►
From YouTube: CORE WG Interim Meeting, 2019-10-16
Description
CORE WG Interim Meeting, 2019-10-16
A
A
C
E
A
A
F
A
D
F
B
B
B
B
Good,
so
let's
do
a
quick
document
status
right
now
we
have
one
document
in
the
RFC.
Editor
cube
and
looks
like
we're
stay
there
for
a
while,
because
the
RFC
in
Turkey
was
pretty
full.
At
the
moment
we
have
percent
mental
edge,
which
I
think
is
waiting
for.
The
authors
actually
Ari
told
me
that
he
won't
be
able
to
make
it
to
projects
on
probation
and
him
here
in
the
list
of
apologies.
So
here
we
are
waiting
for
the
others.
B
I
hope
it
has
been
quite
active
and-
and
there
is
a
whole
devil
upcoming,
with
all
the
the
good
iesg
input,
and
maybe
we
can
do
quick
run
through
that
after
the
document
status,
Sena
mezmur
units
is
entering
ITF
last
call
now
so
Alexei
was
very
quick
in
in
his
a
devaluation.
The
resource
writer
e
is
still
in
ad
evaluation
for
echo
requests
tag.
B
B
B
H
B
The
context,
doesn't
really
explain
how
implementations
that
do
not
implement
hop
limit
handles
the
cash
versus
implementations
that
do
implement
the
hop
limit
handle
the
cash
we
need.
How
was
it
doesn't
a?
Why?
So,
maybe
we
want
to
add
some
text,
never
depart
from
that.
I
think
I'm,
pretty
good
shape
and.
B
F
F
F
B
A
A
A
A
A
A
If
nodes
ITF
workshop
16,
10,
19
I,
don't
know
why
Corey
signal
is
not
in
there
but
yeah,
okay,
so
the
first
discussed
was
the
multicast
for
notification
so
distorted.
If
you
remember
with
a
holy
discussion
about
about
pub/sub
and
how
to
get
multicast
for
notification
working,
they
were
several
proposal
and
one
of
these
was
to
get
multicast
working
for
responses.
A
Just
we
have
gone
through
it
and
we
have
clarified
what.
So,
if
you
remember
that
draft
suggested
to
reserve
ranges
of
tokens
for
these
specific
applications
and
when
we
presented
that
at
the
ETF
we
got
some
very
skeptical
faces,
particularly
from
Klaus
and
Caxton,
and
so
we
have
this
this
workshop
to
make
sure
that
okay,
can
we
not
do
it
in
another
way,
flash
better
and
I
think
we
figured
out
the
better
way
to
do
it,
and
this
picture
or
the
whiteboard
is
supposed
to
explain
that.
So,
let's
let's
go
through
it.
A
D
A
A
So
that
the
biggest
thing
or
the
first
thing
we
did
so
you
have
these
these
boxes
up
here.
The
top
box
is
the
server
and
the
bottom
box
is
the
client.
Oh,
you
can
see
that
there
is
only
one
end
point
in
the
server
the
smaller
box
this
one,
but
the
the
client
actually
has
two
different
end
points.
The
coop
end
points
are
distinct
because
one
of
them
uses
the
unicast
IP
and
the
other
one
uses
the
multicast
IP
multicast
pieces.
Yes,
the
multicast
address
that
the
client
is
listening
on
for
most
agnostic
as
notification.
A
A
Basically,
the
server
is
dialing.
The
client
I
am
going
to
respond,
as
if
you
sent
me
this
request,
and
the
request
is
the
this
little
box
here
and
the
client
is
not
actually
sending
this
request.
So
this
is
a
what
do
we
call
this
F,
either
fake
or
a
virtual
request
that
the
client
is
not
sending
and
the
server
is
actually
responding
to
this
request
using
the
token
phantom
request.
A
A
A
A
B
A
G
I
And
during
the
discussion
we
also
realize
that
the
preserving
or
whatever
the
stocking
value
ranges
wasn't
needed
in
the
first
approach
either.
But
while
in
the
first
approach,
we
were
using
options
to
give
information
back
to
the
client.
Now
we
are
essentially
using
the
payload
of
that
message
for.
A
We
can
go
into
details
of
that.
