►
From YouTube: SES Meeting: Realms Stage 3!
Description
TC-39 advanced Realms to Stage 3. They liked everything but the color.
A
Hello
welcome
to
the
s
meeting.
We
had
a
festive
tc39
meeting
last
week,
where
realms
advanced
to
stage
three,
the
only
remaining,
as
that
was
how
does
it
say
they
liked
everything
but
the
color.
So
we
have
placed
on
our
agenda
the
conversation
about
what
should
we
even
call
realms
and
what's
to
play
some
play
some
name
name
guessing
game
theory,
and
also,
I
think
that
that's
that's
it
does
anybody
have
anything
that
they
wish
to
propose?
We
add
to
our
agenda.
B
I
have
a
just
a
sort
of
a
accounting
and
administrative
item
that
is
realms
related.
That
would
take
me
less
than
60
seconds.
B
Cool
great,
let
me
pick
this
up,
so
I'm
not
looking
down
like
a
weirdo.
I'm
really
pleased
to
say
that
through
a
concerted
preemptive
effort,
there
is
a
pull
request
open
with
a
substantial
set
of
tests
for
the
current
specification
of
callable
boundary
realms
for
for
both
the
evaluate
and
import
value
behaviors.
B
So
it's
I
actually
used
it
to
implement
two
full
polyfills,
one
of
which
is
under
review
in
the
realms
polyfill
repo
that
I
have
made
such
that
it
works
in
javascript,
core
spider
monkey
and
v8
so
far,
and
I
used
the
tests
to
implement
that.
So
as
soon
as
our
plus
from
one
more
person,
I'm
hoping
that
mike
pinissi
will
take
a
look
and
shepherd
that,
through
the
tests,
folks
can
start
working.
A
As
well
done
excellent
yeah,
that
just
leaves
so
that
just
leaves
50
minutes
to
talk
about
names.
D
Well,
so
that
that's
fun
yesterday
I
received
a
visit
from
john
dick
dalton,
who
works
with,
and
my
team
like
rick
and
I
same
team
of,
and
we
were
talking
about
the
names
and
he
explained
like
with
many
reasons,
satisfactory.
D
I
still
have
some
issues
on
using
it
verse
because
some
somehow
I
want
first
to
be
something
that
is
multi-threaded
or
like
the
idea
of
having
multiverse
oh
yeah
and
also
have
shades
of
firefly,
but
to
be
fairly
honest.
There
is
no
reason
for
me
to
really
rename
realms
because,
like
we've
been
working
on
this
for
seven
years,
it's
not
never
been
an
actual
problem
having
realms
as
a
name.
D
I
think
there
might
be
implementation
concerns
and
that's
why
I'm
actually
still
not
shutting
that
down,
and
I
respect
that,
like
that,
we
have
some
of
the
implementation
challenges
or
specs
integration,
but
like
for
end
users
realms
is
just
fine.
E
Jordan
had
one
comment
that
I
think
is
fair
is
that
if
you
search
for
realms
right
now,
you
end
up
finding
information
about
array.
Not
mat
radar
is
array
and
and
and
not
being
able
to
check
the
instance
of
cross
realm
with
iframes,
which
obviously
wouldn't
be
a
problem
with
this,
because
you
don't
get
an
object
so.
E
D
There's
also
the
the
thing
that
you
have
you
have
realms,
but
also
like,
if
you
use
bubble
and
on
the
internet,
especially
if
you
have
like,
if
you
name
it
bubble
and
you've,
let's
say
you
need
to
handle
events
inside
the
bubbles,
so
you
have
bubble
events.
D
A
Yes,
I
I
think
that
actually
in
the
in
this
room,
I
don't
think
that
anybody
wants
to
rename
it.
So
this
is
not
a
conversation
about
this,
isn't
a
conversation
about
better
than
better
names
than
realm.
This
is
a
conversation
about
worse
names
than
realm
that
are
more
likely
to
succeed.
I
see
a
handful.
F
Of
jumping
as
a
member
of
the
library,
so
you
know
I
think,
when
we
when
we
adapted
callable
boundaries,
it
was
a
really
significant
change
from
what
we've
been
calling
realms.
That
was
a
thing
that
was
raised
in
committee
and
it's
and
it's
true
leo
characterized
the
concerns
as
implement
implementer
concerns
or
implementation
concerns.
They're
not
concerns
from
an
implementation
perspective,
naming
doesn't
have
non-trivial
implementation
implications,
they're
design
concerns
that
web
browser
authors
have
raised
and
I
think
they're
legitimate.
