►
From YouTube: Jupyter Community Call - March 26, 2019
Description
Recording from the Jupyter Community Call in March 2019.
The notes from this call can be found here:
https://jupyter.readthedocs.io/en/latest/community/community-call-notes/2019-march.html
Read more about these calls in Discourse:
https://discourse.jupyter.org/t/all-jupyter-community-calls/668
A
A
So
a
couple
of
things
I
want
to
point
out
before
we
get
started
so
I
should
introduce
myself.
My
name
is
Zach
Saylor
I
work,
a
core
developer
for
Jupiter
at
Cal
Poly
in
San,
Louis,
Obispo,
California,
there's
a
few
of
us
on
this
call
from
that
area,
and
so
we
need
just
last
month
discuss
trying
to
revive
this
call.
A
Anyone
can
share
anything
on
this
call.
You
just
add
yourself
to
the
list
and
we'll
give
you
some
minutes
to
talk
about
what
your
site
about
I
think
today
we're
maybe
a
little
bit
core
developer
heavy,
because
this
is
the
first
one,
so
we're
just
trying
to
test
it
out,
but
really
this
is
open
to
anyone.
So
thank
you
to
everyone
who's
here
so
far.
One
other
thing
I
want
to
point
you
to
is
we're
going
to
try
this
out.
A
A
We
have
a
list
or
a
section
in
there
called
the
speaking
queue
with
a
little
checkbox
section,
where
you
can
add
your
name,
and
only
on
the
current
topic
that
we're
talking
about
right
now
and
we'll
try
to
get
through
everyone
on
that.
Well
time
box,
it
a
little
bit
if
that
list
gets
too
long.
We
may
just
have
to
cut
it
off
and
move
to
the
next
topic,
but
at
least
we'll
go
in
order
of
who
adds
their
name
there
first
and
then
that
speaking
to
will
be
refreshed
on
the
next
topic.
A
So
if
we
can't
get
through
everyone
just
for
time
sake,
we'll
we'll
move
to
the
next
topic
and
we'll
clear
the
names
there
and
do
it
for
the
next
one
that
makes
sense
to
everybody
thumbs
up,
perfect
great
okay.
So
with
that
I
talked
a
little
bit
about
the
purpose,
what
they
call
and
a
little
bit
about
the
history.
So
we
broke
our
agenda
down
into
three
these
two
sections.
One
is
kind
of
a
short
reports
which
doesn't
really
have
this
session
I'll.
Just
if
the
people
are
on
the
call.
A
Let
them
read
out
that
point
if
they
would
like
otherwise,
I
can
just
read
through
the
section
just
give
these
shoutouts
and
then
we'll
move
into
the
longer
agenda
items.
We're
gonna
shoot
for
this
call
to
be
about
an
hour.
If
we
start
getting
close
to
that
time
limit,
then
I
will
kind
of
shorten
the
discussion
on
each
topic
and
then
I'll
ask
people
if
the
people
at
the
end
of
the
agenda,
if
we
can
plump
them
to
next
month's
call
cool
all
right.
B
A
C
A
A
Thank
you
so
suicide.
Thank
you,
Tony
Hearst
for
the
weekly
newsletter.
The
tracking
Jupiter
newsletter
definitely
check
that
out,
and
then
I
mentioned
it's
already,
but
we
had
our
odds,
they're
team
meeting
in
Washington
DC
a
couple
weeks
ago
and
that
was
hosted
by
Lorena
Barba
at
George,
Washington
University.
It
was
awesome.
She
was
great,
so
huge
huge.
Thank
you
from
the
team
to
her.
Those
meetings
are
so
critical
for
us,
so
at
least
get
together
once
a
year
in
person
in
all
of
us
to
discuss
kind
of
future
directions
for
Jupiter
and
then
Peter.
B
A
Peter
so
yeah,
so
this
is
definitely
going
to
be
a
work
in
progress
and
hopefully
we
can
iterate
on
this
and
find
out
what
works
best
for
the
team.
But
the
idea
there
is
that
yeah.
This
was
an
avenue
for
the
communities
to
get
engaged
with
the
core
team
and
other
community
members,
and
so,
if
you
have
feedback
about
this
call
at
the
end,
please
just
share
it
with
us.
