►
From YouTube: Jupyter Community Call - November 17, 2020
Description
Recording from the Jupyter Community Call in November 2020.
The notes from this call can be found here:
https://jupyter.readthedocs.io/en/latest/community/community-call-notes/2020-november.html
Read more about these calls in Discourse:
https://discourse.jupyter.org/t/all-j...
A
Okay,
well,
hello,
everyone
and
welcome.
Jupiter
community
calls
they're
back
yay,
so
happy
november
2017.
So
the
first
thing
I
wanted
to
share
really
quick.
I
have
a
little
right
here.
Yes
share
that
just
doing
some
slides,
oh,
can
I
get
sharing
permissions
or
do
I
need
to
there?
Does
that
work
awesome?
Thank
you.
I
made
some
quick
slides
just
so
I
remember
what
to
say.
Every
time
welcome.
This
is
november
17th
we're
here.
It's
a
community
call
and
I'm
super
happy
to
have
him
back
and
it's
great
to
see
everyone
showing
up.
A
So
thank
you
so
much
I
did
want
to
bring
up
where
it
is
there
we
go
give
you
calls
are
recorded
for
anyone
who
doesn't
know
if
you
don't
want
to
be
recorded.
Sorry
we're
going
to
do
that
so
that
we
can
share
them
with
more
of
our
community
and
thank
you
to
everyone
who's
willing
to
be
here
anyway
and
present.
A
Also,
we
do
follow
a
code
of
contact
conduct
at
jupiter.
I
think
people
probably
know
this,
but
I
also
think
it's
worth
bringing
up
at
the
start
of
calls.
So
here
we
go.
It
can
always
be
found
at
jupiter.org
conduct
and
we're
gonna
get
started
so
today
I'm
super
lucky.
I
managed
to
get
three
presenters.
I
need
to
stop
sharing
my
screen
and
oh,
do
we
start
with
shout
outs
said
or
do
we
just
do
them
on
the
agenda?
Does
anyone
have
thoughts
on
that.
A
Thank
you.
We're
gonna
do
our
best
to
make
them
so
that
other
people
can
do
them
really
easy
in
the
future.
So
awesome.
Thank
you
so
much.
I
appreciate
the
thought
and
first
item
on
the
agenda
then
getting
into
it.
We
have
a
share
from,
I
believe
it's
nick
with
ipai
elk
2.0.
So
I'm
happy
take
it
away.
You
ready.
C
All
right,
something's
probably
happening
this
is
a
little
tool.
We've
been
putting
together
it's
up
on
github
stuff,
on
counter
forge
up
on
pipeline.
It's
up
on
npm,
because
we
have
to
publish
in
50
places
the
basic
motivation
behind
knockpail
is
being
the
layout
large
graphs
that
have
complicated
relationships
to
each
other.
So
this
guy.
C
A
little
screenshot
and
I'll
get
into
that
demo
in
a
sec,
but
you
can
see
you
know.
I've
got
a
lot
of
layers
of
nesting.
I've
got
different
kinds
of
gazentas
and
gazadas,
maybe
got
some
interactivity
and
some
custom
animations
that
I
want
to
do,
and
I
want
all
those
things
to
be
interactive
and
responsive.
So
we
started
down
this
road.
We
tried
some
other
things.
First
set
escape,
did
some
stuff
with
data
shader
and
they
just
weren't
doing
it
for
us.
C
So
we
embarked
on
this
quest
on
some
money
after
georgia,
tech.
So
I'm.
C
The
money
one-
this
is
a
graph
that
I'm
creating
in
network
x,
I'm
overloading
a
bunch
of
attributes
and
we
can
watch
it
start
doing
stuff.
And
you
know
it's:
we've
got
some
data,
that's
driving
a
chart.
We've
got
a
data,
that's
driving
this
interactive
graph,
things
go
into
and
come
out
of.
The
graph
edges
disappear
when
they
can
no
longer
occur
and
it's
a
terrible,
terrible
ecosystem,
because
you
know
much
like
in
game
of
thrones,
where
the
ghost
grass
will
consume
all
other
life
forms.
Eventually,
the
grass
just
wins.
