►
From YouTube: ROLL WG Interim Meeting, 2020-06-26
Description
ROLL WG Interim Meeting, 2020-06-26
A
B
C
A
B
A
This
is
the
agenda
for
today
we
are
going
to
basically
route
through
the
status
of
the
documents
that
we
have
the
working
group,
and
there
is
a
complex
about
agenda.
The
capabilities
and
work-based
are
going
to
be
visited
by
round
this
qualification
bar,
tortures
and
a
united
Europe
diet.
Information
bypass
graft.
It's
like.
A
E
A
A
Okay
and
I
wanna
leave.
It
is
a
subtle
change
here.
In
version
17
and
the
there
was
at
the
F
block
in
the
italic
option,
the
world
a
little
bit
as
well
generally,
most
data
as
well
forced
our
feed
as
a
rule.
So
well,
this
is,
let
sees
al
are
no,
but
this
can
get
a
positive
answer
from
the
goal.
First
can
do
you
want
to
add
something
here.
F
F
But
since
we
had
this
review
by
arranged
interesting
proms,
we
ended
up
with
this
huge
simplification,
because
there
was
some
rest
of
our
text
which
induced
some
possibilities
which
were
actually
not
the
animal
due
to
the
fact
that
we
are
using
non
storing
mode
for
a
world
in
any
fashion.
So
this
allowed
those
simplifications
like
remove
the
anonymous
inner
flow,
which
were
always
the
pain
in
the
neck
since
the
beginning,
so
I'm
happy
that
the
ihe
will
give
out
to
review
that
piece.
It's
just
just
erased
it,
and
so
it's
it's
much
nicer.
F
That's
an
additional
optimization,
that's
how
last
fall
and
it
makes
a
lot
of
sense.
We
use
it
in
mobile
IP.
So
it's
that
new
news.
So
we
just
am
elated
here
what
doing
mobile
IP
and
it
appears
to
be
useful.
So
it's
a
side
effect
I
would
say
no
option
without
that
additional
benefit
in
terms
of
protocol.
We
just
removed
cases,
so
we
just
simplified.
We
basically
print
the
tree
of
perseverance
and.
C
D
F
Np
Dow
was
already
modified
just
that
the
text
was
missing.
To
say
that
we
did
so
a
part
as
well
of
the
whole
review
was
to
find
that
the
way
we
use
the
NP
Dow
was
not
consistent
with
the
NP
doubt
draft,
which
was
limited
to
one
case.
So
we
had
to
extend
so
we
don't
change
it
but
weeks.
Then
it's
applicability
to
other
flows,
which
is
actually
the
expectation
for
the
six
low
drafts
as
well.
So
now
the
whole
thing
is
consistent.
F
F
D
A
F
F
F
G
F
I
mean
if
there
is
any
more
challenge
it'sit's
if
somebody
comes
in
with
a
review
between
now
and
then
all
right,
so
you're
basically
telling
us
two
weeks
from
now.
The
duck
needs
to
be
frozen
so
that
that's
I
mean
reinforcing
the
call
that
in
has
just
made
the
community
like,
if
you
want
the
last
chance
to
any
review
of
change.
Anything
in
that
text
now
is
the
right
time
to
say.
H
So
there
has
been
one
major
change
for
in
the
art
field,
extended
control
option
that
was
added
in
the
last
version
of
the
draft.
Basically,
when
one
major
thing
that
I
missed
out
was
the
option
type
had
us,
I
assumed
that
the
option
type
had
a
secure
bit
on
and
hence
I
placed
the
extended
bet
in
the
second
MSP,
as
you
know,
so,
where
there
didn't
Dominic's
review
he
he
he
mentioned
why.
Why
was
it
necessary
to
the
second
image
me?
H
That's
what
I
realize
that
you
know
a
secure
bit
is
not
eat
it
for
option
dying.
So
now
the
extended
fitness
moved
to
the
first
MSB
position
so
that
that
is
only
the
moment
of
the
moment
of
the
bed
location.
One
more
important
point
that
was
discussed
on
the
mailing
list
in
our
last
couple
of
days
was
when
we
set
this
x
bit.
Essentially,
what
we're
doing
is.
We
are
reserving
the
number
of
archaeal
extended
control
options,
so
we
have
if
the
option
type
is
256
by
using
the
expect.
H
One
of
the
things
that
Dominic
mentioned
was
is,
if
me
are
going
to
do,
are
the
extended
control
options
and
in
the
future.
If
we
see
that
most
of
the
time
I
extended
control
options
are
going
to
be
made
user.
Then
why
can't
we
set
two
beds
in
place
of
one?
In
which
case
the
range
could
be
extended
from
0
0
0
to
0
X,
4
0.
H
So
basically,
what
it
would
mean
is
64
64,
option,
control
options
for
regular
options
and
192
control
option
for
the
extended
ones
1
this
so
that
time
I
thought,
maybe
that
might
not
be
needed.
But
after
some
time
you
know
what
it
realized.
I
realized
that
it
is
possible
for
extended
control
option
to
work
as
a
regular
option,
but
not
vice
versa.
So
it
may
not
be
a
bad
idea
to
actually
say
set
a
bigger
range
for
extended
total
option
right
now.
H
So
so
so,
basically
now
I
think
maybe
it
is
better
to
use
the
range
0,
X
4
0,
to
0
x
FF
for
extended
control
options,
given
that
any
extended
control
option
can
still
serve
as
a
regular
option,
Albert
with
an
extra
bite,
but
still
it
works
as
a
regular
option,
which
is
not
possible.
Otherwise,
you
know
a
regular.
E
H
Right
so
next
slide,
please!
So
that's
that's
that
about
mopix
draft
the
next
next
next
updates
are
about
capabilities,
and
we
have
a
major
updates
here.
So
we
have
been
discussing
about
capabilities,
queries
and
response
for
all
the
time.
So
the
the
latest
draft
update
had
added
this
feature.
Basically,
we
now
have
a
new
new
message,
which
can
query
capabilities
and
get
a
response.
H
So
basically,
there
were
three
things:
I
kept
in
mind
is
a
node
should
be
able
to
query
the
list
of
supported
capabilities
and
a
node
should
be
able
to
query
specific
capability
details,
so
these
are
two
different
things
and
a
set
of
capabilities
to
respond
may
exceed
MTU.
So
if
you're,
if
you're
caring
for
all
the
capabilities
it
is,
there
is
a
good
chance
that
a
single
all
the
capability
list,
the
the
set
of
capabilities,
can
or
may
not.
H
F
H
F
If
we,
if
we
start
doing
this,
that
that's
a
question
I
had
for
the
other
draft,
when
we
we
won't
send
something
in
multiple
message:
we
we
don't
have
in
repo
the
capability
to
say
1
of
3
2
of
3
3
of
3
right
here
when
we
receive
something
in
multiple
message.
If
we
got
all
the
messages,
so
yes,
but
it's
not
very
happy
with
that,
there
was
wondering
if
we
could
think
about
you
know
having
that
kind
of
signaling
in
this
draft.
F
H
F
I
F
E
F
H
H
So
this
is
the
so
a
set
of
new
messages
have
been
added.
One.
Is
the
cap
queue
message,
which
is
the
query
message
I'm
calling
it
it
basically
is
it's
very
similar
to
down
message.
It
has
a
cap,
okay
capability,
a
query,
a
sequence
number
and
then
there
are
set
options.
There
is
one
new
control
option
that
has
been
added
which,
which
is
which
is
which,
which
capability
type
list.
H
So
basically
what
a
node
who
is
squaring
the
remote
node
what
it
can
do
is
it
can
specifically
say
that
I
am
querying
you
for
capability
type,
1
type,
2
or
type
3.
So
now,
one
twist
here
is
this
capability
type
list
option
is
not
sent
as
part
of
this
query,
then
the
responder
will
essentially
send
back
this
list
saying
that
I
support
capability,
type,
1
capability,
type,
2
and
capability
like
3,
so
this
is
this-
is
this-
is
one
of
the
one
of
the
things
that
you
know?
