►
Description
This week we learn whatโs new with Qri.io in 2020
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
Alright,
okay,
here
we
are
welcome
to
the
ipfs
weekly
call.
It
is
the
second
of
March
welcome
all
of
you.
Everyone
here
this
is
gonna,
be
fun
times.
Today
we
have
Brendon
and
Dustin
looking
to
us
about
what's
new
in
query
in
2020,
Brendon
and
just
Dustin,
would
you
like
to
take
it
away
or
just
Dustin
or
just
Brendan?
How
do
you
want
to
do
it?
I.
B
C
Thanks
so
much
for
inviting
us,
this
is
so
nice,
it's
so
nice
to
see
everybody
I'm
gonna
share
my
screen
and
solve
this
goes
ok,
but
if
it
doesn't,
you
know
demo
gods
but
yeah.
So
query
in
2020,
we've
spoken
on
the
offense
All
Hands
call
in
the
past
I'm,
so
I
thought
I'd
give
a
quick
catch-up
for
those
who
are
who
have
already
seen
it.
But
if
you
haven't
seen
as
well,
we
do
before
feel
free
to
ask
questions
to
sort
of
fill
in
any
stuff
that
we've
missed.
But
generally
we
are
a
query.
C
We
do.
Data
set
version
control
and
are
sort
of
big
mission
in
the
universe
is
to
make
data
work
open
source.
Hopefully,
I
didn't
ask
everyone
see
my
screen.
Okay,
that
work.
Hopefully
you
can
see
this
thumbs
up:
okay,
cool
yeah
and
so
by
making
didn't
work
like
open
source.
We've
been
on
as
sort
of
like
multi-year
mission
to
sort
of
bring
the
idea
of
data
set
version
control
a
set
of
tools
that
we
think
it
deserves
the
biggest
thing
that
we're
sort
of
hoping
to,
and
that's
like
in
the
world
of
software
engineering.
C
We
have
git
and
github,
and
we
have
really
great
tools
for
collaborating,
and
that
makes
a
sort
of
whole
ecosystem
of
software
development
possible
that
it's
just
kind
of
not
really
the
case.
When
you
look
at
the
world
of
open
data
these
days,
that's
the
way
we
see
open
dinner
right
now
is
very
open.
Data
portal
based
people
are
really
a
sort
of
publishing.
C
Only
really
large
institutions
are
able
to
sort
of
maintain
sort
of
vibrant,
open
data
set
sort
of
you
guys,
that's
a
very
one-way
conversation
and
we're
hoping
to
make
it
a
2,
a
1,
and
since
we
last
talked,
we've
done
a
bunch
of
stuff
and
been
positive
I'm
here
as
a
bullet
list
of
some
positive
things,
we've
done,
we
published
a
new
tool
called
query
desktop,
which
is
an
electron
app
that
runs
on
top
of
our
query.
Cli
I'll
be
showing
you
a
little
bit
of
that
later.
C
We've
actually
started
to
grow,
we've
got
real
users
and
in
many
ways
for
us,
this
sort
of
reflects
a
change
from
being
a
sort
of
R&D
project
to
something
that
is
actually
being
used
in
the
real
world
down
which
comes
with
it.
A
whole
bunch
of
things
and
the
a
lot
of
that
growth
has
been
in
the
last
sort
of
like
3
or
4
months,
where
we're
seeing
lots
of
people
start
to
show
up
and
actually
use
query
for
data
set
version
control.
C
So
the
idea
is
starting
to
sort
of
catch
on,
which
is
very
exciting
to
say,
and
we
also
launched
query
cloud
which
is
sort
of
our
HTTP
bridging
service
that
lets
you
look
at
data
sets
that
are
published
on
query
on
the
regular
internet
without
any
sort
of
like
do.
You
have
magic
and
we
brought
a
lot
of
stability
and
sort
of
like
really
sort
of
brought
forth.
