►
From YouTube: ipfs-desktop on macOS, IPNS and PeerStore in JS, NFC trial - IPFS GUI and Browsers Weekly 2020-03-18
Description
Meeting Notes: https://github.com/ipfs/team-mgmt/pull/1128
About IPFS GUI and Browsers Weekly: https://github.com/ipfs/team-mgmt/issues/790
IPFS Mirror: https://ipfs.io/ipfs/bafybeifgwvqtpgchvub4z65nraop6dxcqvyanhsfr77lc435c5f4mxp46e
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
A
B
A
B
A
Actually,
I
think
so,
because
I
got
clear
smells
before,
but
those
were
most
like
bug,
fixes
and
I've
added
a
lot
of
tests,
for
course,
on
the
HTTP
gateway,
because
the
course
logic
got
changed
across
three
factors
and
I
just
wanted
to
save
guards
that
we
don't
have
any
regressions
there
but
yeah.
This
is
like
pretty
big
for
me,
the
biggest
PR
gratulations
against
go
IPO
first,
which
was
pretty
pretty
fun,
given
that
I'm
doing
this
goal
go
curriculum
in
parallel
at
the
snail
speed,
but
at
least
I
wrote
some
go.
A
That's
it
for
about
me.
Let's,
let's
just
quickly
reiterate.
Why
does
this
matter
this
matters
because
it
adds
additional
modes
to
http
gateway,
the
HTTP
interoperability
layer
that
coati
affairs
exposes
on
which
can
be
queried
by
an
HTTP
client,
including
web
browsers,
and
this
specific
change
is
the
main
consumer
of
this
will
be
a
web
browser.
So
why
do
we
need
some
domains?
A
We
need
subdomains
for
origin
isolation
on
HTTP
gateways,
so
until
we
got
a
native
protocol
handler
or
native
ipfs
support
in
web
browsers,
we
need
to
use
gateways
as
that
intermediate
step,
and
the
problem
with
that
is
entire.
Web
security
context
is
based
on
the
origin.
Origin
is
protocol,
hostname
and
port,
usually,
and
if
you
use
active
as
IO,
you
get
you
get
the
one
one
origin,
so
everything
under
gateway
will
share
cookies,
datastore
subdomain
gateways
of
that
by
creating
a
separate
sandbox,
/
content
route
and
realigning
the
web
security
model.
A
A
A
A
However,
now,
if
you
go
to
Hugh's
Locker
host
hostname,
then
this
request
will
continue
to
work,
but
in
web
browser
it
will
be
redirected
to
sub-domain
version,
and
next
step
would
be
me
adding
supports
to
active,
has
combined,
hopefully
that
will
land
at
the
end
of
this
week.
So
that's
just
a
teaser
I
don't
want
to
derail
this
call
we
subdomains
once
again,
but
it's
like
we
are
getting
there.
It's.
B
Four
four
four
browsers
world
I
mean
this.
This
is
a
it's
a
huge
turning
point
like
it
this
it's
when
the
one
of
the
changes
that
is
required
for
us
to
be
able
to
turn
in
the
project
from
experimental
and
prototype
and
things
I
got
to
like
this
is
the
expected
and
production
level
of
security
that
browser
vendors
will
expect.
A
A
It's
just
pretty
exciting
to
see
it
land
speaking
about
like
exciting
stuff,
IP,
first
desktop
past
Mac,
OS
installation,
I
believe
I
did
not
mention
it.
Last
week
it
was
like
working
progress
last
week,
so
if
anyone
is
using
Mac,
OS
Catalina
or
later
now,
you
no
longer
need
to
deal
with
this
error.
Instead,
you
you
have
an
option
to
run
it,
which
is
good.
C
A
Think
about
it.
The
next
item
is
something
I
sort
of
mentioned
on
the
previous
call,
but
it
feels
typical
for
GUI
and
web
browsers,
one
about
IP
NS
resolution
in
web
browser.
So
it's
something
I
was
not
fully
aware.
That
was
the
card,
current
situation
so
false
from
terminal.
