►
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).
C
B
For
frozen
intrinsic,
we
did
land
the
flag
for
the
command
line.
When
you
fired
it
up,
it
will
freeze
the
JavaScript
intrinsics.
It
does
not
freeze
nodes
core
mm-hmm
provided
functionality,
mm-hmm
we've
already
gotten
reports
of
people
saying
they
can't
run
their
code
if
they
turn
it
on
mm-hmm,
somewhat
expected.
It's
all.
B
A
But
failing
that,
if,
if
node
proceeds
to
you
know,
if
this
thing
gets
some
adoption
and
people
are
failing
specifically
because
the
override
mistake
and
they'll
fix,
then
you
know
that
will
start
a
process.
Well,
they
will
where
they
will
feel
the
pressure
to
fix
those
few
cases
in
their
code,
which
is
good.
B
Yeah,
so
we
do
have
some
fairly
large
libraries
that
don't
work,
namely
Express,
which
is
probably
the
biggest
HTTP
library
for
node.
We
haven't
heard
back
on
the
possibility
of
changing
things
in
expresses
source
code,
but
we
did
have
some
people
objecting
to
requiring
Express
update
their
source
code.
We
had
an
interesting
comment
where
somebody
even
suggested
not
having
the
same
object
model
as
JavaScript
slightly
v8
quickly
shot
that
down.
Okay,.
B
A
Hacking,
those
four
objects
that
Salesforce
identified
or
not
either
way
is
fine.
With
regard
to
all
of
my
goals,
I
prefer
not
doing
the
hack
if
that
results
in
pressure
on
the
ecosystem,
that's
that
eventually
causes
people
to
not
fail
under
the
override
mistake.
That
would
be.
That
would
be
an
awesome
outcome.
B
A
B
A
B
A
A
D
D
D
A
B
Particular
there
was
an
interesting
I
think
undercurrent,
at
least
in
other
discussions
about
get
originals
providing
guarantees.
The
new
get
originals
proposal
is
no
longer
a
set
of
functions
on
a
namespace.
Instead,
it
is
a
set
of
module
identifiers
to
get
the
original
I
did
not
know
that
so
I
think
this
is
actually
an
okay
approach.
B
A
Yeah
I
mean
one:
could
one
could
obviously
provide
some
kind
of
API
for
setting
the
import
map
and
it's
actually
good
that
they're
not
providing
it
because
there's
a
Authority
issue
there,
you,
you
know
until
you
have
the
kind
of
system
versus
user
mode
distinction
that
SCS
gives
us
I
mean
this
baby.
This
is
really
the
same
issue.
Is
the
system
object
which
Dominic
also
resists,
is
that
anything
that
could
modify
the
import
map
is
clearly
privileged
over
anything.
That's
within
the
world
that
was
manipulated
by
that
modification.
B
A
B
B
A
B
That's
perfectly
viable,
so
that's
it
for
first
and
intrinsics
against.
Anybody
else
has
concerns
the
follow
up.
Past
person
intrinsics
was
going
to
be
an
attempt
to
freeze
core
I'm,
currently
not
working
on
that
somebody
else
wants
to
look
into
that
they
can,
but
there
is
a
very,
very,
very
large
amount
of
user
land
marvels
that
mutate
note
score
in
non-trivial
ways,
so
that
would
be
very
breaking
to
turn
that
on.
We
should
probably
work
through
the
frozen
interns.
That's
in
state
that
breaking
code
is
okay
for
the.
A
Business
so
as
far
as
the
SES
goal
goals
are
concerned,
just
freezing
the
JavaScript
intrinsics
is
what
we
need.
We
don't
need
core
frozen,
because
we
consider
core
to
be
all
in
root
terminology.
We
consider
core
to
be
resources
rather
than
sure
shared
objects
anyway,
so
they
would
need
to
be
attenuated
and
and
virtualized
as
they
get
provided
by
more
trusted
code
to
less
trusted
code.
B
A
A
B
A
Yeah
anything
that's
sort
of
inherently
giving
access
to
you
know,
to
cause
effects
or
to
sense
effects
and
in
the
outside
world,
which
note
core
certainly
does
on
the
that
that
that
has
to
go
through
attenuation
on
the
way
and
the
problem
with
with
the
existing
node
API
is
you're
right
that
it's
a
difficult
API
to
attenuate
and
I.
