►
A
A
A
A
So
my
proposal
here
is
also
in
regards
to
the
to
the
new
planning
for
for
the
new
year
for
these
meetings
that
we
cycle
the
leads
on
on
the
meeting.
So
we're
going
to
ask
for
volunteers
for
the
next
meeting,
so
I
would
say
like
cycle
every
two
weeks
or
or
month
or
four
weeks,
and
you
learn
you
lead
so
that
people
can
get
start
getting
used
to
doing
these
types
of
meetings.
If
you,
you
sure,
not
all
right
so.
A
Do
these
formats
we're
going
to
start
with
updates,
which
are
going
to
be
hopefully
focus
on
just
blockers,
and
we
count
the
number
of
people
that
are
going
to
give
their
updates
and
divide
by
the
remaining
time
before
we
hit
the
half
hour
and
that's
the
time
you
have
to
give
your
applets,
and
so
let's
and
the
second
half
will
be
for
the
rest
of
the
agenda,
discussions
or
demos
that
people
want
to
give
or
questions
that
they
have
cool
so
place
your
items
in
the
agenda.
If
you
have
them.
A
B
The
next
steps
for
this
weekend
to
the
next
week
will
be
the
animations
revision
regarding
the
colleague
commented
in
mode
which
weather
off
started
started
today,
and
then
the
farm
animations
within
the
examples
will
use
cases
for
the
entry
and
exit
and
I
will
continue
the
aim
concepts.
They
need
to
be
able
revised
with
the
Moxie
team
and
it's
big
big
Marco
and
ozone
under
occlusion
and
then
apply
the
book
and
field
the
one
that
will
be
shows
and
apply
them
to
the
other
pages.
B
B
C
So
what
I
have
been
doing
and
what
I
will
do
next
is
related
to
discussed
if
I
animations
and
what
I
did
finish
this
this
week
was
the
placeholder
has
some
kind
of
animations
the
skeleton
of
the
comments,
some
some
admission
and
also
when
you
are
scrolling
on
the
extension.
The
comments
that
are
entering
in
view
also
have
some
kind
of
scale,
and
so
I
also
finished
this
this
animation.
C
C
Complex
component
that
I
told
you
about
first
and
also
I,
did
bring
the
the
League
of
game
to
get
the
privacy
policy
so
that
we
can
land
I
specify
on
Firefox,
and
so
what
I
will
do
next?
The
fab
has
some
kind
of
animations
entering
and
exiting
so
I
will
do
that,
and
also
when
you
are
editing-
or
you
are
writing
reply
and
if
you
have
text
on
the
text
area
and
then
you
think
you
can
sell
so.
D
C
A
C
E
So,
building
on
when
I
started
last
week,
I
started
to
build
play
around
with
my
little
I'm
calling
them
sketches
he's
like
sort
of
little
simple
test
cases
that
you
can
drive
interactively
at
the
console.
Last
week,
I
demoed
turn
the
talk
like
when,
with
two
Peters
I
just
tried
to
modify
that
demo
accepts
the
state
machine
a
little
bit
so
I
could
drive
it
all
the
way
this
26
piers
and
surprisingly
nothing
works.
E
It
was
pretty
cool
and
then
I
didn't
go
beyond
that,
because
there's
only
26
letters
and
making
the
Shelf
of
that
I
can
change
my
codes
so
and
then
I
did
another
version
of
it
where
I
did
like
a
control
sample
versus
a
test
sample.
So
I
had
two
different
versions
of
the
pure
based
library
and
great
drove
them
in
synchronization
with
the
keyboard,
and
there
wasn't
really
much
difference
between
the
two
versions
I'd
chose
so,
but
that
might
be
useful
yeah.
E
You
know
in
the
future,
when
we
do
have
some
like
performance
test
cases
and
things
we'll
be
able
to
see
the
performance
like
with
their
eyes.
So
I
was
excited
to
get
that
one
working
and
I
was
really
really
interested
in
swarms.
Yes,
I
see
Victor's
on
the
car,
so
I
actually
did
one
I
watch
a
bunch
of
Vicki's
videos
and
then
I
tried
to
use
swarms.
Yes,
and
it
just
did
really
really
simple
one
and
yeah
that
one
work
so
there's
cold.
Try
that
one
out
this
is
Warren.
E
Jeffs
is
neat,
it's
sort
of
neat
with
the
graph
QL
mixed
in
there.
Okay,
then,
let's
see
yesterday,
because
I
was
working
on
all
this
stuff.
