►
From YouTube: Companion v2.10, shipyard repos, webui e2e, browser ux - IPFS GUI and Browsers Weekly 2019-12-18
Description
About IPFS GUI and Browsers Weekly: https://github.com/ipfs/team-mgmt/issues/790
IPFS Mirror: https://ipfs.io/ipfs/bafybeigm3jso2ysl2kvuedqi3fegzqfvg6ffyel5lfjkdxb6gfxttrorpm/
For more information on IPFS
- visit the project website: https://ipfs.io
- or follow IPFS on Twitter: https://twitter.com/IPFS
Sign up to get IPFS news, including releases, ecosystem updates, and community announcements in your inbox, each Tuesday: http://eepurl.com/gL2Pi5
A
Welcome
to
Glee
and
in
web
browsers
we
call
its
18th
of
December
2019
and
this
week,
I'm
joined
by
Dietrich
and
you
go
their
arsenal,
general
feel
free
to
add
items
to
it.
I
believe
the
first
items,
mine
so
I'll,
just
I,
just
show
my
screen.
The
first
item
on
the
agenda
is
the
stable
release
of
IP.
First
companion
2.10.
It's
a
wrap
roundish
number.
A
First,
one
is
the
experience
of
visiting
website
which
is
backed
by
DNS
link.
The
second
one
is
improved
control
over
IP
FS
integrations
like
enabling
disabling
hit
pair
specific
website,
and
the
third
space
is
the
experience
of
importing
files
to
ideas
so
importing
files
type
EFS
in
the
past
looked
pretty
crew.
It
was
like
a
crude
experience,
users
added
file
to
ipfs
using
hyperfest
companion,
and
it
opened
the
same
file
at
the
gateway.
It
was
either
public
or
local
gateway,
but
it
was
it
user
did.
A
A
The
longer
you
use
it
so
we
switched
our
GUI
application,
including
like
IP,
first
companion
to
NFS
and
now,
when
you
import
file
using
ipfs
companions,
interface
on
the
menu
using
this
one
share
files,
I
profess
or
using
the
context
menu
like
click
right,
click
on
an
image,
select
image
and
import
ifs.
The
file
will
be
opened
in
web
uisce
files
screen
so
in
it,
and
it
will
also
be
put
in
a
directory
with
a
timestamp.
You
can
customize
the
path.
A
So
it's
now
much
easier
to
reason
what
ty,
what
files
I've
added
when
I've
added
them.
If
you
want
to
simply
stop
hosting
some
files,
you
just
remove
them
from
your
NFS
using
web
UI
user
interface,
no
law,
you
no
longer
need
to
worry
about
low-level
pins.
We
no
longer
even
put
low-level
pins.
Just
the
presence
of
the
file
in
your
em
FS
keeps
it
around,
protects
it
from
being
garbage
collected.
So
that's
pretty
big
improvement
for
user
experience
and
also
you
can
customize
the
path
during
the
ad
hoc
imports
for
a
single
file.
A
That's
I
think
that's
it
for
import
improvements.
Then
the
control
of
ipfs
integrations
per
website
is
improved.
Now,
there's
a
single
toggle
under
active
tab,
for
example.
Here
you
can
see
it's
a
DNS
link
website,
loaded
from
local
gateway
integrations
are
enabled.
So
that
means
the
website
was
loaded
from
local
gateway
and
stuff
like
window
ipfs,
all
the
content
scripts
run
got
injected
on
the
page
and
are
running.
However,
if
you
like
disable
this,
the
redirect
will
be
reverted.
Their
website
no
longer
will
be
redirected
to
local
gateway.
A
You
will
get
option
to
open
it
on
your
great
way
of
your
preference.
In
this
case
it's
a
local
host
one,
but
that
will
happen
in
a
new
tab.
It
will
not
like
share
the
origin.
It
will
open
your
tab.
That
would
be
a
separate
an
instance
of
that
website.
So
if
you
prefer
that
you
can
also
change
this
behavior
and
like
disable,
redirect
of
the
enhancing
website
permanently
or
just
like
our
website
using
this
toggle
I,
believe
that's
it
more
or
less.
We
also.
A
If
you
choose
to
disable
redirect
of
DNS
linked
websites,
you
have
option
to
steal,
in
the
background
to
reload
the
data
to
your
local
node.
