►
From YouTube: DNSLink UX in Companion, celebrating ipfsd-ctl v1.0.0 - IPFS GUI and Browsers Weekly, 2019-12-11
Description
About IPFS GUI and Browsers Weekly: https://github.com/ipfs/team-mgmt/issues/790
IPFS Mirror: https://ipfs.io/ipfs/bafybeif7zosi4nxftnndydkoyt2lfa44nhusxcs5yqbwggklr3jr5t7ari/
B
B
I
was
not
sure
if
we
will
make
this
week's
call,
but
there's
one
thing:
I
wanted
to
show
and
Hugo
join
us,
so
I'm
not
alone,
so
I
just
wanted
to
briefly
mention
recent
changes
in
user
interface,
effectiveness,
companion,
it's
related
to
discussion
in
ipfs
dog
sweeper
around
just
like
there's,
like
a
general
discussion
about
hosting
web
apps
on
ideas
and
challenges
related
to
related
related
paths,
and
then
things
like
origin
separation.
It's
bit
longer
and
if
someone
don't
want
to
read
everything,
there
is
a
short
summary
I
believe
here
with
decisions.
B
B
So
DNS
link
is
this
convention
of
adding
DNS
txt
records
with
idea
first
path
and
our
browser
extension
is
able
to
detect
that
there's
a
ipfs
path
for
the
domain
you
are
trying
to
visit
and
we
are
able
to
log
page
from
ipfs
instead
of
original
HTTP
server,
which
is
appointed
by
the
a
record
so
historically,
we've
been
always
redirecting
to
local
gateway
and
in
the
latest
best
of
idea
of
a
companion.
We've
made
a
short
experiment
to
change
that
behavior.
B
There
are
some
concerns
around
that,
mostly
about
the
fact
that
we
stopped
kind
of
distributing
content
on
the
initial
load.
I
proposed
mitigation,
I
believe
its
link
here,
so
I
propose
to
mitigation
that
pre-loaded
data
of
visited
websites
in
the
background.
So
if
the
original
HTTP
server
fails,
there
would
be
a
copy
on
local
on
local
server.
So
we
had
a
design
review,
call
around
that
and
we
sort
of
decided
that
we've
been
doing.
B
What
will
this
redirect
local
gateway
for
a
long
time
and
subdomain
gateways
are
on
the
horizon,
which
will
solve
the
security
issue.
We
wanted
to
stop
you
and
it
does
not
make
much
sense
to
a
break
the
current
user
experience
in
the
intermediate
in
that
time
period
before
today
and
like
the
beginning
of
2021,
we
will
and
subdomain
gateways.
So
that's
long
story
short.
B
We
flip
the
switch
back,
so
we
will
continue
redirecting
to
local
gateway.
However,
I
reworked
the
user
interface
a
little
bit
to
make
it
easier
to
basically
disable
all
I
give
integrations
for
a
specific
website.
So
if,
in
the
past
we
there
was
you,
there
was
a
user
interface
element,
I
believe
I
have
it
here.
Let
me
find
it:
no
I,
don't
have
it.
Maybe
I
do
I
know.
B
Anyway,
there
was
a
in
this
position.
There
was
a
toggle
for
enabling
or
disabling
redirect
for
a
specific
website
and
in
this
PR
I'm
changing
it
to
be
more
generic,
so
we
not
only
start
toggle
redirects.
We
also
stop
all
other
integrations
because
in
the
behind
the
scenes
we
inject
some
content
scripts
with
some
time
change
the
Dom,
depending
on
experiment
user
enabled
there
is
a
window.
Ipfs
object,
injected
two
pages
to
give
our
society
of
SAT
eyes
and
this
one.
B
We
want
to
provide
one
toggle
which
disables
all
those
integration
and
that's
basically,
it
is
under
active
tab
section.
So
it's
enabled
by
default
on
every
page,
but
you
can,
with
one
click,
both
disable
redirects
to
local
gateway
and
shut
down
all
ipfs
integration.
So
if
you
find
that
activist
companion
is
breaking
a
website
or
you
want
to
see
how
website
works
without
idea
for
integration,
you
should
be
able
to
take
advantage
of
this
new
agent
interface
and,
if
you
like,
disable
this
redirect
on
a
website
which
is
backed
by
DNS
link.
B
In
this
case,
it's
dogs,
ipfs
io.
You
can
see
it's
enabled
by
default,
but
when
you
disable
it,
it
opens
page
from
the
original
HTTP
server,
but
gives
you
additional
option
to
open
a
copy
on
local
gay
on
local
gateway
in
a
new
tab,
and
there
is.
There
are
additional
options.
This
entire
DNS
link
section
on
preferences
screen
to
to
tweak
the
behavior.
So
by
default
we
load
website
from
captain
gateway,
but
you
can
disable
it.
B
We
added
this
warning
so
this
we
tell
people
about
possible
risks
and
if
you
disable
it,
you
can
still
trade
out
visited
pages,
so
I
think
that's
flexible
enough
for
everyone,
and
the
important
part
is
that
the
default
state
after
initial
install
that
the
clean
install
experience
is
the
same
as
it
was
before.
