►
From YouTube: 🚀Core Implementations 🛰 2019-07-29
Description
Call Description: https://github.com/ipfs/team-mgmt/issues/992
Notes: https://cryptpad.fr/code/#/2/code/edit/qbcGi7JVeZnzMUJkT0ZlBh-+/
B
Hey
yeah,
okay,
so
benchmarks,
the
old,
not
old,
at
all.
Actually,
the
benchmarks
or
ipfs
top
team
stopped
working
a
few
weeks
ago
and
has
since
been
resurrected.
It
was
it
got
restarted
for
some
reason
and
there
is
a
piece
of
thing
to
to
just
make
sure
that
it
starts
up
everything
that
should
be
running
when
it
comes
back
online.
B
But
one
of
the
things
that
is
in
that
video
is
me
talking
through
the
benchmarks,
where
a
site
and
how
it
works
and
what
infrastructure
is
there,
and
so
the
kind
of
I
guess
to
summarize
that
they
exist
and
they
have
been
focused
on
largely
on
like
Jeff's
ipfs,
and
they
can't
they'd
work
very
well.
If
you
would
like
to
write
a
benchmark-
or
you
have
like
some
some
idea
for
a
benchmark-
that
you
need
to
see
because
it's
something
you're
working
on
and
then
please
come
and
have
look
at
that.
B
B
All
right
I'll
do
that
later
anyway,
so
the
what
I'm
doing
now
is
I
am
just
looking
at
the
benchmark
time,
seeing
what
what
we
can
do
to
make
it
more
useful
to
go
ipfs
because
largely
focused
around
j,
like
the
best
I'm
going
to
be
working
on
that
and
hopefully
came
up
speed
so
that
it
it
will,
it
will
be
a
useful
resource
for
go.
I
profess
as
well
as
json,
and
so
that's
been
fun.
B
D
C
E
A
E
It's
sort
of
like
dog
food,
a
little
bit
delegated
through
thinking
and
by
the
jsf
EFS
training
in
brave,
with
this
0.2
branch
and
seems
to
work
much
better.
There's
like
the
number
of
requests
dropped
significantly
and
I'm
working
on
similar
fixes
for
HTTP
dns
lookups
over
HTTP
and
preload
notes
in
j
sacrifice,
and
I
should
run
that.
But
that's
not
related
to
take
it
through
yep.
F
All
right,
so,
regarding
gossip
sub,
I
wasn't
writing
it
with
difference
last
week
and
I
did
the
problem
with
the
Jessup
FS
test,
which
I
worked,
and
basically
the
vessel
implementation
was
not
supporting
self
subscribe
in
a
way
that
you
subscribe
and
that
your
angular
and
then
you
publish
it
in
your
received
through
your
handler,
is
jessica
fastest
expect.
So,
basically,
I
made
peers
for
that
in
gossip
surveilling
that
as
a
optional
parameter,
so
that
we
can
use
that
in
argumentation
and
from
the
chain
safe
guys.
F
They
basically
can
use
that
as
they
intend
to
and
I,
also
added
that
that
option
to
for
serve
as
well
to
be
similar
in
the
way
we
use
both
and
so
now,
I'm
just
blot
on
the
gossip
sub
release
on
their
ends.
In
order
to
get
it
subtitle
in
package
action,
and
we
can
merge
stuff
Jacob
all
already
reviewed,
Jessup
its
PPR,
and
it
should
be
good
to
go
once
they
release
the
package
and
then
I
will
ping
Alan
for
the
review.
F
A
Okay,
I
I
much
the
only
no
understood
the
main
gateways.
I
have
like
feedback
from
that.
The
the
poppy
team
on
using
PRI
DS
as
see
IDs
or
so
I
can
already
para
to
see
these
macros
I
store
them
in
SC
IDs
in
text,
Vasco's
already
comic
illness,
but
I
guess
no
one
else
here
is
actually
up
on
that
team.
So
actually
I'm
taking
my
fatzke
has
not
Jacob.
Has
we.
