►
From YouTube: IPLD Deep Dives Aug 24, 2017
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
B
Any
selectors
in
is
equal
any
transformation,
and
those
are
the
things
that,
like
once,
we
try
to
include
those
on
the
first
iteration
of
I
Billy
became
like
really
passive,
balloon
fest
and
that's
when
we
agreed
okay,
we're
like
this
can
be
a
layer
on
top
that
we
can
write
first,
implement
the
base,
make
it
work
and
then
experiment
on
top
and
see
what
was
I.
Think
like
there's
a
lot
of
issues
with
a
lot
of
notes
on
this.
B
D
We
were
right.
We
started
an
effort
last
year
this
year
in
February,
to
write
in
the
spec.
It
would
be
nice
to
finalize
the
effort
sometime
soon
is
that
we
got
busy
in
something
else
and
what
I've
been
what
I've
been
working
on
and
I
think
that
I
would
like
to
reboot,
maybe
by
giving
what
I've
done
you
guys,
mostly
because
I,
don't
think
I
will
have
time
in
the
next
couple
of
months
is
to
finish
off
the
selector
work,
which
is
which
has
started
a
while
ago,
and
the
selector
is
literally.
D
And
how
do
I
selects
a
path
of
the
graph
and
in
a
in
an
efficient
way
bow
so
in
a
provable
way,
and
yes,
that
this
man
made
the
I
think
the
selector?
That
is
an
issue
that
I
am
going
to
linking
this
in
this
and
please
chip
in,
and
there
is
also
some
working
code
on.
How
should
we
do
the
selectors
a
few
proposals?
One
is
using
regular
expressions.
One
is
using
a
subset
of
regular
expression.
One
is
using
traditional
XPath
and
Oh
a
even
CSS
selector.
D
B
All
right,
thank
you.
Yeah,
like
you
know
these
things
that
we
know
that
there
has
been
some
discussion
somewhere.
Let's
make
sure
that,
like
they
are
all
linked
from
there,
that's
because
we
need
to
come
back
to
those
issues
at
some
point
cool.
So
next,
up
on
my
screen
is
Emin
I
mean
my
visa
Evan
you're
next
gonna
go
yeah.
E
Well,
I
was
super
resting,
go
and
got
up
to
speed
with
the
go
IPA
film
again
as
Steven
was
my
reviewer
and
I
finally
managed
to
build
a
resolver
for
the
construction
free,
because
the
Atheneum
object
has
3
3
2
pitch
a
twist
storage
state.
The
construction
and
construction
receives
on
the
next
step
is
the
real
goal,
which
is
put
the
whole
porcinis
around
250
gigabytes
into
several
servers
and
then
just
a
plug
in
2.
We
solve
over
the
data
and
be
able
to
use
real-life
clients.
E
So
so
my
goal
in
this
meeting
series
get
around
all
the
interface
all
the
roadmap
of
the
project
to
do
the
right
implementations.
As
far
as
far
as
now,
to
avoid
silly
question,
I
didn't
do
a
lot.
I
need
to
understand,
resolved,
resolved
links.
We
all
the
right
vision.
You
have
guys
so
to
do
the
right
implementations
for
now,
for
example,
the
tree
I
do
is
yes
in
one
level
and
the
rezoning
is
very
intuitive,
but
I
want
to
have
the
real
rules,
they're
real
real
plans.
B
Yeah,
we
definitely
need
to
start
creating
more
examples
that
are
more
than
just
like
fetching
blocks
and
like
a
great
example,
I
guess
right
now
he's
unique
success.
Excel,
because
now
it
uses
the
dag
resolver
to
define
its
walks
and
Traverse
term,
but
attack
still
like
it's
not
very
intuitive,
to
go
to
rebuild
Steven.
A
This
actually
goes
back
to
my
point.
Where
we
go
I
give
us
the
go:
I
killed,
the
api's
are
really
they
don't
follow
IP
LD
very
closely
like
they
have
really
funky
interfaces.
They
just
don't
match
the
concepts
on
multi
like
there
is
links,
method
that
returns.
Would
our
effectively
ipfs
link
objects
that
just
make
no
sense
that
I
feel
deal
and,
like
you
have
names
and
sizes
and
whatnot.
