►
From YouTube: Subdomains, E2E tests, browser design guidelines, stardust - IPFS GUI and Browsers Weekly 2020-01-09
Description
About IPFS GUI and Browsers Weekly: https://github.com/ipfs/team-mgmt/issues/790
IPFS Mirror: https://ipfs.io/ipfs/bafybeicm6vtgxcvpbtzhhpzhs2g6p6a5j65uusiv2glhzxehppih3ykgpm/
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
B
Teacher,
do
you
want
to
add
anything
how
did
I
like
budget,
but
now
that
I
think
I
think
there's
a
good
summary
like
making
sure
that
more
people
can
come
to
this
meeting
to
bring
up
their
critical
issues
and
that
the
teams
can
spend
more
time
doing
the
deeper
work
and
making
bigger
changes.
Happen.
I
have
very
clear
points
at
time
at
which
we
analyze
triage
and
handle
various
reports
and
getting
more
people
to
come
out.
B
A
B
One
thing:
let's:
let's
do
like
a
test
week
of
this
to
a
week
or
two
and
just
see
how
cares
if
people
feel
like
if
it
feels
like
like
things,
are
languishing
and
we
are
concerned
about
things
going
without
a
response
for
too
long
like,
let's
make
sure
like
we
can
do
a
test
like
we
can
have
a
github
action,
auto
responder
on
issues
to
like
hey.
We
usually
check
these
once
or
twice
a
week,
but
you
know
generally
once
a
week
during
this
meeting
feel
free.
B
So
we
can
even
add
an
autoresponder
like
that
to
issues,
but
let's,
let's
play
around
with
it
and
just
see
what
what
works
and
doesn't
see
if,
if,
if
the
world
goes
on
fire
after
we
haven't
answered
stuff
in
two
days,
then
clearly
we
need
to
figure
out
some
other
way
of
communicating
that
before
we
go
full-on
autoresponder.
Let's
try
it
out
for
a
couple
weeks
see
how
it
goes.
That
sounds.
A
A
A
So
the
old-school
gateway
is
our
path
based
when
you
have
our
origin
here
and
all
websites
are
under
that
single
origin.
So
that's
good
for
fetching
content
north,
that's
good!
If
you
want
to
do
application
that
utilizes
web
api
s--,
because
those
who
are
web
api
eyes
are
confined
within
a
single
origin
and
some
api
is
require
additional
permission
and
if
that
permission
is
granted
its
granted
for
entire
origin,
so
a
single
web
site
here
will
get
requested
permission
and
and
every
other
web
site
loaded
from
half
a
gateway
will
get
that
as
well.
A
A
The
default
port
for
HTTP
is
for
for
free,
I
believe
so.
The
proposed
solution,
our
subdomain
gateways
and
subdomain
gateways,
basically
solve
the
problem
by
creating
a
separate
origin
for
each
content
route.
So
content
identifier,
lands
in
subdomain
and
every
sub
resource
under
that
content
route
is
confined
within
a
single
origin.
I
probably
did
a
better
job
explaining
it
in
past
if
I
find
it
I'll
like
it
in
the
notes,
but
this
is
what
you
get
today.
It's
the
place.
A
A
Idea,
stration
at
the
gateway,
so
we
got
this
public
gateway,
a
list
created
by
community
and
there's
a
additional
column
here,
which
shows
which
great
will
support
this
subdomain
mode
right
now,
it's
like
the
web
link,
which
is
provided
by
actavis
project.
Well,
this
one
is
from
cloud
flower.
Someone
contributed
one
here,
but
remaining
ones
are
the
old
school
of
based
gateways.
Those
are
perfectly
fine
gateways.
However,
it's
important
to
understand
that
if
you
are
creating
a
web
application,
which
is
using
api
studs
are
sensitive
through
origin
isolation.
A
You
should,
if
you
are
sharing
links
to
those
websites,
you
should
use
subdomain
gateways
and
those
gateways
exist
right
now,
but
to
create
one.
You
need
to
do
path,
translation,
d'angelico,
reverse
proxy
levels.
