►
From YouTube: Dynamic Data and Capabilities WG — October 2nd, 2018
Description
Meeting notes:
https://github.com/ipfs/dynamic-data-and-capabilities/issues/40
A
All
right,
so,
let's
do
a
round
of
intros
and
updates
arkady
since
you're
your
did.
You
I
think
you're
new
to
the
meeting.
If
you
want
to
know
more
about
someone
someone's
work,
we
all
have
questions
at
the
end
of
each
round,
but
if
you
want
to
know
more
about
the
role
of
someone,
just
just,
please
ask
otherwise
we'll
just
stick
to
the
updates,
so
in
order
of
attendees
here
on
top
going
to
start
so
this
last
two
weeks,
I've
been
focused
on
working
group:
ok,
ours
for
the
quarter.
A
A
Our
app
framework
mainly
discovery
collaboration,
membership
and
connectivity,
and
so,
if
you
will
read
the
RFC
and
chime
in
ask
ask
things
or
propose
a
different
path
is
a
very
simple
document,
very
high
level,
and
it
talks
about
things
like
well
how
to
the
discovery
and
how
to
do
the
connectivity
for
as
a
replacement
for
some
things
we
we
have
now,
for
instance,
what
focused
our?
How
do
we
scale?
A
Sometimes
we
have
difficulty
finding
peer
addresses
if
the
pure
book
is
not
doesn't
have
yet
us
for
appear,
we're
not
able
to
connect
to
it,
and
thank
you
Jim
for
finding
that
out.
So
the
fix
for
that
is
again,
a
temporary
fix
is
to
embed
the
network
address
of
the
peer
and
membership
gossip
itself,
and
for
that
we
needs
going
down
further
down.
A
The
rabbit
hole
is
to
create
that
or
map
which
is
an
embeddable
gal
facility
or
a
map
allows
you
to
have
like
a
map
construct
like
a
JavaScript
map,
but
it
allows
you
to
embed
strategies
inside
that
map.
So
we
can
have
a
map
where
each
values
it's
assigned
to
yourself
and
then
you
can
go
and
do
that.
Recursively.
A
While
keeping
the
causality
keeping
the
the
versality
in
a
causal,
consistent,
causally,
consistent,
safe
state-
and
so
this
this
also
addresses
our
there
is
an
issue
on
on
peers,
our
app
I,
think
and
also
GS
delta
strategies
for
how
to
achieve
an
embeddable,
sanity's
and
I
think
SATA.
Sorry,
men
refer
to
specify.
Yes,
we
have
we've
addressed
this
issue.
A
There
are
other
solutions
for
this,
but
this
one
is
like
a
solution
that
comes
directly
out
of
the
Delta
severity
paper
and
big
shout
out
to
the
third
image,
which
is
a
PhD
student
for
in
the
charity
area
and
he's
been.
Helping
me
out,
is
a
is
a
proficient
proficient
in
Delta
charities
and
he's
been
helping
me
out
a
lot
in
in
making
this
right.
A
C
It's
me
so
for
the
concluded
part
I
finalized
restructuring,
the
extension
to
use
slices
and
message
passing
so
as
I
as
I
said
in
the
previous
meeting.
I
had
some
issues
regarding
CSP
a
horse
and
CSS
here
horse,
so
I've
had
to
basically
iframes
and
shadow
DOM
together,
get
around
this
issues
and
I've
work
on
that
solution.
Solution
and
I
actually
finalized
the
refactor
of
the
extension
to
use
that
strategy.
C
The
second
part
is
the
sidebar
logic,
I've
implemented
the
sidebar
finalized
animation
and
move
the
state
off
the
sidebar.
To
the
background
script
of
the
extension
he's
working
nicely,
I
will
demo
the
state
of
the
application
in
the
demos
part,
so
you
guys
can
see
and
also
I've
implemented.
The
add
new
comments
or
the
create
comment
feature
so
essentially
is
already
implemented.
I
will
demo
that
as
well
so
in
progress,
I'm
working
on
removing
comments,
a
plating
comments
and
reply
to
comments,
that's
essentially
it
I.
Hopefully
the
remove
feature
will
be
ready
today
and
tomorrow.