H
One
has
to
keep
in
mind
during
so
if
this,
if
then,
if
this
control
option
is
not
present
in
the
capability
queue,
it
means
that
the
the
sender
is
squaring
the
list
of
capabilities.
If
the
capability
query
already
contains
this
type
list,
with
the
corresponding
capability
types,
then
it
essentially
means
it
is
squaring
for
those
specific
capabilities
next
slide.
Please.
H
So
the
capability
s
or
the
capability
response
is,
is
sent
in
in
response
to
the
capability
query.
The
capability
query
sequence
number
is
simply
correct
copied
from
the
cap
queue
message.
The
options
here
can
be
to
type
2
things
you
know
one
one
is
the
set.
Is
the
capability
type
list
that
was
mentioned
before?
H
The
second
is
the
specific
capabilities
which
are
being
queried
by
the
response
by
the
query
ur,
so
the
responder
will
send
I
high
support
capability,
type,
1
type
2,
and
the
complete
information
of
those
capabilities
and
send
and
optionally
send
a
capability
type
list.
Now
there
is
a
specific
reason
why
the
capability
type
list
might
be
sent
to
the
sent
by
the
capability
responder
to
the
capability
query.
H
H
H
H
So
here
what
happens
is
this
is
so
that
based
on
the
list,
the
route
or
the
area
or
any
other
node
might
in
turn
query
send
a
capability
type
list
squaring
capability,
1
and
capability
to,
and
then
the
capability
one
then
capability,
two
options
are
sent
back
to
the
to
the
rule.
In
this
case
this
is
the
this
is
the
normal
case
and
then
the
in
the
last
example
in
next
slide.
Please.
H
This
is
where
things
get
interesting.
So,
basically,
what
the
root
is
saying
is
I
want
to
see
capability,
1
capability
to
2,
&,
3,
&
4.
This
snowed
supports
only
2,
&
3.
It
doesn't
support
1,
&
4,
so
I'm
in
this.
If
the
capability
type
list
is
present,
it's
a
it
means
that
cap
1
and
cap
for
is
not
supported
by
by
by
this
node,
so
it
has
sent
only
capability
to
enter
capability
3.
H
So
if
the
this
is
this,
this
is
the
way
in
which
a
root
can
know
whether
all
the
capabilities
that
it
it
is
requesting
from
the
6,
cellar
or
salon
are
responded
back
with
or
not
so
in
this
way,
I
can,
for
this
example
at
least
I
can
avoid
ascending
that
one
1
of
3,
one
of
one
of
three,
were
two
or
three
order:
3
or
3,
so
it
kind
of
bits.
So
this
is
these
are
all
the
examples
call
forms
that
I
have
next
slide.
Please
is
there
an
excellent.
F
F
Other
thing
that
we
asked
how
that
I
have
not
seen,
but
maybe
implicitly
it's
there.
We
said
two
things
we
said
it
could
be
the
parent
pulling
all
the
children.
So
it's
a
multicast,
but
when
just
one
hop
asking
the
parent,
in
which
case
it's
a
unicast,
the
last
thing
it
could
be
the
root
putting
everybody
in
the
network,
in
which
case
we
have
to
propagate
it
down.
H
F
And
if
we
want
to
look
at
this
flow,
it's
it's
like
you
know
there
is
this
capability
and
the
route
wants
to
make
sure
that
everybody
has
it
something
like
that.
Now
you
realize
that
if
the
route
would
really
pull
everybody
like
I
just
described,
that
could
be
a
humongous
amount
of
information.
Yeah.
F
That,
obviously,
is
the
route
can
can
find
that
that's
a
lot
about
option
5,
but
it
may
never
be
sure
that
everybody
supports
of
option
5,
because
somebody
missed
the
message
it's
a
false
negative.
So
but
at
least
the
mechanism
seems
useful
for
the
kind
of
reasons
we
we
introduced.
All
this
discussion,
which
is,
do
we
have
somebody
who
does
not
support?
You
know
I,
don't
know
the
RFC
eight
one
three
eight
or
something.
J
F
H
I
understood
I'm,
just
thinking
whether
to
do
it
with
I
mean
when
Roode
wants
to
query
such
kind
of
information.
Is
it
is
it
should
they
do
it
with
capabilities
or
this
this
this
control
options
as
part
of
tau
or
part
of
capability
query.
You
know,
because
if
we
used
our,
then
we
we
implicitly
get
all
the
other
things
like
act
mechanism
and
all
those
things,
and
you
know
the
through
tack.
The
load
will
even
be
able
to
find
out
whether
the
route
has
actually
got
the
message
or
the
capability
set.
F
E
F
We
try
over
and
over,
and
so
you
you
for
a
long
time
with
expansion
back
off,
so
you
hope
that
everybody
will
get
it,
but
you
never
have
an
actual
acknowledgement.
So
that's
rippled
di
of
style
enough
for
this
message.
You
mentioned
Twitter
earlier,
but
initially
when
we
discussed
it.
One
of
the
reasons
why
I
wanted
to
have
a
different
message
compared
to
dial,
yes,
that
we
wanted
to
use
the
different
style
I
know'd
would
answer
to
his
parent
only
when
all
the
children
evans-
oh,
yes,
we
tried
and
failed,
but
basically
the
idea
was.
F
F
The
diffusion
Stein,
like
I
said,
is,
when
you
have
me,
acknowledge
to
your
parent
when
all
your
children
acknowledged
so
so
the
question
is
flooded
and
then,
when
all
your
children
acknowledges
you
respond
to
your
parent
and
when
all
your
peers
are
acknowledged
to
your
parent,
then
did
your
parent,
and
so
she
untied
it
said
well
when
the
root
gets
all
his
children
or
all
his
first
operators
acknowledging.
That
means
that
he
is
get
the
full
set
of
information.
That's
the
way.
Yeah
Jackie
were
absolute.
The
Jo
algorithm.
F
And
so
there
isn't,
we
did
not
do
it
in
repo
original.
Is
we
never
sure
that
everybody
is?
They
are
listening,
getting
the
message,
etc.
So
we
we
we
based
the
writing
on
the
fact
that
message
is
our
last
and
and
it's
because
we
retry
them
over
and
over
that.
Eventually
they
get
there
for
the
capability.
If
we
want
something
more
reliable,
obviously
it
means
a
lot
more
costly
because
the
parent
would
have
to
retry
to
individual
children
yeah.
H
F
The
routes
unicasting
to
somebody
down
the
deer
more
like
the
route
broadcasting,
the
question
and
the
parents
make
its
making
it
so
that
it's
reliable
multicast
to
the
one
hop
one
way
to
tell
me.
We
have
to
describe
our
the
to
the
world,
but
though
I
do
is
recursively
its
variable.
Much
yesterday
very
help.
Us
we
get
on
the
train
three
or
something
it's
nots,
100%
clear,
even
in
my
mind,
right
I
mean
I
can
see
how
a
diffusion
works.
F
F
H
F
H
F
K
E
F
Noticed
you
mean
anything
you're
fond
of
neighborhoods
I
was
expecting
that
basically
the
recursive
operation
to
be
done
by
parents
down
to
a
children,
so
it
might
be
that
the
parents,
in
a
way
of
a
child
and
a
routine
not
yet
well,
which
case
would
receive
more
information
that
we
expect.
But
I
I
see
that
that's
okay,
okay,.
F
F
Other
case
is
when
you
say:
if
anybody
does
not
support
this
teacher,
tell
me
that
case
the
positing
will
be
funded,
the
top
I'd
be
for
to
negative,
like
you,
don't
get
an
answer,
but
there's
this
guy,
who
does
not
support
the
on
whether
something
they
did
not
share
the
broadcast
right.
So
there's
Chile's.
So
three
depends
on
how
much
we
are
willing
to
pay
farting
a
reliable
multicast
polish
on
three
cows,
a.
