►
A
A
A
A
B
Alright,
so
in
terms
of
what
I've
concluded
I
finished
the
word
floor
plan
for
the
IDM.
It's
basically
a
spreadsheet
that
organizes
the
breakdown
document
into
milestones
and
each
milestone
as
an
objectives
I
like
PRS,
and
they
have
a
finest
for
those
for
those
in
trees
and
also
an
estimate
of
duration
of
those
tasks.
B
B
The
first
paper
is
still
empty
because
it
will
be
the
easiest
one,
just
a
part
of
the
RFC
and
make
it
make
it
smaller.
The
the
spec
paper
already
has
a
few
stuff
in
there,
but
it's
very
drafty
I'm
just
sketching
things
and
experimenting
with
interfaces
on.
It
would
be
exciting
right
now,
don't
shake
it
now,
because
it's
executives
and
I'm
not
not
blocking
anything
and
my
next
things
just
one
which
is
to
create
a
small
presentation
about
a
DM
project
and
by
small
presentation.
B
A
Cool
Thank,
You
Andre,
just
some
people,
know
bit
more
about
about
the
context.
So
very
early.
Next
March
there's
going
to
be
the
eighth
edition
of
rebooting
web
of
trust
conference.
There's
a
call
for
well
to
participate.
You
you
have
to
propose
that
people
can
either
choose
to
vibrate
during
the
the
the
gathering
itself
at
the
conference.
It's
more
of
I
know
it
symposium
and
and
then,
and
then
that's
what
we
and
Ray
is
doing
now,
so
that
we
can
apply
before
the
end
of
the
month.
A
So
yeah
also,
if
anyone
is,
is
willing
to
join
us.
There
is
a
an
event:
entry
on
protocol,
labs,
event,
management,
repo
I,
think
it's
private
mature.
But
if,
if
you
want
your
planning
on
attending
or
if
you
want
to
attend,
please
you
can
also
get
it
get
in
touch
with
with
some
of
us
any
questions
for
Andre
on
this
No
all
right.
So
next
in
line
is
then
congratulations
again.
C
A
A
So
a
great
insight
by
my
researcher,
Carlos
Vaquero,
where
he
gave
the
idea
that
we
could
accumulate
non-mobile
edges
for
for
grj
and
then
try
to
merge
them
and
if
not,
keep
them
in
a
stash
as
part
of
the
state,
and
so
that's
what
I
implemented
and
I
did
a
bunch
of
tests
very
thorough
test
of
random
permutations
on
all
the
messages
and
I
haven't
been
able
to
break
it
so
far,
but
help
is
which,
on
that
ISM
is
appreciated.
Oh.
A
That's
what
Victor
wrist
is
telling
us
that
that
run.
2.0
works
like
like
that.
That's
that's
also
super
super
interesting.
Oh
it's
a
good
thing.
That's
a
good
sign
exactly
so
a
PFS
pops
of
room
I'll
go
get
it
so
there's
a
breaking
change
on
a
private;
well,
not
breaking
change,
but
there's
a
private
API
change
on
JSF
EFS,
which
we
were
lying
on,
which
is
the
lis
PTP
nodes
accessor,
which
now
is
going
to
be
made
public,
because
so
many
people
use
it
and
public
and
undocumented.
A
A
Offline
reconnect
use
cases
in
line
with
that
I
missed
last
week's
local
enough
offline
working
group
meetup
meeting.
You
should
attend
I
missed
last
one
I,
unfortunately
plan
to
attend
the
next
one,
but
that's
something
related
to
that.
So
find
support
for
JSF
EFS,
and
that's
oh
yes
and
next.
Well,
this
weekend,
I'm
going
to
be
at
Brussels
for
for
stem
2019
for
for
catching
up
and
I'm
watching
some
hopefully
attending
some
some
interesting
sessions.
D
Had
a
question
about
the
CR
dt
stuff,
you've
been
doing
so,
you
said:
you've
already
tested
that
out
and
it's
working
is
that
in
a
PR
somewhere,
oh
yeah,.