C
The
features
that
we
think
we
need
to
sort
of
get
into
a
system
where
people
can
actually
start
versioning
and
publishing
data,
and
so
just
to
like
give
you
a
feel
for
what
that
looks
like
materially.
This
is
what
core
desktop
looks
like
these
days.
We've
got
this
sort
of
like
built-in
network
of
data
that
already
exists
and
I
can
sort
of.
These
are
data
sets
that
are
currently
published
on
query
and
all
of
this
obviously
we're
on
a
call.
C
All
of
this
everything
you're
looking
at
is
versioned
ipfs
data,
so
I
can
sort
of
just
come
in
and
look
at
this
data
set
of
presidents
that
are
a
birth
and
death
and
we're
seeing
just
like
a
list
of
commits
I
can
see
them
grayed
out
because
I
don't
have
them,
they
Lily
had
this
clone
button.
This
is
gonna
grab
this
dataset
pull
it
into
my
local
repository
give
me
a
whole
version
history
and
let
me
sort
of
see
everything,
yeah
and
so
like
this
gives
us
like
a
meaningful.
C
We
get
like
our
classic
sort
of
like
github,
desktop
Styles,
just
yellow
dot
store.
What
change
between
versions
we'll
talk
more
about
how
this
is,
but
you
sort
of
see
everything
we
need
to
see
to
sort
of
get
through
version.
Historical
data
get
stuff
over
time.
I
can
pull
up
a
collection
of
it
up.
Creating
data
is
sort
of
almost
a
straight
forward:
I'm
gonna
drag
a
CSV
file
really
quickly
over
a
desktop
application.
This
is
the
process
of
creating
it.
Is
that
there's
a
new
dataset
cool
I
can
hit
publish.
C
C
Engineering
data
and
I
think
is
really
fun
and
so
like
this
is
translation
transitioning
into
stuff
that
we
think
is
kind
of
useful
for
this
people
in
this
call
just
you
a
real-world
use
case
of
how
this
might
work
if
you
are
in
the
engineering
space.
So
for
us
we
had
someone
file
an
issue
a
little
while
ago.
Talking
about
how
long
it
takes
to
add
a
really
large
data
sets
aquarium,
so
we're
sort
of
dealing
with
some
of
our
performance
issues,
and
so
when
people
start
yelling
you
about
your
refer
performance.
C
This
is
really
exciting
and
sort
of
made
for
fun
use
cases
to
demonstrate.
Today
we
took
this
whole
concept
and
then
sort
of
like
rift
on
what
s
command
the
user
sort
of
posted
and
turn
this
into
something
that
runs
in
part
as
part
of
our
continuous
integration
framework,
and
so
now
this
data
set
that
you're
looking
at
it's.
Actually,
the
result
of
I
can
just
cut
to
the
chase.
This
is
like
the
actual
data
that's
coming
out
of
it
and
we
can
see
I
have
that
actually
perhaps
yeah
and
so
then,
as
I
spend.
C
All
of
this
is
actually
posted
in
hosting
ipfs.
So
this
is
a
visualization
that
is
built
into
the
dataset
itself
of
the
current
amount
of
time
that
it's
taking
and
every
time
we
push
to
master.
This
data
set
is
updated
in
CI,
so
it
runs
to
performance
benchmarks,
outputs,
a
in
Arcis,
in
this
case
the
CSV
file
commits
that
to
query,
publishes
that
back
and
updates
this
visualization,
and
so
every
time
we
push
the
master.
We
get
this
really
lovely
sort
of
like
hey.
C
How
well
are
we
doing
in
terms
of
that
sort
of
time,
so
this
is
just
an
example
of
what
it
looks
like
in
CI.
I
can
link
to
any
of
this
if
anyone
has
any
details
or
wants
to
see
any
details
on
this.
On
top
of
all
that,
all
of
this
actually
shows
up
in
Google
dataset
search
because
we've
made
this
sort
of
like
cloud
thing
work
properly.
