►
From YouTube: Revisiting Distributed Wikipedia and topics for Q4 - IPFS GUI and Browsers Weekly, 2019-08-28
Description
IPFS Mirror: https://ipfs.io/ipfs/bafybeib26xf6nqf5vrlyl6h6baxrtbqg7lp6etniubezaegmpa4oeumj3a/
A
A
This
also
translates
to
problems
are
for
public
gateway
operators
when
they
get
DMCA
requests
for
takedowns
of
specific
Wikipedia
pages
in
all
the
versions,
all
the
heavy
stuff,
so
I
started
looking
how
we
were
generating
those
mirrors
and
what
are
possible
paths
for
going
forward.
So
a
quick
background
on
how
we
were
generating
those
mirrors
is
that
there
are
snapshots
of
Wikipedia
created
in
as
in
format.
A
Zine
format
is
like
one
huge
file
created
specifically
for
offline
browsing
of
wiki
wiki,
not
on
the
Wikipedia,
but
any
media
wiki
compatible
wiki's
on
the
public
snapshots
available
for
download
provided
by
key
week's
project.
So
that's
some
the
base,
the
source
of
truth
to
that.
We
pick
the
years
back
for
that
and
we
added
a
thin
orchestration
layer
when
we
unpack
those
theme
files
and
put
those
unpacked
wiki's
on
ipfs,
with
small
changes
such
as,
like
fixing
links
making
sure
relative
linking
works
as
expected,
adding
a
footer
on
a
page.
A
A
Another
one
is
to
like
ensure
good
footer
is
updated,
and
things
like
that
so
I
believe
the
process
of
going
through
those
steps
would
be
like.
We
probably
want
to
do
that.
On
the
English
Wikipedia
manually,
we
will
drop
their
smaller
snapshot
by
let's
say
100
the
most
popular
pages
on
English
Wikipedia,
which
is
just
a
smaller
file,
and
you
can
like
run
your
scripts
on
that.
So
I
believe
that's
something
we
could
do
manually
and
then
we
would
move
to.
A
When
we
make
sure
the
manual
process
works,
we
should
figure
out
a
way
to
automate
the
process.
So
I
like
the
way
I
see
it
is.
There
should
be
automation
in
place
that
the
text
move
snapshots
like
make
sure
it's
not
too
often,
but
let's
say
once
a
month
or
once
a
week,
and
if
there's
a
new
snapshot
for
one
of
our
supported
wiki's,
it
should
automatically
build
it,
put
it
on
ipfs
pin
somewhere
so
there's
at
least
one
source
and
then
open
APR
in
the
suite.
A
A
Just
one
long
file
optimized
for
random
access
and
when
we
put
that
file
on
ipfs,
we
sort
of
split
that
file,
which
is
already
like,
has
internal
it's
already
optimized
for
random
access.
So
we
split
that
huge
file.
It
can
be
like
tens
of
gigabytes.
We
split
that
file,
put
a
build
balance
tree
and
put
it
on
ipfs
and
then,
when
someone
wants
to
access
specific
byte
range
from
that
file
that
needs
to
be
fed
from
ipfs.
We
have
API
for
accessing
specific
by
two
inches
of
files.
A
Putting
Wikipedia
on
ipfs
because,
right
now,
if
we
create
the
new
snapshot,
all
the
existing
images
honestly
like
most
of
them
did
not
change.
So
everyone
who
has
those
old
images
in
cache
they
automatically
are
co-hosting
those
files
for
new
and
future
snapshots.
So
that's
like
an
open
problem,
but
it's
interesting
if
we
are
able
to
put
zim
files
in
ipfs
and
if
it's
like
performant
enough,
what
if
we
have
xeam
reader
in
pure
JavaScript
that
maybe
not
requires,
maybe
it's
not
reading
over
HTTP,
but
what?
A
If
we
have
embedded
J's
ipfs
that
just
fetches
specific
byte
ranges
of
those
theme
files
from
my
peer
files
like
directly
so
that's
interesting
and
we've
got
some
feedback
from
the
qyx
project
that
they
have.
Actually,
they
have
JavaScript
client,
which
sort
of
works
in
browser
extension,
so
those
are
spaces.
We
probably
will
look
at
at
some
point
after
of
course,
like
that.
The
most
pressing
matter
is
too
just
refresh
those
snaps
that
snapshot
in
the
old
way.
A
However,
it's
pretty
interesting
if
we
are
able
to
create
a
Wikipedia
snapshot
which
there's
like
wicked
input,
JavaScript
base
reader
on
ipfs
and
put
zinc
piles
on
ipfs.
So
then
we
people
could
both
download
those
in
files
and
open
them
in
offline
Reader
on
their
machine.
