►
From YouTube: 2023-05-02 IPFS Content Routing Workgroup #10
Description
A moment of silence for the Hydras, and a response plan to get bridging functionality restored. Content blocking from IPFS gateways, and Delegated Routing /Puts(enabling IPFS nodes to advertise to IPNI without running Providers)
A
All
right,
everybody
Welcome
to
the
10th,
ipfs
content,
routing
work
group,
hey
Steve,
welcome
hello,
hi,
sorry
to
interrupt
howdy,
not
at
all,
not
at
all.
We
just
started
recording
I'll
go
ahead
and
share
my
screen
and
we
can
take
a
look
at
the
agenda
I've
put
together
for
us
today.
We've
got
all
kinds
of
exciting
things
to
talk
about
a
lot
going
on
in
the
world
today,
including
a
kind
of
another
surprise
in
the
Hydra
World,
which
will
try
to
jump
to
First,
but
before
we
get
too
deep
into
the
weeds.
A
I
just
want
to
check
in.
If
anybody
has
any
updates,
I'll
go
ahead
and
start
with
the
ipni
side
of
the
house.
First,
so
everybody's
kind
of
up
on
what
we've
been
working
on
yeah,
the
last
supplies,
Surprise
only
more
I
mean
well
I'll,
I'll
jump
into
it,
pretty
quick,
but
just
to
start
the
meeting
off.
A
There
are
presently
no
hydras
as
far
as
we
can
tell
currently
operating
on
the
network,
which
wasn't
the
expectation
presently
so
yeah
yeah
celebration
may
be,
in
some
sorts,
perhaps
say
the
perspective
for
deep
reflection
on
the
other
fronts.
A
But
let's,
let's
start
with
an
ipni
update
so
pretty
exciting
stuff
over
in
the
ipni
side
of
the
house,
we
are
clawing
and
scratching
at
getting
our
reader
privacy
updates
implemented.
We
are
now
serving
as
much
as
30
of
our
production
queries
with
our
100
synchronized
encrypted
key
value
store,
so
we
are
making
great
progress
towards
a
hundred
percent
service
of
the
ipni
key
value
store
with
encrypted
key
value
pairs
of
Sids
to
peer
IDs.
A
For
those
of
you
that
aren't
sure
about
this
I'll
go
ahead
and
describe
a
little
bit
further,
since
we
have
some
new
faces
on
the
call.
The
goal
of
this
project
is
to
make
it
such
that
lookups
that
are
performed
on
the
interplanetary
network.
Indexer
are
private
in
the
sense
that
they
can't
be
observed
for
the
reading
activity,
so
people
who
are
performing
queries
because
we've
encrypted
this
key
value
store,
should
no
longer
be
able
to
be
observed.
A
A
An
instance
of
the
ipni
like
sid.contact,
basically
I've
got
two
work
group
calls
this
week,
primarily
with
storage
providers,
it's
kind
of
a
mix
of
folks
that
are
either
already
attempting
to
run
an
indexer
instance
or
who
are
interested
in
running
one
in
the
future
and
want
to
know
what
it
takes
and
so
we're
taking
those
parties
through
the
process
of
looking
at
what
the
current
state
of
their
indexes
is
working
with
them
to
kind
of
suss
out
any
challenges,
they're
having
and
ingesting
the
ad
chain,
which
can
be
kind
of
difficult,
depending
on
which
version
you
have
and
ensuring
that
they
have
the
most
up-to-date
Advanced
version
of
the
ipni,
because
we've
made
so
many
improvements
in
such
a
short
period
of
time.
A
A
So
that'll
be
the
topic
of
discussion
there.
Nobody
here
is
obligated
in
any
way
to
join
that
I
will
record
those
calls
and
happily
share
them
with
you,
but
if
you
are
interested
in
joining
them
to
understand
a
little
bit
more
about
what
the
storage
providers
who
are
running,
these
things
go
through
what
their
motivations
are
for
doing
so,
you're
absolutely
welcome
to
join
shoot
me,
a
note.
A
I
will
send
you
an
invite
and
then
just
kind
of
catch
people
up
on
the
ipni
side,
we've
been
doing
a
lot
of
AWS
infrastructure
right
sizing
lately
to
kind
of
optimize,
our
behavior
and
traffic,
shaping
with
everything
that
we've
been
wrestling
with.
We
run
into
some
size
limitation
issues
and
we're
not
charting
yet
so
we're
getting
to
actually
test
out
some
of
the
some
of
the
like
horizontal
expansion
tools
or
scaling
tools
that
we've
been
working
on.
A
But
it's
going
pretty
well
so
far,
a
lot
of
monitoring
traffic
and
attempting
to
ensure
that
we're
doing
all
the
right
things
key
I,
don't
want
to
put
you
on
the
spot
too
much
you
just
got
back
from
a
week
off,
but
should
anybody
on
the
call
know
anything
about
problab
of
late
I
can
provide
a
little
bit
of
an
update
from
at
least
giannis's
perspective
because
I've
been
talking
to
him
quite
a
bit
lately.
But
if
you
did
have
anything
you
wanted
everybody
to
know.
Please
jump
in.
B
A
For
sure
I
actually
just
wanna-
you
know
Giannis
isn't
here,
but
I
want
to
give
him
a
shout
out.
He
helped
a
lot
with
the
comms
preparation
for
updating
folks
about
the
state
of
the
hydras,
which
is
really
important.
A
We're
now
going
to
make
that
a
retroactive
communication,
because
we
didn't
get
it
out
before
the
deed
did
itself,
but
we
will
have
a
comprehensive
kind
of
update
to
the
community
on
what
we
consider
acceptable
as
an
outcome
for
this
you,
you
folks
did
a
lot
of
research
and
so
we're
putting
that
to
work
to
identify.