So
we
search
for
query
benchmarks.
We
actually
see
it's
benchmark.
Data
set
is
being
linked
inside
of
query
pod
itself.
C
C
We
sort
of
like
think
that
that's
sort
of,
hopefully
that
sort
of
forms
like
a
concrete
example
of
how
this
kind
of
thing
is
useful
and
where
we
sort
of
see
this
sort
of
like
intersection
between
sort
of
the
hard
engineering
and
why
we
think
sort
of
this
advertising
is
a
valuable
thing.
That
was
a
lot
of
really
fast
talking
that
took
way
less
time
than
expected,
but
just
know
I'm
not
gonna.
Stop
sharing!
Sorry,
once
again,
let
me
go
back,
but
yeah,
just
a
sort
of
closed
houses
Tom
this
sort
of
lets.
C
You
run
your
own
data
portal,
which
we
think
is
really
exciting
and
really
sort
of.
Like
brings
that
conversation
from
just
a
one-way
thing,
to
a
two-way
thing,
in
a
way
that
it's
drag-and-drop
level
easy,
and
if
you
want
to
get
really
hardcore,
you
can
script
it
all
the
way
up,
as
we
sort
of
showed
you
all,
this
is
actually
on
ipfs
and
then
just
sort
of
like
close
out
some
of
this
like.
Where
are
we
headed
this
year
this
year?
C
It
sort
of
marks
for
us
sort
of
just
bringing
more
more
features
towards
this
to
use
because
those
are
sort
of
starting
to
grow.
The
big
thing
we're
gonna
bring
forth
is
a
different
UI,
so
you
can
now
actually
see
the
diff
between
each
version
right
inside
of
the
desktop
itself.
That'll
be
the
next
thing
shift
on
the
sides
on
speaking
to
features
that
are
more
interesting
to
this
crowd.
We
have
the
concept
of
remotes,
which
exist
inside
of
it,
but
in
query
remotes,
are
it
just
ipfs
appears?
C
C
The
other
thing
that
we're
sort
of
like
really
excited
about
is
under
the
hood.
All
of
these
data
sets
are
interact
in
a
way
that
we
can
actually
tell
you
when
your
datasets
are
out
of
date,
so
if
one
dataset
is
combining
to
other
ones,
we
can
actually
tell
you
hey
this
this
to
downstream
did
upstream
days,
that's
from
you
have
been
updated.
C
You
should
rerun
all
of
your
sort
of
scripts
and
get
get
the
latest
version
of
your
stuff
and
the
last
bit
that
we're
sort
of
excited
to
bring
to
on
this
call
and
that
we've
been
really
sort
of
like
what
would
you
say,
like
longtime
reader,
first-time
poster
about
his
been
ipfs
test
ground,
which
we've
been
following
for
a
lot
of
while
for
a
long
while
and
we're
actually
now
starting
to
use
internally
a
query
and
we're
big
big
big
fans
and
test
grounds.
We
plan
to
use
it
a
whole
bunch.
I
was
really
excited.
C
The
first
two
K's
foreplane
be
used
for
is
like
checking
for
read/write
conditions,
risking
race
enos
that
we
think
exists
within
our
app,
but
we
haven't
been
able
to
verify
it
because
created
conditions
where
we
think
that
racing
this
shows
up
is
quite
difficult
and
then
from
there
we're
gonna
move
on
to
more
sort
of
like
but
performance
benchmark,
oriented
stuff
in
the
peer
to
peer
context.
Hopefully
that
answers
everything
or
actually
gets
to
start
a
conversation
started,
but
we'd
be
happy
to
take
any
questions.
A
C
100%
yeah
yeah
I,
hopefully
I
answered
that
Midway
to
beat
but
yeah
hey
nice
to
see
you
and
absolutely
I
was
really
inspired
actually
by
I've,
never
met
Dirk
but
Dirk,
who
has
been
working
on
the
bitts
walk
performance
stuff.