Usually
it's
an
engine
X
configuration
which
takes
CID
from
host
header
and
puts
it
like
as
a
suffix
to
the
path.
A
What
we
want
to
do
is
to
make
it
seamless
for
people,
so
everyone
can
easily
run
their
own
subdomain
gateway,
and
to
do
that,
we
want
to
add
support
for
subdomain
gateways
natively
to
go
up
your
face
and
also
J's
ipfs.
So
there's
an
open
PR
here
and
we
proposed
a
way
for
people
to
configure
the
way
the
gateway
functionality
works
in
a
more
flexible
way.
A
Right
now
you
just
you
just
expose
a
port
and
that's
it,
but
some
people
may
want
to
simply
host
their
own
website,
which
is
backed
by
DNS
link
and
not
return
arbitrary
content.
Other
12
has
by
other
people
some
other
people
may
want
to
run
subdomain
and
half
gateway
using
the
same
hardware,
however,
using
different
domains
and
also
the
nesting
websites.
So
we
came
up
with
this
configuration
scheme
when
you
can
using
this
simple
notation,
enable
subdomain
gateway,
half
gateway
and
also
DNS
link
website
natively
in
go
idea,
fest.
A
So
that's
upcoming
in
the
following
weeks
and
months
on
subdomain
gateways
and
if
anyone
got
any
questions,
we
I'm
happy
to
to
chat
on
IRC
on
freenode
ipfs
in
web
browsers
is
the
name
of
the
channel.
A
link
is
in
the
notes.
Next,
one
is
awesome.
I
should
like
interweave
other
ones.
So
that's
where
I
will
make
it
brief.
So
m2m
test
for
web
UI
landed,
the
pr
is
merged,
so
web
UI
has
end-to-end
test
for
every
screen.
We
test
uploads
of
multiple
files.
A
Yep,
so
we
got
unit
tests
here
and
then
we
build
web
UI
and
then
we
run
and
twenties
against
goal,
and
we
see
you
can
pick
what
version
was
used
for
testing.
So
this
is
the
latest
stable
version
who
go
out.
Give
us-
and
here
we
test
it
against
Jessica
fest,
and
it's
also
you
like
the
stable
version.
The
next
step,
which
should
life
this
week,
is
to
run
to
run.
The
same
test,
however,
enjoys
a
TFS.
So
there's
some
like
open
TI,
which
I
was
debugging
today,
but
it's
the
work
in
progress.
A
The
idea
is
to
of
external
test,
so
the
juris
ipfs
every
time,
someone
changes
something
in
Josiah
PFS,
the
part
of
that
that
suit
will
be
running
the
same
test
that
running
the
web.
You
all
tested
against
the
head
of
joy
psyche,
Affairs
using
that
specific
binary,
so
that
will
make
us,
like
everyone
sleep
better
at
night,
so.
B
I
have
to
thank
you
so
much
for
your
effort
on
this,
both
of
you
and
it
looks
like
you
go,
do
some
reviewing
going
to
this
is
a
this.
Is
a
dream
of
mine
since
I
first
joined
the
project
so
and
you
did
all
the
work
to
make
that
dream
come
true,
so
I
just
wanted
to
give
a
special
thanks
for
happened.
It's
really
important
moment
there
milestone
in
the
project,
helped
the
security
and
for
quality
disability.
Thank
you.
A
Say
to
like
just
to
finish
up
that
environment
variables
that
you
can
pass
to
to
an
end
to
end
tests
against
either
go
out,
Jefferson
Jessup
EFS
from
NPM
or
you
can
just
start
some
developer
version
of
Jace
or
go
and
just
point
testitude
at
arbitrary
endpoint.
So
we'll
probably
use
this
one
for
Ingo
ipfs
because
they
don't
use
edgy.
A
B
So
I
think
I'm
gonna
want
to
schedule
a
separate
session
where
we
just
go
over
this
work
together.
Can
you
see
this
yep
alright?
So
this
is,
you
know
a
table
of
contacts.