However,
they
actually,
if
we
have
this
ipfs,
capable
team
reader,
publishing
ipfs,
we
you
actually
just
need
web
browser
and
then
that
web
browser
would
be
able
to
just
fetch
specific
zip
file
from
ipfs
specific
parts
of
the
same
file.
A
A
Those
files
do
not
update
that
often
I've
seen
snapshots
been
done
like
monthly
or
sometimes
once
a
quarter
or
something
like
that,
so
those
are
fairly
stable
on
the
network.
When
you,
you
have
multiple
people
sitting
them.
It's
just
an
interesting
change
of
approach
to
look
at
or
to
experiment
with
that's
the
update
on
Wikipedia
I'm,
not
sure
I
probably
skipped
some
details,
but
if
you
have
any
questions,
yeah.
B
I
looked
at
em,
it
looks
like
these
snapshots
are
actually
not
really
generated
very
often
like
even
the
newer
snapshots
are
pretty
old.
So,
like
I,
wonder
if
there's
some
way
that
we
could
can
we
support
key
weeks
like
who
do
we
need
to
create
our
own
snapshots?
Can
we
use
their
infrastructure?
Could
we
do
something
they
actually
create
more
updated
snapshots,
very
good?
Really,
yes,
so
that's
sort
of
like
a
snapshot
of
day
would
be
really
huge
improvement
to
what
they
have
going
on
there
yeah.
A
B
Yeah
I
mean:
can
we
give
them
a
grant
or
something
like?
Can
we,
like
it,
I'd
love
to
explore,
maybe
finding
ways
that
instead
of
taking
people
like
you
off
the
core
implementation,
how
can
we
help
accelerate
things?
I
love
the
idea
of
just
automating,
it
too
and
I
think
there's
probably
a
lot
of
other
people
that
would
love
to
be
able
to
see
this
working
there's
a
way
we
can
move
the
bigger
whole
space
forward.
B
A
A
The
first
step
would
be
to
just
add
the
same
data
to
ipfs
like
go
a
gif
as
supports
above
file
store
when
you
can
add
a
pointer
to
existing
file
without
duplicating
data
in
your
repo
or
I
believe
it
was
added
for
Internet
Archive
or
some
other
partner,
the
URL
store,
when
you
add
URL,
and
it's
sort
of
like
fetch.
The
ranges
are
fetched
on
demand
when
someone
requests
requests
those
specific
ranges
so
putting
sim
files
on
a
TFS
and
like
exposing
ipfs
links.
Next
to
me,
torrent.
A
That
would
be
probably
the
first
step
and
we
totally
should
support
them
in
that
and
see
if
this
experimentation
with
like
static,
who
jeaious
based
only
that
which
is
capable
of
fetching
data
from
my
PFS.
It's
interesting
because
when
sim
files
are
already
on
ipfs,
that
would
be
like
the
obvious
next
step
to
check
yeah.
So
that's
more
or
less
an
update
I'll
be
trying
to
support.
A
B
A
B
It
seems
eminently
within
our
our
capabilities
and
reach
and
to
cab
the
set
up
and
I
think
that
would
be
something
that
is.
It
would
be
such
a
powerful
statement
around.
Both
the
values
of
our
organization
and
of
the
capabilities
of
our
technology
is,
if
we
had
this
automated
and
setup,
and
and
was
supporting,
making
Wikipedia
available
and
less
prone
to
this
type
of
attack,
which
is
probably
only
happened
more
and
more.
A
A
That's
more
that's
a
reason
when
I
sort
of
jumped
in
on
the
Wikipedia
cuz,
it's
honestly
the
best
test
case.
I
have
right
now
when
I
want
to
test
browsing
websites
in
Drey
or
IP,
has
companion
or
things
like
that,
because
it
has
everything
it
there's
a
DNS
link.
It
has
it's
so
big
that
it
does
not
fit
in
a
single
directory,
so
you
have
ham
charting.
A
B
A
Yeah,
probably
it's
neverending
story,
however,
like
the
okiya
Kentucky
are
to
have
to
brave
browsers,
to
see
each
other
and
be
able
to
exchange
files
that
should
land
like
local
discovery
works
right
now
down.
The
missing
piece
is
like
expose
port
and
like
announce
port
for
others,
but
I
have
an
idea,
Alan
experimented
with
that
around
Leedy
web
project
last
year,
and
he
stumbled
on
the
similar
block,
I'm
struggling
right
now
so
I'll
probably
reuse.
A
B
Mean
just
a
it
would
that
be
something
like
a
in
a
J,
asleep
PDP
quarter
like
what,
if
we
spent
a
quarter
just
focusing
on
those
on
those
two
issues,
or
even
just
one
of
those
issues,
because
some
DHT
and
sunsetting
web
second
star
like
is
it
been
things
that
have
been
around
for
a
long
time
like
it
wouldn't
even
be
possible
to
say,
let's
dedicate
a
quarter
to
that
and
and
actually
get
it
done.
