►
From YouTube: IETF99-HACKATHON-20170716-1400
Description
HACKATHON meeting session at IETF99
2017/07/16 1400
https://datatracker.ietf.org/meeting/99/proceedings/
B
D
A
E
E
D
D
D
A
D
D
F
F
If
we
get
to
the
end
and
I,
don't
have
yours
well,
we'll
figure
out
what
we'll
do
your
presentation
at
the
end,
there's
probably
like
three
or
four
of
you
at
least
that
I
know
of
who
want
to
either
present
off
your
own
laptop
or,
if
you've
sent
me
an
email,
I'll
I'll,
try
to
access
that
and
we'll
get
you
at
the
end,
but
we'll
start
with
all
the
ones
that
are
in
in
the
github
and
I'll
just
kind
of
go
through
them
in
order
to
try
to
keep
things
simpler
for
me,
but
but
that
gives
you
an
idea
of
when
you're
up
so
I'm
just
going
to
start
from
the
top
and
work
my
way
down,
and
if
you
can
come
on
up,
you
can
use
my
laptop
and
it's
a
handheld
microphone
and.
F
F
F
G
H
H
So
was
the
lotta
thing:
is
a
draft
that
came
out
one
year
ago
in
the
transport
area
is
basically
a
marking
scheme
which
provides
a
super
simplification
over
the
usual
diffserv
model.
The
Assumption
here
is
that
traffic
belongs
to
either
one
of
these
two
classes
slow,
sensitive
or
latency
sensitive,
and
that's
it.
You
mark
the
application
endpoints
mark
the
traffic
appropriately
in
the
network.
Do
the
right
things:
internal
scheduling,
the
difficult
part
for
this
is
convincing
application.
The
endpoints
that
marking
they're
encrypted
traffic
is
provides
an
advantage.
So
we
had
the
discussion.
H
This
discussion
during
the
manual
workshop
and
yakov
off
and
the
discussion
was
told
because
we
didn't
we
had
no
data.
So
the
main,
the
main
reason
behind
this
this
effort
is
to
provide
the
automated
testing
facility
to
to
actually
collect
the
data
and
and
and
and
decide
whether
whether
the
mechanism
actually
works
in
practice
and
provides
the
advantage
it
promises.
So
yesterday,
I've
worked
on
this
and
what
I
had
was
already
I
have
prepared
the
flow
mix
and
a
queueing
implementation.
H
In
practice,
this
effort
was,
we
kept
a
couple
hundred
lines
of
T
shark.
Okay.
Our
later
was
a
first
for
me
and
I'm
sure
the
superpowers
of
it
a
lot
and
then
today,
I've
worked
an
acme
star.
This
is
a
trap
that
we
had
just
adopted
in
the
Acme
working
group.
It.
It
is
an
extension
to
the
Acme
protocol
through
to
me
the
certificate
renewal,
so
Acme
provides
issuance
automate
additions
and
this
one
extends
a
thing
to
actually
automate.
Also
the
renewal
parts
in
general.
This
might
be
a
useful
distinction,
but
what
we?
H
B
H
Needed
to
to
allow
cloud
service
providers
or
Citians
to
obtain
certificates
for
https
delivery
that
bear
the
content
provider
name
instead
of
their
own.
So
today
I've
worked.
This
was
synchronized,
so
we
had
a
participant.
It
was
remotely
providing
the
Acme
star
implementation
as
an
extension
to
Boulder
and
self.
What
why
while
I,
was
working
on
the
other
side
of
protocol,
the
telluric
beats
implementing
them
from
scratch.
Now
we
have
the
two
bits
together.
H
H
G
J
Hi
guys
welcome
to
our
segment
outing,
ipv6
or
net
confrontation
over
the
next
three
minutes.
I'm
going
to
present
our
goal,
which
was
when
you
take
a
DSL
broadband,
allow
allow-
and
rather
it
usually
does
not
or
have
the
capabilities
like
signal,
routing
or
Netcom.
So
we
were
wondering
whether
it
will
be
possible
to
add
this
kind
of
functionality
and
how
hard
that
would
be
also
they
wanted
to
have
a
hands-on
experience
on
segmenting
technology,
so
those
were
our
motivations.
So
what
did
we
do?
We
took
Holland
router,
which
is
very
similar.
J
You
can
get
from
your
broadband
provider
and
it
was
tp-link
and
we
took
VD,
which
is
away
a
firmer
for
a
router
which
you
can
use
to
extend.
The
capabilities
of.
We
took
the
patch
from
David
liebe
room
and
it's
an
signal
routing
enabled
page
which
will
allow
to
tour
of
this
functionality
on
Leedy.
Then
next
thing
we
did.
We
took
people
all
the
young
model,
which
is
storing
information.
We
need
in
order
to
achieve
signaling,
and
in
this
particular
case
it
is
storing
ipv6,
prefix
and
associated
segment
routing
list.
J
J
So,
as
you
can
see
on
the
right
side-
and
there
is
our
topology,
we
have
a
client
which
is
acting
as
a
controller
issuing
net
conf
comments
to
router
a
router.
A
and
a
router
B
are
two
routers
using
liddie
and
all
those
technologies.
I
talked
about
integrity
together,
so
they
understand
that
conf
and
then
understand
segment
routing
or
what
we
are
doing
is
that
at
the
beginning
we
are
pinging,
this
bottom
interface
from
router
a
and
it
is
taking
the
green
path
like
that.
J
Now,
at
some
point,
we
issue
from
the
client,
the
Netcom
common
and
the
new
routing
path
gets
installed
on
the
router.
A
and
the
packet
will
immediately
start
to
take
the
upper
path
and,
as
you
can
see
here
on
the
left,
the
new
route
got
installed
and
then
you
can
also
see
these
two
TCP
dump
TCP
dump
screens
which
are
taking
packets
from
these
two
particular
interfaces,
and
you
can
see
if
you
watch
carefully
that
actually
one
and
the
other
interface
are
receiving
packets
mutually
depending
on
whether
or
not
a
signal
routing
is
live.
J
J
Well,
there
is,
there
is
a
spring
working
group
and
there
is
signal
routing,
so
b2
about
segment
outing
is
discussing
and
we
found
very
how
to
implement
it
in
the
router,
integrating
all
those
technologies
together
and
also
over
I
think
around
peace
c3po.
So
the
short
answer
is
yes.
It's
spring,
there
are
three
people.
Thank
you.
F
K
C
L
L
So
we've
took
all
these
together
and
specified
to
JSON
formats
that
unify
censorship
reports
from
each
of
these
disparate
data
sources.
We
ran
one
of
the
crawlers
in
Turkey
to
see
the
difference
between
countries,
something
that
we
intend
to
explore.
More
difference
between
censorship,
both
through
complicit
actions
by
companies
and
also
independent
filtering,
possibly,
is
an
extension,
but
mainly
this
validates
the
specification
RFC
770
to
5.
So
if
we
move
on
this
is
the
JavaScript
polar.
L
M
N
L
C
O
C
R
S
Yeah,
no,
this
is
fine
yeah,
so
we
had
a
large
crew
and
three
new
ITF,
a
hackathon
participants,
I'm
very
happy
with
it
and
with
nine
18
people.
Your
strength
is
also
a
weakness
out
there,
many
people
with
many
ideas
and
to
work
on.
So
we
come
up
with
a
number
of
themes,
but
we
worked
on
quite
different
subjects.
First,
I
want
to
stress
this.
