►
From YouTube: libp2p weekly sync - 2019-04-22 17:30 UTC
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
A
B
B
So
I
think
Dirk
posted
an
issue
recently
about
following
the
spec
in
terms
of
making
sure
that
we're
actually
republishing
puts
and
I
think
I'll,
probably
look
at
bringing
that
up
as
a
topic
point
for
doing
that
with
provides
as
well
go
before.
We
talk
about
that
later
other
than
that.
I'll
continue
work
on
DHT
performance
this
week,
awesome.
C
Hey
guys
so
last
week,
I
helped
with
my
Cara
and
blocking
her
weight.
Some
issues
in
a
gossip
serve
mostly
in
the
publish
and
subscribe
shed,
was
airing
some
issues
there.
I
divided
and
I
find
out
what
was
the
problem
and
that
story
and
soap.
There
are
a
few
more
issues
and
I
will
tackle
this
week
as
well.
C
Then,
as
Jacob
I
spent
some
time
in
the
interrupt
tests
from
for
ipfs,
it's
related
with
with
HP,
there
are
like
a
penis
or
pub
so
I've
been
eyes
over
the
HD
in
the
DHT
whence
also
I
created
a
proposal
for
our
sink
its
writers
effort
in
JavaScript
for
the
interface
connection.
That
I
will
also
work
on
this
week.
So
this
week,
I
will
finish
helping
my
career
with
the
gossip
serve
up.
Fillet
I
will
have
will
have
all
the
tests
that
she
currently
is
working
perfectly
then
I
will
continue.
C
A
I'm
next
up
on
the
list
mine
was
pretty
focused
last
week,
was
like
really
heavy,
because
that
meant
had
some
pretty
big
wins.
There
I
have
like
dependency
ordering
now,
first
of
all,
schedule
tasks
in
phases.
If
a
given
task
depends
on
another
one
and
got
some
pretty
important
stuff.
Just
for
your
general
usefulness
like
registering
metadata
about
the
nodes
that
we're
spawning
in
the
test
lab,
so
you
can
now
access
their
pure
IDs
and
stuff,
which
is
important
when
you
want
a
bootstrap
to
like
the
gate
right
now
in
or
something
like
that.
A
Next,
up,
like
the
main
deliverable
for
me,
this
week,
is
going
to
be
a
demo
video
on
the
test
lab.
So
that
should
be
interesting.
Get
to
bust
out
my
video
chops
again
and
gonna
be
picking
up
with
a
game
this
week,
set
things
down,
as
I
was
doing
this
last
little
push
for
work
and
then
towards
the
end
of
the
week,
I'll
be
transitioning
to
a
little
sprint
on
the
self
dialing
issue.
Just
to
give
myself
a
break
from
chest,
Labs
stuff,
excuse
me,
and
that's
me
up
next,
we
have
Lisa.
D
Read
the
memo,
but
things
that
were
written
in
there,
we
also
fix
them
well.
I've
also
worked
on
a
couple
of
identify
improvements
her.
Hopefully
this
will
curtail
explosion
and
explosion,
and
he
was
always
have
dinner
and
up-to-date
addresses
for,
of
course.
Basically,
what
we
did
was
replace
other
addresses
reported
to
identify
knowledge
that
we
don't
keep
all
the
addresses
for
nodes
forever
around
and
we
schedule
a
periodic
identify
booths.
D
You
know
like
on
the
basic
cost,
whether
it's
and
that
retains,
so
these
two
could
help
with
the
disposal
and
basically,
we
also
made
some
improvements,
another
relay
which
is
free
of
Chi
Kings
that
went
back
and
forth,
and
eventually
we
added
you
know
like
some
special
logic
programming
as
well
as
body
advocates,
which
we
think
can
be
generalized
would
step
so
it
might
make
sense
generalize.
They
said
I'll
to
the
dialer.
We
could
or
something
like
that
in
order
to
deal
with
address
quality
issues.
D
Let's
discuss
offered
another
point,
so
next,
no
like
I,
want
to
do
something.
I
want
to
make
some
progress,
no
validator
pipeline,
in
fact,
because
there
is
an
open
issue,
let's
say
it
shows
there
is
a
performance
problem.
If
we
have
a
lot
of
messages
incoming
so
this
box
out
reusing,
that's
not
nice,
let's
see
whether
we
can
curtail
there
and
we've
also
been
discussing
with
roll
about
the
ground
refactoring,
where
I
want
to
start
hacking
it.
D
F
E
D
D
There
isn't
one
in
my
appearance,
be
just
introduce
the
dependency
it
section
in
there
in
the
constructor,
so
we
see
how
this
works
out
and
then
on
the
second
iteration
start
removing
the
conducts
from
our
services,
so
that,
if
you
give
background
staff
just
make
your
own
context
and
not
rely
on
external
context,
is
hurt,
support
in
closed
methods
with
on
stop
groups
and
then
we're
gonna
have
to
add
the
event
bus,
and
this
is
going
to
bring
more
changes.