We've
been
looking
at
the
code
actually
output
by
the
bit,
swap
tests
themselves
and
seeing
if
we
can't
turn
that
into
a
data
set,
it's
a
really
extent
of
being
able
to
sort
of
just
refer
to
specific.
It
commit
run
a
whole
test,
grand
suite
and
say
from
commit
to
commit
performance
really
exciting.
C
D
A
It's
it's
I
think
like
it's
safe
to
say
that
everyone
who
is
working
on
test
ground,
the
soup
stoked
that
you
guys
are
starting
to
to
use
it.
It
is
it's
something
that
has
been
in
development
really
heavily
recently,
and
so
it's
very
new
and
there
be
dragons
in
there
so
watch
out
for
them,
but
ya
know
we're
really
excited
that
for
you
to
be
picking
up
as
well.
This
that's
that's
really
good
news
and
we're
really
happy.
That's
that's
happening
so
so,
yeah,
okay,
so
another
question
so
from
Matt
uber.
A
C
B
This
is
let
me
I've
been
looking
at
recently
just
for
kind
of
unrelated
reasons,
but
there
is
a
lot
of
stuff
that
we
were
doing.
That's
just
inefficient,
but
also
you
know,
saving
block
to
my
PFS
can
be
slow.
We're
not
we're
not
really
taking
advantage
of
a
lot
of
a
synchronous
work
that
we
could
be
doing
we're
like
I'm
of
the
opinion.
B
We
should
just
start
saving
and
then
just
like
you
know,
have
our
UI
show
stuff,
and
then
you
know
it
finishes
in
the
background
or
whatever
you
know
like
we're
waiting
for
things,
we
don't
need
to
be
waiting
on
right
now,
yeah
a
lot
of
it.
It's
just
like
we
also
haven't
prioritized
to
work
on
making
some
of
the
large
dataset
processing,
especially
fast,
so
yeah.
There
are
some
theories,
there's
some
ideas,
but
nothing
concrete.
We
can
identify
just
at
the
moment.
B
A
C
Thank
you
for
your
question.
We
have
very
much.
This
is
an
integration,
question
and
I.
Think
it
sort
of
ties
into
tibeats,
question
below
which
is
gonna
plug
versioning
into
any
other
app
integrations,
are
a
massive
part
of
what
we
do
like
query
on
its
own
is
really
just
the
burgeoning
bit
it's
like.
C
So
as
for
making
a
tableau
connector-
yes,
currently
it's
in
its
a
sitting
below
the
Postgres
integration,
connector
and
below
the
sort
of
Excel
export
connector
integration
stuff
that
a
lot
of
people
been
asking
for,
but
as
of
right
now,
you
can
do
a
lot
of
things
that
will
get
you
there
really
quickly
like
where
you
can
take
any
kind
of
data
with
whether
it
be
JSON
or
seaboard
or
anything.
It
supports
and
spit
it
out
as
a
CSV,
which
usually
makes
for
a
nice
fast
sort
of
like
importing
into
other
mechanisms.
C
C
As
for
like
native
support
of
for
query
and
other
stuff,
not
so
much
the
first
thing
that
we're
really
excited
to
do,
there
is
actually
the
Apache
arrow
project
which
is
scheduled
for
later
this
year.
But
Apache
arrow
is
really
comports
well,
with
the
kind
of
technology
we've
been
working
on
and
we've
set
our
two
roadmaps
on
a
sort
of
collision
course
by
using
a
lot
of
the
same
technologies
under
the
hood.
B
Also
I
want
to
add
something
there.
The
the
default
way
that
query
works
is
that
it's
very
much
like
a
database
where
everything
is
back.
You
know
stored
in
my
PFS
and
it's
kind
of
like
an
opaque
system
where
the
user
can't
see
the
data,
we're
storing
it
and
thinking
about
it
internally,
but
we
also
have
a
mode
where
you
can
check
out
files
in
a
kid
like
fashion.
