►
From YouTube: Ethereum Core Devs Meeting #36 [04/06/18]
Description
A
A
Okay,
hi
everybody,
we're
gonna,
go
ahead
and
get
started,
just
a
reminder.
If
you
could
mute
whenever
you're
not
talking.
That's
gonna
help
a
lot
with
the
audio
quality
and,
if
suddenly
I
go
offline.
It's
because
of
my
cat,
who
just
walked
across
my
keyboard
and
hitting
the
Escape
key
but
I,
think
I'm,
safe,
okay,
I'm,
good
anyways.
We
can
get
started
with
the
agenda.
It's
pretty
packed
today.
B
C
B
Yeah
the
issues
that
test
air
PC
me
to
them
me
there's
no
consensus
about
those
made
of
some
still
struggling
with
that
one
method
of
how
to
add
a
new
account
or
is
a
transaction
with
secret
case.
So
the
discussion
with
virtue
was
was
it
should
be
a
new
testing
method
for
executing
a
transaction?
Is
a
secret
key
via
RPC
or
registering
a
new
account?
Who
is
the
private
key,
which
then
could
be
used
to
sign
those
transactions.
B
D
A
A
Okay,
the
next
agenda
topic
is
e,
I,
P
712,
add
s
sign
type
data
as
a
standard
for
machine,
verifiable
and
human
readable
type
data
signing
with
etherium
keys.
So
we
talked
about
this
last
meeting
and
there
was
a
lot
of
support
for
this.
However,
there
was
also
some
some
talk
about
making
sure
that
everybody
in
the
ecosystem
was
on
board
because
it
was
the
decision
was
made
to
not
keep
backwards
compatibility.
Today
we
have
Remco
on
the
call
room
Co.
E
Kind
of
the
interpretation
of
the
data
to
be
signed
to
the
user
agent,
such
that
the
user
would
be
presented
with
a
more
elaborate
view
and
understanding
of
what
is
to
be
signed
as
I
was
working
on.
This
problem,
I
realized
that
it
quickly
splits
up
in
a
number
of
sub
problems
that
need
to
be
addressed
when
it
comes
to
signing
arbitrary
messages,
I'm,
true
user
agents
and
and
quickly
in
order
these
these
would
be
hashing
of
arbitrary
messages
without
collisions.
Any
security
issues,
of
course,
domain
separation.
E
When
messages
do
naturally
collide,
but
you
still
want
to
somehow
make
the
signatures
incompatible,
nonce
s,
which
is
something
that
I,
don't
think,
has
been
addressed
in
the
discussion
at
all,
and
with
that
I
mean
how
do
you
prevent
replay
attacks
so
standard
aetherium
transactions?
Use
nonces?
For
that?
The
sign
messages
do
not
have
that.
E
Currently,
we
require
every
implementation
to
come
up
with
their
own
solution
to
this,
which
might
end
up
being
the
only
viable
solution,
but
it
might
also
be
something
we
can
address
in
some
standard
way
to
prevent
people
from
shooting
themselves
in
the
foot
this
way
and,
of
course,
the
original
issue
of
how
do
we
present
all
of
this
in
a
user-friendly
way
such
that
people
who
are
relatively
novice
to
cryptographic,
signing
schemes
can
still
use
them
securely
and
relatively
trusted.
Now.
E
This
is
a
lot
for
one
standard
and
I
mostly
narrowed
this
standard
down
to
the
scope
of
just
defining
a
hashing
mechanism
that
is
able
to
hash
messages
into
a
bytes
32
in
a
safe
way
such
that
developers
can
basically
just
hash
any
Strutt.
They
want
and
know
that
they're
safe.
It
also
touches
on
the
issue
of
domain
separation,.
A
So
it
sounds
like
you're
still
getting
feedback
on
the
EIP.
In
addition,
have
you
received
other,
at
least
since
the
last
Court
have
called
two
weeks
ago?
Have
you
seen
other
people
or
other
groups
coming
on
board
with
this
and
saying
that
they
support
this?
Maybe
any
hardware
wallets
or
things
like
that.
F
I
was
pre
skeptical
of
the
original
standard
sign
tap
data
with
the
original
idea.
Being
you
know
to
enable
UI
display
of
this,
but
I
recently
reread
the
the
revised
EAP
and
I
think
it's
moving
in
a
good
direction.
I
think
it
could
be
really
useful
standard.
I
just
did
a
review
in
the
bunch
of
technical
comments,
but
in
general
I
think
it's
really
promising.
A
E
Sorry
I
read
through
the
20
comments
in
indeed
I
noticed
some
of
these
issues
and
I
will
definitely
address
them.
I
hope
you
are
generally
on
board
with
the
direction
I'm
taking
it.
I
know
it's
more
complex
than
than
something
like
ABI,
encode
and
yeah.
I
explained
why
I
was
let
in
that
direction.
Yeah.
F
G
So
if
I
may
eirick
question
I
have
not
died
into
the
revised
version,
but
I'm
curious
about.
Is
this
standard
generalized
enough
to
be
kind
of
a
top-level
standard,
or
should
there
be
some
other
form
of
top-level
standard,
something
like
e1
91
and
have
this
as
a
subgroup
of
that
or
is
this?
Does
this
accommodate
old
use
cases
I
I.