C
You
just
watch.
You
just
watch
those
wolves,
they're,
not
lucky
enough
and
eventually
so
we,
you
know,
we've
got
normal
stuff
in
here
that
you
would
put
in
the
simulation.
So
I
can
turn
off
this.
I
can
turn
off
this
and
I
can
crank
the
speed
all
the
way
down
and
I
guarantee
that
the
ghost
grass
will
win
turn
that
back
on
there
we
go
and
then
the
ground
is
covered
in
grass.
You
can
see
how
it
plays
out
here.
So
these
are
the
kinds
of
things
that
we
build
for.
Folks.
C
You
know
simulations
surrogate
models,
you
know
maybe
stuff,
that's
up
on
the
hpc
pipe
and
we
needed
some
pretty
things
that
gave
us
some
power
towards
it.
Some
of
the
other
things
that
we
do
are
in
my
examples,
which
I
could
have
done
smartly,
through
my
notebook
view,
without
having
to
leave
single
integra,
you
know
the
really
basic
stuff
is
you
can
load
in
their
particular
flavor
of
of
else
of
json,
which
is
from
the
the
eclipse
project?
C
C
But
you
know
you
get
your
interactive
control
of
it
plays
nice
with
output
widgets
and
you
know
so.
These
are
the
sharing
the
same
model
so
that
when
I
hover
they're
getting
the
same
thing
but
they're
different
viewports,
we
don't
have
you
know
what
is
visible
in
how
many
viewports
or
things
like
that,
but
we'll
probably
get
there.
C
So
all
the
all
that
viewport
stuff
is
controllable
from
the
animation.
So
we
can
fit
all
the
way
down
to
you
know
just
something
we
got
there.
We
can
fit
to
just
kernel.
We
can
select
stuff
interactively
so
now.
If
I
click
here,
you
can
see
that
that's
getting
updated.
So
we
got
some
of
the
nice
things
here.
We
can
automatically
select
kernel.
We
can
select
multiple
things,
so
I
leave
my
command
keyboard
in
there
and
we
can
clear
things
out.
We
know
when
things
are
hovered.
C
Sprotty
root
is
a
thing
which
is
actually
kind
of
nice
to
know
when
it's
being
hovered,
and
you
can
also
set
that
programmatically.
So
we
can
crank
it
up
right
to
you
know
the
they
link
together.
That's
not
entirely
surprising,
you
can
do
some
fun
things
with
changing
the
viewport.
C
You
know
so
anything
that
I
click
on
here.
That's
gonna,
hover
too,
or
that's
going
to
highlight
too
over
there
there's
some
stuff,
that's
really
hard
to
make
with
their
json.
So
we
made
a
higher
order
thing
on
top
of
it
called
the
app
which
is
a
bad
name,
but
here
we
are,
and
it
supports
things
like
calculating
how
to
collapse
nodes,
and
it
makes
power
edges
out
of
things
that
it
finds
there
so
come
on,
come
on,
so
we
can
collapse
stuff
down.
So
this
gets
really
important
for
us
we're
building.
C
Sometimes
things
with
you
know
thousands
of
nodes.
If
you
want
to
expand
everything,
but
maybe
the
important
story
can
be
told
a
little
bit
easier
and
we've
got
some,
you
know
we
can
put
some
icons
in
there.
Instead
of
that,
the
custom
close
button
array-
ascii
art-
you
know,
so
we
can
close.
This
toggle
collapsed,
close
the
button.
Yeah
we
closed
it,
but
then
what
we
saw
in
that
other
example
is.
We
can
do
really
aggressive
styles
to
different
parts
of
it
and
it's
still
pretty
raw.
C
You
kind
of
got
to
know
what
you're
doing
you
know.
So
we
got
stuff
that
only
happens
when
something
has
collapsed.
You
can
indicate
that
there's
a
collapsed
edge
there
and
you
kind
of
got
to
dig
around
through
the
dom,
so
we're
going
to
be
working
on
some
stencils
and
things
to
do
that.
You
can
also
set
it
a
little
bit
more
pythonically.
C
So
this
is
just
some
examples
that
it
can.
