►
From YouTube: cohosting, ipfs-provider, bootstrapper docs - IPFS GUI and In Web Browsers Weekly, 2019-08-28
Description
IPFS Mirror: https://ipfs.io/ipfs/bafybeiabyc7ob6qgjb3wxems3cbovrpy7y2n7x2oufxplbvvsvkps2p3wu
A
Hello
welcome
to
GUI
and
in
web
browsers
we
click
all
its
28th
of
August
2018
and
and
we
at
the
usual
time,
if
you
are
not
on
the
list
of
participants
at
yourself.
If
you
have
a
topic
that
you
want
to
discuss
a
little
agenda,
I
added
some
items,
I
see
other
people
added
some
items,
so
maybe
we'll
just
jump
right
into
that
Irish
on
screen.
A
A
Files
API
should
be
the
only
thing
you
need
to
use
to
support
this
or
to
manage
this,
and
the
idea
is
to
move
away
from
low-level
pinning
when
it
comes
to
like
co-hosting
websites.
We
had
like
some
discussion.
They
took
asked
some
good
questions
around
use
cases
and
I,
try
to
the
speck
through
that
lens
and
commented,
which
things
are
possible
in
the
scope
proposed
in
this
PR
and
which
are
not
but
could
be
supported
either
by
writing
custom
code
in
user
land
in
your
app
or
tweaking
the
spec
a
little
bit.
A
Especially,
a
important
part
for
me
was
identifying
this
distinction
between
people
who
want
to
save
a
copy
of
a
single
website
and
people
who
want
to
lick
co-hosts
entire
website
right.
So
some
people
care
about
just
one
part
of
website
just
just
this
page
and
other
people
want
to
keep
the
entire
example.com
or
entire
Wikipedia
online
yeah.
So
that's
I,
believe
that's
a
good.
A
A
The
problem
is
that
a
lot
like
effectively,
if
you
see
everything,
is
possible
right,
but
we
we
need
to
decide
where,
at
which
point
we
say:
okay
enough
and
like
personally,
I'm
biased
towards
like
minimalism
and
like
keeping
the
spec
as
small
as
possible,
just
like
just
a
few
directories
like
three
files
commands
for
adding
getting
and
simple
algorithm
for
like
updating
and
that's
it,
and
just
just
for
websites.
I'm
I,
like
I,
noted
here
that
we
could
support
specifically
like
single
pages,
like
single
paths,
but
that
gets
hairy.
Very.
A
Very
fast
websites
can
have
sub
resources
from
different
paths.
So
we
either
like
just
host
a
partial,
the
power
part
of
the
website,
or
or
even
at
the
single
page.
Or
do
we
go
to
a
business
of
like
rewriting
HTML
and
then,
if
we
start
rewriting
stuff,
then
we
lose
the
debt
duplication.
So
you
see
there's
a
rabbit
quickly.
It's
a
just
a
rabbit
hole
which
I
personally
I
would
like
to
not
go
there
and
just
keep
it
around
websites.
B
Yeah
that
sounds
well
first
thanks
for
going
through
and
do
that
scoping,
because
that's
so
helped
me
to
understand
what
the
spec
enables
kind
of
out
of
the
box
versus
tweaks
and
I.
Think
you're
right
like
this,
really
clarifies
what
the
technology
they
already
have
in
hand
enables
easily
versus
having
to
do
a
whole
bunch
of
other
higher-level
application.
Stuff
then
I,
agree,
then
narrower
narrow
and
clearer
in
the
capability
of
the
spec
is
there's
gonna,
be
the
way
that
to
move
quickly
in
and
iterate
on,
something
that
is
useful
in
people
and.
A
A
Let
me
like
make
it
bigger,
so
it's
just
like
a
come-on
line,
told
you
just
pass
webs
like
host
names
of
your
website,
and
it
takes
care
of
fetching
it
and
pinning
it,
but
it's
pinning
it
using
this
low-level
ipfs
pin
command,
which
just
takes
a
Content
identifier
and
pins
it,
but
you
are
not
able
to
assign
like
a
label.