A
There
is
a
link
in
the
the
hack
MD
and
under
that
short
report,
section
with
a
link
to
the
discourse
posts
that
I
opened
up
about
a
month
ago.
Talking
about
this
so
engage
in
conversation
when
there
give
us
feedback.
If
there's
anything,
you
think
we
can
do
better
and
let
us
know
or
anything
you'd
like
to
see
ads
of
these
calls,
let's
chat
it
out.
There
that'd
be
awesome,
yeah
and
I.
Think
a
general
point.
A
The
Jupiter
discourse
is
becoming
kind
of
our
new
way
to
to
engage
in
conversation,
kind
of
a
cross,
Jupiter
project,
and
so
we
really
found
that
we
struggled
out
a
single
landing
place
for
users
to
go
to
to
discuss
topics
in
Jupiter.
You
have
to
find
yourself
going
between
multiple
different
repos
at
a
time,
and
so
the
discourse
lady
provides
a
nice
home
single
home
for
you
to
scan
certain
topic,
be
able
to
search
history
and
conversations
that
have
happened
in
the
past,
so
yeah
so
get
on
Jew
to
discourse.
D
D
I'm,
a
dear
just
for
this
course
I
realize
by
showing
that
to
someone
that
if
you
have
an
ad
blocker
when
you
try
to
log
in
the
first
time,
if
you've
never
logged
in,
you
will
not
get
log
in
with
github
and
log
in
with
Google
buttons.
And
so
it
will
basically
ask
you
to
create
an
account.
So
don't
forget
to
turn
off
your
ad
blocker
to
get
in
refresh
the
page
to
get
disturb
at
coubertin's,
and
then
it's
going
to
link
your
account
with
either
Google
or
github.
A
A
A
Competed
before
before
you
start
auditions,
I
should
say
I
thought
where
this
would
what
you
could
do
with
this
time.
If
you
are
on
the
agenda,
Peter
will
probably
do
this
too
is
use
the
time
to
share
share
your
screen.
If
you
would
like
and
actually
walk
through
demos
or
something
that
you're
announcing,
you
don't
necessarily
have
to
do
that,
but
that
tends
to
be
the.
A
B
Virtual
Jupiter
shop,
one
thing
I
wanted
to
share,
was
Jupiter
dr.
stacks
or
just
a
set
of
pre-built
docker
images
where
we
install
notebook
and
lab
and
Jupiter
of
components.
Among
other
things,
we
still
have
a
lot
of
people
that
use
them,
even
though
repo
the
docker
and
the
binder
and
other
things
has
come
along
since
then,
one
of
the
problems
we've
always
run
into
is
these
are
sort
of
mega.
A
B
With
lots
of
different
packages
in
them
and
versioning
them
in
a
way
that
users
understand
what's
in
them,
has
been
challenging
right
now
we
just
put
a
bit
of
the
gate,
commit
on
the
end
of
the
image
and
then
have
been
telling
users
we'll
come
back
to
the
repo
and
see
what
the
state
of
the
docker
files
or
you
know,
requirements
and
so
on
and
so
forth
work.
We
started
to
automate
a
bit
more
this
to
capture
what
I've
been
calling
build,
manifests
for
images
when
the
images
build
on
docker
hub
post,
build.
B
There
is
information
pushed
back
to
the
wiki
here
on
github,
so
for
each
image
you
can
see
the
tag
you
can
see
the
name
of
the
image
and
then
we're
capturing
things
like
well.
The
gif
is
just
the
full
request.
Technically
that
caused
a
change
to
stay
the
docker
file
at
the
point
that
build
was
made,
but
then,
more
importantly,
we
have
this
build
manifest
concept.
Where,
again,
you
could
see
some
of
the
information
the
git
commit
and
a
shot
and
so
forth,
but
we're
actually
querying
at
Build
time
or
post
build
time.
B
The
state
of
Julia,
for
example,
in
the
image
the
contents
and
versions
of
the
Julia
package,
is
the
Python
packages
Conda
the
Kondylis
so
on
and
so
forth.
Our
is
our
packages
and
the
data
science
notebook
and
then
even
act
because
they're
glued
to
base
images.