F
I
think
it's
legitimate
to
say
that
this
proposal,
by
having
this
kind
of
boundary,
differs
from
what
a
pretty
small
group
of
people
who
maybe
we
don't
have
to
care
about
too
much
we're
thinking
about
realms
for
for
a
long
time,
they
differ
from
what
the
same
words
and
iframe
semantics
are,
and
a
certain
small
group
of
specialists
might
use
the
word
realm
to
to
refer
to
that,
and
I
don't
really
agree
with
the
waiting.
F
But
it's
a
you
know,
but
this
is
a
legitimate
design
concern
and
I
think
you
know
if
we
want
to
make.
F
B
We
didn't
make
jokes,
we
didn't
you
know,
blame
play
in
the
blame
game,
but
at
the
end
of
the
day,
tasks
and
jobs
was
like
a
big
sticking
point
years
ago
and
that
didn't
get
changed
and
everybody
is
still
alive
and
still
productive
and
out.
The
the
262
still
talks
about
jobs
and
web
html
still
talks
about
tasks
and
they
still
manage
to
be
productive
and
effective,
and
I
don't
think
that
the
callable
boundary
change
is
really
that
drastic
of
difference.
I
think
it's
a
super
interesting
and
novel
concept.
B
At
the
end
of
the
day,
that
is
still
a
evaluate.
Just
takes
a
string
and
evaluates
it
that
code
in
in
a
fresh
global
context,
just
as
a
realm
always
would
the
only
difference
now
is
we
don't
have
a
dot
global
property
and
we
can't
pass
in
endowments
yet
yet.
D
So
just
just
want
to
highlight
a
few
bullets.
First,
one
of
them.
D
We
it's
easy
for
me
to
assume
that
I
don't
have
any
specific
problem
with
the
names
or
that
I
it's
really
like.
It's
not
part
of
my
goals,
finding
a
new
name.
I
don't
like
participating
on
that
conversation
in
order
to
be
like
participating
in
good
faith
like
to
help
people
and
it's
more
like
a
sport.
But
it's
not
my
go.
D
My
personal
goal,
finding
a
new
name
for
round,
because
I
am
okay
if
the
name
remains
realm
and
in
in
the
same
sense,
that's
the
other
bullet,
I'm
okay
with
whatever
name
reality.
I
want
a
feature,
but
I
don't
care
too
much
about
the
name
as
like
in
order
to
like
it's
nothing
like
super
precious
for
me,
of
course,
we
can
take
advantage
if
the
name
is
easy
to
be
searchable
on
the
web
and
not
cause
confusions,
I
don't
think
no
name
will
actually
be
intuitive,
saying
like
well.
D
This
is
what
you
can
do,
just
figure
out
like
by
a
single
name
or
double
double
name:
constructor,
you're,
never
going
to
figure
out
that,
just
by
the
name,
you
need
to
take
a
look
and
see
what
it
does
and
how
it
does.
D
So
I
don't
mind
if
we
get,
if
you
call
it
realms
or
if
we
call
it
anything
else,
of
course
we
we
need
that
to
like
name
it
smoosh
on
the
way,
but
we
we
need
to
find
a
name
that
just
works.
D
I
am
doing
this
in
support
and,
of
course,
like
I
think
this
conversation
about
the
names
should
always
take.
Have
it's
always
free
to
have
a
tone
that,
like
it's,
not
super
serious
like
if
we
talk
about
a
name
that
we
find
like,
let's
say,
bubble
a
lot
of
people
like
the
name,
and
I
think
one
of
the
reasons
people
might
like
the
name.
This
is
just
a
guess:
it's
because
it's
cute
in
in
some
sense,
like
yeah
bubble,
calling
it
bobo
is
cute.
D
D
What
do
I
say
a
funny
tone
it
doesn't
make
like
it
should
not
also
mean
like
we
are
making
fun
of
the
problem,
but
also
like
we
are
being
in
a
good
in
a
good
sport
in
a
good
spirit
to
to
move
this
forward,
and
we
are,
I
I'm
still
celebrating
that
we
got
the
stage
three.
I'm
still
like
in
a
very
enthusiastic
mood,
because
seeing
the
next
steps
or
stage
four
are
the
best
part.
A
When
you
proposed
the
name
acrylic
glove
box
until
you
explained
what
you
meant
by
it,
I
thought
it
was
a
joke
and
I
apologize
for
receiving
it
as
if
it
were
it,
but
it
yeah.
