►
From YouTube: Filecoin Core Devs Biweekly #10
Description
Recording for: https://github.com/filecoin-project/tpm/issues/23
For more information on Filecoin
- visit the project website: https://filecoin.io/
- or follow Filecoin on Twitter: https://twitter.com/Filecoin
Get Filecoin community news and announcements in your inbox, monthly: http://eepurl.com/gbfn1n
A
So
that's
good,
perfect.
All
right!
Good
morning,
everyone
welcome
to
the
10th
filecoin
cordov's
meeting
on
the
28th
of
january.
Thank
you
all
for
being
here
our
usual
schedule.
Let's
get
started
with
with
updates
from
the
various
teams,
and
then
we
have
at
least
one
open
fib
to
discuss
and
q
a
and
any
other
any
other
conversation
people
feel
appropriate
for
updates.
Maybe
let's
get
started
with
sorry
this
week.
B
Thank
you.
So
during
this
past
two
weeks
we
were
working
on
establishing
a
file
cable
for
home
node
on
our
virtual
machine
and
make
it
interpret
the
messages
and
I
think,
to
the
current
hit
on
the
mainnet.
B
Single
things,
which
is
not
working
fast
enough,
so
we
have
realized
that
we
have
issues
with
how
purify
proof
is
performed
in
terms
of
like
speed,
then
we
have
a
less
significant
issue
with
single
actors,
so
it
approximately
takes
20
to
30
percent
more
time
than
us
to
interpret
messages
using
sigou,
but
we
are
also
thinking
that
there
are
some
different
reasons
which
we
cannot
see
yet
so
we
are
conducting
a
profiling
and
see
what's
going
on
in
terms
of
how
fast
things
are
working
during
the
interpretation
of
the
blockchain.
B
So
our
goal
is
to
make
our
solution
as
fast
as
possible
to
be
able
to
run
under
my
net
well,
because
the
current
place
is
just
not
workable,
so
we'll
we
are
currently
working
on
optimizing
stuff
to
be
at
least
not
as
bad
as
it's
now
and
well.
Hopefully,
we'll
make
it
we'll
be
able
to
make
it
fast
enough
to
in
reasonable
time
so
one
strategy
we
are
going
to
employ,
for
that
is
to
finish
with
native
actors
as
soon
as
possible.
Basically,
we
have
finished
with
video
actors.
B
Now
we
are
just
polishing
stuff
and
then
we
will
try
to
migrate
to
c
plus
plus
factors
fully.
At
least
we
do,
which
will
mean
that
we'll
have
to
we'll
be
able
to
synchronize
up
to
certain
head
at
least
and
then
see
what
will
be
the
performance,
and
maybe
there
are
some
other
drawbacks.
So
we
are
currently
looking
forward
to
discover
what
other
hidden
issues
between
the
performance
sticker.
I'm
on
the
performance,
degradation
yep.
B
That's
our
main
focus
for
past
two
weeks
and
probably
next
two
weeks
so
and
they're
fixing
some
small
issues
in
parallel.
A
Sounds
good:
what
is
the
current
rough
epoch
validation
time
like
each
upset?
So
if
you're
asking.
B
So
we
can
compare
it
only
to
so.
We
we
have
our
a
single
virtual
machine.
For
now,
it's
not
a
really
a
fast
one.
So,
but
what
we
have
done
is
we
have
established
the
lotus
nodes
there
and
for
the
lotus
node
it
takes
24
hours
to
sink
7000
blocks
all
chipsets
and
for
a
focal
node
node.
It
takes
about
20
times
more
time.
So
this
is
our
baseline.
C
Thank
you
cool.
Maybe
let's
hear
from
venus.
D
Yeah
hi
everyone
for
vilas
because
last
week,
because
we
and
I
think
two
weeks
ago,
we
have
the
first
many
node
on
the
mainland,
so
the
majority
yeah
for
the
venus
team
actually
is
to
maintain
this
node,
because
it's
the
first
one.
We
made
some
issues
and
we
fixed
them,
including
the
main,
the
block,
many
issue
and,
as
you
can
see,
we
we
have
about
more
than
50
database
and
power
and
yeah
we're
still
waiting
for
the
next
chance
to
produce
a
block.
D
This
is
first
thing
and
the
another
thing
is
that
we
are
enhancing
the
documents,
including
the
document
for
the
users
and
also
and
also
for
developers
and
at
the
same
time,
the
team
upgrade
the
speak
actors
to
v3
and
we
run
the
village
load
in
a
calibration
network
yeah,
because
the
calibration
network
is
not
so
stable,
so
yeah.