You
are
not
able.
You
need
to
remember
that.
A
Oh
this
CID
is
for
this
website
and
you
need
to
like
manually
manage
this
so
that
the
idea
basically
is
with
this
pack
is
to
upgrade
at
some
point
tools
like
EFS
co-host,
to
make
it
easier
for
people
to
know
what
they
are
co-hosting
and
make
it
easier
for
people
to
people
to
like
decide.
They
want
to
drop
the
older
snapshots,
or
maybe
they
want
to
just
stop
sharing
this
website
costs
like
takes
too
much
space,
or
they
don't
care
about.
A
This
anymore,
I
believe
like
that,
would
be
beneficial
because
the
toll
would
work
the
same,
but
then
we
you
just
go
to
web
UI
to
the
files
and
you
see
a
hosting
directory
and
it's
likes
it's
basically
like
self
describing
feature.
You
don't
really
need
to
even
know
that
that
thing
exists.
You
go
to
hosting
of
there's
this
domain
I
know,
and
there
is.
There
are
snapshots.
B
B
A
If
you
want
to
implement
this
right,
but
I
believe
we
should
add
like
a
section
of
use
cases,
maybe
not
I'm,
not
sure
it
should
be
spec
or
maybe
in
with
me,
but
basically
take
this
list
and
decide
which
we
include
in
spec,
which
we
don't
and
and
just
write
it
down
what
this
spec
enables
and
which
watch
what
is
not
in
the
scope
of
this
spec
and
just
write,
write
it
down.
I'll,
probably
do
that
and
then
we'll
make
a
final
pass
on
this.
A
B
C
We've
talked
about
it
in
the
past
bill.
It's
one
of
those
things
that
I
like
it's
got
some
real
simple
user
experience,
qualities
about
it
like
it's
very
easy
to
explain
to
users,
but
it
feels
also
like
very
fragile,
or
at
least
like
it's
a
mutable
file
system.
Anyone
can
call
there's,
there's
no
good
act.
There's
no
control
on
this
and
there's
no
guarantee
that
what
you
find
in
the
coast
and
directory
is
in
the
format
that
you
expect,
and
this
may
be
the
right
way
to
go.
C
But
it's
just
like
an
interesting
like
first
example
of
like
trying
up
trying
to
make
a
formal
proposal
to
make
that
a
thing,
and
every
time
we
talk
about
it
with
everyone
like
no
one's
spent
enough
time.
Thinking
about
the
trade-offs
of
using
the
well-known
directory
pattern,
certainly
amongst
us,
no
one's
got
really
strong
everyone's
like
feels
a
bit
floppy,
but
it
is
kind
of
easy
to
explain.
I
merely
offer
this
up
as
like
background
to
the
conversation
I
don't
want
to
I
wanted
to
rail
it
because
of
it,
but
it
it's.
A
A
Agree
and
sort
of
been
thinking
on
this
on
my
own,
namely
I,
know
it's
like
going
into
like
Bagshot
area,
but
we
need
to
go
there
at
some
point
like
this
proposal
of
like
putting
directory
in
the
root.
It's
like
just
like
a
placeholder
cause
like
it's
nice,
it's
in
the
road
and
people
immediately
see
this.
However,
we
could
have
like
actually
like
that
well-known
directory
under
which
we
would
put
stuff
like
that,
and
then
we
would.
We
could
present
that
directory
differently
in
web
UI.
A
That's
a
question
like
how
much
we
want
to
invent
new
things
right
now.
We
just
add
a
directory
to
em
FS
I,
believe
if
we
decide
to
not
put
it
directly
on
the
road,
but
in
like
that
config
or
that
well-known
director
as
a
subdirectory
of
well-known
directory
that
could
be
still
acceptable,
but
anything
like
more
than
that.
It's
it's
probably
diminishing,
returns
and
likely
introducing
multiple
layers
of
directories
for
people
to
click
through
there,
which
is,
like
my
fear,
yeah.