What's,
okay
as
an
outcome
for
this
and
I
think
we
have
a
pretty
confident
answer
to
that
question.
We
actually
expect
a
performance
increase
seems
like
we're
really
kind
of
doubling
down
on
that
that
thought.
A
So,
thanks
Keith
for
all
the
hard
work
you
Dennis
and
Giannis
Duke
over
there
Adeen
did
you
have
anything
you
wanted
to
share.
I
can.
C
Jump
in
there
I
mean
sort
of
we
can
put
this
into
the
pro
black
bucket,
because
they've
been
driving
a
lot
of
the
the
thinking
and
work
here,
particularly
ghee,
but
I
mean
more
to
more
to
come
on
this,
but
there's
some
good
presentations
at
ipfs
thing
you
know,
I
was
like
we've
taken
a
goal
to
get
reader
privacy
happening
within
the
DHT,
there's
different
ways
to
get
there,
and
one
of
the
stepping
stones
that
we
might
do
is
to
actually
first
do
a
composable
DHT
first,
so
that
we
can
basically
kind
of
split
out
get
get
a
public
good
peer
routing,
focused
DHT.
C
That
is
beneficial
for
all
of
lip,
P2P
and
kind
of
pull
out
the
or
differentiate
the
ipfs
specific
portions,
so
that
that
is
a
that's
a
strategy,
that's
being
evaluated.
It
would
kind
of
affect
the
timelines
and
the
sequencing
of
getting
reader
us
to
read
your
privacy
in
the
BHT
app.
But
we're
like
a
lot
of
conversations
were
kicked
off
at
ipfs
thing.
We
haven't
like
fully
estimated
the
work
and
made
decisions
there
or
like
kind
of
sought
as
much
Community
input
on
it.
Yet
but
that's
definitely
that's
ongoing.
C
A
I
think
it's
worth
doubling
down
on
Steve
how
much
we
heard
about
how
important
the
reader
privacy
work
that
everyone's
been
doing
is
I
had
a
few
conversations
with
like
people
that
are
doing
human
rights,
research
and
things
like
that
and
I
haven't
had
a
chance
to
like
kind
of
reflect
that
out
to
everybody,
but
they
are
terribly
excited
about
the
work
that
we're
doing
here
both
on
the
ipn
eyesight
and
on
the
DHT
side.
So
you
should
feel
feel
very
proud
of
it.
A
C
I
I
can
I
can
take
this
one
I.
You
know
no
major
efforts
other
than
making
sure
that
so
we
we
found
a
bug
that
Kubo
right
so
some
while
ago
earlier
this
year,
I
think
even
at
the
end
of
last
year
we
had
announced
that
Kubo
by
default
will
query
cid.contact.
Obviously
that's
configurable
by
a
user.
It
turns
out
that
wasn't
happening
when
the
accelerated
DHT
is
being
used
and
it's
generally
larger
infrastructure
providers,
including
the
ipfest.io
Gateway,
has
the
accelerated
DHT
enabled,
which
means
it
wasn't
querying
cid.contact.
C
So
that
issue
was
found
a
patch
was
was
made
and
it
got
deployed
to
our
to
protocol
Labs
ipfs.io
Gateway
infrastructure
this
morning
and
we're
now
seeing
it
called
cid.contact.
So
that's
that's
good.
That's
now
fixed
that
will
you
know
so.
There's
that's
going
to
be
ongoing
work
to
roll
that
out
across
all
of
the
ipfestyle
Gateway.
We
just
deployed
it
to
like
one
one
box
or
One
bank
and
we'll
get
this
into
the
0.20
release
of
Kubo.
C
It's
already
an
RC
phase,
but
so
we'll
we'll
get
this
included
in
the
actual
release
which
will
be
happening
this
week.
If
we
don't
find
other
performance
issues.
A
Awesome,
that's
so
great
to
hear
Steve
I'm
glad
we
kind
of
identified
like
exactly
what
the
characteristics
of
the
problem
were
and
you
were
able
to
articulate
them
for
us.
It's
super
helpful.
Also
I'll,
step
in
on
behalf
of
beef.
Roast
is
how
everybody
in
Belgium
is
saying
it.
The
bifrost
team,
but
I've
been
talking
to
them
a
good
bit.
So
everybody
knows:
George
is
working
on
behalf
of
the
ipni
team
right
now
to
get
some
of
the
updates
and
testing.
A
So
what
Steve
just
mentioned,
the
bifrost
team
is
basically
rolling
out
on
our
behalf,
so
they've
been
testing
this
morning
and
we
hope
to
see
more
continued,
like
tests
for
deploying
this
across
all
their
infrastructure.
I
haven't
got
a
timeline
from
them,
yet
on
what
the
full
deployment
span
looks
like
I
think
we
wanted
to
observe
and
measure
a
bit
of
the
impact
from
the
smaller
cluster
deployment,
but
it's
been
very
helpful.
A
The
other
thing
is
the
folks
at
bifrost
are
looking
for
some
help
from
this
group,
and
that
is
specifically
related
to
I've
got
this
in
our
topics
for
discussion,
but
they
released
their
content,
moderation,
kind
of
capabilities
that
they've
put
together
and
they're
very
interested
in
whether
or
not
they
could
potentially
get
that
into
like
a
reference,
ipfs
build
or
even
potentially,
in
the
longer
term,
Kubo.
A
So
if
that's
all
the
updates,
we'll
just
go
ahead
and
jump
into
it,
I
put
the
hydros
at
the
top
of
this
list,
because
it
came
up
as
a
bit
of
a
surprise
to
me
that
that
I
I
started
noticing
some
interesting
metrics
on
the
ipni
side.