And
I
and
I
took
the
picture
that
you
sent
and
showed
it
to
my
wife
who
worked
in
science
for
a
long
time
and
asked
her
what
it
was
called
and
she's
like.
B
Lab
box
is
another
word
I
found
for
it.
The
unfortunate
thing
is
that
that
physical
item
that
exists
in
our
world
is
the
is
absolutely
the
closest
analog
to
the
thing
that
we've
created
here
well,
like
visually
mentally
in
every
way
possible.
What
the
hell
is.
It
called
give
give
us
a
a
nice
name
for
that
thing,
that
isn't
glove
box,
because
that's
definitely
the
place
in
my
car,
where
I
put
the
registration
and
and
the
vehicle
manual
yeah
yeah.
D
And
let
me
let
me
bring
my
so
joseph
finish,
something
a
note
about
the
box.
I
bring
my
english
as
a
second
language
card
here
where
jgd
actually
told
me
like,
we
should
not
enab
we
should.
We
should
avoid
our
best
to
name
anything
as
box,
because
of
all
the
connotations
of
these
words,
so
yeah
I've
been
informed
and
I
definitely
gonna
I'm
not
gonna
advocate
for
anything
that
looks
like
glovebox
or
any
derivation
of
that.
It's
unfortunate.
We
have
a
thing
that
yes,
physically,
is
analogous,
but.
B
Doesn't
a
bubble
is
not
what
this
thing
creates.
You
can't
you
can't.
If
you
try
to
enter
a
bubble,
it
will
pop
if
you
try
to
enter
from
the
outside.
If
you
try
to,
if
something
inside
a
bubble
tries
to
reach
through
the
bubble
it
will
pop.
How
does
the
bo,
how
does
bubble
describe
this
thing
at
all.
B
E
You
should
definitely
add
that,
because
I
didn't
think
of
it
and
you're
entirely
right,
it
will
bring
to
confusion
on
the
web.
E
B
Global
global
execution
context
context.
So
when
you
put
new
in
front
of
it,
it's
actually
a
sentence
that
makes
sense
new.
E
E
But
to
me
it's
not
clear
if
that
couldn't
apply
also
to
the
concept
of
compartments.
B
I'm
not
even
here
to
defend
it,
I'm
I
I
that
that
description
came
to
my
mind
literally
while
I
was
writing
tests
and
reading
the
spec
pros
over
and
over
and
over
again-
and
I
was
like
oh
yeah-
this
is
a
global
object
with
the
execution
context,
so
I
just
threw
it
in
there.
But
again,
at
the
end
of
the
day,
it's
a
realm
maker.
G
Yeah,
I
just
wanted
to
make
a
quick
observation.
If
I
may
I'm
thinking
about
this
in
terms
of
well
actually
in
mathematical
terms,
if
we're
going
to
come
up
with
a
new
name
for
realm,
I
would
strongly
suggest
we
stick
to
something
that
has
a
mathematical
meaning
behind
it
maps
and
sets
follow
that
rule
pretty
well.
G
The
definition
of
a
map
mathematically
is
pretty
clear,
ditto
for
set
and
in
computer
programming
we
could
call
it
effectively
name
spaces,
I'm
not
saying
we
should
I'm
just
throwing
it
out
there
as
a
as
a
consideration,
but
if
you
want
to,
if
you
want
to
use
a
name,
I
think
having
it
that
where
it's
based
at
least
somewhat
in
mathematics,
definitions
or
computer
science,
definitions
might
be
helpful.
G
D
A
A
If
provided
that
this
group
succeeds,
there
will
eventually
be
constructors
for
things
that
are
like
workers
in
and
there
will
be
a
constructor
for
things
that
are
like
compartments
and
there's
like
one
they.
They
are
and
contain
the
others
depending
on
you
know
like
like
their
layers,
and
it
would
be
nice
to
come
up
with
a
set
of
names
that,
where
the,
where
the
names
make
sense
together
and
which
isn't
which,
which
might
actually,
depending
on
what
system
of
names
we
find
that
actually
works
for
the
three
concepts.
A
It
might
make
sense
for
realm
to
stay
realm.
If
we
can
find
names
for
better
names
for
agent
and
compartment
that
fit
the
theme,
and
we
need
to
come
up
with
a
better
name
than
agent,
because
agent
is
a
terrible
name.