So
we
sort
of
have
this
full
manifest
of.
What's
in
that
dr.
B
imagine
before
people
go
pull
it
down,
we
can
also
think
about
building
tooling
if
these,
if
something
breaks
from
one
image
to
the
next,
but
it's
really
meant
to
be
a
form
of
like
automated
documentation
to
really
capture.
What's
in
these,
you
know
many
gigabyte,
doctor
images
that
people
still
seem
to
love,
even
though
they're
huge,
so
there's
one.
They
call
light
out,
maybe
just
a
moment
on
how
this
is
automated.
B
If
you
use
docker
hub
docker
cloud
before
you
get
the
idea
there,
there's
the
concept
of
a
post
and
pre
and
so
forth,
build
hooks
and
are
in
a
hooks,
folder
we're
simply
running
a
bash
script
here
and
templating
out.
You
know
what
these
manifest
should
look
like
in
bash
of
ever
worst.
It
might
be
a
nicer
way
to
do
this.
This
was
the
the
first
cut
just
running
docker.
B
You
know
julia
using
interactive
new
tools
to
get
the
bergen
info,
for
example,
same
thing
for
package
status,
same
thing
for
Python
and
Conda
and
our
and
you
can
think
about
its
lending
needs
to
capture
more
information
about.
What's
in
the
images
like
docker,
history
might
be
a
good
one
to
add
to
the
pages
just
want
to
share
that
in
case
it's
useful
to
anyone
and
point
out
to
users.
You
know
now
there's
a
way
to
see.
What's
in
these
massive
images,
without
back
or
pulling
them,
that's
all
I
got.
B
A
E
E
Okay,
so
my
binder
the
org,
if
you
surf
there
in
your
browser,
looks
something
like
this
and
it
tells
you
what
it
will
do
in
the
tagline.
It
will
turn
a
git
repository
into
a
collection
of
interactive
notebooks.
So
the
first
thing
you
need
is
a
git
repository
and
I've
made
one
over
here
as
a
question.
Can
you
guys
see
like
the
little
windows
of
yourself
on
my
screen?
Share?
E
No
good,
then
you
didn't
see
me,
try
to
move
them
out
of
the
way
so
I
made
a
repository
called
my
first
binder
and
it
doesn't
contain
very
much.
It
just
contains
a
readme,
a
hello
dot
py
and
the
requirements
dot
txt,
because
any
even
a
empty
git
repository
is
a
valid
repository
that
my
binder
dog
will
build
for
you
and
launch,
but
that's
not
very
exciting.
E
So
most
people
actually
want
to
install
some
kind
of
software
libraries
before
they
want
to
execute
their
notebooks
or
their
code,
and
one
way
of
doing
that,
if
you're
using
Python,
is
to
create
a
requirements,
txt
file
and
then
you
list
on
one
line,
a
line
per
line.
The
libraries
you
want
to
install.
So
in
this
case,
for
example,
you
want
to
install
numpy
we
list
up
here,
and
then
we
take
the
URL
to
that
git
hub
repository,
and
we
paste
it
at
the
top
here
where
it
says,
get
up
repository,
name
or
URL.
E
You
can
also,
if
you
don't
want
to
use
github,
you
can
use
a
gist
or
repository
hosted
on
get
lab
or
even
any
kind
of
git
repository
which
is
like
accessible
to
the
public
on
the
Internet,
and
then
you
can
specify
which
branch
or
tag
or
specific
commit
you
want
to
launch.
And
if
you
don't
put
anything,
we
will
pick
master
and
there's
a
few
more
other
additional
options
that
you
can
investigate
at
a
later
point.
E
The
most
important
button
is
the
big
orange
one
which,
if
you
click
it,
will
start
the
build
process
and
what
will
happen
if
the
demo
gods
are
good.
Then
you
see
in
this
terminal
window
down
here
all
sorts
of
text
scrolling
by,
and
this
is
telling
you
what
we're
doing,
building
the
docker
image
that
we
are
about
to
launch
for
you,
while
you're
waiting.
E
So
if
you
and
I,
like
you,
know,
take
some
time
to
read
this,
you
will
see
that
we're
installing
mini
condor
and
we're
noticing
that
you
have
a
requirement
so
txt
in
your
repository
and
then
executing
pip,
install
on
that
and
then
very
quickly.
