►
From YouTube: Jupyter Community Call - May 28th, 2019
Description
Recording from the Jupyter Community Call in May 2019.
The notes from this call can be found here: https://jupyter.readthedocs.io/en/latest/community/community-call-notes/2019-may.html
Read more about these calls in Discourse:
https://discourse.jupyter.org/t/jupyter-community-calls/668
A
Come
on,
we
are
now
recording,
so
hello,
everybody
welcome
to
our
monthly
Jupiter
community
call.
This
is
our
third
month
doing
this,
so
we're
still
iterating
on
this
and
learning
from
this.
But
thank
you.
Everyone,
who's
come
and
he's
been
a
lot
of
fun
these
last
couple
of
months.
So
we
hope
you
going
I
want
to
say
one
of
the
things
that
we're
learning
from
this
process
is
just
how
to
effectively
communicate
these
and
get
these
out
to
the
community
and
also
to
get
people
interested
in
giving
talk.
A
So
they
really
are
an
open
format
and
we've
been
very
lucky
that
so
far
we'll
have
this
freely
signed
up
and
we've
been
able
to
fill
these
time
slots
and
I
hope
that
continues
to
happen.
If
that
doesn't
we
will
you
know
if
it
ends
up
being
just
one
person
signing
up
for
that
month,
we'll
continue
to
have
it
that
month.
A
If
no
one
signs
up,
we
may
get
on
the
call
and
say
this
month,
we're
gonna,
you
know,
maybe
make
this
an
open
chat
for
a
few
minutes,
but
for
the
most
part,
we'll
wait
til
the
next
month,
so
we're
gonna
let
these
just
kind
of
organically
evolve.
If
everyone's
cool
with
that,
really
it's
a
good
chance
just
to
connect
once
a
month
and
and
see
people's
faces
too
and
in
those
scenarios,
maybe
we
can
even
open
up
the
floor
for
Q&A
with
with
core
developers
if
community
members
are
interested
in
that.
A
A
If
I
cut
you
off,
it's
nothing
personal,
we
just
try
to
stay
on
task
so,
like
I
said
at
the
end
of
the
call,
if
there
ends
up
being
time
what
kind
of
floor
or
maybe
even
stop
the
recording
and
just
let
people
converse
freely
across
the
call.
So
this
is
an
open
place
to
chat
again.
Anyone
is
welcome
to
sit
in
and
listen
and
participate.
However,
they
like
so
thank
you
all
for
coming
all
right.
So
with
that
we'll
go
ahead
and
we
have
basically
break
our
call
into
two
sections,
a
short
reports
section.
A
This
is
just
people
giving
shoutouts
about
things.
They
want
to
talk
about,
and
really
no
conversation
happens
here.
This
is
just
to
give
announcements
for
the
most
part
and
then
we'll
move
into
our
main
agenda
items
and
that's
kind
of
free
to
speak
time.
So
we'll
just
go
through
our
short
reports,
so
the
first
one
I
think
this
comes
from
Kirsty
and
the
turn
way
team.
You
guys
want
to
give
us
a
shout
out
here:
yeah.
B
Yeah
so
I'm
gonna
bring
this
this
drawing
up.
We've
had
an
book
events
for
this
project
for
the
churring
way,
which
is
which
is
book
to
make
reproducible
research
easy
not
to
do,
which
is
the
hilarious
thing
to
try
and
achieve,
and
and
we've
had
a
group
of
people
one
in
Manchester
a
couple
of
weeks
ago
and
one
today
so
there's
some
folks
who
are
still
beavering
away
at
the
back
and
collaboratively
writing
this
book.
B
It's
a
Jupiter
book
and
so
we've
been
sort
of,
hopefully
getting
some
by
the
work
in
there
and
some
live
demos
and
kind
of
building
it
as
you
go
along,
it's
all
cc-by,
so
anyone
is
welcome
to
focus
reuse
it
remix
it
and,
but
also
contribute
back
in,
would
be
really
amazing.