S
Education
was
actually
also
a
good
part
of
yesterday,
so
we
really
talked
a
lot
with
each
other
and
with
the
new
ideas
about
what
we
have
been
doing
in
the
past
and
why
we
are
heading
to
so
that
I
want
to
stress
that
was
important
and
we've
spent
some
time.
There
work
some
drafts,
sorry
so
important.
S
Open
has
not
and
intersect
chantilas
extensions,
long
name
for
the
draft.
It
allows
TV
check,
authentication
they
all
chain
authentication
in
one
shot.
So
the
transport
here
is
not
dns,
it's
TLS,
so
you
get
all
the
dns
key
material
in
one
shot.
You
can
do
the
validation
by
the
application,
for
example
in
web
browsers
very
useful,
and
we
worked
on
the
seediness
capture
format
and
retrofit
DNS
tab,
which
is
kind
of
a
monitoring.
S
Well,
it's
a
login
format
for
the
internals
of
resolvers
and
name
servers,
and
we
want
to
read
to
fit
insidious
which
is
and
accept
well
and
broadly
supported,
or
it's
in
new
draft,
but
gots
a
lot
of
attention
and
support
here.
So
that's
good
work
here
that
somebody
already
told
you
there's
in
case
k,
rollover
11th
of
October
remember
the
day.
S
Actually
it
started
already
fifteen
days
ago,
no
ten
days,
so
the
new
key
has
been
introduced
on
the
left
of
July,
but
there's
some
extra
testing
of
all
the
software
has
been
already
tested,
so
we
extended
the
testings
also
in
the
venous
test
to
dexhart
their
support
for
not
resolver
and
for
inbound.
These
are
the
timeline.
B's
can
skip
that.
S
What
has
been
forgotten
with
all
the
standardization
in
the
ITF
is
that
we
have
some
automated
update
of
the
church
tanker
of
the
route
for
DNS
operators,
the
chef
cook,
but
we
forgot
to
specify
what
to
do
for
our
end
users,
the
end-to-end
principle.
We
forgot
completely
about
that.
So
how
do
we
deal
with
applications
that
has
to
catch
up
with
the
new
twist
anchor?
So
we
came
well.
The
RFC
fifty
eleven
doesn't
work.
Unfortunately.
S
Well
it
worked
for
the
operators,
but
not
for
the
end
users,
so
we
came
with
the
zeroconf
heck,
so
it
allows
applications
which
depend
on
the
Dena
sec,
etcetera
for
Dane
to
do
kind
of
an
automated
first
anchor
update.
I
won't
go
in
all
the
details,
but
it
will
check
if
there's
updates
on
the
trust
anchors,
if
necessary,
it's
brutal
out
and
can
do
the
whole
authentication
locally
in
the
application.
Some
research
innovation,
those
standards
here
you
experiment
with
opportunistic
TTL.
S
G
S
This
or
yes
this
weekend,
thanks
going
back,
actually
yeah
everything
by
individuals,
different
individuals.
So
there
was
one
two
people
working
on
the
IDI,
the
new
elliptic
curve,
Edward
curve.
One
person
works
on
the
Q
type,
two
people
on
the
T,
less
extensions
for
the
for
this
work.
Also,
three
four
people
also
the
new
people
who
are
working
on
the
deck,
are
testing,
etc.
So
it
was
a
good
energy
of
a
lot
of
work
done.
The
issues
you
found
in
50:11
do
divide
them
up
and
you're
gonna
bring
in.
S
U
Thank
I'm
a
diminishes,
confirm
identity.
We
develop
a
core
implementation
of
thoughts,
I
eat
his
name
go
dogs,
so
most
of
people
in
this
room
could
be
not
familiar
with
that.
So
what
is
that
all
adult
if
the
doors
open
great
Signori,
as
you
saw,
that
is
assimilating
for
standardization
and
automation,
of
the
scenery
for
DDoS
protection?
So
it
is
a,
it
is
to
say,
ask
for
help
from
a
victim
side
to
upstream
provider,
so
it
is
inter-organizational.
So
inherently
it
includes
those
authorization
and
authentication
in
specification
and
vice
versa.
A
protection
status.
U
So
we
opened
the
call
yesterday,
we
just
often
called
as
an
open
source
project,
so
the
this
repository
already
have
a
amount
of
code,
but
it
is
now
open.
So
what
was
developed
in
hackathon?
We
made
the
code
called
easy
to
be
deployed
in
various
environments,
so
we
made
a
kind
of
calm
composure
to
each
services
and
also
we
refined
the
complex
part
of
configuration
and
also
we
clarify
the
documentation,
because
it
is
a
good
for
newcomers
to
this
field
and
also
we
made
a
demonstration
of
a
user
scenario.
So
it
is.
U
The
demonstration
is
on
a
portable
docker
environment,
so
everyone
can
do
that
demo
in
every
year,
so
the
demo
includes
the
triggering
of
the
whole
routing
from
picketing
side.
So
this
is
very
interesting
demo,
so
it
attracted
for
people.
We
showed
them
a
demo
in
this
weekend
as
an
extra
step
we
are,
it
is
now
Oasis,
so
we
directed
ups
to
the
various
deployment
scenarios
and
we
ready
to
keep
going
on
the
implementation
of
a
cube,
trust
and
take
feedback
to
the
specification
and
also
a
you.
U
V
V
Okay,
so
my
name
is
Ron
to
town.
I
will
present
you
something
we
have
done
on
cheek,
so
sche
is
pronounced
cheek
and
it
has
been
done
with
the
milk
bottle
from
a
range
fever
via
from
university
of
windows
air.
I
run
from
an
e-value
from
my
cleo
and
myself
from
a
university,
I'm
tnt.
So
we
work
on
a
new
category
of
network
new
family
of
network,
which
is
called
LP
1,
and
that
pill
has
been
designed
for
Internet
of
Things,
because
we
have
a
very
good
energy,
very
energy-efficient
example.
V
A
device
like
this
can
ruin
for
many
years
on
the
battery,
and
there
is
also
a
long
range,
so
you
can
cover
cities
or
our
countryside
very
easily.
So
that's
a
very
good
technology
for
Internet
of
Things
on.
We
have
different
technology,
for
example,
laura1,
cig,
Fox,
Nirvana,
UTI,
so
next
era,
but
exist
on
cover
viz
requirement.
The
problem
is
that
we
don't
have.
If
you
want
to
write
an
application,
you
have
to
develop
the
application
for
this
technology,
so
at
v--
IETF.
V
We
have
a
solution
when
we
have
different
technology
and
we
want
to
run
application
on
a
single
way
is
to
use
IP
and
to
put
application
on
top
of
IP.
The
problem
is
that
when
you
look
at
LP
one
technology,
you
don't
have
a
lot
of
resources.
For
example,
you
can
send
only
one
percent
of
the
time.
If
you
look
at
one
technology
like
cig
fox,
you
can
send
only
message
of
12
bytes,
so
it
meant
that
to
adapt.
V
Ip
to
this
kind
of
constraint
is
very,
very
difficult,
and
that's
why
at
IDF
we
have
developed
technology
chic,
but
allowed
to
compress
for
this
specific
environment,
IP,
UDP
and
co-op.
So
we
arrived
to
via
cotton
with
to
implementation
of
chic.
One
has
been
done
by
Mt
Atlantic
and
it
was
running
and
Python
in
the
device
like
this
one,
on
JavaScript
on
the
core
network
and
a
creo
brings
its
own
implementation
running
in
goal,
and
so
what
we
have
done
during
this
in
Akutan
is
first
to
test
interoperability
between
all
these
implementations.
V
So
we
fix
some
some
bugs.
We
also
move
the
application
int
application.