Yeah.
A
Yeah
I
I
don't
see
any
other
way.
We
need
to
tackle
it
either
right.
Another
way
would
be
to
just
like
figure
out.
Are
there
other
ways
of
doing
content
and
pure
discovery
in
web
browser?
Are
there
existing
like
services
or
standards
or
api's
available
for
web
browsers
that
we
did
not
think
about
that?
We
could
create
a
little
bit
oopy
discovery
module
for
maybe
perhaps.
B
A
B
A
C
A
B
Yeah
I
already
experimenting
be
interesting
to
look
at
what
the
use
cases
are.
That
would
drive
those
types
of
discovery.
Mechanisms
like
Hong
Kong
is
the
great
example
concerts
like
places
where
you
have
a
lot
of
people
in
one
place,
really
dense
urban
areas
where
that
type
of
physical
ago
signaling
and
discovery
would
be
yeah
really
interesting.
We've.
B
A
A
So
so
that's
there
are
two
pieces
of
that
puzzle.
One
is
the
situation
at
the
gateway
and
people
using
our
like
our
gateway
on
ipfs
io.
Another
thing
is
generally
like
the
way
embedded,
J's
ipfs
on
regular
website.
Does
this
discovery
is
limited
to
honestly,
just
like
WebSockets,
tar
and
being
connected
to
bootstrap
nodes
in
the
past.
It
works
perfectly
fine,
because
those
good
strap
notes,
wear
the
same.
Go
ipfs
instances
that
provide
that
HTTP
responses
on
our
public
gateway.
A
So
actually
every
JSA
PFS
was
connected
to
the
Gateway,
so
the
Gateway
immediately
had
a
connection
to
the
nodes
that
had
data,
so
it
worked,
but
that
is
not
scale.
So
we
detached
bootstrap
notes
from
our
like
gateway
cluster
and
then
it's
not
like
one-to-one.
It's
like
yeah
one
hopped
and
that's
the
problem.
We
just
asked
gears
that
we
are
connected
directly
and
if
there
are
connected
to
appear
that
has
data
yeah.
B
But
I
mean
I
think
that's
a
at
that
point.
You're
talking
about
DHT
algorithm
or
like
we're
not
like.
That's
a
that's
either.
If
that
is
not
a
solvable
problem,
then
then
let
p2p
as
a
whole
is
not
tenable
as
a
solution
to
this
at
all.
Right
like
that,
it
is
designed
for
exactly
that.
One
thing
that
you
just
described
know
is
its
low
purpose
of
existing.
Yes,.
A
A
B
So
turn
to
reel
it
back
in
what
is
the?
What
would
it?
What
is?
It
would
be
a
a
big,
a
big
chunk.
I'd,
you
know,
we've
had
the
last
couple
of
quarters
since
I've
been
here.
I
feel
like
we've
got
a
few
different
things.
We've
got
our
hands
this
and
this,
and
this
business
and
I
would
love
to
be
able
to
maybe
instead
collapse
and
coalesce
our
attention
into
whatever.
We
think
that
biggest
blocker
is
and
take
a
quarter,
push
on
that
and
remove
some
of
these.
B
You
know
long
long,
standing
things
that
that
block
our
progress,
we're
gonna
have
there
within
web
browsers
on
site.
We
can
half
that
let's
have
a
plan
in
hand
better
idea,
I
think
that,
from
what
the
project
operations
meeting
we
talked
about,
the
Oh
care
is
next
Dorothy.
Ours,
we're
gonna
start
doing
next
week
or
the
week
after
next
I
think
week
after
next
is
when
we're
in
the
same
room
together,
yeah.
B
Yeah
and
and
that's
just
browser
land
and
I,
don't
have
the
right
people
here.
We
don't
have
a
hawk
here
to
talk
about
desktop
stuff,
but
from
the
from
the
from
the
desktop
and
GUI
it
around
ensuring
stability
and
performance
and
keeping
up
to
date
with
the
new
release
process
and
making
sure
that
everything
is
still
compatible
working
with
each
new
releases.
Now
that
Rana
cadence
released
becomes
more
important
to
be
able
to
have
things
like
regression,
detection,
so
I
think
like
what
am
I
wonder.
B
My
suggestions
there
in
that
area
was
gonna,
be
around
looking
at
the
test,
matrix
again
figuring
out
what
the
best
high
value
regression,
detection
and
CI
combo
that
we
could
have
to
be
able
to
make
sure
that,
even
if
we're
not
actively
building
a
bunch
of
stuff
into
desktop
and
GUI
that
it's
tested
daily
against
JSF
EFS
and
go
ipfs
and
to
make
sure
that
everything
is
still
functional.
Even
if
we're
down
to
texting,
detecting
our
regressions,
we'd
be
detecting.
There's
sorry
I
said.