We
get
to
the
end
we're
pushing
the
image
to
our
docker
registry
and
then
now
it's
saying
launching
the
server
for
you
and
then
now
you
will
be
redirected
to
a
notebook
interface.
E
E
One
thing
that
you
want
you
might
want
to
look
at
is
lots
of
other
examples
that
we
have
on
how
you
can
prepare
your
repository
so
that
my
binder
org
recognizes
its
content.
Is
there
some
files,
like
requirements,
dot
txt,
which,
if
you're
a
Python
person,
you
already
know
and
love
for
specifying
your
dependencies,
but
we
also
support
julia
and
our
and
for
them
there's
different
ways
of
specifying
their
dependencies
and
because
it's
a
Ubuntu
image
in
the
docker
image
that
we
build
on.
E
And
if
you
want
to
look
at
more
documentation
for
the
tool
that
we
use
to
build
these
docker
images
on
the
slide
for
you,
it's
called
repo
docker
and
it
has
its
own
webpage.
You
can
use
it
on
your
own
computer
and
it
is
what
my
binder
called
defers
to
for
building
the
images.
So
I
think
there
was
a
very
short
Whistlestop
tour
of
all
the
things
you
can
do
in
binder
and.
E
Have
a
look
at
the
questions
that
there
are
and
then
answer
them?
Is
it
true
that
you
personally
are
the
only
user
of
the
open
with
binder?
Yes,
so
if
you
are
eagle-eyed,
you
can
see.
I
have
a
little
binder
logo
in
my
browser,
tab
bar
and
that's
a
browser
extension
for
Chrome
and
Firefox
and
if
you
click
it,
while
you're
on
some
kind
of
page
that
my
binder
door
can
launch
she
just
like
that,
we
will
launch
it
for
you
and
it's
super
convenient,
because
eventually
you
get
bored
of
copying
and
pasting
links.
E
So
we
run
something
right
between
300
to
500,
concurrent
docker
image
or
the
docket
instances
during
the
period
of
the
day.
So
that's
the
kind
of
scale
of
concurrent
users
on
my
binder.
We
limit
every
individual
repo
to
most
100
concurrent
instances.
So
if
you
have
a
course
with
200
people
who
are
all
going
to
launch
it
small
taneous
Li,
we
would
appreciate
if
you
come
and
talk
to
us
before
groups
of
50
people
is
I.
Think
people
do
that
all
the
time
and
we
rarely
notice
that
they're
doing
that.
E
We'd
love
to
figure
out
a
funding
model
for
binder
who
said
that
yeah
we'd
love
to
as
well
I
am
a
TS
yeah.
So
one
question
that
comes
up
often
is
who's
paying
for
my
binder
to
all
currently.
Is
it
cost
a
lot
of
money
in
terms
of
cloud
compute
and
currently
it's
funded
by
sponsorship
from
Google
cloud
and
how
to
turn
that
into
a
ten-year
commitment
is
something
we're
still
working
on
and
with
that
my
10
minutes
are
over,
so
I
will
pass
on
to
the
next
one
and
yeah.
E
A
A
A
F
Okay,
fantastic,
so
I
just
wanted
to
go
through
a
quick
laundry
list
of
things
and
happenings
on
within
the
interactive
project
that
are
going
to
be
going
on
within
the
next
month
or
so
so.
The
first
thing
that
I
wanted
to
share
is
that
version.
1.0
of
paper
mill
is
going
to
be
released
in
early
April.
F
For
those
of
you
who
might
not
be
familiar
paper
mill
is
a
tool
that
allows
you
to
parameterize
and
execute
notebooks
thanks
so
much
to
Matt,
seal
and
Carol
willing
for
the
hard
work
they
put
into
both
developing
the
code
for
this,
and
also
setting
up
the
documentation,
writing
tasks.
All
of
that
good
stuff,
so
version
1.0
release
coming
up
super
soon.
F
The
next
thing
I
wanted
to
announce
in
line
with
the
1.0
release
of
paper
mill,
is
an
update
to
a
tool
that
sits
alongside
paper
mill
called
scrapbook,
which
basically
allows
you
to
kind
of
pick
up
scraps
from
a
notebook,
and
these
scraps
are
generated,
data
values
or
visualizations
and
arrange
them
into
a
scrapbook.