F
Freak
low
as
well
and
you
you
don't-
have
a
real
idea
of
how
long
it's
gonna
take
you
especially
back
up
and
after
the
fifth
of
sevens
back
up,
then
maybe
the
guy
gets
the
question
and
when
he
answers,
but
that's
why
afterward?
You
know,
as
you
said,
the
root
will
time
out
and
think
it's
got
all
the
responses
that
someone
with
trickle
you
caught
redo,
that
arrival,
multicast
and-
and
you
plays
bouncing
in
the
time
to
have
the
question
recursively
and
get
the
answer.
F
H
C
A
Just
have
the
change
law
from
31
to
40.
Basically,
we
have
date
6550,
we
advertise
external
roads
with
no
instruments
in
a
lien.
We
have
eliminated
copper
by
hop
all
the
traffic
coming
from
an
external
target.
They
exist
and
necessary
always
encapsulated
to
the
root.
Therefore,
we
have
the
hope,
I
hope
and
we
have
fixed
some
needs
as
well.
So
please
comment:
the
shape
right
up
was
updated
as
well.
E
F
A
B
G
B
B
G
So
I
also
don't
know
how
much
benefit
we
are
going
to
get
from
ITF
review
and
another
is
evaluation
because
you
know
honestly,
you
guys
here
in
the
working
group
are
the
guys
who
know
right
what
what
the
content
looks
like
and
you
know
all
those
things
and
and
if
anyone's
going
to
catch
something
else.
It's
probably
gonna,
be
you
guys-
and
you
already
went
leader
of
this
document
so.
L
So
so,
first
of
all,
first
of
all
the
the
summary
comes
to
you
that
may
help
you
decide
what
to
do.
I
believe
we
have,
the
unaware
leaves
is
is
know,
is,
as
have
you
done,
your
a
write-up
for
it
I
don't
know,
actually
I
have
it's
in
the
queueing.
So
since
most
of
the
changes
that
we
made
to
this
are
to
synchronize
it
with
unaware
leaves.
L
My
suggestion
is:
take
our
our
our
overview.
Stick
comments
and
pass
that
with
unaware
leaves
to
what,
through
the
process,
the
documents
are
not
going
to
advance
on
their
own
anyway.
They
refer
to
each
other
yep,
so
whatever
review
unaware
leaves
get,
you
could
do
that
or
is
once
one
thought
and
the
other
half
of
that
thought
is
well.
You
could
pass
that
summary
to
the
people
that
did
the
reviews
before.
G
Right
and
yeah
I've
done
things
like
that
before
usually
it's,
you
know
one
revision
later
or
something
right,
eight,
but
so
yes,
yes,
I'd,
rather
not
have
to
run
everything
again
because
it's
this
complicated
document
there's
a
bunch
of
cases-
and
you
know
cetera
so,
yes,
I'd
rather
and
that's
a
really
good
suggestion.
I'd
rather
just
go
and
say
these
are
the
changes.
Does
anyone
have
any
real
concerns?
If
anyone
does,
then
you
know
we
probably
have
to
at
least
do
another
age
evaluation
or
something
or
you
know
something
like
that.
G
F
I
completely
agree
with
the
words
that
you
pronounced
earlier.
The
bottom
line
is
we
change
two
things
in
all
those
versions?
The
summary
of
the
summary
is
we
added
the
support
for
honey,
we're
a
leaf
just
like
Michael
just
said,
and
that's
basically,
one
page
of
text
which
is
kind
of
stand-alone
about
what
we
call
external
destinations.
So
this
is
the
piece
that
you
could
focus
on
if
you
want
to
get
some
help
outside
of
the
working
group
just
review
that
we
didn't
do
something
silly
on
that
particular
page.
F
The
rest
is
fixing
those
and
use
cases
in
cases
that
went
through
the
group
then
went
through
the
iesg
and
there
were
bugs
that
nobody
had
the
first
round
of
the
group.
Nobody
is
through
the
years
she
found,
and
thus
we
did
those
reviews
and
reviews
and
reviews
again
then,
here
on
there
there
was
kind
of
a
typo
mode.
F
If
you
like,
because
there
was
one
spirit-
and
sometimes
there
was
one
mistake
and
like
I
said
going
second
times
for
the
ASG,
do
we
expect
they
will
extract
more
of
those
typos
I
would
say
no
waste
a
lot
at
that
time
because,
like
I
said
it's
an
awful
document
since
so
many,
but
at
the
same
time
that
it
was
fixing
bugs
right.
So
so
we
looked
at
what
we
did
when
we
do
these
changes,
we
fix
the
bugs,
and
you
know
when
you
do
a
back
fix.
F
You
try
not
to
introduce
a
new
bug
or
something
here
in
like
exposure
less
bugs
in
there.
It
was
so
expensive,
it's
partially
less
chances
that
anybody
is.
She
was
not
really
aware
of
those
flows
would
catch
anything
I,
love
all
think.
Well,
you
know
the
thought
process
should
be
concentrated.
Is
this
this
page
which
discusses
with
external
this
nation's
and
which
really
was
introduced
for
the
support
of
repairing
the
ones.
G
G
It
might
be
good
if
we
asked
for
one
now,
you
know
by
the
time
I
get
to,
but
over
leaves
they'll
be
done
with
the
review
just
in
case
they
they
have
something
that
way.
I
can
when
I
go
to
the
AG.
I
can
say
here's
the
summary
we've
got
another
review
from
the
air
T
Directorate
and
we
should
just
keep
going.
A
Thank
you.
So
we
have
a
as
well
the
rest
of
the
active
internet
Rob.
We
have
the
projection.
That
has
a
new
version.
We
have
to
check
if
the
Disney
version
close
all
the
open
tickets
that
they
have
I
see
they
opened
its
artistic
open
source
tab.
Then
these
modifications
want
to
be
discussed
today
by
churchill's
in
video
is
in
the
Q
term
a
priority.
We
have
asked,
forgive
you
and
we
are
going
to
get
you
on
a
what
review
of
a
volunteer
at
the
end
of
July.
A
A
C
A
A
A
A
A
B
A
B
E
J
Each
other,
so
the
one
is
the
recent
one,
but
I
worked
on
the
use
cases.
I
would
like
to
have
your
feedback
about
the
future
of
this
rap,
whether
I
should
continue
working
on
it
or
whether
there
is
interest
or
not
from
the
working
group,
and
the
second
one
is
about
the
solution.
One
of
the
solutions
that
we
may
have
with
the
das
modification
Roth
I
will
give
you
a
brief
overview
for
both
of
these
drops
and
then
I
would
like
to
initiate
the
discussion
at
the
end
of
this
presentation
with
the
polycomb
next
slide.
J
J
E
J
J
He
is
to
trigger
the
DAO
control,
packets
and,
as
a
result,
we
will
have
bunch
of
VI
euros
from
all
its
neighbors
and
reticle
diamond
will
be
reset
it
and
we
will
increase
the
traffic
load
in
the
network
and
if
the
consequence
is
we'll
have
the
energy
consumption
in
the
network
in
extra
use
case
that
it
comes
from
both
of
the
first
two
that
comes
from
the
modification
raft.
It's
it's
about
identifying
the
defunct
dock.
J
So
basically,
the
idea
here
is
that
a
ribbon
node
may
remove
enabled
from
its
parent
set
for
a
dock.
For
a
number
of
reasons,
for
example,
the
neighbor
is
no
longer
reachable
or
the
neighbor
had
advertised
is
an
infinite
rank
in
the
dock,
but
the
problem
is
that
a
ripple
node
may
fail
to
remove
a
neighbor.
The.
E
J
May
fail
to
receive
the
neighbor,
the
divorce,
you're
advertising
and
increase
rank
or
neighbors
membership
in
a
different
dock,
that
a
node
will
continue
to
consider
itself
attached
to
a
dog,
even
if
it's
orange
in
the
product
are
unreachable
or
have
moved
to
different
lecture.