C
Yeah
in
its
kind
of
purist
form,
it
feels
a
bit
like
the
movement
of
convention
over
configuration
like
it's
really
clear.
Everyone
knows
it
just
like
the
coalition
directory.
That's
where
websites
go,
which
may
be
a
strong
reason
to
at
least
like
run
with
this
spec,
see
how
it
will
see,
how
it's
a
what
people
feel
about
it
and
and
then
it
maybe
gives
more
value
to
their
whole.
C
B
Recommend
in
the
spec
talking
about
the
well-known
directory
pattern
and
acknowledging
that
it's
a
vector
for
experimentation
and
for
playing
with
ideas
like
this
in
really
agile
ways
without
requiring
protocol
changes
like
I.
Think
if
you
kind
of
call
that
out
at
the
beginning,
like
we
know,
there's
this
architectural
e
fragile
bit
here,
we
are
taking
advantage
of
the
fragility
in
order
to
get
to
be
able
to
experiment
easily
and
quickly.
A
Things
like
that.
So
like
the
fact
that
we
don't
have
those
things
right
now
and
we
put
a
it
on
em
FS
could
surface
the
need
for
implementing,
or
at
least
discussing
those
things,
so
I
feel
it's
is
probably
like
not
a
super
priority,
but
will
slowly
should
iterate
on
this,
because
it
may
open
interesting
discussions
at
some
point.
A
We
got
a
bunch
of
all
right:
I
prefers
provider
veto
that
this
one
is
mine.
It
will
be
probably
fast
activist
provider
is
library,
JavaScript
library
that
aims
to
make
it
easier
for
people
to
add
ipfs,
as
I
said,
like
a
back-end
service
or
like
a
background
service
of
the
web
app
without
worrying
too
much
about
where
that
actual
ipfs
demon
come
from.
So
lot
of
apps
would
just
want
to
add
some
data
to
ipfs
and
get
cid
and
vice
versa,
so
a
cubist
provider,
it's
a
way
of
introducing
a
very
elegant
heuristic.
A
A
If
there's
none,
then
you
try
some
HTTP
api,
but
if
that
api
is
down,
then
you
fall
back
to
embed
the
JSA
PFS,
which
still
lets
you
access
data
on
ipfs.
That's,
that's
the
that's
the
dream!
We
have!
The
current
version
is
like
a
port
of
all
tool
we
used
internally.
However,
the
issue
I
linked
and
added
to
the
agenda
is
a
discussion
about,
and
programmatic
interface,
which
would
be
very
clear
to
understand
and
very
easy
to
customize,
both
which
providers
we
want
to
enable
and
in
which
order.
A
So
here
we
see
I
first
want
to
try
some
HTTP
API.
Then
I
want
to
fall
back
to
interface
provided
by
browser
extension
and
if
neither
of
those
work
I
want
to
fall
back
to
embedded
J's
ipfs
with
some
custom
configuration
and-
and
this
is
like
an
issue
where
we
discuss
how
this
future
interface
could
look
like.
But
basically
what
we
want
for
people
is
to
just
call
this
get
ipfs
and
they
should
get
working
ipfs
api
instance
that
they
can
interact
with
that's
more
or
less
it,
and
the
questions.
B
To
this
one,
oh
yeah
is
that
I
guess
you
know
I
mean
there's
probably
a
lot
of
reasons
why
you
would
want
to
choose,
be
specific
about
those
fall
backs
and
not
just
have
a
kind
of
like
deterministic
barbacks,
based
on
a
set
of
things
I
deal
in
settings
to
the
time.
It
might
be
good
to
enumerate
what
the
reasons
are
why
people
would
want
to
do
this
in
in
the
readme
for
this
yeah.
A
Totally
the
the
background
on
this
is
Internet
Archive.
It
would
be
one
of
you
of
users
and
the
way
they
want
to
this
to
work.