I
have,
and
the
next
slide
will
say
that
little
just
to
recap:
why
did
we
have
token
ranges
before
we
had
token
ranges,
because
we
did
not
do
this
distinction
between
the
co-op
play,
client,
co-op,
endpoints
that
and
we
saw
the
situation
we
were
worried
about-
was
okay.
What,
if
a
client
oops?
A
Let's
use
this
token
to
send
a
unique
out
requested
server
and
it's
waiting
for
a
no
response.
On
that
token
and
then
suddenly,
the
same
server
is
telling
the
same
client
that
you
should
listen.
For
this
same
token,
so
the
client
would
would
will
be
waiting
on
two
responses
with
the
same
token
and
wouldn't
be
able
to
differentiate
between
them.
Without
with
the
discussion
we
closed
with,
and
we
realized
that
that
is
not
possible,
because
the
co-op
endpoint
is
also
identified
not
only
by
its
identified
by
its
IP
address
and
port
and
the
other
definition.
A
B
I
think
we
learned
in
an
earlier
discussion,
so
we
may
have
been
confused
because
for
requests
to
a
multicast
address,
we
need
to
relax
the
the
matching
word.
So
that
response
and
that
comes
from
a
unicast
address,
matches
the
the
request
that
went
to
a
multicast
address.
But
in
this
direction
we
don't
have
to
relax
anything.
A
A
A
A
Phantom
request
yeah
and
the
client
itself
will
it
would
have
to
do
with
the
exact
stream
thing.
So
so
this
this
phantom
request
information.
However,
that's
formatted
will
be
sent
by
the
server
to
the
client,
every
cuz
for
every
client
that
wants
to
the
wants
to
observe
this
multicast
yeah
resource
I.
Don't
have
to
call
it
multicast
resource
cantata,.
B
A
A
A
A
How
does
it
say
that
it
could
be?
It
could
be
a
gap
to
a
separate
resource
which
can
be
well
known,
slash
the
resource
to
observe
it
could
be
observed
to
that
excel
tree,
sir.
So
there
are
different
ways
and
under
response
the
same
way,
it
could
be
either
205
and
the
content
of
the
response
can
be
like
the
payload
of
the
coop
response
can
be
e.
This
is
the
request
that
I'm
going
to
match
to
this.
A
A
B
G
A
G
So
there's
there
that
be
also
another
option.
That
is
to
to
just
let
the
topic
out.
Let
the
notifications
expire
and
of
not
sending
something
for
some
for
longer
than
the
max
age
and
then
waiting.
Whether
someone
is
asking
for
the
for
the
observation
data
again,
but
that's
just
a
knee-jerk
reaction.
D
And
in
a
completely
different
direction,
one
thought
we
had
was
what,
if
the
server
actually
start
sending
the
multicast
notifications
before
it
even
has
heard
from
the
first
clan
I
just
sit
there,
and
whenever
it
wants
to
send
some
notification,
it
does
it
and
then
you
can
independently
for
that
spread
the
information.
How
to
listen
to
those
notifications
and
intense
can
start
doing
that,
and
they
can
stop
doing
that.
The
service
that
I
do
know
where
this.
A
Thirty
different
ways
and
they're
not
exclusive,
they
could
be
their
options
like
I.
Think
I
personally
think
that
they
both
make
a
lot
of
sense
and
they
should
be
described
and
I
don't
see
a
problem
with
them
coexisting.
This
is
something
like
yeah.
Whoever
deploys
this
would
have
to
decide
how
how
do
I
make
this
work
today
the
data
server
always
and
a
multicast,
or
that
start
standing
on
multicast
if
it
has
too
many
clients
and
I
support
that
and
I
saw
it
something
from
Michael,
Bay
I.
B
B
I
think
the
the
uske
statue
has
always
been
named
for
actually
using
multicast,
not
after
group
communication
rules
use
case,
but
the
medical
use
case
has
been
searching
light,
so
you
can
have
them
all
switch
on
and
off.
At
the
same
time
now,
of
course,
that
only
works
if
the
no
packets
are
lost
and
so
on.
F
F
F
B
A
Yeah-
and
that
was
yeah-
that
was
the
first
bullet.
What
request
what,
if
request-
and
the
second
bullet,
is
what
response?
Because
that's
that's
about
four:
how
do
you
get
this
information
and
what
information?
Exactly
because
here
we
were
saying
we
start
thinking:
okay,
what
what
it
is
not
not
only
the
token
is
request,
one,
including
the
token.