So
we've
got
these
amazing
pieces
of
Arts
from
Matt
who
might
still
be
in
the
camera
and
we're
just
so
I
didn't.
Thank
you.
A
B
Is
where
you
get
together
like
three
to
five
days
to
write?
The
book
is
shorter
than
a
sprint.
It's
just
a
one-day
event.
We
actually
got
together
and
did
a
networking
and
dinner
event
yesterday.
So
we
had
some
Lightning
talks.
There
did
they
showed
off
some
of
their
own
subversive
cross
stitching.
We
showcase
different
things
that
people
Institute
talked
about
last
night
and
had
dinner
and
then
from
9:00
a.m.
and
so
by
DPN
everybody
just
but
it's
acted
like
a
lot.
It's
quiet
in
the
room
and
contributing
chapters
on
Sara
was
great
one.
A
A
E
I'm
sure
he
needs
to
as
well
and
I'm
also
at
the
end,
my
dog,
great
and
so
last
week,
I
did
a
three-day
spring
that
was
jointly
hosted
by
in
theory
and
involved
a
trip
down
to
the
reactor
workspace
in
London
and
hosted
Microsoft.
We
all
got
together
in
like
on
some
cloud
based
projects
and
what
was
really
cool
was
level
of
interest
about
binder
hub
in
the
room
like
like.
We
almost
just
had
one
huge
bundle
pack,
but
eventually
some
people
decided.
No.
E
Maybe
we
should
render
the
project
and
so
I
decided
to
lead
on
a
project
that
I
tried
to
get
started
here
with
the
cheering
by
myself,
with
momentum
on
and
that's
a
set
of
scripts
that
will
auto
deploy
a
binder
hub
to
a
job.
So
there's
like
all
your
deploy
scripts
and
then
a
few
in
then
you
can
pare
it
down,
and
so
all
those.
So
now
let
working
locally
in
a
fork
of
a
fork
and
our
next
goal.
We're
trying
to
reach
is
to
get
push-button
deployments.
E
So
much
like
a
binder
link
in
a
readme.
There
will
be
a
deploy
button
on
a
reach
me
in
a
github
thing
and
within
a
few
minutes.
You'll
have
a
fully
working
by
the
hub
on
a
job,
and
it
was
just
really
cool,
I'm,
still
riding
on
the
excitement
and
enthusiasm,
and
it
was
really
nice
for
they
even
project
and
see
those
people,
country
together
and
another
outcome
of
the
today.
I,
don't
know
how
many
of
you,
with
the
comic.
E
Pull
one
down
the
doctor:
well
that's
floating
on
wild
ocean
and
then
we
have
lots
of
people
coming
up
on
their
little
boats.
So
we've
got
like
a
little
class
of
people
and
they're
all
learning
out
the
software.
This
guy's
decided
he
really
likes
the
software
he's
taking
off
his
business
and
we've
got
kubernetes
up
here,
which
is
the
crane
managing
all
of
the
pots.
B
B
A
D
But
let
me
try
to
talk
a
little
bit,
so
we
did
a
Jupiter
server
design
in
a
roadmap
workshop,
which
was
sponsored
by
Bloomberg
and
hosted
at
IBM
France
on
May,
16
and
17,
some
of
the
folks
that
participated
so
like
Sylvain
Thomas,
you
Roland
myself,
and
we
were
talking
about
the
Jupiter
server,
trying
to
discuss
kind
of
like
a
functionality,
directions
and
and
and
try
to
come
up
with
at
least
a
very
high-level
roadmap
for
those
that
are
not
aware
of
like
what
the
Jupiter
server
is.
D
So
the
goal
of
the
Jupiter
server
so
decoupled,
the
common
back-end
services
from
Jupiter
notebook
server
so
that
it
can
be
usually
used
easily
in
two
different
proteins.
