►
From YouTube: Jupyter Community Call - November 29, 2022
Description
Recording from the Jupyter Community Call in November 2022.
The notes from this call can be found here: https://github.com/jupyter/jupyter/issues/670
Read more about these calls on Discourse: https://discourse.jupyter.org/t/jupyter-community-calls/668/
A
Yeah,
hello,
everyone
and
welcome
to
the
November
Jupiter
Community
call
November
2022
I
need
to
remember
to
say
years,
I'm,
Isabella
I'll,
be
your
host
today
and
yeah
I'm,
so
excited
to
get
started.
I
love,
seeing
people
on
this
call,
I
love
seeing
things
on
the
agenda,
so
thank
you
so
much
before
we
get
any
further.
I
am
recording
right.
Okay,
yes,
before
we
get
any
further,
there
are
a
few
things
I'd
like
to
remind
you
of
so
one.
A
The
call
is
recorded,
that's
and
it
will
be
posted
publicly
on
YouTube,
whatever
that
means
to
you
just
letting
you
know
two.
This
is
a
Jupiter
Community
event
and
as
so,
we
are
held
to
the
Jupiter
code
of
conduct,
everyone
here
that
includes
me
that
can
be
found
at
any
time
at
jupiter.org
conduct
and
finally,
correct
me:
if
I
pronounce
your
name
wrong,
I'm,
trying
to
learn
and
be
better,
so
I'd
really
appreciate
it.
A
I'd
like
to
show
you
proper
respect
without
further
Ado
Jupiter
Community
call,
if
you
haven't
been
here
before
our
goal
of
today,
is
to
share
and
celebrate
across
Community
things.
People
are
working
on
around
Jupiter,
whether
that's
something
they
made
something
they
use,
something
they're
just
excited
about,
or
even
questions
they
have
for
the
community
and
what
we're
going
to
be
doing
is
following
the
agenda
that
I
linked
okay,
the
participant
number
has
gone
up,
though
so
one
more
time
for
anyone
who
showed
up
feel
free
to
sign
in.
A
We
will
be
following
that
meeting
we're
going
to
go
through
the
short
reports,
which
are
things
without
discussion
and
Main
agenda
items
and
kind
of
just
leave
that
room
for
discussion
right
now,
I'm
seeing
three
things
on
the
agenda
that
should
be
pretty
feasible,
but
I
will
try
and
watch
the
time.
So
with
that
I
think
we're
gonna
get
started
for
the
short
reports
celebration.
A
A
Okay,
then,
in
that
case,
I'll
do
the
cheesy
shout
out,
thanks
to
all
of
you,
I
know
we
got
some
different
time
zones
today.
I
think
it's
kind
of
an
awkward
time
for
for
many
of
the
participants,
so
really
appreciate
you
making
the
time
to
be
here
and
then
we're
going
to
go
to
agenda
items.
Gabriella
I
see
you're.
First
on
the
agenda,
are
you
ready
to
share.
A
B
What
I
wanted
to
share
with
you
today
is
a
usability
study.
I
come
back
to
the
long
Jupiter
lab,
so
I,
don't
know
for
those
who
I
haven't
met
before
I'm
a
ux
designer
and
researcher,
and
working
with
quantstec
and
I
did
usability
study
on
some
common
features.
B
So
I'm
not
going
to
go
through
the
whole
thing,
because
it's
a
bit
long
but
just
go
through
I'm
gonna,
go
through
the
main
results
and
then
explain
what
I
was
thinking
of
doing
after
and
how
people
could
participate
in
this
by
having
a
look
at
the
issues
I
created
on
GitHub
that
are
related
to
the
results
from
this
study.
B
So
mainly,
the
goal
of
this
is
to
take
do
little
ux
improvements
on
Jupiter
lab
that
are
easy
to
do
and
can
bring
a
lot
to
the
users
and
to
not
do
them
because
I
or
someone
has
thinks
there's
a
better
way,
but
really
because
it's
user-centered
and
user
validated
by
this
by
a
study
or
by
any
other
kind
of
study.
C
D
B
So
it's
so
basically
it's
taking
participants
through
Alabama
sort
of
tasks
while
telling
them
a
story
around
these
tasks
and
are
they
frustrated
by
some
things?
Do
they
expect
something
to
happen,
and
it
doesn't
happen
and
to
cross
the
information
between
several
participants
and
see
if
something
emerges
all
the
time
and
or
was
it
just
a
particular
case
of
this
person,
and
in
that
case
we
can't
count
it
as
a
result,
so
I'm
going
to
jump,
we
can
come.