So
we
don't
break
the
default
pathways,
which
molly
was
mostly
worried
during
the
call
I.
Think
that's
it.
B
On
my
end.
Regarding
this
change,
it's
not
merge
yet
the
I
link
the
PR
in
the
notes.
So
if
you
find
that
language
or
the
user
interface
could
improve
comment
on
the
issue,
I
want
to
like
merge
it
this
week,
release
to
beta
channel
and
then
publish
tool
to
stores
it
takes
time
to
publish
the
stores.
I
did
not
have
it
on
my
like
agenda,
but
I
think
it's
worth
mentioning
that
it's
getting
like
hard,
maybe
not
harder
button,
takes
more
time
to
publish
to
Chrome
Web
Store
lessons.
B
Yes,
let
me
find
it
because
I
always
try
to
be
very
transparent
about
our
interactions,
with
both
Mozilla
and
Google
around
to
the
browser
extensions
and
like
it's
a
constant
battle
recently
on
Chrome
Web
Store.
From
this
time
they
rejected
our
beta
release
for
some
reason,
with
like
canned
response.
Basically,
I
I
did
not
get
like
human
response.
It's
just
make
out
of
my
death
system,
they've,
updated
some
rules
and
then
our
extension
got
rejected,
but
it's
like
still
available
on
on
the
store
it
just
like.
B
They
rejected
the
latest
version
of
our
battle,
maybe
because
the
screenshot
is
like
pixelized
for
a
reason.
So
people
don't
install
like
the
better
version
by
mistake.
I
need
to
like
do
the
proper
description
screenshots.
Maybe
that
will
make
them
happy,
but
it
is
if
someone
wants
to
test
bet
our
channel
on
Chrome.
They
need
to
like
manually,
install
it
from
from
sources
or
use
the
3ds
version
attached
to
to
release
under
assets
here
all
right.
That's
me
if
you
have
any
questions
related
to
any
topic
side
glad
to
discuss
around.
B
C
A
A
So
some
of
the
improvements
are
these.
As
you
can
see,
it's
probably
much
faster.
Now
the
errors
are
shredder.
We
have
debug
debug
logs
everywhere
we
have
a.
We
have
a
way
to
to.
The
factory
now
can
actually
spawn
several
nodes
and
manage
several
nodes,
so
you
can
in
one
go
clean
all
of
them.
You
can
also
specify
some
default
some
options
for
each
future
spawn
nodes
and
also
some
overrides
per
type.
A
We
have
three
types:
the
go
demon,
Jas
demon
or
a
Jas
in
process
node,
so
you
can
specify
overrides
for
each
type
which
are
all
merged
everywhere
properly,
and
everything
should
be
much
safer
to
deal
with.
We
can
also
you
need
start
and
set
config
in
one
go
at
least
in
JSF
FS
when
I
say
one
go
is
basically
using
one.
A
A
A
We
have
a
create
factory
method
which
has
a
as
you
can
see
a
secondary
parameter
to
overwrite
options
per
type,
so
this
changed.
How
so,
basically
we
had
to
create
and
use
it
to
create
factory
just
to
be
more
explicit
and
also
the
options
that
we
send
to
create
factory
or
previously
create
also
change.
You
can
see
some
of
the
differences
here.
A
A
Ctl
handles
remote
nodes,
so
if
you
want
to
start
the
jst
man
from
the
browser,
obviously
you
can't
write.
You
cannot
spawn
a
process
inside
a
browser,
so
you
need
to
have
a
little
server.
That
scale
provides
that
you
can
connect
to,
and
you
tell
that
servers
for
me,
jst
man,
so
this
is
called
a
remote
controller
and
Rev
is
basically
just
a
require
of
a
specific
package
and
when
dealing
with
remote
controllers,
we
cannot
send
the
reference
to
the
remote
controller
right.
A
So
we
need
to
send
it
send
a
path,
but
also
when
you
we
bundle
for
the
browser
we
need
to
have
a
as
an
explicit
required.
So
the
bundler
includes
that
package
that
we
want
to
use.
So
we
we
need
also
need
a
reference.
We
need
the
reference
to
be
equal
to
a
require
specific
require
explicit
required.
So
that's
the
reason
that
we
need
to
have
not
just
Rev
or
path.
We
need
to
have
both,
so
we
can
work
in
the
browser
and
we
can
also
work
in
in,
like
node
environments
like
Tron
or
whatever.
A
The
same
goes
to
the
other
one,
and
everything
else
is
straightforward.
We
also
previously,
we
had
a
spawn
method
in
the
factory
that
accepted
these
options,
which
are
kind
of
strange
because
they
mimic
some
of
the
ipfs
options
and
others
down
this.
It
was
pretty
strange
so
now
what
is
sent
to
the
spawn?
You
said
it's
basically
the
control
options,
all
all
the
controller
options,
but
you
can
specify
ipfs
options,
which
is
which
are
exactly
the
same
as
the
ones
that
the
ipfs
constructor
accepts.