We
started
with
the
set
of
research,
kind
of
a
review
of
existing
URL
bars,
interviews
of
people
that
work
in
the
space
browser
designers
at
browser
teams,
some
PvP
experts.
We
talked,
we
talked
with
us
about
our
attitudes,
our
understanding.
What
kind
of
the
problem
space
was,
what
the
challenges
are
gonna
be
and
from
there
really
walk
through.
B
B
Instead
of
understanding
what
the
components
are
lots
of
interviews
or
had
with
some
interesting
conclusions
about
URLs
and
addresses
that
I
think
will
inform
our
work
going
forward,
but
that
the
most
important
bit
are
these
kind
of
themes
and
patterns
that
that
we
identified
and
scheduled
meeting.
We
can
do
a
longer
review
of
this
as
browser
team,
but
also
present
this
at
the
one
of
the
upcoming
profess
week
least,
once
it's
finalized.
The
some
of
the
some
of
the
bigger
themes
are,
you
know,
obviously
around
trust,
so
the
existing
model
has
a
of
the
web.
B
Has
a
trust
model
that,
while
it
has
pros
and
cons,
is
something
that
we
need
to
understand
when
we're
loading
resources
from
ipfs
in
a
regular
browser.
This
is
a
question
that
browser
vendors
gonna.
Have
that
we're
gonna
be
able
to
be
able
to
have
advice
and
guidance
on
security,
what
it
was
it,
meaning
it's
tracking.
You
know.
One
of
the
themes
that
came
out
of
the
research
is
that
that
is
the
thing
that's
pretty.
That
seems
to
have
worked
like
people
are
worried
about
it,
even
as
they
use
Chrome,
which
is
this.
B
These
cognitive
dissonance
kind
of
happening
about
the
behaviors
that
people
have
and
we're
pretty
well
as
technologists
people
design
software
if
anywhere
that
phenomenon,
but
we're
like
what
does
it
mean
that
that
something
is
distributed
and
that's
something
that
you
have
to
visually
communicate.
So
some
of
the
you
know,
patterns,
twos
and
outs.
How
about
access
controls?
Network
status
is
extra
important
when
it's
not
just
a
point
to
point.
B
You
know:
request
response
pattern
showing
progress
when
things
are
indeterminate,
so
some
of
these
are
basics,
but
laying
out
these
guidelines
in
a
way
that
design
teams
and
security
teams
can
walk
through.
You
know,
walk
you
through
examples
of
iconography.
These
aren't
aren't
finalized.
These
are
we're
not
even
recommending
a
specific
set
of
economic
be,
but
we're
like
what
are
the
states
that
things
might
be
in
as
resources
they
being
loaded?
The
designers
are
gonna
have
to
design
for
when
they're
implementing
this
number
and
browsers.
B
What
are
the
you
know,
types
of
'invalid
circumstances
that
the
user
might
find
themselves
things
like
ways
of
integrating
availability,
different
states
of
navigational
events
during
a
load
really
important,
so
I
mean
it
goes
on
on.
It's
really
a
lot
of
information
here,
and
so
that
I
and
that's
one
reason-
I
really
want
your
your
feedback
in
particular
both
of
you.
If
we're
having
worked
in
the
GUI
and
browser
worlds,
might
be
FS
for
a
while,
so
I'll
schedule
a
separate
time,
but
then
we'll
also
show
this
with
the
broader
community.
B
It's
once
it's
finalized,
but
I
think
it's
gonna
be
hugely
useful
for
us
going
forward
and
we're
already
talking
about
kind
of
what
our
next
phase
would
be
in
terms
of
doing
mobile,
app
design
guidelines.
So
the
one
of
the
things
that
came
out
of
this
is
that
we
were
focusing
on
desktop
browser
implementation,
since
that's
kind
of
like
the
starting
point
forward
where
we
started
with
I
profess
opinion
and
initial
work
with
brave.
But
the
nature
of
workflows
on
mobile
are
really
different.