D
We're
still
continuing
to
working
on
that
and
another
feature
we
have
done
right
now
is
to
enhance
the
yeah
and
we
have
to
have
security
enhanced
by
adding
they
encrypted
crystal
so
that
and
yeah.
If
you
draw
a
venus,
node
and
yeah,
if
you
need
the
blast
for
yeah
gross
power
and
or
main
broker,
you
have
to
input
as
a
password
and
yeah.
We
have
the
crystal
actually
encrypted,
it's
a
security
feature
for
the
mining
pool
and
we
think
that
is
very
important
and
yeah.
D
We
was
that
the
architecture
design
of
distributed
and
mainly
pool
and
start
to
work
on
that
before
we
do
that
we
we
want
to
separate
the
components.
So
we
need
two
new
projects
and
yeah
to
work
on
this.
One
is
the
venus
miner
and
another
one
is
venus,
thinner,
yeah.
We
have
all
the
services
and
components
and
separate
and
making
yeah,
I
think,
to
make
it
and
understand
that
to
integrate
them
together.
D
Yes,
that's
all
yeah
for
the
new
project
and
added
into
the
file
coin
projects
and
yeah
and
and
and
I'm
not
sure,
I
use
or
money.
I
should
to
ask
to
create.
A
You
can
you
can
yeah
if
you
let
us
know
what
what
you
need.
We
can
make
it
happen.
You
can
just
okay
sure.
D
Yeah,
I
can
okay
and
I
can
send
you
a
message
about
this.
That's
good.
A
Very
cool
yeah
I
mean
obviously
great
that
that
yeah
y'all
are
participating
on
mainnet
and
fixing
stuff
along
the
way
and
jennifer
remarks
in
chat
that
the
community
will
really
like
the
improved
security
of
the
key
store
as
well.
That's
something
that
they've
wanted.
So
nice
progress,
cool!
Let's
hear
from
forrest.
E
Hey
definitely
have
some
exciting
news,
which
is
that
we
are
both
syncing
with
interop
net
and
mainnet
now
so
interop
net.
As
of
two
days
ago,
we
were
syncing,
and
but
now
since
the
network
has
been
stalled,
we're
not
running
a
node
on
the
network,
but
but
still
really
excited
about
that
and
for
mainnet
we
started
from
a
snapshot
and
we
validated
about
20
000
epochs.
E
Now
in
the
process,
there
were
some
discrepancies,
but
we
fixed
these
and
have
been
saying
in
sync
since
yesterday,
which
is
awesome
and
also
we've
been
running
a
local
devnet,
and
this
is
helping
us
uncover
issues
with
our
nodes
and
so
right
now
we're
working
on
solidifying
peer
management,
but
yeah,
just
sort
of
chipping
away,
and
the
next
major
thing
I
would
say
that
we're
working
towards
is
just
the
audit.
E
So
we
are
scheduling
that,
hopefully
for
early
march
and
yeah,
I
guess
like
starting
to
also
think
about
v3,
so
we're
starting
the
v3
update
for
the
hampton
the
amp
and
then
after
that
at
some
point
do
the
actors
but
yeah.
I
think
I
think
that's
it
austin
anything
to
add
or.
F
I
think
just
generally
just
add
on
one
point
of:
is
that
our
node
definitely
isn't
production
ready,
yet
we're
just
like
working
a
lot
on
hardening
a
lot
of
the
stuff
now,
especially
the
networking
stuff.
But
I
guess
the
good
news
is
like
consensus
is
seem
to
be
going
very
smoothly.
Now,
like
there
hasn't
been
a
hard
fork
in
a
while,
and
the
issues
seem
to
be
all
resolved
and
the
validation
is
very
quick.
So
yeah
look
looking
good
so
far,.
A
Very
nice
that
sounds
like
exactly
where
you
want
to
be
good
job.
Folks,
yeah
from
the
from
the
side
of
things
not
not
too
much
to
update.
I
don't
think.
We've
mostly
been
working
on
yeah
the
actors,
v3
lotus
150
release,
which
will
update
actors
to
v3,
implement
the
optimistic
window
post
fip
and
have
a
couple
of
other
changes.
That
release
is
basically
ready,
we're
just
testing
it
right
now,
we've
put
it
on
calibration
net.
A
couple
times
found
some
fun
some
kinks.
A
Basically,
the
biggest
decision
we're
trying
to
make
right
now
is
whether
to
have
the
upgrade
before
or
after
a
bunch
of
east
asia
goes
on
holidays.
We
were
hoping
to
do
it
before,
but
like
we
might
might
set.