I
didn't
catch
a
regression
that
happened
last
week
where
a
pure
base
was
batching,
the
deltas
together
and
was
actually
we
have
these
ideas,
sub
collaborations
and
that
test
case
wasn't
working.
So
the
content
was
getting
mixed
together.
E
Touch
this
with
our
not
sure
why
he
didn't
respond
to
me
and
then
I
think
going
forward.
We
need
to
talk
a
bit
more
about
what
we're
officially
going
to
do
for
a
Pinner
for
pure
pad
in
terms
of
like,
if
you
reboot
like
this
pinning
to
ephemeral
disk
right
now.
So
we
need
to
have
a
better
solution
for
that.
So
we
can
reboot
the
Pinter
node
and
say
not
lose
the
data
and
I've
got
a
bunch
of
ideas
for
more
of
these
little
sketches.
A
Cool
thank
you.
Jim
I've
added
a
question
of
last
discussion
at
the
agenda
for
later
for
junior
for
sorry
right
at
the
end
pin
or
for
pure
pads.
So
let's,
let's
briefly
discuss
what
would
be
the
next
steps
for
that
later
in
this
meeting.
If
you
have
time,
okay,
okay,
thank
you.
Next
up
is,
and
in
hello,
yeah.
D
So
yeah
I've
I've
been
it's
trying
to
work
on
a
few
things
simultaneously.
Mostly
just
you
know,
chiming
a
little
bit
on
some
of
the
other
proposals
that
you
guys
are
not
working
on,
which
are
really
great
and
I'm,
trying
to
think
a
little
bit
more
about
the
peer
based
concepts
and
I
think
every
time
I,
try
and
sort
of
rewriting
a
little
because
trying
to
figure
out
how
to
take
into
account
more
than
just
what
we
already
have
built.
D
Kind
of
things
were
planning
on
and
I'm
also
been
been
poking
around
more
about
and
asking
you
a
little
bit
up
the
Lib
p2p
meeting
yesterday
for
some
advice
on
how
to
make
of
synchronizing
append-only
DAGs
work
on
public
channels
right
now.
I
have
it
set
up.
D
So
if
you
know
who
all
your
collaborators
are
in
advance,
then
you
can
just
to
them,
but
to
have
something
that
works
a
little
more
like
how
pub/sub
works
or
how
peer-based
works
is
is
important
so
trying
to
figure
out
that,
and
also
how
maybe
to
separate
this
out.
It
looks
like
some
of
the
things
that
we're
doing
in
peer
based
are
things
that
Lupita
piece
should
be
or
could
be
handling
for
us.
D
A
F
F
So
what
this
means
is
that
we
had
to
do
some
kind
of
tinkering
in
other
to
support
old
data
structures
and
also
new
ones,
because
the
older,
the
older
ones
was
were
stored
on
DAC
PB
and
any
ones
were
obviously
start
using
black
C
bar
we'd
IPL
degrees,
and
they
wanted
to
have
our
BTB
data
by
his
work
on
previous
data
structures
and
also
new
ones.
So
this
was
kind
of
complex,
but
it
was
merged.
F
Proposal
of
community
in
terms
of
the
data
structures,
so
I've
made
a
kind
of
an
egg
there
in
order
to
do
that,
but
I
also
made
the
proposal
a
more
formal
proposal
on
how
to
handle
breaking
changes
on
the
data
structures
in
the
future
and
also
to
the
public
API.
You
have
the
link
there
to
my
proposal,
but
essentially
is
like
a
migration
by
two
islands
of
some
sorts.
We
runtime
migration,
which
means
that
it
happens
on
on
with
operations
and
not
a
full
migration
of
the
of
the
of
the
data.
F
So
there's
the
proposal
there,
we
have
some
some
discussions
on
pure
base
as
well
related
to
these
topics,
because
as
far
as
I
know,
we
have
also
that
instructions,
internal
internal
data
structures
of
clear
bias
like
the
membership,
CR,
DT
and
so
on,
and
if
you
want
to
support
backward
backwards,
have
backwards
compatibility
on
those
data
structures.
We
also
need
to
address
that
somehow,
and
even
developers
like
like
me
that
that
might
specify
and
also
developers
that
made
a
pure
pure
path.
We
also
probably
want
to
have
some
kind
of
backwards
compatibility.
F
So
this
is
the
thing
that
happens
both
internally
on
peer
base
and
also
externally
for
developers
that
build
build
upon
peer
bytes.
So
I've
made
a
proposal
there
and
perhaps
you
could.