C
I
will
work
on
the
plate
and
and
reply
next
I
I
need
to
refine
the
application
like
see
what
what
can
be
improved
in
terms
of
codes,
called
structure
and
readability,
and
finally,
yeah
typo
here
I
need
to
discuss
with
Pedro
a
few
things
related
to
the
ipfs,
the
way
I
use
ipfs
and
the
data
model.
One
of
the
questions
that
I've
been
thinking
about
is
essentially
the
extension,
injects
a
context.
Content
script
in
each
tab
and
each
tab
as
it
is
now
will
have
its
own
instance
of
a
BFS
running
per
tab.
C
C
So
I
need
to
discuss
with
Bello.
How
can
we
improve
this
I
have
one
suggestion
which
is
to
move
the
ipfs
to
the
background
script,
which
is
just
one
and
all
the
runs,
one
super
chrome
instance
and
but
then
I
need,
if
I
do
that,
I
need
to
use
the
message.
Passing
structure
that
I've
implemented
to
you
know
make
ipfs
available
into
the
content
scripts
or
at
least
data
flowing
around
via
the
via
the
the
content.
C
Scripts
I
know
that
if
you
have
the
companion
accession
and
enable
the
local
IP
Fiesta
min
on
your
machine,
this
is
not
an
issue
but
I
I
think
most
of
the
users
won't
have
that
installed
in
your
computer
and
set
up
that
way.
So
this
is
something
I
need
to
discuss
with
Pedro
and
then
I
have
some
other
concerns
about
the
data
model,
but
too
specific
to
discuss
right
here,
I
think
so.
I
will
I
will
chat
with
you,
Peter,
probably
by
the
end
of
this
week.
If
that's
fine
with
you
and
that's
it.
D
A
I
can,
of
course,
haven't
won
one
question
about
so
this,
of
course,
is
related
to
discuss
if
I
and
the
browser
extension
some
people
that
know
more
about
the
context
that
haven't
seen
other
episodes
of
this
these
meetings,
but
I
think
you
should
should
definitely
talk
to
a
professor
in
web
browsers
working
group
because
they
they
well.
One
question
is:
if
you
you
host
this
in
one
domain,
I'm
mystified
about
or
ignite
say,
is
this
still
a
problem?
Will
there
be
one
of
your
assistants
for
because.
C
This
is
an
extension
so
which
is
an
extension.
That's
that's
different,
but
it
yes,
if
you,
if
there's
a
website,
it
will
be
just
one
because
of
the
Service
Worker
feature
or
package
that
we
now
have
so
in
the
the
website.
It's
not
a
problem,
but
in
the
extension
it
is
because
the
domain
effectively
is
different
in
each
tab
or
mostly
different,
because
you
never
get
in
in
other
domains
and
keep
the
tabs
open,
Iran
and
so
on.
C
The
way
the
way
the
way
is
the
way
I've
been
thinking
is
if
I
move
the
ipfs
instance
and
the
collaborations
part,
and
so
on,
to
the
background
script
and
then
use
a
simple
message
message
passing
it
will
work,
but
there's
an
overhead,
because
if
I
listen
to
the
state
change
event
of
the
collaboration
and
simply
simply
sends
the
whole
States
down
the
wire,
what
not
to
wire
but
via
post
message,
it
will
be
based
on
a
copy
memory
copy.
So
it
will.
C
It
will
copy,
essentially
the
whole
data
structure
and
send
it
to
the
to
the
content
script,
and
this
is
an
issue
I
think
it's
not
very
performance.
I
will
need
to
have
some
sort
of
differential,
DIF
calculator
on
the
top
or
on
the
on
the
macro
script,
and
all
these
send
updates
to
the
to
the
content
script.
C
D
Really
sounds
like
we're
duplicating
work
that
light
all
and
GSA
PFS
folks
you're,
going
to
be
doing
with
their
message,
passing
API
and
I
think
that's
they're
trying
to
follow
the
same
standards
I'm
like
as
a
web
3
things
for
that.
So
it
sounds
like
this
is
maybe
not
not
the
best
use
of
time
right
now,
because
it's
to
two
tracks
once.
A
D
Jumping
and
telling
people
they're
doing
good
work
I,
so
I'm
some
of
you
may
have
worked.
There's
me
a
little
bit
over
the
summer
summits
and
associated
things,
but
now
I'm
really
really
really
full-time
and
I'm
doing
a
couple
of
things.