Yesterday
talking
to
will
about
it
and
he
said
well,
have
you
checked
the
hydras
like
how
many
are
left
right
now
and
I
went
over
to
check?
A
And
lo
and
behold
there
were
no
idras
in
the
grafana
dashboard
and
I
was
a
bit
shocked
and
concerned,
and
a
little
dismayed
when
I
saw
that
so
I
just
wanted
to
make
sure
everybody
here
is
aware
that
that
is
the
present
operating
state
of
the
network.
We've
had
since
then
I've
kind
of
called
it
out
in
one
or
two
channels,
but
I
think
folks
have
been
kind
of
looking
around
Giannis
is
well
aware.
We
should
be
looking
for
any
kind
of
potential
traffic
degradation
across
the
network.
A
We
haven't
seen
I
would
say
major
impact
on
the
network
indexer
yet,
but
we
have
this
kind
of
secondary
challenge,
which
is
closely
related
to
this,
which
is
whether
or
not
the
gateways
are
actually
performing
that
bridging
function,
that
the
hydras
were
already
so
I
think
we
should
expect
a
pretty
significant
traffic
uptake
at
least
as
time
goes
on.
A
We
have
also
seen
that
some
of
our
providers,
like
the
internet
archive
terribly
important
providers,
their
their
lookups,
are,
are
failing
as
providers
right
now
as
a
result
of
this
bridging
functionality
not
being
in
place,
and
so
this
creates
urgency
around
resolving
the
the
break
and
functionality
that
we're
currently
experiencing
I
would
say,
there's
kind
of
a
leveraged
decision
to
be
made
here
regarding
this,
that
I
want
to
try
to
frame
and
I
would
like
you
all
your
all
opinion
on
on
this,
but
I
think
the
question
is:
is
getting
internet
archive,
for
instance,
backup
is
of
definitely
importance,
so
I
think
the
question
is:
is
whether
or
not
we
let
the
hydras
remain
down
and
let
this
play
out
by
resolving
their
ability
to
resolve
their
content
via
the
ipni
gateways
being
updated
to
resolve,
to
send
out
contact,
or
that
this
becomes
urgent
enough?
A
That
we
would
consider
actually
bringing
the
hydros
back
online
I'll,
just
say,
I,
see,
bringing
the
hydros
back
online
is
an
exercise
of
I,
don't
know.
Maybe
this
is
Harsh
terminology,
but
it
just
seems
like
to
me
and
I
would
love
y'all's
opinions
on
this
as
in
exercise
and
futility
that
we
would
be
basically
doing
unnecessary,
redundant
work
by
bringing
the
hydros
back
online
if
we
just
intend
to
take
them
down
again,
I
don't
want
anybody
to
have
to
do
work.
That
is
duplicative.
A
However,
at
what
point
do
we
draw
the
line
in
the
sand
and
say
well,
we
haven't
been
able
to
resolve
these
lookups
through
sidot
contact
from
the
gateways,
so
we
should
observe
improving
this.
You
know
negative
outcome
that
we're
now
experiencing
by
bringing
the
hydros
back
online.
Does
anybody
have
any
thoughts
about
this
subject.
C
Just
just
one
just
to
get
a
little
bit
more
data
here,
I
understood
that
well,
I,
guess
what
is
the
main
way
that
people
are
accessing
the
internet
archive
because
I
like
is
it
through?
Is
it
through
gateways?
Is
it
through
their
own
local
nodes.
A
D
D
Don't
know
last
week
as
hydros
fell
down
like
are
we
seeing
less
successful
resolutions
on
on
Kubo
now
that
they
don't
have
that
set
of
sins
being
responded,
so
how
many
users
are
are
actively
being
impacted
and
that
I
don't
have
the
numbers
in
front
of
me
to
correlate
I.
B
D
Think
we
know
when
the
Tigers
went
down
exactly
like
we,
we
know
a
point
by
which
they
had
gone
down.
If
we
can
identify
what
time
we
think
they
crashed,
we
can
look
for
the
before
after
to
see
if
the
the
success
metrics
changed.
E
E
I
I
guess
I'll
I'll
flag,
that
some
of
this
is
like
you
know,
we'll
call
it
like
sketchy
visibility
territory
so
through
through
the
magic
of
spamming
people.
E
E
So
we're
not
seeing
like
I
suspect
we
would
have
had
a
lot
more
angry
people
yelling
at
us
about
like
why
isn't
my
stuff
working
if
we
didn't
have
pairing
Arrangements
there,
which
means
if
anyone
else
is
doing
this
without
the
pairing
Arrangements
they're
likely,
you
know
fairly
unhappy
some
extent.
The
question
is
how
many
people
are
running
accelerated,
DHT
clients
and
doing
it
in
a
finding
capacity,
rather
than
in
like
a
node.
E
That's
just
sitting
there
serving
up
lots
of
bites
like
cluster
nodes
right
I,
don't
know
what
those
like
I
don't
know,
there's
no
way
to
crawl
the
network
and
find
out
with
you
know
without
a
bunch
of
work.
So
it's
it's
hard
to
know
how
many
people
this
bug
is
affecting.
A
D
A
tricky
problem,
so
we
had
some
updates
from
from
Gus
just
before
this.
D
It
sounds
like
this
was
rolled
out
to
staging
and
we're
pretty
happy
with
the
metrics
that
we're
seeing
there
and
I
think
both
Gus
and
I
gave
thumbs
up
to
roll
it
out
more
broadly
on
the
production,
ipni
there's
a
cherry
pick,
the
19
of
Kubo
with
just
this
change
that
so
it
just
fixes
the
bug
and
it's
relatively
low
risk
and
I
would
be
surprised
if
it
takes
more
than
a
couple
days
for
George
to
be
able
to
roll
that
out,
and
so
that
seems
like
about
as
good
of
a
timeline
as
we
would
get
with
spending
the
effort
to
restart
up
hydros,
so
I
would
be
in
favor
of
just
rolling
that
out.