E
G
F
A
Cool
so
it
sounds
like
there
is
a
lot
of
support
for
this
I
think
it
does
need
to
be
polished
up
a
little
bit
in
my
opinion,
just
because
there's
still
some
few
more
discussion
topics
happening
and
then,
if
we
can
as
much
as
possible,
get
a
feel
for
which
developer
tools,
companies
and
hardware
wallets
and
other
people
who
rely
on
this
get
on
board
if
we
can
get
as
much
of
them
as
possible.
I
think
this
can
be
something
that
is
accepted.
A
E
Can
I
ask
one
final
question
before
you
move
on
to
the
next
subject
so
right
now
the
scope
is
limited
to
a
hashing,
arbitrary
messages
and
it
touches
a
little
bit
on
domain
separation,
but
I,
don't
think
it
fully
addresses
the
problem
of
so
domain
separation.
Yet
that's
an
ongoing
discussion.
It
does
not
touch
on
the
issue
of
non
search
and
preventing
replay
attacks.
Is
this
something
that
should
be
covered?
Is
this
something
that
we
want
to
postpone
to
and
a
future
AIP.
A
A
A
Okay,
cool
in
that
case,
thank
you,
so
much
Renko
for
kind
of
giving
us
that
overview
and
giving
us
an
update
on
it
and
I
hope
this
goes
forward.
Thank
you.
So
the
next
agenda
topic
is
a
IP
665
at
a
pre
compiled
contract
for
the
IDI
two
five
five
one,
nine
signature,
verification
that
is
by
Tobias
Augustine,
who
joins
us
here
today
or
our
Burstein.
Sorry,
who
joins
us
and
he's
gonna
kind
of
explain
just
his
just
real
quick.
H
H
H
H
A
I
I
only
have
a
comment
in
I
can
add
a
couple
of
anecdotes
of
companies
that
have
requested
this
feature
from
paradiso.
Some
just
say
that
there
be
and
precompile
is
too
slow
and
that
Edie
would
maybe
be
faster
and
like
they're,
trying
to
do
some
unchain
crypto
stuff
and
like
either.
They
want
the
IDI
curve,
specifically
or
they're,
just
like
trying
to
find
a
way
to
do
it
in
as
cheap
away
as
possible,
and
so
the
B
and
pre-compiled
is
not
hold.
J
On
so
I
mean
that'd
be
B
and
once
when
he
alteon
once
when
he
aid
is
not
by
itself
like
as
a
curve,
much
slower
to
execute
than
that
look,
there's
nothing
inherently
mathematical
about
the
curve
that
makes
it
slower
to
execute.
Insecty
right,
like
as
I
understand,
is
the
only
reason
why
the
implementation
implementations
have
been
slower
than
sexy
so
far
is
just
because
there
seems
to
be
better
and
more
optimized.
Libraries
for
sex
P
than
there
are
fall
beyond
128
I
mean
so.
I
I
J
But,
like
I
guess,
are
those
benefits
large
enough
to
be
worth
an
entire
new
pre-compiled
I
mean
plot
I
mean
I'm
inclined
to
say,
like
we
not
like,
won't
be
up
with
them
at
the
end
like
have
like
for
that
curve.
That
was
enough
because
they're
like
it,
created
an
entire
new
capability
of
doing
peering
based
cryptography.
But
here
the
games
just
seem
more
marginal.
A
Okay,
do
we
have
any
other
comments
on
that?
My
comment
would
be
in
the
past.
We
sorry
about
that
cough.
In
the
past,
we've
talked
about
and
no
longer
putting
in
pre
compiles,
because
II
was
coming
up
and
I'm,
not
sure.
If
we're
still
of
that
opinion
or
not
just
because
the
time
it
would
take
to
implement
and
test
the
new
pre
compiles
would
take
away
from
the
time
implementing
stuff
like
Casper
sharding,
and
he
was
him
and
the
fact
that
he
was
and
can
sometimes
run.
A
J
Yeah
so
I've
definitely
liked.
My
understanding
is
definitely
that
you
was
was
fast
enough
to
just
implement
all
the
cryptography
and
he
was
I
mean
I
would
be.
Will
I
personally
would
be
willing
to
consider
pre
compilers,
but
for
cryptographic
operations,
but
only
if
they
really
bring
substantial
added
value
and
at
best
points
you
know
we
have
hashes.
We
have
signature
algorithms,
we
have
hearings,
we
have
beginner,
arithmetic
them.
I
would
say
at
this
point
like.
A
A
F
A
G
Yeah
on
the
side
note
victaulic
mentioned
low
in
gas,
cost
for
arithmetic
operations
and
pre
compiles
and
I.
Think
that
I
think
that's
a
good
idea.
I
some
measurements
show
that
I
mean
the
basic
operations
can
probably
be
cheaper
than
they
are
already
with
some
dedicated
libraries
and
optimizations
base
for
the
go,
client
and
I
think
we
all
know
that
it's
basically
the
disk
accesses
that
take
the
lot
most
of
the
time
and
needs
to
have
a
heavy
gas
cost.
They
are
if
Maddox
could
probably
be
a
bit
cheaper.
G
I
A
A
really
good
idea:
Tobias.