I
would
just
just
place
them
into
another
notebook
or
gather
outputs
from
multiple
notebooks
into
one
collection
scrapbooking,
but
for
Jupiter
notebooks,
very
fun.
F
The
interesting
development
here
is
that,
alongside
the
same
1.0
release
of
paper
mill,
scrapbook
we'll
also
be
releasing
support
for
publishing
your
scraps
to
the
cloud.
So
this
is
on
s3
Google,
Cloud
storage,
as
well
as
Asher's
storage
services.
So
all
of
the
outputs
and
data
values
that
you
can
plug
out
of
notebooks
you
can
now
store
and
the
cloud
is
a
post
to
your
local
machine.
So
that's
a
cool
development
on
the
paper
mill
and
scrapbook
front.
F
The
other
announcement
I
wanted
to
make
is
a
tool
that
we've
been
working
on
called
the
curl
relay.
The
kernel
relay
is
essentially
a
graph
QL
service
for
interfacing,
with
one
or
more
jupiter
kernels,
I'm
in
the
github
repo
for
the
kernel
relay
right
now,
which
is
just
interact.
Kernel
relay
here's
the
architecture
diagram
for
those
of
you
who
are
interested
in
looking
at
it.
F
While
I'm
speaking,
we
just
laid
out
a
plan
for
version
1.0
of
kernel
relay
and
some
of
the
work
that
will
need
to
go
on
to
ship
the
first
release
it's
currently
under
a
project
within
this
github
repository.
If
you're
interested
in
this
architecture
or
building
a
solution
like
this
you're,
definitely
welcome
to
contribute
and
help
us
out
working
towards
a
1.0
release
of
this.
F
The
other
development
going
on
is
with
the
respect
to
the
data
explorer.
I
think
I'm
not
to
move
the
window
of
faces
that
is
blocking
this.
Alright
sorry,
I
had
to
move
the
little
visual
of
everyone
on
the
meeting
to
access
these
buttons
over
here.
So
this
is
the
data
explorer
for
those
of
you
who
are
unfamiliar.
The
data
explorer
is
interactive
output,
with
an
interact
with
allows
you
to
render
data
that
is
presented
in
a
json
table,
schema
format
in
a
set
of
automatic
visualizations.
F
It's
designed
to
kind
of
bridge
the
gap
between
you've
got
this
new
data
set
that
you
want
to
play
with,
and
you
want
to
just
kind
of
tinker
around
with
different
visualizations
before
getting
to
the
final
visualization
that
you're
hoping
to
create.
So
this
is
the
tool
and
a
sample
data
set
over
here.
You
can
see
all
of
the
options
that
are
provided.
F
One
of
the
big
efforts
that
is
underway
is
to
modularize
this
tool
so
to
make
it
easier
for
individuals
to
swap
out
the
grid
that
they
might
use
to
render
tabular
data
some
of
the
different
visualizations
and
plots
and
all
of
the
configurations
that
are
available.
The
end
goal
is
that
this
is
going
to
be
much
more
flexible
and
dynamic
tool
for
all
of
the
different
use
cases
that
exist
with
the
data
explorer
I'm.
F
So,
if
you're
interested
in
helping
out
with
that
effort,
we
have
a
data
Explorer
tag
with
or
github
repository,
which
is
just
interact,
interact
and
you
can
explore
some
of
the
issues.
Join
the
conversation,
contribute
or
just
kind
of
watch
and
see
what
we're
working
on
and
then
the
last
update
I
want.
It
to
give
is
that
interact
is
participating
in
Google
of
Summer
of
Code
as
a
non
focus
sponsored
project.
Google
Summer
of
Code
is
a
summer
long
program
where
students
can
participate
and
contribute
to
open
source.
F
They
can
receive
mentorship
from
open
source
materials
and
contributors,
and
they
can
help
build.
You
know.
Software
engineering,
skill
set
or
just
a
development,
skillset
interact
is
participating
this
summer.
So
if
you
know
any
University
students
who
you
think
would
be
a
good
fit
for
this
program,
definitely
have
them
apply.