I'm
going
to
be
managing
the
various
collaborations
that
protocol
is
engaged
in
like
the
internet.
Are
everyone
was
kind
of
a
big
example,
so
I'll
just
be
yeah,
but
in
addition
to
that
I'm
going
to
be
kind
of
the
interim
p.m.
on
peer
path.
This
is
just
helping
it
become.
D
A
Thank
you.
Thank
you
very
exciting
about
about
that.
The
thing
about
that
that
I
was
was
talking
about.
That's
all
it's
totally
true,
and
one
thing
to
have
in
mind
is,
is
whether
we,
if
we,
if
you
do,
have
to
use
message-passing
for
multiplexing,
so
right
now
we're
using
the
p2p
to
to
do
have
a
custom
protocol
on
top
of
the
p2p.
So
if
you
then
may
make
like,
like
a
proxy
up
your
first
node
and
message
passing
for
that,
you
have
to
also
implement
multiplexing
so
streaming
on.
A
C
I'm
more
inclined
into
doing
the
app
specific
stuff
right
now
and
not
wait
for
the
other
solution
in
terms
of
the
the
other
working
group
which
is
I
profess
because
of
one
one
specific
reason,
which
is
remember
that
when
the
user
navigates
in
in
a
website,
let's
say
Google
or
Facebook,
there's
a
lot
of
refreshes,
so
the
page
refreshes
a
lot.
That
means
that
the
content
scripts
also
gets
reloaded
and
the
connections
the
peer
connections
will
go,
go
basically
it's
connect
and
have
to
connect
all
over
it
again.
C
This
means
that
there's
a
huge
boot
time
in
terms
of
the
discussion
being
loaded
and
plated
and
synchronize
so
I
think
doing
the
app
specific
stuff
is
really
straightforward.
I
will
say
it's
like
30
percent
overhead
in
terms
of
using
the
peer
startup
directly
in
the
content
script.
So
I'll
probably
go
with
that
solution.
As
the
initial
approach.
A
C
C
G
G
So
I
could
see
the
in
the
UI,
because
it's
just
a
lot
going
on
underneath
the
covers,
and
that
was
a
really
valuable
exercise
because
it
just
turned
up
I,
don't
know
like
five
or
six
really
major
bugs
and
I'm
surprised
that
it
ain't
even
worked
at
all,
so
I
haven't
yet
done
the
pull
request
to
get
them
into
the
central
pad
at
the
pier
starter.
App
I
will
try
to
do
that
today.
I
just
keep
quitting
more
bugs
just,
and
so
there
was
like
a
major
bug
in
like
the
removal
just
JavaScript.
G
It
was
like
when
you
do
force
thing
in
it.
Actually,
the
index
is
actually
a
string
that
comes
back
and
then
we're
adding
it
to
the
number.
So
it
was
actually
nuking
all
the
all
the
piers
off
whenever
it
would
do
in
an
addiction.
So
that
was
that
was
creating
a
lot
of
instability.
There
is
also
I
was
looking
into
how
the
DSS
work
and
they
weren't
really
doing
what
we
wanted
when
there
were
a
small
number
of
peers-
and
it
was
a
not
a
very,
even
distribution,
so
there
was
easy
fix
for
that.
G
So
I
went
that
yesterday
and
that
greatly
removed
improved
the
stability
of
the
of
the
connections
and
I've
been
playing
around
with
the
painters
they,
my
pinners
weren't,
holding
the
connection
for
a
long
period
of
time.
So
last
night
I
set
up
this
loop
where
they
sort
of
kill
themselves
off
after
10
minutes
and
I
said
it
three
to
be
finished
and
I
seem
to
be
working.
G
A
G
Anyway,
so
if
there's
time,
I
can
give
a
quick
little
demo
of
the
little
appears
display,
works
and
I
think
I
might
be
interesting,
but
if
there's
no
time
I
can
do
that
any
time
it
Candide
in
Scotland,
if
everybody's
there,
so
that
this
this
week
or
next,
two
weeks,
obviously
Scotland
I,
didn't
know,
I
was
going
until
last
Thursday,
so
theoretically,
I
have
clean
tickets.
I'm,
not
sure
the
good
work
on
those
zombie
peers
get
the
pull
requests
in
there's
a
bit
of
corruption
happening.