They
are
experimenting
with
mutability
in
a
web
browser
and
the
problem
was
there
were
not.
There
are
not
able
to
resolve
IP
and
s
addresses
mutable
paths
in
web
browser,
and
that's
mostly
because
it
turns
out
particles
checked
an
hour
ago.
A
The
current
implementation
is
like
hard-coded
bound
to
like
implementation
of
the
HD,
and
we
probably
could-
and
we
don't
have
the
HT
in
J's
ipfs.
Yet,
however,
we
have
delegated
routing
modules.
When
you
can
delegate
ADHD
queries
effectively,
not
like
specifically
tht,
but
just
like
Rudy,
you
can
do
content,
routing
queries
and
Peele
routing
queries
like
JSF
EFS
can
delegate
them
to
some
Kawai
TFS,
which
exposes
HTTP
API
endpoints,
basically
like
even
default
delegate
servers
provided
at
like
if
sio
gateway.
A
D
Basically,
we
would
only
have
to
change
the
config
for
browser
in
for
a
lipid
field
to
use
the
content,
routing
nodes
which
I
think
currently
it's
not
using
by
default,
but
I
think
with
that
it
shouldn't.
If
it's
our
camera
thing
works
immediately,
it
should
be
like
a
day
of
work
or
something
it
should
be
pretty
easy.
I
can
even
talk
on
it.
Next.
A
Level
when
you
ask
for
peers-
and
you
get
or
when
you
ask
about
peer
addresses
and
to
get
those
addresses
back,
those
usually
will
be
TCP
addresses
I'm,
not
sure
it
feels
like
there's
like
a
missing
step.
We
may
be
needing
some
sort
of
like
a
real,
a
mechanism
for
dialing
those
TCP
ones.
If
we
want
to
truly
truly
unlock
this
delegated
routing
cuz
right
now,
it's
kind
of
limited
you
ask
the
HT
for
PID
or
like
which
period
is
have
this
content.
You
get
peer
at
least
then.
E
D
D
D
Can
you
just
open
the
issue
and
scroll
to
the
milestones
swing?
Yeah?
Basically,
this
issue
contains
a
that's
good.
Thank
you.
This
issue
contains
all
the
stuff
that
we
discussed
in
the
past
in
week
regarding
the
opportunities
and
everything
that
we
can
achieve
from
the
real
factor
of
beer.
Sir.
Basically
we
have
been
discussing
after
the
team
week.
We
should
schedule
the
the
workmen
Jacob
and
basically,
we
decided
through
something
like
the
five
milestones
that
we
we
get
there.
D
The
basically
the
the
first
three
should
shouldn't
be
blocked
now
in
the
I,
can
immediately
start
to
work
on
them
and
that's
where
I'm
currently
starting
to
do
while
I'm
waiting
for
a
feedback
answer
list.
So
basically
the
tree
of
death
for
the
first
three
we
don't
provide
for
the
two.
The
first
two
are
just
basically
refactoring
and
there
will
be
no
bigger
gains,
but
it
will
unblock
the
remaining
ones
and
that
the
third
one
will
be
the
first
one.
D
It
will
really
bring
the
improvements
for
the
browser,
that's
when
we
will
get
back
to
peer
start
and
then,
of
course,
in
K
the
mem.
That
story
will
be
the
default,
but
if
the
user
provides
or
I
suppose
yes
ffs
will
provide
so
in
GFS,
it
will
be
just
the
default.
The
data
store
for
the
browser,
and
basically
you
will
get
the
peers
that
you
already
know
available
since
you
start
and
you
don't
need
to
go
and
discover
new
peers
once
again.
So
that
will
be
the
first
game
after
that.
D
But
we
are
limiting
also
the
parallel
dials
that
we
can
do
and
basically
we
can
start
for
dialing.
Imagine
generally
the
first
three
multi
others,
but
the
coolest
move,
the
others
would
be
the
fourth
or
the
fifth.
And
basically,
we
will
need
to
discuss
and
decide
the
scoring
thing
to
make
a
multi
other
confidence
approach
and
also
those
move.