A
It
was,
it
was
merged
on
the
posterity's
that
there's
a
PR
can
I
can
find
that
it's
like
I
can
not
are
not
interrupting
throughout
this.
For
now,
I
can
lie
to
send
you
the
PR.
That's
part
of
the
RGA
itself,
you're,
probably
asking
about
the
current.
The
current
state
of
the
the
thing
that
Jimmy's
is
is
working
on
on
vitae
clocks,
synchronization,
that's
that's
a
different
different
issue.
I!
Think
it's
not
related
the
desynchronisation.
That
Jim
is
is
saying
not
sure.
If,
if
that
was
okay,
cool
and
dry
photos
are,
you
were
raising.
B
Your
hand,
it's
not
really
a
question
is
just
suggestion,
which
is
there
will
be
a
session
on
Folsom
I
posted
on
our
C
in
the
epiphyses
identity
channel,
which
is
this
one
I
will
put
on
on
soon.
You
might
want
to
attend
that.
It's
not
really
in
line
with
the
discussions
that
we
are
going
to
adopt
in
terms
of
you
know
we
are
talking
concerning
soil
I,
don't
think
the
solution
area
is
really
what
we
are
going
to
go
through,
but
it
might
be
useful
to
attend
and
see
what
they
are
doing.
A
E
Okay,
so
notes
so
I
went
to
three
different
working
groups
last
week,
but
a
lot.
Last
last
week,
I
gave
a
little
demo
and
I
showed
that
hey
everything's
working
and
then
immediately
afterwards,
I
discovered
that
actually,
everything
isn't
working
so
I'd
merged,
Pedroza,
CR,
DT,
fixes
and
I.
Think
the
new
Sierra
TT
is
working
really
good
at
the
CR
DT
level.
The
RTA
yeah
it
before
it
has.
E
These
unmerged
edges
have
been
added
to
the
CR
DT
and
it
sort
of
keeps
them
around
then,
but
then,
when
randomly
so,
the
we
have
one
test
in
pure
base
called
collaboration,
random
and
it's
a
fuzzing
test.
Basically
like
it's.
It's
this
whole
purpose
is
to
try
to
get
random
combinations
of.
Basically,
you
set
so
X
number
of
peers
and
they
all
type
random
characters
at
the
end,
and
it
all
merges
it
together
and
you
run
it
were.
You
know
so
many
characters.
E
So
if
you
have
like
I
was
testing
with
three
peers
in
each
peer
testing
or
typing
30,
characters
at
random,
with
random
delays
and
then
at
the
idea
of
90
characters
and
all
three
of
the
peers
should
have
the
same
90
characters
at
the
end.
But
what
was
happening
was
most
of
the
time
it
would
work,
maybe
like
five
percent
of
the
time,
ten
percent
at
the
time
it
would
just
stop
it
would
never
one
of
the
peers
would
never
ever
get
the
full
set
of
characters
and
I
thought.
Oh
well,
that's
easy!
E
That's
gonna
be
one
line
fixed
somewhere
so
a
week
later,
that's
where
they
spent
most
of
the
week
on
so
there's
some
issues.
I'm
gonna,
just
take
a
minute
just
to
show
what
I
was
doing
visually
because
it's
gonna
be
impossible
to
explain,
but
basically
the
idea,
ideas
trace
everything.
Every
single
thing
that's
happening
in
each
of
the
peers
and
log
it
all.
So
it's
more
than
just
logging,
it's
tracing,
because
you
can
reconstruct
what
actually
happened
from
that
data.
So
that's!
Why
did
this
week
so
yeah?
If
I
can
get
past
this
next
week?
A
Let
me
cool
thank
you
Jim,
so
I
added
a
I
had
a
comment
on
that
issue
which
I've
been
I've,
been
sketching
out
and
I.
Think
I
know
I,
I
well,
I
think
I
know
but
famous
last
words
the
solution,
but
yeah.