It's
funny.
G
I'm
watching
crypt
pad
pat
has
the
same
sort
of
problems,
but
I
think
it's
just
a
problem
in
my
front
end
code
because
it
shouldn't
they
shouldn't
even
be
possible
for
this
corruption
to
be
getting
in
service.
It
just
tells
me
there's
a
bug.
I
made
I'm
gonna
get
the
pin.
Er
said
after
we
start
through
every
ten
minutes.
G
I
also
want
to
take
I,
also
wanna
play
around
with
actually
just
doing
really
really
crude,
like
every
ten
minutes
or
so
dump
the
the
state
into
the
DHT,
so
that
we
can
rehydrate
it
and
also
do
full
rendering.
So
you
can
turn
javascript
off
and
we
just
rerender
the
index.html
like
or
we'd
have
to
do.
It
would
be
specific
to
each
app,
but
for
like
my
little
demo,
app
that
would
be
trivial
to
do
so.
Just
be
like.
G
Can
you
pull
it
in
JavaScript
turned
off
like
the
first
pull
from
the
server
you'll
get
the
text
that
you
were
editing
and
then
maybe
hit
a
button
to
load
in
all
the
JavaScript
I
automated
tests,
cuz
I
was
supposed
to
do
automated
tests
and
I
just
they're
headed
to
many
like
really
major
things.
The
automate
tests
are
depend
what
working
into
place.
So
we
don't
regress
and
Scotland
not
really
block
on
anything.
G
A
H
So
I
finished
the
user
journey
for
disgust
if
I
log
in
right.
Now
it's
make
with
a
different
decision.
Such
you
can
see
the
comments
and
then
you
need
to
be
inside
the
to
log.
In
imagine
that
you're
seeing
any
comment
you
want
to
stay
and
leave
your
comment
and
then
you'll
be
triggered
with
a
modal
and
then,
if
it
asks
for
login
instead
of
every
time
that
you
need
to
open
the
app
here,
we
need
were
required
to
log
in
that's
a
use
case
that
I've
changed
another
thing
in
following
up
Andre.
H
H
Okay,
so
I
finished
the
ID,
my
frames
for
the
POC
in
progress
right
now,
I'm
studying
and
trying
to
start
sketching
different
approaches
for
the
IDM
visual
language.
The
brand
is
mostly
define
it,
but
then
we
need
to
standardize
everything
like
a
brand
Manuel
define
the
typography
and
all
that
starts
stuff
associated
with
a
brand
itself
and
then
try
and
run
a
few
tests
for
the
app
concept
visually
speaking
and
then
something
that
I'm
working
as
well
right.
Now,
it's
the
component
system.
H
H
So
it's
like
the
react
system
would
apply
to
the
design
workflow,
so
I'm
not
blogging
and
on
anything
right
now
and
next
will
be
yeah
define
with
Andre
the
shadows
are
the
required
flows
for
the
IDM
PLC
to
make
sure
that
we
can
have
that
specific
scenarios
promised
to
be
we
implemented
and
then
deploy
and
for
now
yeah?
That's
it.
I
A
C
Can
I
get
this
sort
of
screen?
I,
don't
I,
don't
think
so.
Anyway,
let
me
move
this
around
I'll
say:
okay,
so
I
will
turn
on
the
extension
I'm
going
to
open
it.
As
you
see,
I
already
have
comments
set
up
so
right
now
we
can
comment
like
say
something
you
can
comment.
You
can
also
do
multiple
lines
as
well,
and
this
is
the
create
feature
and
right
now,
I
was
working
on
the
delete
comments.
I
had
it
set
up,
but
there's
some
CSS
issues.
C
H
C
Hopefully,
by
the
end
of
the
week,
I
will
have
all
the
the
features
the
essential
features
implemented.
I
probably
will
take
one
day
more
to
implement
the
nesting
of
the
replies,
because
there's
a
lot
of
specific
rules
like
there's
a
maximum
indentation
of
replies,
and
then
it
shows
replying
to
something
right
and
another.
H
C
I
can't
show
it
in
a
huge
monitor
because
I'm
my
laptop,
but
you
know
like
a
4k
display
the
sidebar
is,
is
a
little
bit
bigger
and
the
spacings
spacing
as
well.