Does
it
sound
alright
if
we
just
start
to
kind
of
collect
more
use
cases
and
add
them
to
the
EIP
yeah?
That
sounds
perfect
great.
Thank
you
all
right.
Thank
you
very
much
all
right.
Getting
on
to
the
next
agenda
topic.
We
have
modified
block
mining
to
be
a
sick
resistant,
that's
something
that
Piper
initially
started
and
has
now
grown
into
a
large
topic
of
discussion
in
the
community,
so
Piper.
If
you
can
just
give
a
quick
background
on
how
that
AIP
started
and
where
it
is
now.
D
D
We've
got
an
EP
that
was
merged
as
a
draft
a
few
days
ago.
That
proposes
some
very
low
touch.
Changes
to
the
Etha
hash
algorithm
things
that
I,
if
I,
understand
correctly,
should
be
very
easy
for
clients
to
modify
their.
You
know
internal
block
ceiling
to
accommodate
these
without
any
you
know,
really
significant
changes.
D
The
idea
behind
this
is
not
to
improve
a
sick
resistance
as
much
as
to
an
attempt
to
break
existing
hardware,
and
at
least
from
my
perspective,
that
approach
is
intended
to
be
a
short-term
fix
to
push
the
you
know,
assuming
you
agree
that
Asics
are
a
problem
or
that
we
don't
want
them
on
the
network.
The
hope
is
to
push
them
out
far
enough,
that
we
have
time
to
get
proof
of
stake
into
place.
D
D
First
of
all,
sorry
of
italic
I
want
to
throw
one
more
thing
in
I
have
had
one
piece
of
back-channel
communication
that
indicates
that
we
can
expect
these
to
not
be
programmable.
The
the
current
Asics
that
are
in
production
there's
no
way
to
verify
that
information,
and
so
you
know
you
can
you
can
take
that
as
you
will,
but
I
just
wanted
to
toss
that
one
more
thing
in
there.
J
It
would
still,
but
you
would
still
the
the
speed
increase
would
still
be
limited
because
it
would
have
to
like
the
bottleneck
would
be
ramp.
Well,
at
least
up
to
some
points
right
like
if
the
computing
launched
it
was
extremely
fast
and
it
could
all
Democrat
just
like
pseudo
like
verification
process,
but
it
all
fakes
there
even
close
to
doing
that.
J
So
those
things
together
basically
implies
the
the
AO
is
that
it
seems
likely
that
the
ASIC
is
just
like
a
regular
computer
was
a
was
a
better
piece
of
ram,
was
up
with
anything
non-essential
like
that's
in
the
GPU
or
that's
in
the
RAM
stripped
out
and
like
the
other
thing.
Yes,
that
I
heard
I
mean
something
about
like
basically,
these
like
Asics,
using
some
new
kind
of
Ram
like
either
gd-r
gd-r,
3
or
and
like
something
that's
otherwise
fairly
hard
to
get.
J
So
that
would
imply
that
you
know
like
it
is
just
basically
a
better
like
a
computer,
that's
optimized
to
word:
I'm
IO
hard
group
of
upper
fork,
algorithms,
in
which
case
the
immediate
concern
is
basically
that,
like
there
is
nothing
that
we
can
do
you
know
and
because,
like
even
if
we
just
replace
some,
if
hash
entirely
with
like
equi
hash
or
whatever,
that
you
know
what
the
hell
Munira
use.
This
or
something
similar
than
it
would
like.
J
Basically,
the
ASIC
would
still
just
be
able
to
do
it,
and
it
may
be
the
case
that
that's
not
true,
but
we
just
have
no
idea.
You
know
at
least
right
now
what
yeah,
what
specific
protocol
change
actually
would
manage
to
make
a
difference
and
like
given
that,
like
the
only
thing
that
I
think
would
reliably
stop,
these
particular
Asics
is
switching.
The
proof
of
works
is
something
that's
just
not
even
IL
balanced
at
all.
So
one
example
that
I've
heard
up
is
just
switching
it
to
shots
Reid
but
like
with
the
chain
I.
J
Will
they
changed
that
drastic
idea
inclines
not
to
do
it
because,
like
first
of
all,
if
we
switch
to
something
like
that
and
that's
the
last
asic
resistance,
which
would
mean
that
we
explore
it
again
in
something
like
six
to
twelve
months,
so
we're
only
buying
a
bit
of
time
and
also
the
actual
development
effort
of
making
good
GPU
miners
for
sha-3
and
just
getting
everyone
upgrade,
is
likely
to
be
fairly
chaotic
and
really
anti
detract
from
work
for
more
important
things.
So
at
this
point,
I
get
in
it.
J
I
personally
do
mean
quite
significantly
to
word
no
action
I
mean
the
other.
The
other
reason
to
lean
to
where
no
action
is
that,
if
you
just
don't
get
a
worst-case
scenario,
look
worse
case
scenario
was
basically
that,
like
you
know,
bit
main
controls
like
a
very
large
portion
of
the
etherium
network
for
some
period
of
time
and
like
the
day
they
and
they
can
end
up
trying,
like
you
know,
look.
This
is
not
Bitcoin
right,
like
miners
are
not
in
control
here.
