►
A
A
C
A
F
F
Ids
talk
to
each
other,
I
feel
like
I've,
been
reinventing
some
reinventing
some
of
the
wheel
here
and
that
some
of
the
stuff
must
already
exist,
but
I'm
trying
to
locate
the
people,
who
can
only
find
that
working
right
now,
I
put
together
some
examples
of
four
candidates
for
a
graph,
synchronization
algorithm,
so
I
have
a
graph
of
deltas.
You
have
a
graph
of
deltas.
F
How
do
we
put
them
together
next
up,
I'm
gonna,
go
implement
them
and
and
see
how
they
work
and
I
think
try
and
put
together
in
the
next
couple
of
weeks
a
basic
chat,
client
that
takes
into
account
you
know,
don't
don't
show
a
message
until
you've
seen
the
previous
message
in
that
chain
or
other
ones
that
I've
known
about.
So
that
is,
that
is
the
direction
that
I'm
currently
going
in
I've
also
been
reaching
I.
A
F
C
F
A
E
Comparing
to
the
previous
meeting
so
I've
concluded
this
year,
key
data
model
for
the
comments
on
the
specified
so
essentially
I
have
on
the
Capt
Thank
You
team.
I
already
have
the
the
nested
comments
in
terms
of
the
data
model.
It
supports
heading
comments,
updating
comments,
removing
comments,
the
basic
stuff,
basically
so
I
also
co-hosted
a
workshop
with
Pedro
almost
most
ways.
We
just
came
back
this
for
this
weekend.
E
Basically,
I
think
it
went
good,
so
basically
we
presented
their
pure
startup
and
how
can
people
use
that
foundation
to
develop
dynamic,
apps
or
dubs
without
a
blockchain,
and
also
it
culminated
also
with
a
publish
the
the
publish
of
the
alpha
version
and
office
crucified,
which
essentially
allows
you
to
install
and
experiment
with
application?
Now
it's
not
finished
is
a
very
awful
version,
but
it
allows
you
to
try
it
out.
So
if
you're
interested
it,
we
will
be
in
it.
You
can
go
into
the
story.
E
E
So
right
now
in
progress,
I'm
working
on
the
replies
feature
in
terms
of
the
user
interface,
as
I
said,
it's
already
permitted
in
another
model,
but
not
yet
easily
I'm,
working
on
that
and
also
I'm
working
on
the
comments
history.
So,
for
instance,
if
you
edit
a
comment
or
delete
the
comments,
you'll
be
able
to
to
show
or
to
view
the
whole
history
of
the
comment
itself
and
my
next
steps
after
that
is
to
have
persistence.
E
Persistency
of
the
data
by
using
the
Pinner,
probably
it's
it's
table
and
I
can
deploy
it
on
on
some
place.
I
talked
with
Pedro
about
having
it
deployed
somewhere,
Pedro
or
maybe
someone
responsible
with
the
infrastructure
and
also
I
would
like
to
to
take
some
time
to
do
some
formatting
for
most
improvements,
but
it
like
it's
more
internal
stuff
between
the
the
background
script
of
the
extension
and
the
slices
themselves,
the
clients
of
Taxation.
B
A
B
B
Just
get
into
the
flow
of
everything
after
Scotland,
but
I
got
a
bunch
of
small
little
things.
Fixed
and
I
did
a
the
highlight,
I
think
was
I
did
a
live
editing
cus
with
the
slack
users
in
protocol
labs,
I
got
them
all
to
log
in
and
do
a
live
editing,
and
there
was
still
some
corruption
like
if
people
type
something
and
then
another
peer
with
like
modify
what
they
typed,
it
was
I.
B
Most
of
my
last
week
was
just
playing
around
with
pinning
and
with
Dirk
stuff
the
persister
and
I
integrated
in
writing
in
just
just
the
pen
or
not
in
the
web
client,
but
in
the
node
G
spinner
and
I
got
it
successfully.
Writing
in
and
just
yesterday,
I
got
it
actually
reading
back
out
the
there's
a
couple.
Little
little
fixes
there,
but
as
long
as
I'm
running
the
node
on
a
disk
like
on
my
local
machine
that
has
access
to
the
disk
I'm.