We
also
include
loja,
SiC,
Fox
and
lower
one.
So
now
we
have
devised
that
application
that
can
run
on
these
all
these
kinds
of
device
and
we
for
range.
We
also
test
the
Python
implementation
in
the
core
network.
So
it's
what
we
we
have
done
on
the
result
of
the
accattone
is
that
now
we
can
test
the
solution
on
multi
technology.
We
also
resolved
some
problem.
We
have
in
the
draft
we
so
in
the
graph
is
on
the
padding
management.
V
Q
G
G
W
As
you
may
know,
mile
is
a
working
group
working
for
recent
information
exchange
and
this
work
group
has
completed
to
define
format
for
each
different
exchange.
So
we
wanted
to
have
a
kind
of
converter
that
can
be
useful
for
the
Styx
community
and
also
I
or
the
committee.
So
in
this
huxton
work
we
did
a
to
work.
First,
Rockies
are
building
a
converter
for
our
converter
between
iodine
and
sticks,
and
other
work
is
a
extension
to
our
own
situated
role
that
can
output
IOT
alert.
W
So
these
two
works
are
there
and
all
works
are
now
available
on
github,
and
this
is
the
status
of
the
output
for
the
ionic
converter.
We
are
able
to
provide
the
POC,
so
if
we
take
a
look
at
the
draft
draft
has
a
lot
of
example
of
our
dev
document,
so
those
examples
are
supported
by
at
all,
but
I
know.
Different
sticks
have
a
lot
of
classes.
It
was
impossible
to
implement
everything
in
these
two
days.
So
in
these
two
days
we
were
only
able
to
cover
the
convert
converter.
W
X
X
On
the
left
side,
you
see
the
right
networking
system,
integrating
the
our
own
network
stack
called
GRC,
and
the
challenge
was
to
adapt
it
to
the
northbound
interface,
so
called
sock.
Here
you
can
see
it's
it's
con.
It's
just
the
duplicated
name.
Sock
is
a
generic
network
API
and
the
southbound
interface
so
called
net
F
is
a
generic
heart
abstraction
layer
for
network
devices
such
as
8
or
250,
got
4
radios.
X
So
what
if
I,
what
what
kind
of
efforts
have
to
be
targeted
here?
First,
there
must
be
an
automatic
way
to
download
the
open,
WS
n.
Stick
as
a
software
package
and
included
into
riot
then
include
past
mag
files
need
to
be
adapted
and,
as
I
already
mentioned,
the
adoption
of
native
interface
for
the
southbound
adoption
and
the
northbound
interface
sock
yeah,
and
this
is
what
we
worked
on
the
last
two
or
one
and
a
half
days.
X
We
didn't
finish
everything
yet,
but,
however,
we
had
the
cooperation
of
different
open
source
projects
so
guys
from
the
open,
WS
n
community,
external
community
members
from
the
riot
team
and
me
internal
members,
core
developers
of
the
riot
operating
system.
We
integrated
and
migrated
open
access
protocols
develop
a
few
guys,
such
as
six
open
insta,
and
we
used
standard
cooperative
coding
platforms.
So
in
this
case
it
was
github
yeah,
and
that
was
it.
Thanks
for
your
attention.
X
Partly
I
guess
who's
suited
of
you,
honestly
I,
don't
know
who
student
and
whose
research
assistant
and
who's
just
community
member
I'm
a
research
assistant
at
the
University
of
Applied
Sciences.
X
Y
Hi,
so
sorry,
so
my
name
is
jenkum
I'm
from
Hamburg.
This
is
Chris
from
Basel.
This
is
Claudia
and
Palace
also
from
Basel.
So
we
worked
on
the
year
and
then
well,
it's
better
ICN
experience
in
the
IOT
there's
already
ICN.
For
those
who
don't
know
I
see-
and
this
is
an
information
centric
networking,
it's
a
paradigm
shift
and
moving
away
from
addresses
and
on
the
network
layer
like
we
have
an
IP
away
and
giving
names
to
to
everything.
So
we
have
names
of
the
network
layer.
This
is
ICN
and
what
we
do.
Y
What
we
did
is
yeah
so,
like
I,
said
the
this
ICN,
the
IOT,
but
what
we
want
to
do
is
compress
had
a
packet
headers
because
currently
and
in
ndn
formats,
this
is
one
of
these
well.
Implementation
for
ICN
approach
are
kind
of
nervous,
so
they
are
type
length
value
based.
So
this
is
your
always
have
variable
length,
names
and
header,
packets
and
in
the
IOT
we
have
link
layers,
have
really
small
size
didn't
used.
So
we
think
that
this
phone
scale
and
what
we
do
is
we
have
an
with
what
we
did
this.
Y
We
yeah
design
the
name
compression
for
these
packet
formats
and
we
also
omitted
fields
in
the
packet
formats
like
it's
done.
For
example,
it's
in
the
sixth
open
for
sixth
open
in
IP
networks,
the
IOT,
the
results
that
we
yeah
gain.
This
I
mean
we
did
prototype
implementation
in
CC
and
night,
which
is
an
implementation
with
a
very
lightweight
in
fermentation
of
this
ICN
approach
from
the
Basel
guys-
and
we
did
this
in
the
right
environment.
So
I
am
a
core
developer
Brian.
These
are
core
developers
of
CC
and
I.
Y
So
this
was
a
really
great
match
to
me.
Today
we
have
a
basic
or
a
simple
calculation
that
we
achieved
with
our
compression,
so
we
have
like
a
twenty
eight
byte
packet,
which
is
sent
in
the
NBN
well
as
an
end
and
the
end
packet
in
the
IOT,
and
we
compressed
that
by
50%
so
down
to
12
bytes,
and
this
is
just
a
simple
packet.
So
we
already
yeah
drafted
up
a
really
small
text
which
we
are
going
to
put
Submit,
maybe
tomorrow,
as
a
0-0
graph
to
the
IC
energy,
which
also
works
on
this.
Y
And
so
what
are
the
next
steps?
We
want
to
finalize
the
implementation
of
this
compression
format
or
scheme
and
further
work
on
down
the
draft
and,
furthermore,
we
are
going
to
meet
again
so
in
the
next
week's
at
other
hackathons
and
probably
at
the
right
summit,
which
is
soon
yeah,
and
we
think
that
we
can
can
yeah
give
back
to
the
community
and
submit
the
drafts
and,
let's
see
what
happens
then,
after
receiving
the
feedback.
So
any
questions.
Q
Q
Y
Y
Z
Ok,
so
now
I'm
going
to
introduce
about
our
project,
it's
we're
working
on
SD
applications
for
microwave
radio
link
to
RIT
under
the
model,
so
our
table
is
just
a
right
there
and
the
project
is
deformities
of
members
from
the
following:
5
5
campaigns
and
university
from
Karen,
sweet,
Midori
and
microwave
link.
It's
a
technology
which
is
widely
used
for
backhaul
poppers,
and
nowadays
there
are
more
than
60%
of
base
stations,
work-
Islamic
livings.
So
it's
a
very
necessary
to
find
a
young
model
for
the
microwave
link.
Z
This
is
what
we
do
in
ITF,
C
camp
group
and
master
journey.
There
was
a
design
team
set
up
to
work
on
this
work
through
reading
and
we
have
to
draft
the
framework
drop
the
end,
the
young
draft,
those
were
adopted
in
Sikkim,
and
we
also
have
a
article
publishing
unit
of
general
and
now
coming
into
the
Nara
days.
We
feel
it's
the
right
time
to
have
the
high
connectivity
to
validate
the
model.