Yeah
I
think
I
think
that's
that's.
It
I
also
need
to
implement
this
part.
This
at
Mamata
static,
but
I,
read
to
I
need
to
read
the
the
domain
extracted
of
mine
from
the
location
and
also
extract
description
from
the
meta
tags
and
also
an
icon.
That's
essentially
the
favicon
or
something
similar
to
present
before
the
the
domain.
C
I!
Can't!
Let
me
see,
I
can
show
you
the
Redux
store,
because
the
Redux
store
is
not
connected
at
the
moment,
but
because
of
the
extension
the
relaxed
star.
Does
it
work
directly
but
yeah
it's
essentially.
If
I
pick
the
image
you
see
that
tits
come
coming
from
me
flora
from
a
profess,
so
it's
set
up
already
and
are.
C
No,
no,
the
signing
part
will
only
be
made
as
soon
as
we
complete
IDM
yeah
because,
as
as
you,
as
you
might
know,
the
the
the
current
your
ports,
mobile
application,
doesn't
allow
any
signing.
So
you
can't
sign
stuff.
What
we
could
do
is
much
I
I
mean
that
that's
the
overall
IDM
strategy,
but
what
we
could
do
is
to
generate
the
session
key
or
an
app
key
or
something
like
that
and
then
sign
it
with
a
mobile
with
a
mobile
application
because
they
have
like
claims.
C
You
can
make
claims
and
we
can
do
it
as
a
claim
and
use
that
session
key
to
to
sign
stuff
here.
But
then
you
can't
prove
that
you
are
that
the
signing
key
is
in
the
in
the
ID
document,
because
you
don't
have
it
so
it
sort
of
arts
for
outsiders
to
verify
the
ownership
of
the
comments
and
another
issue
of
your
part
is
that
it's
not
peer-to-peer
friendly
because
I
have
this
thing
here.
C
Unfortunately,
I
have
this
thing
here,
which
is
in
directions
which
I
can't
have
it
here.
This
is
the
secrets
secrets
a
shared
secret
between
you,
ports
and
my
up.
Then
is
that
it
doesn't
belong
to
the
salty,
can't
be
in
the
source
code
of
front
end,
so
it
doesn't
work
correctly
or
I
mean
it's
not
suitable
for
peer-to-peer
ups,
because
I
I
have
to
have
this
code
somewhere
in
the
code
base.
Yes,.
C
B
C
I
opted
out
for
you
port
as
a
stopgap
between
the
viewport
and
IBM,
essentially
because
it's
already
implemented
is
easy
to
log
in
easy
to
create
an
account,
and
so
I
can
so
I
can
move
on
to
the
the
data
model
and
mention
the
author
and
profile,
the
name,
the
picture
and
so
on,
which
is
already
real.
But
once
we
got
EDM,
we
win
the
application
of
psychic
signatures
and
signing
stuff,
and
so
on.
I
It's
really
cool
and
it's
exciting
to
see
that
you
can
implement
to
an
actual
identity.
You
know
just
something
we're
gonna
want
to
do
for
IG
is
so
I
have
a
couple
questions
I've,
actually
written
a
Chrome
extension
before
and
I've
gone
through
all
this
post
message.
Well
that
you
are
not
dealing
with
in
the
moment
and
I'm,
just
wondering
how?
How
is
this
performing
so
far?
Just
in
your
development,
environment,
I'll.
I
C
So
what
it
does
is
essentially,
when
you
edit
a
file,
it
waits
for
the
webpack
compilation,
and
then
it
reloads
the
page
with
the
app
to
date
code
automatically,
so
that
so
that
you
don't
have
to
go
into
the
Chrome
extensions
and
reload
it
to
yourself.
It's
it's
do
do
that
automatically
for
you.
Basically,
besides
that,
it's
pretty
fast,
the
web
compilation
is
fast,
I
mean
I,
don't
have
any
any
issues
in
terms
of
performance.
I
C
G
So
mostly
I,
just
wanna
show
the
the
UI
I
built
for
debugging.
So
everybody
see
this
yep
okay,
so
this
is
basically
pure
pad,
but
I
stripped
away.
All
the
the
the
extra
UI
and
I
just
boiled
it
down
to
just
the
text.