B
B
F
H
I
I
D
Someone
is
typing
and
clicked,
but
this
really
sucks
when
two
people
type
so
I
was
taking
notes
and
someone
is
typing
and
I
cannot
take
notes
anymore.
If
I,
maybe
can
you,
let's
edit,
that
clip
pad
after
we're
done?
I,
don't
know
it's
really
odd.
They
never
happened
before,
but
I
stuff,
the
typing
and
someone
I
saw
someone
else
typing
and
all
of
my
sentences
got
mixed
up.
I
see.
B
B
B
B
B
A
So
we're
gonna
need
some
way
of
adding
types
to
Jason.
I
think
the
way
will
or
one
way
I
want
to
do.
This
is
to
basically
use
the
same
thing,
we're
doing
with
C
IDs,
where
we
have
a
special
object
and
say
like
/type,
instead
of
just
/I
dang,
it
there's
an
issue
about
this
somewhere,
but
I
can't
find
it
interesting.
B
B
I,
could
we
could
go
ahead
within.
J
The
other
thing
I
had
a
question
about
is
yeah
with
the
work
Oh
with
the
graph
builder
I
was
wondering
if
there's
any
interest,
to
sort
of
like
standardize
a.
J
Adjacent
representation
of
a
a
part
of
partial
graph,
where
you
leave
out
the
intermedia
ashes
and
include
the
encoding
information
all
in
Jason
and
then
recently
I've
been
I
implemented
a
binary
radix
tree
on
a
type
of
graph
builder,
and
this
opened
up
an
interesting
question
as
to
where
things
like
this
should
live
on.
The
stack,
because
I'm
not
sure
if
it
should
go
under
should
be
a
result
or
itself
or
should
live
on
top
of
IP
LD
so
really
quickly.
J
J
J
The
whole
point
of
a
binary
radix
tree
is:
it
generates
proofs
for
key
for
values
that
are
much
smaller
than
like
aetherium
hexadecimal,
Braddock,
Street
and
I
found
some
major
performance
issues
when,
when
writing
large
amounts
of
keys
reading
large
amounts
of
keys
and
di
also
been
the
issue
of
making
a
change
to
the
way.
A
D
Hello,
so
I
have
a
little
parentheses
on
the
work
on
the
radix
three
in
the
pathing
I
think
this
is
also
something
that
we
start
a
conversation
on
a
long
time
ago
and
for
a
year
so,
and
this
called
the
IP
LD
transformation.
I
know
if
Steven
also
wanted
to
mention
that,
but
to
me
that
the
the
simplest
I
killed
II
will
will
give
you
some
form
of
pathing,
which
is
the
basic
path
scheme
that
you
will
get,
would
stand
IPL
B.
J
D
J
K
J
B
At
then,
that's
basically
on
me,
like
spinning,
becomes
SPR
every
night
you
trying
to
get
back
to
my
backlog
this
week.
I've
got
most
of
this
stuff,
but
I'm
use
that
sorry
for
that.
Oh
I
promise
to
the
way
this
thing
I'll
do
it
after
these
calls,
just
like
I.
Think
like
what
you
did.
This
like
done,
like
I,
don't
have
any
other
comment
and
yeah.
Thank
you
for
like
doing
all
the
veg
Friday
and
catching
a
problem.
B
Right,
like
you,
you
have
like
these
three
memory
and
they
just
like
flush
it
out
because,
like
the
tree
is
big
enough,
that
it
starts
consuming
a
lot
of
memory
or
because,
like
you,
just
want
to
checkpoint
it
like
you
want
to
be
able
to
just
fudge
it
to
disk.
So
that's
like
if
your
process
dies
for
some
reason,
you
can
pick
it
up
from
where
you
left
off
and
like.
If
you
don't
flush
it,
you
don't
know
what
it
was.
This
was
also
an
issue
that
we
had.
B
We
meet
em
FS
like
so
em
FS
keeps
a
virtual
directory
tree
structure
in
your
regular
stuff
and
when
you
add,
like
thousands
of
files
and
like
thousands
of
the
right
trees
at
some
point,
the
node
might
run
out
of
memory
because,
like
it
is
something
I
personally
to
solve
it
like
it
is,
but
it
might,
and
so
we
had
to
create
mechanisms
to
go
in
the
middle
and
just
like
flash
the
entire
tree.