You
could
read
it
and
perhaps
have
some
sort
of
discussion
and
later
on
on.
How
can
we
address
that
on
on
pure
base,
as
well?
F
Also
I've
been
I've
been
reviewing
a
few
pull
requests
on
on
a
previous
work
related
to
the
things
about
Versa
dog
versus
FS,
for
that
many
contributors
or
if
you
consider
making
I've,
also
had
at
that
time
out
support
for
reading
operations
on
on
FS
law
because
they
didn't
have
the
time
out
so
I
added
that
I'm
not
really
satisfied
on
the
solution.
That
is
there
because
it's
too
complex,
perhaps
to
explain,
but
it
will
probably
change.
F
So
this
is
a
very
important
workshop
and
me
and
patrol
alongside
with
two
hands.
One
will
go
there
in
terms
of
what
I'm
doing
now
I'm
integrating
on
that
aims
to
break
down
dog
and
I'm
I
want
to
make
an
apologize,
because
I
haven't
had
many
time
to
actually
respond
to
all
the
feedback
that
I
have
in
the
document,
but
hopefully
next
next
few
days.
I
will
have
that
time,
because
I
I've
been
like
switching
context
between
ffs
log
and
identity
and
other
things.
F
So
next
time
and
I
think
we
should
create
issues
for
each
section
on
that
entity,
breakdown
dog,
so
that
we
can
like
migrate.
What
we
have
done
in
the
document
to
eiv
shoes
and
start
working
and
there
from
there
and
I-
will
also
schedule
some
brainstorm
sessions
on
a
few
topics
of
the
identity
based
on
topic,
because
I
think
those
topics
actually
need
to
be
discussed
with
the
team
and
evolve
involves
some
people
in
order
to
conversion
some
on
some
topics.
F
A
A
G
B
G
So
all
the
relevant
stuff
is
here
examples
and
so
on.
So
basically
it
is
innovation
which
has
like
a
long
port
like
ooh,
logical
things,
terms
attached
in
different
pieces
of
data,
so
it
is
able
to
always
measured
using
civilities
and
it
also
has
annotations.
But
basically
the
miracle
tree
is
derived
from
the
tree
of
creations
and.
G
G
What
can
we
do
useful
this,
so
basically
I'm
trying
to
make
little
embedded
database
for
mobile
devices,
which
may
also
work
using
rep
assembly
in
the
browser
so
basically
embedded.
The
degrees
in
C
is
a
very
hardcore
topic,
so
I'm
using
the
proxy
B,
which
is
specifically
like
Allegro
database,
which
is
like
a
database
for
building
other
databases,
so
also,
first,
con
stuff
is
done
by
as
I
smart,
well-paid
people
and
I'm
just
doing
the
CDP
and
hash
apart
so
currently
is
in
very
prototype
basic
condition.
So
it
kind
of.
G
G
G
So
this
mode
is
what
I'm
working
on
and
currently
I'm
kind
of,
making
it
dimmable
so
I
have
the
basic
engine
working
Mallis,
so
I'm,
adding
like
data
formats
and
everything
today
like
I,
did
a
very
preliminary
version
of
the
global
array
we
just
like
our
main
city
by
virtually
for
a
collaborative
editing
and
everything
like
that.
So
this
is
not
modest.
G
G
G
He
basically
explains
it
very
fine
details,
convergent
evolution
of
jellies
and
like
this
brand
of
Serapis,
is
with
disabilities
and
I
mean
initially
it
was
like
obvious
or
state
by
solidity,
and
they
started
conversion.
So
this
wrong.
It
is
kind
of
outpaced,
but
at
the
same
time
built
enabled
if
you
wish,
so
it
is
like
very
close
to
them
disabilities,
because
it
has
the
same
data
structure
for
such
as
hundred
state
and
operations.
So
it
is
not
exclusively
operational,
more
consumption,
so
it
is
like
one
of
those
commercial
schemes,
so
so
more
or
less.
E
G
G
G
G
G
G
It
was
a
binary
version,
because
these
days
it
is
not
really
difficult
to
parse
binary
in
JavaScript
and
and
mas
will
be
updating,
hopefully,
small,
ingest,
2.1
I
think
we'll
be
using
the
binary
version
of
the
product
on
that
and
the
text
version
is
much
easier
to
eat
and
embark
on
a
mission
so
but
well
in
jars
country.
It
is
a
bit
painful
to
parse
it
in
terms
of
performance
and
everything
in
sepals,
Pasco
and
others.