Let's,
let's
stop,
let's
not
at
all
I'd
love
to
see
what
what
you
have
for
tracing.
Oh
you
had
you
had
a
demo
right.
You
have
a
yeah
Jim
yeah
nice.
Thank
you
any
question
for
Jim.
So
far,.
F
D
D
F
F
F
F
G
F
F
Items
because
I'm
following
the
atomic
design
I
will
include
some
missing
textiles
for
you
for
larger
scales
and
then
in
progress.
I
have
right
now
the
the
wire
framing
and
designed
the
other
pages
for
IDM,
considering
the
concept
and
IA
decisions
and
then
yeah
until
maybe
next
week.
I
will
collaborate
with.
G
A
F
H
B
F
A
B
So
bear
in
mind
that
the
features
that
you
are
seeing
in
the
screen
is
not
the
purpose
of
this
image.
Just
for
you
to
feel
how
the
white
looks
will
look
like
and
the
concept
that
we
are
trying
to
explore,
so
we're
trying
to
explore
this
passport
alike
or
a
document,
a
logic
fuel
to
the
to
the
application.
So,
if
you
actually
to
actually
zoom
in
you
see
what
I
mean
by
these
illustrations,
so
that
people
can
can
associate
this
type
of
application
to
fill
the
day,
they
are
used
to
basically.
B
F
A
B
Is
Christian?
Yes,
all
right,
so
let
me
open
that.
So
this
is
a
diagram
that
we
might
like
I'm
zooming,
because
it's
it's
quite
big.
So
essentially,
what
we
have
here
is
all
all
the
user
journeys
and
all
the
features
inside
of
small
pages
that
are
outlined
here.
You
might
check
it
out
and
give
us
feedback
as
well.
It's
quite
complex
but
I
think
it
outlines
well
what
IBM
really
will
have
in
terms
of
the
MVP.
F
A
B
It's
just
opening
issues
so
that
we
can
expose
what's
going
on
at
the
moment.
Of
course,
we
will
be
convincing
every
everything
into
a
single
into
a
single
document
and
so
on.
That's
why
I'm
doing
the
papers
and
all
that
in
terms
of
whole
features
there
you
see
here.
These
are
the
MVP
that
we'll
have
for
for
June
end
of
okay.
B
A
G
I
G
D
G
H
D
D
So
I
was
kind
of
looking
at
the
existing
membership
protocol
and
some
small
changes
there
that
improved
improved
that
time
and
I've
done
a
little
bit
of
experimentation
to
demonstrate
how
much
it
improves
and
then,
while
I
was
thinking
about
that.
I
realized
that
there's
quite
a
lot
of
round
trips,
even
just
to
get
to
that
stage
where
we're
talking
to
the
discovery,
server
and
then
we're
talking
through
flood
sub.
D
So
we
have
to
set
up
a
flood
sub
connection
and
then
send
some
messages,
and
so
I
came
up
with
a
concept
of
how
we
could
build
our
own
kind
of
sub
protocol
in
order
to
reduce
the
number
of
round
trips.
That's
specifically
about
discovery
and
membership,
so
I
wrote
a
github
issue
about
it.
I
just
put
it
in
the
crib
headshot
there.
D
A
I
J
Okay,
so
yes,
so
yeah
I
was
mostly
focused
on
collaborations
work
last
week,
but
there's
a
couple
of
things
that
I
want
to
say.
Oh
so,
first
of
all,
I
really
like
the
identity
manager
like
visual
work,
it
looks
very
trustworthy
and
and
but
also
kind
of
friendly,
but
I
guess.
One
thing
that
I
want
to
point
out
is
that
the
way
that
the
documents
work
it's
it
reminds
me
very
much
of
like
kind
of
European
style
like
more
modern
design
but
I
think
for
Americans.
J
It
might
actually
not
make
them
think
of
like
a
passport
or
something
because
our
passports
look
very
different.
