►
From YouTube: 🚀IPFS Core Implementations 2020-04-27 🛰
Description
Meeting notes: https://github.com/ipfs/team-mgmt/issues/992#issuecomment-619278770
For more information on IPFS
- visit the project website: https://ipfs.io
- or follow IPFS on Twitter: https://twitter.com/IPFS
Sign up to get IPFS news, including releases, ecosystem updates, and community announcements in your inbox, each Tuesday: http://eepurl.com/gL2Pi5
A
Hi,
everyone
welcome
to
the
ipfs
core
weekly
sink
from
Monday,
the
27th
of
April
2020
I.
Am
your
host
I
can
bring
on
the
Internet.
Now
it's
in
your
life
food
now
high
priority
initiatives
and
give
updates
on
what's
going
back
very
exciting.
So
first
one
is
upcoming
and
shipped
releases
of
big
news
is
Gloria.
Feldt
0.5
tell
us.
What's
going
on.
B
C
The
Tod
are
is
that
we
tried
to
monitor
and
try
to
get
particular
to
the
network
yeah
on
Friday
and
don't
like
a
Thursday
and
Friday,
and
it's
not
take
a
little
metrical.
We
feel
like.
We
could
still
make
this
work
in
theory,
but
it'll
take
more
work
to
get
some
extra
scale
up
to
where
we
can
actually
receive
all
the
captions,
because
yeah
one
support
constant,
providing
lots
of
content,
get
hundreds
of
thousands
of
connections
a
second
or
request
a
second
yeah
all
right.
C
First,
let's
go
on
so
the
field,
ers
new
code,
but
same
old
protocol
version.
That
means
the
performance
in
the
RCS,
unfortunately,
will
be
the
same
as
the
performance
on
live
network
until
upgrading.
So
when
we
cut
this
really
is
for
gonna
start
telling
people
please
upgrade.
Please
upgrade.
Please
upgrade
please
upgrade
and
beyond
that.
We
may
start
also
like
pushing
out
nodes
that
are
running
the
old
version.
C
So
example,
like
the
next
release,
may
have
some
rules
front
like
saying
hey
like
if
you
don't
appear
to
be
a
good
DHT
node
like
you're
running
an
old
version
like
ping,
you
you're
behind
in
that
I
fixed
your
routing
table,
and
it
appears
to
be
mostly
empty
Arecibo
bad
beers,
that
kind
of
stuff.
Then
we
just
like
droppers
in
traffic
table.
C
This
is
tricky
to
do,
because
if
we
do
too
much
of
this,
then
like
we
know
actually
breaking
Network
more
than
we
fix
it,
because
if
some
peers
have
a
view
of
network
where
they
see,
though
there's
a
network
emoji
or
something
feel
the
network,
they
don't
think
there's
a
network.
You
can
end
up
like
putting
and
getting
context
different
peers
yeah.
C
We
may
also
do
the
protocol
switch
in
a
future
release,
it's
necessary,
but
basically
pumped
it
up
for
now.
The
other
thing
you
have
to
say
about
der
by
five
is
that
we
have
a
release
party
on
Friday.
Please
check
your
calendars.
Everyone
is
invited.
It
is
a
virtual
release
party,
unfortunately,
but
mix
together.
Ideally
tasty
things.
Talk
with
Magnus,
hang
out
different
stations.
C
Well,
don't
get
annoyed,
so
it's
a
lot
of
people
don't
upgrade
then
they're,
probably
the
people
who
are
not
upgrading
will
continue
to
join
the
DHC,
even
if
they're,
behind
hacks
and
also
like
the
t's
not
properly
maintained
their
battery
tables
so
like
putting
bad
actors
I
feel
even
built
around
around
them.
So,
like
other
changes
we
are
making
in
the
next
release
should
help
with
this
a
bit.
C
One
is
like:
if
we
start
having
quick
by
default,
it
means
we
can
start
dialing
more
and
basically
having
more
parallelism,
just
change
how
we
use
our
parallelism.
The
DHT
served
like
these
notes
faster
again,
there's
a
saying
before,
like
we
could
have
good
notes
or
like
ping,
each
other
and
determine
like
if
like.
If
other
knows,
the
networks
are
like
misbehaving
it.
C
If
they
are
urgently
item
drop
them,
if
necessary,
for
like
really
old
nodes,
we
could
start
like
projecting
them
as
if
it's
Jeffers,
we
don't
want
to
do
that,
because
that's
not
very
nice.
That
means
they
can't
use
it
period,
but
we
do
have
some
notes
that
are
still
using
like
there
are
not
4.14
and
if
you've
ever
seen
it
up
here.
That
has
like
a
million
addresses,
usually
this
this.