B
We've
thought
about
trying
to
fit
them
into
this
design
guidelines
to
start,
and
it
really
didn't
work
out,
because
not
only
is
the
user
interface
of
mobile
browsers
very,
very
different
in
desktop
browsers
and
differs
wildly
between
mobile
browsers,
but
there's
also
this.
Those
apps
are
a
subset
of
native
apps
on
mobile,
where
we
have
things
like
textile
and
birdie
and
bunch
of
other
different
projects
that
are
building
native
mobile
apps
on
top
of
IP
FS.
B
That
might
need
to
be
able
to
communicate
these
things
and
a
browser
is
a
subset
of
that
total
set
of
mobile
apps.
So
I
think
that
phase
two
is
probably
going
to
be
mobile,
focused
set
of
design
guidelines
that
apply
to
native
apps
as
a
whole
web
browsers
as
a
specialized
subset
of
that
type
of
application,
so
more
more
coming
soon
from
there,
and
this
will
probably
be
finalized
by
the
end
of
the
month
and
then
prison
I,
because
we
clean
read
blog
post,
like
that.
A
A
Yeah
see
see
how
I
change
the
topic.
We
all
know
talking
about
WebSocket
star
and
that's
important
to
start
here,
because
in
the
browser
we
don't
have
the
HD
yet.
So
the
question
is:
how
do
I
discover
other
peers?
How
do
I
ask
other
peers
for
content
that
I'm
not
able
to
connect
directly,
because
those
people
are
maybe
also
in
web
browser
and
they
cannot
open
TCP
ports?
So
it's
a
WebSocket
starts.
It
was
like
a
temporary
solution
for
connectivity
in
web
browser
context.
A
It
provides
means
of
both
announcing
ourselves
to
our
to
the
rest
of
the
ipfs
network.
A
WebSocket
star
is
server
and
the
transport,
so
images
ipfs,
it's
transports
that
you
can
enable
and
the
cell
it
uses
a
rendezvous
server.
So
the
rendezvous
server
is
service
that
you
are
somewhere
and
installations
are
on
a
port
for
WebSocket
connections.
That's
useful,
because
web
browser
can
connect
to
it.
So
when
Jessica
first
starts
in
web
browser-
and
it
has
this
wonderful
server
configured
somewhere,
it
will
connect
to
it
and
it
will
announce
itself
hey.
A
B
A
Yep,
so
you
get
an
address
which
starts
with
the
address
of
rendezvous
server
in
the
beginning.
The
wonderful
server
address
is
the
suffix
of
the
address,
and
then
you
got
the
PID
of
your
note.
So
basically,
it's
like
your
PID
is
announcing
itself
at
that
wonderful
server,
so
it
just
like
an
address.
Anyone
can
connect
to
it.
Even
other
nodes
running
in
the
browser
can
connect
it.
So
that's
why
WebSockets
is
pretty
old.
It
was
created
in
the
beginning
of
jsapi
first
project
to
just
give
a
base
like
give
a
basic
connectivity.
A
A
Websocket
style
made
by
an
M
kg
with
log
number,
it's
much
much
a
contributed,
a
better
version
of
WebSocket
start
some
time
ago,
and
we
I'm
not
sure
if
it
landed
or
not,
but
it's
it's
a
better
version
of
WebSocket
style.
It
addresses
some
problems
we
with
which
you
could
start.
The
problem
is
that
we
have
this
plan
to
to
sunset.
A
Websocket
star,
like
totally
the
problem
with
WebSocket
star,
is
that
you
need
to
hard
code,
this
random
server.
In
your
configuration,
it's
a
single
point
of
failure.
If
that
wonderful
server
is
down,
you
lose
the
connectivity.
Jfs
supports
multiple
random
servers
and
at
least
one
needs
to
work,
so
you
can
have
like
free.
That
gives
you
more
robustness,
but
still
it's
like
a
single
like
multiple
points
of
failure.
Those
are
centralized.
Signaling
servers
are
just
a
temporary
thing
that
we
use.
We
there's
this
issue
in
just
p2p
report
that
is
created
with
more
details.