He's
like
very
garish
and
horrible
suggests,
there's
something
you
know
the
kind
of
a
side
side
note
I
guess
the
two
things
that
I
wanted
to
point
out
is
that
we
now
finally
have
a
real
local
working
group.
As
of
last
week,
local
was
an
O
instead
of
a
for
collaboration.
J
So
that's
that's
really
good
and
the
other
thing
that
I
want
to
draw
some
attention
to
is
a
discussion
of
the
head
of
the
reconnection
API.
That's
I
guess
Pedro
mentioned
already.
Yes,
oh
please,
chime
in
with
comments
on
that,
because
I
think
that's
like
pretty
important
to
define
that
API
well
and
kind
of
push
things
in
to
look
p2p
lands
so
that
we
don't
have
to
worry
about
them
in
user
space.
So
please
give
that
some
support.
Yeah.
H
A
A
H
A
A
A
J
A
Yeah,
of
course,
that's
a
good
good
good
question
cool,
so
we
cannot
think
step
into
unless
there
are
any
questions
for
Akkad
II
know
we
can
step
into
the
next
part
of
the
of
the
agenda.
So
first
I'm
going
to
ask
for
volunteer
to
lead
the
next
month's
meetings
going
on
with
the
with
the
product,
make
these
meetings
more
more
productive
and
also
so
to
rotate
the
the
the
leader
of
the
meetings.
A
A
Arkady,
thank
you
between
you,
you
and
ray.
We
cannot,
I
mean
we
can
do
alternate,
for
instance,
if
no
one
else
is
interested
or
can
alternate
between
anyone
that
is
interested
in
volunteering
and
then
and
then
we
we
can
still
do
we
evaluate
this
set
by
later.
If
availability
is
not,
it
is
not.
What
expected
then
are
then
also
volunteered
Thank
You
Jane
yeah
I'll.
B
B
I,
don't
think
that
one
thing
to
consider
is
to
give
authorization
to
the
zoom
account,
because
there's
an
account
and
for
the
DBC
is
I
think
you
are
the
owner
and
she,
if
you
wanna,
if
you
want
to
coordinate
the
next
meetings,
we
won't
be
able
to
use
the
same
zoo,
mailing,
ID
and,
for
instance,
I.
Don't
really
have
a
zoom
account,
for
instance,
I
think
it's
a
pain
by
the
count
of
something
so
I.
Don't
really
web
pen
won
the
yeah.
A
A
E
E
So
can
everybody
see
that
so
so
this
is
this
is
pure
base
here
and
if
I
look
in
the
test
directory,
here's
our
tests
and
the
one
particular
test
which
I'm
talking
about
here
is
the
collaboration
random
and
basically
it
starts
up
a
number
of
peers.
So
here
I'm
running
an
OG
s,
LD
3,
peers
and
type
some
30
characters
prepared
and
then
there's
a
down
here.
There's
a
this
is
the
delay
between
each
character
that
gets
typed.
Now,
when
your
end,
the
test
doesn't
really
look
like
much.
E
So
I
I
shortened
the
test,
so
it's
only
three
piers
here,
so
it's
quicker
to
run
so
in
this
case
it's
passing
so,
but
what
I
was
seeing
before
was
that
it
wasn't,
they
would
run
and
then
it
would
stall
out
so
I
in
order
to
debug
that
I
turned
up
the
debugging.
They
add
a
bunch
of
console.log,
so
it's
not
silent
anymore,
but
to
get
this
sometimes.
A
E
E
E
E
B
E
E
E
That
paid
no
attention
that
I'm
organizing
my
sauces
it's
a
mess,
so
this
is
what
it
would
actually
look
like
when
it
failed.
What
would
happen?
Was
it
would
stall
here?
It
wouldn't
actually
finish.
It
would
never
get
to
the
part
where
you
succeeded,
but
what
I
was
doing
was
I
was
tracing
back,
so
I
could
so
I
just
added
a
little
bit
of
color,
because
it's
just
this
this
line.