You
know
offense
so
that,
like
if
the
nerve
crashed,
we
will
be
able
to
pick
up
from
there.
B
So
so
then,
that
tells
me
that
there
is
an
interesting
area
which
is
grab-grab
builder,
that
market
has
been
developing
and
it's
like
very
robust
Sun,
like
20%
best
coverage
and
so
on
and
like
try
to
develops
the
same
graph
building
thing
that
we're
doing
mimics
FS,
we've
graphed
builder,
for
market.
This
would
give
us
two
things.
First,
like
it
will
like
converge
in
a
way
that
that
creating
a
tree
in
memory
words,
it
will
use
the
same
language
and
second,
it
would
enable
UNIX
access
to
both
create
trees.
B
J
I
J
B
B
I
see
so
yeah
that
then
right
now,
at
least
from
memory
I
think
like
if
we
have
a
note
that
I
want
to
store
because,
like
you
already
have
to
pay
the
cost
of
hashing,
the
thing
like
we
are
like
catching
a
thing
and
trying
to
store
it
to
to
the
repo
and,
if
I'm
a
little.
Currently,
you
actually
don't
check.
B
So
what
you
need
to
do
exactly
is
read
the
data
that
go
through
a
repo
action
also
and
then
compare
the
actions
they
are
like
yeah
I
had
to
every
store.
It's
correct.
The
this
couch
that
we
have
right
now
is
every
time
we
want
to
write
something.
We
always
ask
you
to
the
right.
We
don't
confirm
if
the
data
that
was
there
was
correct
or
not
right.
J
Yes,
so
let
me
know
yeah
the
the
way
I
would
implement.
It
is
keep
a
weak
map
between
all
the
leaf
nodes
and
their
original
hashes,
and
then,
when
flushing
the
graph.
If
a
leaf,
node
half
matches
the
cache,
cache
then
return
false
and
that
entire
branch
underneath
it
would
not
be
rehashed.
B
C
B
Need
to
be
implemented
on
our
Calvi
level,
it
can
be
on
the
resolver,
though,
like
you
can
be
the
resolver
understanding
that
there
are
trees
and
quitting
it.
We
clap,
and
then
we
can
ask
like
just
options
when
we
are
doing
put
so
that
it
knows
it's
part
of
like
a
certain
tree
years
or
something
yeah
I'll,
be
totally
fine
to
include
that
as
a
feature
of
the
economy.
Resolver
slash
service
yeah
by
the
way
I
keep,
might
be
confusing
forego
people,
so
I
kill
the
resolver
that
are
I
peel.
B
J
I
have
a
partially
implemented
in
graft,
folder
and
there's
like
maybe
they
should
actually
belong
somewhere
lower
level
in
the
stack
so
I
quit
I.
The
left
forward
here
might
be
to
look
at
resolver
a
little
bit
more
and
see
where
I
could
add.
This
series,
Oliver
yeah.
J
C
B
B
Like
that
would
be
an
excellent
excuse,
slash
an
excellent
way
to
convince
everyone,
just
like
start
adopting
women
simply
for
IPL
B
and
start
developing
those
resolvers
with
webassembly.
Right
now.
We
are
also
like
specially
Friedel,
which
not
here
he
start
started
implementing
some
critter
primitives
with
rest
and
I,
compelling
you
to
allens
yes,
because
I
really.
C
B
The
Decrypter
performance
in
jessup,
if
s
is
right
now,
like
80%
of
the
time
expand.
My
ideas
of
the
test
notes
is
on
second:
do
I
like
increasing
the
packets
on
the
browser.
So
it's
come
it's
kind
of
like
a
ridiculous
chunk.
So
if
we
could
like
join
the
efforts
together
and
I
just
level
up
the
good
book,
relatives
used
by
most
template
for
everything
improve
our
ylb
works
and
making
more
performant
that'll
be
possible.
Would
you
like
to
to
to
lead
the
charge
there
or
does
it?
This
is
how
good
some
reason
yeah.
J
That
sounds
good
yeah.
Could
you
open
that
issue
for
me,
though,
for
the
awesome
stuff?