Is
they
probably
they
are
running
their
own
notes,
I
believe
so
they
wanted
to
try
API,
first
and
then
fall
back
to
and
by
the
J's
ipfs,
or
vice
versa.
I.
Never
remember,
but
I
know
it
was
like
very
unusual
and
in
the
old
library
it
was
not
possible.
A
Similarly,
window
ipfs
is
interface,
experimental
interface
provided
by
browser
extension,
but
some
people
want
it
comes
with
limitations.
It
comes
with
like
sandbox
for
filesystem
API,
so
some
application,
which
need
access
to
like
regular
full
files,
API,
might
decide
to
just
disable
this
provider
because
it
does
not
provide
the
full
likely
first
experience
so
don't
select
to
from
the
top
of
my
head,
but
you.
B
A
A
A
The
problem
is
that
actually
we
have
like
regular
dogs
in
go.
We
have
regular
dogs
about
configuration
in
Jessica
fares
and,
if
you
read
all
those
dogs,
you
may
connect
to
dots
and
figure
out
how
to
set
it
up,
but
effectively.
A
lot
of
people
would
like
to
run
their
own
ipfs
note
on
the
backends,
be
that
JSF
EFS
or
go
ipfs
and
have
that
backend
service
exposed
WebSocket
endpoint
for
the
thin
for
the
J's
ipfs
running
in
the
website
to
use
it
as
a
bootstrap
node.
A
The
problem
is
it's
like
a
list
of
eight
servers
which
which
we
always
try
when
we
try
to
reach
out
to
to
the
swarm,
and
if
you
are
building
a
web
app
with
ipfs,
you
probably
want
to
at
least
add
the
some
additional
bootstrap
nodes
to
ensure
that
your
application
remains
functional
and
is
able
to
connect
to
the
swarm,
even
if
the
public
default
boot
shot.
Nodes
are
down.
Another
reason
why
you
want
to
use
custom
bootstrap
nodes
is
our
limitations
of
JC
DFS
in
the
browser
context.
A
We
want
to
access
ensuring
it's
already
connected
to
a
node
with
that
data
improves
that
content
discovery
speed.
So
that's
not
very
long
provides
why
this
matters
case
matters
and
the
problem
is
we.
We
are
missing
a
good
like
one-stop
tutorial
for
setting
this
up,
because
there's
a
K
dot.
The
caveat
is
that
WebSocket
transport
is
supported
by
go
ahead
of
us,
but
it's
disabled
by
default
and
it's
disabled
by
default,
because
by
default
it's
not
encrypted.
It's
just
WebSockets
without
any
encryption.
A
On
top
of
it,
and
most
of
deployments
will
run
on
8-hd
PS,
which
means
JSI
PFS
won't
be
able
to
connect
to
unencrypted
web
socket.
If
you
want
to
use
JSI
DFS
on
a
website
on
HTTP,
you
need
your
web
socket
port
to
be
also
wrapped
in
TLS,
and
that's
what
the
problem?
That's
the
problem.
It's
not
easy,
cuz
go
FS
does
not
support.
A
Http
does
not
support
like
WebSocket
over
TLS,
so
we
need
to
put
nginx
or
other
reverse
proxy
in
front
of
go
ideas
and
setup
certificates
there
and
sure
WebSockets
are
wrapped
in
TLS
and
all
and
then
like
multi
others
for
this
endpoint
will
be
different
like
unencrypted
is
/ws
encrypted
is
/w
s
s
stuff
like
that,
so
I
believe
that's
a
big
hole
in
our
documentation
that
we
don't
have
like
one
stop
for
that.
A
lot
of
people
would
love
to
run
their
own,
but
right
now
it's
really
really
tricky.
A
I
was
looking
at
the
ways
we
could
anything
we
could
link
or
anything.
I
could
like
him
extract
into
like
quick
tutorial
and
I
was
not
able
to
so
for
now.
I
just
ask
anyone
excited
about
this
or
who
already
said
this
up
on
their
own
to
either
comment
on
this
issue
or
post
like
if
there
are
tutorials
out
there.