This
is
far
and
out
comes
from
like
older
nodes
that
are
like
remembering
too
many
like
observed.
C
A
C
I
know
we
don't
see
Krause
right,
we
won't
see
any.
Do
you
see
a
gradation,
so
we
will
see,
see
will
operate.
The
GC
can
only
go
up
from
here.
Basically,
like
I
was
working
up
quicker
by
five
it'll
get
better,
and
the
certified
code
against
the
current
HC
is
already
much
much
better
than
the
journal
for
code
against
the
current
DHD.
So
like
we're,
definitely
seeing
lots
of
big
improvements,
but
when
it's
a
logical,
don't
upgrade,
then
it's
not
going
to
get
as
good
as
our
light
orchestra
house
saku-dono
to
test
this
thing.
C
Okay,
what?
If
everyone
is
running
video
everyone,
a
dirty
cook,
it's
great!
Yes,
like
not
everyone's
running
the
new
code
and
but
we're
all
using
same
protocol.
Then
it's
basically
as
good
as
like.
The
current
test
got
black
pepper.
Give
you
try
to
like
resolve
content
on
your
local
node.
Originally.
D
1K
pretty
stable,
a
2k
has
been
working
last
time.
I
tried
it.
We've
done
some
tests
on
the
live
network
as
well.
Obviously,
that's
it
is
both.
It
is
both
different
and
and
in
some
ways
more
to
get
it
and
more
indicative
of
what
it
will
be
on
the
during
the
release.
Now
that
we
know
we're
not
doing
the
protocol
rollover
most
of
the
test
ground
tests
were
done
either.
D
Unlike
pure
old
protocol
new
protocol,
we
had
some
that
were
done
with
a
mix
of
the
two,
but
there's
less
of
an
emphasis
there,
because
the
plan
was
to
do
the
rollover
turns
out
that
is
you
scale
the
network
as
more
people
upgrade
performance
will
improve
for
everyone,
not
just
the
people
who
have
upgraded.
So
even
if
you're,
selfish
and
don't
feel
like
upgrading,
because
you
have
some
independency
tell
your
friends
it'll
make
your
thing
bastard.
E
D
C
C
F
C
E
C
A
Coming
releases
for
jockey,
first
he's
getting
a
patch
release
and
like
eminently
which
put
some
questions
of
DHT,
the
DFC
API
over
HTTP
man,
I'm
just
trying
to
figure
out.
Why
wasn't
really
weird
test
Bailey's
in
CI
they
probably
and
possibly
so
it
looked
fairly
good
investment
and
there's
something
TC
tests
and
comedy
monkey
and
then
days,
whichever
manner
hash
one
man's
replicate
locally.
So.
B
B
After
the
auto-5
lunch
and
then
I
will
jump
into
noise
in
82
five
five,
four
nine.
These
were
its
separate
line
item,
but
their
content,
routing
related,
at
least
for
the
team.
So
I
pulled
them
up
into
this
for
now,
I
added
a
tracking
issue
for
DJ
s,
side
of
things
and
Jay
subdue
P,
so
I'm,
just
tracking
everything
there
for
both
of
those
because
I'm
just
creating
myself
a
crypto
Jay
s,
work
stream
to
fix
and
do
things.
So
if
you're
interested
following
over
there,
you
can
look
at
that
for
noise.
B
There
is
a
blocking
issue.
The
goal
of
PGP
Damon
needs
to
get
updated
with
the
latest
noise,
so
I
will
coordinate
with
somebody
to
make
sure
that
that
gets
done.
Yeah
and
then
my
goal
is
to
get
Interop
testing
for
noise.
I
would
like
to
I
give
s
testing
to
start
tapping
by
next
week
so
that
we
are
ready
will
in
advance
of
the
6rc
and
that's
the
update
there.
G
G
These
things
are
partly
because
of
the
250
ahead
thing,
but
when
we
scaled
them
down
to
50,
we
were
still
seeing
that
issue.
So
there's
something
up
that
needs
to
be
debugged
and
that
I'm
hoping
to
spend
a
whole
load
of
time
on
this
week
and
aside
from
that,
the
share
Postgres
provider
store
is
now
looking
stable.
That
means
that
all
hydras
that
we
have
so
not
just
hydra
heads
and
all
hydras
will
share
the
same
data
store
that
has
to
provide
a
records
in
it.
G
So,
rather
than
sharing
like
for
every
Hydra,
we
had
like
a
million
records
in
the
data
store,
but
now
we've
got
all
of
the
hydras
sharing
around
four
million
records
and
there's
no
duplicates.
So
that's
that's
good
and
also
having
a
separate
data
store.
Allow
us
to
have
a
persistent
in
datastore.
So
if
the
hydras
went
down
and
came
back
up
again,
then
the
data
would
still
be
there.