A
A
D
D
G
Yeah,
so
it
is
not
about
the
formal
model
grandma
it's
about
the
formal
model,
because
it
it
is
effectively
embedded.
It
was
a
database,
so
it
needs
to
be
really
very
strict
to
make
sure
it
works,
always
no
matter
what.
So.
There
is
a
very
strict
formal
model.
What
is
allowed,
what
is
not
allowed,
how
it
works,
so
any
limitation
must
work,
did
mystically
exactly
display
and
actually
that
formal
model
it
is
very
upset.
There
is
some
internal
for
Hamilton,
which
is
very
simplistic.
Like
everything
is
16,
bytes
I
will
fill
the
16
bytes.
G
Everything
is
fixed,
we
stand
for
boy,
smallest
corresponds
to
in
memory
layout
and
when
it
calculates
caches,
it
is
using
that
internal
yeah,
but
that
we
don't
know
Oh
yard
is
basically
painful
to
see
realized
because
it
takes
too
much
space.
So
there
is
text
based
civilizations.
There
is
binary
civilization,
protoman
civilization,
absolutely
no
problem.
Jason
Bay
sterilization
no
problem.
Basically,
there
is
some
click,
normal,
no
internal
format,
each
kind
of
the
definition
of
fiends
and
any
serialization
that
can
be
derived
from
sin
turn
off
or
bottom.
G
Parts
back
into
internal
format
is
a
good
sterilization,
so
binary
Jason
when
literature
is
also
absolutely
no
problem
and
well
and
in
case
of
the
text
by
it
based
serialization.
Well,
formal
grandma
role
is
necessary
to
make
sure
it
always
passes
correctly.
Deterministically,
basically
I,
using
a
poinsettia
generator
to
make
the
parser,
which
is
5,000
lines
and
which
is
most
likely
correct,
because
I'm
using
the
regular,
which
is
like
heavily
used
in
high
load
applications
by
CloudFlare
and
others.
G
D
G
Kind
of
yes
sure,
I'm,
actually
more
reliant
on
your
IDs
and
I,
mostly
use
hashes
for
integrity,
checking
because
hashes
have
no
internal
structure
and
it
is
difficult
to
parasite
well
Natasha's,
but
you
can
lead
to
around
hoc.
You
need
a
double.
You
need
a
table
of
also
has
a
side
like
distributed
hash
table.
G
Tell
about
the
you
need
all
of
them
in
memory
and
if
you
can
see
like
IP
addresses
or
anything
like
that,
they
have
was
the
structure
which
allows
to
like
aggregate
them
in
batches
and
greatly
demise
of
the
screens,
so
I'm,
actually
using
video
ad
for,
like
routing
or
different
purposes
and
hash
for
integrity,
checking
and
doing
my
best
to
like
use
hashes
implicitly
without
actually
storing
them.
Most
of
the
time.
I
like
it
very
derive
exam
on
the
fly.
A
H
Gratefully
I
shouldn't
working
on
the
IP
FS
whip.
You
re
I
have
another
thing
particular
relevant
to
the
dynamic
they
agreed
to
add,
but
we've
been
fixing
bugs
with
uploading
directories
and
multiple
directories
from
dragon
draw
and
I
hope
you,
and
this
week,
I've
had
to
think
of
the
Lib
p2p
ground
and
hedgerow
with
the
dynamic
data
stuff
and
hopefully
Andres
Souza
Petrov
about
the
the
UI
concerns
that
your
team-
that's
it
so
maybe
more
next
and
next
Jacob.
A
So
next
up
there
is
a
bunch
of
questions.
When
will
we
discussed
in
new
name
of
pure
base?
First,
one
yeah
right
away.
We
should
be
discussing
that,
and
not
necessarily
in
public
forum,
so
because
of
all
these
obvious
reasons,
but
we
will
we
will
that
really
soon
trademark
yeah
really
soon
prioritize.
A
F
You
want
to
scrub
that
yeah
yeah,
it's
related
to
the
Pinner,
which
is
the
ability
to
pin
any
cid
which
is
currently
blocking
us
of
having
the
pin.
I
intuited
on
this
classified
because
the
CR
dt
itself,
even
if
that
speciality
itself
would
be
pins.
The
comments
themselves
wouldn't
be
loaded
because
they
are
not
peanuts.
So
we
need
that
kind
of
feature
on
the
Pinner
so
that
we
could
type
in
DC
ad
for
me
and
the
Pinner
would
think
that
cid
so
yeah.
This
is
come
easily
coming
like
a
blocker
for
us.
