►
From YouTube: Eth2.0 Call #31 [2020/1/9]
Description
A
Okay,
the
stream
should
be
transferred
over
people
in
the
YouTube
chat
box.
Let
me
know
when
yes,
I've
just
shared
the
agenda
in
the
zoom
chats
and
on
the
youtube
check
and
we
will
get
started
happy
new
year.
Everyone.
Thank
you
for
coming
step.
One
else
has
been
step,
one
for
many
months.
Testing
and
release
updates
as
I
believe
you
are
aware,
and
please
meet
when
you're
not
talking.
A
Cool
as
I
believe,
you're
aware
we're
working
on
b0
10.
This
is
since
the
since
Def
Con.
We
made
some
breaking
changes
to
facilitate
a
altered
phase.
One
there
has
been
a
lot
of
churn
and
the
face
of
respect.
D0
10
is
to
be
a
stable
target
both
for
you
on
test
nets
and
for
some
audits
that
are
being
conducted
so
excited
to
get
it
back
to
a
stable
place,
thanks
for
bearing
with
us
on
that
journey.
A
So
again
that
is
to
be
released
tomorrow,
bye,
bye
tomorrow,
potentially
today.
This
contains
some
small
modifications
to
network
spec,
a
couple
of
different
configuration
ability
of
grammars
to
stay
zero
and
generally,
at
least
from
many
angles.
Non
breaking
changes,
not
I,
don't
think,
there's
anything
really
substance
to
in
the
phase
due
respect
other
than
being
able
to
configure
some
additional
rooms
anyway.
That
will
drop
soon
and
we
can
Danny.
A
A
Bls
is
going
in
BLS
is
the
major
breaking
change.
Thank
you.
Everyone
that's
been
working
on
the
standardization
effort.
Thank
You,
Karl
and
Kirk
who've
been
working
on
integrating
the
changes
into
the
spec
and
thank
you
everyone
for
doing
the
work
that
you're
going
to
need
to
do
to
get
that
up
to
spec.
I
know
there
is
some
movement
on
integrating
these
standards
into
libraries,
but
there
might
be
some
like
work
that
we
have
to
do
on
our
with
them.
Okay,
so
that
is
a
major
breaking
change.
Thank
you
other
than
that.
A
There's
some
additional
configuration
parameters,
there's
a
slight
modification
to
the
networking
spec
that
we
discussed
on
the
last
call,
but
nothing
crazy,
but
the
BLS
is
crazy
and
if
you
haven't
seen
Ben's
BLS
intro
digest
document,
he
shared
the
other
day
check
it
out.
It's
really
good,
in
addition
to
that,
along
with
testing,
I,
think
there's
a
handful
of
additional
tests,
I
think
on
the
deposit
processing
to
catch
a
couple
of
other
cases.
A
E
A
F
Here
and
yeah
we
we
have
published
Alex
has
published
some
of
some
of
the
updates.
Let
me
pose
just
this
issue:
it's
on
the
github.
You
can
check
the
format
and
example
of
test
two
actors.
This
is
an
integration
test
and
not
a
unit
test
that
we
used
to
have
for
state
and
other
things.
Yeah.
So
take
a
look
at
that
and
you
will
get
back
in.
A
F
C
C
We
are
focused
right
now.
On
the
networking
side,
we
have
a
first
PR
on
discovery
v5,
so
it's
a
halfway
done
and
that's
the
module
blocker
to
join
prison
and
lighthouse
test
nets.
We
are
currently
verifying
interrupts
with
Virgo
leap
p2p
as
well.
As
you
know,
we
need
to
be
version
introversion
three
weeks
ago
and
we
are
reevaluate
or
since
strategy
looking
into
what
deterrent
it's
too
safe
and
bandwidth.
Also
under
the
tooling
sides,
we
are
planning
a
PC
endpoint
for
become
load
for
all
kind
of
a
proceed
purposes.
A
H
We've
been
working
on
over
the
last
month
includes
each
one.
Data
management,
including
bowling
client,
needs
one
client
for
both
data,
and
there
is
a
question
about
this.
That
I'll
ask
in
a
moment
other
things
that
we've
also
been
working
on.