J
If
there
comes
a
day
when
they
have
been
a
majority
hash
power
and
they
try
to
use
it
for
evil,
then
you
know
the
net
basically
will
just
speed
up
like
Casper's
development,
and
you
know
like
to
hell
with
any
remaining
bugs
and
just
one
and
try
really
hard
to
launch
the
thing
within
a
week
and
mining
reward.
So
it's
down
by
90%.
J
J
In
progress
of
like
finalizing
the
code
for
X,
the
second
stage
of
the
test
net-
and
it's
once
the
the
one
of
our
goals
for
the
second
stage
of
the
test,
that
is
an
algorithm
freeze
and
like
full
specification
of
the
algorithm.
So
that's
something
that
you
know
is
totally
ready
for
gas
and
parity
to
theoretically
start
implementing.
Now
it
is
in
progress
of
also
being
formally
verified
and
also
being
checked
by
yet
multiple
groups
of
academics.
But
if
we
really
wants
to
like
like
if
51%
attack
concerns,
do
me
I
demand
it.
J
D
K
So,
as
you
all
know,
the
runtime
verification
guys
received
a
grant
and
are
doing
formal
verification
of
the
Casper
contract.
You
always
working
really
closely
with
them.
I
rereleased
the
implementation
guide.
Last
week
with
some
of
the
updates
we
had
made
and
contacted
some
of
the
clients
I
think
we
jumped
the
gun
a
little
bit
of
that,
because
it's
actually
a
little
bit
under
specified
in
terms
of
the
mechanics,
so
we're
actually
writing
the
formally
IP
that
specifies
Emily
what
the
clients
need
to
do
at
fork.
K
This
is
we're
finalists,
smaller
implementation,
details
and
I
expect
you
have
the
IP
ready
for
review
next
week,
so
I
mean
generally
the
planners.
Have
the
runtime
variation
people
form
over
normally
verify
we
launch
the
test
net
with
IP
specification
so
that
clients
have
something
that
tests
against
finalize
the
Casper
contract.
They
see
anything
that
comes
up
with
the
test.
Informal
analysis
set
block
numbers
release
on
tests
main
that
generally,
a
few
questions
I
have
is
where
C++
tests
come
into
this
the
yellow
paper
and
KVM.
In
terms
of
like
does
that
stuff
happen?
K
A
Okay,
great
thanks
for
the
update
to
answer
a
few
of
your
questions.
Danny.
What
generally
happens
is
there's
an
EIP
that
gives
the
motivation,
abstract
specification
and
implementation
details
and
after
that
is
produced,
then
the
KVM
team
in
the
yellow
paper
team
generally
take
that
EIP
and
then
translate
it
and
to
pull
requests
towards
the
yellow
paper
and
kab
m
in
order
to
keep
up
with
the
EIP
as
developed.
That's
my
understanding,
Nick
I
think
we've
talked
about
this
a
little
before.
Is
that
how
you
understand
it
as
well?
A
F
A
K
A
Yeah,
so
it's
usually
how
these
things
go
down.
Testing
wise,
sometimes
tests
are
prepared
well
before
the
announcement
of
a
hard
fork,
but
they
get
shirred
up
as
we're
announcing
dates
and
stuff
and
giving
timelines.
So,
for
instance,
during
Metropolis
we
had
tests
developed.
You
know
over
the
past
six
months
because
we
knew
what
he
IPS
were
going
in
and
just
as
the
case
with
her
sorry
about
Byzantium
is
what
I
meant
we
have
Metropolis
coming
up.
Byzantium
with
that
instance,
we
had
test
cases
made
and
some
have
a
comment.
B
If
you
want
to
implement
a
new
test
and
you
test
VR
pcvue,
it
will
require
at
least
a
couple
of
major
clients
to
support
it
and
I
still
haven't
finished
as
a
protocol
itself.
So
if
you,
if
we
really
need
to
start
the
test
for
the
Casper
already
I,
think
it's
reasonable
to
work
there.
All
four
months
that
we
have
working
is
high
and
there
is
and
test
each
conceivably
client.
B
K
B
B
A
Okay,
great
sounds
like
we're
a
shirt
up
there
and
I
want
to
make
note
that
we're
trying
to
coordinate
here
between
you
know
the
testing
and
the
client
implementations
and
the
Kasper
research,
but
we
shouldn't
be
rushing
any
of
the
research
efforts
unless
absolutely
necessary
in
worst
case
scenarios.
Since
you
know
we
like
things
to
be
very,
very
assured
to
not
break
and
not
have
any
issues
as
much
as
possible.
So
I
think
we
can
continue
to
do
a
good
job
of
that.
M
It's
been
Edgington
from
the
pegasus
team,
pointing
back
to
your
yellow
paper
remark
earlier
so
I
raised
a
few
months
ago.
The
idea
that
a
IPS
or
two
contain
a
pull
request
to
the
yellow
papers
before
they
are
merged
or
accepted
and
I.
Just
wonder
if
that's
relevant
here
I
don't
to
derail
the
discussion,
but
just
wonder
if
they'd
have
been
thought
about
or
whether
others
feel
that's
a
good
idea.
They
needn't
been
written
by
the
author
of
the
EIP.
A
A
F
J
You
know
definitely
agree
on
that,
like
this.
Okay,
I.