F
So
I
was
suggesting
that
we
could
talk
about
arenas
where
that's
preceded
by
like
what
it's
an
arena
of,
so
a
realm
could
be
a
callable
boundary.
Realm
could
be
like
an
object
arena
and
then
a
compartment,
or
at
least
the
maybe
the
subset
of
compartments.
That's
about
module
loading
could
be
like
a
module
arena
and
then.
B
Arena
yeah
object
arena
is
sort
of
misleading
because
the
the
the
realms
as
we've
designed
them
so
far
disallow
the
passing
of
objects
back
and
forth.
So
I
mean
you
could
make
you
say
like
yeah,
but
you
have
all
the
objects
inside
it.
It
is
indeed
an
arena
filled
with
objects.
I
won't
disagree
with
that,
but
I
think
folks
would
find
it
surprising
that,
like
a
thing,
that's
referring
to
objects
can't
actually
talk
in
objects.
F
And
module
space
if
we
wanna,
if
we
wanna,
have
compartments,
be
under
a
similar
naming
scheme,
so
space
is
a
is
a
math
word.
If
you
want
map
and
set
or
math
word,
space
is
like
a
you
know:
you
have
like
vector
spaces
or
in
general,
different
kinds
of
algebraic
spaces
that,
where
you
have
you
know,
values
and
operator
operations
on
the
values
and
they're
all
like
closed
inside
of
the
space.
E
Well,
I
actually
want
to
get
back
to
a
little
bit
the
relation
between
compartment,
realm
and
agent.
The
way
I
see
it
is
what's
specific
to
a
compartment.
Is
that
it
it
gives
you
a
new
global
object.
The
fact
that
you
can
load
that
it
has
its
own
module
graph
also
is
kind
of
consequence
of
that,
but
the
most
defining
feature
is
that
it
has
a
new
global
scope.
E
Go
ahead,
yeah
in
my,
in
my
opinion,
what
defines
realm
is
that
it
has
a
new
set
of
intrinsics,
not
that
it
not
not
anything
else
and
for
the
agents.
The
defining
feature
is
that
what
is
it
execution?
All?
The
execution
is
synchronous
inside
an
agent
right.
A
E
And,
and
by
extension
and
whatever-
and
the
consequence
of
that,
is
that
whatever
reference
that
you
can
get
your
hands
on,
you
can
you
can
access
the
value?
But
if,
if
you
have
a
global
vendor,
that
doesn't
mean
you
can
you
can
reach
by
value
the
objects
in
the
other
in
the
other
realms
inside
the
agents?
So
I
think
the
defining
feature
really
is
synchronous.
E
E
And
if
you
apply
those
like
global
space,
intrinsic
space
and
sync
space,
you
lose
the
relation
between
them.
Now.
B
E
Inherit,
what's
your
upper
level
has
so
because
you
have
sync
access
in
in
an
agent
realms
and
compartments
also
have
sync
access,
because
you
have
a
new
global
on
in
a
realm
the
compartment.
Also,
actually
you
know
they
share
the
global
of
the
well
so
realm
and
compartment
share
the
sync
environment
of
the
agents.
Compartments
share
the
global
of
the
the
realm
and
the
global
context
is
different
for
every
one
of
them,
but
each
of
them
also
is
one
of
the
under
it.
C
I
think
we
might
be
well
served
to
draw
a
chart
or
diagram
of
this,
because
we
have.
We
have
this
layering
of
of
things,
which
are
all
sort
of
vaguely
compartment-like,
but
each
one
has
has
you
know,
is
isolates
certain
aspects
of
itself
from
its
peers
and
shares
certain
stuff
with
them
that
that
it,
that
is
then
governed
by
whatever
the
higher
level
thing
is
and
and
and
even
though
we're
all
kind
of
know
all
of
this
stuff.
B
I
would
get
behind
compartment
but
like
I'm
also
fully
behind
realm
forever,
so.
D
There's
a
one
of
the
suggestions.
Also
john
jdg
also
talked
about
it
yesterday
at
home,
and
he
sent
a
link
there.
There
was
a
weak
video
alex
already
answered
like
suggesting
brain,
but
I
very
like
that.
D
Wikipedia
article
has
interesting
things
about
it,
like
with
the
mathematical
references
and
at
some
point
it
also
leads
to
two
other
words
they're
mathematical
analog
analogous
to
this
one
of
them
is
that
what
general
suggests
like
space
as
a
mathematical
space,
but
also
this
one-
I
don't
like,
had
a
manifold
but
on
the
but
like
in
some
sense
I
want
to
like
brain,
but
I
I
I
agree
with
alex
in
advance
that,
like
this
will
confuse,
will
be
confusing
with
the
main
brains.