Our
the
sync
between
Artemis
clients,
including
the
startup
of
these
different
clients,
we're
also,
and
we've
also
done
some
integration
at
the
disk.
Very
module,
as
I
mentioned
previous
calls,
we've
taken
the
harmony,
discovery
module
and
we
have
adapted
it
and
put
it
into
Artemis,
and
that's
almost
done.
H
I've
also
been
working
on
the
beacon
plus
integration
of
Artemis
into
the
be
confessed
project
and
that's
almost
complete
as
well,
and
there
is
a
question
about
that
which
I'll
ask
any
moment
and
of
course,
Ben's
Java
implementation
of
the
IETF
standard
hash.
The
curve
method
432
has
been
merged
into
Organists
very
recently,
as
well,
thanks
Ben
for
that.
H
So
the
two
questions
that
I
had
the
first
one
was
about
the
eath
one
data
just
wondering:
if
there
are
any
resources,
we
can
look
at
on
how
people
are
syncing
eath,
one
main
that
data
to
the
chain.
Is
it
one
peer
that
there
it's
polling
or
more
and
people
are
interested
in
answering
that
question
they
can
ping
me
later.
The
other
question
which
perhaps
is
some
people
can
talk
about
now,
is
for
the
beacon
photos.
I
know
the
project.
That
project
is
that
the
zero
nine
one
spec
not
sure
if
it
will
be
updated.
A
There
were
two
approaches:
broadly
one
is
that
the
beacon
node
is
talking
to
some
eath
one
RPC,
employing
gathering
and
caching.
This
data
or
there's
like
that,
a
separate
kind
of
side
module.
That's
doing
the
work,
I
believe
that
everyone
is
generally
talking
to
one
RPC
endpoint
operating
under
the
assumption
that
this
is
either
a
node
that
is
being
run
by
the
validator
or
a
node
that
the
validator
is
connecting
to
in
which
they
have
some
sort
of
trust
relationship
with.
Are
there
any
other,
more
fancy
methods
going
on
that
I'm
not
aware
of.
A
Cool,
if
you
have
anything
to
share
ever
shot
on
on
methods
check
it
out,
we
have
a
there's:
a
change
in
B
zero.
Ten
that
uses
time,
stamping,
rather
than
block
height,
to
cast
votes,
which
has
some
benefits
in
terms
of
caching.
I
call
it
generally
backwards-compatible,
because,
even
if,
even
if,
like
a
handful
of
people,
are
doing
the
votes
differently
towards
the
end
of
the
period,
they
would
agree
on
votes
in
99%
of
the
time
anyway.
A
H
A
I
We
are
testing
it
with
the
test
provider,
and
the
third
thing
is
that
about
the
spec
thinking
we
are
following
following
up
the
zero
point,
nine
point
three
version
and
we
have
a
funny
route
PR,
which
is
almost
done,
and
after
that
we
will
think
with
now
they
were
pulling
nine
point.
Four.
Thank
you.
G
Happy
new
year,
everyone
excited
for
an
awesome
year
with
you,
so
most
of
you
guys
have
heard
we're
relaunching
our
test
net.
The
Genesis
time
is
4:00
p.m.
Pacific.
Today
it
implements
everything
up
the
to
the
V
0
V
0,
9,
3
spec.
We
do
use
the
V
0,
9,
force
pesto
and,
most
importantly,
this
uses
the
many
config
with
one
tenth
of
the
deposit
requirement,
which
is
3.2
e.
G
So
we
do
in
Spanish
singing
time
to
be
relatively
slow
early
on
and
it's
a
p0
priority
for
us
to
work
on
for
the
next
coming
weeks.
So
come
one
key
progress
is
from
the
last
few
weeks
we
implemented
everything
up
to
you
busier
busier
than
three
of
this
back.
There's
a
bunch
of
optimizations
that
went
into
so
forming
a
country
such
as
probably
handling
of
the
deposits.
I
know
how
we're
doing
and
how
we
don't
really
look
at
the
given
data.
G
We
had
to
refactor
some
of
the
belt
at
her
assignment
and
committee
logic
for
the
belt.
In
her
obvious
response
to
be
us,
it
could
be
as
efficient
as
possible.
Raul
also
the
allowed
reward
on
implementing
a
custom
version
of
of
SSH
rehash
state
so
yeah.