This
is
really
Houston's
dock,
which
is
characterized
defunct
from
be
not
perspective.
J
Here
is
that
if,
if
the
node
will
continue
maintaining
the
state
about
a
large
number
of
the
defunct
docks,
it
may
consume
considerable
of
the
total
memory
in
the
node.
So
the
potential
idea
here
will
be
like
the
IES
message
and
inside
this
des
message
it
will
include
the
slack
no
inconsistency
flag,
prevent
the
turtle
timer
to
be
reset
it,
and
it
will
include
some
solicited
information
options
to
identify
the
duck
questions,
for
instance
the
flux
IND
for
the
ripple
instance
and
the
like
idea
to
find
appropriate
products.
J
Then
we
have
the
the
use
case
that
comes
from
the
ripple
observation
from
the
browse
draft
which,
by
the
way,
I
saw
that
role
that
solved
this
issue
from
from
the
draft
from
the
preservation
drove.
So
it
will
be
addressed
here
know
here,
is
that
to
reduce
the
control
traffic
overhead,
reproduces
the
trickle
timer
to
update
configuration
parameters
in
the
absence
of
regular
traffic.
The
Johnson
cities
cannot
be
tested
and
prepared.
If
Brogan
we
do
have.
The
unicast
D
is
to
query
a
node
for
its
di,
oh,
where
I'm
not
receiving
this
unique
as
di.
J
Yes
must
respond
with
the
unicast,
the
IO,
with
configuration
option
and
now
a
question
here
comes
if
this
mechanism
could
be
as
well
made
use
for
probing
the
edge
on
senses,
then.
Finally,
we
have
the
fourth
use
case,
which
I
need
to
update
the
draft
with
zero
one
version.
So
this
drug,
this
uses
comes
from
postcard
from
the
our.
B
J
Meeting
so
where
the
idea
is
okay,
we
have
a
sudden
power
shutdown
in
the
in
the
network.
What
will
happen
is
that
when
the
the
nodes
will
reboot
all
of
them,
all
of
the
devices
would
like
to
discover
the
network
that
will
start
setting
all
of
them
directly
di
es,
but
it's
here
and
there
and
will
be
a
little
bit
chaotic.
So
we
may
have
some.
We
may
need
some
solutions
in
order
to
mitigate
the
problem
with
multiple
di
es
packets
and
the
collision
will
do.
We
may
have
an
additional
little
transmissions
and
so
on.
J
J
Thank
you
so
here
here
comes
the
di.
Yes,
modification
trust,
and
it
comes
with
several
enhancements.
Let's
say,
and
here
I
give
a
big
overview,
so
we
have
somehow
a
solution
to
to
reduce
the
number
of
di.
Your
responses
on
D
I
guess:
request,
adding
some
flux,
for
instance
the
N
empty
flag
and
constants
for
no
inconsistency
and
t
stop
stands
for
type
so
with
n.
J
Basically,
the
node
will
inform
the
receivers,
do
not
reset
your
triple
timer
and
with
n
it
will
identify
with
to
the
receivers
whether
the
di
use
that
they
will
send
in
the
response
will
be
in
broadcast
or
in
universe
if
they,
if
we
want
to
reduce
even
more
the
number
of
responses
via
your
responses,
there
is
a
what's
called
selectivity
of
multicast.
Yes,
messages.
J
The
third
point
here,
what
we
it's
called
the
information
carried
by
the
D,
is
basically
this
is
related
to
the
second
use
case
that
I
talked
before
about
the
identifying
the
defunct
dodok
value
inside
the
DI.
Yes,
we
could
have
some
information,
solicited
information
about
the
route
or
the
prefix
information,
or
about
the
dock
and
so
on.
If
we
want
to
find
the
problematic
dodok.
J
The
fourth
point
here
is
the
response:
training
where
we
some
oh
right.
We
try
to
randomize
the
response
of
the
di
us
when
we
are
sending
NDIS
requests.
So
basically,
what
the
problem
here
is
that,
when
a
node
sends
adias
requests,
which
neighbors,
if
two
or
more
nodes
will
reply
with
di
your
control
pages,
we
make
up
collisions
having
this
response
spreading.
We
will
randomize
the
response
time
and
like
this,
we
will
avoid
to
have
collisions
which
in
turn
makes
it
to
have
retransmissions
if
we
have
collisions
in
video
responses.
J
J
J
J
If
we
want
to
even
more
optimize
the
number
of
nodes
that
will
respond
on
this
di,
yes
control
packet,
we
may
include,
as
I
said
earlier,
the
meter
container
option
where,
if
some
of
the
devices
we
fulfill
specified
around
in
constraints,
nodes
will
reply,
for
instance,
in
this
case,
only
node
a
and
our
will
respond
to
f.
So
we
are
reducing
by
half
the
number
of
vio
packets
in
the
network
show
the
traffic
which
translates
into
energy
next
slide.
Please.
J
So
in
this
slide
we
have
a
theoretical
count
of
the
I/o
packets
in
response
on
di
es
control
budget
as
a
function
of
the
number
of
neighbors
receiving.
So
basically
in
the
x-axis
we
have
total
number
of
neighbors
and
on
the
y-axis,
is
the
total
number
of
the
I/o
responses.
So,
as
you
can
see
with
the
default
version
of
ribbon,
which
is
the
gray
color,
we
are
the
more
we
have
neighbors
traffic
the
control
packets
we
may
get,
can
go
above
1000
and
control.
Packets,
yes,
is
request.
J
So
here
I
have
a
series
of
slides
where
I'm,
showing
some
detailed
depiction
of
behavior
with
different
flag
with
different
flag
with
different
flags,
the
configuration
and
with
some
metal
container
options
or
response
breathing
options.
So
basically
we
did
a
small
setup,
small
network
of
10
nodes
where
we
had
node
6
that
was
joining
into
this
network.
J
J
What
happens
when
we
do
not
have
any
flag
any
option?
So
basically
we
have
the
default
of
6550.
Node
6
will
join
around
time,
equal
30.
So
basically
we,
the
red
line,
means
that
the
radio
is
off
and
then
green
line
indicates
that
radio
is
on
and
node
6
will
send
the
DI
yes,
hello,
guys
and
want
to
join.
Please
somebody
send
me
a
dao
in
order
to
get
these
necessary
information
so
that
I
will
join
mr.
ripple
Network
and
if
you
can
see
all
the
neighbors,
we
are
resetting
their.
J
J
J
So
if
we
include
the
N
flag
plus
the
responsive
reading
in
the
bottom
subplot,
you
will
see
that
the
good
thing
here
is
that
we
don't
have
any
more
collisions
or
very,
very
low
chance
to
have
collisions,
because
the
nodes
now
will
respond
with
a
variable
timer
and
around
2
milliseconds.
So
what
happens
is
that
we
are
avoiding
the
collisions
and
by
avoiding
collisions
with
we
avoid
the
retransmissions.
So
this
is
the
response
spreading
functionality
here
next
slide.
Please.
J
So
we
have
n
flag
response,
printing
and
the
metric
container,
where
the
third
item
here
basically
filters
out
the
the
nodes
that
will
fulfill
the
rican
strain
that
is
specified
inside
the
DI.
Yes,
and
as
you
can
see
here
when
node
6
sense,
the
D
is
only
three
or
four
nodes
respond
with
the
IO,
which
is
sent
only
once
it
is.
They
are
sent
in
different
times
because
of
the
response
burning
next
slide.
Please.
J
So
here
we
have
to
flux,
the
no
inconsistency
and
the
type
flux.
So
basically
we
add
one
more
line
in
the
legend,
so
we
have
the
you
unicast
VI,
u
DX!
So
basically
what
you
see
is
that
at
around
time,
30
again
we
know
Six
Senses
da
yes,
but
because
we
it
sends
the
n
NT
flags
together.
It
will
receive
only
unica's
dieö
in
response
and
these
are
the
green
cross.