Editing
component,
not
the
other
bits
of
the
collaboration
and
what
I've
done
is
I've
built
this
little
area
here
and
I
know
if
I
should
label
these
different.
But
this
is
this
is
like
the.
G
The
computers
that
are
currently
in
the
collaboration
I
just
taken
the
last
three
characters
of
the
pure
ID
and
use
them
in
there
and
I
also
should
display
the
vector
clocks.
This
shows
that
this
particular
web
browser
is
a
n2
and
then
I
have
a
little
legend
here.
So
you
can
see
the
ones
that
are
have
the
outline
around
them
are
in
the
DSS,
which
means
that
those
are
the
ones
that
wants
to
connect
here
and
you
can
see
as
he
connected
this
connect.
Two
people
come
and
go.
G
You
can
sort
of
watch
it
all
shut.
So
there's
like
other
there's
other
collaborations
that
might
be
happening
that
aren't
part
of
this
collaboration
and
and
when
new
things
show
up.
They'll
show
up
here
in
this
other
line.
First
and
there's
also
a
way
so
you'll
keep
track
of
peers
that
are
no
longer
connected
at
all,
but
they've
they're
in
the
vector
clock,
they've
typed
characters
into
the
document
before
but
they're
just
like
gone,
so
it
all
shows
up
in
the
UI
here.
G
D
F
G
H
G
Issues
with
corruption,
I
discovered
that
and
I
think
that's
just
in
the
front
end.
So
it's
like
when
it's
applying
these
patches
from
the
editor
against
the
the
CR
DT,
so
I
think
if
the
here,
if
there's
some
asynchronous
thing
happening
wrong
there,
where
the
CRT
T
is
being
updated
before
the
patch
gets
applied.
F
G
He
should
show
up
and
then,
if
everybody
was
to
disconnect
this
thing
would
sink
the
document
and
then
even
then,
and
when
somebody
joins
up
again,
this
I'll
have
a
copy
of
it
and
it'll
keep
it
alive.
So
so
there's
there's
little
issues.
I
just
discovered
it.
If
you
last
night,
because
I'm
I
actually
have
three
of
these
Pinner
set
up
on.
My
is
pure
podcast.
Yet
version
and
I
know
this.
G
For
some
reason
it
wasn't
I
can't
leave
them
if
I
leave
them
running
for
a
long
time,
they
they
sort
of
loose
the
plot
and
they
stop
connecting
to
people.
So
I
put
in
a
little
thing
where
I
have
three
times:
I,
just
restart
them
every
10
minutes,
and
hopefully
only
when
restarts
at
the
time.
So
then
they
they'd
keep
the
data
alive,
but
that
died
actually
triggered
another
bug
so
that
one's
actually
sort
of
funny
I'll
show
it
to
you.
Cuz
we
go
to
is
pure
pad
passed.
Yet
when
it
first
loads
up.
G
A
G
I
decide
that
this
morning,
so
I
think
it's
because
I
had
these
things
we
starting
all
night
so
but
it
seems
to
be
okay
now
so
anyways.
It's
there
were
some.
The
burglar
fixed
the
last
week
were
really
were
deadly
to
the
stability
of
the
connections.
So
hopefully
I'll
get
these
bugs
in
and
that'll
be
also
a
useful
useful
for
discuss
the
fire,
I
think
as
well
and
I.
Don't
we
could
even
deploy
them
on
the
public,
your
pad
I
guess
so.
G
A
G
G
Figure
out
how
to
like
the
way
it
is
now
like
I
could
do
all
the
you
know.
If
the
work
is
like
on
the
Pinner
app,
which
is
certain
OGS
apps,
so
we
can
add
all
sorts
of
features
into
that
one.
So
right,
yeah
there's
a
question
there
like.
How
do
you
put
these
it
like
assigning,
maybe
domain
names
to
things
and
that
that's
a
whole
extra
level
there's
something.
G
A
G
Should
be
yeah
I
would
I
would
like
to
actually
maintain.
They
have
an
app
that
maintains
the
list
of
content
I'm
pinning
into
my
ipfs
node
and
just
have
a
C
or
D
key
to
manage
that
Oh
get
back
it
update
for
multiple
machines,
I
ripped
out
all
the
encryption
on
this
one
like
the
keys
and
that
so
that's
enough.