I
will
be
mentoring
and
I
think
I'm
a
generally
a
pretty
cool
mentor.
So
there's
that
there's
also
other
numbaz
projects
participating.
If
you
worry
what
I
might
do
to
the
children
there's
other
projects
they
can
participate
in,
but
yeah
interact
will
be
participating.
F
Looking
forward
to
working
with
some
brilliant
students,
the
supper
feel
free
to
send
this
link
out
to
anyone.
You
know
who
might
be
interested
or
who
might
not
be,
but
you
think
this
would
be
a
great
opportunity
for
ya.
That
is
all
for
me.
I
know
there's
some
other
stuff
that
needs
to
be
presented,
so
I'll
jump
to
questions
and
then
hop
into
the
next
person.
Beckon.
C
F
You
want
to
talk
to
more
in
depth
about
some
of
the
things
that
were
mentioned
here.
The
best
place
would
probably
be
the
interacts
slack
where
we
have
kind
of
informal
conversations.
We
also
have
a
weekly
planning
meeting
that
we
do
on
Mondays.
It
was
previously
on
Monday
mornings,
but
we're
now
shifting
it
to
Monday
afternoons.
But
generally
we
have
a
weekly
meeting
on
Mondays
that
you
can
find
out
about.
By
going
to
our
slack.
F
F
G
Hi,
you
haven't
used
this
on
this
operating
system
before
where's
the
share
screen
share
screen.
It's
big
and
green
I.
Have
lots
of
desktops
come
on
desktop
share
screen?
Can
you
see
my
stuff
yeah
all
right,
this
deathbeds?
This
is
a
thing
that
Tony
fast
and
I
do
for
fun,
and
we
just
have
a
lot
of
stuff
in
here
there's
almost
a
hundred
notebooks
in
deathbeds
github
I/o.
If
you
were
really
bored,
that's
probably
the
most
dangerous
kondeh
environment
to
create
ever
takes
about
a
clock
time.
G
I'm
going
to
talk
about
two
little
little
projects
day,
we've
been
kicking
around
this
one's
called
Linda
type
lint
to
type
lets.
You
take
tools
that
are
built
for
Python
source
and
use
them
for
the
contents
of
your
notebook.
So
a
while
ago
we
got
access
to
cell
IDs
when
you
actually
do
execution
and
that
opened
up
a
whole
new
world
of
things.
So
pretty
easy
right
now
this
one
has
wired
up.
G
G
So
it's
actually,
it's
actually
evaluating
all
the
cells
at
once,
because
you
have
to
do
that
for
a
lot
of
these
annotation
tools.
So,
even
though
that
this
this
line
right
here
had
already
been
lent
'add
we're
seeing
here
that
you
can't
add
a
string
time
as
a
subway
or
another,
but
I
think
if
you
switch
around
it
works
fine,
so
right
now
it
does
this
stuff.
It
is
interactive
which
means
that
I
can
pop
into
one
and
generate
a
bunch
of
options
that
I
can
configure
and
see
different
things
about
my
application.
G
So,
while
I'm
working
on
it,
I
can
turn
step
off
if
I
don't
want
to
see
things
work
in,
but
if
I
do
not
care
to
be
schooled
by
the
cruel
mistress
that
is
the
winter.
So
it
goes.
Let
me
turn
those
back
on
and
one
of
the
funny
ones
that
we
did
is
a
class
diagram
thing.
So
this
uses
something
built
into
pile
inch,
soap
out
of
Class
A
and
it
doesn't
like
to
think
old
stuff
and
I
got
a
Class
B,
which
subclasses
a
and
you
know
so.
G
I
look
there's
a
little
button
over
here.
If
I
hit
class
diagram
there's
a
two
B.
This
is
also
looking
across
the
whole,
the
whole
shooting
match.
So
if
I
was
to
Class,
C
extend
B
again
and
update
that,
so
that's
pretty
fun.
It's
a
you
know.
It's
not
fully
aware
of
types
like
some
of
the
other
tools
are,
but
it's
still
pretty
good,
so
this
Lindsay
type,
the
other
ones
WXYZ.
This
is
for
those
of
you
that
have
been
watching
at
home.