The
timeline
until
before
would
be
around
feb
8,
but
we
might
start
the
timeline
too
after
those
holidays,
which
would
be
sometime
the
last
last
portion
of
february,
so
trying
to
figure
that
out.
I
am
curious.
Does
that
affect
planning
for
folks
here
at
all?
A
Or
do
you
have
a
stake
in
this.
F
It
doesn't
really
matter,
but
just
like
we
probably
if
it's
done
sooner
than
later,
it
might
be
a
little
tight
to
for
us
to
join
v3.
But
it's
not
a
big
deal
like
if
we
just
won't
be
on
the
on
main
net
until
we
upgrade
b3,
because
we
we've
done
part
of
the
upgrade,
but
we're
definitely
not
in
a
position
to
be
joining
and
staying
in
sync
with
mainnet.
If
it,
you
know
changes
in
a
week,
you
got
it.
B
A
D
Yeah
for
villas-
and
I
would
think
that
yeah
we
hope
we
can
have
the
villas
upgraded
and
the
same
time
as
dota
stars,
and
we
have
done
some
integration
and
upgrade
and
running
on
the
recovery
network
and
we
think
we're
okay.
If
we
do
that
on
february
5th
yeah,
if
everything
is
fine
and
yeah,
it
works.
A
Okay
sounds
good
yeah.
Well,
we'll
keep
you
posted,
obviously
we're
just
trying
yeah,
we'll
see
where
we,
where
we
end
up
with
that,
it's
just
a
matter
of
how
much
confidence
we
have.
But,
apart
from
that,
I'm
trying
to
think
I
don't
think,
there's
anything
else
exciting
from
from
lotus
end.
Yeah
lots
of
ongoing
work
on
improving
the
deal
making
cycle,
because
you
know
that's
that's
kind
of
our
focus
for
the
year
is,
is
like
the
network
is
seeming
generally
healthy.
A
We
just
want
to
make
it
more
and
more
useful
and
easier
to
use
so
that'll,
that'll
kind
of
be
the
running
theme
for
at
least
the
first
couple
quarters,
but
yeah
that
that's
kind
of
a
an
evergreen
goal.
Yeah
did
anyone
else
from
from
lotus
and
friends
want
to
share
anything
cuba
nicola.
Do
you
all
have
any
updates?
You
wanted
to
share.
G
Nothing
major
on
our
end
this,
just
maybe
let's
look
in
the
future.
What
will
happen?
We're
working
really
hard
on
this
aggregation
of
proofs
and
we
have,
as
of
last
week,
unblocked
most
of
the
work,
so
we
we
did
find
a
way
for
which
provers
could
be
batching.
All
of
the
many
proofs
commit,
so
miners
can
batch
can
aggregate
most
of
the
proof
commits
together.
So
instead
of
posting,
one
message
per
prove
commit.
G
G
1000
cost
almost
as
much
as
costing
one
and
too
early
to
say,
so
we
still
have
to
integrate
this
into
the
protocol.
We
have
the
piece
of
technology
now
and
we
need
to
now
find
what's
the
best
way
to
integrate
into
the
protocol
and
what's
the
crypto
economy
impact
on
this,
and
what
I
think
will
be
important
is
that
solutions
like
this
wouldn't
have
just
one
solution
with
many
solutions
like
this
massively
reduce
the
gas
requirements
for
miners,
the
the
gas
cost
for
miners.
G
So
we're
working
hard
to
make
this
as
soon
as
possible.
I
don't
have,
I
don't,
have
a
specific
actor
upgrade,
but
this
is
the
major
update
that
we
have.
A
Cool
sounds
good
thanks
and
did
folks
from
the
foundation
have
any
updates.
They
wanted
to
share.
H
Yeah,
that's
good,
nothing,
nothing
bigger
than
I
said,
but
still
really
getting
up
to
speed,
but
this
is
great
cool
sounds
good
thanks
again
for
being.
A
Here
alrighty
next
up,
maybe
we
have
cooper
here.
Let's
talk
about
fip
11..
This
is
a
draft
ship
that
was
opened
a
while
ago.
If
you
can
just
go
over
kind
of
the
the
idea
today
and
then
we
can
think
about
it
for
for
a
while
and
raise
concerns
and
questions
as
needed.
I
All
right
so
fit
11
or
the
title
is
removing
a
reward
action
from
reporting
consensus
faults.
So
currently,
when
you
report
a
consensus
fault
when
you,
when
a
consensus
fault
happens,
as
in
someone
produces
consensus,
fault,
then,
based
on
that
timing,
a
dutch
auction
is
started
on
chain.