That's
another
work
to
do
is
to
go
back
in,
and
how
do
you
make
these
things
so
only
you
can
edit
it
or
edit
it.
C
It's
not
really
a
question
is
about
the
encryption
parts
which
which
I
tried
like
weeks
ago
when,
when
we
try
to
use
peer,
pass
to
actually
put
the
notes
of
this
meeting,
but
it
was
really
slow
in
my
in
my
MacBook
because
of
the
the
time
it
took
to
encrypt
and
decrypt
the
payloads.
So
essentially
it
was
blocking
the
white
thread,
which
is
you
know,
causes
decide
to
be
very
sluggish.
G
Yeah
I
did
some
profiling
and
I
saw
like
the
most
of
the
CPA
was
being
used
by
the
encryption
so
but
there's
a
it
could
just
be
that,
like
the
things
compiled
wrong
or
something
so
like
that
would
be.
The
first
thing,
I
would
look
at
because
I've
gotten
stuck
with
that
using
different
encryption
library,
it's
not
that
slow.
So
you
should
be
possible
the
encryption
faster
than
that.
A
There's
also
one
singer
on
a
PhD
I
just
tested.
This
was
a
few
months
ago.
I
tested
the
performance
on
some
things,
sanity
related
and,
if
I
just
disable
DHT
it
will.
It
was
much
faster
because
well
at
least
one
when
you're
creating
records
I'm,
not
sure,
if
that's
the
case,
but
something
to
look
look
out
for
is
the
background.
Dht
yeah
that
it's
all,
of
course,
that
is
also
encryption
and,
at
least
for
the
psyche,
is
or
the
channels
that
between
between
all
the
peers,
I.
G
Have
turned
out
it
turned
out
the
encryption
and
it's
still
sluggish,
so
there's
things
that
are
not
encryption
which
are
making
it
feel
sluggish,
I.
Think
it's
more
perceived
performance,
so
I
think
it's
like
a
synchronous
like
throttle
this
or
debounce
that
or
probably
in
the
wrong
place,
so
like
sometimes
I
type
characters
and
it
shows
up
in
the
vector
clock
but
doesn't
show
up
in
the
UI.
So
we're
gonna
be
looking
at
that
today.
So
I
think
there's
just
problems
at
the
top
level,
so
so
very
cool.
C
This
case,
if
I
I
will
not
have
that
problem,
because
I
will
turn
off
encryption,
because
all
the
discussions
are
public
right
now
and
also,
if
even
if
I
in
the
future
have
private
discussions,
it
will
run
on
the
background
script.
So
it
won't
block
the
any
white
thread
on
the
any
white
handling
thread,
helping
happening
in
the
browser.
So
idea.
I
A
The
app
is
really
small,
so
the
club
with
the
charity
is
doing
all
the
work
and
then
you
can
inspect
membership,
and
then
you
can
get
up
the
file
on
certain
events,
but
just
just
think
the
clicking
way
ahead.
Maybe
this
this
could
be
long
in
something
off
threads
entirely
and
then
peers
are
at
with
with
interface
with
with
that
Service
Worker
in
somewhere.
C
A
A
C
A
C
Yeah,
if
it's,
if
it's
something
that
it's
opt,
opt
in,
it's
it's
pretty
easy
to
put
that
functionality
that
you
know
into
a
service
work
and
use
post
message
to
communicate
between
the
lemons,
red
and
whites
red
via
a
library.
That
will
just
do
that.
Exactly
that,
and
you
will
install
the
library
in
the
serviceworker.
A
I
Like
you
know,
you
can
actually
visually
see
what's
going
on
in
in
terms
of
the
networking,
which
is
something
I
really
get
unless
I
can
actually
see
it
with
my
eyes,
and
it
reminds
me
of
I,
don't
know
if
you
guys
have
seen
the
latest
file
coin,
videos
in
which
they
actually
have
a
like
a
visual,
animation
or
I.
Think
it's
actually
real
time.
What's
going
on
between
the
different
nodes,
so
I'm
gonna
could
be
worth
creating
something
something
like
that.
That's
very
specifically
for
showing
how
networking
happens
within
collaborations,
yeah.
G
A
I
know
what
you're
talking
about
yeah,
it's
it's
the
thing
but,
like
you
can
see
the
ultimate
in
the
entire
message,
like
a
wire
graphical,
Wireshark
protocol,
specific
graphical
Wireshark,
saying
it's
very
cool.