E
Is
there
a
way,
I
wonder
if
it's
worth
cutting-
and
maybe
this
is
just
a
separate
Kubo
conversation
cutting
like
a
zero
19
like
patch
release,
alongside
zero?
Twenty
just
to
like
in
case
just
be
like
you
should
be
updating.
You
should
really
be
updating
and
fixing
this
problem
if
you're
not
ready
to
take
like
the
next
update
with
whatever
else
it
comes
with
like
just
just
do
this:
it's
really
it's
really
small,
because
it
will
impact
people.
C
Think
I
think
it's
a
fair,
probably
I
Dean.
We
should
yeah,
let's
talk
about
that
during
stand-up,
but
yeah
I
think
having
a
019.2
that
has
this
patched
and
a
0.20
so
that
it's
like
hey
that
way,
our
response
can
be
if
anyone
you
you
complains
about
this,
it's
like
we've,
given
them
two
ways
to
to
unblock
themselves.
E
A
This
is
kind
of
like
a
jarring
process
and
like
a
clutchy
way
to
like
go
through
it,
but
also
simultaneously
healthy
for
the
network
right
like
we're
we're
forcing
ourselves
through
like
something
that
we
have
to
do
right
to
be
operating
in
a
extensible
way.
I
think
thank
you
Steve
for
grabbing
that.
A
Let
us
know
what
you
all
decide,
so
we
can
kind
of
keep
an
eye
out
and
I'll
keep
keep
the
Line
open
with
George
we've
been
talking
quite
a
bit
to
see
what
it'll
take
I'll
drop
notes
in
the
content.
Writing
work
group
on
what
their
timeline
looks
like,
but
well.
My
intuition
was
that
this
seems
like
the
obvious
solution
to
me
is
just
to
persist
by
going
this
direction.
A
But
I
just
didn't
want
to
presume,
on
behalf
of
the
entire
group
that
that
was
the
way
to
go
so
just
to
clarify
we'll
we'll
proceed
by
attempting
to
resolve
this
by
getting
the
gateways
hard,
pointing
towards
Sadat
contact
as
our
method
of
resolution
no
need
to
consider
trying
to
bring
hydros
back
online
presently
we'd.
Rather,
let
them
stay
down
and
have
their
honor
undisturbed
resting
in
their
grave
for
for
all
time
to
come,
possibly,
hopefully
any
other
thoughts
on
that
subject
before
we
move
on
to
content
blocking
yeah.
C
I
mean
it
seems
to
me,
like
the
only
other,
I
mean
right.
The
thing
we
have
missed
in
this
process
is
the
community
comms
about
it
right,
and
so
the
thing
that
could
happen
here
is
other
people.
Just
you
know,
wake
up
and
realize
this
and
you
know
weren't,
you
know
hadn't
prepared
yet
to
get
things
into
ipni
or
create
special
pairing
arrangements
or
whatever,
and
so
we're
reliant
on
this.
And
you
know
we
don't
really.
C
We
can't
really
Point
them
to
a
fact
that,
like
hey,
we
We
messaged
that
this
was
going
to
happen.
One
month
ago,
like
we're,
trying
to
move
the
network
forward,
so
I
feel
like
that's.
Maybe
one
of
our
sticking
points
I'm
good
with
the
plan
of
record
but
I'm
just
trying
to
think
of
what
could
derail
that?
That's
maybe.
C
It
that's
a
good
point:
okay,
yep
here
and
then
I
guess
have
I
know.
Problab
was
kind
of
last
week
are:
have
we
been
seeing
any
yeah
we've
been
observing
any
performance
or
I?
Guess
it
changes
as
a
result
of
the
Hydra
is
being
fully
down
like
I.
Think
that
was
one
of
the
action
items
from
the
ipfs
thing
meeting,
as
we
were
going
to
like
look
to
see
to
do
some
of
the
analysis
and
I.
Don't
know
if
that
has
been
done
done
yet
so.
A
What
I
gathered
from
talking
to
Giannis
was
that
there's
been
some
like
kind
of
like
cursory
like
watching,
but
I
wouldn't
consider
it
like
like
a
full
Deep
dive
yet,
and
they
are
like
looking
at
it,
but
I
just
dropped
a
link
here
and
oh
I.
Don't
want
to
drop
that
I
want
to
drop
the.
Let
me
grab
the
link
to
the
notion
dot,
but
the
comms
dock
that
I've
got.
A
He
dropped
like
kind
of
a
lot
of
I
would
say
flavor
on
the
subject
regarding
their
confidence
that
bringing
the
hydras
down
is
definitely
the
way
to
go
and
kind
of
supporting
that.
All
of
our
observations
during
that
meeting
were
actually
maybe
even
a
little
conservative
in
regards
to
whether
or
not
we
would
be
okay,
I
think
from
their
perspective.
Everything
outside
of
these
like
peering
relationships
that
were
presently
concerned
about,
obviously,
but
that
we
were
okay
to
do
it.
The
comms,
Steve
I,
don't
want
to
waste
that
doc.
A
Like
we
put
like
this
great,
you
know
communicate
to
the
community
yeah
script
together.
I
am
going
to
go
through
today
and
revise
it
to
be
kind
of
a
more
retroactive
like
a
heads
up
everybody.
This
is
going
on
right
now,
type
of
Doc,
and
now
that
I've
got
giannis's
eyes,
I
should
be
able
to
get
it
out
today.
I,
don't
see
why
we
wouldn't
so
I'll
I'll,
revise
that
we'll
get
it
out
today.
I
totally
agree
with
you
Steve.