You
know
in
our
services
because
we're
gonna
have
to
change
how
we
write
code.
D
You
know,
take
advantage
of
the
event
path
and
then
letting
time
eliminate
time
dependencies.
So
this
it's
gonna,
be
you
know,
I
get
for
a
long
deferred,
so
rather
break
it
down
in
pieces.
So
anyway,
I'm
gonna
read
doctor
outline
my
intentions
for
cracking
the
scenario
for
parent
code
base
in
order
to
bring
it
to
something
more
modern
and
are.
D
A
D
F
The
context
there
is
a
period
2.0
is
gonna,
be
using
gossip,
so
and
I'll
talk
about
a
bit
about
this
in,
in
my
update
as
well
cause
they're
going
to
be
doing
performance,
stress,
testing
and
so
on.
They
one
of
the
one
of
the
things
that
they
want
to
do
is
have
the
ability
to
offload
advantage,
easy
jobs,
because
they
have
this.
F
This
does
requirement
that
whenever
a
message
arrives,
they
might
not
have
enough
data
to
actually
validate
whether
that
message
is
breaking
up
being
gossiped
further,
so
they
want
to
be
able
to
spawn
in
a
synchronous
job
that
is
then
potentially
gonna
fetch
data
from
other
notes
to
acquire
enough
stage
event
validate
that
message.
So
it
really
is
a
fully
a
synchronous
pipeline.
F
D
F
But
discovery
concern
I
had
was
whether
it
would
make
sense
to
control
service
life
cycle
by
events
themselves,
so
having
services
publish
events
when
they
were
started
when
they
were
stopping
and
so
on,
and
services
that
depended
on
other
services
would
be
listening
for
those
events,
instead
of
actually
explicitly
modeling
a
service
tree
or
joining.
You
know.
A
G
It's
just
a
different
way
of
modeling
how
how
different
layers
interact
with
each
other.
So,
for
example,
right
now
we
have
trance
like
we
have.
You
start
with
a
bear
transport,
and
then
you
can
upgrade
that
by
adding
security
and
muxing,
and
then
you
can
connect
create
streams.
On
top
of
that,
so
my
my
thought
is:
we
can
generalize
this
a
bit
and
have
a
chain
of
nodes
that
are
owned
by
something
on
our
side
that
that
essentially
form
of
pipelining
that
data
flows
through-
and
this
is
still
very
preliminary.
G
This
is
just
the
design
I
came
up
with
last
week.
I
had
a
discussion
with
Stephen
about
it
on
Friday
and
I.
Think
it
didn't
go
as
well
as
it
could
have,
because
I
was
using
the
term
message
in
two
different
ways
that
are
somewhat
ambiguous.
So
what
why
I
was
thinking
of
messages
as
a
computational,
primitive
we're
essentially
coming
out
of
the
the
object
oriented
world
where
a
message
is
essentially
a
function
call
but
a
message.
G
So
so
the
point
of
what
is
this
whole
graph
trying
to
actually
represent
well?
What
it's
trying
to
represent
is
each
node
is
a
stateful
part
of
a
stateful
component
of
lit
p2p
that
transforms
data
in
some
way.
So
this
is
still
a
very
preliminary
design.
I'm
going
down
to
Rivendell
to
talk
to
Steven
about
it
more
today,
I
know,
Aiden
was
interested
I
think
initially
in
some
discussion
about
Avenger,
then
stuff
too,
so
I
think
I.
G
H
Stephen
Justin
I,
don't
know
how
it's
changed.
Implementation
side
absolute
because,
like
this
is
already
effective.
What
we
do
or
we
have
transforms
on
transforms
on
transforms
it's
more,
that
we
will
be
exposing
that
more
we're
like
currently,
it's
kind
of
ad
hoc
or
like
connector
swarms
are
parts
of
connections,
and
then
they
basically
have
one
transform.
That's
the
protocol.
You
apply
to
the
connection
yeah.
This
will
make
it
I
think
slightly
or
maybe
slightly
easier
to
do
it.
H
A
H
H
We
would
do
instead
is
we
just
cut
except
the
connection
and
then
just
let
the
other
side
tell
us
what
to
do
so
like
they
would
say.
I
want
to
speak.
This
word
left.
So
if
I
don't
speak
that
vertical
whatever
and
we
just
give
like
the
only
restrictions
we
play
so
well
before
security,
we
only
speak
these
verticals
after
we've
negotiated
security
protocol.
There
would
be
speak,
anything
you
want
and
if
you
don't
want
to
speak
of
all
the
flexor,
oh
well,
that's
fine.