B
Just
writing
out
the
the
head
references
to
my
heart,
hard
directory
on
my
hard
disk
and
then
when
I
I
can
shut
down
the
Pinner
start
it
back
up
and
all
the
data
gets
loaded
in
and
still
there.
So
that's
really
really
big,
because
before
I
was
running
pinners,
but
they
were
only
persisting
in
memory.
So
if
all
the
Pinner
shut
down
you
lose
the
data
and
so
I
think
that's
a
pretty
big
milestone.
B
E
B
B
The
I
got
it
so
it
wasn't
happening
when
I
was
just
myself
typing
on
my
own
little
instances,
but
when
we
had
a
whole
bunch
of
people
typing
at
once,
multiple
keyboards
happening
at
once,
then
it
did
pop
up.
So
that
was
a
hard
thing
to
test
without
automated
automation.
So
Pender's
been
testing
it
with
his
test
suite.
So
hopefully
his
version
works
better
than
mine
and
I'll
just
switch
so.
B
So
imagine
there's
two
writers
and
I'm
typing
in
one
and
the
other
writer.
Even
nobody,
even
though
nobody's
at
the
keyboard,
it's
modifying
the
document.
So
it's
more
like
timing
issues
and
like
it's
getting
the
deltas
in
the
wrong
order
and
then
how
the
binding
editor
bindings
work
as
it's
doing
discs
between
the
CR
DT
and,
what's
in
the
editor
that
particular
time.
So
it's
very
time
so.
B
A
F
B
This
is
strictly
a
code
mirror
and
how
we
bound
it
into
code
mirror
and
using
the
certainly
creative
things
algorithm
and
it's
when
it's
written.
It
needs
a
little
bit
of
locking,
because
it's
like
when
you,
when
you
reading
the
CR
DT
state
it
it
essentially
types
in
to
code,
mirror
and
then
that
creates
another
more
Delta's,
which
should
be
the
same,
but
they
never
the
deltas
arrive.
At
the
same
time.
It's.
A
Clarify
I
think
I
think
it's
exactly
that.
Yeah
I
think
it's
that
the
syncing
between
the
nasarah
T
and
the
code
mirror.
That
was
a
default,
at
least
from
what
my
tests
and
I
think
that
they
should
mostly
be
resolved.
I'm
very
keen
to
see
if
Jim
can
can
put
the
pin
er
into
that.
If
you
do
introduce
any
entropy
into
the
system
or
not,
but
I'm
expecting
not.
But
let's
see
yeah.
B
And
another
another
point
on
that:
one
is
like
we're
sort
of
doing
it:
sort
of
crudely-
it's
just
like
it's
adding
edits
in,
but
like
you're,
getting
edits
that
somebody
else
typed
put
into
your
code
mayor.
So
if
you
hit
undo
that
undoes,
the
other
person's
edits,
which
is
surprising
so
people
views
code,
mayor
for
lots
of
collaborative
editor,
so
I'm
sure
there's
support
in
there
and
we
probably
have
to
use
the
API
differently
on
coding
here.
B
F
A
H
Okay,
I
got
it
I
feel
like
there's
an
issue
off,
because
the
deltas
are
virginal,
you're
kind
of
undoing
a
merged
set
of
operations.
That
includes
most
the
intended
actions,
and
then
they
actually
distract
her
to
be
at
the
same
time,
I
mean
yeah.
You
know,
yeah
I,
think
I
think
we
should
have
a
better
way
of
thinking
about,
like
probably
refer
to
sets
of
things
sets
of
Delta's,
because
they
think
that
also
applies
to
like
snapshotting,
because
is
it
like?
Do
you
have
like
what
does
it
mean
if
you're
on
the
same
version?
A
Right,
that's
right.
We
could
you
could
look
into
and
do
in.
You
know
more
in
the
in
the
in
the
Delta
context.
Yeah
that
makes
sense
the
thing
yeah
we're
losing
a
bit
of
information.
We
may
lose
a
bit
of
information
as
we're,
reducing
the
the
deltas
and
so
I.
Think
in
terms
of
the
history
of
the
document.
I
think
you
have
to
revisit
sorry.
I
might.
E
Have
one
more
question
for
Jim,
and
perhaps
it's
more
a
generic
question
also
perhaps
involves
you
and
which
is
the
Pinner
at
the
moment.