Definitely
don't
think
that,
like,
even
if
we
once
too
hard
for
the
836
or
any
good
like
it's
any
kind
of
emergency
situation
that
deserves
an
arc
even
when
bringing
anything
things
ahead
of
schedule
or
like
bringing
or
I
bring
an
immediate
heart
fork
ahead
of
schedule.
D
A
So
my
take
on
this
from
a
community
perspective,
I've
been
talking,
I've
been
seeing
a
lot
of
discussion
back
and
forth,
and
it
kind
of
comes
down
to
the
people
who
feel
that
it
is
actually
as
a
security
risk
that
they're
not
entirely
sure
how
people
who
think
we
should
do
nothing
and
people
who
are
very
emotional
about
bidding
and
just
are
like
down
with
bit
main
down
with
a
six
stuff
like
that.
I.
Don't
know
where
that
comes
from
I.
Think
it's
like
a
cross
over
from
the
Bitcoin
community
or
something
but.
C
N
N
That
these,
these
particular
Asics,
which
are
probably
just
2.5
times
more
expand
5
times
cheaper
per
hashing
speed,
are
probably
not
much
of
an
issue
but
I.
Think
an
interesting
and
important
question
would
be
like
what
kind
of
development
would
require
an
emergency
response.
What
case
would
say
that
there
is
a
meet
word
transaction
which
does
not
actually,
which
is
not
actually
the
case
now.
J
So
one
thing
to
keep
in
mind
right
is
that
a
bit
the
any
ASIC
manufacturer
has
an
incentive
to
downplay
the
amount
of
influence
that
they
have
over
the
network,
because
they
know
that
the
warrants
won't
stay
out
over
the
network.
The
more
political
pressure
there
is
gonna
be
at
a
heart
for
clim
away.
J
So
it
seems
to
me
extremely
likely
that,
even
if,
like
that,
like
bit
mean
or
whoever
else
when
you
know
what
eff
to
pool
you
know
like
some,
you
know,
I
created
North
Korean
government's
like
some
people
in
Iceland,
whoever,
even
if
they
have
51%
the
chance
that
we
will
find
out
before
they
say
they
actually
start
doing.
A
51%
attack
is
quite
low.
So
it's.
J
J
It
seems
to
me
that,
like
just
seeing
someone
by
any
having
someone
buy
an
ASIC
and
being
able
to
see
it,
it's
not
a
sufficient
condition
and
a
51%
attack
actually
happening
clearly
is
a
sufficient
condition
and
they
don't
really
see
anything
in
bits
like
we
in
between.
There
could
be
used
as
a
flag,
so.
F
J
D
A
K
J
J
So,
okay,
in
that
case,
like
as
the
protocol
currently
exists,
what
would
happen
is
basically
that
a
51%
attacks
can
effectively
prevents
the
finalization
of
new
blocks,
but
it
cannot
prevent
cause
a
final,
but
you
cannot
break
finality.
One
simple
fix
that
could
probably
like
seriously
mitigate
the
power
of
a
51%
attack.
J
There
is,
if
we
just
add
a
consensus,
rule,
that's
where,
if
we
had
to
the
fortress
rule
that
a
had
like
it
should
be
preferred
if
it
has
a
larger
number
of
kaspar
votes
for
the
current
epoch,
you
would
inside
of
it
okay.
So
what
that
would
do
is
that
would
basically
say
that
if
there
is
some
like,
if
there's
an
attacking
mining
chain
that
refuses
to
include
phone,
so
then
there's
a
good
guy
mining.
J
J
A
J
So
what
the
eat
basically
does
is
it
says
that
we
added
to
decide
on
some
max
a
maximum
either
supply
cap,
and
I
recommended
once
120
point
4
million,
and
it
basically
makes
all
rewards
inside
of
ethereum.
So
this
would
be
current
proof-of-work
rewards
and
also
it
includes
a
commitments
to
be
call
future
rewards,
including
you
know
I
casper
floating
rewards
like
starting
call.
N
But
that
raises
the
question
that,
for
now,
smart
contracts
have
been
pretty
diverse
in
their
way
of
burning
eater
when
they
wanted
to.
For
example,
they
just
lock
the
top
somewhere
unaccessible.
But
if
you
have
a
limited
supply
that
mean
that
would
mean
that
there
would
be
a
formal
way
of
burning
ether
that
would
actually
remove
it
from
the
calculator
supply
like
something.
J
Yeah
and
like
there
are
different
ways
of
doing
that.
So,
for
example,
one
way
of
burning
supply
that
exists
in
the
casper
contracts
is
that,
if
I'm
in
the
task
for
contracts
the
if
a
fellow
dieter
gets
get
slashed,
then
their
penalty
is
basically
just
stay
inside
the
Casper
contract
as
part
of
the
Casper
contract
balance,
and
so
in
that
case
it
would
be
pretty
like
if
it
was.
If
the
Casper
contract
was
the
only
thing
issuing
rewards,
then
it
would
be
pretty
easy
to
calculate
because
the
Casper
contract
could
just
be
assigned.
J
You
know
a
twenty
million
if
and
then
it
would
set
any
rewards
to
be
proportional
to
or
20
million
the
balance
remaining
inside
of
the
contract.