This
is
a
from
my
perspective.
A
Probably
one
of
the
most
painful
observations
is
that
we,
we
could
have
been
a
lot
well.
I
I
could
have
been
a
lot
more
ahead
on
this,
so
I'm
gonna,
I'm,
Gonna
Roll,
with
that
yeah.
C
No,
that's.
That's
all
I
I,
like
this
I
I'm
some
degree
viewing
this
as
a
blessing
in
disguise.
It's
forced
the
hand.
It's
moved
this
forward.
Obviously
it
would
have
been
great
to
have
comms
on
the
front
end
of
it,
but,
like
I,
think
we're
doing
the
best
we
can
I
was
just
trying
to
think
about.
Like
your
question
of
any
like
about
you,
you
proposed
a
plan
I'm
good
with
the
plan
I'm
just
trying
to
think
about
like
what
would
change
it,
but
it
sounds
like
we
don't.
C
We
don't
have
anything
in
yet
right
now
that
we
change
it.
I'm
good,
just
I'm
good,
to
support
with
what
we're
suggesting
of
assuming
we're
going
to
be
able
to
roll
out
IP
test.io
Gateway
updates
across
all
the
fleet
that
we
just
let
the
Hydra
stay
dead,
sweet.
A
That
feels
like
such
a
heavy
term
I
feel
like
they
deserve
better
I
feel
I
feel
a
little
wince
every
time.
I
say
it
like
they've
done
so
much
for
us,
but
maybe
it's
the
most
apps
terms:
okay,
we'll
we'll
jump
into
content
blocking
in
the
gateways,
which
is
a
request
from
bifrost
I'd
like
y'all,
to
take
a
look
at.
If
you
haven't
it's
ipip383.
A
Which
is
worth
reading
it
kind
of
gives
you
a
sense
of
what
they're
attempting
to
do,
but
basically
it's
configurable
I,
don't
know.
Maybe
people
would
wince
if
I
use
this
terminology,
but
I
look
at
it
like
an
access
list,
kind
of
control
that
they've
put
together
and
their
requests
from
this
group
is.
A
Is
they
want
support
for
integrating
this
into
like
a
Kubo
release
for
testing,
but
they're
longer
term
request
is:
could
we
get
a
native
implementation
in
Kubo
they're,
hoping
you
all
could
take
a
look
at
it
weigh
in
on
your
perspective?
They
see
it
as
incredibly
valuable
and
I
could
see.
A
Why,
like
there's
I,
didn't
get
to
like
ask
them
about,
like
the
I
think
like
the
elephant
in
the
room
when
I
see
this,
and
maybe
this
is
like-
maybe
I'm
over
assuming
some
things
but
balancing
like
the
ability
to
kind
of
not
be
prevented
from
communicating
via
ipfs
nodes.
As
a
like
a
censorship
resistance
perspective
versus
like
enabling
us
such
that
you
can
recognize
like
illegal
or,
like
you
know,
whatever
type
of
content,
and
make
the
decisions
about
moderating
that
that
content,
via
your
node,
for
instance,.
E
I,
don't
think
there's
much
stress
there.
Sometimes
that
comes
up
in
the
community
is
like.
Oh
no,
you
can't
censor
things,
but
it's
like
you're
going
to
tell
me.
I
can't
have
an
ad
blocker
on
my
on
Chrome,
because
it's
like
censoring
the
internet
like
like.
Are
you
out
of
your
mind
right,
like
I,
it's
my
node
I
can
BL
I
can
block
whatever
data
I
would
like
to
not
happen
on
my
node
yeah
right
I.
A
A
Had
Cameron
had
like
some,
you
know,
I,
don't
know
what
they
see
on
the
bifrost
side
of
the
house,
but
he
had
some
pretty
strong
examples
for,
like
you
know,
we're
gonna
find
like
terrible,
terrible
stuff
at
some
point
in
time,
and
particularly
like
from
the
perspective
of
like
pl's
operation
nodes.
We
want
the
ability
on
all
of
our
nodes
to
like
not
be
forced
into
the
supposition
of
Hosting,
that
content
right
so
or
or
amplifying
that
content
earlier
across
the
network,
which
makes
perfect
sense
like
I,
get
that.
A
E
So
I
think
Hector
has
I'm
trying
to
remember
the
name
of
it
right
now.
He
has
a
effectively
like
a
Kubo
plug-in
that
will
do
this.
E
So
if
the
question
is,
can
we
bundle
it
by
default
and
expose
it
via
config
file?
Then
I
I
suspect.
The
answer
is
as
long
as
there's
a
sane
way
to
deal
with
the
formatting
and
there's
a
spec,
for
it
then
sure.
E
Yeah
no
PFS
there
y'all
I,
like
the
the
awkward
thing
here,
is
that.
E
E
They
I
think
want
this
too.
Some
way
to
like
communicate
and
pull
in
each
other's
lists
and
combine
them
and
stuff.
So
I
would
I
guess
push
that,
like
maybe
push
back
a
little
and
say
maybe
this
plug-in
thing
isn't
so
scary
and
that
if
you
want
this
to
like
Upstream
to
places
make
it
usable
by
other
by
other
people
and
I
I
sort
of
suspect.
E
It's
not
right
now,
because
you
want
some
story
whereby
you
know
FPL
and
inferior
each
get
a
bunch
of
stuff
that
they
get
abuse
reports
for,
and
they
reasonably
trust
each
other
to
not
be
like
overly
censorshipy,
for
no
particular
reason
that
they
can
choose
to
include
each
other's
lists
so
that
they
would
reduce
the
number
of
emails
they
get
with.
Takedown
notices
in
in
the
meanwhile.