Correct
me
if
I'm
wrong
doesn't
support
any
kind
of
sharding.
That's
by
charting
I
mean
I'd
say
that
this
case,
if
I
has
a
lot
of
collaborations
or
discussions,
a
single
Pinner
is
responsible
for
the
whole
set
of
discussions
happening
at
the
same
time,
right
so
interesting
to
have
like
configuration
where
it
could
specify
a
sharding
function
so
that
we
can
split
the
the
pinners
among
assets.
B
I
think
that
would
be
great.
It
was
already
a
task
to
split
off
the
Pinner
into
a
separate
repo
and
I
can
see
like
look.
I
can
immediately
see
that
people
are
going
to
have
all
sorts
of
different
types
of
pinners
with
different
strategies,
and
so
there's
just
the
simple
single
spinner
and
but
then
there's
like
the
situation
where
you
don't
even
need
a
pinare,
but
there
might
be
a
situation
where
you
want
to
have
a
cluster
of
pinners,
so
I
think
they're
all
directions
worth
exploring.
B
Did
a
different
just
a
stupidest
naming
strategy:
I
could
do
I,
just
write
out
the
collaboration,
ID
and
then
I
just
write
out.
Let's
see,
I
lit
the
last
CID
into
a
file
on
in
the
directory
and
that
my
we
start
at
the
Pinner
when
it
sees
new
collaboration.
It
just
reads
that
and
closed
in
so
I
wouldn't
mind
doing
the
code
review
with
you.
If
you
have
a
chance
whether
it
seems
to
work
for
my
really
trivial
case,
I
just
want
something
to
work
with
my
little
demo,
so
that
was.
B
C
I
Yeah
I'm
here,
hello,
everyone,
I'm,
Victor,
I,
used
to
be
working
mainly
on
infrastructure
stuff,
but
now
we
have
people
who
actually
want
to
work
with
infrastructure,
so
I'm
gonna
move
to
other
things.
One
of
the
things
that
I
am
very
interested
in
in
having
is
tearpad
as
default
application
or
not
taking
for
protocol
apps
people
in
the
beginning,
but
I'm
in
the
future,
everyone
in
the
whole
world.
So
what
I'm?
What
I'm
gonna
try
to
focus
on,
or
a
few
things
like
the
essential
thing
that
we
need
to
make
pearpad
great?
I
Basically,
so
some
of
the
things
that
I
will
be
helping
with
is
deploying
the
Pinner
I
have
a
lot
of
infrastructure,
stuff
experience,
so
I
can
help
deployed
the
the
Pinner
and
we
can
figure
out
the
the
architecture
for
for
all
of
that
and
I.
Also.
The
first
thing
that
caught
my
eye
is
that
we
don't
have
a
lot
of
metrics
and
health
checks
and
and
other
various
numbers
and
and
red
and
green
bottoms
to
understand
if
your
pad
is
actually
working
as
it
should
be
working.
I
So
this
helped
in
the
deployed
version
that
we
have
that
we
can
see
more
errors.
We
can
catch
things,
they
were
broken
earlier
and
stuff
like
that,
and
then,
of
course,
I
can
help
a
lot
with
the
testing,
like
the
testing
cluster,
that
we
just
been
talking
about.
Pedro
is
doing
and
try
to
fold
in
the
folding
like
metrics
in
there
as
well,
and
we
can
start
doing
some
benchmarking.
Another
interesting
thing.
E
I
So
one
thing
I
guess
we
have
to
make
sure
is
the
difference
between
the
pearpad,
specific
things
and
the
the
peers
our
stuff.
So
the
painter
should
probably
work
with
both,
but
just
in
things
in
some
way
not
being
specific
to
Pierre
part,
and
then
we
can
figure
out
the
deployment.
So
we
can
stay
scale
it
down
depending
on
the
application
which
one
is
seeing,
the
more
you
search
and
so
on.
But
yes,
please
bother
me
and
I
will
try
to
help
you.
F
F
A
So
I
can
say
that
Victor
I've
ever
excited
to
have
you
here.