So,
even
though
you
would
use
the
original
HTTP
server
or
flag
of
the
website,
outers
choice
if
the
server
goes
down,
I
prefers
companion,
would
try
to
recover
and
open
it
from
local
Gateway
and
local
Gateway
would
not
have
to
waste
time
for
finding
the
content
on
the
network,
because
that
would
be
already
on
your
local
node
and
there's
new
web
UI,
which
has
some
improvements
around
sharing
files.
A
Now,
when
you
share
a
single
file,
the
link
contains
the
file
name
and
some
interesting
developments,
including
restored
ESR
compatibility
for
Firefox
users.
So
if
you
are
using
Firefox
CSR,
this
latest
version,
we
is
now
compatible
again
and
you
can
get
all
those
new
features.
I
believe
that's
it
for
companion,
updates,
Oh
last
last
thing
to
mention
is
you
can
install
it
today
from
Firefox
store
at
Firefox
add-on
store,
or
you
can
wait
about
one
week
until
it's
accepted
to
Chrome
Web
Store.
A
If
you
are
like
chromium
browser
user
details
are
here,
I'll
probably
mention
it.
Multiple
times
of
people
already
tired
of
me
mention
it,
but
if
you
install
from
Chrome
Web
Store,
you
will
get
the
older
version.
Eventually
it
will
automatically
get
updated.
But
for
now,
if
you
want
latest
and
the
greatest
maybe
consider
switching
folks
who
knows
I
think
that's
it
and
the
questions
are
on
this
release.
Yeah.
B
B
The
directory
structure
of
the
resulting
of
MF
s
adds
here
end
up
with
these
directory
a
file
directory
file
directory
file
directory
file,
which
is
kind
of
a
pain
and
not
really
how
the
experience
that
people
have
in
finder
or
Windows
Explorer
today
and
there
we
get
their
file
system.
So
what
is
the
path
forward
for
utilizing
core
support
for
metadata?
When
will
that
lammed?
What
part
of
that
future
will
be?
And
how
will
that
is?
There
is
the
goal
here
to
migrate
to
that
type
of
metadata
system,
as
opposed
to
this
directory
structure.
A
It's
a
good
question:
it's
a
first
tab
and
the
simplest
way
was
just
to
make
it
like
time
stamp
based.
So
at
least
you
know
when
you
uploaded
it,
and
you
can
reason
that
oh,
it's
like
a
year
old
or
like
half
a
year
old
I,
probably
don't
need
to
this
data
around
when
it
comes
to
metadata.
We
like
we
are
technically
able
to
store
things
like
if
it's
imported
from
a
specific
website
like
the
original
URL
and
stuff,
like
that,
we
don't
have
a
good
place
for
storing
that
metadata.
A
Unique
surface
as
far
as
I
know
simply
just
allows
like
simple
attributes.
Maybe
new
surface
v2
would
allow
more
advanced
metadata
to
be
embedded
within
the
dock,
but
right
now
the
only
way
to
add
they
like
method.
Additional
metadata
is
just
like
to
add
additional
a
file
to
the
directory.
It's
an
open,
I'd
say
it's
an
open,
open
problem.
A
Is
it
like
refining
this
user
experience
of
how
how
file
import
should
look
like
I,
believe
I
prefers?
Companion
is
the
first
GUI
application
in
IP
for
second
system,
which
actually
does
this
so
because,
like
we,
we
had
to
figure
it
out
where
to
put
files
and
like
putting
everything
on
the
road
was
probably
not
the
best
idea.
So
this
is
like
the
second
worst
worst
idea.
I,
don't
have
good
answer
to
this,
just
that
it's
a
first
iteration,
even
the
like
timestamp
times
time
based
path.
It's
just
like
a
proposal.
A
We
can
change
it
in
the
next
version.
We
can
migrate
users
like
the
new
users
to
a
new
path.
It's
it's
not
like
it's
breaking
anything.
If
we
change
the
default
destination,
it's
just.
It
just
needs
much
more
thought,
especially
if
we
make
this
change
of
adding
files
automatically
imports
them
to
NFS.