F
A
E
A
It
has
to
be
apt
level
because
the
the
Pinner
doesn't
know
the
the
contents
of
the
strategy
not
able
to
to
decrypt
if
it's
a
private
collaboration,
it's
so
it's
up
oriented,
but
even
the
app
could
be
peer.
Bias
could
be
some
of
some
help
on
on
there
right
so
as
it
just
try
to.
If
he
finds
the
Pinner,
it
does
try
to
pin
the
charity.
A
It
could
also
somehow
help
pinning
the
into,
but
if
there
was
a
reference
to
nappy,
FS
idea
to
could
try
and
pin
and
pin
that
or
even
get
that
locally
and
then
and
replicate
that
locally,
so
that
there's
there's
there
are
some
finer
points
there
that
we
should
also
we
don't
want
to.
Well,
we
want
you
don't
want
to
recreate
that
EFS
infrastructure
on
our
side.
A
F
Think
from
the
from
the
consumers
of
pure
bytes,
the
developers
that
used
the
pure
base
module.
It's
kind
of
kind
of
weird
to
not
have
this
feature,
but
I
agree
that
the
solution
that
we
will
use,
we
could
like
take
a
few
parts
different
parts
and
we
must
discuss,
discuss
the
advantages
and
shortcomings
of
each
of
each
of
one
and
just
choose
one.
For
now
and
the
moment.
F
A
D
Some
googling
around
I
saw
that
there's
a
project
on
github
of
trying
to
create
like
a
sort
of
standard
interface,
for
a
created
for
doing
like
ipfs,
pinning,
which
you
may
want
to
take
a
look
at
right.
I
think
perhaps
the
way
to
do
this
is
we
have
our
you
know
peer
based,
spinner,
which
then
is
allowed
to
call
out
to
an
IP,
FS
Pinner,
and
that
way
we
don't
have
to
replicate
anyone
elses
work,
but
we
also
have
an
interface
that
we
can,
you
know,
switch
around
and
then
we
can
have
different.
D
A
That
was
also
I
guess
there
was
no
someone
else.
I
proposed
that
and
I
think
it's
a
good
approach
where,
where
it's
just
just
a
proxy
for
well
the
pure
base
spinner
is
a
proxy
and
that,
in
that
part
of
the
of
the
persistence,
is
just
a
proxy
to
to
an
actual
Pinner
which
could
be
the
public
IP
FS
cluster.
A
A
What
this
topic
anymore,
I
didn't
know
all
right.
Let's
move
on
so
next
one
is,
should
measure
the
benefit
and
how
hard
it
would
be
to
use
any
kind
of
charity
pasted
by
Delta
charity
run
basically
the
benefit
and
how
hard
it
would
be
to
use.
F
This
was
my
question
as
well
in
terms
of
the
pure
lines,
so
in
terms
of
fees,
but
now
we
just
supports
Delta
by
these
charities
and
if
we
really
see
the
value
of
using
the
operational
base,
charity
for
some
kind
of
situations
or
even
run
Bay
is
I.
Don't
know
if
that
makes
sense.
But
let's
say
that
new
kind
of
Sharon
he
comes
over
in
the
future.
A
That's
a
good
question.
My
intuition
is
that
is
that
we
shouldn't
just
just
confine
ourselves
to
daughter
and
by
charity,
because
it
may
not
fit
so
so.
I
guess
the
basic,
primitive
would
be
a
causal
tree.
Cause
of
a
console
tree
would
be
the
basic
primitive
for
up
biased
causes,
realities
and
that
would
fit
and
and
easily
you
could
adapt
any
use
any
existing
library
for
a
based
charity
that
is
more
or
less
pure
I.
Thinking
about
the
Jason
charity,
one
most.
A
A
On
top
of
that,
so
I
think
that's
why
I
want
to
start
trying
I
profess
what
converse
the
deck
and
see
how
well
it
fits
with
with
pure
base,
because
that
could
be
a
good
foundation
for
for
for
that
right
right
now,
there's
no
there's!
No
reason
I
mean
the
user
is
does
right
now
the
the
pure
base
user
does
not
perceive
that
it's
a
delta
by
dt
or
it's
not
based
charity
is
real
transparent.
So
there's
no
really!
No!
Nothing
locking
us
into
delta
based
charities.
A
A
Just
show
you
yes,
because
yes,
because
were
right
now
we're
extending
the
one
only
the
extent
it
takes
they're,
extending
the
the
austerities
singleton
library,
library,
singleton,
yeah,
no
you're
right,
but
that's
easily.
That
could
be
easily
configurable
III.