A
A
We
did
also.
We
now
have
a
specific
option
to
set
the
FS,
the
CTL
to
test
mode,
and
we
only
apply
some
of
the
default
configuration
that
we
always
applied
previously.
We
only
apply
it
when
we
are
in
test
mode
and
also
the
default.
Others
option
was
removed
because
it
was
just
like
it
was
superfluous,
because
we
we
now
only
apply
the
some
of
the
configuration
when
we
are
in
test
mode,
because
we
really
need
it
for
tests.
A
A
The
others
now
actual
actually
take
into
account
to
environment.
So
now
we
that's
why
I
was
able
to
enable
a
bunch,
a
bunch
of
browser
chests
in
JFS
and
jsep
client
that
were
skipped
with
this
new
version.
I
was
able
to
enable
all
almost
all
the
tests
in
the
browser,
and
they
were
skipped
from
one
simple
reason:
we
couldn't
connect
when
we,
when
a
test
needed
needs
to
connect
several
nodes
to
do
some
stuff
in
browser.
All
of
those
tests
were
not
enabled
because
TL
didn't
didn't,
set
the
proper
addresses,
and
now
it
does
so.
A
Everything
is
wired
up.
So
remote
nodes
open
web
sockets
enable
WebSocket
server,
so
the
browser's
can
connect
to
remote
nodes
and
everything
is
properly
wire
and
the
connectivity
in
can
happen.
The
only
situation
where
we
can't
we
cannot
do
connect
that's
by
design
is
when
we
have,
or
by
it's
by
designer
limitation.
We
can
call
imitation
of
arousal
when
we
have
to
in
process
browser
nodes.
A
They
cannot
connect
to
each
other
because
we
have
no
way
inside
the
browser
to
connect
two
nodes,
everything
else
it
works,
and
this
made
all
the
tests
much
easier
to
end
of
the
hoax.
The
before
hopes,
the
the
after
hooks
all
the
spawning
it's
much
easier
to
do
all
the
overriding
for
specific
situations
in
all
of
our
repos.
We
have
different
configurations
in
specific
tests
and
we
can
now
have
said
we
are.
B
A
We
also
learnt
what
that
that
did
was
that
he
started
uncovering
stuff
that
we
mean
know
was
failing,
because
previously
we
were
always
running
inside
interface,
the
interface
core
tests.
We
were
always
using
nodes
of
the
controller's
of
the
same
type
and
that's
why
the
browser
tests
the
tests
were
skipped
in
the
browser,
because
in
the
browser,
when
we
set
like
type
in
process,
they
couldn't
connect
by
design.
So
we
just
someone
just
skipping
the
tests.
A
So
now
what
I
did
I
started
kind
of
specifying
a
primary
secondary
situation
where
at
least
one
is
in
process
and
the
others
need
to
be
either
j-s
or
go
so
what
I
uncovered
was
like?
We
have
pretty
a
pretty
serious
situation
where,
like
interrupting
between
all
of
these,
all
of
these
three
types
in
some
combinations
just
work,
so
there's
already
issues
about
it
and
we
will
work
on
it
and
it
will
fix
everything,
hopefully
as
soon
as
possible.
B
Was
just
sending
compliments
in
your
direction?
Thank
you.
Now
I
was
saying
that
it
is
very
important
that
we
start
to
try
mean,
like
those
different
context,
cause
a
lot
of
stuff
when
I
was
working
on
like
brave.
A
lot
of
stuff
that
was
passing
interest
did
not
work
in
browser
context,
so
it
will
be
easier
now
to
catch
those.
Those
cases
yeah.
A
Almost
everything
is
enabled
for
fun,
it's
just
a
matter
of
action.
Actually
we
don't
have
one
situation
that
I
think.
Well,
we
already
identified
the
source
of
the
error.
We're
in
pubs
are
subscribing,
doesn't
work
in
Firefox
specifically,
is
Firefox
and
that's
the
we
have
for
tests
all
around
subscribing
that
file
in
Firefox.
A
Everything
else
is
enabled
all
the
interface
core
tests
are
enabled
in
HTTP
client
in
jessup
EFS.
All
of
them
I'll
enable
HTTP
client
still
has
some
that
I'm
not
which,
but
less
than
than
previously,
but
JSF
infests.
All
of
them
are
enabled,
besides
this
specific
one
that
we
needed
to
skip
even
though
Chrome
passes,
the
Firefox
still
has
the
issue,
but
I
think
we
already
identified
the
source.
So
it
should
be
pretty
easy
to
fix.
B
C
B
We
are
at
the
end
of
agenda
all
right.
So
besides
thanks
so
much
for
dialing
this
week,
I
pretty
sure
for
the
next
two
weeks.
We
won't
have
this
call
like
double
check:
ipfs
community
calendar
but
I
think
those
are
already
removed.
The
next
one
will
be
in
January,
hopefully
with
with
more
more
people
after
everyone
will
be
refreshed,
so
see
that
SIA.