We
are
seeing
a
lot
of
like
like
notebooks
was
the
initial
one.
Jupiter
lab
interact.
We
Jets
boil
and
up
orders
that
are
basically
using
the
notebook
or
using
the
backend
functionality,
but
providing
different
you
eyes
and
user
experience
in
general.
So
the
idea
is
to
start
splitting
the
Jupiter
server
so
that
it
can
be
used
by
all
of
this
in
a
reasonable
way.
D
Also,
how
do
we
want
to
accomplish
that?
So
that
is
that
the
Jupiter
server
starts
providing
extension
points
and
those
UI
they
become
like
front-end,
interactive
applications
in
excuse
me
and
the
existing
front
ends
will
become
the
application
extensions
providing
UI
on
top
of
Jupiter
server.
The
adductors
server
will
also
continue
to
provide
back-end
extensions
so
where
we
can
update
existing
handlers
or
customize
the
behavior
of
those,
but
also
add
the
new
handlers
from
the
REST
API
that
are
being
exposed
by
the
server
at
a
high
level
roadmap.
D
I
think
we
were
discussing
kind
of
like
two
things:
a
phase
one
is
we
start
introducing
kind
of
like
the
Jupiter
server
and
start
splitting
or
or
integrating
with
the
existing
UI
so
like,
for
example,
notebook
in
lab
will
start
becoming
an
epoch
station
and
using
the
server
it's
kind
of
like
the
backhand
for
for
the
services
that
it
needs,
then,
as
a
face
to
you,
which
then
the
the
xsplit
already
happened.
We
were
thinking
on
starting
expanding.
D
Some
of
the
functionality
so
Thomas
had
provided
like
karna
provided
PR,
which
provide
a
little
bit
more
flexibility
to
integrating
different
kernel
management
in
on
the
backend,
and
we
already
did
something
like
that
in
the
enterprise
gateway,
where
we
have
this
notion
of,
like
process
proxies
to
provide
remote
kernels
and
all
of
that
kind
of
like
integrating
and
giving
the
Jupiter
server.
Also
the
ability
to
become
kind
of
like
kernel
services
in
a
more
flexible
way.
So
those
were
kind
of
the
things
that
we
discuss
it
in
very
high
level.
D
The
phases
that
we
are
thinking
I
think
the
next
step
I
mean
Zak,
is
already
working
on
on
the
the
PRS
for
the
splitting
I
have
to
provide
a
blog
post
with
this
summary,
similar
to
what
we
are
discussing,
but
a
little
bit
in
more
details
and
weave
some
illustrations
to
help
folks
understand
a
little
bit
more
in
socialize.
We
will
start
also
like
trying
to
come
up
with
bi-weekly
calls
like
some
of
the
projects
are
doing
weekly.
D
D
G
D
A
This
doesn't
change
the
server
at
all
Nick,
it's
just
separating
it
out
into
a
separate,
a
separate
repo
or
trying
to
make
the
transition
essentially
unfilled
by
the
users,
and
you
then
make
it
so
that
the
front
ends
don't
really
feel
like
they're
behaving
any
different,
like
classic.
Notebook
doesn't
feel
like
it's
changing,
but
it
really
is
it's
the
it's
being
loaded
as
a
server
extension.
It's.
A
G
Yeah,
it
just
seems
like,
with
the
with
the
the
three
sunset
and
some
of
the
growing
support
around
ASCII.
There
may
be
a
point
to
becoming
less
opinion
about
what
actually
serves
their
stuff,
so
our
stuff
becomes
more
embeddable.
You
know,
I
mean
how
many
times
like
everyone's
put
envy
viewer
inside
their
application,
I'm
not
being
a
terrible
idea.
It's
not
really
convenient
to
do
that
because
of
our
our
specific
tornado
tool
chain,
which
is
actually
not
that
broadly
used.
G
A
This
possibly
opens
the
door,
though,
for
other
server
implementations.
You
can
imagine
that
what
we
do
is
we.