Previously
we
were
just
using
a
level
DB
in
a
docker
container
that
was
wiped
with
when
they
went
down
and
up
so
that's
good
and
so
yeah.
H
This
week,
I
post
the
J's
because
work
for
a
few
days
to
focus
on
IPS
desktop
to
be
ready
for
0.5,
which
will
ship
subdomain
gateways
to
test
abusers.
I
guess
that's
related
I
think
that's
it.
We
fix
a
small
UX
issue
with
directory
listing,
but
I
think
Stephen
scheduled
that
20.6
right.
So,
okay
yeah,
that's
it.
I
Cool,
that's
what
not
dates
yes,
so
bits
up
is
pretty
much
ready
to
go.
So
we
we
found.
We
had
an
issue
where
we
were
creating
sessions
for
every
single
block
and
that
resulted
in
us
broadcasting,
every
single
block.
So
that's
been
fixed
now
and
we
also
had
a
timing
issue.
So
we
were
getting
a
lot
more
duplicate
blocks
and
we
should
have
been
getting
so.
We've
also
fixed
that
so
I
think
that's
pretty
much
ready
to
go.
A
I
J
J
I
am
currently
chasing
one
more
bug
in
the
library
that
implements
an
async
read
at
least
interface
of
sorts,
which
is
celebrity
that
has
been
extract
already.
Chrome
dagger
also
am
also
finalized,
both
the
general
chunker
interface
and
the
general
lincoln
interface,
everything
kind
of
fell
to
the
other
ones.
J
J
There
are
a
couple
more
things
that
are
behaving
strangely
when
I
spin
this
on
a
vertical
system,
so
yeah,
that's
kind
of
what's
take
most
of
the
time,
but
yeah
looking
clicking
great
and
the
final
version
of
the
interface
is
actually
something
that
is
going
to
be
easier
to
integrate
in
the
white
resin.
I
originally
thought
so
Goodwyn
there.
We
can
talk
this
over
with
Steven
already,
but
it
might
be
that
some
of
this
stuff
will
even
learnt
in
207.
So
so
I
have.
K
Hey
folks,
over
the
next
few
weeks,
we're
gonna
be
moving
towards
UNIX
FS
support,
which
is
something
that
we've
got
from
some
community
people
that
want
to
use
rusty
FS,
particularly
not
in
the
I,
want
to
say,
embedded
but
not
like
firmware
but
more,
like
really
constrained
small
Linux
environments
that
run
on
like
industrial
IOT
devices
and
stuff.
So
that's
what
we'll
be
working
on
next
and
getting
that
support
in
and.
A
K
F
So
peers
are
ingest
the
milestone
which,
bearing
for
him
mobile,
is
not
complete.
All
the
peers
were
merged
in
the
Liberty
modules
and
they
are
now
released
as
bitter
for
a
time
being.
The
gesture,
PT
PPR
was
also
merged
in
the
0:28
branch
and
we
will
basically
release
all
of
them
at
the
same
time,
also
related
with
this
beautiful
removal.
F
Basically,
we
were
using
before
the
period
for
providing
the
listening
addresses
for
a
loop
HP,
which
was
kind
of
strange
or
not
that
cool,
so
I
created
also
related
to
this
an
address
manager
component
PR,
so
that
basically
we
can
provide
similarly
to
go,
live
HP
and
you
go
ipfs
listen
announced
in
no
announced
addresses
for
a
lipid
feed.
This
is
also
currently
being
reviewed
by
Jacob,
so
this
is
probably
merge
this
week
as
well.
Then,
in
the
milestone
tree
with
the
persistence
for
address
book
inverter
book,
the
initial
implementation
is
also
ready
for
hope.
F
You
I
think
there
are
a
couple
of
open
questions
that
we'll
need
to
make
some
decisions,
but
I'm,
confident
it
will
probably
get
these
merge
this
week
as
well.
So
also,
meanwhile,
this
week,
I
will
start
to
work
on
the
fourth
in
the
last
milestone
of
these
initial
improvements
for
the
piercer,
which
will
include
the
cubic
and
also
integrating,
let's
make
a
chain
on
it,
and
that's
it.
A
A
A
So
it's
the
timeouts
are
done
at
the
very
at
the
call,
so
the
CLI
and
everything
calls
into
the
call
time
answer
there.
But
the
signal
is
passed
all
the
way
down
to
at
least
the
bookstore
right
now,
and
so
there
is
Piazza
been
emotional,
IPL
t
and
the
blobstore
and
then
the
next
movie
two
bits
after
that.
A
D
You
know
I'd
like
to
get
feedback
on
both
like
the
general
idea
and
implementation
for
how
to
do
sort
of
a
gradual
DHT
protocol
rollover.