It
can
apply
to
there
we
got
x,
we
we
have
svg
exporting,
so
that's
pretty
fun.
A
browser
does
have
to
be
in
the
loop
which
is
tiresome,
but
here
we
are
so
you
know.
If
I
open
this
in
the
new
window,
it
embeds
all
the
svg.
You
get
everything
you
can't
do
subsets.
You
can't
do
a
lot
of
things.
C
It
does,
for
whatever
reason
bring
along
the
selection
state.
So
you
you
can
actually
highlight
stuff
like
that,
so
that's
kind
of
fun,
I'd
love
to
have
a
way
that
this
worked
entirely
offline,
but
we're
not
up
to
that
task.
Yet
yeah,
that's
really
about
it.
There's
a
there's
a
lot
of
other
complexity
inside
of
here
that
that
we
needed
for
some
some
different
kinds
of
tools,
but
the
performance
is
good
enough
for
our
use
case.
C
We
can
crank
it
to
about
a
thousand
some
nodes
with.
You
know
twice
that
number
of
ports
and
edges
between
them
and
it
kind
of
does
the
job
and
then
the
only
other,
exciting
thing.
Oh,
it's
gonna
break
in
binder
because
I
didn't
put
my
x11
headers
in
there
yeah,
let's
at
least
see
it
break,
we've
been
trying
to
not
do
a
lot
of
heavy
unit
tests
and
whatnot.
So
most
of
this
is
all
driven
by.
C
I
don't
have
it
in
here.
I
don't
have
the
interaction
most
is
driven
by
acceptance
tests
that
we
do
against
the
browser.
So
you
know
all
those
little
activities
that
you
saw
up
on.
There
were
slowly
adding
the
the
rigor
to
make
sure
that
that
stuff
keeps
working
and
that's
really
important
for
us,
because
we,
you
know,
we
go
to
the
effort
of
shipping,
something
and
then
it
breaks
for
us
immediately
downstream.
We
don't
really
have
time
for
that.
C
C
Noise,
I
will
look
at
the
chat,
so
this
is
ipad,
widgets
and
hall
views.
There
are
no
hollow
views.
There
is
no
bq
plot.
Are
the
examples
tested?
Yes,
what
do
you
need
to
know
to
use
elk
network
x
and
yes,
there's
import
nb?
So
those
are
those
examples
and
those
were
all
very
leading
points
that
tony
threw
in
there
because
he's
kind
of
a
jerk
like
that.
C
Got
you
know
we
got
ci
we
use,
we
use,
do
it
to
manage
all
of
it,
because
it
is.
You
know
on
like
a
bunch
of
different
buses.
We
do
use
conda
to
install
everything
because
there's
a
lot
going
on.
You
know
really
the
key
build
step.
Is
I'm
not
logged
in
so
I
can't
see
what
those
logs
are,
but
it
really
just
fires
up
our
our
dodo
pie
and
it
does
a
monstrous
amount
of
things.
C
There's
a
lot
there's
a
lot
going
on
here,
but
it
works
and
it
helps
us
release
software.
So
I
don't
care
and
it
helps
us
when
we
step
away
from
it
for
a
while
and
to
know
that
it's
just
gonna
keep
working.
You
know
the
jail
up
output
area.
Do
you
identify
any
other
ways
to
interact
with
the
jlab
interface
structure?
When
writing
widgets?
I
try
and
have,
unless
they're,
very
special
widgets.
C
C
So
I
can't
think
of,
I
can't
think,
a
way
of
a
way
that
elk
would
want
to
use
lab,
but
I
can
think
of
a
lot
of
ways
that
lab
would
want
to
use
elk.
So
if
we
got
it
to
a
mime
renderer
at
some
point
that
took
all
these
these,
who
gaws
and
what's
it
and
what
not
work
in
there.
The
other
kind
of
weird
thing
anybody
that's
played
in
the
debug
adapter
space
or
the
language
server
space.
C
C
This
is
all
we
had
to
do
to
implement
their
fake
stuff
right,
but
that
that
actually
could
be
very
powerful
in
the
long
run,
so
long-running
jobs
give
me
a
graph
of
that
installation
procedures,
environments,
all
those
kinds
of
things
we
got.