G
G
So
all
these
widgets
I
don't
have
them
fully
wired
up
yet,
but
all
these
are
isomorphic
in
that
they
are
implemented
in
both
Python
and
JavaScript,
so
they
could
be
implemented
in
other
back-end
languages
as
well.
So
this
little
particular
one
lets
us
do
some
templating
with
ginger,
which
is
also
implemented
in
nunchuks,
and
so
we
can
look
across
some
stuff
here
and
we're
actually
looking
at
the
data.
G
The
template
you
have
put
some
widgets
aren't
all
in
the
same
thing
here,
so
it's
pretty
fun
it'll
be
good
when
it
works,
when
I
actually
have
an
example
we're
working
outside
of
there.
In
addition
to
templates,
we
brought
in
the
whole
JSON
stack,
so
we
got
JSON
document,
JSON
schema
and
JSON
pointer,
they're
super
cool,
so
I
can
ask
for
everything.
That's
in
there
or
I
can
or
just
the
I
get
nice
errors
in
there.
G
So
this
is
kind
of
the
beginning
of
some
some
more
interactive
dashboarding
work
that
we
that
we
want
to
see
happen.
These
also
can
escape
from
the
editor
I'm
in
single
wow.
That's
a
slider
holy
moly
yeah,
so
you
can
actually
start
pulling
your
your
pieces
of
stuff
off
of
where
they
used
to
live
and
draw
them
outside
of
it.
If
you
really
don't
like
that,
then
you
can
just
go
ahead
and
pop
stuff
right
into
the
dock.
Just
make
some
more
sliders
Relic
sliders,
all
those
things
as
well.
G
If
you
oh,
it
looks
a
lot
better
than
dark
theme.
So
this
lets
you
just
fullscreen
some
stuff.
Everybody
likes
that,
and
that
also
happens
to
work
on
the
dock
panel.
So
you
can
pop
into
one
of
those
takes
a
sec
for
it
to
redraw,
but
it's
still
kind
of
kind
of
cute.
You
can
play
around
with
things
like
how
much
outlay,
how
much
border
it
has
things
like
that
and
pop
out
of
there
and
the
last
one
is
an
oldie
but
a
goodie.
G
So
this
lets
you
load
up
an
SVG
and
use
the
SVG
is
the
background
for
a
bunch
of
other
widget.
So
you
say:
hey
I
got
an
SVG
I
want
to
see
it
in
a
box,
and
that's
described
by
a
thing
that
I
built
over
and
hang
skate,
and
so
these
are
just
really
nasty
test
issues
for
that
I
think
that's
about
all
I
got
both
these
are
all
up
on
github
and
they
have
binders
and
they
work
black
doesn't
work
right
now,
so
I'm
working
on.
That's
all
question.
G
Someone's
typing
the
SVG
fast
with
my
eyes,
yeah
I,
know
I,
know
and
I
was
moving.
It
really
fast,
yeah
it's
just
because
of
black
and
white
I
need
to
do
test
stuff
on
light
and
dark
and
I
know
things
broken
yeah
there
you
go
I'm,
also
working
on
a
slider
widget,
so
starting
to
replace
the
built-in
stuff
that's
made
out
of
HTML
Dom
so
that
you
can
make
like
you
know,
games.
F
G
So
this
is
super
important
for
her
type
languages
and
stuff
like
that,
where
you're
trying
to
build
a
consensus
document,
it's
not
a
stream
of
things
like
the
whole
document
has
to
be
looked
at
it
once
it's
in
the
metadata
on
an
execute
request
and
you
just
got
to
go,
hunt
around
for
it
and
find
it
and
then,
if
you're,
implementing
a
front
end,
you
got
it
provided,
but
it
has
a
it
has
no
meaning
to
the
colonel.
Unless
you
decide
to
make
it
up.
Yeah.
A
Sorry,
I'm
in
a
lab
coffee
shop,
so
I'm
just
wondering
what
is
like
the
tools
you
just
showed
up.
Are
these
meant
to
be
like
extensions
for
Jupiter
lab
or
were
they
like
future
features
of
Jupiter
lab
or
I'm
just
gonna
curious?
What's
the
relationship
between
the
specific
things
you
mentioned
and
like
the
kind
of.