A
What
I
said,
but
better
explains
about
history
of
WebSocket
star
and
why
we
want
to
move
away
from
both
centralized
services.
So
the
problem
is,
that's
like
Stardust
is
better
than
this
old-school
WebSocket
style.
The
problem
is
that
it's
not
fully
drop
in
the
replacement.
It's
it's.
This.
The
protocol
is
the
little
bit
different,
so
it
requires
a
different
set
of
random
servers
and
we
cannot
like
simply
shut
down
old
ones
and
on
use
the
same
host
names
to
start
the
starburst
servers.
A
So
it's
a
it's
a
it's
a
replacement,
but
he
logisitics
are
migrating
away,
given
that
we
have
users
who
rely
on
WebSockets,
tar
are
not
easy
and
I,
don't
think
we
like
committed
so
much
to
doing
that
change
if
we
plan
to
replicate
those
star
protocols
anyway,
I'm
not
sure
if
that's
useful
explanation,
but
that's
the
gist
of
it.
So.
B
It
is
the
the
essential
of
connectivity.
Approach
is
still
the
same
as
WebSocket
star,
but
it
has
some
improvements
around
fault,
tolerance
and
reliability,
and
they
do
you
have
an
understanding
of
who
the
existing
WebSocket
star
users
are
like
how
many
people,
or
who
are
we
breaking
if
we
just
shut
those
around
every
service
service
that
we
ran
down
today,
I'm.
A
We,
if
you
want
to
replace
website,
replace
WebSockets
dive
status,
we
would
have
simply
create
a
new
set
of
Stardust,
really
servers
while
still
maintaining
their
the
old
ones.
Can
they
exist
in
parallel,
yeah
yep,
because,
like
a
starburst
protocol,
it's
like
as
separate
protocol,
so
they
can
coexist
at
a
single
node
can
use
both
the
I
think.
There's
some
other
work
in
like
a
plan
that
we
like.
We
could
simply
like
in
one
that
release
of
JCAP
ifs.
A
A
If
it's
using
WebSockets
and
that's
the
problem
with
WebSockets,
it's
not
only
signaling,
it
also
is
used
for
pushing
bytes.
So
the
is
web
RTC
star,
which
is
be
better
because
it's
using
or
it's
using
this
relay
only
for
like
actually
level
and
signaling,
and
then
there's
peer-to-peer
WebRTC
connection
established.
So
that
would
be
much
better
to
get
that
to
the
point
where
it's
enabled
by
default,
because
it
would
put
less
load
on
our
infrastructure
on
our
like
star.
A
B
Okay,
thanks
for
thanks
for
clarifying
I,
wonder
what
like
what
is
the
I
just
saw
some
references
to
to
the
star
test
bit
and
so
I
was
curious.
If
that
actually
ended
up
helping
us,
it
was
that
laying
the
groundwork
for
eventually
getting
like
client-side
WebRTC
DHD
things
like
that.
But
it
sounds
like
it's.
A
The
way
I
see
it
is
that,
like
after
Jacob
b2b
project
are
done
with
three
factors,
I
think
it's
worth
checking
in
with
them
about
this
peer-to-peer
signaling.
What's
the
time
frame
for
actually
seeing
that
and
then
we
can
reevaluate,
is
it
worth
switching
to
star
based
in
the
meantime,
that's
like
an
open
question
about
benefit.
Does
it
provide
enough
benefits
to
to
like
make
it
worthwhile?
A
B
A
It's
a
Monday's,
but
Jacob
is
on
call,
implementations
call
as
well,
so
I
cannot
use
that
question
there
and
that's
pretty
good,
because
Iran
is
also
there
and
Alan
from
JSI.
Dfs
will
be
extremely
interested
in
that
change,
because
that's
a
little
bit
the
key
connect
like
the
WebSocket
style
right
now.
It's
a
key
connectivity
feature
for
Jays
idea
facing
web
browser.
Basically
cool
next
one
may
be
compiling
on
message
about
missing.