This
file
is
now
2,000
lines
long,
but
there's
basically,
three
pairs
and
I
collected
some
manual
notes
here.
E
So
basically,
in
this
case,
I
have
I
just
took
the
last
two
characters
of
the
pier
ID
just
make
it
simple
and
then
I
sing
bark
us
is
they
they
have.
The
the
parodies
are
consistent
every
single
time
because
we're
regenerating
the
appear
keys
so
anyways.
In
this
particular
situation.
We
had
three
piers,
so
you
a
q90
got
to
this
state
and
there
were
90
characters.
If
I
scroll
back
up
UMS,
which
was
a
different
pair,
they
also
got
to
90
characters.
If
I
scroll
back
up
here
SH
see,
they
got
only
88
characters.
E
So
for
some
reason
these
two
little
characters
went
missing.
I
did
some
like
a
accents
and
stuff
because
I
wanted
to
have
90
different,
separate
symbols,
and
it's
like
why
it's
like
they'll
never
ever
converge,
and
then
the
interesting
thing
was
I
was
looking
here
is
I'm.
Also
printing
out
I
wrote
a
little
routine
to
take
her
our
vector
clocks,
which
is
basically
pure
ID
and
an
incrementing
number,
and
for
each
character.
That's
typed
and
Rebecca
Clark
Scott
complete,
so
30
30
characters
there
for
each
of
them
30
characters
there
30
characters
there.
E
E
E
Little
JavaScript
program,
which
is
actually
not
that
little
now,
but
that
takes
takes
the
the
dump
that
the
trace
and
then
pretty
prints
it.
Basically
so
if
I
run
and
if
I
run
that
one
so
I
had
the
debug
output,
which
was
just
the
extra
console
logging
that
I
added
and
then
if
I
run
this
one
it's
now,
it
was
2000
triggers
before
now
it's
15
2009
s.
Now
it's
15,000
lines,
but
but
I've
got.
E
So
somewhere
in
here
so
one
of
the
peers
type
9.
So
in
this
case
it
was
s
type
9
and
then
I
can
say
before
this
was
from
the
log
file,
but
then
I
pretty
print
what
the
CR
DT
look
like.
So
in
this
case
there
was
nothing
in
the
CR
DT.
It
was
an
empty
document
at
that
point
and
then
he
creates
a
delta,
so
the
delta,
basically
is
it
edged-
is
from
nothing
to
9
and
I've
got
like
the
little
internal.
E
E
In
this
case,
there's
only
one
Delta
in
the
batch,
but
in
this
investigation,
I
was
going.
I
was
able
to
give
any
sort
of
see
the
CEO
duties
are
growing
here
and
you
can
see
this
was
the
Delta.
So
this
is
a
CV
before
there's
a
delta,
so
you
can
see
like
at
the
end
here
with
the
one
where
we
had
the
problem.
E
Eighty-Eight
if
I
look
the
Serie,
C
Rd
T
final
dump
this
unmerged
edge,
so
we
can
see
the
e
and
unmerged
edge
so
II,
but
it
wanted
this
value
tomorrow.
E
So
is
only
missing
a
single
element,
so
anyways
I'm,
not
gonna,
go
I.
Think
this
meeting
and
go
on
for
an
hour
if
we
tried
to
actually
debug
this
right
now.
But
that's
that's
the
basic
technique
that
I'm
doing
and
I
think
we
could
actually
make
it
more
of
us
standard
part
of
pure
pad
in
terms
of
being
able
to
I.
E
Don't
think
you
want
to
do
the
logging.
The
way
I
do
but
like
send
a
log
off
to
like
a
side
channel
where
it
could
be
like,
say
like
archive
to
disk
or
something
and
then
have
some
tools
that
could
bring
that
in
and
do
these
reconstructions
and
then
you
can
bring
them
into
visualizations
or
tools
that
can
reconstruct
not
just
one
peer
but
like
multiple
peers,
each
of
each
peer.