Rico
has
been
a
longtime
contributor
to
coach
internet
protocol,
apps
product
protocol
I,
deploy
and
for
a
long
time
longer
than
much
longer
than
I've
been
so
I'm
very
excited
to
have
you
in
here
helping
us
to
drive
this
thing
further
and
just
wanted
to
say.
Welcome.
J
We
had
some
some
some
discussion
between
some
user
experience
and
UI
that
were,
will
rework
and
right
now,
I'm,
looking
at
I
did
and
already
provided
a
new
demo
without
all
of
those
changes.
It
will
be
a
good
exercise
to
try
to
share
with
all
of
you,
the
demo
and
just
try
to
gather
feedback,
maybe
in
visually
individually,
since,
if
I'm
already
showing
at
the
demo
I
will
be
influencing
each
one.
It's
one
of
you
with
my
with
my
word,
so
I
will
try
to
do
if
this
exercise,
at
least
for
for
this
time.
J
J
I
mean
I
may
send
an
email
with
some
of
the
features
that
were
predicted
and
some
of
the
changes
that
were
created
from
the
implemented
version
for
the
new
version
that
I'm
suggesting
and
if
I
will
create
something
right
right
straight
away
with
a
high
level.
So
it
will
be
easier
to
understand,
what's
changed
and
try
to
come
up
with
questions
for
me.
So
I
can
understand.
This
is
like
a
user
testing
experiment
but
yeah
just
just
for
us
for
now.
Perhaps.
G
K
But
cozy
very
I'm
here
as
a
meeting
so
I'm
mark
Oliveira,
I'm
I'm,
helping
out
mainly
with
discuss
a
fight
and
I'm,
also
helping
with
the
design
working
group.
So
I'll
start
by
saying
that,
on
the
discuss
of
I,
think
whoops
I
finished
reviewing
the
UI
and
UX
with
Andre
Sosa
suggested
a
few
changes
to
make
the
UI
cleaner,
and
we
reiterated
on
the
threaded
conversation
which
ended
up
creating
a
few.
A
few
changes
right
now
in
progress,
as
I
said,
I'm
also
mediating
with
the
design
working
group
which
is
just
coming
up.
K
It
has
no
final
shape
I'm
trying
to
do
my
best
to
represent
the
needs
from
the
dynamic
data
and
capabilities
working
group
there
right
now
we're
discussing
mainly
process
goals
and
collaboration
with
our
teams.
I'm
also
trying
to
get
the
next
of
UGS
going,
which
will
be
a
good
chance
for
us
to
to
redo
the
building
gaps
on
the
permanent
web
workshop.
This
time
a
little
shorter
and
create
a
good
recording
of
that,
so
that
we
can
share
that
with
with
the
community
more
in
a
more
effective
way.
K
K
I
need
to
help
drop
in
the
the
roadmap
as
well,
and
I
need
also
to
direct
you
to
help
me
out
here
and
see
what
what's,
what
sort
of
help
he
needs
from
me
here.
I
will
be
providing
feedback
on
keypad
220,
a
Andre
Sosa
I
already,
provided
some
that
I
I
want
to
see
that
video
we
mentioned.
I
haven't
seen
it
yet
last
IDM
concepts
and
overall
look
and
feel
I
have
and
Ray
I
have
a
feeling
that
time
will
be
short
to
be
well
to
do
this
in
the
next
couple
of
weeks.
F
As
I
said,
just
a
discuss
a
PHY
question
because
I'm
still
just
new
and
learning
about
it,
the
I
got
invited
recently
to
like
the
file
claim
like
discuss
it,
like
you
know,
discuss
group
which
I
imagine
is
supposed
to
be
doing.
A
similar
sort
of
thing
is
the
idea
we
could
like
potentially
use
them
as
guinea
pigs,
and
you
know,
I,
don't
flooded
that
out
there
as
an
option
of
trying
to
you
know
get
you
know
some
of
the
dog
fooding
and
using
our
own
stuff.
K
K
G
E
K
But
by
the
way,
I
want
to
make
notes
every
time
identity
comes
up
in
a
discussion,
a
lot
of
people
show
interest
and
and
well
coming
from
very
different
groups,
so
one
we
might
need
to
eventually
bring
those
people
in
because
each
each
person
is
trying
to
defend
a
certain
agenda
and
a
lot
of
that
is
very
reasonable.