Think
that
just
a
constraint
we
have
to
live
with
and
I,
don't
think
I.
Don't
think
freezing
really
does
much
to
to
help
that.
B
We
also
had
a
request
to
open
up
a
PR
about
fixing
promised
instances
having
a
doc
domain.
Own
property,
rudimentary
looks,
doesn't
look
like
will
break
anything
if
we
move
it
to
a
getter
setter
with
a
weak
map
or
something.
But
we
haven't
made
a
PR.
They
don't
really
have
the
energy
to
try
to
push
that
through.
Okay.
B
B
B
A
A
C
B
A
A
B
B
B
B
A
A
Designed
the
cess
proposal
in
the
session
to
accommodate
was
running
trusted
shims
that
do
modify
the
primordial
x'
before
freezing
the
primordial
x'
and
that's
you
know
more
compelling
for
the
browser
than
for
node,
because
for
in
node
you
have
the
the
person
running.
The
the
code
has
much
more
control
over
what
version
of
node
it's
running
on,
but
for
the
browser
being
able
to
shim
it
up
to
a
given
version
before
freezing
it
and
creating
the
the
the
shared
secure
environment
did
seem
like
did
seem
important
enough
that
we
didn't
have
didn't.
A
Have
the
realm
born
frozen,
we
had
the
realm
born
unfrozen,
you
run
the
registered
shims
and
then
after
the
registered
shims
are
run.
Then
you
freeze
does
any
of
that.
If
you're
not
hitting
that
as
an
adoption
barrier,
the
the
inability
to
shim
before
freeze
I
would
not
bother.
Then,
because
it's
you
know
it's
a
trickier
mechanism
and
we
thought
it
through
under
the
assumption
that
needing
to
shim
is
it,
you
know,
would
be
necessary.
A
E
E
Language
spec
evolution
prior
to
complete
implementation
by
the
by
the
engine,
implementers
and
I,
don't
see
that
being
any
less
of
the
concern
in
node
than
in
the
browser.
Probably
there's
a
lot
more
stable
stuff
in
in
node
that
that
you
know
it
works
in
whatever
version
it
works
in
and
anybody
who's
trying
to
work
in
a
future
version
is
doing
experimental
stuff.
And
so
you
know
it's
less
of
a
big
deal
in
that
sense.
B
A
A
B
So
the
big
thing
for
at
least
the
policies
that
we've
had
very
few
complaints
about,
but
I,
think
very
little
usage
is
they
are
URL
based
a
couple
meetings
ago.
I
also
did
mention
that
I
opened
up
to
the
security
working
group,
the
possibility
of
adding
data
URLs
and
HTTPS
URLs
to
what
we're
allowed
to
import
and
was
asking
for
feedback
there.
I
haven't
seen
any
complaints
about
them
outside
of
some
cache
consistency,
issues
for
data
URLs,
so
I.
A
A
But
I
mean
in
general,
the
I
don't
want
URLs
to
become
part
of
the
JavaScript
language.
We
succeeded
at
essentially
keeping
them
out
of
the
language
and
just
talk
about
specifier
strings
and
leaving
it
up
to
the
environment
to
provide
the
means
by
which
the
default
means
by
which
specifier
strings
are
dereferenced
as
a
source
code.
So.
B
Well,
there's
three
topics
here:
one
is
layered:
api's
are
now
having
fallback
specifiers
within
string
specifiers
for
browsers,
yes
and
then
so
that's
a
little
bit
complicated.
If
we
do
not
sync
with
them
and
have
the
exact
same
mechanism
there,
even
if
it's
not
URLs
with
the
pipe
character
and
then
for
arbitrary
strings
within
there,
that
are
not
URLs
their
paths
or
their
names
of
some
kind.
I'm.
A
B
A
A
B
C
A
There's
no
special,
there's
no
way
to
specify
percent-encoding.
That
does
not
recognize
a
URI
that
just
says
in
the
string
you
can
present
and
code
and
it
will
%
decode
it
before
interpreting
it
as
a
specifier
string.