E
If
they
can
be
running
on
different
machines,
you
can
capture
separate
file
and
all
the
different
machines
bring
them
all
together
and
do
sort
of
like
forensic
reconstruction
x'
to
figure
out
when
things
go.
Go
strange,
and
this
is
I.
I
was
looking
at
the
CR
DT
things
now,
but
I
think
the
CDs
working
the
issues
with
our
replication
protocol.
A
That's
very
interesting,
thank
you,
Jim,
but
I
think
the
if
you
could
think
of
extracting
some
some
very
simple
command-line
tools
that,
like
like
I,
did
a
coupled
with
some
capabilities
of
logging,
that
you
can
add
some
environment
variable
that
we
can
activate
for
to
enable
logging,
like,
like
the
current
debug
usage
that
we
have
in
in
pure
base.
That
would
be
very
interesting
and
also
this
distributed.
Logging
is
something
for
p2p
is
for
decentralized
edge
edge.
A
Applications
is
something
that
I
believe
it's
it's
kind
of
hard
to
do,
but
every
step
that
we
can
I
think
can
get
into
that.
That
will
make
this
easier
to
I
mean
I'm,
perhaps
in
a
lab
environment
or
even
in
the
production
environment.
For
now
a
lab
environment
would
be
would
be
very
helpful
because
we
haven't
had
that.
That
story
is
not
not
great
so
far,
but
sometime
later
in
in
the
production
environment,
where
we
can
activate
traces
and
the
user
can
volunteer,
Tracy
is
like
typically
I.
E
A
C
C
C
You,
how
do
you
find
the
latest
version
of
your
di
D,
which
relates
a
little
bit
to
the
conversation
you
guys
had
about
how
you
can
have
two
different
users
or
two
different
devices
owned
by
the
same
user,
both
making
a
change
to
the
D
ID,
and
this
thing
is
solved.
This
thing
is
solved
by
anything
with
an
agreement
protocol
right,
so
a
blockchain
where
you
make
sure
to
add
all
the
blocks
to
the
chain.
You
can
make
a
CR
DT
out
of
the
blocks,
and
then
you
can
get
your
your
structure
and
you're
good.
C
You
can
use
the
leadership
stuff
that
Dirk
has
been
working
on.
That
will
also
do
it.
You
can
use
the
sort
of
multi
writer
append-only
graph
thing
that
I've
been
working
on.
That
will
also
do
it
right.
Any
of
those
things
will
solve
this
problem
I
because
none
of
those
those
are
all
still
in
development
and
have
ups
and
downs
to
them.
I
added
in
sort
of
a
bonus,
other
options
for
solving
that
problem,
which
is
the
way
I
understand
how
ipfs
IP
NS
use.
C
The
DHT
is
that
there's
this
bear
distributed
hash
table
thing
and
then
there
are
validators
on
top
of
it
that
apply
the
specific
rules
of
the
system.
So
for
ipfs,
the
rule
is
something
like
take
your
top.
You
know
take
anyone
who
will
host
all
the
multi
addresses
of
people
who
will
host
the
data
and
take
the
top
20
and
and
keep
those
around
and
for
IP
NS.
It's
you
can
only
push
if
your
peer
ID
is
the
key.
C
So,
for
instance,
you
could
say
if
you
receive
two
updates
concurrently,
instead
of
doing
the
IP
NS
thing,
which
is
choose
a
winner:
instead,
you
could
just
set
a
flag
that
says
you
know,
error
key
has
been
compromised
or
errors
simultaneous
right,
and
then
you
would
sort
of
failover
into
stage
two
of
figuring
out
what
to
do
next.
Now
that
the
bad
thing
has
happened.
C
You
also,
you
know
you
can't
really
keep
both
of
the
changes
around
in
the
arbitrary
case
in
Glasgow.
I
spoke
with
Stephen
and
Pedro
about
this.
How
if
you
need
to
keep
all
of
the
simultaneous
conflicts
than
the
amount
of
data
that
you're
asking
this
very
kind
DHT?