B
So
what
what
you
normally
think
of,
like
the
data
set
body,
can,
you
know,
just
be
the
file
in
your
file
system
and
that's
kind
of
like
our
easier
version
of
integration.
You
know
you
can.
You
know,
move
a
check
that
file
into
excel
or
whatever
I'm
not
familiar
with
tableau
WDC,
but
that's
something
a
look
into
but
yeah.
The
idea
is
like
we're
basically
watching
files
and
as
they
change
we
can.
A
Thanks,
thank
you
Dawid.
Does
that
answer
your
question
as
well?
Weird
yeah,
yeah,
okay,
all
right
Brad!
There
are
no
more
questions
in
the
chat,
but
I
have
a
question.
Desktop
looks
slick.
Did
its
March
now,
have
you
just
been
grinding
on
that
for
two
months,
or
has
it
been
in
development
for
a
lot
longer
and.
E
Sure
versions
of
the
best
of
the
desktop
app
happens.
It
stood
for
a
while
I
think
in
October
November
we
decided
we
wanted
to
go
a
sort
of
coalesce,
some
ideas
that
we
had
and
came
up
with
a
new
version
of
the
app
and
then
in
the
past
two
months,
I
think
was
the
redesign
that
Brendan
Brendan
is
responsible
for
it.
Looking
so
good
I'm
responsible
for
it
functioning
red
to
dry,
it
looks
so
good
but
yeah.
Thank
you.
A
F
A
G
C
It's
good
to
see
you
a
great
question:
yeah.
No,
we
are
not
using
em
FS.
We
are
actually
still
writing
an
IPF,
sv1
Dex
and
that's
for
a
number
of
reasons.
We've
had
it.
We
have
a
whole
conversation
about
links
and
IP
LD
to
be
had
that
it's
a
long
conversation
but
so
yeah.
Right
now
we
we
have
a
sort
of
like
hacked
management
system
that
actually
creates
the
deck
as
the
deck
is
being
created.
C
We're
pulling
the
hashes
of
the
sub
files,
then
like
pulling
them
up
into
actually
do
this
a
definition
itself
and
then
closing
it,
and
so
it's
sort
of
like
a
purpose-built
thing.
It's
using
a
little
regular,
chunker
and
stuff,
but
not
using
em
FS,
some
of
those
right
now.
Sorry,
it's
a
sort
of
bespoke
versioning.
G
A
B
Been
a
road
map
for
a
while,
just
like
we
haven't
be
able
to
prioritize.
There
is
a
bit
when
we
were
starting
to
make
the
move,
but
then
just
other
things
came
up
with,
because
you
know
we've
got
a
bunch
of
pieces
coming
down
here
and
are
in
our
kind
of
environment.
Our
platform,
like
we're
kind
of
doing
things
that
are
manually
imitating
out
of
FS,
is
like
you
don't
working
yeah
like
we're
doing
things
the
hard
way.
B
I
guess
you
could
say
nothing
has
happened
yet
where
we've
been
like
forced
to
do
it
so
I
think
that's
that's
because
the
thing
we're
just
kind
of
kicking
the
can
down
the
road
for
now,
but
we
do
play
to
do
it
once
you
know
it
seems
appropriate.
I,
don't
know
right.
Did
you
get
bored
details
I'm?
What
your
thoughts
are
here,
yeah.
C
There's
one
bit
that
I'd
like
to
service
here,
it's
go
and
I
think
it's
a
really
interesting
question
for
this
group.
We
like
UNIX,
FS
v1
decks
because
we
can
put
a
file
in
there
called
index.html
and
when
you
visit
it
that,
from
a
from
the
gateways,
you
see
the
visualization
right.
So
we
were
demoing
that
today
from
the
knife
LD
perspective,
we
don't
get
that
right.