A
A
Yeah,
like
I,
believe,
like
a
lot
of
people
in
community,
already
did
that
the
problem
is
that
you
dug
you
did
that
at
you
and
you
just
moved
on,
and
other
people
have
the
same
problem.
So
I
was
really
surprised.
We
did
not
recommend
this.
It's
something
I
want
to
improve
and
that's
it.
I
promise
to
shut
up
shut
up
now
and.
B
B
B
D
D
B
D
B
That
would
catch
the
same
issue
then
maybe
it
would
be
better
to
spend
the
time
in
that,
but
if
the
unit
tests
are
an
area
that
we
are
regularly
funny
get
issues
that
they
were,
these
tests
would
catch
like
looking
back
at
some
of
the
bugs
that
we've
discovered
or
shipped
or
found.
If
he,
if
he
unit
testing
like
they,
would
catch
those,
then-
and
it
seems
like
a
thing
worth
doing,.
D
Like
the
the
same
main,
I
mean
now
probably
wouldn't
have,
but
most
of
the
bugs
we
we've
had
in
the
past
are
like
really
stupid
things
just
like
a
myth
me
in
this
time,
like
I,
misspoke
ins,
something
like
that
word
in
the
code,
so
I
believe
it
will
work.
We
have
a
son
and
twin
tests,
but
they're
only
test
launch
process
I
like
to
test
more
of
the
web
UI
integration,
although
I
haven't
had
time
to
check
that
out
and
also-
and
it
is
relatable
to
the
next
max
point
in
the
agenda.
B
B
D
Hello,
hey
yeah,
we
can,
you
know
so
I,
don't
know
what
you
heard
so
I'm
going
through
it.
What
I
was
saying
I'm
just
concerned
about
the
size
of
the
box
folder
directory,
because
it
seems
that
if
I
continue
to
make
unit
tests
for
all
of
the
code
of
IPs
desktop
I,
don't
need
to
mark.
Almost
all
external
packages
will
require
and
a
lot
of
internal
modules
of
a
business
desktop.
B
That
I
mean
that's,
that's
always
the
the
risk
with
mocking
right
is
like
you
will
have
an
infinite
list
of
things
that
need
to
be
mocked
in
in
the
system.
Changes
and
you
have
to
rebuild
the
end
up
doing
this.
This
non
non
trivial
slice
of
work
is
permanently
managing
the
mocks
at
that
point,
so
I
think
what
right
before
you
cut
out.
B
You
were
talking
about
how
this
is
related
to
kind
of
the
build
issues
that
you're
talking
about
below
in
the
next
item,
and
so
I
think
you
know
you're
the
best
person
to
kind
of
make
no
decision
as
to
whether
or
not
the
work
doing
unit
tests
and
mocking
will
solve
or
catch
the
type
of
issues
that
come
up
first,
but
it
sounds
like
the
the
build
issues
are
probably
a
higher
priority.
Okay,
do
you
want.
D
To
skip
down
to
that
item,
boo-boos
yeah
sure
I
would
shut
my
screen.
I'm
Internet
is
not
evil.
People
was
doing
that.
So
I
asked
you
to
open
the
shoe.
Please
I
have
a
link.
Yeah
me
notes.
So
the
first
problem
is
touching
web
UI
like
right
now,
when
we
type
NPM
install,
we
fetch
the
web
UI
from
ipfs.
That
I
owe
the
problem
is
that
the
latest
web
UI
version
is
not
pin
to
the
Gateway,
but
it's
into
the
in
normal
synchronous
sentence.
D
It
would
work,
but
right
now
the
gate
we
can
is
having
some
connection
problems.
It's
Saturday,
so
the
builds
are
all
failing
on.
All
CI
is
because
of
timeouts,
and
there
are
two
switch
things.
I
was
told
we're
all
about
them.
He
suggested
to
make
whether
you
either
like
yesterday,
all
the
way,
through
his
version
and
like
if
I
stay
up
late
this
coming,
so
they
would
both
get
into
the
Gateway
and
the
other
option
would
be
to
add
a
module
to
act
as
a
repository,
so
good
build.
