►
From YouTube: CRDTs in IPFS by Pedro Teixeira
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
A
So
if
you
try
to
use
the
PO
box
without
any
plugins
no,
so
this
is
what
happens
when
any
plugins
on
a
fly
he's
trying
to
connect,
and
you
cannot
edit
the
documents,
while
he's
trying
to
connect
with
planes,
there's
a
plugin
that
you
can
use
for
that,
but
spreadsheets
for
assistance
in
support
of
one
at
all.
So
this
is
makes
me
sad.
A
I,
don't
know
what
you
and
also
makes
me
a
bit
worried
that
Google
is,
has
access
to
at
least
the
method
you're
right,
so
it
has
to
synchronize
between
years
what
the
changes
are,
and
so
for
that
they
have
to
know
something
about
your
document,
if
not
the
document
at
all.
So
is
that
all
about
relative
to
to
this?
So
let's
just
press
some
some
goals.
First,
so
I
want
to
have
a
net
word.
A
I
want
for
so
much
to
enable
decentralized
collaboration
between
two
or
more
tiers
for
editing
any
type
of
data
like
a
document
or
a
set
a
Mac
and
in
this
network
I
didn't
want
it
to
be.
No
note
should
be
special,
so,
although
notes
should
be
only
users,
so
if
I'm
collaborating
with
the
user,
I
just
want
to
note
in
network,
and
no
one
has
special
privileges,
so
I
want
in
real
soft
real-time
update.
So
if
you
make
a
change,
I
eventually
get
it.
If
I
make
a
change,
we
eventually
get
it
and
really
important.
A
This
I
want
the
stage
to
convert
eventually
converge,
so
I
want,
as
if
two
people
are
collaborating,
I
want
to
see
if
we
stop
collaborating
and
we
connect
each
other
or
three
people
are
collaborating
and
they
all
eventually
connect
to
each
other.
I
want
them
to
end
up
on
the
same
same
right.
So
cities
are
a
heart
problem.
That's
why
Google
has
a
centralized
service
to
handle
the
synchronization
between
interviews
right
also,
I
want
to
support
offline
first,
so
I
want
to
be
able
to
like
I'm
gonna
train
on
the
plane.
I
need
more
flying.
A
I
want
to
be
able
to
edit
my
document
on
my
data
structure
and
I
want
it
when
I'm
reconnected,
I
want
the
my
changes
to
sync
to
all
the
other
peers
in
the
seamless
way.
I
want
to
get
the
way
to
connect
message
or
I
want
to
be
able
to
to
worry
about
I'm
offline,
so
I
also
want
the
system
to
support
different
types
of
then
I
want
to
edit
text
I
want
to
edit
may-maybe
god.
A
Relevance,
encounters
really
simple
data
structures
or
more
complex,
like
maths
or
maps
of
maps
of
maps
of
arrays,
and
and
enable
this
decentralized
applications
to
define
data
types
that
can
be
synchronized
between
peers,
complex
data
types.
So
this
is
a
very
vicious
code
right.
So,
first,
let's
then
there's
a
there's.
Google
Google
Docs
implements
something
that
addresses
this
more
lens,
which
is
called
operational,
transforms
anyone
heard
of
operational
transforms
okay,
so
they
don't
feel
all
all
the
tips
on
that
list.
So
it
is
centralized.
It
requires
some
centralization
of
coordination.
A
So
it's
I'm
not
going
to
explain
that
really
thoroughly,
but
it
has
another
set
of
the
clients
in
order
to
order
the
operations
in
the
right
way,
so
that
everyone
sees
the
same
thing
at
the
end,
all
right.
So
this
is,
if
other
centralized
and
doesn't
he
doesn't
a
lot
for
offline
first
uses
right.
So,
let's
see
how
we
can
map
this
on
top
of
what
way
to
describe
IP,
offense
and
actually
I
feel
Lee.
Is
you
remember,
David,
the
vid
sorry
shows
some
grass,
so
graphs
are
either
play
their
link
there.
A
A
7
rights,
so
let's
say
that
our
edits
this
book
and
I
create
a
new
version
of
the
document,
but
so
since
they
have
changed,
has
a
new
batch
right.
So
this
doesn't
work
because
I
have
different
hashes
I'm,
not
able
to
collaborate
in
any
way
because
I'm
producing
new
data,
so
they've
been
also
ended
into
what
the
solution
for
this
you
have
an
end
so
be
points
to
the
head
of
my
heart.
B
A
Ips
and
now
I
point
to
the
inversion
right.
So
does
this
support
multiple
writers,
because
I
want
to
collaborate
with
other
people
right
so
I
want
to
collaborate.