No,
that's
storing
data
for
free
for
you
to
have
could
balloon,
but
it's
possible
that
for
the
IP
ID
case
we
can
make
an
exception
because
you
could
say
if
you're
an
IP,
ID
hosting
DHT
node,
then
you
can
actually
do
the
merging
yourself.
C
You
can
pull
the
down
the
documents,
do
the
merge
and
and
do
it
like
that
I
can
clarify
that
more.
If,
if
there
are
questions
that
was
the
the
suggestion,
I
had
as
an
alternative
to
all
these
consensus
things
that
we're
still
working
on
alright,
so
then
the
next,
the
next
layer
of
the
next
question,
which
you
guys
went
through,
was
what
happens
when
you
want
to
revoke
a
key
so
solution,
one
was
basically
don't
lose.
Your
key
have
a
single
key.
You
Sameer
secret,
sharing
two
moving
around.
C
C
The
other
option-
I
wrote
in
the
note,
was
basically
the
same
as
I
had
those
two
world.
You
know
also
there
and
also
part
of
this,
but
there's
the
possibility
for
another
extension
or
two
extensions.
One
is
for
the
thing
that
has
the
two
layered
approach
it
sounded
in
the
video
like
you
guys,
moved
away
from
that,
because
the
fear
was
that
you'd
have
to
do
to
IP
NS
requests
and
those
can
take
a
long
time
to
resolve
I.
C
Okay,
yeah
say:
yes,
I
just
wanted
to
clarify
that
you
don't
actually
need
to
IP
NS
requests,
because
if
you
had
something,
if
you
had
a
way
of
like
flagging,
you
know
or
solving
that
problem,
one
that
first
problem
of
the
multi
writer
thing
or
having
a
flag.
You
could
just
wait
until
the
layer,
two
key,
the
non
primary
key.
C
You
could
wait
until
that
key
has
been
compromised
and
then
go
look
at
the
other
address.
Instead
of
looking
at
the
primary
address
and
forwarding
to
the
secondary,
you
could
look
at
the
secondary
wait
for
a
conflict
and
then
go
back
to
the
primary.
So
that
means
that
you
actually
only
need
one
IP
n
s
request
until
a
key
has
been
compromised,
so
that
helps
you
out
there.
C
The
the
other
thing
is
secret,
sharing
is
cool,
but
if
the
problem
is
that
you
actually
do
end
up
reconstructing
a
key
and
that
key
could
get
compromised,
but
you
can
use
group
signature
schemes
to
say
instead
of
giving
you
know
part
of
my
key
to
pedro
in
parts
you
know
under
and
part
to
gym
right.
Instead,
I
could
just
say
if
two
out
of
Pedro,
Andhra
and
Jim
sign
my
request,
then
that
one's
the
real
one
right
and
now
there's
no
key
to
get
compromised.
B
A
friendly
process
will
be
like
pairing
the
device
with
the
with
similar
to
how
you
pair
with
care,
cause
or
even
son,
when
you
pair
an
Apple
TV
or
something
like
that,
it's
very
protocol,
then
the
thing
just
just
passes
through
that
would
be
a
better
user
experience
and
that
is
actually
possible
by
doing
a
second
version.
Second
duration
of
IP
ID
and
your
ideas
could
be
incorporated
in
the
internet
spec,
but
because
it's
more
complex,
we
kind
of
opt
for
the
first
one.
B
For
now
just
to
get
the
thing
working
and
then
we
can
explore
those
things
later
on
so
I
think.
The
point
here
is
in
the
future.
Like
that's
July,
we
could
have
a
new
conversation
regarding
this
matter,
so
improve
the
fkd
spec
to
be
more
more
secure
and
more
more
more
user
friendly
as
well,
but
I.
Now
on
this
understand,
what
do
you
mean
with
those
comments?