G
Yeah
so
I
started
the
interface
migration
for
lib
p2p
ransom
issues.
I'm
just
trying
to
make
that
same
for
people
to
require
should
be
able
to
finish
that
this
week
and
I've
got
a
PR
out
for
starting
with
the
switch
pulling
that
into
J
slippy
to
be,
instead
of
it
standing
alone,
I'm
just
trying
to
work
on
avoiding
polluting
the
J
slippy
to
be
commit
history,
because
we're
pulling
in
so
many
modules
that
could
get
nasty
pretty
quickly,
but
also
trying
to
make
sure
that
we
maintain
the
contributors
as
part
of
that
pull
over.
G
G
One
of
the
things
that
we
talked
about
is
doing
a
SDP
exchange
over
circa
relay
connection.
So
one
of
the
things
I
want
to
try
to
look
at
this
week
is
getting
a
proof
of
concept
of
that
SDP
exchange
happening
over
circuit
relay
and
then
work
on
flushing
out
the
spec.
For
that
I
think
there
are
variety
options
that
we
can
look
at
for
that,
but
that
seems
to
be
the
the
quickest
win
that
we
can
get
is
going
over
a
circle
relay.
I
So
mostly
still
working
on
pub/sub
PRS
so
get
to
this
arc
and
I'm
going
to
be
moving
some
of
the
the
name
sis
code
into
its
own
package,
so
that
people
can
use
it
outside
of
talk
of
us.
If
you
have
thoughts
about
that,
I
think
the
issue
I've
also
been
doing
yeah
so
back
to
the
purpose
of
PRS.
I
realize
that
I'm
not
sure
I'd
like
you
know
detailed
the
high
overview.
I
What's
going
on
implementation
wise
for
anyone,
j
s
land
who
is
interested
mostly
there's
gonna,
be
some
notifications
for
what
peers
join
and
leave,
join
and
leave
a
channel.
We're
gonna
have
discovery
being
called
internally
and
we
have
this
Golub
p2p
pub/sub
router,
like
a
value
store
on
top
of
pub/sub,
that
we
can
use
for
things
like
in
US
and
that's
gonna
have
a
custom
protocol
for
asking
for
the
latest
updates
for
stuff.
I
J
So
the
status
on
some
of
that
stuff
on
my
end
is
kind
of
halted
at
the
moment.
Just
because
we
need
to
come
to
some
agreement
like
calm,
discuss
some
of
that
stuff
and
figure
out
how
it's
going
to
work
across
projects.
That
is
just
a
conversation
we
need
to
have
specifically
I
think
you're
going
to
be
in
with
that.
So
we
can
talk
about
it,
I'm,
Hugo,
I,
don't
know
if
you
have
anything
to
add
to
that.
K
Right
now,
I'm
still
going
through
some
pr's
and
need
to
line
on
this
to
properly
shape
everything
and
when
I'm
ready
to
actually
shape
the
router
I
will
probably
set
up
meeting
everyone.
Interesting
I
mean
I've,
been
asked
to
figure
out
everything
we
need
to
this,
the
most
important,
important
train
and
then
meaning
there's
going
to
be
out
to
handle
multiple
roads
in
a
fraction
endo
on
Jess.
K
The
coordination
issue
we
have
a
couple
of
things
in
the
IP
Ness,
the
English
specs
before
I-
think
again,
I'm
right,
I
think
I'm
right
about
it.
But
it's
not
like
formal
thing
about
like
integrating
everything,
just
some
thoughts
that
we
dumped
in
there
about
naming
and
about
multiple
errors
and
everything.
But
we
will
probably
start
from
there.
The
discussion
is
how
to
intuit
everything.
I
D
Think
it'd
be
really
useful
if
we
took
if
we
issues
ruffling
current
state
of
things
and
what
needs
to
happen
next
and
because
I
think
we
might
end
up
in
a
state
where
we're
not
clarifying
things.
Sorry,
I
can't
do
this
sorry
I
drives
me
crazy,
I,
don't
know
what
it
sounds
like
for
you
guys,
but
like
I.