It
doesn't
really
start
it.
It's
it's
implicit.
I
So
what?
What
is
that
dutch
action?
That
action
is
an
auction
where
the
normally
the
price
decreases
over
time?
In
our
case,
the
reward
increases
over
time.
The
aim
of
it
initially
when
it
was
introduced
long
long
time
ago,
before
even
actress
v0,
was
to
reduce
the
contention
on
chain
about
reporting
consensus
faults,
because
back
then
the
reward
for
reporting
consensus
fault
was,
I
think,
10
of
total
minor
collateral
and
the
miner
was
completely
slashed
as
in
it
cannot.
I
Its
sectors
are
completely
removed
from
from
from
the
chain,
which
meant
that
rewards
were
very,
very
high,
which
required
this
this
auction
mechanism
to
to
make
it
such
that
not
everyone
like
jumped
right
on
it
as
it.
It
was
super
like
you,
you
were
looking
at
gaining
like
thousands
of
falcons
right
from
from
reporting
consensus,
faults
right
now,
the
right
along
the
way,
from
very,
very
early
allotus
to
where
we
stand
today,
the
reward
from
reporting
the
census
falls
was
changed.
I
It
changed
from
ten
percent
of
total
minor
collateral
for
that
minor
to
a
quarter
of
a
block
reward,
which
is
value.
That's
about
300
to
a
thousand
times
lower
than
you
were
looking
for
at
in
comparison,
and
the
argument
is
that
a
reward,
a
dutch
auction
doesn't
make
sense
for
for
that
law
of
reward.
It
just
means
that
the
consensus
faults
are
reported
way
way
later
than
we
would
like
them
to
be
reported.
Just
just
why
we
are.
I
I
To
for
for
gain-
and
not
so
far,
mostly
we've
seen
consensus
faults
be
caused
by
someone
disabling
consensus,
fault,
filter
and
in
their
minor,
but
for
security
reasons
we
want
to
see
consensus
faults
reported
as
early
as
possible
because
it's
possible
to
gain
an
advantage
by
producing
a
consensus
fault
and
the
longer
this
consensus
fault
starts
stands
unpunished,
as
in
the
miner
isn't
stopped
for
mining,
which
is
one
of
the
penalties
from
from
consensus
faults,
the
bigger
the
possible
advantage
and
harm
to
the
network,
a
minor
producing
consensus,
false
can
can
do
and
in
essence,
you
get
an
advantage
over
the
network.
I
You
you're
more
more
likely.
You
can
try
to
be
more
likely
in
producing
blocks
you
can.
You
can
be
more
likely
in
getting
your
blocks
included
in
the
canonical
trial
and
so
on.
So
the
argument
is
to
remove
completely
the
dutch
auction,
which
simplifies
the
code.
It
just
removes
a
big
chunk
of
code
that
specializes
in
that
and
always
pay
out
the
reporter
the
maximum
reward.
We
were
pairing
paying
out
so
far,
which
is
a
quarter
of
block
reward.
I
So
currently
it's
currently
the
the
reward
raises.
Gross
exp,
reverse
exponential,
no
exponentially
reverse
exponentially
from
zero
to
from
almost
zero
to
from
sorry
from
point
one
percent
of
the
reward
of
the
maximum
reward
to
the
maximum
reward
over
the
period
of
508
bucks,
which
means
that
it's
currently
not
profitable
to
report
consensus
falls
before
70
epochs,
which
is
like
30
minutes
and
and
that's
the
break-even
point,
given
the
gas
fees
and
then
like
to
gain
like
even
one
falcon.
You
have
to
wait.
Another
120
epochs.
A
Make
sense
also,
do
you
have
rough
stats
on
exactly
how
frequent
consensus
falls
have
been
on
midnight
so
far,.
I
I
don't
have
like
precise
stats
but
like
if
I
had
to
say
we
had
like
20
consensus
faults
since
the
start
of
the
magnet
or
something
like
that
there
were
more
consensus
faults,
but
so
there
were
more
consensus,
faults
committed,
but
usually
they
were
within
900
epoch
window
and
you
cannot
report
multiple
consensus
defaults
within
the
same
like
window.
So
if
you
report
one
consensus
faults,
then
it
deactivates
all
the
future
ones
for
900
epochs.
I
So
yeah
this
also
combats
this,
because
what
in
essence
happens
is
someone
starts
committing
consensus
faults
and
they
keep
committing
it
until
you
report
them
and
they
can
commit
as
many
as
they
want
in
that
period.