Believe
we're
still
in
time
to
to
to
help
on
that.
A
F
Know
when
I
came
I'm
asking
this
is
this
just
because
if
we
want
to
like
adapt
or
adopt
or
even
use
other
as
TLC
oddities,
that
you
mentioned
not
emergent
and
so
on,
I
would
see
like
several
modules,
several
adapter
modules
like
say
pure
base
autumn
urge,
or
something
like
that.
That
will
be
like
the
adapter
between
the
outer
merge
and
our
our
replication
strategy
and
an
adapter
for
the
pure
base,
and
we
will
have
like
a
weak
ecosystem
of
adapters
for
all.
These
kinds
of
civt
is
on
tie
possibilities.
Victor.
G
We
see
it
in
a
way
that
there
is
directed
acyclic
graph
of
changes
right
and
these
are
like
static
data
pieces
which
replicated
over
the
network,
and
that
is
like
the
lower
interface
and
the
upper
interface
is
some
sort
of
user
consumable
data
structure,
subjects,
Jason
and
everything
in
between
well
is
some
sort
of
certainty
in
our
case,
so
these
are
like
very
clear
in
their
faces.
Exactly
I
didn't
you
wanted
yeah.
D
I
think
the
different
types
I
don't
know
I've
been
noticing
so
I
was
thinking
about
sort
of
the
deltas
in
the
office,
then
state
based
CRT
T's
and
how
we
can
sort
of
make
use
of
all
of
them.
If
we
need
to
I
think
they
start
to
affect
the
replication
strategy
as
well
and
I
was
trying
to
think
how
to
use
sort
of
the
lid
p2p
style.
Like
we'll
have
a
you
know.
D
You
know
uptick,
you
know
off-base
CR
DT
to
do
their
to
do
what
they're
doing
and
that's
that
only
works
because
they
sort
of
have
these
always
online
Message
Queuing
servers,
yeah
I
may
have
misunderstood,
but
that
was
what
I
understood
from
our
phone
call.
You
know
whatever
a
month
ago,
so
I
don't
know
if
it's
gonna
be
quite
as
simple
as
like
create
this,
you
know
create
a
causal
tree,
put
things
on
top
of
it.
I
think
we
can
have
user
I
think
we
can
have
something:
that's
nice-looking
to
the
users.
D
A
You're
absolutely
right
the
the
replication
strategy
there.
It
has
to
be
different
because,
because
there's
a
lot
of
underlying
assumptions
that
mean
that
that
Delta
Charities
implements
that
tapas
charities-
just
just
just
don't
need
well,
they
have
the
different
early
from
representations.
So
when
they
have
different
replication
strategy,
your
your
your
stats
right,
I,
guess
that
replication
strategy
has
to
match
somehow
somehow
match
the
reality
type.
A
I
would
say,
because
if
it's
a
causal
merkel
bag,
you
could
do
use
IPL
the
if
s
upheld
the
graph
swap
or
graph
sync
or
whatever
replicate
the
job
set
well
the
battery,
but
there
could
be
other
strategies
for
faster
if
you
want
like
real
time.
Perhaps
this
is
real
time
and
a
lot
of
fine
grained
operations
like
if
you're
doing
a
text
editor.
A
Perhaps
this
isn't
a
lot
the
best
strategy,
and
so
you
could
plug
different
strategies
that
we
could
define
or
our
plugin
different
strategies
for
for
application,
which
is
something
that
I
believe
could
be,
could
could
be
a
solution
to
to
to
this
and
they
could
live
well.
We
all
have
to
invent
every
when
some
type
of
grammar
and
that
will
allow
for
both
of
this
to
coexist
but
or
grĂªmio
some
some,
some
language
that
will
offer
this
to
co-exist.
A
You
could,
in
theory,
use
use,
Delta,
state
based,
replication
and
and
then
Merkel
diagram
location
on
top,
because
we're
already
doing
doing
that
for,
for
instance,
the
misclassify
we
have
like
a
delta,
c
oddity,
and
then
we
have,
if
you
want.
We
have
history
like
a
causal
history,
replicated
as
it's
like
a
side
chain,
and
so
there's
nothing
preventing
this
from.
I
F
Just
mentioned
there
and
I
think
it's
worth
discussing.
These
are
resuming
this
discussion
when
we
make
and
I
think
he's
adding
that
he's
doing,
that
the
document
that
breaks
down
periods
on
on
interfaces
and
so
on.
Perhaps
this
is
a
point
I
want
on
the
collaboration
parts
or
our
section
to
discuss
Travie
further
discussion
brainstorm
session.