D
Just
keep
hearing
like
a
blip
of
myself
BAM
to
do
reflects
like
the
current
state
of
things
and
what
what
needs
to
be
done
left
because
right
now,
I
think
the
the
checklist
there
is
like,
maybe
a
little
bit
out
of
date
in
the
2:05
issue.
So
if
we
can
clarify
like
the
next
steps
or
what
the
question
is
that
we
need
input
on
I
think
we
can
get
you
guys,
a
better,
better,
unblocked
versus
just
knowing
that
there
is
some
decision
that
has
to
happen,
but
not
having
clarity
on
what
it
is.
J
K
Only
through
the
things
we
need
on
each
side,
especially
like
me
on
the
DNS
side
and
j
SI
anna
de
mon
of
subside
on
the
go
side
and
after
that
we
need
to
either
start
with
implement
in
pubs
or
on
Jas
or
dnas
on.
Oh
then,
we
need
to
work
it
out,
but
right
now
we
just
finish
our
work.
Two
hours
ago,
I
go
pops
up
and
Jas
DNS.
K
Okay,
that's
that's
that
one!
It's
fine!
It's
not
I
have
played
it
a
little
bit
as
we
did
the
okay
ours
and
that's
basically
up
to
date.
That's
the
stuff
I
need
to
do
before.
We
can
get
a
more
brother
conversation
about
implementing
of
savages
and
that
one
I
think
it
includes
getting
the
goal.
Dns
working,
you
would
assume
it's
not
there.
You.
K
A
Okay,
but
for
a
penis,
everything
has
to
be
really
nice.
Let
us
back
so
that
we
can
talk
about
it.
A
moment
like
it's
kind
of
like
reverse
engineering,
I
guess
fine
I'm
being
hypocritical
here,
because
this
is
have
JavaScript
was
built
just
missed.
Okay,
it's
reverse
engineering
from
GS.
Yes,
it's
hard
to
talk
about
it.
Yeah.
K
B
I
can
do
this
a
little
bit,
so
we
have.
We
have
Adam
on
the
call
who's
been
building
a
repo
migrator
for
Jas,
so
that
we
can
move
between
different
repo
versions
in
the
browser
which
would
be
super
cool
and
the
it's
this
close
to
being
ready,
ready
to
go
and
merge.
It
just
needs
a
few
few
polishing
I
think
and
the
natural
progression
of
that
work
is
to
then
write
a
write.
A
migration
for
the
repo
and
one
of
the
things
that
I
know
eventually
needs
to
be
done
is
migrate.
B
The
repo
to
using
multi
ashes
instead
of
C,
IDs
and
so
I
think
the
the
idea
of
putting
it
here
was
just
to
kind
of
raise
awareness
of
of
the
fact
that
this
now
exists
and
thank
you
Adam
for
sorting
it
out,
but
also
the
we
could
now
start.
We
could
do
this
migration
and
maybe
get
it
done,
and
so
the
idea
was
to
do
it
with
a
new
repo
version.
B
Have
a
migration
in
place
to
do
that,
and
we
just
wanted
to
check
with
the
NGO
side
of
T
of
things,
doing
a
reaper
migration
for
this
incrementing.
The
version
number
was
kind
of
a
good
way
to
go,
or
if
we
wanted
like
I
mean
an
alternative
could
be
to
do
it
like
you
know,
as
we
go
like,
we
continue
to
go
along
and
we.
A
Yep
we
would
like
to
do
migration.
There
are
two
outstanding
issues
that
we
never
really.
Finally
finish:
one
is
at
fist
raps
local
is
cred,
expects
return,
C
IDs.
We
have
to
make
the
mature,
like
raw
block,
see
IDs
so
pipette
to
retain
compatibility
order,
molting
hashes,
but
then
they
may
prevent
people
correct.
The
second
issue
is
that
provider
records
you
tease
wrong,
all
the
hashes.
A
They
correctly
you
see
IDs
and
not
just
a
change
where,
if
I
make
the
network-
and
it
may
bring
some
things,
but
luckily
90s
on
the
content-
is
still
CATV
zero.