So
what
does
it
mean
collaboration?
It
means
that
multiple
people
are
writing.
So
if
Google
Docs,
for
instance,
you
are
writing
I'm
writing
let
this
work.
So,
let's
see
if
we
have
like
two
actors.
A
A
B
A
So
here
comes
charities
which
are
hopefully
replicated
data
types.
The
big
words,
then,
is
that
this
is
the
fruit
of
kind
of
recent
research
done
by
some.
Some
very
interesting
academics
like
Marcia
period
in
Ferguson,
comes
back.
I
know:
let's
leave
the
Seminole
the
initial
paper
published
in
2011.
A
So
it's
a
70
is
a
data
structure
which
can
be
replicated
across
multiple
computers
in
the
network
where
the
refugees
can
be
able
to
live
independently
and
apparently
without
coordination
to
fit
the
replica
some
depth
of
special
notes
right
and
where
it
is
always
mathematically
possible
to
solve
conflicts
right.
So
this
is
kind
of
what
we
want
right,
the
news.
So
there
are
many
different
types
of
charities
as
charity
is
not
something
very
concrete.
It's
just
always
a
bunch
of
mathematical
rules
right.
A
A
A
So
we're
all
operating
I
am
a
doctor,
but
you
have
your
movement
that
someone
else
has
documents,
so
we
all
have
different
replicas
of
the
same
data
set
right,
the
1
0
2
3
each
replica
has
their
own
internal
state,
okay
and
they
all
communicate
with
each
other,
using
in
a
synchronous
network
to
exchange
messages
between
them.
Okay,
so
this
replica
part
is
kind
of
leasing.
The
the
rest
part
would
be
the
network
layer.
So
what
are
the
network
requirements
for
this
to
be
able
to
work?
A
What
happens
if
I
lose
the
message,
for
instance
off?
If
the
message
is
delivered
first
and
then
another
one?
What
happens?
It
depends
on
the
type
of
sanity.
There
are
two
main
types
of
sanity's:
you
have
the
convergent
ones
and
it's
the
conversion
seventies,
which
have
CP
r
DS,
is
why
the
state
is
broadcasted
on
every
message.
So
those
messages
that
you
saw
before
contain
the
entire
state
of
the
strategy.
A
For
instance,
if
I'm
sharing
a
Google
Doc,
it
means
that
and
the
Google
Doc
is
represented
using
the
CBI
VT
means
that
I'm
setting
the
document
on
every
message
right.
So
this
allows
you
this
makes
the
decision
property.
So
it's
commutative,
which
means
that
messages
can
be
swapped
in
order
associative,
so
messages
can
be
moved
together
in
a
form
another
message
and
are
important.
So
we
fight
really
put
the
same
message
twice:
it's
ok
right!
Ok,
so
it
supports
message,
loss,
duplication
and
after
water
delivery.
A
A
paragraph
to
the
document,
I
generate
the
message
to
all
the
viewers
that
describes
them,
inserting
this
character
at
this
position,
for
instance.
So
this
is
this
is
more
fragile
LED,
because
it
only
has
the
commutative
property
I
can
change
the
order
of
the
messages,
but
I
cannot
loose
or
duplicate
messages.
A
Okay,
so
which
means
that
if
I
in
CMR
T's,
if
I
have
r1
created
an
operation
and
21st
transmitting
that
operation
to
r2
and
the
message
is
struggling
through
network
and
then
there's
operation
to
nethers
operation,
3
r1
is
making
some
changes
and
if
the
overlocked
operation
is
changed
and
they
aren't
assumed
by
r2
received
by
m2
and
r2
of
less
the
states
accordingly,
they
both
get
the
same
state.
They
both
converge.
Okay,
now
that's
what
gets
done
because
sometimes
in
C
marketÃs
you
can
have
causal
dependency,
which
is
very
simple
thing.
A
So
I
had
like
three
operations
but
happens
that
operations
really
depends
on
operation.
One
having
had
been
received.
Some
strategies
have
this
built
in
I
can
explain
why.
But
it's
the
city's
requirements
for
almost
all
complexities,
because
my
face
is
fine.
Removing
something
this
is
a
remove
message
means
that
I
must
have
had
it
before
right,
for
instance,
so
to
solve.
There's
some
similarities.
Embed
in
the
message
cause
dependency,
those
links
between
operation
3
in
version
1,
which
makes
it
real.
A
So
you
have
to
implemented
a
network
later
all
the
SE,
a
monetary
like
this
in
a
specific
way,
because
you
only
can
process
you
have
to
determine.
If
that
message
has
already
has
a
dependency
and
incident
legacy
has
already
been
processed
before
you
can
process
that
message,
so
it
makes
it
really
hard.
