
people.
Think.
Oh,
look,
there's
a
you
know
two
strings
working,
so
you
know
I'll,
try
has
owned
property
that
should
be
working
too
and
I'll.
K
If
we
wanted
to
do
a
hard
deprecation
like
sooner
or
if
we
wanted
to
go
through
like
a
doc
deprecation
and
follow
through
that
kind
of
trajectory
or
whatever,
and
there
were,
there
was
also
some
discussion
about.
You
know
if
we
eventually
want
to
transition
to
using
an
es6
map,
instead
of
just
a
plain
object
as
the
backing
store.
K
Obviously
that's
going
to
be
a
lot,
probably
a
significant
change,
because
the
API
just
completely
different
as
far
as
accessing
and
setting
properties
and
all
that
kind
of
stuff,
and
also
the
rod
had
showed
he
has
a
PR
that
shows
the
ESX
maps
seemed
to
be.
If
we're.
If
the
benchmark
is
correct,
that
they're
saying
they
seem
to
be
faster
for
them.
K
Just
plain
objects
for
storing
of
values,
so
recent
versions
of
v8,
and
so
basically
we
need
to
I
guess
find
out
what
we
think
is
the
best
solution,
at
least
for
the
time
being,
you
know
whatever
the
plants
may
be
for
when
v6
goes
in
the
LTS
or
for
v7
or
whatever
so
yeah.
That's
that's,
basically,
the
gist
of
it.
A
A
couple
things
I
used
I
think
we
should
be
extra
careful
of
maps
they
when
I
did
a
lot
of
testing
long-timers.
That
was
a
little
bit
ago,
but
they
were
definitely
not
faster
unless
you
were
deleting
a
lot
of
properties,
but
many
applications
do
not
specifically
use
the
delete
operator
on
such
objects.
So
you
do
not
incur
those
sort
of
performance
penalties.
A
O
Looks
like
question
is:
is
double
underscore
proto?
Does
that
even
a
valid
header,
oh.
K
I
think
it
is
because
I
think
well,
yes,
well
I.
Take
that
back.
I
think
that
there
was
an
issue
sometime
back
I'm.
I
can
remember
if
it
was
in
time
for
v
forever.
It
was
I
ojs.
At
the
time
there
was
an
issue
where
somebody
was
sending
a
header
name
with
the
name
constructor
or
something
like
that.
I
think
I
had
to
actually
tweak
the
check
when
the
header
names
are
set,
so
that
makes
sure
that
it
doesn't
try
to
override
a
function,
I
think
or
something
because.
A
K
A
B
A
Something
we
could
do
since
I
mean
one
thing
we
could
do
is
we
could
make
a?
U
tilde
format
check
if
something
has
the
you
doesn't
have
a
two-string
if
it
doesn't
already,
if
it
doesn't
already
check
for
that,
because
then
we
can
just
use
object
to
string
that
might
help
mitigate
some.
The
problem
of
people
like
trying
to
log
out
headers
when
they're
trying
to
like
look
at
stuff
for
stuff,
like
we.
D
A
K
B
But
my
I
support
that
too,
but
just
before
we
we,
you
know,
don't
like
fully
commit
to
that.
I
recommend
that
you
have
a
look
at
the
thread.
The
in
that
issue.
There
is
some
arguments
against
that,
and
particularly
from
Jordan
hub
and
whose
tc39
so
go.
That
thread
have
a
look
way
in
if
we
need
to
bring
it
back
for
a
vote
again
Brian.
Let's
do
it
next
week,
but
for
now
back
to
github,
please
go
and
weigh
in.
If
you
find
this
discussion
interesting.
B
O
L
O
You
past
the
chrome,
dash
extension
scheme
has
its
own
way
of
parsing
that
so
I
honestly
my
opinion,
adding
support
for
specific
schemes
is,
is
a
non-issue.
We
should
be
able
to
add
those
without
about
worrying
about
it.
There
is
slight
difference
in
functionality.
I
didn't
realize
till
today,
which
is
URL.
L
O
O
B
Think
the
katay
to
inspire
this
is
where
it
would
live,
because
the
proposal
is
not
to
remove
the
or
replace
the
old,
the
current
implementation,
but
is
to
add
a
new
thing
and
so
there's
multiple
options,
even
though
the
doc
says
hanging
it
off
the
URL
module
as
all
caps.
Url
there's
also
discussion
there
about
making
it
a
global.
That
I
think
is
the
most
important
thing
that
we
need
to
it's
the
most
important
hurdle
to
jump.
First,
one.
A
Yeah
good
point:
didn't
we
talk
about
something
really
similar
to
this?
In
terms
of
like
the
HCP?
Pars
are
the
DNS
logic?
It's
definitely
got
at
the
time.
The
discussion
was
well
since
we
didn't
want
to
have
two
implementations
forever
was
to
use
a
flag
to
ease
in
and
then
also
use
a
flag
to
ease
out
the
people
that
still
needed
the
old
implementation
later
down
the
road
so
be
around.
A
B
O
B
Last
thing
on
the
agenda
is
miles.
We
put
this
one
on
last
and
let
just
recently
within
the
last
few
hours
issue,
number
seven
688,
which
is
poor
request
for
version
4.2,
5.0,
so
release
candidates.
Are
we
rolling
out
but
miles
wanted
to
raise
awareness
here?
Do
you
want
to
tell
us
miles
what
you'd
like
from
us
yeah.
N
N
So
I
think
the
intention
is
that
we
want
to
ship
with
that.