C
C
F
Let
me
take
that
Hadron,
so,
as
you
do
so,
the
long-term
answer
for
this
is
that
an
IP
of
the
ADL
will
contain
essentially
a
rendering
component
for
the
web
as
well.
So
you
will
say
this
stuff
when
you
plug
it
into
an
IP
into
and
something
that
then
in
space
as
a
gateway.
You
also
render
this
other
stuff
with
it
based
on
you
know,
selection
and
stuff
like
that
inside.
F
C
Into
a
composition,
there's
there's
lots
to
be
explored
here,
and
maybe
we
should
sort
of
make
a
point
of
opening
that
conversation
up
more
because
towards
the
end
of
this
year,
will
actually
be
the
time
where
we
start
to
ask
these
questions
ourselves
where
we're
walking
from
a
functionality
perspective
is
like.
We
need
the
notion
of
a
link
inside
of
query.
Many
of
our
users
are
asking
for
attachments
right,
like
can
I,
add
arbitrary
file
data
and
to
query
data
set
and
then
right
now
the
answer
is
no.
C
Currently
we
have
no
way
of
saying
that
and
the
answer
that
is
obviously
I,
peeled
D,
but
right
now
we
we
kind
of
can't
we're
it's
not
that
we
can't
do
it.
It's
just.
We
want
to
architect
that
carefully,
we
link
resolution
for
us
it's
a
very
important
sort
of
like
performance
characteristic
and
so
there's
a
whole
bunch
of
stuff
to
talk
about
there,
which
could
be
really
fun
to
read.
Restart
that
conversation
absolutely.
F
Just
one
more
thing
about
what
they
said:
there
definitely
will
be
some
sort
of
like
middle
middle
ground
or
how
to
get
from
point
A
to
B
is
not
going
to
be
like
you
have
to
wait
for
this
thing.
You
know
that
allows
for
pluggable
HTTP
gateways.
They
is
going
to
be
something
in
between.
We
just
don't
know.
D
What
it
is
here,
also
one
one
middle
ground
would
actually
just
be
to
use
mesh
data.
No,
it's
like
when
you
finally
get
units
best
to
we
will
have
like
arbitrary,
just
link
to
anything
so
then,
like
the
main
link
could
be
linked
to
like
the
some
root
node
bit
points
to
both
of
an
HTML
file
for
like
to
display
the
data
and
then
also
points
to
the
actual
data
set.
So
he
was
trying
to
kill
him.
D
C
Would
be
delightful,
we've
had
a
number
of
users
ask
for,
like
basically
query
j/s,
where
you
have
the
same
thing:
it's
like
window
down
ipfs,
where
you
can
sort
of
interact
with
and
get
a
query
like
system.
We
have
an
api
for
interacting
with
data
sets,
and
this
has
been
a
that
kind
of
thing
sort
of
it
eludes
us,
because
we
have
to
figure
out
link
resolution
inside
of
hashes,
and
so,
like
we've
been
doing
that
again
in
the
UNIX
F
has
to
be
one
world.
D
C
And
that's
I
mean
I
think
we
have
a
nice
use
case
then,
because
we've,
the
way
we
have
handled
all
of
this
so
far
has
been
through
convention
right.
We
every
single
file,
has
the
exact
same
name
and
has
the
exact
well
tags
have
the
same
layout
structure,
not
from
the
perspective
of
blocks,
but
from
the
perspective
of
the
naming.
C
A
A
C
A
And
open
issues
and
ask
questions
and
and
have
it
check
out
their
logo
because
it's
rad
and
yeah
just
just
thank
you
very
much
for
coming
and
presenting
it's
been
super
interesting
and
I'm.
Looking
forward
to
all
the
stuff
you
do
in
2020.
Thank
you
again.
Everyone
for
coming
and
we'll
see
you
next
week
for
another
exciting
edition
of
ipfs
weekly
call
bye,
bye.
Everyone.