C
C
Whereas
what
we're
trying
to
do
is
capture
the
distinction
between
different
kinds
of
you
know:
different
different
members
of
this
species
of
thing
or
different
levels
in
this
hierarchy,
and
we
should
be
focusing
on
focusing
trying
to
find
language
which
emphasizes
the
distinction
between
the
layers
as
opposed
to
what
any
one
of
these
layers
is
in
isolation,
because
if
you
take
them
in
isolation,
the
same
word
will
work
equally
well
for
any
of
them.
C
E
I
yeah,
I
think
we
should
really
try
to
find
words
that
capture
the
difference
between
those
three
entities
first
and
then,
and
then
see
if
there's
a
taxonomy
that
could
apply.
A
Yeah
it's
it's
likely
that,
as
you
suggest,
we
just
need
a
word
that
that
implies
a
region
that
has
a
boundary
it's
any
of
the
words
that
are
region
that
has
a
boundary
and
then
maybe
find
adjectives
to
to
prefix
that
make
the
distinction
clear
yeah.
But
I
don't
know
what.
A
Yeah
so
in
this
diagram,
using
arbitrary
names
made
up
for
the
purposes
of
me
of
having
a
diagram,
everything
inside
of
a
realm
communicates
synchronously
everything
else
by
the
thread
communicates
to
synchronously.
Can
it
well?
Okay?
So,
let's,
let's
put
it
this
way,
everything
within
a
realm
has
a
single
event
loop
that
that
they
share.
So
so
there
are,
there
are
distinctions
like
do.
We
have
shared
intrinsics.
Do
we
have
a
shared
event
loop
and
having
a
shared
event?
A
Loop
means
communicating
synchronously,
do
they
have
shared
memory
right
like
can
a
shared
array
buffer,
be
shared
between
between
things
that
are
within
that
within
that
boundary
and.
B
E
A
They
would,
they
would
be
in
a
shared
memory.
Yeah
well
realms
would
would
still
be
within
a
shared
memory,
even
with
especially
with
endowments
right.
You
don't
need
to
resort
to
asynchronous
message
passing
for
communication,
as
you
do
as
like,
somewhere
between
threat
and
process
and
javascript
kind
of
makes
this
messy,
because
workers
are
used
for
both
of
those
terms
or
isolates,
and
and
such
I
mean
to
be
clear,.
E
Realms
callable
rounds
today
are
already
shared
memory
because
you
have
function
references
so.
A
Yes,
yeah
yeah.
It
is
certainly
conceivable
that
a
shared
array
buffer
could
be
shared
directly
between
two
realms.
They
might
have
two
ins,
they
would
have
distinct
instances
of
the
of
the
shared
array
buffer,
but
they
would
be
views
into
the
same
memory
and
that's
what
I
mean
by
shared
memory
daniel,
I
see
daniel's
hand,
anybody
who's,
unable
to
post
their
hand,
feel
free
to
just
say
you
you
want
to
place
in
the
conversation
in
in
chat
and
we'll
make
a
space
for
you,
daniel.
F
Oh,
by
the
way,
I
recently
learned
how
to
raise
your
hand
in
the
new
version
of
zoom
and
it's
you
click
on
reactions,
and
then
you
click
raise
hand
on
the
thing
that
pops
up,
which
I
didn't
know
anyway.
So
what
do
people
think
about
space?
I
mean
I
feel
like
that,
could
lend
itself
to
this
hierarchy,
where
we
could
put
words
on
the
front
of
it.
I.
A
Think
I
I
think
that
to
chip's
point
space
realm
compartment,
verse
all
are
words
that
could
fit
in
that
position
and
space.
A
Or
any
of
them
are
equally
suitable
for
that
role.
F
E
I
actually
do
like
the
boundary
more
than
space
because
it
it
gives
where
the
the
limit
is,
and
so,
if
you
apply
it
with
the
prefixes,
I
think
it
actually
works
where
my
problem,
if
you
apply
the
prefix
to
something
like
space
or
environment,
it
doesn't
capture
the
the
limit.
So
at
that
point,
a
synchronous
band
area
for
agent,
an
object
boundary
for
a
realm
and
a
global.
F
E
For
me,
it
means
whatever
is
in
it.