That
demon
is
too
serious.
A
B
Is
price.I
strides
file?
This
is
an
issue
we
can
skip
it.
Okay,
it
was
like
I
stopped
competing
on
Firefox
nightly
and
it's
not
working
I'm
stuck
here.
I
was
like
no
you're,
not
you're,
not
stuck
that.
That's
exactly
where
you
should
be
so
maybe
yeah.
Maybe
we
should
revise
that
message
to
make
it
clear
and
like
communion
still
provides
benefits.
Even
if
there's
no
local
node
there's
a
whole
set
of
features
that
companion
provides,
but
it's
not
really
clear
from
that
page.
A
The
thing
is:
that's
on
the
screenshot
in
that
tweet.
Let
me
lecture
especially
it's
interesting.
It's
a
mobile
version
of
Firefox.
It's
actually
not
Firefox,
not
it
maybe
Firefox
for
Android
beta,
but
it's
not
firefox
nightly,
so
that
I
think
I
think
it's
just
firefox
android
yeah.
So
the
thing
is
that
it's
the
same
UI
is
the
same
extension.
That's
pretty
cool.
A
You
can
run
Firefox
for
Android,
not
Firefox
preview
Firefox
for
Android
the
but
separate
apps
on
the
snare
original
one,
the
original
one
and
you
can
run
the
same
browser
extensions
and
as
in
desktop,
including
help
of
his
companion.
But
the
problem
is
that,
like
the
screen
is
created
for
desktop
user,
you
cannot
let
go
to
the
console
or
unless
you
wrote
at
your
phone
and
start
the
daemon
you
can
install
like
there
I'll
give
as
demands
for
all
on
the
Play
Store
for
Android.
One
is
a
dual
effects
which
ipfs
the
combines
like
IP.
A
First
droid
I
tested
it.
It
works
if
you
start
to
them.
Companion
in
this
Firefox
for
Android
will
detect
it
and
start
here.
It's
just
sort
of
like
a
mixture
of
message
that
sent
that
is
centered
around
desktop.
We
probably
could
detect
that
you
are
running
on
Android.
The
problem
is
like
Firefox
for
Android
is
deprecated
and
five.
Then
the
future
of
Firefox
on
Android
is
something
called
Firefox
preview.
It's
a
different
like
engine
in
different
code
base
and
that
code
base
does
not
support
browser
extensions.
They.
B
Actually
more
for
Android,
so
on
Android
they
want
to
be
able
to
sometimes
use
whatever
the
native
rendering
engine
is
and
then
sometimes
use
gecko
on
different
Firefox
based
products.
So
you
might
have
a
Firefox
based
product
that
uses
the
native
in
or
you
might
have
a
fire
fixed
based
product
that
uses
gecko.
Okay,.
B
A
To
mention
so
you
don't
need
to
actually
on
this
ipfs
droid
are
like
a
diamond.
You
can
go
to
preferences
of
an
extension
in
Firefox
400
and
switch
them
better.
Jcap
FS.
You
won't
be
able
to
use
it
for
browsing
websites,
but
you
will
be
able
to
like
upload
a
photo
and
get
a
link
and
share
it
to
someone
on
mobile.
A
B
A
B
A
B
This
is
an
example
of
what
I
think
would
be
cool
to
basically
ditch
all
the
companion
just
for
as
an
example
like
for
companion,
remove
all
the
branding
make
it
an
IP
FS
share
and
only
highlight
that
one
feature
publish
anything
to
ipfs
using
the
images
boom
all
does
this
year.
That's
it
share
freely.
Whatever
you
call
it
like,
you
know
unstoppable,
chair
yeah.
C
A
Yes,
yes,
good
screenshot,
and
just
like
that,
there's
this
spec
called
sub
origins
and
it
had
like
a
long
life
like
it
had
a
good
run
years
past.
But
everything
comes
to
an
end
and
I
closed
this
issue.
So
now,
let's
take
three
seconds
of
silence
in
memory
of
this
spec.
That
was
never
meant
to
be.
Oh.