At
the
same
time,
we
have
to
balance
that
with
how
effective
we
are,
we
don't
we
don't
want
to
be
frozen
just
because
we
are
getting
way
too
much
feedback
or
briefing.
K
B
A
No
okay,
I
can
share
my
update
quickly,
so
we
just
released
your
pure
star
at
pushing
0.9.0
and
0.1
this
weekend
for
for
most
fast
fixing
some
issues
and
also
adding
memorization
for
the
shared
value
so
repeated
shared
value
access
is
just
Nicastro
and
recomputing.
The
states
all
the
time
but
into
any
load
test
is
slow
tests.
A
Preserving
intention,
as
we
discussed
so
that's
something
that
I
want
to
do
to
work
on,
to
make
sure
that
that
that
users
are
are
okay,
acidity,
that
that
converges
to
everyone
covers
the
same
value,
but
the
volume
it
doesn't
have
anything
to
do
with
with
what
the
input
was.
Is
that
a
very
useful
see
oddity
so
I
want
the
it's?
Not
not
testing
the
charity
itself
and
I
see
more
that
the
sync
between
the
binding
and
sorry,
the
seam
between
the
editor
and
charity,
interest,
payment
and
and
so
yeah?
That's
my
update.
K
A
A
J
A
A
D
B
B
And
then
and
anybody
wants
to
join,
you
can
join
and
then
oh,
but
then
I'd
have
to
pass
you
the
URL
better,
not
join,
actually
cuz,
then
the
Pinner
one.
So
it's
going
to
sink
should
sink
up
here.
Hopefully
yeah
so
I
got
it.
So,
of
course
the
problem
is
if
I,
if
I
reload,
both
of
these
at
the
same
time,
I'm
going
to
lose
the
data,
so
I
lost
the
data
so
test-tube.
B
B
B
B
You
can
see
that
and
then
it's
got
that's
a
delta,
so
Delta's
always
have
a
parent
and
then
the
parent
was
a
snapshot,
so
that
has
all
the
data
in
yet
originally
I.
When
I
was
doing
this,
the
it
wasn't,
saving
the
snapshots
properly.
So
that
might
be
something
I
have
to
talk
to
Dirk
about,
but
anyways.
B
Also
when
I'm
typing
here,
I'm
also
I've
modified
this
I'm,
not
using
IP
NS,
although
I
could
I'm
just
writing
out
little
files
that
look
like
I'm,
just
writing
the
last
CID
for
its
natural
or
Delta
into
it,
and
so
so
that
means
that
when
I
shut
this
down
and
I
restart
it.
Okay
I'll
do
a
real
test
here.
So
if
I
reload
both
of
these
at
the
same
time,
I've
lost
the
data.
B
B
E
So
it's
cool.
It
could
be
like
pure
pure
pods,
pure
pad
Pinner,
like
that
entity
name
per
se,
but
it
won't
have
the
same
key
ID.
So
the
application
could
identify
the
pin
earth-2
to
see
if,
if
it
pushed
so,
if
it's
synced
with
that
peer,
which
is
the
peer
node,
the
pinning
mode
in
order
to
present
saved
to
the
user,
that's
something
that
I've
been
thinking
about
and
I
wanted
to
say
that,
so
that
everyone
here
can
also
think
about
that
as
a
possible
solution
to
this
problem.
Mm-Hmm.
E
Yeah,
it's
a
it's
a
there's.
Two
scenarios
actually
there's
a
scenario
where
the
application
itself,
as
a
pinging,
a
pinning
over
one
or
more
pinning
nodes,
and
there
are
community
pinning
nodes
that
people
can.
You
know,
contribute
to
the
network
in
order
to
provide
availability.
So
we
need
to
think
about
that.
Just
wanted
to.
You
know
to
mention
that
and
possibly
you
will
be
thinking
about
that
solution.
We.
F
B
Aren't
you
for
my
for
my
thing:
it's
mostly
the
data.
So
it's
just
you
know
if,
if
everybody
shuts
down
their
web
browser
and
then
they
joined
the
collaboration
again
later
as
the
data
still
there
but
Jim
easy,
the
data
or
the
deltas,
how
dark
stuff
works
is
it
saves
Delta,
Delta,
Delta
and
then
I'll
do
a
snapshot.