Is
that
specific
thing
so
synchronous
battery
everything
inside
that
binary
is
going
to
be
synchronous,
object,
binary
all
the
objects
inside
that
boundary
are
going
to
be
related
and
a
global
binary
every,
and
maybe
you
can
find
something
else
in
global,
because
it's
very
overloaded,
but
all
the
the
context
in
there
or
yeah
context
context
boundary.
I
don't
know
the
compartment.
E
One
is
a
little
bit
iffy,
but
it
I
I'd
like
to
find
something
that
captures
that
everything
inside
the
compartment
is
is
in
the
same
global
context.
D
On
the
space
thing
I
have,
I
take
this
part
of
still
from
the
brain
wikipedia
page,
I'm
saying
on
the
objects
being
a
mathematical
structure,
so
that
such
as
sets
vectors
spaces
topological
spaces.
That's
one
of
the
things
that
I
would
actually
like
support
usage
renaming
it
to
space.
D
A
Mind
so
we
could.
We
could
rationalize
space
using
a
taxonomy
from
an
astronomical
metaphor,
especially
since
we
have
globe,
as
in
global
system,
is
already
something
that
we've
thought
about,
using
as
a
name
in
javascript
for
module
system.
Actually,
that's
not
good,
and
you.
F
Yeah
yeah,
I
don't
when
I
think
about
the
word
space.
I
think,
if
that's
supposed
to
be
in
outer
space,
then
they're
what
it
means
to
to
make
a
new
space.
I
meant
space
more
like
well,
there's
I
don't
know,
there's
a
bunch
of
stuff
in
this
room
and
I'm
going
to
make
a
space
over
here
and
a
space
over
there,
and
those
are
two
places
that
I
can
you
know
operate
in.
E
So
somebody
shut
down
sandbox
but
as
its
own
might
be
a
problem,
but
if
we,
if
we
qualify
it,
we
did
make
more
sense.
Then
object.
Sandbox,.
C
A
There's
so
so,
we've
we've
talked
about
naming
this
after
the
thing
that
it
encloses
we
haven't
talked
about
naming
after
what
can
pierce
the
boundary.
A
So,
let's
so
suppose
that
we
suppose
that
we
take
an
old
idea
of
callable
boundary
and
we
keep
boundary
and
callable
denotes
what
can
pierce
the
p,
what
can
pierce
the
boundary
and
then,
if
we
named
it
systematically
around
that
idea,
it
would
be
a
message,
boundary,
a
callable
boundary
and
I
guess,
object
boundary
or.
F
A
C
Yeah,
the
the
word
primitive
is
is
troublesome
because
it's
both
we
use
it
both
as
an
adjective
and
as
a
noun
and
when
somebody
says
primitive
boundary
they're
using
it
in
the
noun
sense,
with
the
noun
being
turned
into
an
adjective.
And
so
it's
ambiguous
as
to
whether
you're
talking
about
a
boundary
which
is
itself
primitive
or
a
boundary
which
separates
primitives
from
each
other
and-
and
I
just
don't
think,
there's
any
clean
way
out
of.
C
A
Dan,
I
still
see
your
hand
is
up.
I
don't,
I
assume
you're
not.
A
Okay,
well,
we
have.
I,
I
want
to
reserve
some
time
off
the
record
in
five
minutes
and
I
know
that
daniel
had
hoped
to
reserve
15
to
talk
about
web
integration.
Let's,
let's
give
that
five.
D
F
F
That
hosts
should
be
able
to
add
globals,
but
on
the
web,
they're
going
to
add
a
reduced
set
of
globals
and
in
particular
we
want
the
web
to
add
globals
that
have
less
power.
So
the
most
important
thing
which
we
already
have
accomplished
in
the
current
draft
spec,
is
that
all
the
globals
have
to
be
configurable.
A
So
I
want
to
recap:
the
position
from
the
from
the
perspective
of
the
lockdown
proposal
is
that
that
we
do
not
that
having
a
configurable
global
is
sufficient
for
the
needs
of
anybody,
constructing
a
locked
down
environment
and
that
if,
if
realms,
are
created
with
powerful
mechanisms
to
communicate
amongst
themselves
or
others
yeah,
the
configurability
is
sufficient
in
order
to
reduce
their
power
down
to
a
locked
down
equivalent
and
that
I
don't
think
that
we
need
to
impose
a
stronger
constraint
than
that.
F
Okay,
that's
good
to
have
established.
I
think
that
means
that
the
current
specification,
which
doesn't
specify
anything
more
than
configurability,
is
sufficient
on
the
javascript
side.