A
Multicast
IP
address.
That
was
something
that,
in
the
previous
version
of
the
draft
we
moved
out.
So
it
was
there
in
the
beginning
of
what
we
removed
it
out
because
we
thought
okay.
This
is
negotiation
multicast
addressing
too
complicated,
but
in
this
case
maybe
it's
I
think
it
makes
sense
to
put
it
in.
A
Yeah,
it's,
for
example,
request
one
includes,
for
example,
D,
request,
URI
or
the
options
in
the
request,
and
then
for
security.
It
would
be
necessary
to
have
the
Oscar
if
Oscar
is
used
group
of
score
external,
a
B.
So
this
is
the
a
a
B
that
is
external
Adidas.
This
use
in
to
in
response
in
all
the
notifications
that
are
linked
to
the
registration
request
and
since
yeah
that
does
not
it's
not
sent
to
the
server
is
telling
okay
I'm
gonna
I'm
gonna
link
all
the
reputations
to
that.
D
We
were
thinking
of
literally
putting
the
quad
wire
format
into
inside
the
payload
of
response
for
chrysalis,
and
then
we
were
thinking,
maybe
not
and
doing
something
nice
based
on
keyboard
instead
and
then
we
were
thinking.
But
what,
if
request?
One
is
actually
protected
by
off
score
and
then
taking
out
the
information
would
break
the
Earth's
core
protection
of
the
information.
So
then
we
were
back
at
maybe
literally
transferring
the
ausco
protected
request,
one
and
inside
this
payload
operator,
yeah.
D
I
A
I
A
J
I
D
I
I
I
A
B
A
B
A
I
Coop
husband,
as
an
analogy
in
7390
and
the
beasts
we
are
building,
we
are
describing
both
the
insecure
case
as
a
starting
point,
and
we
don't
really
recommend
to
go
for
unless
you
really
know
what
you're
doing
and
then
how
to
make
it
secure
with
Google
score
I
suppose
we
can
think
the
same
way
here.
I.
A
C
H
F
I
A
I
G
I
A
No,
you
cannot
no
you're
not
having
a,
but
what
we
did
in
real
life
is
that
it
was.
It
is
a
no
for
this
co-op
endpoint.
This
court
on
point
is
not
doing
absorption,
but
if
the
SBS
client
supports
the
specification-
and
this
client
will
know
to
start
or
to
how
to
do
the,
how
to
have
this
second
coop
endpoint
do
the
observation
so.
F
A
D
D
G
G
So
if
you,
if
you're
aware
of
what
what
that
document
does
or
just
just
to
summarize
briefly
that
document
that
we
could
put
the
request
message
in
pull
into
an
option
and
if
we
just
replaced
that
word
and
by
the
way
for
the
request
message
see
there,
then
you
could
use
the
same
things
that
we
put
in
message
for
over
there
just
again
and
have
basically
the
same
situation.
So
this
is
a
response
to
what
has
been
defined
over
there.
A
So
the
answer
is
yes:
how
did
we
get
there?
So
this
on?
The
figure
on
the
right
here
is,
is
pub/sub,
with
a
distinction
between
configuration,
server
and
content
broker.
So
there
is
the
collection
of
the
topic,
configuration
then
configuration
and
then
the
content
which
can
be
on
another
another
node
and
there
are
different
ways.
Club
was
saying
there
are
different
ways
of
of
distributing
the
notifications.
A
Publish-
and
so
these
right
now
are
not
defined,
but
there
will
be,
and
when
we
looked
at
this
picture-
and
we
looked
at
to
realize
that
this
match
with
our
multicast
notifications
and
the
answer
is
yes,
you
could
still
have
a
subscriber
that,
as
a
unicast
IP
address,
it
would
reach
for
the
configuration
there
is
some
something
that
needs
to
be
defined
there
about
how
you
do
the
discovery
of
where
the
public
is.
What
is
the
configuration?
What
are
the
different
ways
that
you
could
actually
receive
the
notification?
A
But
the
point
is
that,
assuming
that
the
topic
supports
the
the
notification
multicast
notification,
then
you
can
get
the
same
response
as
number
four
over
here
and
then
from
this
multicast
IP
endpoint
you
can,
you
can
do
the
same
thing.
You
can
listen
to
the
responses
on
the
multicast
IP.