A
There
was
the
there
is
a
proposal
by
Alain
I
believe
to
like
change
like
ipfs,
add
command
the
command
blind
one
I
PFS
import
and
make
it
instead
of
like
the
adding
low-level
ping
by
default,
importing
it
to
a
directory
in
EMFs
like
slash
imports,
and
it's
also
like
the
same
open
question
there.
How
should
that
directory
structure
be
look
like
what
type
of
metadata
do
we
want
to
store?
Do
we
want
to
store
like
the
path
from
which
we
imported
the
file?
A
So
it's
I
feel
those
are
valid
questions,
but
we
should
not
like
evaluate
them.
Excuse
me
evaluate
them
only
like
from
IP,
first
compile
perspective,
I'd
say
it's
more
important
to
think
about
them
from
that
perspective
of
this
command-line
core
API
v1,
which
will
replace
the
current
one,
not
sure
if
it's
useful,
that's
I
see
a.
B
A
B
I
think
from
it
from
it
from
a
design
perspective
too
I
think
we
kind
of
having
a
goal
like.
Is
it
an
explicit
goal
to
have
a
files
8
as
close
to
a
file
system
like
experience
or
not,
will
help
us
kind
of
figure
out
what
the
requirements
are
and
I
guess
you
know
the
TLDR,
like
my
ask
of
this
work,
is
that
we
think
about
what
that
we
decide
what
that
diet.
Design
goal
is
so
that
we
can
feed
the
requirements
back
into
the
core.
A
B
A
Yeah
sorry,
I
I,
probably
I
skipped
a
few
sentences
which
I
had
in
my
mind
what
I
mean
is
the
the
key
problem
that
this
release
of
ideas
compile
sauce
is
the
situation
when
someone
learns
about
type
EFS
and
they
like.
Oh,
they
add
a
file
from
the
command
line
like
a
facade
file
and
they
don't
see
it
in
web
UI
right
yeah.
A
So
the
user
experience
I
have
in
mind.
Is
this
entire
like
this
path?
When
someone
adds
from
the
command
line,
it
should
be
visible
in
the
web
UI,
and
it
should
not
feel
too
different
to
what
companion
is
doing
right
now,
so
it
should
be
like,
like
remove
complexity,
it
just
added
stripe.
The
first
people
should
not
think
that
it's
like
em
FS,
it
was
just
imported
and
I
can
like
manage.
A
A
Simply
check
switch
gateway
is
up
which
gateway
supports,
chorus,
color
and
stuff.
Like
that
I
see,
probably
my
PR
got
merged
cuz.
It
looks
different
yep,
so
there
are
like
there
have
been
a
list
of
peers
which
were
not
reviewed,
but
we
will
do
our
best
to
get
to
those
in
more
timely
manner.
I'm,
not
sure
if
there's
more
to
add
to
that,
it's
on
my
list
to
to
check
into
this
repo,
just
like
we
probably
should
for
awesome,
IP
of
SEO
and
other
websites.
A
We
just
need
to
figure
out
a
framework
how
to
ensure
there's
always
like
someone
checking
on
those
we
pose.
It's
just
like
yet
another
report
that
we
we
have
it's
like
under
ipfs
org
for
historical
reasons,
but
it's
actually
like
ipfs
CPR.
It's
like
community,
driven
provided
by
community.
We
don't
maintain
those
gateways.
Those
are
just
like
volunteers,
providing
their
server
servers
and
bandwidth,
and
we
should
like.
We
should
make
a
better
job.
I'll
figure
it
out
how
to
be
a
better
steward
to
those
public
resources.
B
Is
there
do
we
need
to
maybe
have
like
a
a
weekly
just
like
a
weekly
triage,
where
we
have
a
list
of
repose
cuz?
It's
definitely
like
this
meeting
like
we
talked
about.
I
saw
my
PFS
being
in
this
meeting
also,
and
that
was
kind
of
just
like
us,
stepping
in
to
fill
the
gap,
but
there
might
be,
from
a
community
perspective,
like
a
a
broader
ipfs
community
issue,
triage
to
make
sure
that
these
repos
are
are
not
falling
behind
I.
A
We
could
have
won
that
this
dedicated
board
for
those
ipfs
shipyard
and
community
repos,
which
we
really
want
to
attend
to,
but
not
always
have
time
or
there
are
like
other
fires.