The
others
should
have
a
time
to
live
and
will
shoot
like
evaluate
once
in
a
while.
If
we
should
improve
them
or
not.
D
There
are
a
lot
of
notes
that
we
grab
our
it,
but
it's
still
like
discussing
discussing
stage
so
far
and
with
that
we
also
want
to
do
that
peers,
priority
thing
which
basically
will
interact
with
connection
manager
in
order
to
trim
properly
the
connection
that
we
have
in
the
event
of
having
too
many
connections
open,
which
now
is
a
problem:
okay,
just
close
them
almost
weeds.
No
logic
is
just
if
I'm
not
wrong.
D
We
have
some
metrics,
but
it's
just
like
almost
blind
or
trim
connections
and
these
scars
plus
the
multi
other
scars,
and
we
are
also
considering
the
protocols
because,
for
example,
if
we
have
a
peer
running
pub
sub,
this
beer
will
have
more
priority
to
know
other
peers
which
are
running
pub
sub.
So
the
score
eventually
for
this,
those
beers
will
be
higher,
but
it's
also
a
work
in
progress
to
define.
I
will
decide
the
priority
of
peers
and
that's
it
for
now
on
the
piers
darlin.
A
Thank
you.
That's
super
useful,
especially
I'm
excited
about,
like
multi
other
confidence
and
improvements
around
connection
manager
for
a
bit
of
background
in
in
the
brave
browser.
If
you
switch
to
embedded
JSI
TFS
node,
that
node
has
access
to
TCP
transport
and
also
we
run
delegated
routing
module
and
I,
like
sort
of
like
youth
connection
manager,
just
to
limit
I,
believe
it's
hard-coded,
250
connections
just
to
keep
a
browser
happy
and
don't
kill
brow
extension
if
it
runs
too
many
or
it's
too
many
resources,
but
it's
been,
like
always
kinda
sketchy.
D
I
think
the
three
milestones
should
be
pretty
fast.
At
least
I
would
expect
them
to
land
may
be
middle
q2.
The
other
two
I
would
expect
to
at
least
have
a
decision
by
the
end
of
q2
and
I
will
already
eventually
start
working
on
them
so
that
in
q3
like
beginning
middle,
we
should
have
all
of
that.
Okay,.
B
For
the
I'd
love
to
have
a
clearer
idea
of
you
know,
I'm
reading
through
the
goals
here
and
I,
see
like
within
the
context,
it
sounds
like
it
might
be
performance
related,
but
it
would
be
great
to
be
able
to
have
a
summary
of
exactly
which
parts
this
makes
faster.
For
example,
does
this
make
faster
reconnections
like
which
we're
looking
at
so
many
loads
of
webpage,
which
exact
bits
would
be
faster
when
they
start
making
transfers
or
if
its
initial
pure
collection
and
things
like
that,
it's
initially.
D
It's
basically
the
initial
setup
of
the
node
there
is.
It
will
be
faster
for
the
persistence
peer
store
itself
because,
basically,
we'll
just
start
and
already
know
the
peers.
Instead
of
like
needing
to
wait,
because,
for
example,
in
the
in
decentralized
parts
like
WebRTC,
star
server
and
the
other
signaling
servers,
we
would
need
to
make
this
initial
jump
on
gather
the
discovered
nodes
and.
F
D
D
B
And
is
this:
is
this?
Is
the
plan
to
is
the
plan
to
back
this
data,
persistent
in
our
existing
local
IQ
fess
repo
and
data
store?
So
then,
if
somebody
does
something
like
either
a
clear
private
data
or
they
do
or
we
hit
the
storage
quotas,
it'll
all
get
blown
away
with
the
rest
of
the
repo
I
think
so,
okay.
D
B
D
B
B
All
right
should
probably
be
different
for
these
things,
but
if
they're
all
stored
in
the
repo,
they
can
all
get
at
any
time
with
that,
especially
if
another
app
hits
the
you
know,
a
totally
entirely
different
domain,
whatever
hits
the
storage
quota
and
all
this
stuff
goes
away
as
well,
because
we
don't
press
data
at
this
point
so
I.