We
have
some
issues
open
for
for
kind
of
things
that
we're
interested
in
yeah
right
here
for
the
dog
food,
you
know
so
yeah
source
code
it'd
be
great.
I
just
throw
up
a
tree
of
my
whole
project
and
I
can
jump
to
where
the
file
is.
That
sounds
fun.
C
Architecture
diagrams.
We
need
those
everywhere
in
jupiter.
How
does
my
github
actions
work?
How
do
my
automation
processes
work?
What
do
I
have
installed
show
me
some
profile
stuff.
I
mean
I've
been
doing
these
graph
things
for
you
know
on
and
off
for
years.
So
we're
we've
been
thinking
about
a
lot
of
this
stuff,
but
we
don't
have
any
of
it
really
spun
up
yet
yeah.
C
There's
so
many
to
choose
from
if
you
ever
go,
look
into
the
network
x,
read
write.
That
is
true.
You
know,
so
the
things
that
you
find
in
there.
So
you
know
it's
graph
links
one.
If
you
ever
did
d3
right,
that's
what
you
always
find
in
there
and
why'd
you
find
them
in
there
because
they
came
out
of
network
x
or
or
whatever
that
that
particular
lineage
is
lost
to
time.
Cytoscape
json
is
valuable.
C
This
elk
json
is
kind
of
a
second
class
citizen
to
their
java
implementation
so
whatever,
and
that
is
not
currently
supported
by
network
x,
but
that
would
be
a
really
sweet
pr
if
we
had
a
better
way
to
run
this
in
python,
and
not
only
in
the
browser
but
yeah
I
mean,
like
we've
done,
work
with
rdf
and
turning
it
into
those
formats
and
and
a
bunch
of
other
things.
I
mean
that
once
you
say,
network
graph,
like
everybody's
network
graph,
looks
very
different
right.
Is
it
nested?
C
A
D
It's
the
jerk's
turn,
apparently
thanks
nick
well,
let's
see
what
am
I
gonna
do
here?
Not
the
you
know.
I
broke
the
demo
this
morning,
but
I'm
going
to
share
the
intent
of
this
work.
D
So
dgaf
is
the
deathbeds
generalized
automation
framework
and
the
idea
behind
it
is
that
it
should
be
easy,
easier,
simpler
to
use
your
python
tools
within
different
environments,
whether
that
be
a
local
development
environment,
a
github
action
or
a
jupiter
hub,
for
example-
and
you
know,
I
think
all
a
lot
of
us
have
a
lot
of
ideas
and
it
gets
really
hard
to
manage
projects
at
scale,
and
some
people
are
better
at
it
than
others.
D
So
I
think
that
there's
an
opportunity
to
build
a
lot
of
publishing
content
from
partial
information
and
repositories
here,
so
the
deathbed,
generalized
automation
framework
goes
and
does
some
of
these
things
for
us,
so
it
does
inference.
So
let's
say
you
had
a
bunch
of
markdown
content
and
some
notebooks
in
there.
What
we're
going
to
do
is
we're
going
to
go
and
discover
the
dependencies
that
you
reference
and
then
we'll
build
your
requirements.
D
File
we'll
build
your
pi
project
file,
we'll
build
your
content
environment
and
then
you
can
use
these
things
on
the
different
services.
When
we
think
about
installing
things.
You
know
I
work
locally
in
conda,
but
when
I'm
deploying
to
give
actions,
I'm
working
in
a
pip
place
so
using
environment
variables
and
files
that
exist,
we
should
be
able
to
manage
how
environments
are
installed
easier.
We
should
be
able
to
develop
and
install
stuff
easier
and
having
a
higher
level
api
over
top
of
the
tool.
D
Churn,
that's
happening
in
python
specifically
on
these
is
really
important,
and
then
we
should
just
be
able
to
test
stuff.
We
should
be
able
to
generate
docs
and
I'm
doing
this
with
jupiter
book
and
jupiter
books
awesome.
I
love
doing
that.
I
love
using
jupiter
book
for
this
part
and
then
we
have
post
build.