If
we
had
like
separate,
could
be
part
of
community
call
or
maybe
a
separate
call
I
don't
want
to.
But
it's
something
to
think
about
you.
B
Then
I
profess
community
manager,
Job
Description,
should
go,
live
the
next
couple
of
days
and
I
think
that
would
be
like
the
definitely
the
role
that
would
maybe
help
drive
some
of
this.
But
there's
this
projects
like
this
are
kind
of
in
between
like
they're.
It's
not
awesome,
I
profess,
there's,
definitely
more
on
the
community
end
of
things.
Public
Gateway,
checker
areas
like
okay
are
people.
People
operating
gateways
are
definitely
part
of
our
community,
but
it's
a
very
different
roles
right
and
then
maybe
not
contributing
to
people
that
are
contributing
to
the
core.
A
A
If
we
see
the
build
the
way
it
works,
it
runs
into
and
tests
both
go
and
Jas
and
don't
stir
so
super
fast,
like
10
seconds,
because
it
there's
not
much
just
like
I
used
the
latest
ipfs
the
control,
which
is
super
cool
and
I'm,
just
a
puppeteer,
and
that's
it
and
the
way
we
maybe
not
here,
I
just
wanted
to
show
how
how
its
how
easy
it
is.
If
I'm
not.
I.
A
It's
as
easy
as
this,
but
you
you
just
run
tantrum
test
with
this
command
by
default.
It
will
run
against
go
ideas.
However.
You
can
customize
and
run
it
against
jails,
and
if
you
want
to
customize,
either
go
on
J's
version
that
you
are
running
against
you
do
that
with
you.
Do
that
with
within
package.json,
the
go
version
is
here:
it's
the
go
idea.
First,
step
is
a
go
binary
that
we
published
when
p.m.
and
Jace
ipfs
is
here
so
in
case
of
Jace
ipfs.
B
Okay,
I'm
curious
about
the
the
and
adhere
to,
if
you
have
any
opinions
and
thoughts
on
whether
package,
that
JSON
is
the
right
way
to
be
able
to
swap
different
versions.
So
what
does
the
workflow
look
like
for
a
go
engineer?
Who
wants
to
run
this
test
against
changes,
they've
made
to
Co,
IP,
vests
or
one
of
those
repos,
so
does
that
make
it
easy
for
them?
Does
that
work
without
they're
our
CI
works?
Oh
I,.
A
A
But
the
third
type
to
this
set
up
when
you
pass
URL
of
the
HTTP
API
that
you
want
to
test
against
and
that
will
enable
people
to
test
run
end-to-end
test
against
like
any
node
they
want
like.
If
they
build
custom,
go
ipfs
on
their
local
machine.
They
don't
have
it
published
anywhere.
They
just
run
it.
A
C
C
C
A
C
A
C
Jsf
defense,
you
basically
have
three
options:
one
for
the
binary
one
for
the
HTTP
clients
and
one
for
Jessica
first
score,
and
you
should
set
up
all
of
them
for
each
time
that
you
want
to
run
and
then
the
stuffy
you
are
using
the
environmental
variable.
It
should
go
into
spawn
and
not
in
create
factory.
A
C
A
A
A
B
A
With
a
test
for
our
original
selection,
I
yeah
I
simply
wanted
to
see
what's
under
window
location
object.
So
that's
a
good
smoke
test.
Is
there
actual
original
selection
because
you
can
the
URL
object,
so
you
can
check
origin
attribute?
Is
it
like
the
root
of
the
Gateway
or
is
it
like
the
same
thing
conduct
as
the
content
root
in
ipfs?
So
it's
a
good
place
to
check,
because
if
that
place
is
set
correctly,
nearly
it's
like
probably
everything
else
works,
because
that,
like
it's
like
the
key
indicator
for
hours
in
isolation,
all.
B
C
A
Is
to
have
unique
origin
per
ipfs
content
words,
so
ipfs
content
route
is
basically
like
the
cid
right
so
yeah.
If
they
use
under
like
under
the
UI,
if
they
just
use
HTTP
a
gateway,
it
needs
to
be
subdomain
gateway
if
they
implement
native
protocol.