It's
just
good
to
know
where
we
have
these
dependencies
on
other
large
system
level.
Changes
that
we
need
to
figure
out,
like
especially
around
storage
quota,
make.
B
This
this
seems
I'm
less
worried
about
blowing
away
this
data
because
you
can
recollect
it
every
time.
It'll
just
be
like
the
status
quo
right,
but
it's
good
good
good
to
know,
for
example,
yeah
I,
wonder
if
like
if,
if
we
actually
did
things
like
persistent
storage,
we
still
might
even
want
this
in
non
persistent
storage.
A
A
B
This
is
awesome
now
I
wonder
if
we
could
get
I'm
trying
to
think
of
like
who
would
be
people
who
would
help
us
out
with
some
like
telemetry
or
metadata
around
initial,
like
node
and
peer
instantiation
time.
So
like
people
like
three
box
people
who
have
like
larger
deployments,
maybe
even
orbit
to
be
folks
who
might
be
able
to
do
some
and
be
able
to
like.
Let
us
know
like
how
just
how
big
is
this
change?
They
could
be
given
all
this
work.
B
One
thing
I
would
recommend
to
is
setting
up
some
initial
metrics
so
that
even
in
our
tests,
we
can
do
a
measurement
they're
like
just.
How
much
does
this
matter
like
set
up
a
couple
different
nodes
in
different
geographic
regions
and
then
see
how
much
they
a
cold
start
with
no
peer
peer
backing
at
all?
How
long
that
takes
to
get
to
to
say
a
hundred
peers
or
whatever
and
then
do
stuff
with
warm
start,
where
it's,
the
data
stack
to
be
able
to
measure
what
the
difference
is.
Yeah.
B
B
Yeah
can,
can
you
add
some
some
measurement
or
metric
bit
to
the
plan
for
this.
So
this
would
be
a
great
story
to
tell
like
just
how
much
we're
improving
the
performance
of,
especially
if
people
are
building
web
applications
with
it,
where
people
are
going
to
keep
coming
back
to
the
same
domains
over
and
over
sure.
B
If
we
can
get
a
lead
web
and
ipfs
working
together
again
really
really
cool
opportunity
there.
But
he's
been
working
on
taking
some
of
the
changes
that
that
iraq
we
made
to
do
to
be
able
to
get
it
to
perform
better
and
be
more
stable
and
taking
those
from
the
Gecko
core
patch
that
iraq
they've
done
and
putting
those
into
the
live
d.
B
That's
like
shipping,
the
feature,
even
in
experimental
mode,
still
super
important
and
creates
a
really
great
support,
but
also
people
are
interested
in
hacking
on
some
of
this
and
that
people
are
interested
in
like
building
up
their
gecko
development
skills,
doing
things
like
bugging
learning,
the
ins
and
outs
of
web
extension
architecture
and
infrastructure
in
Firefox.
This
is
some
place
where
it
would
be.
B
A
A
Ensure
those
IP
eyes
do
not
differ
too
much.
Oh
I
didn't
need
that,
though,
sake
cause.
Those
are
browser
extension
api's
for
manifest
video.
So
in
the
perfect
world
those
both
cases
would
look
similar.
That's
my
dream
to
have
a
single
manifest
for
all
browsers.
It
will
never
happen,
but
we
probably
like
both
people
working
on
this
grant
and
people,
including
linear,
could
learn
from
each
other
compare
notes.
A
B
A
Speaking
about
browsers,
what's
new
with
brave
and
the
updates
and
local
discovery
right
now,
we
believe
the
local
discovery
grant
was
signed
by
the
contributor
I,
don't
think
the
work
started
yet
because
it
took
some
time
on
our
end
and
that
person
is
working
like
had
other
projects
line
up.
I'll
probably
check
in
this
week
get
like
ETA
when
the
work
can
start
I
believe
it
was
signed
like
last
week.
A
A
I'll
make
it
more
interesting
by
opening
relevant
issue.
So
the
problem
with
the
existing
experiment,
which
is
embedded
JSF
EFS
in
brave
one,
is
performance,
it's
jazz
and
no
matter
how
much
we
try.