But
that's
what
you
seem
that
it
is
time
long
next
slide.
Please.
J
J
E
J
The
ultimate
we
have
no
inconsistency
unicast
in
the
response,
metal
container
to
filter
out
and
response
spreading
sort
of
the
nodes
will
send.
It
will
not
send
at
the
same
time,
so
to
avoid
the
collision,
C
response
and
in
the
bottom
figure
you
can
see
that
the
four
devices
that
they
are
transmitting
there
they
are
having
some
time
differences
differentiation.
So
we
are
avoiding
the
collision
in
response,
as
well
as
where
it
leaked
using
the
number
of
responses
in
the
area
next
slide,
please.
J
This
is
a
summary
of
different
flux
and
different
combinations
of
between
flux
and
options.
So,
basically,
what
you
can
see
here,
we
are
reducing
more
than
40%
the
number
of
di
your
responses
when
we
are
including
some
flux
and
options
inside
the
das
requests
when
we
compared
to
the
default
RP,
which
is
a
gray.
The
gray
line
in
the
bottom
next
slide,.
J
So
this
is
the
slope
where
I
would
like
to
initiate
the
discussion
from
my
side.
My
next
to
do
is
to
include
the
use
case,
what
was
given
by
Pascal,
so
that
I
will
have
the
zero
one
with
all
four
huge
cases,
question
to
the
working
room
will
be.
You
have
more
use
cases
in
mind
and
the
second
question
will
be:
is
there
an
interest
on
this
draft?
I
J
My
last
word
and
then
discussion,
so
the
last
part
will
be
whether
it
will
be
okay
to
continue
with
the
is
modification
draft,
or
you
want
to
have
one
single
solution
with,
because
I
know
that
the
lighting,
the
idea
information
cuz
some
some
some
some
options
for
the
D
is
as
well.
So
whether
you
want
to
have
one
single
draft
or
two
separate
in
this
subject.
That's
all
from
my
side
and
I
will
be
happy
to
have
a
nice
discussion.
F
The
first
thing
is
when
we
include
these
metrics
specifically
filter.
We
will
have
less
answers,
but
we
might
be
missing
the
right
answer
right.
So
the
drawing
seem
to
indicate
like
oh
by
magic,
the
best
parent
and
the
guy
on
the
Left
would
answer.
Since
we
don't
I
mean
unless
the
metric
is
a
rank
and
we'd
say
but
I'm.
Sorry,
your
rank
is
better
than
this,
but
since
we
don't
know
where
to
start,
I
mean
sorry
hard
to
do
something
like
that.
Actually,.
B
E
B
F
You
say
you
say
we
need
to
expect
magic
from
this
right,
just
I'm,
saying
files
and
see
if
the
metric
is
you
need.
This,
no
must
be
better
than
this.
You
might
get
only
the
responses
for
guys
who
are
even
farther
than
you
are
from
the
route
just
because
the
other
one
close
to
you.
So
we
have
to
be
careful
just
saying
days.
Yes,
we
have
less
answer,
but
we
might
be
missing
the
right
answer.
Well,
the
filter
is
wrong.
Then
we
might
not
have
any
answer
at
all,
for
instance,
better
break.
F
If
you
don't
have
any
idea,
but
what
the
right
run
cares
to
start
with.
Then,
if
you
ask
for
too
low
and
too
far,
then
everybody
will
answer,
the
filter
will
be
useless.
If
you
ask
for
something
too
high
to
any
other
route,
then
there
will
be
nobody
because
you're
far
from
there,
so
it's
very
hard
to
know
what
to
start
with
it's
a
chicken.
So
I
was
just
that's
just
you
know.
Food
I
felt
I'm,
not
saying
you
have
no
answer
for
that.
I'm.
Just
saying
hey,
it's
not
easy!
F
So
so
that's
that's
point.
One
point.
Two
ones
about
the
use
case
for
which
is
where
the
node
sends
the
node
reboots,
but
but
say
it's
a
blackout
and
you
are
not
talking
about
your
electrical
meter
and
so
so
now,
after
the
blackout,
power
comes
back
and
all
the
meters
blew
it
and
guess
what
that
pretty
much
the
same
time,
they
would
send
this
message.
Obviously
it's
not
something
you
want
for
the
reasons
Joe
just
has
explained
like
they
will
collision
with
one
another
or
making
a
cacophony.
F
J
F
F
We
doesn't
see
what
we
call
the
same
thing,
for
instance,
if
ice
sound,
III
reboot,
and
they
want
to
send
it
this
if
I
trick
all
that,
but
I
consider
that
any
Dias
that
I
would
receive,
which
would
have
the
same
flags
as
mine,
would
be
the
same
as
mine.
That
I
would
basically
cancel
out
my
own
transmission,
because
I
got
ideas
from
somebody
else,
so
so
using
it's
basically
explaining
and
saying
what
what
the
D
s
is,
which
is
the
same
as
yours,
and
the
lowing
trickle
to
cancel
them,
consult
your
own
selling.
F
That
could
probably
help
a
lot
in
the
same
fashion.
If
the
some
answers
and
yo-yos
are
the
result
of
a
dis
answer
and,
for
instance,
we
have
a
bit
in
the
yo-yo
which
says
I'm
announcer
to
adea's,
then
somebody
would
also
answer
ideas.
But
you
know
everything
is
trickled
in
in
ripples,
so
that
di
use
themselves,
art
record,
even
if
you
she'll
put
your
trickle
value
back
to
the
minimum.
F
Still
it's
trickled,
meaning
that
if
anyone
answers
for
the
same
question,
you
should
not
have
to
answer,
but
we
need
to
explain
that
this
is
actually
a
duplication
and
that
it's
it
makes
it
unnecessary
for
you
to
answer
what
I'm
saying
is
basically
trickle
bass
very
naturally,
this
delayed
answer
at
least
some
of
it
yeah
I'm,
not
saying
we
don't
need
the
flags,
no
all
right,
I'm,
saying
hey.
We
need
to
look
at
exactly
what
trickle
does
how
we
can
use
it
yeah
and
is
it
in
a
format
so.
B
Can
I
just
take
three
points
as
a
response?
Pascal
the
reason
work
we
have
the
delay
spread.
Option
is
precisely
in
the
case
where
we
request
a
unicast
answer,
because
if
it's
the
multicast
answer,
it
will
be
tricking.
We
you
don't
need
to
have
this
spreading.
Add
it
on
top
of
trickle,
but
if
you
request
a
unicast
and
then
it
will
be
sent
immediately
quote-unquote
and
that's
why
we
have
the
spelling
option.
B
B
And-
and
the
other
stuff
is,
when
you
say
trigger,
will
do
the
job
and
you
will
eventually
get
to
the
area.
But
the
point
is
that
issue
have
acquired
networks
that
has
a
very
long
maximum
period
for
Strickland.
Then
it
said
the
the
expectation
for
the
time
you
have
to
waste
before
you
join
the
network,
and-
and
so
there
are
some
cases
where
the
technician
is
not
going
to
wait
so
that
vacation
of
times
that's
why
we
want
to
send
the
DEA.
So
the
question
just
waiting
to
get
the
CIO
through
trickle,
then.
F
We
have
two
choices
right.
We
said
when
we
only
discussed
that
we
said
that
could
be
a
button
on
the
meter
or
something
and
you
press
it
that
would
force
the
VSS
apply.
So
that
was
one
we
had
the
question
I
have
was:
could
we
could
we
trick
what
we
we
have?
So
we
don't
even
need
a
button
right.
A
button
is
additional
complexity.
B
F
If
trickle
could
not
only
decide
that
you
don't
have
to
sound
message,
but
also
that
you
should
not
even
reset
the
bit,
do
you
mean
or
something?
So
you
know
you
receive
these
tears
as
we
receive
it
today,
so
today
this
multicast
one.