There's
no
way
to
do
that
and
have
it
actually
do
the
right
thing
for
you
are
eyes
because
you
are
eyes
only
accept
the
percent-encoding
at
particular
places
in
particularly
you
can't
present
and
code
the
the
protocol
name
and
probably
can't
present
and
code
the
colon.
C
B
A
B
C
Right,
which
is
which
is
also
the
reason
I
remember
now,
one
of
the
meetings
that
I
was
saying
very
strongly.
They
should
be
careful
about
how
to
identify
namespaces,
because
that
first
:
following
an
initial
sequence
of
alphanumerics,
is
very
important
if
it
is
to
be
potentially
interpreted
as
a
URI,
the
set
of
all
possible
schemes-
it
is
not
fixed
and
can
change
over
time.
So
if
you
have
like
STD
:
key
value,
then
that
is
a
potential
collision.
If
STD
is
ever
introduced
as
a
URI
scheme,
yeah.
A
C
C
C
I
want
to
say
it's
like
it's
the
difference
between
a
URI
and
a
URI
reference
coke
to
your
lie.
You
get
the
same
experience
when
a
service
worker
and
the
service
worker
API,
like
the
URLs.
You
get
our
scope
to
a
particular
base,
the
scope
of
the
service
worker
and
the
you
could
drop
the
leading
part
and
it
becomes
relative
to
just
like
when
you
have
a
base
on
a
document
like
an
HTML
document,
you
could
start
to
refer
to
relative
URLs
without
the
leading
slash,
except
that
only
doesn't
apply
for
module,
specifies
weight.
A
C
B
A
B
A
B
C
A
B
A
A
A
Is
anybody
there's
a
tc39
meeting
coming
up?
Does
anybody
have
any
expect
to
be
doing
anything
at
that
meeting?
That
would
be
of
interest
in
this
group.
A
One
is
the
whole
issue
about
the
system.
Well,
the
thing
that
was
originally
the
system
object,
which
is
more
generally,
the
issue
of
having
the
platform
provide.
A
white
list
of
of
rules
are
pure.
A
That's
the
the
new,
less
stressful
form
of
the
system,
object
less
stressful
in
the
sense
of
more
adoptable,
the
because
of
resistance
from
Google
to
the
idea
of
introducing
the
system
object
as
a
namespace.
So,
instead
of
introducing
one
global
there,
instead
going
down
a
path
where
you
introduce
for
every
new
resource
introduces
its
own
new
global.
So
a
new
week
map
constructor
as
their
that
week
map
week,
ref
constructor,
a
new
finalization
group,
global,
a
new,
get
stack
global
and
then
for
these
Global's.
A
That
are
resources
that
are
inherently
resources
and
should
not
be
part
of
a
safe
environment
of
a
pure
environment
that
we
would
have
a
purity
whitelist
that
the
resources
would
not
appear
on
so
that
a
SES
like
mechanism,
any
SES
like
mechanism,
could
consult
the
whitelist
so
that
an
old
mechanism
could
know
for
new
abstractions
created
after
the
old
mechanism
was
written,
which
things
were
allegedly
pure.
Not
the
I
expect
there
to
be
a
fight
specifically
with
regard
to
allowing
we
regress
to
go
forward,
not
blocked
on
that
issue,
and
I
am
I.
A
But
the
reason
why
I'm
thinking
of
allowing
it
to
go
forward
without
this
safeguard
that
I've
always
previously
asked
for,
is
that,
hopefully
we
are
winding
down
rapidly
in
introducing
whole
new
abstractions
by
introducing
new
Global's
into
JavaScript
and
we're
going
to
shift
to
introducing
new
built-in
modules.
There
has
to
be
some
built-in
module.
That
goes
first,
that
that
pays
the
cost
of
being
the
forcing
function
on
built-in
modules.
I
expect.
That
will
be
a
moment.
A
But
in
any
case,
if
we
do
get
the
distinction
in
built-in
modules,
which
is
the
one
that
is
worth
fighting
for,
then
we
basically
have
a
fixed
whitelist
for
these
stragglers
on
new
Global's.
That
happened
before
built-in
modules
and
it's
a
whitelist
which
means
that
if
new
Global's
do
get
if
new,
pure
Global's
do
get
added
after
that,
the
mechanism
failsafe
it
we
fail
to
what
you
know
old
code.