D
This
should
actually
be
binder
now,
but
the
idea
behind
this
is
that
when
we
go
to
a
binder
binder
recognizes
our
post
build
file
and
what
we'd
want
to
do
is
we'd
want
to
just
do
dgaf
build
me
a
binder,
and
that
would
go
and
infer
your
environment.
It
would
update
your
content,
environment
it'll,
go
and
install
your
setup
tools
or
install
your
pi
project,
but
it
should
be
a.
I
think.
Binder
becomes
a
little
bit
more
interesting
of
utility.
D
If,
right
now,
we
we've
proven
that
we
can
do
reproducible
work
on
binder,
but
if
we
can
go
further
and
say
hey
what,
if
binder
actually
set
up
as
a
development
environment,
and
now
somebody
can
go
and
use
it
and
do
development
like
things
and
actually
potentially
make
changes
and
figure
out
how
to
upstream
them
afterwards.
D
So
we've
got
our
post-build
phase
here,
and
the
idea
is
is
that
we
want
to
build
off
of
partial
information.
So,
for
example,
if
we
were
to
start
with
a
pi
project
tomo
like
this,
it's
got
some
extra
entry
points
and
stuff
like
that,
but
it
doesn't
have
the
key
names
that
we
need
to
generate
stuff.
So
if
I
go
over
into
a
cli
here,
actually
this
is
very
unbecoming,
but
and
we
look
at
git
right
now.
D
Let
me
revert
this
file
for
the
sake
of
the
demo
and
then
where'd
my
console
go
all
right
there.
It
is
okay,
good
all
right.
So
now,
if
we
go
and
do
dgf,
djif
infer,
we're
yeah,
that's
a
superfluous
output.
We
updated
our
pi
project
here
and
now
it's
updated
and
this
will
work
under
a
flit
install
and
that's
a
nice
feature.
D
It
used
to
generate
a
conda
environment.
I
must
have
broken
that
along
the
way.
But
oh,
no,
if
there's
a
cond
environment,
condo
environments
are
generated
and
it
actually
goes
and
does
a
solve
in
the
middle
wakanda,
and
then
the
allows
it
to
split
it
up
with
the
pip
dependencies
and
the
conda
dependencies
here
and
there's
some
nice
mapping
tools
that
depth
finder
has
that
lets
me
stay
updated
with
these
things
yeah.
This
is
not
a
good
environment
right
now.
D
I've
gotta
fix
something
clearly,
but
then
yeah,
so
some
partial
information
and
then
when
we
go
to
something
like
github
actions,
what
we
would
do
is
we
would
do
our
pre-install,
this
installs,
our
environment.
We
do
our
install
this
builds
our
package
and
then
we
go
and
run
our
test
and
without
having
too
much
opinion
and
only
applying
config
generating
configuration
files.
D
We
can
do
a
lot
with
a
little
bit,
so
that's
djf
and
one
last
thing
is:
I
am
very
happy
with
how
I've
written
it,
and
even
this
is
new
to
nick-
and
we
talked
about
this
last
night,
but
djif
is
running
a
lot
of
shell
commands.
So,
first
of
all,
this
is
our
readme
file
here
and
we've
got
some
imports
and
we've
got
some
all
saying
these
are
the
things
that
are
coming
out
of
it.
D
Then
we've
got
our
infer
function
and
that
is
going
and
create
reading
our
dependencies
and
creating
our
requirements
when
needed,
and
it's
got
some
flags
and
some
logic
here,
but
then
what's
fun
is
when
we
get
down
to
things
that
are
actually
running
shell
commands.
This
document
is
written
in
a
literate
form
in
co
that
uses
conch
also.
So
basically,
this
actually
works
as
my
code,
and
we
can
see
some
more
fun
shell
commands
here,
but
this
is
still
working
in
progress.
It's
just
an
idea.
D
If
anybody
has
any
feedback,
I'd
love
to
hear
it
and
the
business
sort
of
happens
here
in
so
nick
had
import
mb
in
his
demo.
This
is
an
overloaded
version
of
import
mb
that
actually
lets
me
go
and
import.