A
This
is
this
is
a
really
good
point,
I'm,
going
to
take
an
action
item
to
follow
through
on
that
I'm
going
to
dig
through
this
back
as
well.
A
I
I
haven't
seen
like
the
technical
side
of
it
of
what
I've
seen
is
largely
like
kind
of
the
the
scope
of
The
Proposal
side,
which
makes
a
lot
of
sense
like
it's
really
well
put
together
and
I.
Think
they've
they've
built
quite
a
bit
of
this
out
already
so
I
I
do
think.
A
The
component
like
I,
haven't
I,
haven't
looked
at
the
scope
of
what
they've
built
enough
to
determine
whether
or
not
they've
kind
of
got
that
type
of
functionality
I'm
guessing
they
wouldn't,
because
that
would
be
an
additional
layer
of
work
that
the
I
think
we
lost
you,
but
I'll
I'll
take
an
action
item
to
follow
up
with
Emma,
Dean
and
kind
of
get
a
look
at
it.
If
you
all
can
kind
of
toss
it
on
your
backlog
to
consider
a
reference,
implementation.
A
Okay,
all
right,
the
last
thing
we're
making
record
time
today,
everybody
it's
great.
A
The
last
thing
I
wanted
to
kind
of
get
people
thinking
about
here
is
so
we're
we're
kicking
off
this
index,
operators,
work,
group
and
I'm
going
to
be
meeting
with
all
these
index
operators
this
week
talking
about
running
their
ipni
instances.
Our
goal
is
to
get
them
to
this
Federated
mesh
of
network
indexers,
which
is
it's
a
pretty
high
goal
like
it's
going
to
be
pretty
tough,
but
one
thing
that
we're
kind
of
looking
at
is
like
to
make
this
sustaining
I.
A
For
me
at
least
right
now
is
the
delegated
routing
puts
and
one
challenge
that
will
brought
up
that
we
were
kind
of
iterating
on
a
bit
in
our
IP
and
I
team
Colo
was
how
we're
going
to
resolve
Distributing
those
puts
across
all
the
ipni
instances
which
there's
a
bit
of
a
trick
here,
because
if
we're,
this
is
going
to
be
very
powerful
for
us
to
be
able
to
advertise
essentially
directly
from
ipfs
nodes
to
to
ipni
getting
that
work
done.
A
I
think
we
don't
really
know
when
that's
going
to
happen
or
how
much
of
the
scope
like
falls
on
which
side
of
the
table
between,
like
the
stewards
and
the
ipni
team
I
think
there's
like
a
little
bit
of
collaborative
effort
that
needs
to
happen
there
and
I'd
like
to
like
start
the
process
of
like
engagement,
where
we're
like
looking
at
like
who
needs
to
do
what
and
what
the
timelines
look
like
to
be
able
to
do
those
things
together,
but
secondarily,
I
think
there's
this
like
bigger
design
problem,
which
probably
is
going
to
have
to
go
into
a
design
Dock
and
get
beat
up
there,
but
is
how
we're
going
to
wrestle
with
that?
A
And
if
anybody
had
any
thoughts
on
that
subject
it's.
This
is
definitely
an
ipni
problem
to
figure
out,
but
definitely,
if
you
all
had
opinions
on
this
I
think
it's
it's
a
Content
routing
problem
for
sure,
and
the
DHT
wisdom
in
the
room
is
always
appreciated
when
thinking
about
these
kind
of
things,
because
they're
definitely
some
parallels
in
mind
for
sure
does
the
the
problem
that
I
just
described
like
resonate
with
you
that,
like
yeah,
this
is
kind
of
a
pain
to
figure
out.
E
So
we'll
say
yes
and
no
so
the
awkward
thing
about
delegated
puts
with
the
DHT
is
you
can't
do
them
for
provider
records?
You
can
only
do
them
for
ipns
records.
You
can't
do
them
for
peer
records
either.
E
E
So
IPM
is
the
only
thing
that
allows
us
at
the
moment
and
to
some
extent,
the
the
questions
like.
How
do
you
want
this
to
go
right?
So
do
you
want
people
to
be
able
to
do
this
in
individual
puts
ipni
tends
to
work
in
batches?
It
wants
you
to
do
batches
and
then
stamp
them
with
a
with
a
metadata
label
or.
A
E
E
Is
it
because
somebody's
writing
an
ipfs
implementation
in
Python
and
they're
serving
data
over
HTTP,
and
they
would
like
to
ask
ipni
to
please
advertise
that
they
are
serving
the
data
over
http
and
like
They,
Don't
Really
Wanna
operate
any
of
that,
but,
like
they're
still
totally
there,
they
just
there's
just
a
whole
bunch
of
code.
They
don't
want
to
write
and
and
go,
and
they
don't
feel
like,
including
your
thing
with
like
ffi,
and
that's
why
they're
doing
this
like?
Why
is
somebody
yeah?
Why.
A
I'll
tell
you
my
perspective:
Adine
and
I'm
I'm.
Sad
well,
actually,
I'm
really
happy
that
mossy's
actually
on
a
week
off
because
damned
if
he
doesn't
deserve
it.
So
let
me
let
me
rephrase
that
real,
quick
I'm,
just
sad
that
he's
potentially
not
here
because
I
know
he's
gonna,
have
very
strong
feelings
about
this
and
I'll
look
forward
to
his
recap:
we'll
we'll
catch
up
with
him
next
week
on
it,
but
Andrew
joined
us
today.
Andrew.
If
you
have
opinions,
please
throw
them
in
there.
A
I'll
tell
you
my
opinion
is
that
the
ability
to
gain
advertising
minus
the
necessity
of
running
a
provider
side
car
I
think
is
the
most
powerful
Paradigm
that
could
change
how
we
interact
with
a
lot
of
our
present
clients
and
so
like
I.