The
aim
is
such
that
reporters
always
report
right
after
this
and
a
high
gas
premium
message
with
high
fee
cap
right
after
the
the
consensus
fault
was
committed.
A
The
yeah
we
discussed
it
on
the
flip
sweeper,
so
yeah,
sorry,
that
sounds
good
cool.
I
forgot
to
mention
that
raul
from
from
testing
and
test
vector
fun.
I
was
available
to
join
today,
but
he
said
he
didn't
have
any
super
exciting
updates,
but
if
anyone
has
questions
for
him
feel
free
to
share
them
here
and
we'll
take
them
async
in
the
film
planters
channel.
A
B
I
have
a
question,
so
do
you
have
any
updates
on
interop
net?
A
A
Basically,
in
short,
it's
been
a
long
while,
since
we
spun
up
a
new
network
and
had
to
maintain
it
and
some
stuff
has
gone
so
some
of
that
code,
it
looks
like
is
a
little
out
of
date
because
since
we
had
mainnet
and
the
latest
calibration
that
we
just
haven't
had
to
do
that,
we're
working
on
it-
we'll
probably
dedicate
some
time
to
it-
hopefully
next
week,
because
it
shouldn't
be
too
difficult
but
yeah
right
now
the
network
is
having
a
lot
of
difficulty
staying
alive
and
it's
it's
like
it's
not
an
infra
thing.
I
Yeah,
it's
it's
one.
One
part
is
proof
proof
thing
where,
with
small
sectors
you
there's
a
lot
of
proving
the
challenge,
windows
are
shorter
and
so
on,
and
it
turns
out
there's
a
gpu
lock
somewhere.
That
is
being
very
contented.
Even
though
you
have
no
gpu,
it
doesn't
show
up
unless
you're
running,
like
15
15
proofs
in
30
minutes,
which
normally
you
don't
and
yeah.
A
F
It
doesn't
really
matter
to
us.
I
mean
it's
just
beneficial,
especially
it's.
It's
a
lot
more
important
for
us.
If
the
v3
upgrade
happens
like
next
week,
because
then
it
would
just
give
us
a
network
to
sync
against.
But
if
that
doesn't
happen,
then
it
doesn't
matter
we're
just
using
mainnet
to
test
and
verify
a
lot
of
stuff.
A
Have
a
fun
one:
what's
if
you
had
to,
if
you,
if
the
three
teams
had
to
like
guess
the
date
when
you
know
they'll,
have
a
note
on
maynet
and
it'll
stay
alive,
no
problem
without
anyone
monitoring
it
for
at
least
a
week.
When
do
you
think
folks
will
be
at
that
point.
J
Oh,
that's
a
tough
one,
we're
quite
close
because,
like
our
node
was
like
running
for
like
many
many
hours,
and
I
I
think
so-
we
don't
have
any
consensus
issues
right
now,
but
there
are.
It
appears
to
be
something
in
our
syncing,
algorithm
or
somewhere,
some
sort
of
scheduling
thing
where
we
may
potentially
have
some
lock
issues.
J
F
It's
not
replicatable
and
it's
like
it
happens
every
like
if
the
node's
running
for
like
12
hours
at
a
time,
it'll
randomly
do
it.
So
it's
it's
going
to
be
a
fun
one,
but
I
mean
like
if
I
had
to
throw
it
at
a
dart
of
the
board,
maybe
like
a
couple
weeks
and
we
might
be
staying
in
sync,
but
it
just
depends
on
again.
If
there's
the
v3
upgrade,
then
it's
definitely
gonna
be
a
lot
longer.
A
F
B
Yep
for
us
realistically,
I
would
say
about
a
month
so
we'll
spend
probably
one
more
week
on
figuring
out
what
what
are
the
reasons
since
we'll
spend
some
time
on
fixing
the
issues
and
amman
seems
to
be
a
realistic
estimate.
Nice.
D
D
D
We
hope
that
the
code
is
yeah
kind
of
stable
and
we
could
have
the
block
rewards
to
compromise
all
this.
J
A
J
A
Yeah,
okay
sounds
good.
I
think
we
can.
We
can
wrap
up
here
thanks
thanks
for
the
conversation
but
yeah
just
based
on
that.
It
kind
of
sounds
like
within
the
scope
of
two,
maybe
three
of
these
meetings.
We
might
all
be
at
a
place
where
we're
seriously
talking
about
either
already
being
on
mainnet
or
joining
mainnet,
and
you
know
trying
to
get
various
groups
to
start
diversifying
and
using
different
implementations,
which
is
great
it's
great
for
security,
it's
great
for
satisfaction
for
all
of
us
and
for
all
the
work.