But
do
I.
D
D
So
also
for,
like
we
don't
use
a
web
UI
for
the
for
testing,
it's
only
to
add
to
the
final
binary
to
it
show
up
this
way,
so
it
couldn't
matter
much
but
on
say
it's
I
believe
into
the
burn
to
the
traffic
get
sub
module
because
right
now
quickly
get
is
a
bit
part
reliable,
then
target.
So
it
wouldn't
be
an
issue.
B
B
B
D
B
D
Yeah,
just
let's
move
for
the
second
topic.
Election
builder
is
counting
on
Linux
releases
on
Travis,
so
custom
builder
seem
stupid,
but
this
version
that
makes
it
feel
on
Travis
cinder,
opened
an
issue
under
the
repo
20
days
ago.
They
said
nothing
the
report
didn't
reply.
It
have
no
activity
on
the
repository
since
18
days
ago
or
something
I
could
downgrade
the
livestream
builder
version,
but
that
would
require
me
to
downgrade
election
version,
which
would
also
require
me
to
change,
make
something
and
reversing
changes
of
the
code
and
I
don't
know.
No.
D
This
is
only
happening
on
Travis
I
by
typing
a
by
running
this
one
with
machine.
It
works,
so
I
believe
it's
a
Travis
specific
problem
and
it
would
work
on
other
Sierra
like
circle
and
I
totally
on
Monday
about
this,
and
he
suggested
me
to
try
circles
in
sustain
our
support.
Windows
and
Mac.
Well,
I.
Think
that's
a
good
idea,
but
I'm
not
sure.
If
the
price
of
circle
see
I
would
be
worth
it
because
circle
needs
a
paid
plan
for
Mac,
OS
and
windows
and
I.
Don't
know
if
it'd
be
it
costs
like.
B
D
But
it's
only
testing
on
Linux,
okay,
it's
not
actually
my
fault,
it's
gateways
fault,
because
you
can't
download
what
UI
and
service
fail.
So
I
can
really
test
the
tests
and
the
build
process,
but
this
is
only
for
Linux.
I
I
need
to
be
able
to
run
this
on
Windows
in
my
quest
to
to
make
sure
it
will
work
closely.
I.
B
D
B
D
A
Instance
of
our
gateway.
We
have
like
multiple
instances
in
different
parts
of
the
world
and
that's
why
same
thing
works
for
one
person
but
does
not
work
for
Travis,
because
it's
using
like
effectively
a
different
machine
which
might
not
have
the
same
data
in
the
cache
and
stuff
like
that.
But
for
you,
specific
use
on
our
CI
is
that
we
actually
know
have,
and
we
have
like
direct
aliases
for
each
geographical
instance
of
our
gateway.
A
So
I'm
not
sure
if
it's
a
pattern
on
an
anti-pattern,
but
it's
a
pattern
to
just
try
them
in
order
and
basically
like
sidestep
the
Anna
cast.
So
instead
of
being
routed
to
that
one
specific
instance
of
our
gateway,
I'm
I,
wonder
if
it
would
be
benefit
solving
this
problem
to
just
try.
Okay,
the
u.s.
one
does
not
work.
Let's
fall
back
to
the
one
in
Frankfurt
I.
A
You
may
ask
only
about
like
details
how
to
request
specific,
like
request
data
through
our
gateway,
from
a
specific
like
geographical
instance,
I'm,
not
sure
if
it
will
help,
but
maybe
it
feels
like
something
worth
looking
at
before
we
go
into
the
rabbit
hole
of
moving
to
a
different
CI
infrastructure,
or
things
like
that.
Maybe
that
will
be
enough.
Yeah.
D
D
A
D
Not
not
being
able
to
start
if
it
says
desktop
on
Linux
and
the
betta
releases,
and
it's
my
fault,
because
I
built
the
binaries
on
a
virtual
machine
and
the
podium,
because
it
wasn't
working
on
Travis
and
they
I
should
have
said
likes
different
some
weird
option
for
sandboxing
on
Linux
that
I
didn't
know
about,
and
this
is
not
a
good
way
to
make,
for
this
is
definitely
also
on
Linux.