We
have
the
finding
in
C
camp,
and
this
is
what
we
plan
to
do.
Z
We
are
the
two
key
points
is
that
first
of
all
you
need
to
implement.
Is
that
a
young
model
within
the
audio
controller
then,
based
on
that
away?
We
do
have
developed
to
use
case
wine
store
energy
efficiency,
which
is
to
save
the
power
consumptions.
Another
one
is
to
control
the
frequency.
The
purpose
is
to
control
the
interference
between
the
radio
links.
Z
The
result
is
that
we
have
a
successful
implementation
as
a
young
model,
so
we
cooperated
with
our
idea
of
drafts
together
with
the
open
source,
so
we're
and
then
we
developed
two
applications
and
it
was
successful.
We
have
run
the
test
or
you
could
say
the
fig
on
the
right
and
we
also
a
successful,
brings
and
Industry
vendors.
Even
we
are
competitors
are
in
the
market
and
also
the
University
into
this
project,
and
there
was
three
of
us
is
first
idea.
Z
Another
part
is
that
our
exam,
the
model
might
need
additional
attributes,
for
example
the
aesthetics,
and
we
also
thinking
about
to
use
rest
IP
I
instead
of
the
rest
count
for
the
next
step
is
wires
to
cooperate
with
some
other
packet
in
an
interface
and
also
to
improve
to
the
model
with
more
aesthetics
tributes
and
finally,
well
also
thinking
about
to
contribute
back
to
the
audio
project,
and
that's
one
I
would
love
to
thanks
for
the
team
members
without
your
efforts
to
make
this
haxon
project
to
happen.
Thank
you.
Q
T
G
G
N
It's
alright,
so
John
and
I
did
a
proof
of
concept
on
delivering
multicast
in
in
a
network
simulating
his
Comcast
network,
so
we're
using
the
emts
a
new
standards
track
specification,
RFC,
7450
and
the
spring
segment.
Routing
work
is
the
way
he
would
like
to
transmit
in
his
network.
This
is
our
topology
and
basically
I've
got
the
the
problem.
N
N
John
has
some
technical
difficulties
with
with
doing
native
multicast
within
his
network,
but
he
is
interested
in
in
transmitting
it
multicast
over
the
cmts,
but
what
he
wants
to
do
is
have
it
done
by
segment
routing
within
the
network.
So
what
we
wrote
was
was
I.
Have
a
sender.
I
have
no
multicast
connectivity
from
my
sender,
but
I
opened
an
AMT
tunnel
to
a
gateway,
and
this
is
using
the
open
source.
Amt
project
I've
been
working
on
and
then
within
his
network.
N
The
ingest
point
I've
act,
the
relay
so
that
I
have
a
unicast
kind
of
connection
to
the
home
router,
but
the
unicast
connection
to
the
home
router
as
it
gets,
joins
and
that's
receiving
the
traffic
natively.
It
will
push
it
out
with
with
the
proper
shim
so
that
it
will
get
routed
through
his
network.
The
way
he
wants
it
to
so
it's
really
that's
the
basic
idea
so
for
this.
For
this
weekend,
I
ran
it
on
open
wrt
for
the
fur
and
the
Gateway
on
open
wrt.
For
the
first
time
that
went
fairly
smoothly.
N
Some
of
the
setup
was
a
new
thing
and
then
the
shim
layer.
It
was
a
new
implementation,
so
it's
just
John
and
I,
and
it's
not
done
yet.
His
boxes
got
delayed
on
the
shipping.
That's
but
we'll
be
setting
it
up
in
the
lounge
and
finishing
it
up.
Hopefully,
this
week,
yeah
I
I
don't
have
a
whole
lot
to
talk
about
it.
Any
questions
so.
N
Oh
yeah
yeah,
so
so
we
want
to
ingest
from
the
sender
just
once
and
then
have
it
sent
throughout
his
or
maybe
a
small
number
of
times
and
have
it
sent
throws
network
with
the
fan-out
happening
within
his
network.
So
it's
gonna
get
sent
once
to
each
cmts
within
his
network
and
then
we
don't
want
it
to
have
to
come
from
our
sender
all
the
time.
Yeah.
F
AA
Thank
you.
So
this
project
was
about
another
hour,
IOT
related
thing,
where
you
might
have
many
devices
that
sleep
most
of
their
time
and
you
have
to
wake
them
either
periodically
or
by
radio
messages.
And
when
you
do
that,
we
get
the
problem
that
that,
if
you
have
the
last
hop
router
are
here
in
the
middle
that
every
packet
that
it
forwards
to
the
sleeping
device,
D
may
wake
up,
will
wake
up
that
that
device
and
then
drain
its
battery.
AA
What
we
had
before
the
hackathon
started
was
some
proprietary
software
on
the
control.
Node
see
the
the
node
that
generates
the
packets,
the
some
sort
of
routing
software
on
our
that
uses
some
radio
protocol
to
wake
up
the
the
USD
which,
which
was
an
entire
black
box,
and
what
we
did
was
that
we
kept
this.
AA
G
AB
Alright,
for
those
of
you
that
don't
know
quick
is
a
integrated
security,
transport
and
application
level
protocol
all
sort
of
rolled
into
one.
So
the
working
group
actually
tried
to
organize
around
this
hackathon
as
a
Interop
event,
so
we
got
a.
We
made
a
wiki
page
up
there.
That
said
what
aspects
of
the
current
draft
specifications
we
were
interested
in
testing
Interop
on,
and
you
know
we
had
a
wiki
page
yay,
but
we
had
zero
tested,
interoperable
implementations,
although
we
knew
people
were
starting
to
do
that.
AB
We
didn't
know
what
happened
when
we
showed
up
showed
up
here
today,
but
one
to
find
out
whether
oh
four
was
reasonable
or
just
kind
of
crazy
talk.
So
we
came
and
we
had
somewhere
around.
You
know,
18
folks,
representing
clients,
servers,
CD
ends
wipers
and
that
kind
of
thing
come
and
we
managed
to
create
that
interrupt
grid.
So
that's
you
know,
clients
and
servers.
You
know
in
each
direction.
Greet
and
the
letters
down
at
the
bottom
are
certain
aspects
of
what
we
were
testing,
so
that
was
all
blank
and
empty.
AB
When
we
started
that's
pretty
great,
we
came
up
with
one
new
issue
for
the
working
group
to
consider
that's
in
the
CrazyTalk
area
and
a
whole
bunch
of
editorial
certifications
along
the
way.
You
know
the
Interop
problems
we
had.
We
were
able
to
sort
of
amongst
ourselves,
largely
related
to
versions
of
TLS
great.
If
we
got
a
working
group
final
call
on
that
one
to
come
out
and
the
various
configurations
there.
Q
So
what
are
you
doing
with
the
broken
ones?
Are
the
stuff?
That's
not.
Are
you
gonna
continue
this
for
the
other
DNC
stuff
everywhere
and
yeah.
AB
So
actually,
we
expect
to
keep
doing
this.
This
format
and
the
number
of
the
things
with
tests
will
grow
in
scope.
So
the
wiki
page
was
what
we
call
the
first
implementation
milestone
and
we
will
create
a
second
and
a
third
and
the
intention
is
to
have
interrupts
in
front
of
every
working
group
meeting.
We
do
so
that's
the
three
times
a
year,
IETF
fundraise,
but
we're
also
doing
interim
so
we're
going
to
incorporate
essentially
an
Interop
hackathon
like
event
at
the
front
of
each
one
of
those
as
well.