Yes
for
the
web,
we
still
need
to
specify
the
exact
set,
and
I
think
the
guideline
that
I
mentioned
even
though
power.
F
F
Is
that
something
you'd
kind
of
agree
with
that's
correct,
yeah,
okay,
great,
so
the
the
mechanism
for
this
will
be
that
in
in
web
ideal,
we'll
add
a
like
exposed,
equals
realm
thing
analogous
to
exposed
equals
window
or
exposed
equals
worker
and
yeah
salesforce
has
been
sponsoring
egalia's
work
on
on
realms
in
general
and
now
on
realm
html
integration
with
with
this
stuff,
hopefully
soon
we'll
be
working
on
on
browser
implementation
as
well
and
yeah.
F
E
It's
I'm
trying
to
understand.
Is
there
a
reason
you
want
to
restrict
what's
available
on
a
realm
and
not
inherit
anything
that
was
from
the
incubator
realm?
So
if
you
were
in
a
window,
you
get
all
the
globals
of
the
window,
but
as
configurable,
if
you
were
in
a
worker
and
you
create
a
realm,
you
get
all
the
globals
of
the
worker
but
make
sure
they
are
configurable.
E
My
my
understanding
is
that
one
of
the
use
case
is
test
suites.
It
should
be
being
able
to
to
create
an
environment
so
that
you
can
run
a
test
in
it
and
then
tear
it
down
and
if
that
environment
doesn't
have
doesn't
replicate
the
incubator
you,
those
those
use
cases,
will
run
into
heavy
complications.
F
I
imagine
that
test
suites
could
make
things
work
by
using
a
membrane.
Didn't
they
no.
E
Not
not
for
everything,
if
you
don't
have
fetch
in
it,
and
you
can't
like
the
performance
of
of
of
array
buffers
and
things
like
that.
Is
it
it's
going
to
be
extremely
complicated.
You
can't
do
post
message
through
a
membrane
either
or
because
you
don't
have
access
to
structured
cloning,
so
I
mean
it.
It
would
be
a
huge
nightmare
to
do
it
through
through
a
membrane.
F
Yeah,
I
can
see
how
those
things
would
be
nightmares,
I'm
still
a
little
skeptical
of
just
giving
everything.
I
think
I'll
have
to
think
about
it.
Some
more,
I
feel
like
we're
gonna
run
into
problems
if
we
just
put
everything
there,
but
I'm
not
sure
exactly
what
the
problems
are.
Maybe
maybe
it's
just
because
we've
been
talking
about
a
review
set
for
so
long
that
it's
stuck
in
my
head.
That
way.
A
It's
I
I,
I
think
that,
having
it
being
expressed
expressly
marking
things
that
are
that
are
that
can
be
shared
with
a
realm
or
can
or
or
have
that
are
constructed
with
a
default.
Realm
is
good
and
it
gives
us
an
opportunity.
A
It
would
give
the
web
folk
an
opportunity
to
explicitly
think
about
the
the
problems
case
by
case,
which
is
good,
especially
because
some
of
those
things
like
location,
if,
if,
if
we,
if
it's,
if
it's
just
do
whatever
you're
doing
for
your
existing
realm,
that's
clearly
going
to
run
into
cases
where
the
configurability
constraint
might
not
be
reviewed.
A
And
some
of
these
things,
I
imagine,
are
not
possible
to
emulate.
Well.
F
We
could
expose
lots
and
lots
of
things
to
realms
with
an
eye
towards
just
excluding
things
that
have
these
configurability
issues
like
window
dot
top,
and
the
other
thing
is
that
I
don't
know
if
specs
are
all
written
to
make
it
so
that
it's
okay,
if
it
has
another
copy
of
things
I
mean
some
specs
are
when
they
use
web
ideal.
It
should
should
be
okay,
but
other
specs
like
streams,
at
least.
F
A
Yeah,
I'm
I'm
in
favor
of
doing
a
careful
review
and
expressly
marking
anything
that
that
should
be
in
every
realm
yeah.
I.
E
E
I
would
I
would
suggest
that,
instead
of
doing
an
exposed
window
exposed
worker
and
exposed
realm
equality,
you
had
it
as
a
flag
like
allowed
in
sub-realms,
or
something
like
that
for
cases
where
you
have
an
api
that
might
be
only
on
a
window
and
not
in
a
worker.
But
you
want
to
allow
on
a
realm
that's
created
by
a
window
but
not
created
by
in
a
realm,
that's
created
by
a
worker.