It
would
be
really
great,
especially
for
like
security
and
back
porting
purposes,
to
be
on
a
more
modern,
Libya
V,
but
with
the
number
of
different
edge
cases
and
problems
that
we
felt
with
the
upgrade
in
six.
Even
though
most
of
them
were
based
around.
You
know
like
the
real
path
implementation
which
we
are
not
back,
porting
we're
just
back
porting
the
upgrade
to
live
EV.
N
N
B
My
suggestion
in
the
the
meeting
six
hours
ago
was
too
we
have
a
new
release
candidate
cut
soon
and
then
to
have
some
heavy
formulation
around
in
their
path
of
that
on
the
various
social
channels
that
we're
all
connected
to.
So
maybe
just
have
everyone
here
tuning
into
that
release
and
growing
up
and
promoting
people
testing
it.
N
B
B
N
J
N
The
the
release
candidates
are
available
on
nodejs,
calm,
/
download,
/
RC,
so
it's
available
from
the
same
download
site.
As
you
know,
our
other
releases,
you
can
actually
get
it
working
in
nvm
by
just
changing
an
environment
variable,
and
I
can
share
the
information
on
how
to
do
that.
If
people
are
interested
in
getting
a
little
bit
happier
with
it.
P
B
Let's
reach
out
to
handle
office
maybe
miles
you
want
to
take
that
if
you
have
a
relationship
with
Hunter
I
do.
B
N
B
C
That
ok,
so
if
it
sorry
there's
one
thing
just
like
I
mentioned
and
stand
up,
Oh
four
modules:
we're
going
to
have
to
rethink
some
things.
Just
because
we
got
back
from
vm
implementation.
We
can't
get
poisoning
so
named.
Imports
on
common
J's
modules,
just
can't
be
supported
by
the
VMS,
for
a
variety
of
reasons,
so
we're
going
to
need
to
figure
out
a
solution
to
that,
but
it
just
means
that
we
will
have
major
breakage
whenever
we
do
implement
es
modules,
which
we
weren't
expect
such
a
big
problem.
O
C
No
so
like,
if,
if
I'm
using
Babel
I
can
do
import,
read
file
from
FS
and
that
works,
what
that's
doing
is
something
either
called
plucking
or
hoisting,
depending
upon
specific
things.
Basically,
it's
doing
property
access
and
there's
some
code
generation
problems
and
some
optimization
problems.
The
VMS
face
when
doing
that,
so
it
would
basically
mean
any
usage
of
a
imported
variable
would
need
to
be
guarded
during
codegen,
which
would
cause
d
ops,
and
so
that's
just
not
something
can
be
done.
C
Like
I
can
import,
you
can
import
anything.
That's
an
identifier.
That's
been
marked
for
export,
so
the
way
that
the
model
spec
works
in
reality
is
it's
giving
direct
access
to
the
modules
scope
and
it's
giving
access
to
anything.
That's
been
flagged
for
export.
So
export
is
more
like
a
boolean
flag,
saying
this
variable
is
available
for
other
people,
then
saying
I
want
to
store
it
on
an
object.
C
Well,
there's
that
and
then
this
has
an
interesting
thing,
because
we're
going
to
use
plucking
to
get
transitivity
to
work.
So
if
you
have
an
ESX
module,
that's
re-exported
by
a
common
J's
module
and
then
loaded
into
an
es6,
or
vice
versa,
CJ
s
tes
to
CJ
s.
We
want
it
to
maintain
the
exact
same
shape,
even
though
it's
gone
through
the
other
module
system
as
a
result
of
things
uncovered
during
implementation
problems.
C
While
they
were
looking
at
it,
we
actually
have
to
figure
out
some
way
to
fix
that
problem,
which
could
be
done
if
we
could
do
hoisting,
but
we
can't
do
hoisting,
so
I've
got
to
figure
out
some
way
to
do
it
and
it'll
probably
be
something
like
put
a
frozen
object
into
a
week
map
somewhere.
If
you
need
something
to
be
in
a
different
mode,
which
sounds
really
weird,
but
I
haven't
fully
figured
it
out
yet.
B
Okay,
I
don't
have
much
time
left,
but
Q
&
A
from
the
people
listening
in
on
the
live
stream.
If
you
want
to
leave
comments
on
YouTube
and
or
IRC
and
all
the
the
issue
for
this
meeting,
please
do
so
now
we
might
pause
for
a
minute
or
so
to
let
that
come
in
unless
there's
anything
pending
Jeremiah
or
else,
nothing.
B
Okay,
let's
wrap
up
now
we
have
a
fantastic
new
section
at
the
end
here,
I
think
Josh
you're
responsible
for
this
upcoming
meetings,
so
the
next
CTC
making
is
on
the
twentieth
of
this
month.
The
tsc
meeting
is
tomorrow:
you
can
tune
into
that
they'll
be
an
issue
on
the
TSE
repo
a
bit
later.
Today,
diagnostics
working
group
meeting
has
the
next
meeting
next
month
or
third
post-mortem
has
one
on
the
18th.
B
So
that's
next
week,
api
working
group
tomorrow
LTS
working
group
in
two
weeks
and
the
build
working
group
next
week
on
the
19th,
so
that
each
of
those
issues
will
show
up
on
the
for
those
means
will
show
up
on
those
various
posit
or
ease,
and
if
you
have
suggestions
for
making
that
more
accessible,
please
do
so
I.
Think
on
the
tier
cereal
is
a
such
thing
in
there
about
meeting
calendar
so
yeah.
That's
it
thanks
for
everyone
for
joining,
and
we
might
have
a
good
private
section
after
this,
but
yeah
but
I
for
now.