AC
Hello,
everybody,
so
this
is
the
last
presentation
or
another
one
all
right,
so
this
project
was
about
provisioning
domains,
captive
portals
and
having
fun
with
this
bunch
of
people
from
different
companies
and
including
to
ITF
newcomers.
So,
as
you
know,
hosts
today
are
luteum
that
you
can
connect
to
Wi-Fi
wired.
They
can
connect
to
smartphones,
and
even
when
they
connect
through
a
single
interface,
they
can
have
multiple
ways
to
actually
access
the
networks,
who
is
p's,
VPN,
cetera
et
cetera.
AC
So
when
an
application
actually
comes
in
as
I
want
to
connect
you
there
well,
the
networking
stack
is
like
okay.
Well,
I
do
best,
I
can,
but
they
cannot
really
do
much.
So
really
what
the
provisioning
domains
or
pvd's
are
about
is
to
identify
the
different
ways
you
have
to
connect
to
the
internet,
not
only
the
interface
but
the
way
to
connect
to
further
internet
and
their
capabilities
and
the
particularities
of
each
way
you
have
to
connect
to
the
internet.
AC
So
the
first
step
step,
one
is
to
put
in
the
Urata
advertisement
the
ipv6
water'll
advertisement,
a
new
option
which
includes
fully
qualified
domain
name
and
that's
how
you
identify
the
PVD
you
are
connected
with
that
is
specified
in
a
draft.
It's
the
draft
in
entire
working
group,
and
there
is
also
the
point
that
you
may
want
to
send
multiple
arrays
on
a
single
interface
in
order
to
support
multiple
provisioning
domain
on
a
single
link
and
that's
in
a
different
raft.
AC
The
step
two
is
to
actually
retrieve
from
a
pbd
server,
JSON
blob,
which
includes
all
the
information
that
you
want
to
know
about:
capabilities
of
capabilities
of
the
PVD
you're
connected
to
so
you
get
one
HTTP
blob,
it's
PVD
JSON
from
each
server
and
then
the
project
I
mean
part
of
the
project
was
to
put
within
each
of
those
JSON
object.
Uri
to
know
if
there
is
a
captive
portal
and
if
there
is
what's
the
URI
to
connect
to
the
captive
photo.
So
what
we
did
was
first
to
implement
that
in
open
wrt.
AC
Actually
in
later,
it's
obviously
open
source.
We
implemented
it
in
the
router
advertisement
daemon
as
well.
The
possibility
to
have
multiple
arrays
sent
on
a
single
link,
as
well
as
the
user
interface,
to
configure
all
that,
so
you
can
configure
it
with
CLI
or
or
the
UI
Luci.
Then
we
implemented
a
server
I
mean
implemented.
No,
we
configure
the
server
to
be
able
to
deliver
the
PDG
JSON
objects
and
a
very
basic
captive
photo.
Also,
you
would
have
multiple
pvd's,
so
multiple
captive
bottles
in
our
demo.
AC
Maybe
you
will
see
it
at
the
bits
and
bytes
for
whether
you
are
smart
on
them
and
you
can
access
different
access
depending
on
that
there
was
support
in
linux
like
well.
Actually
we
did
an
interrupt
with
open,
wrt
and
realized
that
what
we
did
in
linux
was
not
really
good.
So
we
fixed
Wireshark,
we
fixed
array
DVD
and
we
fixed
the
linux
patch
that
we
did.
AC
You
can
see
a
screenshot
of
Wireshark
I,
don't
have
time
to
go
through
this
and
we
implemented
a
new
ziwei
kind
of
web-based
UI
to
be
able
to
decide
what
pbd
you
want
to
use
with
firefox,
and
this
is
kind
of
a
continuation
of
Nakasone
we
had
one
year
ago
already
and
which
was
using
a
let's
say,
simpler
way
to
do
it
less
mature.
So
this
is
a
continuation
of
what
we
did
one
year
ago,
India
chasm.
AC
Finally,
we
had
as
well
some
iOS
support
with
the
capability
ability
to
pass
the
option,
as
well
as
some
UI.
Here
you
see
to
retrieve
the
information
from
the
pvd's
and
display
to
the
user
on
the
iOS
platform
and
finally,
actually
we
get
support
from
needs
as
well
to
put
some
information
in
the
in
each
information
based
finally,
very
quickly.
Some
feedback
from
people
in
the
room
that
we
are
going
to
integrate
in
the
draft
and
time
is
out
okay,
ten-second
late.
Thank
you.
Q
AC
AC
AC
F
Ok,
so
there
are
several
people
who
actually
before
we
go
to
yours,
but
some
people
sent
me
slides
that
they'll
still
be
presenting
from
my
laptop,
so
I'm
gonna
try
to
hit
the
ones
where
you
are
still
presenting
off
of
mine.
The
hackathon
picture,
sockem
team-
you
guys
here
and
the
idea
is
I
just
put
this
up
and
all
you
need
is
this
picture
right.
F
AD
Thank
you
so
so,
for
the
last
couple
of
days,
we've
been
exploring
something
that
we,
we
call
a
vulnerability
assessment
scenario,
so
it
starts
with
effectively
a
vendor
or
security
researcher,
publishing
a
vulnerability
bulletin
and
then
taking
that
information
enriching
it
with
a
query
that
you
could
use
to
match
that
against
software
inventory,
to
determine
if
one
or
more
endpoint
devices
are
vulnerable,
given
the
information
that's
available
in
that
and
that
bulletin.
AD
So
what
you
see
here
is
the
rough
architecture
that
that
we've
been
building
out
over
the
last
couple
of
days.
On
the
left
hand,
side
you
see
a
rolly
server.
The
roli
server
is
is
based
on
a
mile
specification
that
we've
been
developing,
which
is
really
just
a
profile
of
Adam
pub,
and
it
effectively
allows
us
to
serve
up
arbitrary
resources
that
that
can
support
various
security
automation,
applications
on
the
right
hand,
side.
You
see
an
endpoint
that
was
the
target
of
our
assessment,
the
endpoint
that
for
which
we
were
collect
software
inventory
turns
out.
AD
In
this
case
it
was
in
a
bun
device
and
then
we
leveraged
one
of
the
the
specifications
we've
been
developing
in
sac
I'm
called
Schwimmer,
which
is
software,
inventory
messages
and
attributes.
So
what
Schwimmer
allows
us
to
do
is
it
allows
us
to
transport
to
collect
the
software
inventory
and
transport
it
to
a
server.
So
the
second
box
to
the
right
is
the
server
that
the
the
software
inventory
was
transported
to
the
the
box
immediately
to
the
left
of
that.
AD
The
the
second
one
from
the
left
is
where
we
we
leveraged
a
product
that
allowed
us
to
query
the
software
inventory
information
from
the
strong
TNC
server
to
query
the
the
vulnerability
detection
information
from
the
Roli
repository.
So
it
act
as
an
Orchestrator
to
do
that
work
and
it
allowed
us
to
actually
perform
an
assessment
to
match
the
vulnerability
description
to
the
software
inventory
that
was
collected.
To
do
that,
we
used
something
called
the
called
oval
the
opened
vulnerability
and
assessment
language.
So
we
were
able
to
get
all
of
that
working
together.
AD
But
in
doing
so
we
we
found
a
few
things.
So
we
identified
a
persistence
problem
and
our
rolling
implementation,
and
it
turns
out
that
we
could
post
resources
to
the
server,
but
if
we
tried
to
update
those
resources
they
weren't
getting
properly
updated.
So
that's
something
that
we
need
to
fix,
but
overall
we
were
happy
with
the
way
the
Roli
implementation
was
performing.