So
today
we
would
reset
to
him,
and
everybody
would
do
that,
but
know
there
is
something
it
will
fire
the
DAO
first
and
because
of
say,
you
know
the
disc,
a
number
of
the
areas
which
are
fired
during
this
short
window
of
time.
The
others
also
reset
did.
F
We
said
there
are
3
min,
so
they
will
do
him
in
time
too
often
time,
but
I
was
like.
Oh
could
we
say
I'll
go
back
to
where
you
were.
Don't
don't
stay
with
this
more
I
mean
you
had
your
chance.
So
just
you
know,
the
bottom
line
would
be
to
play
the
trick
or
with
the
shroud
duration
just
once
and
then
resume
the
normal
duration.
K
F
F
B
B
B
E
F
F
H
So
so
my
I
mean
was
set
to
12,
which
means
send
the
initial
here
after
four
seconds
and
in
four
seconds
roughly
four
seconds
between
two
and
four
seconds,
and
then
the
dis
is
configured
to
be
sent
randomly
between
zero
to
four
seconds.
Now.
If
my
note
density
is
sufficiently
high
like
like
I,
have
a
root
and
then
there
are
like
at
least
10
or
12
neighbor
neighbors
to
the
root,
and
this
neighbor
starts
sending
this
all
the
time
and
my
I
mean
is
for
four
seconds.
H
Then
there
is
a
good
chance
that
the
root
might
keep
resetting
the
DIY
worship
trigger
timer
and
won't
send
the
T.
However,
so
this
is
I
am
assuming
the
proposition
here
will
solve.
That
issue
is
that
is
that
assumption?
Correct
I
mean
I
know
for
sure
that
we
had
a
discussion
before
and
we
targeted
this,
my
I
mean
is
12.
By
default,
the
I
mean,
as
per
RFC.
6550
is
256
millisecond
with
256
millisecond.
H
It's
very
difficult
to
get
this
problem,
but
if
I
increase
the
I
mean
to
four
seconds
and
keep
the
default,
this
timer
to
four
I'm
in
zero
to
four
seconds,
in
which
case
I
have
this
problem.
It's
sufficiently
higher
with
higher
note
density,
I
want
to
keep
an
eye
mean
which
is
slightly
bigger
inside.
So
that's
when
I
face
this
problem.
F
H
So
the
problem
is
before
the
dialectical:
timer
get
start,
someone
is
sending
keep.
If
someone
will
keep
sending
this
and
there
is
no
way
the
root
will
know
not
to
reset
it
still.
So,
at
least
for
the
first
time,
at
least
for
the
first
first
till
the
first
D
I
was
sent
there.
The
root
should
not
be
be
sitting
it's
tickle
time.
F
F
H
H
F
Because
the
problem
is,
everybody
stands
at
this
at
the
same
time
and
nobody
considers
that
the
dis
served
by
the
neighbors
is
a
duplicate
of
the
dis
that
they
would
like
to
sell.
So
if
everybody
wants
to
send
a
multicast,
this
I
mean
they
should
be
sensitive
to
the
fact
that
all
the
sort
of
matters
are
rebooting.
Next
to
that,
so
trickling
sending
yes.
So
if
for
metals
are
on
me
centered,
this
multicast
I
would
expect
that
that
will
be
the
ions.
They
don't
need
to
send
multicast.
H
J
F
F
So
my
case
was:
it
is
natural
to
capture
all
those
deaths
asking
for
multicast.
It
looks
natural
to
consider
that
your
neighbors
doing
the
same
thing
is
to
placate,
and
so
you
can
sell
your
own
and
I'm
back
to
okay.
Should
I
then
do
with
I
equal
x
to
or
do
I
do,
I
go
back
to
IMAX
or
something
which
is
even
for
the
distant
I'm.
If
I
want
salad,
yes
and
I
see
people
around
me,
son
India's,
isn't
it
didn't
have
the
date
yet
so
it
should
I.
B
H
I
F
F
F
F
F
B
B
K
B
F
H
H
B
B
J
B
On
that
and
I
was
one
to
a
schedule
just
so.
This
use
case
draft
was
to
draw
attention
from
the
working
group
to
the
use
cases
so
that
we
discussed
use
cases
first
and
then
we
discuss
solutions,
and
so
because
we
had
a
hard
time
drawing
attention
draft
for
the
use
case.
Who
draw
it
and
I.
Think
now,
with
the
attention
of
all
the
people
who
care
and.
B
F
This
is
about
aging,
so
the
use
case
is
also
interesting
in
that
it
not
only
says
what
the
situation
you
want
to
solve,
but
also
what
kind
of
this
you
want
to
sound?
Then,
if
you,
if
we
start
spitting,
if
you
are
in
this
situation,
you
send
this
this,
then
it's
more
than
the
use
case
draft.
Is
it's
actually
your
specification?
F
B
F
B
F
B
F
F
Bottom
line
is
either
we
make
it
so
that
whatever
is
done,
it's
not
too
bad,
and
that
includes
what
I
mentioned
about
you
know
sending
the
disk
and
we
call
to
the
DS
of
your
neighbor.
So
you
don't
sound
the
same
so
this
would
make
you
know
my
situation
less
harmful,
mostly
harmless,
but
if
we
don't
solutions
like
that
and
there
are
ways
of
using
bits
which
can
really
arm.
We
also
must
specify
not
to
set
the
bits
in
that
situation
and
that
becomes
data
track.
F
And
that's
when
I
say,
if
you
we
start
having
this
and
then
it's
not
an
appendix,
but
we
need
to
describe
the
situation
where
you
must
not
send
those
bits.
Basically
in
that,
so
the
use
cases
helped
a
lot
sorting
out
the
various
situations
and
compiling
them.
Obviously,
when
you
send
say
must
almost
not,
you
also
need
to
describe
something
more
more
generic,
then
a
technician
with
a
blue
coat
and
burrowed
pressing
a
red
button,
not
a
yellow.
B
F
F
J
J
Hundred
percent
clear,
whether
it's
going
to
appendix
or
the
main
of
the
solution,
rough,
because
if
it's
going
to
the
main
of
the
solution,
rough
I'm,
not
saying
that
I'm
against
the
thing
is
that
the
protocol,
who
proposes
something?
But
if
we
are
proposing
during
the
use
cases,
sections
some
argumentation.
J
F
If
there
is
a
chance
that
the
network
fully
reboots,
then
you
should
be
doing
this.
For
instance,
the
list
could
be
a
setting
of
trick
or
could
be
whatever,
and
then
you
could
say
in
the
appendix
there
is
Appendix
A,
for
which
actually
gives
you
an
example
of
such
a
case
for
the
whole
network.
Reboot,
and
that
example,
is
my
smoker
ate
after
a
black
hole.
J
F
Cannot
apply
to
the
specific
case
of
a
smart
crate
right.
The
most
applies
to
situation
when
there
is
a
global
reboot
of
the
network.
People
would
say:
Oh
an
IT
network,
never
izotova
reboot,
because
people
will,
you
know,
put
the
battery
in
London.
Yes,
but
if
you
have
a
smart
grid
network,
yes,
you
have
at
the
moment.
F
B
The
question
is
much
into
which,
because
now
we
have
the
question
at
the
bottom
of
the
slide,
which
is
where
do
we
put
the
new
modifications
to
this
right
now?
There
are
some
modifications
in
the
lighting,
the
aircraft
and
some
others
in
the
distance
education
soft.
So
how
do
we
do?
We
merge
those
two
as
a
big
one,
the
lighting,
the
air
you
and
all
the
modifications
to
this
or
do
we
keep
them
separate
the
initiations
to
this
earth
of
Li
Jing
Bao
into
a
separate
one,
so
it's
better
contained
and
make
sure
they're
synchronized.
F
H
H
H
A
A
D
F
B
D
F
Okay,
so
I
published
in
March,
but
the
change
changes
are
very
minor
same
prime,
as
I
was
mentioning
earlier.
We've
we've
had
a
lot
of
of
work
on
the
number
of
documents
like
this
fall
and
now
we'll
turn
down
and
use
a
free
Pineville.