G
Well,
I
mean
I,
don't
new
start
new
notebook,
widgets,
so
they're,
just
widgets
and
they're
just
built
on
top
of
things
that
are
already
in
lab,
so
code
mirrors
and
lab
phosphors
and
lab
so
yeah
I'd
like
them
to
at
some
point
they
would
I
mean
we
should
talking
to
merge
in
the
lab,
but
you
know
they
could
go
into
core.
Those
ones
could
go
into
court.
A
D
Plenty
of
time
I
was
trying
to
be
fast
outside.
We
will
be
like
hunger.
So
one
of
the
thing
we
discussed
two
weeks
ago
at
the
meeting
that
was
more
profitability
of
what
entities
are
being
done
and
so
on
and
so
forth,
and
so
I'm
going
to
try
to
do
monthly
release
of
my
Python,
which
I
mean
trying
to
do
for
quite
some
time.
D
If
you
can
subscribe
to
725,
there
are
a
lot
of
small
tasks
when
the
actual
resistant
to
update
a
lot
of
of
things
like
come
back
for
a
quickie
yeah
with
a
new
version
that
you
can
and
you
can
help
with.
Otherwise,
the
development
of
a
Python
has
slowed
down
a
bit
because
we,
we
don't
have
many
people
contributing
and
I
have
well
at
times
and
before
so,
don't
expect
as
many
new
features
as
before
and
there's
a
budget
for
ipython.
Not
much
not
much
have
AB
change.
D
D
So,
basically,
while
you're
waiting
for
a
camel
to
start,
your
different
places
can
do
other
stuff
without
having
to
you,
rather
than
everything,
it's
quite
a
complex
problem.
It
will
be
interesting,
for
example,
for
HPC
HPC
system.
So
so
far
when
you
use
HPC
system,
you
basically
need
to
start
a
full
notebook
on
the
compute
node
to
do
something,
and
you
would
like
to
move
to,
for
example,
having
server
at
some
place
and
then
having
to
compute
away
on
some
other
machine.
I.
D
Think
in
a
way,
I
would
love
to
have
your
your
feedback,
or
at
least
your
input
and
your
help
to
reduce
some
of
the
requests
in
the
discussion.
I
linked
to
the
main
one
here,
which
is
on
Jupiter
client
but
I
quest
for
28,
give
incentive
requests,
I'm,
trying
to
review
and
fix
bugs
so
far
trying
to
get
em
to
get
the
best
bass
and
that's
about
it.
There
is
no
well,
there
is
a
past
for
a
while.
D
A
A
All
right
thanks
everybody,
so
that
brings
us
to
the
end
of
our
agenda.
We
actually
are
finishing
early,
so
great
job,
everybody.
Thank
you
to
everyone
who
presented
and
like
I
said
I
want
to
for
those
of
you
who
weren't
here,
maybe
at
the
beginning
of
the
call
I
want
to
point
everyone
to
the
discourse
post.
That's
actually
open
on
this
video
call
about
this
video
call
it's
under
in
the
hack
and
the
agenda.
A
It's
like
the
fifth
item
in
the
shorts
report
just
says:
leave
your
feedback
on
the
algebra
community
call
go
ahead
and
click
on
that
link.
Leave
your
feedback.
Tell
us
what
you'd
like
to
see
more
of
maybe
what
you
thought,
work
didn't
work
and
we'll
continue
to
iterate
on
this
each
month
and
then
in
the
meantime
one
other
thing
is:
we
are
going
to
try
to
have
these
monthly
and
so
it'll
be
great
one,
just
to
connect
and
see
everybody,
but
two.
A
If
you
have
something
you
want
to
share
in
the
next
month,
we're
gonna
try
to
open
these
agendas
a
month
early.
So
add
yourself
to
that
list
and
try
to
do
that.
Both
24
hours
before
the
call
and
then
will
help
us
organize
these
calls
and
set
them
up
for
that
next
call
so
yeah
other
than
that.
I'm
gonna
go
ahead
and
stop
the
recording
if
people
want
to
stay
on
and
talk
after
recording
some
that'd
be
totally
fine,
but
other
than
that
thanks,
everybody
was
great
to
see
everyone.