F
If
we
want
to
allow
realms
to
do
io
in
this
way,
then
yeah,
I
agree
that
this
rebuilder
is
a
separate,
extended
attribute,
but
I
think,
probably
in
a
future
ses
meeting,
we
should
talk
more
about
the
criteria
for
which
things
are
exposed
in
realms,
because
I
was,
I
just
had
it
in
my
head
for
a
really
long
time
that
we
would
omit
these
things.
I
guess
we're
we're
disagreeing
and
we
should
come
to
a
conclusion
and
then
we
then,
based
on
that
we
can
do
all
this
spec
logic.
Yeah.
D
There
is
one
a
note
about
this
I've
I
at
least
from
what
I
understand
from
my
use
this
case.
D
What
I
want
for
realms,
I
we
currently
have
just
have
the
constraint,
the
the
the
new
global
days
of
the
realms
being
an
ordinary
object
and
all
the
properties
being
added
there,
they're
always
going
to
be
configurable
that
help
us
doing
configuration
that
we
need
so
having
a
yo,
it's
not
recommended,
but
it
still
can
be
avoided
after
configuration.
D
So
this
is
a
constraint
that
we
have
and
it's
set
in
manuscript
so
so
far
I
am
fine
and
not
worried
too
much
anything
after
that
is
more
like
an
assumption
that
I'm
making
the
more
that
we
have
results
and
more
configuration
that
we
need
in
when
we
set
a
new
realm
to
give
a
perspective
for
everyone.
D
We
have
like
a
very
large
usage
of
realms
even
today,
and
we're
expanding
that
to
even
much
much
more,
and
that
means
like
one
of
the
problems
today
is
that,
like
the
iframes
are
overloaded
with
everything-
and
there
is
a
big
configuration
of
iterating
over
everything
from
coming
from
the
iframes
and
deleting
everything
because
we're
to
be
replacing
creating
the
virtualization
on
top
excuse
me
and
that
that
makes
it
like
a
huge
load
of
configuration.
So
my
assumption,
if
we
have
much
more
at
some
point,
this
might
become.
D
This-
might
take
a
toll
on
memory,
performance
or
whatever.
But
it's
assumption
it's
assumption.
It's
too
early
to
talk
about
it.
It's
like.
I
cannot
measure
that
I
can
only
imagine
like
if
we
have
like
thousands
of
methods
to
delete
and
remove
and
iterate
over
them.
E
Quick
question:
I
thought
any
of
these
type
of
system
usually
just
have
an
allow
list,
and
anything.
That's
not
on
there
it
gets
removed
so
is,
is
the
concern
that
more
things
need
to
be
added
to
the
allow
list.
A
A
There
has
to
be
a
mechanism
for
endowing
I
o
capabilities
to
your
child
process
so
that
it
doesn't
have
to
route
all
of
its
messages
through
you
for
performance
reasons,
if
nothing
else-
and
I
think
that
that's
going
to
establish
a
precedent
for
having
an
options
bag
that
says
any,
I
o
capability
granted
to
a
worker
at
least,
would
have
to
be
expressly
granted
and
have,
and
each
of
those
would
have
an
impact
on
the
startup
time.
A
For
example,
having
access
to
the
shared
having
access
to
a
dom
itself
is
even
something
that
is
optional
for
a
child
process
and
extremely
expensive
in
the
cases
where
considerably
expensive.
In
the
cases
where
it's
actually
needed.
E
A
A
Yeah
and
like
with,
if
creating
a
new
realm
with
no
options,
would
would
confer
no,
I
o
capabilities,
and
then,
if
we
want,
if,
if
a
realm
on
the
web
needed,
I
o
capabilities
having
them
be
flagged
seems
totally
reasonable
to
flag
it
in
the
constructor
options.
Back
there.
F
Proposal
right,
it
would
be
like
we
would
be
kind
of
throwing
the
testing
use
case
that
you're
talking
about
under
the
bus
initially
and
then
maybe
doing
it
later,
because
there's
there's
a
ton
of
different
things
that
we're
doing
later.
For
example,
the
whole
loader
api,
which
we
all
agree,
is
important
yeah,
even
though
we
we
know
that
we
want
it.
We're
not
we're
going
step
by
step.
A
We're
at
time
I
I'm
going
to
stop
the
recording
and
I'm
hoping,
oh
no
rick.
Anyhow,
that's
that's
a
meeting
then
thanks
for
coming.