H
G
F
We
move
before
we
move
topics.
This
looks
a
lot
like
the
architecture
of
nutty,
for
example,
apache
nutty,
as
is
not
a
patch,
is
not
an
Apache
project.
It's
Netezza
is
java.
Networking
stack
that
is
basically
about
the
foundation
of
a
networking.
Application
in
Java
nowadays
quarter
Java
and
it
is
completely
event-driven
and
is
based
on
pipelines
as
well,
where
you,
actually
you
know
it's
basically
at
the
heart
of.
F
It
is
just
an
event-driven
architecture
that
is
just
listening
as
close
to
the
socket
as
they
can
get
in
Java
and
is
push
a
you
know.
The
raw
networking
events
into
a
into
a
pipeline
that
is
just
doing
performing
stuff
and
feeding
it
into
one
another
until
the
chain
until
it
reaches
the
application
and
that
much
right
and
then
there
are
other
projects
that
are
probably
worth
looking
into
as
well
for
inspiration,
apache,
camel
and
apache
cxf
onto
projects
that
build
up
on
that
kind
of
architectures,
but.
D
H
G
H
F
We
think
that
Monson
stream
2.0
is
really
doing
two
things
and
we
need
to
t
couple
of
them.
One
of
all
is
train
both
connection
bootstrapping
and
the
other
one
is
protocol
negotiation.
So
we
really
want
to
think
about
connection
bootstrapping
in
one
like
in
one
paradigm
and
like
a
trauma.
Goshi
ation
and
obviously
this
is
up
for
discussion.
I
know
that
you
probably
have
thoughts.
F
What
is
connection
bootstrapping,
so
who's
driving
is
the
process
of
actually
performing
to
be
getting
before
you
get
to
open
before
you
get
to
the
place
where
you
can
actually
start
so
so
that
we
think
that
encryption,
and
so
the
upgrade
process,
sorry
upgrade
process
would
be
bootstrapping
right
before
the
upgrade
process.
Currently
we
do
use
multi-stream
select,
but
in
practice
there
are
a
number
of
requirements
coming
to
us
from
different
directions
like,
for
example,
TLS
1.3.
F
H
Like
this
is
basic
any
time
you
could
have
negotiated
particle,
that's
it's
the
same
thing.
It's
upgrading,
so
I
I,
really
don't
want
to
split
that
too
much
because,
like
there's,
no
difference
between
an
initial
like
it's
like
this
protocol,
that's
like
this
protocol
suck
the
protocol
and
a
later
on.
It's
like
this
protocol
is
like
this
particle
yeah
I
do
agree
that
yeah,
like
ideally,
would
seek
TLS
like
maybe
just
are
tweaking
TLS
with
nothing
out
in
front
I.
F
So
really
Steven
is
just
like
in
your
proposal,
it's
kind
of
like
taking
Siri
on
stream
right
and
then
thinking
about
it
and
sort
of
like
in
the
same,
like
you
know,
picture,
but
also
in
a
special
way,
because
nothing's
here
in
stream,
is
really
intended
for
that
bootstrapping
right.
So
so
we
want
to
think
of
it
and
like
wire
it
it
with
requirements
such
as
being
able
to
specify
you
know
encrypted
channels.
Let's
go
on
to
multi
adders
stuff,
like
that,
so
I.
F
Know
before
hats
like,
if
we
have
some
information
beforehand
of
what
the
peer
supports
may
be
by
using,
for
example,
and
DNS
records,
or
by
using
bluetooth,
discovery,
service
discovery,
or
you
know
other
mechanisms,
then
we
might
be
able
to
take
the
uncertainty
out.
But
so
we
will
need
to
provide
serial
serial
stream
in
some
way,
because
we
won't
have
this
information
at
all.
Right.
Yes,.
H
Well,
yes,
so
like
in
the
fast
path,
I
think
what
should
happen
is
like
you
know,
someone
speaks
to
us,
so
you
just
connect
to
the
TLS
end
point.
Then
you
during
the
TLS
handshake
you
give
them
like
the
protocol.
They,
you
exchange
your
both
a
flexi
particles.
So
then,
like
we
want
to
be
couple
things
for
Cle.
Just
do
it
immediately,
but
serial
stream
is
just
if
they're
for
situations
where
you
can't
do
this
or
potentially
pact
the
protocols
for
like
use
like
behind
it's.
F
A
and
then,
if
you
add,
for
example,
a
potential
noise
and
put
a
channel
into
the
picture,
then
it's
you
know
doing
that
push
data
is
even
it's.
It's
actually
realize
about
today.
If
not
that's
1.3,
you
still
have
to
wait
for
go
to
so
so
yeah.
So
we
kind
of
like
thinking
about
these
things
and
we
want
to
like
get
our
IDs
trade
before
we
start.