The
readme
markdown
document
with
the
conscious
syntax
and
the
hybrid
ipython
syntax
in
there-
and
I
actually
am
right
like
dgaf-
is
a
literate
program.
D
A
D
Yes,
several
times
until
I
break
it
intermediately
and
then
I
then
I
solve
it
again,
but
yeah
djif's
at
the
point
where
it's
almost
working
on
github
actions.
It
works
on
binder,
it
works
locally.
So
like
short-term
goals,
it's
getting
really
close
and
then
hopefully
adding
more
opinions
to
it.
A
Okay,
I'm
also,
I
forgot
to
post
the
agenda
in
the
chat,
so
I
just
did
that
sorry
about
that
everyone,
if
you
were
looking
for
it,
it
was
hiding
on
the
discourse
post
all
along,
but
we
actually
have
one
more
share
today
from
eric,
and
I'm
super
excited
to
hear
this
one.
What
is
it
jovi
and
eric
ready
to
take
it
away.
G
Yeah,
this
is
exactly
my
question.
I
didn't
know
I
I
don't
know
what
the
jovian
is.
The
first
time
I
I've
seen.
That
word
is
when
I
I
launched
a
docker
container
with
the
notebook
I
connected
to
the
container,
and
my
surprise
was:
oh,
I'm
not
root,
I'm
I'm
not
I'm
not
ubuntu,
I'm
jovian
and
I
typed
jovian
in
google
and
it
was
not
an
existing
world.
G
I
don't
think
it
is
so
I
I
I've
been
driven
to
some
good
issues
or
stack
of
the
overflow
questions
and
also
to
the
jupiter
main
documentation.
I
think
so.
This
is
the
link
I
have
put
there
and
it
looks
like-
and
this
is
my
share-
it
looks
like
we
are
all
jovian,
but
what
what
who
are
we
actually
and
in
jupiter
ecosystem?
G
Sometimes
you
see
ikey
people.
Sometimes
you
see
phd
people.
Sometimes
you
see
designer
and
so
on.
So
it's
very
very
diverse.
So
this
was
my
share.
I
guess,
and
if
you
didn't
know
what
jovian
was
so
you
know
at
least
it
is
some
user
of
the
jupiter
ecosystem,
something
like
that.
F
E
And
I
think
we've
we've
always
encouraged
and
been
really
happy
with
the
the
puns
and
plays
with
the
around
the
the
planet.
Name
right
in
the
sense
that
the
the
original
motivation
behind
the
name
was
actually
a
conversation
and
a
quote
around
julia
python
and
r
being
the
three
open
languages
of
data
science
that
inspired
the
name
not
meant
as
an
acronym
but
meant
as
an
inspiration.
E
But
there's
there
was
always
the
connection
to
galileo
from
day
one
and
and
to
to
his
discovery
of
the
moons
of
jupiter
and
to
the
notion
both
of
the
the
scientists
who
built
builds
their
own
tools,
but
also
the
long-running
connection
to
astronomy
and
the
scientific
python.
Community
astronomers
were
some
of
the
earliest
adopters
of
python.
E
The
astronomy
community
has
been
very
open
to
sort
of
tool
building
in
science
rather
than
only
doing
science
and
giving
the
people
who
build
the
tools,
sort
of
respect
and
recognition,
partly
because
obviously,
astronomy
can't
be
done
without
building
telescopes
and
building
big
big,
complicated
tools.
E
So
and
so
there
there's
there's
multiple
threads,
both
in
the
history
of
science
and
the
history
of
python
and
the
sci-fi
community
and
folks
like
perry
greenfield
are
the
space
telescope,
whose
work
funded
early
developments
of
matpot,
lib
and
early
developments
of
namurai,
which
led
to
numpy.
So
there's
kind
of
multiple
reasons
why
we
have
these
connections
and
that
that
naming
pattern
has
picked
up
a
lot
of
momentum.
E
Some
of
you
may
have
seen
the
syzygy
project
in
canada,
for
example,
to
offer
jupiter-based
resources
to
hp
on
hpc
infrastructure
in
canada.