Don't
forget
absolutely
and
really
quickly
on
benchmarking,
there's
there's
two
big
bottlenecks:
there's
the
one
with
the
hashing
that
I
keep
bringing
up,
but
the
other
big
one
as
somewhere
in
JSI,
PFS,
I'm,
pretty
sure
it's
with
the
the
the
the
file
system,
repo,
but
I
need
to
dig
into
that
a
bit
more
okay,.
B
Thank
you
awesome.
So
I
would
like,
if
like,
if
there
are
more
questions
on
this
topic,
I
would
like
to
move
now
to
Steve
initial
question,
which
is
a
KPI
as
an
integration,
see.
What
I
would
you
like
to
like?
Just
bring
up
the
problem
a
little
bit
this
morning,
like
with
more
depth
and
I,
can
explain
what
you're
taking
in
the.
A
B
Often
right
like
there
was
a
lot
of
changes
in
the
early
beginning
of
IP
offense,
because
we
also
figure
out
a
lot
of
things
and,
like
me
and
Jeremy,
the
one
which
is
like
sit
down
and
like
this
guy's.
Okay,
this
part
needs
to
change
because
it's
consuming
that
same
epi.
This
part
needs
to
change
because,
whatever
and
like
as
we
refuse
like
the
dependency
free,
we
also
realized.
Oh,
we
can
also
improve
this
and
that
commands.
So
if
it
be
awesome,
if
you
could
do
that
for
go
for
DC,
never
editor
pushing
yeah.
A
A
Yeah
I'll
continue
working
on
that.
That
doesn't
actually
include
a
lot
of
the
other
eye,
peeled
he
related
auto
fiying,
the
actual
IP
LD
node
api's
is
mostly
a
local
branch.
My
computer
I
need
to
now
like
clean
up
a
bit,
but
I
was
like
all
these
emails
related.
We
have
to
a
personal,
don't
want
to
discuss
like
how
we're
going
to
reorganize
the
block,
store
and
dad
services,
something
that
was
it
guys.
Mmm-Hmm
are.
A
B
A
B
A
Problems,
the
exchange-
they
can't
actually
go
there
in
the
future,
because
the
exchange
needs
access
to
IP.
All
the
objects
no
blocks
because
integrated
with
them
I
be
sorry
to
have
IP
LD
selectors.
It
needs
to
understand
that
building,
so
it
actually
needs
to
go
up
a
level
on
top
of
a
dag
service,
or
something
like
that.
However,
now
we
also
needed
to
sing
some
tune
of
services
that
simply
store
I,
feel
the
nodes
I'm
going
to
call
them
node,
resolvers
or
node
stores,
and
things
actually
understand
like
wearing
IPL
D.
A
B
K
B
Interested
with
you
here,
questions
that
people
have
then
but
like
right
now
from
all
the
discussions
like
all
the
development
progressing,
just
at
the
best,
we
also
have
the
blog
serviced
about
services.
I
could
be
seen
veneer
that
might
clog
the
bookstore.
We
big
change,
and
that's
because,
like
it's,
what
historically
has
only
exchanged
blocks
not
like
that
we
had
in
the
past
back
service
but
like
when
we
moved
to
IPL
B
I
replaced
the
bag
service
with
the
idea
of
a
resolver.
B
A
J
A
Then
there
are
functions
that
perform
resolving
the
hazard
actually
at
the
moment,
but
that's
approximate
how
that
works
yeah
the
DAGs
dag
started.
Thinking
of
is
there
specifically
needs
to
exist
to
execute
I
PLD,
quick
or
sorry
yeah
I
peel
the
selectors.
If
we
also
been
ideally
transforms
that
would
be
done
that
level
or
that
layer
as
well,
probably-
and
at
that
point
probably
right.
Parham,
like
resolution
functions
in
terms
of
I,
feel
the
selectors
and
IPL
de
transforms,
which
case
we
don't
have
to
be
in
special.
I
B
What
does
it
sounds
like?
Is
that
yeah
so
like
when
my
building
was
implemented
in
BFS
the
transition
from
that
service
to
like
IPL
the
service
didn't
really
happen?
You
just
got
like
an
arboreal
and
it
is
sounded
like
somehow
like
the
jessica's.