F
It
was
also
a
proposal
for
a
sequel
light
as
a
take
tester,
which
is
very
immaturely
just
an
idea
right
now,
but
we've
discussed
this
in
several
menus
and
I
wanted
to
put
it
out
there
and
also
this
is
this
relates
to
to
what
I
said
earlier
about
no
system.
If
theorem
2.0
is
gonna,
be,
is
thinking
about
using
also
sub
and
probably
there's
gonna
be
a
high
chance
if
they
find
that
gossips
gossip
stone
is
performant
enough
and
is
secure
enough
for
them
right.
F
So
there
is
a
there
is
a
team
within
the
iam
2.0
community
that
has
been
mandated
with
performing
stress
testing
on
the
go
implementation
of
gospel
stuff,
so
I've
set
up
we've
set
up
a
shared
channel
with
them
to
make
sure
that
we
support
them.
Are
you
would
be
yes
didn't
make
sure
these
EDT?
That
I
were
like
okay.
F
Already
have
their
test
lab
these
are
you
remember,
Paul
I
told
you
about
a
team
that
had
built
a
network
simulation
framework,
so
they
already
have
that
we've
been
negotiating
with
them
to
open
stories
that
they
are
open
sourcing.
It
I
think
I
have
private
access
to
it
already,
but
but
yeah
they
kind
of
have
their
own
test
lab
because
they've
been
like,
they
do
conduct
this
type
of
testing.
For
me
here
in
committee,
for
compliance,
Muslim.
F
I'll
bolster
them
the
documenting
in
a
few
minutes.
Let
me
just
finish
yeah.
So
next
I'm
gonna
be
investing
some
time.
Data
version
to,
hopefully
you
get
it
yeah,
that's.
Finally,
in
a
good
state,
because
users
actually
and
I
want
to
hopefully
get
the
abstractions
merge
pretty
quickly.
How
do
you
guys
feel
about
it
like
I
tell
them
if
s
is
the
right
place
to
ask
like
what
is
your?
What
is
your
sense
right
now,
I've
got
a
HUD.
How?
F
H
A
H
H
Yes,
it
wasn't
issue
affecting
Mars,
unfortunately,
but
updating
to
this
I've
brought
another
bug
with
a
new
Mars,
apparently
we're
like
so
we
recently
in
the
swore
we
recently
started,
storing
all
of
the
heiress's
that
have
a
lately,
one
of
them,
which
is
really
nice
for
good
best
cuz.
Nobody
actually
see
why
I
was
failing
instead
of
just
getting
at
random
error.
H
I
Yeah
I
hate-
this
has
been
a
great
great
discussion,
we're
at
time
so
and
the
Dean
is
still
has
to
go
so
I'll
be
very,
very
brief.
I'm
all
about
security
audit.
This
week
and
next
week
sometime,
we
have
more
time
I'll
explain
to
you
guys
the
whole
there's
like
a
whole,
going
to
be
a
sort
of
sequence
of
audits
that
will
culminate
in
a
in
a
public
report,
and
this
only
affects,
but
maybe
they
have
people
on
the
call.
I
E
So
just
I've
been
working
a
little
bit
of
like
benchmarking,
a
penis
over
pub/sub
or
persistent
pub/sub,
it's
sort
of
predictably
fast.
Once
you
find
the
nose
that
you're
supposed
to
talk
to,
which
sort
of
then
brings
us
to
discovery
and
and
rendezvous,
and
all
of
that
sort
of
stuff
I'm
not
trying
the
grading
with
the
rendezvous
sort
of
PR.
That
is
out
there
now,
but
also
may
have
some
time
to
help.
If
you
guys,
you
know,
need
some
help
moving.
Some
of
that
alone,
you
know
yeah.
E
You
know
some
some
asks
either
for
like
getting
someone
to
review
the
persistent
pub/sub
configuration
as
opposed
to
just
making
sure
I
didn't
break
anything
by
adding
configurability,
making
sure
that
the
new
thing
that
I've
added
is
actually
good
and
also
something
about
a
multi
address,
based
content,
routing
that
it
looks
like
there's
some
interest
in
for
my
PLD,
and
all
the
data
seems
to
be
in
the
right
place
already
on
the
DHT.
A
it's
just
a
matter
of
what
the
interfaces
look
like
on
the
go.
Implementation.
A
I
I
just
wanna
say
it's
it's
it's
not
just
that.
It's
also
that
we
have
external
people
who
don't
come
with
these
calls
like
the
theory
in
people
but
may
want
to
view
them.
So
it's
kind
of
promoting
YouTube
I
know
it's
a
centralized
platform.
It's
ironic,
but
yeah
I'm,
just
kind
of
promoting,
like
let's
try
do
both
of
them.
Yeah.