If
we
want
to
do
something
more
standardized
that
we
could
just
do,
what
based
on
you
know,
exuberant
ease
by
sending
it
to
the
zero
address
and
then
rewards
I
could
potentially
even
be
paid
out
of
the
zero
address
and
proportional
to
the
amount
of
money
in
the
zero
addressed
or
you
know.
F
J
It's
really
worth
noting
right
that,
like
the
transaction
caused,
some
costs
in
in
aetherium
are,
in
the
long
run,
not
proportional
to
ether
fees,
right,
they're,
proportional
to
a
supply
and
demand,
and
so
like
in
general.
It
seems
like
they
go
up
when
it
goes
up,
because,
when
eastcoast
up
like
demand,
an
adoption
goes
up,
but
really
if
the
price
of
eath
were
to
go
up
for
some
exogenous
reason,
then,
which
would
happen
if
it
which
in
general
is
likely
to
happen.
J
That
stuff
tends
to
apply
much
more
when,
like.
First
of
all,
the
thing
that
you're
is
deflationary
is
the
unit
of
accounts
for
an
entire
Iike
for
an
entire
economy.
And,
second,
when,
like
it's
something
it's
something,
deflation
II
read
that
kind
of
a
long-term
sticky
prices
are
being
set
in
and
I
think
that
you
know
in
the
past.
It
was
definitely
the
case
that
ether,
gas
prices
are
stinky,
but
now
you
know
what
we
have
said:
Amex
key
calculation
for
pretty
much
everything,
and
so
I.
F
Didn't
mean
deflationary
spiral
necessarily
and
the
traditional
economic
sense,
but
in
the
sense
that,
and
if
we
assume
people
adjust
their
behavior
based
on
how
much
transactions
gonna
cost.
And
if
we
assume
that
there's
some
minimum
number
amount
of
fees
that
we
have
to
pay
in
order
to
secure
the
network,
then
a
high
transaction
costs
could
lead
to
fewer
transactions
which
leads
to
higher
transactions
notice.
J
So
I'm
suggesting
well
it's
so.
First
of
all,
like
the
mean
place
that
in
the
long
run,
the
kind
of
the
money
would
come
from
in.
This
change
is
from
a
lower
revenue
page
to
us
to
a
participants
who
provide
security
so
in
the
shirts
or
miners
and
like
in
the
long-term
Casper.
Validators
right
like
in
the
long-term
I,
do
think
that
no
a
Casper
validator
revenue
should
be
roughly
equal
to
DM
amount.
The
people
in
transaction
fees.
F
J
No
minimum
there's
like
a
level
right
like
so
like.
Basically,
if
we
have
a
smaller
reward,
then
you
know
instead
of
like
20
million
years
taking,
we
might
end
up
having
10
million
he's
taking
which,
on
the
other
hands,
though
like
there
is
a
risk
that
like
if
we
have
a
cryptocurrency
which
is
who
is,
which
is
means
inflationary,
then
that
could
lead
to
its
value,
dropping
which
by
itself
leads
to
a
less
capitalistic
securing
the
network.
J
That,
basically,
a
transaction
transaction,
a
few
levels,
are
capable
of
providing
enough
revenue,
it's
who
was
secured
to
a
secure
ablution
and
in
the
long
run,
if
we're,
if
they're
not,
then
like
there's
the
question
of
well.
How
about
how
valuable
is
this
system,
like
is
a
system
that
we're
building
in
the
first
place,
I
mean.
F
Personally,
I
I:
don't
think
that
transaction
fees
have
to
be
the
thing
that
supports
a
blockchain.
You
know
if
we
need
some
amount
exceeds
a
day
of
money
to
incentivize
miners
or
stay
cos.
You
can
take
that
from
inflation
or
fees
and
personally
I
think
it
makes
for
a
more
useful
system
as
we
take
it
for
inflation
because
it
imposes
the
cost
on
everyone
who
is
invested
in
the
system,
just
those
who
are
transacting-
and
it's
also
just
transacting
right.
J
If,
if
becomes
this
unique
token
inside
of
aetherium,
that
has
the
ante
privilege
of
being
being
inflated
to
pay
for
security
expenditure,
and
you
have
the
ability
to
just
like
print
out
ERC
twenties
on
top
of
aetherium
and
like
market
them,
and
these
tokens
don't
have
this
disadvantaged,
then
you
know
like
it
may
well
be
the
case
that
you
know
like
eventually
there's
going
to
be
this
tragedy
of
the
comments
or
even
though
eath
is
necessary
for
now
for
network
security.
No
one
wants
to
support
youth.
J
2%
of
2%
of
your
inflation
is
a
lot
it's
like
the
difference,
but
so
think
of
it.
This
way
right,
like
I,
believe
it's
the
case
and
that,
like
in
the
long
run,
people
expects
to
be
able
to
earn
something
like
4%
a
year
returns
reliably
from
the
stock
market
and
if
you
take
2%
off
of
that,
then
what
you're
basically
saying
is
that
the
amount
of
money
someone
if
someone
needs
to
retire
goes
up
by
a
factor
of
two,
because
you
go
down
from
4
percent
to
2
percent.