So
it
was
just
magical
work,
but
the
longer
moving
on
this,
the
more
painful
things
get
so
we
should
do.
This
can,
unfortunately,
have
no
one
working
on
this
currently
go,
but
if
J
it
starts
meeting
on
this,
like
I'll,
just
make
sure
that
happens.
It
go
so
basically
one
side
to
start
pushing
on
this
and
then
the
other
side
I
think
we'll
just
follow.
B
A
There
is,
you
will
forget
the
first
you'll
forget
the
SE
ID
type
so
like
if
this,
if
the
data
store
just
stores
up
their
blogs,
there's
no
get
plots.
You
don't
know
who
that
codec
anymore,
so
some
walking
to
destroy
sir
yeah
Gotha,
although
actually
the
wax
local
may
be
fine.
Is
that
no
no
yeah?
That's
all
yeah!
That's
all
your
local
blocks,
yeah!
No!
So
that
would
look
like.
A
Pins
are
fine
because
pins,
we
saw
that
the
codec,
but
for
just
like
arbitrary
blocks
that
we
happen
to
have,
especially
because,
like
with
the
way
providers
work,
we
like.
We
don't
record
this
information,
it's
kind
of
annoying.
Now
we
do
have
some
work
in
I
profess,
like
record
providers,
so
that
we
could
then
expose
this
as
my
diskeeper,
the
CID,
but
that
actually
like
that's
the
exact
opposite
of
what
we
want
from
like
a
CID
D
1
v
2
perspective
we're
like
for
that
perspective.
A
A
Refs
is
pretty
easy:
I
think
we
can
just
like
return
route.
Cid
is
just
return
base.
30
to
see
you
18
feet,
one
for
like
like
roblox,
so
because
yeah
like
yes,
it'll,
be
breaking,
but
it's
technically
correct
and
it's
less
breaking
than
just
returning
like
Rommel
dashes.
Well,
let's
keep
using
different
walk
rubble.
This
is
the
pose
like
a
romualda.
Hash
is
also
a
valid
ab1
CID,
which
I
think
will
confuse
users
more
than
is
returning
like
Rossy
IDs.
B
A
But
before
we
do
the
roof,
migration
we'll
have
to
have
a
solution
for
a
certain
place
and
we
will
also
have
to
switch
over
providing
to
providing
Rossi
ideas
which
will
require
a
bunch
of
changes
for
non-star,
but
I
think
like
if
one
side
starts
pushing
on
this
I
think
vos
I
will
follow
so
one
invitation.
For
so
that's
what
the
open
food-ish
will
follow.
So
as
long
as
someone
starts
working
on
this
applicable
but
take
off.
Finally,.
M
L
A
The
main
implications
at
least
and
goes
like
pipes
everywhere,
oh
yeah,
so
like
we
have
to
now
change
all
the
I
guess
actually
may
be
easier
and
go
because
we
have
like
Claire
type.
So
we
can
easily
change
things
but
yeah.
Basically,
everything
that
was
the
CID
will
now
be
mostly
hash.
But
when
you
get
into
the
block
store.
A
So
I
mean
I
guess
we
can
also
start
moving
forward
on
this
and
go
I'll
start
moving
for
this
on
this
in
the
content
reading
go,
it
seems
possible
that
we
can
get
that
first
to
think.
That
is
like
the
hard
dependency
for
all
this
getting
assembly
contrib
running.
Once
we
have
done
in
content
routing,
then
we
can
start
migrating
everything
else,
eternally
okay,
any
other
questions
on
this
or
can
afford.
E
Yep
so
there's
a
very
short
proposal
to
allow
like
in
web
browsers.
We
will
need
to
have
a
own
protocol
handler
that
stays
in
location
bar,
and
the
idea
is
that
the
idea
is
to
allow
DNS
link
under
ipfs
protocol
handler,
as
well
as
like
write.
The
initial
initially
had
like
ipfs
protocol
handler
for
immutable
content
and
separate
IP
NS
handler
for
immutable,
sorry
for
mutable.