If
anyone
is
interested
being
a
prison
validator,
the
charity
time
is
4
p.m.
today.
Oh
I
will
drop
the
participation
link
at
Sharada,
4
p.m.
pacific.
A
K
Yeah,
hey
everyone
from
us:
we've
been
running
a
test
net
Fri
I
think
we
started
on
like
the
22nd
of
December.
We
haven't
actually
announced
it
publicly,
but
we
should
be
doing
that
in
the
next
few
days
and
it's
it's
on
a
main
net
spec
that's
been
running
fairly
stable,
so
it's
time
to
get
some
public
people
in
there
to
try
and
break
it.
K
So
since
then,
we've
been
trying
to
focus
on
some
of
the
stability
of
the
test
name
and
and
performance
in
particular
block
processing
times,
because
syncing
obsolete
takes
a
while
to
process
all
the
blocks.
We've
actually
made
some
some
pretty
significant
gains,
which
should
hopefully
be
good
when
we,
when
we
really
start
to
the
public.
So
we've
been
looking
at
some
of
the
networking
stuff
as
well
in
terms
of
I'm
fixing
and
stuff
and
stability
for
our
test
net.
K
The
Biff
Floyd
team
also
found
like
a
lock.
You
know
you
know
API,
so,
thanks
to
them,
they've
helped
us
kind
of
debug
that
which
has
been
really
useful,
we're
close
to
like
a
version
1.1
release
which
adds
all
these
kinds
of
stability
fixes
and
what
processing
updates.
So
we
hopefully
release
those
in
the
next
few
days
with,
along
with
the
new
tests.
Now
we
remove
the
quiet
with
the
requirements
for
like
a
full
f1
note,
so
we
don't
need
a
you
can.
K
We've
started
where
we've
add
Harumi
BLS
Urumi,
which
saw
think
about
two
times
gains
in
that
which
is
passing
all
the
tests.
We
haven't
actually
added
it
into
master
yet
and
we're
currently
in
the
process
of
adding
the
naive
aggregation
strategy,
which
should
hopefully
be
soon
which
we
will
release
in
a
in
an
entirely
new
chestnut.
Oh
yes,
thank.
A
F
Harmony,
we
have
updated
this
back
to
zero
point.
Nine
point
four
and
alex
has
generated
those
test
vectors
according
to
this
version,
also
mitri
working
on
this
crew
force,
innovation
Anton
helped
him
to
extend
his
simulator
that
he
crafted
for
gossip
sub
simulation
and
now
he's
bringing
some
sorry
to
bring
in
disco
five
to
to
this
simulator.
A
A
L
Hey
so
we're
getting
back
up
to
speed
after
the
holidays
as
far
as
spec
tests
as
far
as
the
spec,
we
are
in
the
final
stages
of
merging
our
0.9
branch
into
master,
and
then
at
that
point
we
should
be
up
to
zero
point.
Nine
point:
three:
we
have
been
working
on
refactoring
our
SSC
library
to
allow
for
something
like
proto's,
remarquable
library,
something
that
can
abstract
over
a
Merkle
tree,
backed
structure
and
give
you
an
interface.
It
looks
like
a
kind
of
like
a
normal
object
would
and
as
part
of
that
I
guess.
F
D
The
idea
is
make
part
of
the
to
see.
Philosophy
is
objects
can
be
very,
very
large,
so
some
kind
of
data
sharing
as
part
of
your
SSE
library
is
just
very
nice
to
have
if
you're
working
this.
This
is
like
a
focus
more
on
proofs
or
more
like.
These
could
be
continuous
at
this,
where
you
want
to
know,
and
if
all
of
this
has
to
route
based
on
cached
sub
trees,
and
so
this
is
all
nicely
obstructed
in
remarkable
marriage.
It
can
just
act
like
it's.
J
Okay,
this
slide
here
so
we're
still
progressing
I
mean
we're
behind
everyone
else.
We've
got,
we've
just
got
the
validator
block,
signing
working,
we're
still
on
0
going
1.