A
F
J
So
my
personal
preference
for
implementation
is
that
we
just
basically
pre
mine,
like
100,
max
cap
currency
supply
into
some
particular
address,
and
you
know
this
could
be
address
zero
and
we
just
basically
say
that
whenever
we're
paying
rewards,
we
subtract
the
balance
for
me
from
that
amount.
From
from
that
account,
and
then
that
accounts,
our
meaning
balance,
is
basically
by
itself
just
used
as
the
constant
multiplying
factor.
J
A
J
A
A
Okay
sounds
good:
well,
let's
just
scrap
this
for
now
and
bring
it
up
next
week
once
we
have
a
little
bit
more
feedback
from
the
community
post
meeting,
okay.
So
the
next
agenda
topic
is
AIP
process
updates
we'll
go
through
that
real
quick,
basically
Nick
made
a
really
badass
site
called
VIPs
dot,
aetherium
org.
That
is
automatically
generated
using
Jekyll
Nick.
Do
you
want
to
kind
of
just
go
over
that
real,
quick
and
also
your
automated
bot
sure.
F
I
think
I
gave
a
brief
overview
of
the
site,
the
last
two
lands,
but
the
basic
idea
is
that
it
takes
all
of
the
amount
down
files
for
the
heaps
and
automatically
indexes
them
by
type
and
category
and
generates
individual
pages
for
them.
And
so
as
soon
as
a
new
leap
to
the
repository
and
it
gets
updated
on
the
site
with
github.
F
Of
course,
any
time
you'd
ever
build
something
slips
through
and
in
general
I'm
trying
to
work
towards
set
up
where
editors
merge
drafts
sooner
and
quicker,
and
then
make
it
easier
for
authors
of
drafts
to
to
continue
working
on
changes
until
they're
ready
to
take
it
to
work
or
divs
or
a
standardized.
If
it's
not
a
quarry.
So
that
in
mind,
I've
also
written
a
bot
that
keeps
an
eye
on
EP
RS
and
when
it
sees
one
that
is
just
only
of
its
two
existing
IP
drafts,
it
will
allow
it
to
be
merged.
F
And
so
any
suggestions
on
remaining
the
process
are
welcome.
I
really
want
to
get
it
to
the
point
where
open
Piazza
will
require
action
either
from
the
submitter
or
from
any
that's
a
rather
than
the
current
situation,
where
drafts
can
survive
indefinitely,
and
also
to
get
it
to
the
point
where
drafts
of
misurkin
github
URI
for
their
Adeeb
almost
immediately,
and
that's
it.
For
me.
A
Thanks
Nick
I
also
wanted
to
point
out
that
there
is
a
pull
request
9:56
that
Nick
made
to
kind
of
clear
up
the
draft
process.
So
if
anyone
wants
to
give
feedback
on
that,
it's
basically
making
the
EIP
one
less
ambiguous
with
how
drafts
are
handled
and
it's
getting
a
good
amount
of
support,
except
for
from
some
people
who
think
that
they
don't
want
to
take
out
the
words
not
keeping
with
the
etherium
philosophy.
So
that's
one
of
the
points
of
contention,
so
anyone
who
wants
to
chime
in
there
feel
free
to
join.
F
Was
bringing
up
the
other
issue
is
debate
between
mostly
nicely
between
myself
and
Greg
Colvin,
about
whether
we
should
member
deep
drafts
immediately
or
only
when
they
propose
past
the
bath
stage.
I
wrote
the
book
with
the
idea
of
either
working
with
either
system
the
feedback
I
meant
appreciated,
and
for
anyone
who
wants
to
insist
on
keeping
the
in
keeping
with
the
etherion
philosophy
bit
I
would
invite
them
to
submit
a
description
of,
if
they're,
in
philosophy
that
they
can
get
a
hundred
percent
approval
on
and
if
they
can
do
that.
F
A
J
Guess,
on
the
sharding
side,
we've
been
just
looking
at
like
more
ways
to
make
the
like
the
entire
setup
of
like
how
I
of
how
collisions
work
and
alcohol
use
get
verify.
It's
like
somewhat
more
efficient,
since
no
panda
supports
a
implemented,
a
bunch
of
ways,
but
like
so
far,
it's
still
unlike
ideas,
information
I
mean
I,
expect
I
expect
the
implementation
algorithm
to
end
up
becoming
simpler.
After
the
end
of
all
of
it,.
A
Cool,
so
just
to
give
a
quick
research
update,
prismatic
labs
has
started,
recording
their
meetings
and
they've
also
been
releasing
blogs
on
their
updates
they're,
the
ones
who
are
trying
to
incorporate
sharding
in
together.
So
anyone
interested
can
look
that
up
and
then
James
Ray,
one
github
is
doing
drops
of
diamond,
which
is
a
rust
implementation
of
of
Casper,
I,
believe
and
he's
working
on
the
SMC
and
Viper
right
now,
and
someone
else
is
working
on
the
proposer
and
call
later
to
interface.
A
A
M
Know
just
to
add
Pegasus
to
your
list
of
people
working
on
shouting
with
us
quickly.
They
on
the
each
research
site
of
the
current
proposal
around
proposes
and
Co
laters
metallic
scissors
have
changed,
we're
waiting
with
eagerly
to
find
out
what
the
new
model
looks
like.
So
you
know
that
would
be
great.
A
Thanks,
Ben,
okay,
so
if
that's
it
for
research,
the
next
item
is
metropolis.