If
this
is
really
supposed
to
devote
to
debug
protocols-
and
it
consumes
not
the
problem,
it
consumes
a
very
specific
format
which
we
would
then
have
to
pipe
somewhere
central
to
get
a
visualization
of
the
whole
at
least
a
whole
collaboration.
But
no
no,
if
the
entire
network
would
be
read.
But
this
is
the
whole
collaboration.
I
Think
the
way
that
they
have
a
lot
of
logging
going
on
and
so
basically
it
just
they
just
tap
into
the
log
stream
and
they
use
the
log
stream
as
a
way
of
driving
this
UI.
So
you,
if
you
implement
the
longing
layer,
then
you
can
derive
the
UI
from
so
I'm
wondering
I,
don't
know
what
logging
there
exists
already
within
the.
A
A
Or
we
could
check
into
that
BFS
layer
and
next
time
well,
instead
of
hard
for
a
TFS
later,
because
it's
it's,
it
will
all
be,
it
will
not
be.
We
will
not
be
able
to
inspect
the
semantics
off
of
the
packets.
It
will
be
harder
to
interpret,
but
yet
peers
are
at
level
yeah.
We
could.
We
could
absolutely
do
do
that
at
CDC,
ality
Delta's
traveling
from
beaut
appearance
II
like
to
prove
whether
also
debug
a
problem
would
of
thought
to
prove
whether
those
are
many
most
of
the
other
stuff
like
that.
A
C
C
Do
a
post
post
message
API
around
it,
I
pick
to
a
postmessage
api
and
have
like
dev
tools,
tab,
for
instance,
listening
listening
on
those
events
and
drawing
graphical
interfaces
right
on
dev
tools,
for
instance,
or
even
you
can
develop
an
extension
and
that
extension
will
be
listening
on
those
events
as
well
and
draw
draw
stuff
that
is
happening
under
the
hood.
It
will
be
awesome,
that'd
be
really
good.
I
like
that
idea.
C
C
No
I
mean
it
could
just
develop
a
simple
like
you
just
you
just
define
the
events
and
the
data
of
those
events
and
then
just
do
a
post
message.
Call
saying
these
events
happen
with
these
arguments
and
then
the
the
chrome,
the
chrome
tab,
a
developer,
developer
extension
tab
will
listen
to
that.
Whatever
extension
will
con.
Listen
to
that
and
we
will
be
drawing
all
that
stuff
in
those
in
those
tools.
Basically
I.
I
I
A
A
C
A
C
I
So
I
found
a
bug
in
a
is
Delta
C
oddities,
with
the
RGA
implementation
that
I
submitted
a
PR
for,
and
it
was
managed
and
to
be
honest,
it
wasn't
really
a
bug
in
inspection.
It
was
actually
a
bug
in
the
persistence
layer.
That
is
the
way
that
the
information
is
encoded
and
decodes,
and
so
I'm
wondering
if
you
guys,
have
any
tests
around
sort
of
testing
stuff
in
jazz,
Delta
C
IDT's.
But
you
know
with
the
persistence
layer
in
mind.
You
know.
A
A
So
definitely
so
it's
it's
just
they.
They
internal
state,
has
maps
and
sets
and
new
javascript
niceties,
but
it
doesn't
translate
really
well
to
to
jason,
and
so
we
do
encoding
and
decoding
at
the
burette
level,
using
message
back
and
we
message
back
extension
for
sets
and
and
maps,
and
but
you
don't
we
don't
test
that
the
correctness
at
that
level.
We
don't
know
encoding
and
decoding.
So
that's
a
good
question.
A
A
A
Right
so
one
thing
latin
is
that
you
will
migrate
to
a
key
LD
point
in
the
future.
So
I
don't
know
if
that
that
internal
state
of
deltas
are
going
to
be
represented
in
IPL
d
itself
or
or
they
will
just
be
blobs,
transparent,
apple-pie,
coops,
sorry,
but
that's
the
kind
of
a
related
thing
encoding/decoding,
because
archaeology
already
does
some
of
that
will
have
to
do
to
dig
into
more.
But
can
you
point
me
to
to
the
issue
I'm,
not
sure
if
I
got
that.