B
C
C
B
The
second
thing
is
that
once
we
applied
to
my
PID
version,
2
with
a
better,
better
spec
and
so
on,
we
can
also
issue
a
notification
there
saying
hey,
you
should
approve
operate,
your
IP
ID
to
version
2
and,
and
the
user
will
follow
through
the
process
of
MiG
migrating
from,
like
I
give
a
person
one
to
person.
Basically,
so
we
can
leverage
that.
A
Alright,
thank
you,
then
yeah
I
think
yeah.
That's
that's
good
called
in
that.
We
should
start
discussing
this
and
thinking
about
things
so
that
we
don't
remember
to
start
on
developing
software
that
that
only
on
on
June
or
July,
but
that
we
can
plan
that
that
before
she
we
need
motive,
attracting
happiness
and
all
the
good
things
cool
anyway.
By.
B
The
way,
let
me
just
say
one
thing
about
the
multi
writer
I
penis-
we
all
kind
of
need
that
for
for
the
IP
ID
and
also
the
identity,
not
that
entity,
but
the
sinking
of
debris
firewall
credentials
as
well.
So
in
case
of
the
IP
e
ID.
What
happens
is
that
you
might
be
heading
to
devices
at
like
concurrently,
and
this
means
that
two
entries
on
the
public
key
array
will
be
concurrent
like
those
new
addition
additions.
So
we
must
solve
that
as
well.
B
B
As
you
said,
you
can,
you
can
do
like
a
deck
deck
good,
three
and
just
keep
the
heads
and-
and
we
start
actually
the
heads
on
the
on
IP
NS
on
the
penis
record,
and
when
we
resolve
the
GID
document,
we
are
resolving
those
heads
and
so
on,
which
are
kind
of
operations,
operation
base
deck,
and
we
just
reconstruct
the
the
document
based
on
that.
But
if
we
go
this
way,
we
are
complying
with
IP
IDs
back
because
IP
IDs
back
size
that,
if
I
resolve
that
many
snipers
it
should
give.
B
C
Yeah
I,
just
to
clarify
briefly
I,
think
I
think
we're
on
the
same
page,
the
when
I
say
like
multi
writer
IPS
like
basically
all
I
want
is
away
because
it's
gonna
be
some
CRT
T
thing
right.
All
I
want
or
need
is
some
way
to
actually
keep
track
of.
To
actually
keep
all
of
the
heads
right.
If
it
was
a
blockchain
I
would
get
all
the
heads,
because
someone
would
push
them
all
in.
It
was
a
consensus
thing
like
Derek's
been
working
on
it's
okay,
cuz
only
one
person
is
pushing
all
of
the
heads.
C
B
That's
correct:
if
we,
if
we
go
that
way,
as
I
said,
we
won't
be
compatible.
We
like
PID,
unless
we
propose
a
change
to
the
IP
IDs
back
or
maybe
we
for
can
give
it
another
name.
I,
don't
know
there
are
several
ways
to
do
that,
but
at
least
we
could
or
we
can
have
two
records.
One
is
actually
your
IP
IAE
and
that's
that
map
directly
to
the
final
Jason
and
also
you
might
have
a
sides.
A
That,
if
you
yeah
that's
interesting,
so
there
there's
some
solutions
where
you
have
like
a
by
that
time:
fault:
tolerant
cluster,
that
that
is
the
source
of
truth.
But
there's
always
the
eventual
consistent
player
for
for.
If
you
want
to
have
be
highly
available,
highly
available
rights
and
and
and.
A
A
D
A
A
B
Hair
on
the
banister
somehow
related
yeah
interesting.
It
is
this
okay
I
like
I,
like
the
view
of
having
died,
penis
record,
which
is
your
IP
ID,
basically
resulting
to
the
Jason,
because
it's
very
easy
and
and
if
we
change
this
back,
we
won't
be
compatible
with
the
website,
which
is
yeah
in
solver
Universal.
The
ID
result,
basically
the
side
nav
ID
and
they
already
have
the
IP
ID
spec
methods
implemented
there.