Think
from
like
a
storage
provider
perspective
from
the
file
coin,
Network,
for
instance,
giving
them
the
flexibility
to
advertise
to
us
through
their
own
devices
without
having
to
run
like
our
provider
on
their
their
tools
chain,
like
whatever
they've
got
running.
A
It's
just
an
incredibly
like
freeing
Dynamic
that
they
could
potentially
approach
advertising
to
us
in
any
way
that
they
want.
Not
that
running
a
provider
is
like
some
kind
of
major
overhead
or
challenge,
but
that
this
would
like
free
them
up
to
have
like
a
lot
of
different
approaches.
That
would
make
things
easier
for
them,
I
think
Andrew.
How
much
does
that
put
sour
or
something
in
your
coffee?
Well,.
F
It's
interesting
because,
essentially,
what
it's
doing
is
asking
some
other
entity
us
I,
guess
in
this
case,
to
be
the
advertiser.
So
when
we
say
an
Advertiser,
it's
really
that
someone
who's
maintaining
some
sort
of
knowledge
of
what's
currently
being
advertised
as
available
at
some
location,
and
you
know,
instead
of
just
having
anybody
post
to
indexers
right
now.
F
Why
do
we
have
them
advertised
well,
because
that
puts
it
on
them
to
keep
track
of
what
they
currently
have
available
and
to
retract
that
when
it's
no
longer
available
and
things
like
that,
when
we
change
this
Paradigm
of
being
able
just
to
put
things
that
sort
of
diminishes
the
respond,
the
responsibility
of
maintaining
this
thing.
So
now
they
so
there's
not
really
a
path
for
cleaning
things
up
or
retracting
things.
People
can
just
put
things
unless
we
have
or
some
sort
of
a
built-in
cleanup
mechanism
like
expiration
times
and
things
like
that.
F
We
need
to
be
very,
very
careful
and
also
what
is
the
reason
why
anybody
would
want
to
advertise
versus
just
putting
things
into
indexers
and
thereby
making
the
indexers
responsible
for
doing
all
of
this
maintenance.
F
So
another
way
to
look
at
it,
so
I'm,
not
I'm,
not
person
I'm,
not
prescribing
any
particular
solution.
I'm
just
saying
those
are
some
things
to
be
careful
of,
but
also
are
we
really
just
setting
up
another
public
advertisement,
publisher
and
letting
people
just
use
that
essentially.
A
This
is
you
know,
the
the
other
thing
I'm
thinking
about
is
like
one
of
our
big
goals
is
to
get
like
caches
of
these
index,
like
as
close
to
the
lookups,
as
as
they
could
happen
and
I've
seen
this
like
having
HTTP
puts
as
kind
of
like
enabling
to
that
caching
story,
because
you're,
basically,
in
my
opinion,
like
lowering
the
bar
to
entry
to
be
able
to
perform
these
advertisements
you're
making
it
so
that,
basically,
anybody
could
do
it
and
I
don't
know
I.
A
Think
there's
like
a
lot
of
like
potential
use
cases
that
are
unlocked
by
this.
That
just
won't
exist
so
long
as
we
require
providers,
but
it
does
create
this
like
huge
garbage
collection
problem
that
we're
kind
of
changing
the
Paradigm
from
batching
to
potentially
like
small
instances
of,
and
it's
not
going
to
be
much
less
efficient,
Network
traffic
pattern.
I
didn't
please
jump
in
also.
F
A
E
Dean,
please
jump
in
yeah,
so
I,
don't
know
how
how
different
those
things
are.
The
thing
that
you're,
but
the
two
things
you're
trying
to
be
careful
of,
is
that
you
want
to
have
some
like
traceability
on
what
happened,
because
you
want
to
be
able
to
deal
with,
like
someone's
just
sending
me,
like.
You
know,
a
billion
cids
that
are
garbage
and
they're
just
having
me
hold
them
for
a
while
and
they're
sending
me
a
ping
every
two
weeks
to
let
them.
You
know
that
I
should
keep
all
the
garbage
right.
E
If
that
traceability
is
expensive
for
you
or
too
expensive
to
do
when
you
have
individual
puts-
and
you
want
to
encourage
people
to
do
batches
okay,
then
then
you're
encouraging
people
to
do
their
puts
and
batches
and
given
that
people
have
to
do,
puts
right
and
they're
doing
their
puts,
whether
individually
or
batches
and
even
in
the
normal
ipni
thing
you
have
to
inform
them
every
so
often
hello,
I
still
have
my
stuff,
so
I
have
put,
and
then
I
have
I
still
have
it,
as
opposed
to
the
DHT
where
I
have
put
and
then
I
have
put
again
instead
of
put
and
I
still
have
it
right,
that's
the
distinction
you're
asked
like
it
seems
like
if
you're
delegating
to
somebody
else
to
do
the
work
for
you.
E
E
If
you
have
a
DH,
if
you
have
a
DHT
that
lets
us
do
delegated
puts
after
we,
you
know,
do
some
RPC
changes,
and
now
we're
like
oh
I,
would
like
a
way
so
that
someone
could
abstract
over
this
and
hit
both
of
them
all
right.
Well,
now
we're
talking
right,
but
yeah
I
feel
like
I
feel
like
if
there's
a
painful
lesson.
Anyone
who's
looked
at
the
Go
code
for
ipfs
things
has
seen
it's.
Please
don't
make
abstractions
without
two
implementations
yeah,
because
we're
bad
at
it.
A
That
that
sentiment
is
not
lost,
I
mean
I
mean
I.
Can
I
can
imagine
so
this
this
topic
will
will
not
go
away
from
this
discussion
for
sure
I
My
Hope
in
this
meeting
was
to
kind
of
get
you
all
to
add
your
brain
power
to
the
idea.