We
provide
the
connection,
essentially
define
the
connection
points
between
clients
and
the
server
and
then
whatever
other
pieces
like
we're,
really
turning
the
server
into
a
another
building
block,
and
then
we
could,
you
know
someone
could
create.
It
could
replace
that
building
block
the
server
building
block
with
another
implementation,
a
server
that
yeah
I
don't
know
man
I
mean.
Maybe
that
makes
that
possible
and
we
can
transition
that
direction.
But
yeah.
G
F
D
Yeah
I
think
what
we
were
thinking
at
least
for
the
notebook.
If,
if
the
community
all
agrees
is
probably
like
the
next
major
release
so
like
on
sick,
so
I
would
say,
I
think
the
next
question
will
be
like
how
about
Jupiter
lab
I
think
what
we
were
talking
is:
it's
gonna
be
post
Jupiter
lab
lab
dr.
lab
1.0,
but
it
should
be
transparent,
so
it
might
not
need
like
a
kind
of
like
major
release
or
anything
like
that.
It
can
be
like
a
expect
or
something,
but
we're
still
discussing
some
of
those.
A
A
G
G
G
This
isn't
much
better
but
different,
and
maybe
a
little
bit
more
predictable
and
trying
to
make
your
stuff
work
in
that
environment
is
hard.
So
qts
been
part
of
the
jupiter
story.
For
a
long
time,
we
had
the
the
QT
console,
which
has
been
donated
to
the
spider
community
project
and
it's
not
really
directly
supported
much
anymore
and
there's
not
that
many
QT
developers
in
the
science
world.
G
It's
gonna
draw
some
PQ
plot
and
some
bouquet
and
it's
gonna-
do
some
some
SVG,
matplotlib
and
I
don't
know
whatever,
and
so
it's
running
takes
a
long
time
under
the
hood.
It's
spinning
up
at
Jupiter
lab
and
then
it
connects
the
QT
browser
directly
to
it,
and
then
they
talk
back
and
forth
and
wait
for
things
to
be
done.
So
it's
actually
faking
dev
mode
because
it
gets
access
before
the
Dom
is
finished,
rendering
so
I
get
direct
access
to
the
lab
command.
G
G
Okay
and
it
finished
so-
we
almost
ran
out
of
this
free
space
in
our
memory
move
anyhow,
so
we
got
144
Kb
PDF
out,
which
is
fine.
That
is
a
great
size
and
here's
what
it
looks
like
in
the
side
of
the
browser,
PDF
viewer,
so
there's
just
a
bunch
of
stuff,
not
great
but
great,
Marv.
The
same
token:
here's
some
bouquet
bouquet
loads,
here's
some
more
code!
This
is
what
it
was
complaining
about,
that
it
couldn't
draw
the
Pats
but
I'll
take
it,
and
this
matplotlib
is
a
widget
I.
G
G
So
you
get
like
a
web
channel
thing
that
you
can
use
to
talk
back
and
forth
directly
over
PI
QT
slots,
but
it's
really
not
that
much
code.
The
the
next
place
that
you
can
go
from
this
is
actually
I.
Think
they
got
here.
Can
you
guys
see
that
yeah
yeah?
Yes,
this
might
not
be
working
right
now,
but
no
all
right,
I
killed
that
thing
off
I
was
doing
too
much
depth.
Let
me
see
if
I
can
fire
back.
G
So
this
is
taking
it
a
different
direction
where
oh
come
on
it's
taking
a
different
direction
where
instead,
the
entire
app
web
application
is
served
up
through
there,
so
the
other
one
is
all
about
making
sure
that
it
runs
correctly
inside
of
a
headless
environment.
This
is
where
you
know
what
actually
want
to
do
like
stuff
with
it.
So
this
is
running
inside
of
the
chromium,
5
9
and
it
looks
like
I'm
gonna.
Imagine
that's.
The
screenshare
is
making
something
go.