AD
The
the
search
functionality
also
worked
quite
well.
The
lessons,
though,
that
we
learned
was,
it
was
very
difficult
for
us
to
to
develop
a
query
that
was
actually
useful
in
in
in
querying
the
software
inventory
using
oval.
So
one
of
the
things
that
we
would
like
to
this
decease
AK
I'm
work
on
is
to
define
a
better
query
approach
that
would
allow
a
some
arbitrary
data
store
to
be
queried
I.
So
that's
something
that
we're
going
to
look
at.
AD
AD
AE
AE
So
this
is
the
posture,
so
I
turn
SF
is
for
on
that
of
security
services.
Framework
er
such
as
the
Faculty's
comes
in,
and
then
we
can,
depending
on
the
policy,
a
cadet
the
power
to
a
buyer.
If
the
fire
think
this
package
should
be
invested
more
such
as
Iraq
filter,
so
it
validated
by
computer
and
then
the
Tonto,
our
gateway
and
port
to
destination
host.
So
this
is
I
NSM
framework.
So
this
time
hackathon
a
core
is,
we
are
implementing.
AE
The
fire
were
using
the
open
source
ideas,
IPS
open
source
code
name
is
ricotta,
or
so
we
implemented
the
Web
Filter
or
so
we
can
combine
a
fire
and
web
filter
using
accepted,
see
the
architecture
or
so
this
time
we
you
use
the
extensible
stylesheet.
The
languages
are
transformations
to
translate,
however,
young
policy
into
the
low
level
young
policy,
so
we
implemented
postings
this
we
can
so
we
proved
the
our
concept
I
to
NSF
framework
using
open
sources.
AE
So
this
shows
the
open
source.
Of
course,
based
on
that,
we
implemented
our
I
to
NSF
project
now
this
time
this
figure
shows
to
another
comparation
this
enterprise
network,
and
we
can
prevent
that
the
accessible
or
YouTube
during
the
long
time
so
based
on
the
port
number,
we
ask
we
prevent
access.
Also,
this
time
we
order
implemented
our
ideas.
AE
Dpr
dpi
are
based
on
the
content
based
the
security.
If
the
some
are
web,
the
content
includes
a
Facebook
YouTube.
We
can
are
built
out,
so
we
implemented
this
over
a
weekend.
So
these
figures-
this
is
network
topology.
So
if
this
access
is
some
Facebook
case,
we
can
filter
out
using
the
total
number
and
the.
Secondly,
based
on
the
content,
we
also
a
picture
of
a
web
filter.
So
if
this
is
relevant
us
on
Facebook
or
Google
or
YouTube,
we
can
compute
out
so
we
implemented
this.
AE
So
you
can
find
the
board
information
we
documented
and
uploaded
our
source
code
or
github,
or
so
we
have
a
demonstration
video
clip
in
our
YouTube.
This
is
the
dista.
We
learned.
Okay,
we
proof
concept
of
alternate
framework,
use,
open
source
host
and
also
we
validated
alternative
interface
design,
especially
parsed
polished,
NSF
net
function,
page
interface
for
fire
and
repeater,
or
so
we
prove
low
gain
young
based
data-driven
approach.
Okay,
thank
you
for
your
attention.
AE
AE
AF
AF
This
work
focuses
on
use
of
young
subscriptions
because
we
already
heard
that,
for
example,
strong
Swan
and
strong
teen
see
is
work,
flowed
for
vulnerability
assessment
in
the
scope
of
the
security,
automation
and
continuous
monitoring,
but
there's
also
this
huge
amount
of
predefined
stuff.
You
might
be
interested
in
for
security.
This
comes
from
ops
in
defined
via
yang
modules,
so
I
will
start
with
a
ridiculously
small
diagram.
So
I'm
not
sure
if
this
everybody
can
read
this.
This
is
a
lie.
AF
I
had
no
time
to
recreate
any
diagram,
so
I
will
just
say
that
there
is
the
lower
location
which
is
here.
This
is
Prague,
and
this
is
the
lower
boxes.
The
second
domain
we
created
and
then
we
have
multiple
international
domains
we
acquire
information
from
which
is
one
in
the
USA,
which
is
a
switch
that
provides
a
yang
sub
interface
and
it's
from
the
People's
Republic
of
China,
which
also
supplies
real-life
data.
We
got
flowing
here
and
separated
in
the
second
on
life,
so
both
are
using
Yang
subscriptions
the
left
side,
the
u.s.
AF
location
we
decided
to
use
unchanged
subscriptions
and
for
the
right
side
we
chose
to
do
periodic
subscriptions.
Both
are
viable
methods
to
get
subscribed
information
we
are
yang
push,
and
the
goal
here
was
to
create
Sokka,
so
called
components
that
are
collectors
which
are
these
boxes
here.
One
collector
taking
charge
of
PRC
one
collector
taking
charge
of
the
USA
and
then
because
the
pops
up
is
a
one
to
one
subscription.
AF
We
want
to
broker
this
vital
information
into
a
broker
domain
that
is
facilitated
by
the
XMPP
grid,
also
drafted
in
my
the
thing
is
that
I
have
a
lot
of
requirements.
We
did
not
address
yet,
and
this
is
the
first
time
we
had
actual
real
data
flowing
through
the
system.
Addressing.
This
is
just
an
excerpt,
some
of
the
requirements
of
the
second
workgroup,
for
example,
data
partitioning
in
regard
to
geographic
location,
which
has
actually
been
done
because
our
demo
spans
the
globe.
AF
So
we
are
able
to
partition
the
information
acquired
in
the
broker
domain
of
the
XMPP
grid
via
topics
and
subtopics
or
partitions.
We
composed
the
second
collectors,
which
are
multiple
functions.
One
is
a
subscription
to
the
yang
one
is
the
publication
to
the
XMPP
grid
according
to
specific
topics
and
content.
We
also
do
the
information
discovery,
so
there
Piaget's,
server,
XMPP
server,
can
identify
the
information
and
relay
it
to
the
specific
subscribers.
AF
Added
to
that
we
do
the
push
access,
which
is
g11.
We
do
actual
component
setups,
so
we
have
a
generalized
way
now
to
combine
functions
in
different
languages.
I
think
at
the
table
was
the
Python
C
Java
and
even
some
Ruby
glue
code.
So
we
are
now
able
to
pull
all
these
functions
together
in
a
single
component,
and
this
component
architecture
is
so
generic
that
we
will
now
have
basically
no
fish
at
all
for
the
next
second
phones
to
add
future
functions.
AF
P
All
right
so
good
afternoon,
so
I'm
brunoise
and
since
the
very
first
hackathon
we've
been
having
like
a
young
table
and
I'm
very
happy
about
that.
We
start
with
the
team
of
people
who
are
coming
specifically
for
the
I
mean
not
specifically,
but
someone
specifically
for
this
hackathon,
so
that
makes
me
happy
as
well
now
what
have
we
done
so
in
the
world
of
automation
and
operation
management?
We
have
to
understand
that
we
might
have
data
models,
but
is
not
enough.
P
We
need
to
have
the
to
chain
as
well
and
then
all
the
metadata,
along
with
those
data
models.
So
basically
this
is
what
we're
trying
to
do
with
this
yang
catalog.
It's
an
inventory
of
all
the
yang
models
from
the
ITF
from
different
CEOs
from
different
open-source
project,
even
from
vendors.
We
want
you
have
the
to
chain
over
there
and
we
want
to
be
able
to
search
you
see
the
impact
analysis,
etc,
etc.
So
it's
a
one-stop-shop
for
yang
models.
P
What
have
we
done?