A
D
B
A
A
B
I,
in
fact,
I
think:
let's
do
that
before
at
all
I'll
put
30
minutes
on
the
calendar
and
let's
go
over
those
things
and
do
that
triage
before
we
get
together
yeah,
then
we
get
together,
we
can
be
as
we
can.
We
can
do
the
things
instead
of
talking
about
doing
things
cool
thanks
for
creating
this
board
I'm
still
like
github
project
boards.
It's
like
they're,
not
as
easy
as
gel
boards,
but
but
they
are
also
not
another.
Senator.
A
A
B
Yeah,
but
I
also
do
like
the
fact
that
this
has
like
a
this
is
built
into
the
place
that
we
already
work.
I
did
I
was
looking
at
some
bots
that
do
automation
around
these
types
of
things
too,
so
you
can
pre-populate
columns
with
issues
based
on
so
when
you're
managing
different
issues,
the
columns
autofill.
B
B
Okay,
so
it
sounds
that's
that's.
An
interesting
idea
is
to
spend
a
quarter,
especially
since
we
have
lab
week
during
the
quarter
and
2020
planning
during
the
quarter,
maybe
maybe
now
to
say
better
approaches
to
list
all
of
those
kind
of
stability
and
testing
infrastructure
requirements
that
we
would
always
love
to
have.
But
there's
never
a
good
time
to
do
the
house
also
the
holidays
and
q4.
B
D
B
B
A
Honestly,
we
already
are
using
Wikipedia
puffs
in
our
tests,
all
right
honestly
yeah
for
testing
things
like
like
special
characters
in
paths
and
the
way
you
escape
those
okay,
because,
like
I
thief
as
files
are
not
like
URL
paths,
but
we
sort
of
started
to
be
using
similar
conventions.
So
that's
interesting
problem
space,
so
yeah
Wikipedia
is
a
part
of
our
test
suit.
Already.
C
Yes,
we
do
there's
a
guy.
Thank
you.
Person
I
have
lined
up
at
the
moment,
is
a
guy
who's
focused
on
like
what
the
internet
limitations
are
like
four
people
on
native
lands
in
the
States,
so
people
so
that's
I-
think
maybe
next
Wednesday
or
something
like
that,
but
yeah
and
then
on
the
protocol
side.
We
just
deal
go
left
about
a
couple
weeks
ago
and
now
there's
a
guy
named
Jill
who
joined
us
who's
also
at
Moxie.
He
was
working
on
the
IBM
project
on
the
front
end
of
it,
so
not
I.
C
So
it's
part
inventory
and
in
part,
doing
the
work
to
make
it
happen.
So
at
the
moment,
I'm
working
on
the
inventory
side
of
it
but
I
definitely
see
some
fast
connections.
One
of
the
things
that
I
think
would
be
cool
would
be
to
use
that
this
CID
inspector
thing
that
somebody
built
I,
don't
know
if
it
was
Allen
who
built
it
or
just
showed
it.
But
right
now
you
can
sometimes
when
the
results
is
a
dag
thing.
You
can
click
a
button
that
says
for
you
and
I
failed.
C
B
C
B
B
So
I
just
shared
a
one
of
the
one
of
the
carrots
that
we
had.
The
project
Operations
Group
was
too
was
like
communications
plan
and
be
able
to
share
out
cam
content
and
Jonathan
Victor
spent
a
couple
days
helping
out
putting
together
this
master
list
of
just
it's
a
sheet
of
just
everything
that
happened
to
camp
and
links
to
the
things
and
then
links
to
whether
or
not
somebody
like
to
already
did
a
blog
post
about
it
or
already
shared
something
about
it
or
and
then
Jonathan
is
actually
going
to
move.
B
All
of
this,
we
talked
with
Zack
yesterday
and
Zack
has
an
air
table
of
all
the
video
stuff,
and
we
are
talking
about
using
that
as
the
single
source
of
truth
for
all
of
this
content,
so
Jonathan's
gonna
migrate,
the
contents
of
the
sheet
into
Zacks
air
table
and
then
insects
air
table.
You
will
have
a
single
database
that
has
all
of
the
stuff,
not
just
camp
related
but
being
able
to
search
by
topic,
be
able
to
search
for
a
keyword
you
want
to
find
all
CID
related
stuff
be
able
to
find
links
to
it.
C
I'm,
like
I,
don't
see,
I
clearly
have
an
idiot
yeah
so
and
then
we'll
need
to
add
the
video
thanks,
but
the
other
place
that,
like
one
of
the
things
that
we
can
do
very
easily,
is
now
that
we
have
those
resources
pages
at
the
end
of
each
tutorial
stuff,
that's
related,
but
either
we
haven't
had
time
to
shove
it
into
proto,
school
or
never
format
correctly.
We
can
be
leaking
out
to
those
resources
more.