Yeah
I,
don't
think
it
likes.
The
screen
share.
I,
don't
know!
G
What's
going
on,
you
know
desktop
chart
so
anyhow
and
yeah.
Here's
some
yeah,
it's
I,
haven't
tested
it
under
screen
share
before
so
anyhow,
but
the
basic
idea
is:
if
I
wasn't
running,
screen
share.
All
these
things
would
work
inside
of
here.
Crazy
stuff,
like
I,
might
WebRTC
just
works,
which
is
insane
geolocation
works
and
screen
sharing
will
work
when
you
know
whenever
QT
512
comes
up,
so
this
is
yeah.
It
is
it's
very
angry
I.
Don't
my
window
manager
is
fighting?
Oh
maybe
it's
gonna
work
here.
G
G
You
know,
so
you
have
to
use
a
really
old
chrome
driver
to
talk
to
it,
and
then
you
got
to
do
things
to
make
it
work
properly,
and
then
you
got
to
make
it
work
headless,
but
on
the
end
you
know
if
we
can
test
it
and
we
can
run
it,
then
this
presents
an
option
for
folks
that
may
not
be
able
to
control
which
browser
they're
running,
but
they
are
allowed
to
install
the
scientific
Python
suite
which
includes
QT.
So
let's
get
around
some
of
those
things.
That's
all
I
got.
Oh
there's
repo.
G
F
F
G
You
know
the
gentleman
earlier
on.
The
call
was
saying
that
he
can't
use
NPM,
so
I
don't
want
to
bring
electron
into
this.
It's
using
QT,
that's
already
in
an
install.
If
you
installed
Jupiter,
if
you
say
Conda
install
Jupiter,
you
get
QT,
and
so
it's
it's
just
a
lot
more
accessible
from
the
Python
side,
which
will
make
it
easier
to
even
agree
with
the
rest
of
our
kit
that
we
actually
want
to
support
because
I
I
can't
I
know
it's
tough,
isn't
saying:
I
can't
do.
F
I
G
J
G
Yeah
I
mean
it's
got
a
work
offline
right,
I
mean
it's
got,
it's
gotta
work
headless.
It's
got
to
work
inside
of
CI
and
stuff,
like
that.
That's
a
that's
a
each
limiter
for
me.
So
you
know
if
I'm
supporting
20
classes
and
I've
got
you
know,
each
each
class
has
20
slide
decks
that
all
need
to
be
distributed
as
PDF
and
we
changed.
You
know
the
copyright
date.
I
can't
go
through
lab,
for
you
know
two
hours
and
and
press
buttons
like
it
has
to
work
offline
right.
So
that's
yeah,
I,
don't
and
I.
G
J
G
G
But
the
fact
that
it's
actually
Jupiter
lab
and
not
going
through
HTML
bootstrapping,
all
those
other
things
fighting
means
that
you
only
will
have
to
write
stuff
for
the
UI
that
you
want
to
dump
that
you
don't
have
to
make
it
also
work
in
the
static
HTML.
You
can
make
it
work
in
the
dynamic
full
environment.
You
can,
you
know
twiddle
widgets,
all
that
kind
of
stuff,
so
right,
yeah
right
now
just
works
with
lab,
but
it
would
work
just
fine
with
interact
or
class
or
any
other
stuff.
A
G
E
G
A
All
right
thanks,
everyone
for
making
Nick
for
presenting
and
sharing
that
extra
filled
in
the
questions
and
comments
with
that
I'm
going
to
go
ahead
and
conclude
the
video
call,
or
at
least
the
video
recording
and
then
feel
free
to
stay
on
the
call
if
you
like,
and
we
can
continue
to
chat.
Thank
you,
everyone
who
came
this
month,
we're
gonna
have
us
again.
Next
month,
you
can
put
on
your
calendars:
that'll
be
the
last
Thursday
of
June
June
25th
same
time.
That's
pretty
much!