That
was.
A
D
A
D
Since
we
are
thinking
about
this,
a
different
way
rocketing
notifications
to
subscribers,
like
over
MQTT
and
so
on,
there
must
be
something
defined
in
the
pub/sub
proposal
where
you
can
get,
and
here's
how
you
get
innovative
notifications
or
web
qgt.
Here,
though,
is
how
you
get
your
notifications
over
multicarts.
Yes,
how
you
can
supply
a
call
back
your
I
now,
where
the
broker
can
put
two
for
example,
and
so
yeah
all
of
these
options
could
be
horrified
by
one
of
the
broker
or
just
one
of
them,
so.
A
The
part
about
the
current
broker
and
the
subscriber,
and
only
the
part
about
responses
are
as
notifications,
and
it
would
be
very
simple
to
add,
notifications
or
app
requests
and
mqtt
publish
and
they
could
even
like
be
under
specified
and
and
I
would
say
like
this
can
be
done,
and
you
know
someone
else
will
do
it
and
then
have
a
separate
document
for
the
configuration
with
all
the
like
API
or
interfaces
that
needs
to
be
defined.
How
do
you
get
the
configuration?
How
are
these
formatted
and
everything.
A
My
suggestion
was
that
one
that
it's
up
to
you
to
fill
in
to
the
working
group
to
decide
like
what's
the
best
way,
I,
definitely
see
that
these
cause
all
together
but
yeah,
and
also,
of
course,
that
whatever,
however,
we
format
this
message
for,
we
need
to
make
sure
that
it
fits
with
the
interface
on
the
configuration
topical
figuration
in
the
in
that
document.
When
that
will
be
specified.
D
D
A
H
D
Lord,
it
is
not
really
to
be
read
that
I
can
try
to
him
to
tell
a
bit
about
it,
and
we
have
talked
about
in
particular
in
Montreal,
a
lot
about
different
components.
We
have
resource
directory
and
we
a
table
saw
broker
and
and
so
on,
and
one
hope
was
or
what
observation
was
that
these
are
all
structurally
very
similar
in
some
way
most
avoiding
on
some
kind
of
collection,
resource
pattern.
D
And
so,
if
you
have
all
these
different
fits
of
small
components,
some
small
pieces
of
components
in
core
and
I
hope
would
be
that
we
could
define
them
avoiding
some
kind
of
loose
decoupled
way
and
then
simply
arrange
them
and
combine
them
into
these
applications
like
resource
directory
and
observe
program
and
so
on
and,
for
example,
collector
resources
is
this
happen
on
this?
Data
is
based,
and
it
have
is
a
much
more
than
that,
and
then
we
have
an
octuple
for
the
collection
results
of
configurations
and
in
resource
directory
of
collection,
resource
of
resource
descriptions.
D
And
so
we
could
actually
use
data
up
to
upload
the
configurations
as
an
interface,
all
those
resource
descriptions
without
defining
anything
specific
to
two
possible
and
resource
directory.
But
then
we
would
need
to
define
example,
resource
discovery
interface
in
addition
to
data
up,
so
that
clients
can
actually
look
up
resources
that
the
we
are
looking
for
and
when
we
have
create
topics
that
publishers
and
of
subscribers
need
some
kind
of
topic,
discovery
interface
and
then
can
check
upon
dissolve
that.
Aren't
those
two
interface
not
really
the
same
thing,
because
topics
are
just
resources.
D
And
then
we
have
these
other
public
certifications
by
a
unique
out
notifications
by
multi-card
specifications
by
a
foot
call
back
and
so
on,
and
that
is
not
really
specific
about
certifications
by
a
MQTT,
and
that
is
not
really
specific
toolbox
for
brokers
could
be
simply
used
by
some
device
to
deliver
telemetry
to
to
solve
to
the
other
devices
or
to
the
cloud.
Also.
D
We
have
lots
of
reusable
parts,
Ian,
I,
think
and
Francesca
are
taking
photos
of
me
and
the
focus
that
we
can
move
those
parts
independently
forward.
So
they
don't
gawk
each
other
and
one
at
the
same
time
getting
synergies
from
aligning
them
within
each
other
so
that
you
don't
have
to
do
the
whole
thing
again.
If
you
meet
the
same
thing
in
a
different
application.