It's
not
all
distribution,
distribution,
twerkin
or
distributions
or
all
the
soft
environments,
and
look
that's
what
they're
called.
A
However,
the
newest
norm,
environment,
which
is
unlike
unity
in
in
Ubuntu
or
maybe
non-free,
either
way
some
of
the
latest
ones
just
removed
that
tray
section
and
that's
the
problem,
because
when
we
start
IP
first
desktop
the
user
interface
was
just
just
that
one
icon
in
the
tray,
and
now
when
there's
no
tray,
we
are
not
able
to
open
anything
I.
Believe
that's
one
of
the
problems
were
there.
A
Like
we
already
have
a
warning,
it's
just
a
matter
of
someone
spending
time
in
like
Ubuntu
and
finding
out
the
way
of
providing
alternative
user
interface
for
like
opening
the
main
menu,
because,
right
now
the
menu
is
opened
through,
like
the
tray
area
and
in
some
desktop
environments
on
Linux,
there's
no
tray
area
or
you
need
to
enable
it
and
most
of
people
don't
have
it
so
they
install
a
Kieffer's
desktop.
They
click
on
the
icon
and
nothing
happens.
I
might.
If.
B
A
A
D
A
And
I
generally,
like
maybe
other
note
that
we
would,
we
are
like
looking
for
Linux
maintainer
or
something
like
that,
because
that
effectively
it's
problematic,
it
is
much
easier
to
support
Mac
and
Windows,
because
those
are
just
two
desktop
environments
on
Linux.
You
have
like
multiple,
so
yeah
I
believe
we
have
with
lost
deitrick,
had
to
go
to
a
different
meeting
so.
B
A
Just
quickly
open
open
the
last
item
on
on
the
agenda
and
we've
wrapped
up,
there's
an
open
question:
should
we
exclude
gateway
paths
from
being
crawled
by
machines,
so
things
like
Googlebot,
which
is
indexing,
the
Internet
and
creating
Google
search
index,
there's
a
way
of
excluding
specific
paths
on
your
website.
I
open
them
pyaare.
A
Just
to
start
this
discussion.
Should
we
like
exclude
gateway
paths
from
being
crawled,
and
an
argument
for
that
is.
Those
results
are
already
dropped
from
the
Google
search
I,
believe
they,
those
were
sort
of
spanning
search
results
because
anti
Wikipedia
got
indexed.
So
you
know
that
were
real
typed.
There
was
a
page
on
ipfs
which
shown
on
the
federal
verse
the
first
page,
and
it
was
just
a
mirror
of
Wikipedia,
which
was
kind
of
duplicated,
duplicating
content
and
I
believe
Google
just
removed
all
all
those
duplicates
of
Wikipedia
mirrors.
A
So
when
the
crawler
is
browsing,
the
Wikipedia
mirror
on
ipfs
it's
effectively,
causing
our
gateways
to
load
content
and
put
it
in
the
cache.
Even
though
no
one
asked
for
this
content-
and
it
may
be
kinda
triggered
other
content
that
people
actually
cared
about
to
be
garbage
collected
because
it
was
like
older
and
stuff
like
that.
So
it's
like
an
open
question:
what's
the
value
in
being
crawled
by
those
search
indexes
and
other
automated
tools,
and
if
should
we
on
IPF
SEO,
should
we
disallow
those
paths
and
there's
a
separate
discussion?
A
Should
all
public
gateways
to
the
same
or
is
it
something
we'll
live
up
to
the
Gateway?
Should
it
be
a
part
of
the
spec
or
is
it
like
an
individual
decision
of
gateway,
maintainer
we've
run
out
of
time,
but
no,
but
it
was
called
cool.
Our
thanks
for
being
the
being
here
and
the
last
last
minute
items.