So
the
validator
talks
to
Beacon
note,
so
you
actually
run
both
and
generated
ongoing
chain,
so
we've
got
their
core
there,
but
there's
still
no
attestations
still
mock
if
one
data
weird
between
just
about
to
start
peering
work
on
the
viewing
to
publish
the
the
block,
it's
already
started
some
research,
but
no
actual
code,
yet
most
of
the
teens
being
busy
with
their
eighth
one
work.
J
A
Ok,
ok,
next
line
agent
network
updates,
tell
you
know:
service
off,
I
meant
research
up
did.
L
Research
updates-
let's
see
so
that's
mostly
phase
one
and
Phase
two
and
figured
out
some
input.
Some
improvements,
the
data
to
idea
the
availability
of
routes
that
makes
the
instruction
slightly
cleaner
I
mean
we
spent
a
bunch
of
time
looking
through
looking
through
alternatives
to
the
2ds
date.
Availability,
construction
and
it
seems
like
the
consensus
view-
is
that
there
are
there's
a
bunch
of
cool
ideas,
but
there,
but
not
an
immediate
term
enough
time
to
be.
M
L
L
Trying
to
rule
recent
things,
I
mean
on
the
on
the
face
to
front
I
mean
there
were
also
some
I,
some
ideas
around
how
to
do
them,
execution,
environments
and
probably
the
most
recent
kind
of
actually
new
thing.
Actually
new
thing
that
at
least
I
published
was
this
proposal
for
switching
from
e1
to
e2
or
moving
eath
one
over
inside
of
eath
to
on
an
earlier
schedule
than
before.
L
So
this
was
the
idea
that
we
would
basically
figure
out
the
stateless
client,
verifier
and
then
and
have
actually
just
include
that
into
the
e3
system
and
then
come
up
with
this
mechanism,
where
each
one
of
our
theaters
can
voluntarily
declare
themselves
as
being
too
friendly,
and
if
they
are,
you
too
friendly,
then
like
it's
only
from
that
will
know
you
would
select
proposers
and
proposers
would
have
to
have
the
youth
one
state.
But
then
the
committee
members,
that
validate
would
still
be
taken
from
the
global
set
and
they
would
still
validate
stay
alone.
L
So,
actually,
that's
something
the
problem
we
could
be
in
interesting
to
discuss
at
some
point
like
in
terms
of
the
prerequisite.
That's
I
wrote
that,
like
I
wrote
up
that
post
on
protocol
changes
for
Steelers
clients,
but
the
main
pre
record
is
pretty
much
is
getting
sales
clients
working
correctly
and
so
that
and
that's
something
that
I
know
that
people
are
working
on.
A
J
L
So
the
numbers
that
I
figure
it
out
for
witnesses
are
that
we
can
realistically
bound
to
us
tailless
Kawai
witnesses
through
a
worst
case
of
about
4
megabytes
and
an
average
case
so
about
six
hundred
kilobytes,
and
it
just
so
happens
that
625
kilobytes
is
the
maximum
size
of
a
block,
that's
possible,
and
these
one
chain
host
system.
Also
one
thing
I
suggested
it's
just
repeating
the
the
tests.
L
And
I
feel
like
uncle
Reid
is
like
the
biggest
direct
sign
of
trouble
happening,
because
everything
else
can,
like
the
only
other
thing.
Negative
thing
that
would
happen
is
just
clients
like
regular
nodes,
syncing
those
blocks
more
slowly,
but
that's
something
that
we
could
kind
of
measure
entirely
other
fairly
easy
ways
as
well.
M
One
exciting
update
from
me
is
that
so
supranational,
which
is
the
the
team
that
is
working
on
basic
kind
of
independently
kind
of
as
a
side
project,
started
thinking
about
accelerating
DL
s12
381
and
they
actually
working
with
the
collaborator,
who
is
working
on
kind
of
an
implementation
from
scratch
of
obviously
one
for
four
CPUs
and
he's
he's
already
implemented,
point
addition
and
point
doubling
and
he's
actually
faster
than
Harumi,
so
he's
about
10%
faster
for
these
two
operations
and
he's
going
to
start
implementing
a
new
pairings.
M
You
know,
in
addition
to
the
performance
improvements
which
I
guess
I'm
nice.
This
guy
is
also
a
formal
verification
kind
of
expert
and,
like
he
had
a
look
at
the
halloumi
code
and
his
impression
is
that
formerly
verifying
the
Hiromi
stuff
will
be
difficult
and
how?