Well
now
we
have
passed,
we
just
guessed
about
doing,
ripple,
v2
and
so
the
questions
a
group
is
to
have
this.
F
This
consistent
and
overall
view
of
where
we
want
to
go
with
ripple
v2,
and
if
you
remember
that,
in
my
view,
and
at
least
the
main
difference
between
ripple
and
Whipple
meet,
you
is
that
ripple
was
designed
for
I
would
say
one
use
case
with
a
fixed
configuration.
You
have
it
four
times
and
it
will
never
change
and
and
the
network
lives
exactly
like
that
voice,
laughs
and
now
we're
saying
a
ripple
is
actually
successful.
F
There
will
be
a
brown
field,
there
will
be
upgrades,
there
will
be
changes
in
the
configuration
and
then
we
can't
bring
down
the
network
and
do
a
flag
day
every
time.
Something
changes
not
acceptable
in
the
use
cases
where
actually
repo
is
deployed.
So
we
need
to
think
Republican
in
a
way
that
will
allow
to
improve
the
network.
Operation.
Add
features
our
capabilities,
while
the
network
is,
is
deployed,
active
and
even
very
large
possible,
and
we
will
not
be
able
to
afford
having
to
bring
down
the
whole
network
and
restart.
F
So
this
is
the
reason
why
we've
got
the
keys.
This
is
also
the
reason
why
we
have
this
draft,
because
this
draft
is
all
about
making
sure
that
the
configuration
information
not
just
a
configuration
option,
but
all
that
we
could
call
the
configuration
information
that
would
node
needs
to
have
to
operate
report
correctly.
All
this
configuration
is
up
to
date.
F
F
Okay,
so
for
now,
I
mostly
got
one
big
review
and
actually
quite
questions
in
that
review
from
home
and
solely
for
the
type
when
your
neighbor,
who
I
typed
es
twice.
So
basically,
how
kind
of
asked
about
the
mechanism
that
we
have
with
this
FCSS
Sony,
our
CSS
is
sequence
counter
right.
It's
there
to
basically
give
a
numerical
value
to
a
database
which
has
all
the
possible
information.
That
means
a
number
which
is
actually
split
in
the
number
of
options
in
report,
but
all
those
options
there
are
five
of
them.
F
F
If
you
don't,
you
ask
that's
why
we
have
these
these
things,
but
when
the
complexity
comes,
is
I
tried
to
make
it
so
that
you
could
get
a
compressed
value,
a
compressed
content
of
a
second
option
if
it
has
not
changed
like
before
a
CSS
going
from
five
to
six
say
the
ripple
configuration
option
has
not
changed
and
it
would
be
possible
to
just
sign
it
and
say
here
is
the
TR
CSS,
where
it
was
not
changed?
And
if
you
have
this
a
CSS,
you,
okay,
that's
the
sort
of
thing.
F
And
so
this
way
the
node
needs
to
maintain
the
database
of
for
each
option,
which
is
the
FCS
s,
why
it
was
last
changed
and
that
enables
to
to
basically
make
the
DI
use
much
smaller,
because,
instead
of
having
this
problem
of
sending
something
in
multiple
yo-yos,
as
we
discussed
earlier
here
option
that
does
not
change.
We
can
just
said
said
the
four
bytes
which
abbreviated
it
remains
same
thing
as
last
time
with
these
are
CSS.
F
If
we
don't
do
that,
which
means
we
increment
the
SS
of
each
option
each
time
the
our
CSS
is
incremented,
so
considering
all
the
options
that
are
separate
fifths
of
the
configuration,
but
but
all
the
options
together
on
the
configuration,
then
that
means
that
each
time
is
some.
You
upgrade
your
CSS.
You
need
to
resend
the
full
content
of
all
the
options
like
you
need
to
dump
your
folding
state
database
if
you
like,
and
it's
a
choice
because
versus
more
code
and
that's
where
we
are
now.
F
The
draft
allows
for
that,
but
the
thing
is
we
would
need
probably
a
capability
to
leave
in
the
bran
field.
We
have
some
notes
on
that
and
some
don't
because
if
if,
for
instance,
most
of
the
nodes
tuple,
but
one
node
does
not
for
that
node
each
time
the
our
CSS
is
incremented,
we
need
to
resend
everything
for
the
other
nodes.
We
can
send
the
abbreviated
versions
of
things
which
have
not
child
to
know
and
that's
a
capability
game.
So
we
basically
have
three
choices
here.
F
F
H
Actually,
I
prefer
option
3
here,
the
reason
being
so
you've
mentioned
that
all
the
options
must
be
sent
when
RCSS
increments.
The
artists.
This
will
increment
only
when
either
of
the
option
actually
changes
and
if
you
see
prefix
information
or
configuration
option
or
any
of
those
options
are
sort
of
sort
of.
You
know
the
they
rarely
change
in
my.
So
what
I'm
trying
to
say
is
even
if
they
change
and
if
the
whole
whole
set
has
to
be
sent
again,
it's
it
still
as
compared
to
the
overall
control
over
edits.
H
You
know
it's
still,
it's
still
a
very
less
overhead.
Now
when
it
comes
to
complexity,
if
you
see
you
when
it
comes
to
complexity,
maintaining
and
our
CSS
in
context
for
every
such
configuration
option
is.
This
is
a
big.
It's
quite
a
quite
a
fighter.
Quite
a
big
deal.
Now
you
mentioned
third
option,
which
is
which
is
having
capability
to
indicate
that
now
what
I'm?
What
I
feel?
Is
that
eventually
having
the
capability
it?
It
does
not
you
know
it.
It
actually
further
complicates
the
specification.
It
does
not.
H
You
got
the
point
and
one
more
thing
is
one
more
thing
is
late,
let's
say
in
the
future:
at
some
point
of
time
we
we
see
that
such
kind
of
thing
is
necessary
because
not
because
of
the
current
set
of
options
that
we
have,
but
because
of
some
new
options
that
gets
introduced
later
on
I
feel
it
can
be
an
add-on
later
on.
If
really
we
have
that
kind
of
basis.
Yes,
that's
what
I
think
you
know
I
feel
it
the
bed.
H
F
F
So
because,
unless
you
said
there
will
be
one
option,
changing
very
rare
and
it's
going
to
be
very
rare.
So
when
there
is
one
there
is
only
one,
and
so
you
can
compress
all
the
others
basically
again,
so
that
means
that
will
always
fit
within
a
yo-yo
and
apart
from
the
very
beginning
of
the
network,
but
the
rest
of
the
time
it
will
fit.
So
we
we
save
a
lot
of
energy
and
bandwidth
by
doing
this
or
never
done
here
you
even
if
they
are
not
so
often
they
are
still
are
many
of
them.
F
So
you're
the
trade
off
he's
here
bad
with
a
lineal
energy
in
the
one
hand,
and
having
to
maintain
500s
instead
of
one
and
usually
usually
when
we
design
ripple,
we
tended
to
accept
a
little
bit
more
complexity,
like
I,
mean
5
sequence
hunters,
instead
of
one
having
to
sound
more
bytes,
so
you're
just
allocating
for
the
river.
So
I
want
to
point
out
that
that's
what's
happening,
and
so
I
would
really
really
like
to
do
to
consider
how
many
times
the
options
are
sent,
even
just
because
somebody,
you
know,
asks
for
something.
F
Yeah
much
that
who
presents
so
that's
the
big
question:
I'm,
not
sure
that
we
have
the
real
view
of
often
do
the
options
would
be
sailed
and
what
it
costs
to
to
not
be
able
to
relate
some
up.
So
that's
what
the
GI
yoga
now
I
believe
that
your
command
only
applies
to
the
I/o,
because
there
is
also
the
Dao
I
will
also
allows
you
to
say
you
know
all
those
storing
more
downs,
the
past
tense.