Since
the
whitelist
is
a
user
code.
E
A
V,
so
the
so
that's
one
topic,
it
sounds
like
nobody
here
wants
to
stop
me
from
accepting
that
compromise,
so
I
will
I
will
expect
to
do
that.
Another
one
is
the
import
expression.
A
It
is
now
the
case
that
that
realms
and
SES
and
all
of
its
variants
and
shins
realistically
have
to
do
something
for
module
code
for
the
shims.
We
have
to
do
it
by
rewriting.
We've
talked
in
this
meetings
about
how
we're
actually
planning
to
do
that,
but
a
realms,
API
or
SAS
API
that
only
traffic's
any
valuable
strings
is
not
is
no
longer
a
linguistic
surface
to
present
to
normal
programmers.
A
Asked
for
a
delay
of
one
meeting,
so
we
could
explore
the
issue
that
was
a
lot
of
and
we
and
you
know-
and
we
did
explore
a
lot
of
module
issues,
including
my
trying
to
engineer
a
rewrite
that
was
adequate
for
running
modules.
On
top
of
the
shim.
So
I
don't
see
a
problem
with
it.
Does
anybody
else
see
a
problem
with
it
and
did
anybody
else
spot
any
assumption
I
was
making.
That
seems,
over-optimistic
can.
E
I
can
I
echo
some
of
that
back
to
see
if
I
understand
it
back,
but
just
try
to
capture
some
understanding
that
I
have
yeah
issue.
Is
that
once
you've
admitted
a
construct
that
can't
be
dealt
with
sort
of
parsing,
okay,
well,
you've
already
signed
up
to
do
parsing,
so
you
know,
what's
the
cost
of
you
know,
you
know
marginally
more
parsing
or
more,
or
rather
the.
A
Well,
that's
the
idea
on
the
shim
side
right
I
mean
the
idea
on
the
spec
side
is
also
that
the
realm
spec
and
inherited
by
the
suspect
will
be
providing
import
hooks
and
the
import
hooks
will
clearly
hook
the
import
expression
and
write,
write
and
anything.
That's
and
and
clearly,
even
if
they're
divided
up
between
two
different
classes.
For
other
reasons,
because
of
you
know
where
scripts
get
evaluated
versus
where
modules
get
evaluated,
the
nature
of
the
hook
would
be
the
same.
E
E
Parser
access
into
the
language
API
where
you
had
a
standard
ast
and-
and
you
could
just
ask
the
engine
itself
to
parse
things
and
give
you
back
the
parse
tree,
which
I,
which
would
which
would
be
helpful
here,
but
certainly
proposing.
That
is
not.
It
would
not
be
helpful
here
because,
obviously
this
this
wants
to
go
immediately
and
that
would
be
several
years
down
the
line
at
a
minimum,
but
I
kind
of
wonder
about
it,
because
this
meta
issue
keeps
coming
up
over
and
over
again.
E
A
I
say
clever
because
of
my
purposes,
the
straightforward
thing
about
it
from
their
purposes
is
that
it's
not
proposed
to
serve
the
purpose
of
being
a
programmatic
ast
for
purposes
of
inspecting
or
manipulating
code.
The
purpose
for
which
it's
being
introduced
is
to
have
a
significantly
more
compact
way
of
conveying
code
from
the
server
to
the
browser
and
a
and
a
way
in
which
the
browser
can
ingest
code.
That
is
faster,
so.
E
C
A
E
C
A
E
Left
a
chat
message
saying
that
he
was
be
back
in
a
while:
okay.
D
A
A
As
a
matter
of
fact,
I
remember
I've
forgotten,
that's
the
critical
thing.
Dean
brought
that
up
as
well
he's
correct
I'm
going
to
remove
it
from
the
agenda.
I
just
completely
forgot
about
this,
thanks
for
reminding
me
I'm
going
to
remove
it
to
threat
from
the
agenda
and
if
it
comes
up
I'm
going
to
say
that
propose
that
we've
discussed
its
next
meeting
and
to
many
of
the
people
who
are
interested
cannot
make
it
to
provide,
and
that's
a
perfectly
good
reason.
Okay,.