The
native
protocol
itself
should
use
the
cid
as
the
outer
attic
component
of
the
URL
both
solve
the
security
problem
and
the
subdomain
gateway
is
probably
most
more
realistic
if
they
are
using
clicking
genre
and
like
they
need
to
like
provide
like
things
around
or
something.
Okay,.
B
B
Browser
design
guideline
so
Jim
Co,
Simmons
I'm,
working
on
that
they
passed
the
research
phase,
we're
now
into
the
final
set
of
recommendations
and
formulating
what
this
looks
like.
The
results
could
be
a
deck
and
that
will
probably
like
I'll
do
it
and
that
I
confess
weekly
will
like
main
a
test
back
by
giving
a
presentation
of
the
final
results
it
was.
It
was
really
interesting.
The
kind
of
the
interviews
of
we
interviewed
designers
at
a
couple
of
different
browser,
vendors
that
lead
designer
for
Firefox,
a
UX
designer
for
the
different
Firefox
Mobile's.
B
We
talked
to
the
lead
that
one
of
the
lead
designers
at
brave
and
then
he
it
also
interviewed,
but
a
number
of
non-technical
users,
kind
of
about
ideas
and
attitudes
around
the
location
bar
or
the
address
bar,
as
we
are
now
calling
it
because
they're
not
just
locations
there
anymore,
and
it
was
really
interesting
conclusion
or
findings
around
that.
The
human
readable
names
like
we're,
like
we
kind
of
in
the
abstract,
say:
hey
whatever's
in
there
should
be
a
human,
readable
thing,
but
it's
a
little
deeper
than
that,
like
people
actually
do
make.
B
Unlike
relative
value
judgments
on
on
how
much
they
understand
what
something
is
based
on
the
path
structures
of
addresses,
you
know
things
like,
even
if
not
technical,
you
understand
what,
like
blog,
slash,
2020,
slash
a
1/13,
slash
my
new
blog
post,
like
you,
can
look
at
that
and
at
a
glance,
understand,
get
some
understanding
and
value
from
what's
happening
there.
Even
if
you
don't
know
what
it
means
so,
there's
like
when
people
look
at
the
URL
bar,
they
might
not
understand
what
it
means
in
terms
of
a
network.
B
Operation
happening
request
response,
whether
it
comes
from
one
server
or
many,
whether
it's
of
what
a
404
means
of.
Why,
like
so
there's
a
lot
lack
of
understanding
from
the
technical
aspect,
but
there's
also
a
lot
of
cognitive
processing
that
happens
of
the
location
bar
people
know.
Padlock
is
good.
B
Green
is
good,
but
they
don't
understand
that
underpinnings
of
why-
and
this
makes
the
design
question
around
what
padlock
alternatives
would
be
really
important,
and
that's
kind
of
where
we're
at
in
the
in
the
process
now
is
synthesizing
the
results
of
that
research
in
order
and
then
using
that
in
the
context
of
what
ipfs
means,
when
it's
ipfs
colon
slash,
slash
in
the
browser,
you
know
we're
talking
about
willing.
What
is
the
distinction
between
transport
level
encryption,
but
an
absence
of
a
DNS
based
trust
model?
B
Well,
it
combined
with
the
cryptographic
verification
that
means
that
nothing
was
a
man-in-the-middle
so
or
also.
How
do
you
visually
communicate?
The
fact
that,
if
something
loaded
at
this
address
with
ipfs,
it
is
the
only
thing
that
ever
will
learn
at
this
address
or
nothing
might
load
at
this
address.
It's
either
this
or
nothing
yeah.
A
B
Like
so
we're
talking
about
things
like
each
part
of
this
might
even
be
like
put
going
to
the
thesaurus
and
giving
a
little
word
cloud
for
each
one
of
these
things.
So
we're
like
you,
know
currently
here's
what
the
lock
means.
How
would
you
show
this
differently,
like
with
the
lock,
have
a
a
little
badge
on
it
with
the
number
of
peers
connected
at
the
time?
You
have
a
constant
view
of
how
many
peers
are
connected.
You
only
show
where
the
peers
are
connected
when
somebody's
looking
at
this
resource
to
you.
B
Do
you
make
changes
to
the
lock
color
to
identify
the
transport
level?
Encryption
happens,
but
there's
no
business
registered
with
an
SSL
certificate.