So
the
idea
was
one-to-one
mapping
from
paths
like
ipfs
paths
and
IP
address
box.
E
I'm,
not
sure
if
I
did
a
good
job
now
for
explaining
it,
but
I
think
I
made
a
better
job
in
the
issue
I
linked.
So
if
anyone
is
interested
in
how
we
would
load
websites
in
browsers
which
support
ipfs
protocol
handler
natively
would
be
very
useful
to
have
multiple
points
of
view
in
that
particular
well.
A
I
think
I
think
the
witness
now
is
everyone
reads.
The
issue
reads
the
discussion
for
request
and
then
we
schedule
a
call
where
we
can
all
get
on
it
and
like
make
a
final
decision
in
say:
okay,
like
regroup,
these
are
all
the
arguments
we've
had.
This
is
the
current
state
of
things
where
we're
going.
If
you
are
interested,
please
put
your
name
down
or
attendees
I
know
we
can
schedule,
I'll,
send
an
email
on
that.
A
And
we
can
organize
a
call
there.
Okay,
anything
else
on
that
before
I
move
on
okay.
Next,
one
final
decision
on
death
in
as
a
TLD
in
DNS
link.
The
basic
idea
here
is
the
DNS
wants
to
be
able
to
go
to
like
Sasha,
hi
P
and
s,
slash
something
got
eath
and
have
the
automatically
redirect
to
use
the
link
because
they
don't
have
a
TLD,
because
I
can
is,
like
you
know
so
yeah.
A
The
real
Italian
question
here
is
really
like.
We
want
to
do
this.
I
do
want
to
do
some
other
wakes.
Ideally,
we
would
have
everything
like
slash
something
really
big
/e
/
GN
s,
link
/
whatever,
but
that
that,
just
because
the
way
wants
to
work
they
want
to
interoperate
with
DNS
and
like
how
does
this
basically
actually
watching
you
to
be
able
to
like
go,
do
something
dot
eat
instead
of
something
calm
and
it
to
serve
magic?
That
works
even
without
the
custom
protocol,
and
so
that
ability
we
may
just
once
for
this
anyways.
A
H
With
a
new
binary
decoding,
encoding
and
decoding
and
I
turned
out,
they
were
incompatible
because
in
IPL
DC
bore
when
it
encodes
C
IDs
in
in
binary.
It
puts
a
single
zero
in
front
of
them,
and
it's
a
the
comments
is
something
about.
This
is
a
hack
until
we
support
binary
multi
base
and
I
have
no
idea
all
the
way
back
to
2017.
I
was
just
wondering
if
maybe
anyone
on
the
call
had
any
context
for
that.
We.
H
A
H
A
H
A
A
H
A
A
H
A
H
A
I
I
guess
just
maybe
maybe
silly
question
I
found
like
whatever
a
little
while
ago,
a
boggy
and
like
the
gullet,
PAP
testing
stuff
that,
like
wasn't
gonna
work
on
Windows,
because
it
like
timing
resolution
and
feels
like
maybe
is
there
like
a
way
we
can
do
CI
testing
that
also
does
Windows,
alright,
so
I
think
you
wanted.
A
A
K
A
K
A
It
may
make
sense
to
do
this
for
just
kill
the
p2p,
like
the
master
ego.
I'm,
not
sure
you
can
talk.
Thinking
about
that.
I
like
in
general,
I
would
like
to
switch
to
circle.
Also
cuz
like
to
have
these
they,
since
they
got
warps
the
driver
likely
is
where
you
can
like
share
configs
between
multiple
repositories
and
the
club
from
the
center
location.
K
I
Mean
I
I,
don't
know
if
it's
like
people
have
more
experience
with
working
with
DCI
tools,
probably
no
no
more
than
more
than
more
than
I
do
this
one.
This
is
sort
of
flag
it
as
like.
This
is
something
that
was
like
pretty
easily
avoidable.
If
we
had
had
this
anyone
in
the
stock,
at
least
on
the
p2p
side,.