A
And
before
we
move
to
the
next,
I
actually
wanted
to
bring
up
that
since
we
have
been
working
on
sub
sub
I,
and
since
we
have
the
one
of
the
eight
chairs
in
the
meeting,
hi
Jim
I
think
it
will
be
time
to
adopt
the
pub/sub
profile,
because
I
think
we
need
to
work
on
sub
sub
security.
At
the
same
time
as
we
work
on
pub/sub.
A
Okay,
so
I
was
hoping
we
could
do
an
adoption
call
before
ITF,
106
and
I
mean
even
now.
I
talked
to
people
who
were
you
here
and
they
are
all
supporting
and
I'm
asking.
Maybe
like
okay,
we're
not
that
many
today,
but
please,
if
you
think
it's
important,
go
to
a
mailing
list
and
hand
voice.
So
in
the
middle
list.
A
C
A
F
A
F
A
A
A
A
So
this
will
be
step
zero
and
then
the
client
would
have
the
group
manager
address
and
join
resource
address,
and
it
would
try
to
get
probably
either.
It
knows
what
the
duration
server
to
talk
to,
or
it
would
talk
to
the
group
manager,
get
back
an
error,
saying:
hey,
you
need
to
talk
to
the
authorization
server
and
then
step.
I
It's
in
two
steps,
the
joining
node
knows
only
one
thing,
the
name
of
the
application
group,
so
it
has
to
find
out
the
name
of
the
security
group,
use,
identify
the
score
group
link
to
the
joint
resource
at
the
group
manager
and
the
multicast
IP
address
used
to
communicate
in
the
group.
In
order
to
do
that,
mostly
because
of
the
way
the
reason
directory
works
requires
to
perform
two
queries.
I
This
is
the
first
one,
and
this
is
actually
need
content
brought
up
by
this
document.
So
through
this
query,
the
joining
node
gets
in
the
payload
of
the
response.
In
fact,
the
link
to
the
generations
of
the
group
manager
friendly
read
the
name
of
that
security
group,
the
name
of
the
application
group
again
plus
a
number
of
things
we
define
in
the
latest
version
of
the
draft,
actually
a
number
of
parameters
describing
during
time
the
algorithms
and
related
parameters
used
in
do
school
group.
I
That's
something
that
the
joining
up
node
will
end
up
knowing
from
the
group
manager
eventually
anyway,
but
knowing
that
so
in
advance
makes
it
possible
to
avoid
trial
and
error
process
with
the
group
manager
or
M,
and
explicitly
ask
him
to
the
group
manager
upon
access.
Token
posting,
like
we
described
in
the
ACE,
joining
draft
also
in
case
in
this
response.
The
joining
not
get
indication
of
algorithms
using
your
score
group
that
the
join
node
doesn't
support
at
all.
I
It
can
just
give
up
here
and
not
continue
with
trying
to
get
an
access
token
or
whatever.
So
it's
a
hint,
but
it's
fair.
It's
possible
later
pain
to
the
joining
node
next
slide,
and
this
is
the
second
query
we
added
in
the
examples
of
the
drug
for
the
sake
of
completeness,
but
this
is
a
pattern
taken
from
the
resource
directory
document.
I
B
B
A
B
A
I
By
the
way,
sex
GP
end
up
GP.
Well,
there
will
be
strings
the
values
they
have,
depending
on
your
particular
mutation
and
setting,
of
course,
while
for
the
other
ones,
we
are
not
going
to
register
any
value
for
them.
Values
are
going
to
be
taken
by
the
poker,
consider
histories
and
some
just
about
the
parameter.
Concerns
quote.
B
D
Link
format
was
like
the
rock
solid
solution
that
we
are
convinced
of.
Is
the
right
and
best
way
to
move
forward
or
the
security
of
the
future.
Then
I
would
be
all
in
for
creating
this
registry.
At
least
I
am
trying
to
convince
myself
or
I'm,
not
convinced
yet,
but
I'm
trying
to
convince
myself
that
link
format
is
nearing
its
end
of
life
and
therefore
I
have
some
reservation
on
spending
too
much
time
on
growing,
more
life
into
link
format,
but
I
am
NOT.
A
B
B
B
D
D
A
I
D
A
D
We
don't
know
if
there
are
any
one
out
there
with
their
own
link
target
attributes
and
if
we
created
three,
maybe
they
register.