So,
right
now
like
there's
a
lot
of
like
the
way
that
browsers
approach
communicating.
This
is
really
relies
on
DNS
as
the
trust
model
for
that
encryption.
B
So
you
have
transport
level
encryption,
but
ultimately,
like
trust
model,
ties
back
to
the
fact
that
the
SSL
certificate
ties
to
a
DNS
registration
and
the
DNS
restoration
is
tied
to
a
physical
address
that
you
can
mail
something
to
or
that
you
could
drive
your
car
to
you
and
walk
up
and
knock
on
the
door
and
also
believe,
that's
fake.
That's
false
right,
but
that's
the
only
thing
that
browsers
have
and.
A
B
I
know
like
so
we
have
this
collective,
like
from
us
from
a
from
a
browser
security
engineering,
though
an
internet
security
engineering
standpoint
they
they
do
lean
on
that,
like
that
is.
That
is
all
we
got
and
they
understand
that.
That's
like
we're
working
with
what
well
we
got
like
and
that
that
that's
we
don't
have
another
alternative,
and
the
alternative
that
kept
us
presents
is
is
challenging
because
well
we
guarantee
transport
level
encryption.
B
So
we
have
transport
level
security
from
that
perspective
by
default,
90
bucks
right.
We
don't
have
content
level
encryption
and
that's
fine,
there's
no
expectation
of
guarantee.
Now
we
don't
have
the
old
user
identity
and
key
system
no
problem,
but
the
trust
there's
a
non,
a
nonzero
value
to
the
trust
model,
that's
implied
by
having
a
SSL
register.
Dena
SSL
encryption,
that's
tied
directly
to
the
DNS
registration
of
a
specific
right,
and
we
don't
have
anything
to
offer
there.
We
say:
there's
no
man-in-the-middle
risk
great,
but
what
trust
Authority
does
the
does?
B
The
address
itself
come
with
we're
like
we're
done
mainly
there's
zero
and
our
path
based
addressing
model
kind
of
underscores.
The
risks
of
that,
which
is
everything,
falls
into
this
kind
of
global
scope
of
data,
and
you
hate,
wave
your
hands
and
say
if
it
came
unencumbered
with
risk
of
man
in
the
middling
that
you
should
be
able
to
trust
it
like
yeah
I.
B
Don't
think
anybody
really
actually
thinks
that
we
don't
think
that
really,
but
we
haven't
presented
another
model
for
evaluating
the
relative
trust
of
an
address
that
was
given
to
you
out
of
band.
If
that
address
came
to
you
through
a
chat
app
and
you
clicked
that
link
and
it
opened
up
in
your
IP
FS
based
mobile
browser,
how
do
you
evaluate
the
relative
trust
of
that?
Well,
it's
from
where
you
got
it
right.
B
So,
ultimately,
what
our
trust
model
comes
down
to
is
the
fact
that
we
say
users
have
to
take
the
responsibility
and
ownership
right
now
of
trusting
addresses
that
received
out-of-band.
Where
did
it
come
from?
That's
really.
All
we
have
to
lean
on
and
put
browsers
is
actually
doing
better.
You
know
they're
like
then:
we
have
this
whole
separate
system
for
for
placing
that
trust,
even
though
it
might
be
misguided
and
not
ultimately,
that
much
value,
because
you
can
hide
your
ID
either
for
the
little
low
cost
of
one
dollar
a
year
right.
B
So,
but
there's
a
little
bit
of
there's
like
this
gap
that
side
it's
pretty
useful
exercise
in
just
kind
of
thinking
through
where
those
gaps
are,
we
can't
pretend
like
we
don't
have
them
so
and
it
in
order
to
be
able
to
close
those
gaps,
or
at
least
minimize
them.
It's
really
important
for
us
to
acknowledge
that
they
exist
and
this
exercise
is
really
good.
B
Pre
in
January
well
polished
up
and
done
so
maybe
that
one
of
the
I
could
weeklies
in
January
will
be
a
walkthrough
of
this
work.
But
again
there's
this
like
it's
very
interesting
cuz.
This
is
the
really
it
jump
forward
in
thinking
about
what
native
IP
bus
looks
like
and
as
we
found
in
the
CID
and
subdomain
stuff
is
for
going
on
in
the
IFS
hosting
web
apps,
and
how
that
the
people
in
our
own
community
are
very.