B
B
So,
who
will
talk
to
I
have
two
kinds
of
science
to
user,
the
beginner
user,
being
people
who
have
been
using
Jupiter
lab
for
less
than
two
months
on
a
regular
basis,
but
very
low
level
like
first
year
or
second
year,
students
and
they
experienced
user
is
more
teachers
that
have
been
using
it
for
more
than
10
years.
So
this
is
information
about
the
beginners
persona,
and
this
is
about
the
experienced
user
so
and
they
experience
of
more
than
10
years
on
digital
lab.
B
Unfortunately,
for
the
study
there
is
a
lot
more
participants
in
versus
two,
so
most
of
these
results
concern
the
beginner
user.
B
The
protocol
that
we
go
through
is
I,
introduce
myself
and
the
context
of
the
study.
Then
I
explain
what
is
expected
from
them
or
not
like
I.
Don't
expect
them
to
say
if
they
find
things
that
look
nice
or
not,
but
I
expect
them
to
really
do
the
tasks
I
ask
to
need
to
and
there's
never
a
wrong
answer
to
anything,
and
then
we
did
the
user
testing
so
for
that
I
shared
with
them
a
prototype,
I
create
which
looks
like
Jupiter
lab
but
isn't
Jupiter
lab,
and
so
not
everything
is
clickable.
B
B
So
I
want
to
go
through
the
limitations
that
they
were
a
couple
of
ones
concerning
so
the
first
task
that
I
asked
participants
was
to
open
a
notebook
that
had
been
previously
shared
with
them.
So
this
wasn't
clear
at
all
on
the
Prototype.
The
Prototype
here
had
two
tabs
like
the
files
all
and
the
files
shared
with
me,
so
it
wasn't
at
all
understood
all
right
on
the
Prototype.
There
was
also
a
way
to
see
other
people
that
collaborate
on
the
same
notebook.
B
B
Displaying
text
was
very
to
create
cells
below
above,
like
collaborators,
and
so
this
had
a
couple
of
problems
links
to
mainly
the
design
and
the
UI
of
what
I
had
done
in
the
Prototype,
so
I
I'm
going
to
document
all
this
I
am
I've,
actually
put
a
couple
of
wedding
issues,
and
so,
if
you're
interested
in
Sun,
you
can
go
and
look
more
detail
and
what
are
the
results
and
what
do
I
suggest
to
improve
it?
B
How
do
people
expect
to
move
us
out
this
also,
it
was
understood,
but
it
created
some
frustrations
on
because
when
you
move
a
cell
up
using
this
icon,
for
example,
it
holds
that
we
move
it
up
a
second
time
and
then
again
and
again,
and
it
can
be
really
frustrating
when
you
want.
You
have
a
long
notebook
with
lots
of
cells,
and
you
want
to
take
your
cell
all
to
the
top
or
to
the
bottom,
for
example.
B
For
example,
I've
tried
to
make
them
as
small
as
possible,
so
they,
for
example,
here,
is
just
creating
a
disabled,
State
full
move
up
and
move
down
icons
on
the
Uber.
They
don't
exist
today
and
it's
just
a
little
ux
principle
that,
even
though
it
could
seem
obvious,
it's
nice
to
not
have
a
button,
that's
enabled
when
actually,
you
can't
click
on
it
like
when
a
cell
is
or
all
the
way
things
like
that
that
I've
been
putting
on
the.
B
A
D
That's
cool
we
need
more
of
this,
were
you
testing
features
that
not
everybody
has
some
of
that
stuff
looked
a
little
bit
unfamiliar?
Are
you
all
testing
4.0
or
like
a
late
3x.
B
C
B
Couple
of
new
things
on
it,
and
so
it's
a
lot
quicker
than
actually
creating
it
for
a
wheel,
and
then
we
can
test
them
and
before
actually
like
implementing
them,
we
can
do
quick
tests
so
yeah.
It
was
things
that
don't
exist.
D
D
Folks
doing
design
work
in
that
space
are
going
to
need
a
lot
of
help
and
allies
in
it.
It's
I
still
believe
there's
a
lot
of
fundamental
research
in
this,
like
interaction
that
we're
doing
that
is
going
to
be
a
challenge
so
yeah.
This
is
exciting
to
see
and
it
seems
really
challenging
to
scope,
but
I'm
glad
that
people
are
talking
about
this
in
the
community.
Hopefully
we
see
more
of
this
stuff.
E
Not
really
I
mean
okay
parks
on
the
sidewalk
yeah
I
mean
like
there
there's
a
couple,
there's
a
couple
box
that
we
did,
that
you
and
I
you
know,
went
back
and
forth
on
a
bit.