So
what
we
need
is
to
make
this
demand
right
if
you
need
to
have
a
simple
way
for
people
to
be
able
to
create
similarities
that
have
caused
dependency
in
a
very
simple
way.
Okay,
so
basically
we
want
generalizing
this.
A
So
it's
replicate
has
their
operation
mode
right
so
now,
if
the
each
replicate
gives
ahead
so
a
link
into
the
current,
the
the
heads
you
know
ever,
do
you
know
gifts
that
you
give
raise
your
hand
if
you
have
used
it?
Okay,
some
people
happen
so,
okay,
the
operation
one,
has
the
cash
rats
and
head
points
to
the
current
hash
of
the
latest
operation.
If
I
add
another
operation.
What
happens
is
that
I
move
heads
into
the
next
operation
and
so
forth?
If
I
have
an
operation
head,
so
it's
a
permanent
permanent
we
broadcast
there.
A
They
have
content
ID
variable
that
are
able
to
transmit
what
is
the
current
state
and
then
the
other
notes
can
retrieve
through
a
key
FS,
can
retrieve
all
the
missing
operations
right,
make
sense
yeah,
so
you
can
have
like
two
replicas
communicating
over
network
each
one
of
them
has
their
own
operation
log,
which
is
in
this
case
linear.
So
this
has
operations
for
one
or
two
country
and
each
one
of
the
nose
pointed
to
their
lettuce,
heads
okay,
so
far,
people
on
us,
okay.
So
what.
A
Each
one
of
the
replicas
creates
their
own
operation
modes
with
different
entries,
so
amending
documents
right,
I,
insert
a
character,
but
someone
else
r2
is
also
a
certain
character,
so
they're
creating
entries
here.
So
what
what?
How
can
we
resolve
this?
So
basically,
if
I
get
a
new
question
from
other
modes
and
and
I
have
so
backtracking
are
too
great
oppression.
Three
right,
but
I
get
operation
for
from
a
remote
node
which
links
to
the
same
old
operation.
Basically,
I
have
to
merge
those
two
operations
right,
so
I
create
a
new
entry.
B
A
So
this
is,
you
could
go
deeper
into
into
this
subject,
but
basically,
I've
I
did
a
proof
of
concept.
Implementation
in
java
script
that
uses
a
DFS,
which
is
pure
charity
and
Bill's.
Charity
ipfs,
which
you
can
install
with
using
NPM,
can
go
to
beat
up
a
BFS,
GPR
splash
pure
strategy
and
check
the
documentation.
It
supports
a
bunch
of
stuff
like
composing,
the
name
of
composition.
That's
some
building
types.
A
A
Also,
there
is
a
network
and
story
limitation
for
so
this
package
is
like
an
abstract
package
right,
so
there
will
not
be
a
specific
package
that
creates
the
store
and
network
layers
for
pr70.
That
would
be
others,
but
this
is
a
specific
for
DFS
and
you
can
also
follow
instructions
here.
If
you
want
to
create
a
personality,
using
a
DFS
should
be
straightforward.
I
find
this
is
experimental,
so
there
may
be
dragons,
but
especially
the
deputy
first
implementation
is
still
very
fresh
and
also
there's.
B
A
C
A
It's
really
been
so
pimp
at
what's
on
the
browser,
so,
whatever
network
you
have
available
on
on
your
laptop,
if
you're
ready
to
connect
to
your
to
your
college,
okay,
fine,
it
works
because
the
defense
that,
if
they
find
both
notes,
we
will
start
collaborating
with
each
other
okay.
So
if
they're
findable
through.
C
A
Chances
of
our
design
for
exactly
that
this
case
for
what
people
are
offline
and
when
they
are
reconnected,
they
this
they
state
sink
okay,
so
which
means
that
depth
and
without
any
conflicts,
but
so,
which
means
that
if
you
were
editing
different
parts
of
the
document
versus
if
it's
purified
for
parts
of
the
documents
should
be
fine.
But
if
you're
editing
the
same
to
same
sentence
sentence.
For
instance,
you
will
have
some
someone
will
well.
You
will
have
been
depending
on
the
type
that
they're
using
you
know.
A
A
D
D
A
Middle
in
certain
positions
and
in
some
resistors
in
the
position
he
wants
that
ready
to
grow
have
those
two
assertions
in
the
same
order
or
not
replicas.
So
the
guarantee
is
that
they
will
convert.
So
everyone
will
get
the
same
copy,
but
in
the
series
of
year
of
pear
pad.
Yes,
so
get
the
all
insertions
at
the
same
place
in
a
specific
order,
which
is
kind
of
fun.