It
sounds
like
we
have
some
more
things
that
could
be.
Eip
is
especially
if
we
do
that
either
supply
cap,
and
if
we
were
to
do
that
before
Casper,
we
could
include
some
of
the
other
smaller
entities
we've
been
wanting
to
include
so
that's
a
possibility.
Now.
Does
anyone
have
any
other
opinions
after
last
meeting
thinking
about
it
about
when
metropolis
should
be
or
the
second
part
of
Metropolis
Constantinople
should
be
or
if
it
should
even
happen
before
Casper.
K
A
Could
actually
cancel
metropolis
if
we
were
not
metropolis
constants
in
Oakland
we
wanted
to.
There
were
some
eeap's
that
we
wanted
to
get
in,
though
so
I,
maybe
those
could
get
in
during
the
Casper
Fork
I.
Don't
think
that
would
be
too
difficult
to
kind
of
bundle
them
all
together.
Any
thoughts
on
that
anybody.
A
Okay,
cool
well
I'll,
put
it
on
the
agenda
for
next
time
and
we
can
bring
it
up
again
because
there's
not
like
a
super
rush
on
it
or
anything
to
bring
up
Constantinople
since
no
one
has
a
strong
opinion,
either
way.
Okay,
we're
gonna
zoom
through
client
updates.
Now
let
me
start
with
Jeff.
Is
anyone
here
from
guess
that
can
speak
speak
to
any
updates
or
important
news.
G
C
Yes,
we
:
to
neighbor,
wasn't
smart
contracts
been
pretty
smooth,
and
now
there
are
some
first
contracts
deployed
as
we
have
we.
We
had
some
issues
with
boxing
and
edit
now,
like
some
kind
of
walk
barrier
to
the
clients,
users
can
like
configure
a
minimum
block
to
war,
to
the
issues
that
many
outdated
snapshots
out
there
and
the
major
major
annoyance
that
people
walk
to
like
local
4
million
and
tends
to
have
to
synchronize
a
million.
I
I
That
and
actually
doesn't
sync
normally
like
it
forces
a
warp.
Sync
so
usually
like
there's
been
a
problem
of
you,
try
to
warp
sync,
it
fails
or
it
can
appear
with
a
snapshot,
and
so
it
reverts
back
to
normal
sync,
and
that
will
take
a
week
to
finish,
and
people
aren't
really
aware
of
this.
So
if
you
set
this
warp
barrier
now
like
actually
force
a
warp,
see.
F
O
P
There
is
no
major
updates
since
last
release
keep
working
on
Kasper
and
playing
in
the
next
release.
Now
one
of
the
things
which
will
different
could
be
included
is
proxy,
be
fine
tuning
because
during
last
profiling
of
import
I
know,
I
notice
that
reading
database
reads
are
taking
about
40%
processing
so
and
I'm,
pretty
confident
that
it
could
be
improved
by
Justin
rocks
to
be
options
and
parameters
so
yeah,
that's
all
I,
guess
no.
A
Q
Great
I'm
in
a
bit
of
an
unstable
internet
connection,
I'll
keep
it
super
brief,
not
a
whole
lot
to
report.
We've
been
doing
a
lot
of
work
on
EVM
to
azam,
which
is
sort
of
spyler,
which
would
allow
us
to
continue
to
run
legacy
EBM
bytecode
in
kind
of
a
la
CIMA
world,
and
so
our
current
kind
of
task
there
is
getting
state
tests
running
via
a
VM
to
Wasson.
We've
got
the
first
few
passing,
which
is
pretty
exciting.
Q
We've
been
fixing
some
issues
around
things
like
endianness
and
then
I'll
just
read
here:
I've
got
an
update
from
Everett
Hildebrandt
is
working
on
K
EVM.
He
says,
he's
been
going
back
and
forth
with
he
ohe
about
removing
some
tests
from
the
vm
tests
to
make
it
only
test
core
vm
infrastructure.
It's
a
good
step
in
the
direction
of
separating
the
etherium
specific
stuff
out
of
the
EVM.
In
the
same
way
that
you
know
the
e
separates
the
etherium
stuff
from
wasman.
He
was
him.
A
A
A
A
Okay,
we're
just
hitting
an
hour
and
a
half
so
I
had
on
the
agenda
that
if
we
had
time
we
would
get
to
e
IP,
908
reverbs
for
clients
and
full
nodes,
validating
transactions,
but
we're
just
hitting
the
max
time
for
the
meetings,
so
we'll
have
to
push
that
to
next
time
at
the
bottom
of
the
agenda
or
actually
in
the
agenda.
It's
item
number
10.
A
If
you
all
could
like
read
that
and
get
ready
for
next
meeting,
we
can
have
some
opinions
on
that
one
and
and
kind
of
talk
about
some
of
the
ideas
other
than
that
thanks
everybody
for
joining
today,
we'll
see
you
in
two
weeks:
I
might
not
be
there.
I
think
it'll
be
the
first
chord
F
meeting
I've
missed
since
January
2017,
but
I
think
if
I'm,
not
there
lanes
gonna
be
helping
out
so
go
easy
on
him
and
we'll
see
you
in
two
weeks
bye.
Everybody
thanks
had
some.