It
will
never
be
as
fast
in
chunking
and
other
things
as
go
in
this
specific
context,
cuz
I
know,
Alex
will
hunt
me
down
and
say
that
it's
faster
I
know
it's
faster,
but
not
in
browser.
A
Problem,
another
problem
is
that
Google
is
deprecating,
something
they
call
Chrome
apps,
so
Chrome
apps
will
be.
Deprecated
are
like
the
deprecation
process
started
and
they
plan
to
simply
stop
supporting
them
in
2022,
and
that
means
those
chrome
API
switch.
We
got
access
which
enable
us
to
expose
TCP,
port
and
other
stuff.
It's
not
like
those
IDs
will
disappear
in
2022,
but
probably
no
one
will
maintain
them
and
if
they
break
no
one
will
care
apart
from
us,
because
it's
no
longer
will
be
used
in
upstream
Google
products.
A
It
will
be
just
left
over
in
that
chromium
code
base,
so
it
will
be
up
who
lets
in
vendors
like
brave
who,
if
they
decide
it,
there's
a
value
in
maintaining
them.
The
maintenance
burden
will
be
shifting
away
from
like
Google
engineers
to
other
organizations
which
have
commit
access
and
commit
rights
to
chromium
project,
so
both
for
those
both
reasons
will
probably
not
like
immediately,
but
we
will
be
moving
away
from,
maybe
the
GSA
PFS
or
to
be
more
specific.
A
We
will
add
another
option,
another
experiment:
we
will
not
remove
them
by
the
GSA
PFS
as
long
those
api's
are
there
and
work.
However,
we
will
probably
move
to
more
performant
or
more
native
implementation.
That
could
be
go
ITF.
S
brief
is
bundling
binary
right
now
and
there
are
fine
with
pumping
co
hf
s.
That
requires
additional
orchestration.
However,
we
initial
discussion
we
had
around
this
were
that
we
could
reuse
ipfs
combined
as
user
interface
for
managing
that
load,
so
that
will
simplify
implementation
on
their
end.
A
Basically,
if
you
start
companion,
if
you
enable
that
experiment
in
brave
settings
which
you
can
do
here
like
this,
so
we
probably
could
have
this
power
button
which
enables
and
disables
compile
in
the
browser
it
would
also
like
start
and
stop
like
in
the
initial
proof
of
concept.
That
would
be
that
one
additional
API
that
we
need
to
make
that
happen.
A
That
depends
like
timelines
are
still
uncertain,
but
it's
possible
that
if
we
have
this
like
native
protocol,
180
I
and
the
rust
ipfs
is
good
enough.
Just
because
we
don't
need
like
entire
go
activist
arrow
implementation,
we
just
need
to
be
able
to
connect
to
activist
Network
and
get
specific
cid.
That's
basically
it
like
everything
else
we
can
do
in
companion
and
we
don't
need
all
the
AP
ice
and
stuff
like
that.
So
that's
the
option.
B
and
right
now
we
are
in.
We
are
sort
of
like
trying
to
evaluate
what's
the
timeline.
A
What's
the
complexity
for
each
it's,
not
it's
not
even
they
exclude
each
other,
but
we
probably
don't
want
to
make
those
efforts
to
overlap.
Maybe
that
one
will
happen
first
and
then
we,
when
the
rustic
FS,
assuming
it
will
learn
later,
we
may
switch
to
that.
That's
the
landscape,
not
sure
if
it's
useful
or
if
it's
answering
the
question,
but
that
does
that
the
current
situation.
A
For
now
we
will
improve
companion
for
sure
note
that
companion
is
still
using
old
libraries
before
asking
iterators
prefactor,
so
the
next
like
in
the
q2
when
we
switch
to
those
this
and
other
JCAP
first
note,
will
for
sure
be
much
more
performant.
That
is,
it
is
right
now,
so
that
will
be
like
a
immediate
win
for
users
of
that.