So
well,
syzygy
is
a
planetary
analogy
planetary
term.
The
the
kalisto
project,
also
in
canada,
to
offer
k-12
educational
resources
in
jupiter
again
is
a
name
kind
of
drawn
drawn
from
this,
and
it's
it's
kind
of
fun
to
see
to
see
that
intersection
of
naming
puns
around
the
history
of
astronomy
and
our
our
place
in
the
solar
system
and
whatnot
and
people
find
ways
to
use
the.
E
Why
sort
of
misspelling
in
to
give
them
a
twist
and
to
make
them
now
the
downside
is
sometimes
you
can't
find
these
things.
The
good
side
is,
they
have
a
basically
a
disjoint
search
space
right,
which
was
very
handy
when
we
were
establishing
the
jupiter
name
in
the
community
like
well.
You
know
what
that
word
actually
doesn't
exist,
and
so
there's
a
nice
and
clean.
You
know
in
a
world
that
is
overcrowded
with
overlapping
names
and
it's
hard
to
find
a
unique
term
that
isn't
some
sounds
like
incomprehensible
gibberish
to
anyone.
E
This
actually
ended
up
working
quite
well
for
us.
So
that's
that's
a
little
bit
of
sort
of
backs
long-running
commentary
from
from
the
early
days
of
the
name.
A
D
C
We
were
walking,
we
were
walking
back
to
the
yeah.
We
were
walking
back
to
my
weird
catholic
hostel
and
just
talking
about
like
how
it
was
mathias's
idea
he's
like
we
need
a
mascot,
because
I'm
tired
of
people
using
the
jupiter
trademarked
logo
that
we
have
to
maintain
protect
or
we
lose.
C
We
needed
a
we
needed
a
toy.
We
needed
a
figure
that
could
be
used
very
flexibly
and
we
were
talking
about
like
that's
really
hard,
because
what
does
you
know?
What
does
the
jovian
do?
What
is
and
who
is
it
jovian?
What
do
they
look
like
and
we
ended
up
with?
We
ended
up
with
this.
This
space
space
person,
space,
robot
space
monkey
space,
whatever
that
could
hide
inside
of
those
things
and
do
all
these
fun
things.
A
A
So
that
is
actually
what
we
have
on
the
agenda
for
today
and
it
went
faster
than
I
thought
it
would,
but
yeah
we
can
continue
discussion
on
any
of
those
if
anybody
has
anything
that
they
want
to
show
totally
off
the
cuff
there's.
Never
any
pressure
with
these
about.
I
don't
know
how
formal
they
should
be.
We
just
love
to
have
people
share.
So
if
there's
anything,
people
want
to
throw
in
the
ring
here.
D
I
I
can
throw
something
in
there.
I'm
gonna
do
I'm
gonna
off
the
cuff,
something
I
I
always.
I
love
that
I
I
have
thank
you
for
doing
this.
I
like
yelling
at
people
if
anybody
wants
to
leave,
go
ahead,
but
there's
this
line
of
code
that
I
want
to
show
you
all
all
right.
So
I'm
just
going
to
run
this,
and
I
should
return
me
a
data
frame
here.
Oh
I
don't
have
pandas,
that's
crazy.
D
I
got
a
new
mini
condo
the
other
day
and
man
all
right,
import,
pandas.
Now,
we've,
okay,
great!
Oh,
that's
why
import
pandas!
It
actually
needs
pandas
there,
all
right
df!
So
now
what
this
is
here
if
we
go
to
workflow
runs,
I
didn't
know
this
part
of
the
github
api
existed
until
recently
pandas.series.
D
So
now
this
gives
us
a
data
frame
and
what
this
is
is
this
is
the
github
api
returning
you
information
about
your
github
actions
and
your
workflow
runs.
So
like
it's
a
conclusion.
It
was
a
failure.
What's
the
status
so
you'll
actually
get
get
information
about
intermediate
runs
while
they're
still
going
on.
It's
got
date,
time,
information
there
are
links
to
log
files
and
stuff
like
that.
So
this
is
just
a
pretty
interesting
trove
of
data
that
I
wasn't
aware
of.