They
are
attributes
at
some
point,
but
exactly
not
immediately.
What
we
know
of
is
all
the
attributes
in
our
sieves
and
contrast,
so
we
can
always
keep
a
list
in
some
wiki
somewhere,
where
we
keep
track
of
what
we
personally
know
then
anytime
created
registry,
when,
when
we
feel
that
the
wiki
is
not
good
enough,
that
the
couple
that.
D
D
D
I
A
E
E
D
That
is
not
something
that
can
be
done
and
within
one
week
it
requires
a
bit
more
thinking
how
we
want
to
set
up
the
registries,
put
more
guidelines
for
designated
experts.
That's
the
main
reason.
Currently
we
have
too
few
guidelines
designated
experts
and
so
to
avoid
a
flood
of
registry
creating
updating
drafts.
We
could
also
say
that
this
new
registry
here
can
be
part
of
this
part
of
that
effort,
but
that
would
be
moving
more
slowly
than
and
send
Alondra
I.
A
B
The
direction
I
can
give
here
is
it's
really
easy
to
to
define
just
tight
attribute
registry
and
I'm
all
for
moving
small
documents
unless
we
are
doing
something
that
actually
does
require
coordination
and
I'm,
not
sure
that
that
would
be
actually
the
case
here.
We
do
have
to
define
the
registry
policy,
and
that
would
take
a
little
bit
of
discussion,
but
I,
don't
think
that's
going
to
be
very
complicated,
I
mean.
A
A
A
D
A
B
I
Okay,
yes,
then
I
had
a
nice
poem
which
next
bullet
point
I.
Remember
the
core
meeting
in
moral
Jane
proposed
the
possibility
to
have
this
document
actually
moved
to
the
edge
working
group
calls
all
in
all
it's
about
advertising
and
discovering
information
that
are
useful
in
ace.
In
fact,
and
I,
agree
and
I.
Remember,
Carson
be
positive
on
this.
I
We
started
in
core
because
this
is
about
rigid
directory
that
that
has
caused
it
home
I'm
fine.
With
that
it's
a
slapping
I
wanted
to
check
again
because,
of
course,
we
are
working
on
on
an
athlete
of
this
draft
and
if
that's
something
we
can
confirm
now
or
soon
we
can
submit
that
update
place
right
away.
B
A
I
I
I
Okay,
we
talked
about
that
at
the
very
beginning
we
discarded
adoption,
which
resulted
in
that
second
query:
we
showed
before
because
we
wanted
to
have
the
oscar
group
manager
in
principle,
agnostic
of
the
multicast
IP
address
using
the
group
cause.
He
should
care
about
key
material
all
in
all,
and
also
because
the
multicast
IP
addresses
are
in
fact
associated
with
application
groups
that
have
their
own
different
ways
of
being
registered
in
the
result
directory.
C
I
C
I
D
I
By
the
way,
a
few
meetings
ago,
what
was
raised
by
the
chairs
that
before
formally
raising
a
dog
Hong
Kong,
they
should
go
to
reduce
and
aware,
if
you
promise
review
where's.
Yes,
so
yeah
we're
waiting
for
those
reviews
and
otherwise
they
seem
pretty
advanced
and
close
to
be.
Consider
for
or
doctors
actually.
A
D
It's
something
we
should
definitely
do
in
my
opinion
find
employed.
It
should
be
done
if
we
want
them
to
standardize
this
approach
a
lot.
It's
a
secondary
question,
I'm,
not
100%
sure,
but
one
is
requite
and
we
know
they
believe
it,
but
you
should
definitely
do
the
exercise
on
finding
out
how
it
looks
like
if
you
use
coral
and
coral
reef,
for
example,.
B
J
C
C
A
B
You
so
what
one
interesting
thing
about
Tuesday
and
Wednesday
excuse
me:
Wednesday
and
Thursday
factions
and
14th
w3c
is
not
going
to
meet
and
Singapore.
We
have
our
workshop
on
the
15th,
but
we
are
not
going
to
have
a
w3c
Weber
things
meeting
on
the
13th
and
the
14th
people
generally
have
had
budget
problems
adding
another
meeting
that
late
in
the
year.
So
that
is
not
going
to
happen.
So
that's
great
because
those
of
us
arriving
early
are
going
to
have
more
time
doing
things
like
these.