It's
configured,
bollock
I
set
it
to
10.
B
C
A
C
So
that
makes
it
pretty
difficult
to
do
snapshots
you
sort
of
have
to
like
ask
another
node
for
a
snapshot,
so
one
of
the
advantages
of
having
leader
election
and
having
a
single
node
who's
already
part
of
the
collaboration,
be
fact
as
the
Penner
is
that
you
don't
have
to
worry
about
those
issues.
Yeah.
B
Had
to
modify
my
version,
so
it
just
has
a
like
a
and
there's
an
I
just
changed
the
encryption
function,
so
they
just
an
identity
like
there's.
No
encryption
happening
so
right,
that's!
Okay!
For
this
particular
it's
a
fairly
permissive!
You
know
it's
it's
okay
for
collaborations,
where
you're,
basically
keeping
the
key
of
the
class,
the
the
name
of
the
collaboration
secret.
Of
course,
that
gets
all
shared
on
the
pub/sub
things.
B
B
Issue
I
think
it's
interested
in
like
how
fast
is
it
with
the
encryption
turned
off
because
we
want
to
just
see.
What's
the
performance
impact
is
with
it,
so
we
can
benchmark
it,
so
he
can
go
yeah.
This
is
how
fast
it
is
was
it
turned
off.
Then
we
turn
it
on.
What's
overhead,
like
that's
gonna
impact
scalability,
that's.
A
L
F
A
Good
yeah,
of
course,
which
we
I
was
thinking
that
we
in
terms
of
this
is
more
more
pure
plans,
but
I
think
it
applies
also
to
others
which
the
persistence
of
today
we
don't
need
to
persist
on
every
keystroke
right.
So
we
could
have
the
deltas
saved
without
without
being
able
to
actually
extract
the
timing
of
the
title
from
the
user,
and
so
we,
you
could
have
less
so
again,
I
like
things
like
really
good
things,
or
something
like
that,
instead
of
just
just
having
every
stroke.
H
Being
said
so
so
from
a
UX
perspective,
I
think
it's
important
to
keep
in
mind
that
we'll
probably
want
like
adaptive
like
Delta
formation
or
like
snapshotting,
in
terms
of
like
get
like
how
fast
you're
typing
playing
basically
like
rest
periods
of
what
you
stop
typing
for
a
little
bit.
That
should
generate
a
snapshot
even
if
it
doesn't
kind
of
like
fit
the
right
number
or,
if
you're
typing
really
fast,
that
you
can
probably
try
to
consolidate
that
into
like
fewer
Delta's
right.
B
I
think
it's
gonna
depend
like
applications
are
quite
different,
so,
like
not
all
applications,
these
appear
star,
APRA
gonna,
be
collaborative
editing
like
you
might
want
to
collapse.
A
bunch
of
deltas
together,
even
though
they're
different
authors
and
everything
just
see
for
like
an
undo
history
type
of
thing.
Early
good,
like
I,
want
to
see
what
what
the
document
state
was
two
days
ago
or
something
great,
but
you
know
sometimes
you
just
wanted
to
be
able
to
undo
a
few
characters.
So
the
the
best
thing
is
that
it's
all
policy.
Basically,
so
that's.
B
A
G
A
A
A
H
Right,
do
we
want
to
do
some
prep
for
that
asynchronously,
because
I
feel,
like
there's
a
bunch
of
stuff
that
we
can
put
down
like
the
length
of
the
polygon
thing.
I
think
we
can
do
offline
pretty
much
and
then
a
three
things
we
can
like
put
a
bunch
of
suggestions
down
and
then
we
can
discuss
it's
real
time.
I
think.
A
That's
very
very
useful.
We
should
we
should
first
do
that
like
have
our
input
and
then
we'll
have
a
meeting
on
on
on
thinking
that
everyone
I
agree
so
yeah.
If
everyone
could
good
start
and
on
on
on
that
with
the
people
that
were
at
the
meeting,
that's
we
can
then
like
tomorrow
or
the
day
after
we
could
join
and
join.
We
will
put
it
did
a
lot
with
with
the
an
try
and
schedule
something
for
tomorrow
or
day
after
good
all
right.
Well,
thank
everyone
for
coming,
see
you
soon.
Okay,
thank
you.