D
So
basically,
the
form
of
the
end
point
here
is
your
repo's
and
then
your
organization,
your
project
actions
and
then
runs,
and
it
gives
you
all
the
workflow
runs
that
have
happened.
I
think
it's
up
to
a
hundred,
but
you
can
paginate
through
that
and
yeah.
I
thought
some
folks
might
be
interested
in
this
if
they're
doing
a
lot
of
build
work
and
want
to
introspect
it
at
some
point
ever
so
that
was
my
very
very
quick
demo
here.
D
D
But
it's
good
data,
it's
pretty
cool
data,
especially
if
you
can
just
dump
it
into
some
like
graph
or
diagnostics,
and
you
know
export
an
artifact
or
something
it's
kind
of
a
nice
tool.
A
Great,
I
don't
want
to
keep
everyone
if
there's
not
further
discussion
to
be
had
a
few
wrap
up
things
first.
I
do
want
to
thank
everyone
for
being
here,
because
I
really
wasn't
expecting
this
many
people
at
all
to
be
to
be
honest
with
you
or
this
many
shares.
So
thank
you
and
thank
you
for
the
enthusiasm.
A
I
hope
we
can
keep
carrying
it
on
next
month.
Call
is
december
15th,
that
is
on
the
calendar,
but
I
will
be
promoting
it
again
and
if
you're
interested
in
presenting,
I
think
I'm
just
going
back
to
throwing
it
on
the
agenda,
because
that's
what
people
did
anyway,
so
I
will
make
that
appropriate
discourse
comment
and
the
agenda
will
be
on
there
and
yeah.
A
If
you
have
any
interest
in
sharing
things
in
the
future,
please
do,
as
you
saw
here
like
right,
doesn't
have
to
be
whatever
you
think,
polished,
technical,
anything
that
you
would
expect
here.
I
think
anything
involved
in
the
jupiter
jupiter
ecosystem
would
be
awesome
to
hear
about
and
yeah.
Thank
you
so
much
isabella.
Do
you
have
an
exit
form
I
do,
but
I
need
to
throw
it
is
the
chat
the
best
place
to
throw
it.
D
Let's
put
in
the
chat
and
let's
also
put
it
in
the
hack
md,
so
we
can
keep
making
the
converter.
It
would
be
really
nice
if
you
all,
could
give
us
some
feedback
about
how
today
went
and
yeah.
That
would
be.
E
Thank
you
so
much
isabella.
This
is
fantastic
and-
and
I
really
appreciate
it-
I
I
I
was-
I
didn't
realize
that
the
community
call
was
happening
today.
It's
my
bad,
I
had
lost
track
of
it
and
I
sort
of
hopped
on
to
the
the
the
governance
documentation
work,
but
this
was
the
best
surprise
of
the
day.
So
this
was
wonderful
and
I
really
really
appreciate
what
you
folks
are
doing.
This
is
wonderful.
A
E
D
E
Oh,
that,
actually
that
actually
exists.
One
thing
you
know:
one
thing
that
could
be
cool
would
be
the
demo
I
put
it
up
online
a
while
back.
I
made
a
copy.
E
That
was
a
little
hacking
exercise
at
brian's
place
in
march
2010.
I
just
drove
down
for
a
week
into
his
place
in
san
luis
obispo,
and
what
we
did
was.
It
was
kind
of
cool
because
we
were
hacking.
This
was
using
zero,
mq
brian
had
just
written
up
syphon
bindings
for
it,
so
we
could
use
it
in
python
and
we
didn't
use
ipython
proper.
E
If
we
were
playing
with
something
on
the
client
side,
we
would
keep
like
killing
it,
making
edits
and
restarting
it,
but
leaving
leaving
the
server
up
on
the
other
laptop
and
just
like
it
would
reconnect
and
zero
and
q
was
seamless.
So
it
was
a
really
fun
experience
to
play
with
that
and
the
code
is
lying
around.
So
one
thing
we
could
do
is
look
at
the
old
ipython
and
look
at
this
like
baby
micro
prototype
of
what
became
sort
of
what
became
my
python
and
what
became
the
jupiter
architecture
they're
both
on.