The
reason
is
that
you
know
he
Rumi
has
this
custom
compiler
and
it's
it's.
You
know
it's,
but
the
whole
code
base
is
quite
complex
and
and
all
over
the
place,
so
I
mean
he's
taking
a
different
approach,
which
is
you.
M
M
So
like
we
could.
You
know
we
could
imagine
like
a
small
like
a
PCIe
card
or
a
small
box,
that
you
connect
to
your
computer
and
it
would
do
about
a
billion
exponentiation
in
like
two
seconds
or
something
like
that.
I
mean
still
early
days,
but
I'm
quite
excited.
The
other
thing
that
I've
been
working
on
is
vdf
day.
So
there'll
be
a
vdf
day.
One
day
prior
to
SBC
the
Stanford
banking
conference
I
was
there
will
be
on
the
18th
of
February.
N
Awesome
so
we
have
a
few
things:
there
is
a
phase
two
column
14th
Tuesday,
so
that
that
is
coming
and
we'll
be
brainstorming
and
talking
about
a
few
different
things,
we're
a
lot
of
things
are
continuing
from
updates
from
from
last
time.
So
we
have
a
you
know
a
ganache
like
tool
that
can
run
ease
on
multiple
shards.
So
we
have
that
running
for
multiple
shards
in
parallel
right
now,
and
so
it's
early
on
will
release
this
next
week,
a
full
roadmap,
if
doc
and
readme
around
it
as
well.
N
We're
also
now
focus
with
tx/rx
work,
collaborating
to
build
an
end-to-end
testing,
suite
and
so
kind
of
the
move
here
is
that
it's
library
based?
So
you
would
pick
your
proof
back
end
or
memory
back
end
and
it
kind
of
connects
all
these
pieces,
and
then
you
can
develop
the
Yui
to
run
on
top
of
that
so
doc.
A
roadmap
doc
is
coming
coming
soon
on
that
as
well.
N
So
looking
looking
to
get
feedback
on
that
and
hoping
people
will
go
through
that
before
the
call
continue
to
look
at
some
of
the
cross
shard
schemes
and
things
around
that.
The
talk
did
a
interesting
right
at
Baumann
netting
format
recently,
which
seems
really
interesting
and
seems
to
remove
some
of
the
need
for
nonces
and
fields.
So
I've
been
playing
around
with
some
I
guess
some
ideas
on
that,
and
then
we
have
some
other
write-ups
coming
and
one
of
those
is
on
the
upgrade
ability.
Another
one
is
on
a
fee
on
the
fee
market.
N
The
discussion
for
phase
one
and
laying
out
the
different
options.
I
think
there's
three
that
have
been
stated
so
far
and
one
includes
kind
of
a
minimalistic
Phase
two.
So
there's
some
interesting
discussion
around
that
and
wanting
to
get
some
yeah
get
some
answers
around
that
as
well,
and
then
also
some
interim
interim
or
iteration
plans
for
Phase
two
as
well.
Thinking
of
you
know,
can
we
start
off
with
a
really
restricted,
simple
phase,
two
that'll
test
that
and
then
you
build
on
top
of
that
so
deserves
our
updates.
O
A
O
P
A
A
A
N
F
A
F
A
L
How
funny
expensive?
What
do
you
finally
get?
That's
a
good
question.
I,
don't
know
like
I
think
a
finality
gadget
as
a
voluntary
thing,
like
I,
think,
could
be
done
earlier
than
a
full
switch,
but
but
like
enshrined
fine
before
curious
rule
like
if
we're
willing
to
do
that,
then
why
would
we
not
be
willing
to
just
fully
merge
the
two
together
yeah.
A
A
A
Great
okay
cool
any
other
spec
discussion,
general
items:
we
have
phase
one
call
coming
up:
we're
gonna,
have
a
networking
call
coming
up.
We
have
some
relaunches
of
main
net
SPECT
test
nets
on
the
OU
9x
release
and
we
have
a
Oh
10
release
coming
out.
We
also
have
a
number
of
audits.
We
have
some
economic
analysis
going
on
the
deposit
contract
right
up
on
the
floor.
Verification
is
being
done
right
now.