Implementation
might
be
a
little
bit
more
towards
the
future
that
you
are
looking
for
enjoy
with
us,
and
so
it
will
be
very
useful
and
I
can
like
just
guide
you
through
that
all
the
things
I
didn't
JSF.
It
does
okay,
just
a
Bridget.
B
Your
understanding
yeah
like
to
get
you
under
sitting
here
because,
like
I
think
like
once
you
and
just
there
now,
it
just
depends.
You
probably
will,
like
you
know,
defy
problems
that
you
can
I
conclude,
but
you
might
also
get
like
I
just
helped
from
mapping.
Why
is
it?
Woman
didn't
get
as
plans
to
go
here
and
what
yeah
go
ahead?
One.
A
Thing
I
would
like
to
change.
I,
don't
know
how
this
works
might
be
out
there.
Ip
sorry,
ejs
land
is
the
job
I.
Don't
think
that
swap
should
be
like
a
special
service
affected
like
two
pieces.
There's
the
piece
that
watches
the
dag
and
serves
up
blocks
and
then
there's
the
piece
says
effectively:
a
dag
store
or
a
block
store,
whatever
want
to
call
it.
A
It's
alittle
just
exports,
the
block
store
interface
and
you
just
ask
it
for
Glocks
there's
like
at
least
equally
that's
the
way
it's
implemented
is
effectively
two
different
services
that
do
two
very
different
things,
but
then
we
mash
them
all
together
and
stick
it
in
this
weird
spot
and
just
doesn't
make
sense.
I
agree.
B
The
only
thing
that
who
I
would
argue
is
they're
like
going
by
doing
the
full
transition,
so
I
going
from
deck
service.
That's
now
very
weird
and
one
key.
The
integration
part
to
like
I
killed
his
service
plus
ref
swap
and
I
say
he
breasts
have
not
that
to
be
able,
I
couldn't
beat
swap
anywhere,
but
sometimes
just
like
I
wait
to
say,
I
mean
this
is
going
to
be
an
exchange.
They
like
supports
selectors
and
that's
yeah
like
it
might
be
too
much
and
that
it
might
be
more
wise
to
do.
B
C
B
A
B
A
Would
love
to
have
other
people
commenting
on
it
they're
helping
me
design
things
and
answer
questions.
There
is
a
again
I
wrote
up
an
issue
in
I
feel
denotes
repository
at
the
F
s,
notes
repository,
that's
rather
large.
You
wanted
to
try.
Reading
that
you
can
I
will
reread
it
once
I've
gotten
a
bit
more
sleep
because
it
may
not
make
any
sense,
but
unless
who
was
written
while
I
had
sleep.
B
B
I'll
take
from
the
silence:
that's,
we
should
just
didn't
find
it
to
lose
any
next
week
and
I
was
just
like
go
through
the
notes
here
that
Nicola
so
kindly
take
it
for
us
yeah
so
like
we
need
to
set
up
a
call
to
discuss
IPL
VM
json-ld,
probably
before
the
car
would
be
good
to
a
proposal
in
an
issue
like
just
like
a
context:
I'm
Tony
step
people
can
drill
and,
like
you,
we
can
even
like
use
this
call
to
discuss
this.
Also.
B
It
would
be
great
if
you
could
show
like
a
prototype
of
that,
like
what
we
can
use
from
jason
LOL
vo
and
that,
by
having
the
context,
key
to
IPO
vo
ends
would
be
possible.
That
is
not
possible
right
now,
if
you
would
like
take
some
notes
there
and
post
it
as
a
issue
posted
on
the
same
issue
that
we
have.
B
B
B
B
That's
the
semester,
maybe
twelve,
true
yeah,
okay
cool.
So
did
we
miss
to
capture
something
here
only
to
this
for
next
week,
all
right
look
looks
good
here.
Yeah.
Thank
you.
I
think
this
was
productive.
There's
a
lot
of
still
surface
area
to
go
through
like
a
lot
of
things
on
that,
be
sure
that
we
haven't
talked
about,
but
it
seems
like
we
have
some
tubers
and,
like
people
are
working
on
some
stuff
that
we
can
achieve
and
yeah.
Let's
keep
this
meeting
happening
like
so
I
asked
Shirley
asked
in
the
beginning
of
the
meeting.