Instead
of
this,
like
switch
everything
all
at
once,
in
particular,
I
want
to
have
a
like
effectively
a
Testament
that
will
operate
and
be
compatible
with
the
existing
networks
so
that
we
can
get
people
testing.
D
A
A
A
A
Question
from
the
beat
with
the
HTV
one
versus
V
to
the
multitude
of
plans
that
we
had
over
times,
present
epoch
engineering
war
story
and
something
that
is
not
trivial
to
understand
what
level
of
detail
where
we
explained
it
to
users,
we
be
sure
they
don't
interpret
it
as
we
are
breaking
backwards.
Compatibility.
I
E
D
D
C
Yes,
with
small
ambiguity:
if
the
old
nodes
are
on
the
same
private
network,
but
don't
have
public
addresses,
then
in
the
past
you
would
route
requesters
best
question,
though
using
the
private
to
work,
but
that
was
broken
anyways
because,
like
you
would
until
house
it
will
try
to
connect
his
private
notes.
That
would.
E
C
C
It,
but
to
have
a
sort
of
a
break
where
they
would
the
old.
The
new
DC
was
a
subset
of
the
old
DHT.
That's
it
like
it
wasn't
it
was
overlapping
enough.
Is
it
any
put
the
old
age
to
it?
Oscar
the
new
da
seen
any
puts
the
DHD
would
always
be
something
the
old
eh
D
yeah,
but
we
couldn't
get
to
the
point
where
we
could
have
I
could
pick
it
a
subset.
Really
do
that
properly
sure.
E
E
What
if
there
was
like
a
server
mode
like
an
infrastructure
mode
where
we
could
like
enable
in
our
bootstrappers
our
gateways
like
if
you're,
at
cetera,
where,
if
you
have
a
node
connecting
to
you
and
you
can
see
that
you
cannot
buy
all
the
nodes
in
another
report,
because
that
node
is
behind
a
net.
You
will
not
include
that
nodes
in
your
DHT
routing
table,
yeah.
C
C
Up
doing
we
put
it
in
there,
so
we'll
get
them
to
upgrade
now
we
might
actually
do
them
like
one
is
battle
back.
The
problem
is
like
we
don't
wanna
do
this
for
every
single
pair,
but
you
see
we
could
do
it.
Eight
like
if
we
decide
Mead
and
you
want
to
add
them
to
a
routing
table.
The
other
thing
we
can
do
is
like
we
can
make
queries,
I,
guess
summit,
a
like
hey,
like
these.
Your
routing
table
folk
appears
if
it
queries
the
charge
that
appears
that
we
got
from
you.
C
B
C
E
C
C
If
an
in
traverse
the
network,
three
difference,
it
appears,
I
will
try
to
put
or
get
from
a
different
set
of
peers,
which
means
that
you
won't,
like
you'll
benefit
grade
and
service
the
network.
So
when
you,
when
you
make
these
changes,
you
still
need
to
get
like
a
large
force
in
the
network
to
start
like
following
these
rules,.
D
It's
also
worth
noting,
like
some
of
these
rules
are
like
hard
and
we'll
have
to
like
figure
out
later.
In
particular,
China
is
hard
government.
Has
this
Great
Firewall
thing
that
makes
it
so
that
convictions
between
the
country
and
places
outside
of
it
operate
differently
than
connections
inside
of
the
country?
And
so
you
could
end
up
in
scenarios
where,
like
some
people,
think
other
people
are
diable
and
others
do
not,
which
could
lead
to
like
on
network
network
shapes
I.
Think
you
guys
that's.
D
We
have
a
yes,
so
you
can
law
there.
I
started
on
a
crawler
I
last
week
to
sort
of
graph.
The
network
distribution
will
has
started
on
some
crawler
stuff
to
show
like
where
all
the
nodes
are.
Hopefully
we
will
be
able
to
glue
these
two
things
together
and
then
we
can
see
if
the
network
distribution
in
shape
has
any
correlation.
The
notes
actually
are
and
then
we'll
get
a
better
picture
of
how
it
looks
sounds.
A
Good,
and
so
we
are
who
the
time
now
so
we
can
cool
this
museum
closed.
Thank
you
for
coming
everyone.
It's
been
the
I
profess
core
Luigi
stink
for
Monday,
April
28,
the
27th,
so
a
2020,
it's
been
amazing,
really
great
has
been
awesome,
progress
but
they're
very
much
looking
forward
to
go
up
here.
First,
one
point:
five
amazing
and
also
like
you
know,
enjoy
your
isolation,
haircuts
that
bloom
on,
because
we're
gonna
be
added
like
a
gel.
Some
crazy,
crazy
shapes
I'm
sure.
But
you
know
on
the
Internet.