I
like
this
theme
very
much
because
they
were
active
yesterday
we
had
an
AI
Tripoli
meeting.
We
show
our
tool
and
they
said
you
know
what
you're
showing
like
old
ethertype
there.
We
do
an
example:
ok,
I
just
call
the
guys
there
and
say:
ok
down
done
now.
You
see
on
the
right-hand
side,
give
me
only
the
latest
revision
of
the
draft.
Now
we
keep
the
old
ones.
P
If
you
want
to
see
the
history
of
what's
happening,
so
that's
a
typical
direction
of
the
team,
whether
another
one
receive
like
a
request
to
say.
Well,
you
know
in
yang
we
have
like
reg
X.
These
are
double
three
C
reg
ex,
but
POSIX.
It
would
be
great
to
have
like
a
kind
of
reg
ex
re
danger
to
see
if
the
pattern
that
we've
got
in
yang
are
working.
P
Fine,
okay
done
it's
they're
going
a
young
validator,
so
you
just
type
the
the
pattern
at
top
right
hand,
side
and
then
or
multiple
ones,
and
you
see
if
it
buddies
correctly
right.
So
this
is
pretty
useful
for
yang
with
your
designer.
While
some
of
the
two
are
good
for
yang
module
users,
we
created
an
API
for
this
catalog.
P
These
are
all
different
things
that
we
have
on
a
young
model
right:
okay,
the
name,
the
revision,
any
deviations
or
disorganization
maturity
level,
and
we
provided
that
for
BBF,
for
example,
because
the
maturity
level
and
BBF
is
different,
the
one
in
ITF.
So
what
else
we
have
did
compile
right?
What
is
the
error?
Who
is
the
contact?
P
Is
there
like
code
so
we're
trying
to
populate
all
of
these
direct
in
the
catalog
and
we've
been
writing
not
only
an
API
to
write
in
the
catalog,
but
to
read
it
so
typically,
we
put
all
the
non
extractable
metadata
from
BBF
in
the
catalog.
That's
a
great
way
to
inter,
inter
operate
with
some
SDS,
but
there
are
many
other
things
we
did.
P
We
had
a
request,
for
example,
from
chin
to
say:
can
I
get
all
the
yang
models
that
are
done
in
what,
in
good
black,
we
created
the
API
directly
to
carry
all
information
from
the
catalog,
including
all
the
metadata
done.
We
improve
the
validator
the
same
time.
We
work
an
integration
of
some
huawei
and
juniper
models.
These
are
Prophet
remodeled.
At
the
same
time,
we
improve
the
validator,
which
is
a
good
thing.
P
We
did
the
full
inventory
of
all
these
model,
the
full
compilation
and
if
you
see
Peter
there,
it's
at
iPod
right
we're
working
on
a
GUI
as
well.
The
last
time
we
worked
on
a
GUI
to
generate
some
code,
some
Python
code
to
go
and
discuss
with
a
router
or
switch.
The
feedback
was
yes,
it's
flash
based.
So
you
know,
maybe
not
so
we're
working
on
this.
We
were
supposed
to
get
this
integrated
not
done
today,
but
will
be
done
like
the
next
days.
We
worked
on
new
man,
one
two:
three.
P
We
improve
the
Yankee
later
that
you
could
go
and
just
get
post
your
draft
and
see
if
it
validates
we
work
on
it.
Ok
execute
the
detail
of
the
tree
type
form
you
needed,
but
yet
another
meta
data
we
need
to
have,
and
we
validated
all
the
models
from
math
and
improve
them,
and
the
next
steps
is
that
you
know
it's
not
for
us.
It's
not
like
with
two
days
hackathon,
it's
something
that
we've
been
continuing
to
do
for
a
long
time
and
we
will
continue.
P
F
Great,
thank
you
so
now,
I
believe
everyone
who
sent
me
presentations,
I,
think
you
got
a
chance
to
present.
Is
that
right,
Wow
amazing!
You
actually
got
it
right,
so
we
do
have
one
more
presentation
which
is
going
to
be
extra
special
because
there's
a
demo
and
everything,
and
so
it
can't
be
on
my
laptop.
R
AG
AA
AG
Hey
cool,
so
I'm
Tom,
Jones
I'm,
going
to
talk
about
what
we
did
at
the
the
NEET
team,
which
is
over
in
this
corner,
and
so
we
worked
on
quite
a
few
things.
We
did
some
bug
fixes
to
SCTP
iData
code,
which
is
a
draft
in
TS
vwg.
We
got
the
the
python
bindings
to
our
library
working
whoo
and
we
did
some
work
on
provisioning
domains,
as
you
saw
before,
and
we
did
work
on
our
web
RTC
implementation,
which
we
have
a
demo
of
at
the
end.
AG
So
what
is
neat
me
is
a
socket
API
replacement.
It
presents
a
single
callback
driven
API
to
the
network,
the
supports
load
of
different
transport
protocols
to
make
that
work.
It
offers
an
automatic
selection
for
transports
and
interfaces
and
offers
fallback
to
make
sure
that
everything
comes
together
and
works.
A
neat
is
a
tapped
implementation.
AG
So
this
weekend,
I
worked
on
adding
PVD
support
to
meet
for
me
to
offer
selection.
It
has
to
learn
things
about
the
network
and
all
of
the
selection
decisions.
We
do
are
informed
by
dynamic
policy.
The
dynamic
policy
is
fed
characteristics,
so
an
e
enable
system
can
get
these
characteristics
from
its
local
network
interfaces,
and
now
after
this
weekend,
we
can
look
up
over
HTTP
characteristics
fed
by
PVD
internet.
So
the
upshot
of
this
is
the
lease
enabled
application
can
automatically
for
free,
take
advantage
of
a
PVD
enabled
Network,
which
is
a
big
step.
AH
So
we
also
added
support
for
peer-to-peer
communication
in
particularly
web
RTC
for
need.
It
is
implemented
in
C,
so
we
have
a
native
rtcdatachannel
implementation
written
and
see
and
it
makes
use
of
the
unified
need
api.
So
you
can
write
a
web
RTC
application
in
the
same
way
as
you
write,
a
tcp
or
UDP
application,
and
it's
interoperable
with
any
web
browser
we
tested
so
far
except
safari.
We
just
did
not
test
yet,
but
we
plan
to
do
so,
and
I
would
like
to
give
you
a
short
demo.
AH
So
we
have
a
Raspberry
Pi
here
which,
in
the
res
reply,
has
an
attached
sense
hat.
Since
that
means
there
are
several
sensors
and
LED
display,
and
we
read
the
sensor
data
of
the
Agora
scope
and
we
displayed
in
the
browser
and
the
communication
is
done
by
a
web
RTC.
The
sensor
data
are
sometimes
a
bit
weird,
but
if
you
see
on
the
screen
as
I
move
the
PI
a
bit,
this
sense
of
data
that
the
graphs
changed,
so
it
works.
Q
F
F
F
So,
just
a
couple
of
things
before
we,
we
close
I'll
get
those
out
of
the
way
while
they're
doing
their
judging
so
I'm.
If
you
notice
that
the
t-shirt
I'm
wearing
it's
actually
from
the
first
hackathon
that
we
did,
it
was
in
Dallas
just
like
a
little
over
two
years
ago
and
I
remember
going
around
and
you
know
like
basically
begging
people
like
hey.
You
know
please
we're
trying
to
do
this
thing.
F
Can
you
show
up
maybe
lead
a
project
or
help
some
people
and
we
managed
to
get
I
think,
like
maybe
40
people
which
we
thought
was
great
and
we
thought
wow.