They
lean
heavily
on
success.
B
Looking
like
a
local
host
URL
that
companion
redirects
to
a
local
gateway
that
redirect
that
is
desktop
and
that
looking
like
success
and
if
we
wanted,
if
we
want
to
leapfrog
that
to
like
being
able
to
set
ourselves
up
for
a
mass
adoption,
we
need
to
be
able
to
visualize
what
native
I
professor
actually
looks
like
and
actually
what
their,
what
the
relative
risks
are
and
designing,
for.
It
is
good
exercise
and
doing
that
so
I
feel
like
we're.
Gonna
have
well
there's
a
lot
of
side
benefits
of
doing
this
work
in
helping
us
reorient.
B
Our
own
community
thinks
about
what
ifs
success
looks
like,
and
it's
it's
complicated.
It
really
points
out.
Some
of
the
you
know
glaring
bits
in
the
in
the
overall
application
model
from
a
privacy
and
security
and
surveillance
like
we
don't
have
good
answers
for
surveillance
like
especially
if
I
every
ipfs
addresses
its
own
origin
like
you
can
still
set
cookies
and
stuff,
and
you
could
still
make
network
requests
to
two
ways:
people
they
could
track.
A
B
So
we
need
to
be
able
to
have
and
it
we
need
to
have
a
plan
and
and
a
story
around
how
we
communicate
those
relative
risks.
If
we
want
to,
you
know,
have
have
browsers
at
the
billion
user
or
even
hundred
million
user
level,
consider
adoption
ipfs.
So
this
is
some
front
work
way
in
advance
of
solving
those
problems
later
and
hopefully
setting
us
up
for
a
more
a
clearer
and
more
honest
understanding
of
where
the
gaps
are
in.
B
I
think
there's
a
there's,
an
exercise
to
be
done
in
trade-offs
to
like,
like
all
of
security
and
privacy,
is
a
question
of
trade-offs
anyway,
and
I
want
us
to
be
able
really
clear
about
when
we're
making
when
users
are
presented
with
the
options
of
trade-offs
there.
So
a
user
might
say
I
and
I
am
aware
that
this
content
is
not
sensitive
and
therefore
I
am
taking
down
my
tracking
shields
I,
don't
care
how
many
people
I
get
it
from
just
want.
B
The
image
to
load
and
I
am
choosing
local
network
reliability
and
stability
over
broader
trucking
guarantees.
But,
like
that's
a
that's
an
option,
we
should
let
people
make
and
I
looking
at
I
confess
choices
through
the
lens
of
those
types
of
trade-offs.
It
will
help
us
understand
the
user
better
and
what
their
needs
are
and
what
the
context
under
which
their
decision-making,
it's
gonna,
it's
gonna
happen.
A
It's
it's
interesting.
How
some
lessons
could
be
learned
from
from
the
old
world
like
from
the
existing
HTTP
system,
when
we
did
there's
this
problem
of
like
man
in
the
middle
or
like
centralized
servers
and
the
projects
like
tor
and
tor
browser
like
providing
a
different
set
of
guarantees
to
like
users
who
need
them,
and
they
there's
even
like
a
slider
which
is
like,
gives
a
convenience
versus
like
speed.
A
A
B
B
A
Think
they
also
made
some
experimental
enhancements
of
the
gateway,
so
they
like,
they
provide
the
gateway
which
you
can
like
pass
parameter
to
right.
Custom
content
type,
which
is
interesting
thing,
I,
want
to
look
at
because
it's
both
interesting
from
the
security
standpoint.
What
are
unknown
like
risks
related
to
that.
However,
we
had
historically
problems
like
people
wanted
to
load,
webassembly
files
from
our
gateways
and
our
gateways
were
not
returning.
A
When
we
participated
in
the
origin
trial,
we
had
to
overwrite
content
type
at
the
entry
next
level
and
a
lot
of
people
who
want
to
run
like
webassembly
under
stuff.
They
basically
need
to
override
that
at
the
nginx
level.
So
we
probably
could
improve
that.
Oh
I
can
merge.
I
can
merge
them
question.
Do
we
merge
or
do
is
caution
large?
We
probably
merge
thoughts,
feelings,
I,.