B
Now
that
we've
kind
of
got
beyond
some
of
the
other,
either
high
priority
buyers,
it's
exciting
to
see
us
be
able
to
kind
of
push
this
problem,
these
challenges
again
and
it's
a
nice
transition
I.
Actually,
the
the
protocol
handler
bit
I
did
I
poked
golly
again,
so
hopefully
we
can
get
some
movement
there
and
q2
on
that
protocol
handler
different
and
like
every
every
single
one
of
these
bits
will
each
one
of
these
are
a
little
little
pieces
of
that
bigger
puzzle
and
start
to
reinforce
each
other
right.
It's
just
yeah.
A
B
It's
coming
soon,
I
think
there's
one
one
final
issue:
we're
still
waiting
to
hear
back
from,
but
we're
close
enough
that
the
my
p0
by
the
end
of
this
week
is
to
draft
a
blog
post,
so
I'll
I'll
share
it
with
flix
here
to
be
able
to
get
feedback
on
that.
But
things
are
things
are
looking
good
right
going
out.
The
next
version
you.
A
Cool
III,
muted
myself,
so
it
will
be
exciting,
especially
that
it
will
go
like
with
this
original
selection
out
of
the
box.
So
that's
a
small
step,
but
it's
setting
the
proper
precedent
that,
when
the
support
it
ships,
it's
not
like
botched,
it's
like
respecting
the
web
security
model
and
I
feel
that's
sending
a
good
signal
for
the
next
vendor.
Who
will
do
something
signal?
A
B
A
A
Is
now
is
pure
language,
more
or
less,
and
it
works
again
on
Debian
10
there
were
issues
with
sandboxing
of
chromium.
Chromium
is
like
IP.
First
desktop
is
built
in
electron
with
electron.
Electron
is
basically
built
on
top
of
chromium
and
chromium
made
some
changes
about
the
default
mode
or
like
the
defaults.
A
B
C
A
A
B
A
B
I'm
kind
of
curious,
like
is
there
any
precedent
for
something
that
is
a
specialized
API
for
an
initial
like
peer
discovery,
but
doesn't
actually
do
a
whole
transport.
It's
like
well
Betsy,
for
example,
would
be
something
that
we
might
support
for
that
initial
connection
or
pure
discovery,
but
it
isn't
actually
what
we
use
for
a
transport
yeah.
D
D
We
are
also,
we
also
tried
the
web
loo
to
a
while
ago,
but
at
the
time
there
issues
regarding
transmitting
the
type
of
betrayed
data.
They
only
allowed
us
to
transmit
data
according
to
the
type
of
data
that
they
were
expecting
and
at
the
time
we
couldn't
do
the
discovery
where
it
might
have
changed
again.
B
A
What's
happening
with
me
today,
I
usually
have
dotted
the
like
the
keyboard
shortcut,
so
I'm,
pretty
efficient
with
this
today,
I
was
just
mentioned
that
it's
like
a
testament
how
how
slow
some
things
on
the
web
move
change
of
all
it's
pretty
useful,
even
like
documenting
it
and
looking
back
to
understand
that
it's
not
happening
in
one
day
week
or
quarter.
It's
a
pretty
long
process.
B
A
A
It
was
like
origin,
trial,
I,
believe
it's
like
this
year
or
at
the
end
of
the
last
year,
but
yeah
it's
important
yeah.
It's
for
sure
importance
to
I'll
give
us
be
part
of
those
conversations,
even
if
we
finally
don't
decide
it's
a
good
match
for
some
reason.
It's
very
important
to
to
be
part
of
that
conversation
I
this
week
we
got
local
and
offline
call,
which
is
happening
right
after
this
call.
So
unless
there
are
like
very
good
questions
or
topics
for
the
next
eight
minutes,
I
propose.
B
F
A
Dev,
which
is
more
more
like
technical,
mostly
because
the
IP
fs1
is
it's
always
like
a
wall
of
text
and
type
if,
as
def,
is
more
about
people
who
actually
build
or
discussing
technical
details.
So
it's
on
freenode,
it's
also
on
matrix,
and
there
are
discussion
forums
on
discuss,
active
sio.
But
those
are
you
should
like
less
less
technical.