If
it's
worth
or
not,
and
if
it
does,
then
how
can
we
approach
this
right?.
A
Yeah,
some
of
you,
the
the
developer,
experience
the
the
the
API
I
think
it
should
be
really
intuitive
for
for
a
beginner,
so
that
people-
just
just
maybe
if
they
don't
know
shiera
t
they
don't
care
about.
What
I
saw
I
saw
thought
about.
This
want
to
create
a
just
like
structure
or
an
array
or
assets
or
something
they
should
be
really
straightforward,
and
they
shouldn't
have
to
think
about
that.
E
I've
been
thinking
about
this
too,
because
I
was
playing
around
with
I,
worked
on
the
Auto
merge
project
with
that
previously,
and
then
you
know
obviously
the
Delta
C
rdt
stuff
and
then
playing
around
with
sworn
GS,
and
they
they
all
three
of
those
have
like
radically
different
sort
of
programming
interfaces.
You
know
swarm
GS
is
exposing
a
graph
QL
type
interface,
which
is
really
nice
for
saying
react.
Yes,
programmers.
E
You
know
the
Delta
C.
Our
duties
is
sort
of
nice
because
you
can
pick
and
choose
specifically
which
CRT,
T's
and
there's
a
whole
menu
of
different
Sierra,
T's
bt's.
You
can
use
and
the
auto
merge
sort
of
it's,
because
this
just
seems
like
JSON,
so
like
a
JavaScript
programmer
here
knows,
JSON
can
sort
of
sort
of
croc
it
fairly
easily.
So
I
don't
see
like
there's
a
lot
of
benefit
in
having
like
a
unified
programming
interface.
A
Absolutely
right:
yeah
I,
don't
I,
don't
believe
that
they're
unified!
Well,
that's
it's
just
my
view,
but
I
don't
know
that
the
unified
interface
is
something
to
strive
for
I
I,
don't
I,
don't
really
care
I'll,
just
care
about
the
ease
of
use
for
first,
first
comer
and
and
yeah
you're
right,
there's
the
in
terms
of
legally
PGP,
this
all
being
something
that
perhaps
lit
b2b
should
be
solving.
There's
a
gossip
sub.
Well
in
the
in
that
net
chapter,
there
is
gossip
sub
implementation
that
happened
long
ago
and
should
will
probably
happen
on
Jas.
E
Like
I
think
like
because
we're
ipfs
and
we
have
serve
a
unique
advantage
in
that,
we
can
drop
things
into
the
ipfs
daemon
that
will
be
on
thousands
and
thousands
of
things,
and
the
problem
with
running
things
in
the
browser
is
browsers.
Can't
truly
talk
peer
to
peer.
They
have
to
go
if
your
gateways,
so
we
can
we're-
probably
the
only
people
in
the
world
that
has
this
ability
to
do
like
a
massive
sort
of
the
public
infrastructure
that
people
can
use
in
browsers,
or
is
this
stuff.
A
A
A
A
D
You
yeah
yeah,
so
of
all
of
the
types
of
synchronizing
of
graphs
and
poor
names
for
them
there
for
at
least
a
few
of
them.
There's
a
meeting
tomorrow
to
figure
out
sort
of
get
an
implementation
underway
so
that,
instead
of
IP,
instead
of
just
using
bits
law
and
asking
for
the
hashes
that
you
need,
if
you
ask
for
and
I
peeled
the
objects
right,
the
root
you
should
be
getting
the
leaves
with
it
too,
and
and
also
the
ability
to
have
other
sorts
of
you
know,
queries
with
that.
D
D
E
This
started
getting
ahead
of
where
I'm
at
but
I
was
doing
that
experiment
where
I
was
trying
to
represent
at
in
IP
LD,
and
you
know,
I
wasn't
worried
about
replication
yet
because
you
know
bit
swap,
but
obviously
like
that
has
its
replication
algorithm,
which
is
just
you
know,
pulling
on
bits
of
binary
tree
and
syncing
it.
So
obviously,
but
I
would
think
that
graphs
link
with
just
if
the
graph
sink
is
doing
his
job
at
all.
It
would
cover
like
a
simple
binary
tree
syncing
situation,
so.
A
Food,
the
full
subset
I,
think
our
use
case
or
simple
to
use
case
is
get
you
get
the
root
and
then
then
get
get
the
whole
the
whole
tree,
because
I
believe
getting
a
part
of
the
tree
is
not
really
useful
to
to
a
CRT.