A
little
bit,
maybe
think
about
it.
A
I,
don't
know
how
much
you
all
are
like
me,
but
I
think
about
these
things
and
then,
when
I'm,
having
a
cup
of
coffee
like
three
days
later,
like
a
moment,
comes
and
I'm
like
well
wait
a
minute
what
about
what
about
this,
and
if
we
can
ever
get
that
kind
of
imagination
from
you
all
towards
these
kind
of
things?
We
we
want
it
for
sure,
but
we'll
get
Mossy
to
weigh
in
on
this
subject.
Next
time
we
we
get
to
catch
up.
F
I
did
just
one
to
ask,
though,
is
there
any,
is
there
any?
Does
it
make
any
sense
at
all,
or
is
it
a
useless
abstraction
the
idea
of
having
something
of
an
ephemeral
indexer
so
whereby
and
I'd
be
able
to
advertise
that
as
a
repository,
a
temporary
repository
for
all
of
my
puts
and
somehow
that
would
be
the
entity
that
was
discoverable
on
a
top
level
as
opposed
to
this.
A
How
persistent
is
that
going
to
be
like?
Does
it
even
need
to
be
persistent,
we're
talking
about
thousands
of
nodes
right,
but
I
I
suspect
that
we're
going
to
need
to
assume
some
level
of
like
failed
persistence
with
some
of
these
right,
like
they're,
going
to
be
adding
in
depleting
from
the
network,
probably
kind
of
constantly
right,
at
least
I
think
maybe
that's
a
wrong
wrong-headed
assumption,
but
I
I
think
it
makes
sense
to
to
be
thinking
about.
In
these
terms,.
E
Ephemeral,
so
if
you
do
an
ephemeral
thing
right,
then
then
you're
sort
of
you're
just
doing
puts
again
right
instead
of
having
like
put
and
I
still
have
it
you
just
do
puts,
which
is
fine.
It's
just
that!
That's
like
that's
one
of
the
one
of
the
larger
distinctions
that
you
you
got
from
the
like.
Having
a
separate
mechanism
from
how
the
DHT
was
doing.
This
was
that
you
were
like
well
we're
going
to
reduce
the
number.
We're
gonna
reduce
the
upload
bandwidth
from
the
providers.
F
And
there's
a
that
does
that
we
did
solve
that.
Specifically,
we
did
that
for
that
specific
reason,
but
also
that's
looking
at
indexers
as
a
global
resource
versus
something
that's
much
smaller
and
more
local,
where
maybe
we
don't
really
care
about
that.
So
much
you
know
is
that
something
that
we
want
to.
F
In
other
words,
it's
forcing
more
of
a
hierarchical
structure
instead
of
just
having
everything
flat,
put
indirectly
into
a
global
in
index,
or
we
have
more
of
a
hierarchy
where
we
have
something
a
sub
index
or
if
you
want
to
look
at
a
tree
of
indexers,
and
it's
the
this.
This
sub
index
versus
this
femoral
indexer,
which
is
the
repository
of
temporary,
puts
for
some.
E
F
Yeah,
that's
that's
either
that
or
or
it's
discoverable
like
a
DHT
node
and.
F
In
other
words,
hey
I'm,
looking
for
this
thing,
okay!
Well,
let's
here's!
The
next
hop
in
this
discoverable.
E
E
F
In
that
one,
so
basically
all
I'm
really
doing
is
making
it
hierarchical,
and
so
now
we'd
still
go
back
to
the
original
semantics.
So
if
you
have
puts-
and
you
have
I
still
have
this,
but
the
put
would
be
for
the
whole
index
really
I
still
have
this
is
it's
still
up
and
that
way
the
garbage
collection
is
a
single.
No
I.
Don't
have
this
whole,
so
it's
basically
it's
just
a
mechanism
for
making
a
hierarchy
is
all
that
comes
up
down
to,
and
is
that
and
so
I'm
asking?
F
Is
that
something
that's
inherently
valuable,
or
is
that
just
more
of
the
same,
an
abstraction?
It's
not
really
giving
us
any
any
benefit
and.
B
Yeah
exactly,
and
so,
if
you
already
know
when
you
won't
need,
if
you
already
know
the
moment
when
you
won't
need
the
data
anymore,
then
it's
perfect
to
advert
I
mean
to
share
it
when
you
advertise
the
content
so
that
they
don't
need
to
give
it
longer
than
needed.
F
F
We
have
that,
so
we
actually
do
have
this
sort
of
expiration
time,
so
it's
not
as
as
clean.
In
other
words,
we
could
still
have
things
that
are
advertised
when
they're
no
longer
available
at,
and
it
requires
all
these
ttls
to
be
enforced,
but
yeah
we
so
we
do
have
ability
to
to
do
some
garbage
collection
that
way
by
expiration.
F
We
also
have
the
idea
of
a
context
ID
whereby
you
can
have
a
whole
bunch
of
data
mapped
to
a
the
same
context,
and
then
I
can
say
this
whole
context
is
no
longer
there
that
can.
That
has
the
same
effect,
but
also
currently
is
kind
of
an
expensive
operation,
because
that
means
you
have
to
keep
a
reverse
mapping
of
all
the
individual
things
that
context
map
to
I
still
think
it's.
It
seems
to
me
it's
very
viable
just
to
have,
rather
than
any
of
those
mechanisms.
A
We
are
officially
out
of
time
on
another
good
discussion,
look
forward
to
this
every
two
weeks,
which
isn't
the
same
for
every
other
meeting.
So
I
really
appreciate
you
all
joining
it's
a
pleasure
and
thanks
for
Lending
us
your
brain
power.
Congratulations
again,
gee
go
enjoy
being
a
married
man
have
a
good
one.
Y'all.