You
learn
the
non
stirring
everything,
essentialist
I'm,
still
saying
these
saves
on
everyday,
oh,
when
they
are
periodic
right.
F
F
H
Yes,
so
sorry
for
the
doubt,
for
the
DAO
part
also,
is
it
is
it?
Is
it
not
the
same
point
like
if
options
are
not
getting
changed?
For
example,
in
most
of
the
cases,
the
target
option
is
going
to
remain
the
same
as
as
well
as
the
transit
information
option
might
also
remain
the
same.
So
the
the
point.
What
I'm
trying
to
understand
is
won't
a
single
other
CSS
work
in
that
case,
as
well.
F
H
B
F
H
You
know
here
will
also
require
the
information,
so
these
the
the
are
CSS
in
context
you
have,
the
configuration
option
has
to
be
maintained
in
a
database.
It
has
to
be
persistent,
you
cannot
afford
to
lose
it,
and
things
like
that.
You
know
on
on
reset
again
so
I'm
wondering
all
those
implications
will
also
be
there
like.
F
The
option
in
persistent
storage,
in
which
case
we
would
you,
would
have
to
keep
the
CSS
for
it
for
each
option.
If
you
don't
keep
the
options
in
our
system
storage,
then
you
lose
the
option
a
on
the
associated
our
CSS
together,
and
that
means
you
don't
have
it
and
you
start
like
from
scratch.
Starting
from
scratch
is
okay
right,
so
I
don't
see
that
there
is
a
personal
storage
game
here
you
saw
everything
is
in
Boston
to
the
right
you're,
nothing,
that's
the
bottom
line
the
option.
F
The
difference
is
you
have
one
LCSs,
five
different
values
for
five
different
options:
all
you
have
the
same
value
for
all
the
options,
and
if
you
have
five
different
values
for
the
five,
then
you
can
you
can
in
lied.
When
you
ask
unique
asked
for
some
options,
you
can
alight
the
others.
Where
has
fueled?
F
H
F
H
F
Agreed
it's
more
like
if
you
lose
well
I'm,
not
convinced
in
need
of
fashion
right,
you
seem
to
be
convinced
on
you.
You're
further
than
I
am
because
I'm
not
convinced
either
way
right.
My
intuition
was
it's
always
better
to
say
bad
word
and
then
five
counters,
and
now
you
come
back
to
me
insane.
Maintaining
five
constants
instead
of
four
one
is
a
lot
more
complex.
I
did
not
see
that
coming
to
be
honest
because
for
me
one
contour
of
five
contours
is
just
like.
If
you
bites
and
I'm
not
saying
I
said.
L
K
L
Good
clarification,
but
but
I
think
what
we're
trying
to
say
is
that
it's
the
rate
at
which
we
have
to
erase
flash
blocks
and
if
you
have
to
erase
a
flash
block,
then
everything
in
that
flash
block
can
be
updated.
So
so
it
actually
doesn't
so
so
fewer
it's,
not
how
many
counters
we
have
or
how
many
values
we
have.
It's.
The
frequency
of
updates
that
that
and
the
effect
of
failing
to
update
before
you
crashing
before
you
update,
is
the
issue
as
well.
The.
F
Else
you
assess
covers
the
whole
thing.
The
question
you
remember
the
FCSS
of
the
last
change
of
the
given
option,
which
means
that
you
maintain
five
values
and
they
are
stored
with
the
options.
So
when
you
go
to
flash
should
write
the
option
in
flash,
then
you
write
your
CSS
for
it
in
flash
as
well
at
the
same
block.
F
So
I
don't
see
that
as
as
representing
any
difference,
unless
that
makes
your
block
bigger
of
sudden
it'll
sustain
one
page,
awesome,
otherwise
I
don't
see
difference
then
again,
maybe
you
need
to
be
very
clear
on
the
mechanism.
The
mechanism
is.
There
are
five
options
today.
If
you
change
any
of
the
MS
ESS
goes
up,
bang
bang
on,
but
did
so
each
option
as
the
last
change
of
CSS
associated
to
it.
This
option
was
change
versus
is
5,
and
this
option
was
generous.
Yes
and
6.
E
F
F
You
could
do
Saturdays
and
say
the
last
I
know
is
five,
and
so
all
you
all
you
need
to
get
is
the
others
right.
So
what
the
dis
will
contain
is
the
one
that
changed
for
six
and
four
seven
and
the
other
one
would
be
just
the
abbreviated
versions.
Saying
oh,
it's
not
that
we
don't
sell
it.
It's
just
that
it's
still
with
the
same
old
value
of
five,
and
so
so
that's
what
it's
used
for,
but
in
terms
of
storage,
it's
you
just
have
to
store
it.
Where
you
store
the
option.
F
F
I
H
Pascal,
it
is
possible
to
theoretically
analyze
how
much
of
course,
savings
we
can
get
assuming
an
update
frequency
of
X
for
all
this
five
configure
five
configuration
option
now,
even
if
we
assume
that
either
of
this
Co
five
configuration
option
gets
updated
once
per
week,
which
is
which
is
which
is
not
possible.
You
know
once
per
week
is
too
frequent
an
update
for
such
kind
of
configuration,
so
it
is
possible
to
theoretical
and
allies
how
much
of
control
over
its
savings
we
can
get
and
and
I'm
pretty
sure
that
we
will
be.
I
H
F
H
Sure
so
yeah
I
would
at
the
least
what
you're
suggesting
is.
If
we
keep
the
a
oh,
we
will
keep
it
completely
decoupled.
We
will
still
have
an
option
of
having
only
our
CSS
or
a
configuration
option
of
our
and
our
CSS
per
configuration
option
right.
So
if
that
is
the
case,
then
okay,
I
guess
then,
as
an
implementer
I
have
an
option
to
choose
to
use
either
of
it
right.
F
H
F
You
mean
the
question
that
this
all
the
set
of
all
the
options
that
are
about
configuration
5.
Yes,
yes,
yes,
there
is
an
a
CSS
fall
5
right.
There
is
a
single
counter,
that's
easier.
F
F
F
Oh
and
the
DI
yo
will
either
be
split
in
several
and
of
all
the
options
in
for
if
some
of
them
would
be
replaced
by
a
oh
ho,
which
makes
it
yeah
you're
much
smaller
and
avoid
splitting
in
several
messages
and
now
you're
saying:
okay,
I
prefer
of
several
messages
and
of
all
the
options
info
they're
having
some
of
the
options
replaced
by
the
a
oh.
That's,
that's
what
you're
telling
me
right
something!
F
H
Not
correct
because
yeah
yeah,
so
you
mentioned
the
case
of
multiple
di
use
on
the
dl
fragmentation
that
can
happen,
not
fragmentation
in
character.
You
might
be
eventually
there
might
be
a
reason
to
send
multiple
d
is
if
we
are
not
able
to
compress
well,
but
these
multiple
di
URLs
will
get
cheese
sent
only
if
there
is
that
change
in
either
of
the
configuration
option
right
and
that's
when.
E
F
H
Not
about
the
new
option,
so
it's
a
it's
a
new.
This
new
option
plays
a
part
for
every
individual
configuration
option
so
in
the
in
the
RAM.
I
need
to
have
this
configuration
option.
Has
this
version?
It
has
a
direct
implication,
although
on
the
RAM
right,
as
well
as
on
the
flash,
so
so
so
that
that's
what.
F
F
Like
I
said,
you
might
be
right
on
dancing,
you're
wrong.
I
still
need
to
be
convinced.
I've
not
found
that
the
argument
that
says.
Oh,
it's
so
complex
tons
of
code
that
that
we
don't
want
to
save
the
energy
and
whether
I
see
that
the
code
is
not
very
complex.
There
is
no
static,
Ram,
prime
or
I.
Don't
see,
I
still
see
the
power,
but
but
it's
a
group
size
go
with
it.
I
mean
I'll
go
either
way
because
I'm,
not
convinced
by
the
arrow,
is
either
not
convinced
is
aware.