This
is
actually
pretty
cool.
We
should
continue
this
and
he
now
fast
forward-
and
here
we
are-
and
you
know,
I
think
everyone
had
a
good
time.
But
you
know
the
complaints.
I
heard
was
that's
too
hot.
It's
too
crowded
and
I
mean
it's
just
to
me.
It's
a
sign
of
phenomenal
success.
F
There's
almost
200
people
here
we
have
almost
as
many
project
presentations
as
we
have
people
total
at
the
start,
so
just
phenomenal
to
see
all
of
you
coming
out
and
the
hackathon
really
doesn't
work
unless
we
have
the
champions
volunteering
to
leave
the
projects
and
all
of
you
coming
and
still
crazy
to
me
that
we
all
spend
our
weekend
doing
this.
But
it's
fantastic
and
glad
that
you,
you
feel
passionate
about
it,
and
so,
thanks
to
all
of
you,
so
you
know
we
will
be
having
the
next
hackathon
in
in
Singapore
and
I
hope.
F
One
tough
thing
or
challenge
with
the
hackathon
growing
is
that
it
continues
to
cost
more
money
to
run
it
because
we
we
eat
and
drink
a
lot
and
take
up
space
and
breathe
air
and
use
a
lot
of
power.
So
we
are
looking
for
sponsors
and
we
actually
don't
get
any
sponsors.
This
time
we
got
seeing
a
few
sponsor.
You
get
the
privilege
of
like
putting
your
company
or
logo
up
there
and
you
can
show
it,
but
really
not
all.
Joking
aside,
we
do
need
some
some
sponsors.
F
Could
you
come
up
here
and
just
say
a
few
words
like
you
guys
didn't
get
to
know
Joe,
because
he'll
be
in
charge
of
this,
and
even
if
you're,
not
the
right
person
to
help
with
getting
to
sponsorship,
maybe
put
him
in
contact
with
people
at
your
company
who
are
and
we'd
appreciate
it,
but
turn
over
to
you.
Joe.
AI
Hello,
I,
don't
have
much
more
to
add
really
about
that.
I
mean
it
is
a
great
event
and
that
we
do
need
some
support
financial
support,
as
you
said,
as
well
as
the
participation.
If
anybody
has
an
employer
who
you
think
might
be
in
a
position
to
do
something
and
I
think
don't
presuppose
what
the
nature
of
the
contribution
would
be,
it
could
be
tools,
it
could
be
infrastructure,
it
could
be
hardware,
it
could
be
money,
it
could
be.
All
kinds
of
things
could
be
prizes
yeah
exactly
so.
AI
We
are
open
to
any
kind
of
support
at
all,
because
I
think
you
know
in
some
sense
a
diversity
of
different
kinds
of
support
is
better
than
just
a
lump
of
cash.
So
you
can
tailor
this
to
whatever
suits
your
organization.
So
if
you
do
have
any
thoughts,
you
want
to
talk
more
or
you
know,
either
at
the
meeting
I'm
here
till
Friday
or
any
other
time.
I
have
just
received
what
I
think
may
be
the
first
personal
email
address
at
IETF,
or
you
can
reach
me
at
Jabba
Lee
at
IETF
dot
org.
AI
F
D
F
I'm,
the
other
thing
I'm
always
looking
for
suggestions
and
help
to
make
things
better
in
the
future.
So
any
any
comments
and
I
don't
mind
some
some
real
criticism
as
long
as
it's
going
to
help
me
do
better
next
time.
What
did
you
guys
like?
What's
what's
getting
to
be
a
problem,
if
you
want
to
share
something
now,
that's
great.
If
you'd
rather
meet
me
for
our
beer
and
talk
that
way,
that's
fine
too,
but
does
anyone
have
anything
they'd
like
to
share
or
suggestions
or
or
gripes?
F
Maybe
you're
wondering
what
you
have
a
chance
to
win
so
just
to
speed
that
up.
Like
I
said
we
don't
have
much
in
the
way
of
prizes
this
time,
but
I
do
have
five
social
tickets,
and
so,
if
you're,
looking
for
a
social
ticket
and
you're
one
of
the
winners,
that's
something
you
might
want
to
consider
other
than
that.
It's
kind
of
warm
and
instead
of
the
shoes
you're
wearing
you
might
want
some
flip-flops
and
we
got
a
bunch
of
flip-flops
in
the
corner
over
here.
F
F
T-Shirts
too,
but
the
t-shirts
like
I
said
we
will
be
getting
they'll,
probably
just
be
here
tomorrow
and
as
soon
as
they
do
get
here
and
I
have
them
and
I
know
like
they're
in
hand,
I'll
send
something
out
to
the
hackathon
list
and
that
you
guys
know.
But
this
area
here
turns
into
the
lounge,
and
so
we
will
have
an
area
here
in
the
lounge
for
people
like
for
the
hackathon
to
basically
continue
right.
F
So
you
mean,
if
you
want
to
continue
to
work
on
your
projects
or
any
project
and
just
be
able
to
meet
and
talk
with
others
who
are
working
on
code.
We're
also
going
to
have
the
networking
equipment
that
that
the
Comcast
folks
have
brought
and
that
we
usually
have
it
bits
and
bytes
it's
going
to
be
here
on
the
lounge
in
case
you
wanted
to
take
advantage
of
that
we
can
spin
up
VMs
on
it.
F
C
F
It's
not
for
today
we
have
to
get
out
of
here
today,
anyways
so
tomorrow,
it'll
be
here
at
some
time,
so
we'll
let
you
know
when
that's
available
too
and
then
bits
and
bytes.
So
several
of
you
have
told
me
that
you
want
to
demonstrate
something
at
bits
and
bytes.
Assuming
that's
I.
Tell
you
what,
because
I'm
gonna
forget
things.
Send
me
an
email.
If
you
haven't
already
saying:
hey
I
want
to
demo
our
project
at
bits
and
bytes
and
the
way
it
works
is
bits
and
bytes.
F
It
starts
at
7:15
on
Thursday
and
runs
to
9:15,
so
the
doors
won't
open
until
7:15,
but
if
you're
demoing
you
can
get
in
there
like
around
6:30
6:45
to
do
any
setup,
you
need
to
do
so.
What
you
do
is
you
send
me
a
single
slide,
that
kind
of
summarizes
what
it
is,
or
at
least
announces
what
your
demo
is
and
then
you
can
come
and
set
up
whatever
you
need
on
the
table
and
hopefully
I'll
be
able
to
get
enough
space
for
everyone
who
wants
to
present
to
do
just
that.
E
O
O
O
O
Four,
four
four
best
new
work:
we
have
the
HTTP
HTTP
error
code
for
51
people
thanks
for
coming
and
doing
that
and
engaging
the
human
rights
stuff
in
the
hackathon,
which
is
kind
of
novel,
I,
think
best
University
work
we
have
I,
think
we've
given
this
to
them
before
is
the
I
to
NSF
people
wherever
they
are.
Thank
you,
I
to
NSF.
O
O
For
Best
continuing
work
from
previous
work,
the
Schick
people
SCH
see
thank
you
and
finally,
Best
Overall
work
to
the
SDN
microwave
radio
link
people.
So,
and
so
you
can
all
not
pick
up
your
trinkets
that
we
do
not
have
Thanks.
So
we
have
flip-flops,
we
actually
do
have
some
social
tickets,
so
should
we
give
so
the
the
best
the
best
overall
group?
If
you
have
people
who
need
social
event
tickets,
we
have
a
few
that
Charles
can
give
you
so
go,
see.
Charles
and.