I
may
be
wrong,
but
I'm
not
seeing
an
use
case
for
us
to
be
to
get
the
partial
replication
or
be
at
least
be
notified
or
over
an
intermediary
state
of
the
replication
yeah.
E
E
So
yeah,
it's
just
it's
just
the
it's
it's
model,
it's
just
it's
like
it's!
It's
really
not
much
different
than
I.
Think
what
we're
talking
about
with
ipfs
login
that
where
it's
like
you
know
it's
a
tree,
and
you
want
to
just
get
sparse
to
get
a
sparse
set
of
records.
You
know,
maybe
a
million
records,
the
only
one
you
know
secured
500.
You
know
you
don't
want
to
sink
the
entire
there.
A
Very
nicely,
my
my
requirement
would
be
sink:
an
entire
bag
of
of
operations,
which
could
be
potentially
be
a
huge
tree
or
very
granular
fine-grain
operations
and
sink
that
I
could
do
the
single
across
as
as
quick
as
possible.
Some
nodes
I
may
have
locally
some
nose.
I
may
not
have
so.
There
is
like
there.
There
may
be
some
concurrent
root
root.
Hashes
and
I
want
to
get
all
of
the
concurrent
root
hashes.
E
D
A
A
F
Also
I
would
like
to
say
that,
and
he
specify
or
one
of
the
goals
of
the
specify,
was
to
have
the
data
of
the
comments
in
the
discussions
open
to
them.
So
at
the
moment
they
aren't,
they
are
the
aren't
really
start
on
ipfs
in
terms
of
having
like
a
tree
a
graph,
but
it
was
one
of
our
goals
in
terms
of
like
get
this
Eid
and
you'll
have
like
the
full
discussion
link
through
the
to
the
tree
of
comments,
because
threads
are
linked
to
that
the
parents
and
so
on.
A
I
think
that
this
also
deserve
deserves
a
more
discussion
of
were
ahead
of
time,
but
one
thing
I
I
would
like
to
add
to
that,
and
we
can
discuss
this
later,
is,
is
the
having
a
if
possible,
having
it
streamable.
So,
if
I'm
getting,
if
an
you
know,
he's
getting
a
set
of
operations,
I
could
thought
the
Dussehra
T
works,
the
opposite
if
he
works
by
reducing
the
all
the
operations
into
into
the
state
right.
A
A
Blocking
should
be
should
be
streaming
every,
but
but
we
we
could
just
if
you
have
like
like
something
that
you
want
to
propose
and
then
like
a
very
quick
document,
you
want
to
propose,
go
and
convey
to
tomorrow.
For
for
the
meeting,
we
could
also
be
on
tomorrow's
meeting.
If
you
could
just
just
point
us
to
the
issue
where
they're
planning
that
that
would
be
also
be
useful.
D
Yeah
I
think
they're
and
they're
trying
to
just
keep
the
meeting.
It's
I
think
I'm
sort
of
in
the
meeting
Sam
I
by
accident,
from
when
people
thought
the
original
graph
synchronization
stuff
that
I
was
working
on
was
the
same
as
this
there's
a
few
different
proposals
that
they've
got
everyone
sort
of
looking
at
all
the
context
ahead
of
time,
and
they
just
want
to
do
that.
I
think
the
things
that
we've
asked
for
all
things-
people
other
people
have
already
asked
for
which
is
good.
D
Okay,
I
think
I'm,
mostly
yeah
a
fly
on
the
wall
in
that
meeting,
because
there's
a
lot
of
people
who
spent
like
they've
been
graphs.
Think
proposals
going
back
for
like
two
years
or
something
so
yeah
I
want
to
I
think
I'm,
just
gonna,
mostly
let
them
let
them
run
the
show.
Although.
A
It's
a
matter
of
prioritization
if
there's
like
a
very
small
subset
of
of
a
feature
that
we
may
want
that
from
that,
and
if
we
could
get
if
the
fly
on
the
wall
could
could
just
push
four
four
four
four
four,
those
or
at
least
say
that
we
we
need
them
now
in
order
to
do
the
efficient
replication.
That
would
be
fantastic.
D
A
So
a
were
way
had
a
lot
of
time
were
past
eleven
minutes
from
from
the
tower
I'm
going
to
end
this
here.
Thank
you,
everyone
for
coming.
Thank
you
ollie,
for
coming,
welcome
and
doing
the
notes
and
thank
you
for
everyone
to
do
for
for
their
the
contribution
and,
let's
see
each
other
next
week
same
time.