I
think
it
was
specifically
in
the
context
of
the
read-only
cells
and
it's
just
they're.
So
when
you
get
to
these
documents
that
are
not
that
are
not
playing
text,
there's
so
many
places
to
hide
things
and
I'm.
E
Just
not
sure
if,
like
the
sidebar
is
not
right,
because
you
want
to
be
able
to
use
the
sidebar
for
good
stuff,
I
feel
like
we
need
a
junk
drawer
in
the
notebook
UI
itself
to
put
stuff.
You
know,
and
it
needs
to
get
out
of
your
way
when
you're
not
sharing
and
hard
tagging,
but
whatever
whatever.
A
B
Well,
yeah,
that's
a
good
question.
Actually
I
can
make
it
available.
I,
don't
know
where
I
could
put
it
well.
A
There's
one
option:
oh
sorry,
there's
one
option
that
I
don't
know.
Let
me
get
it
for
you,
but
we
do
have
I
believe
it's
Jupiter
surveys,
because
it
was
started
with
surveys,
but
I
believe
it
holds
kind
of
research
beyond
the
limits
of
surveys
at
this
point,
so
I
think
that's
kind
of
a
place
or
that's
at
least
a
place
I've
treated
as
kind
of
like
a
record
of
of
jupiter-centered
research.
That's
open
someone
correct
me
if
they
think
that's
wrong,
but.
A
Thanks
I
use
it
I,
don't
know
what
to
say
yeah.
So
if
you
want
to
put
things
there,
there's
multiple
ways
you
could
do
it.
You
know
like
you
could
actually
put
stuff
in
a
directory,
but
if
it's
like,
oh
I,
just
want
to
share
a
link,
I
could
probably
do
it
in
an
issue,
but
that
might
be
a
good
place
unless
somebody
else
is
like
actually
I'd
prefer
it
here.
So
that'd
be
my
recommendation.
A
D
Secretly
My
Hope
Is
that
the
usability
studies,
foreign
help
us
figure
out
if
we
take
up
white
space
or
not
keep
feeling
there's
a
lot
of
things
on
the
UI
these
days
and
I
know
one
of
the
things
that
we
don't
talk
about
from
an
accessibility
perspective
often
is
the
cognitive
aspect
of
the
computational
thinking
part,
but
yeah
I
like
to
hear
that
one
of
the
first
steps
is:
let's
get
rid
of
some
buttons
because
I'm
all
for
that.
B
A
Where
would
we
go,
I
think
I'm
going
to
switch
agenda
items
unless
there's
any
last
comments?
I'll
be
quiet
for
a
few
seconds.
A
F
Oh
okay,
sure
so
I
I
thought
I
should
be
the
last
because
I
say
on
the
agenda.
It
seems
my
topic
is
as
they
lost
so
but
okay
I'm
ready
to
go.
F
So
may
I
ask:
how
is
my
network
connection?
Can
you
hear
me
clearly
so.
F
Oh
okay,
I'm,
not
very
confident
with
my
network
connection,
so
please
feel
free
to
interrupt
me
if
there
are
some
network
issues.
F
Okay
I
have
a
very
brief
slides
and
to
short
demo
videos,
and
my
name
is
Erica
and
I'm.
Currently
working
for
Alibaba
Cloud,
previously
I
was
working
for
Amazon
as
a
software
engineer
and
nowadays,
I'm
also
very
interested
in
contributing
to
open
source
project.
Actually,
I
am
one
of
the
maintainers
of
our
budget
domain
schedule
project
and,
of
course,
I
am
a
Jupiter
user.
So
recently,
I
am
exploring
the
possibilities
and
what
fantastic
stuff
I
I
can
do
by
combining
these
two
together.
F
So
today,
I'm
going
to
give
a
very
short
introduction
on
what
is
Apache
topic
scheduler
and
a
short
demo
for
how
Apache
domain
scheduler
is
integrated
with
Jupiter.
F
So,
okay,
so
obviously
doping
schedule
is
a
big
data,
workflow
orchestration
tool,
but
you
can
also
schedule
different
other
tasks,
such
as
stupid
tasks
or
python
task
or
different
kind
of
machine
learning
tasks,
and
one
of
his
most
amazing
feature
is:
it
is
a
local
development
platform,
for
example,
if
in
production
environment
the
the
platform
team
can
deploy
this
scheduling
tour
within
the
cloud
or
in
some
kind
of
distributed
system
and
those
data
analysis
or
data
scientists,
they
only
need
to
focus
on
their
their
data
tasks.
F
And
if
you
need
to
do
more
configurations
for
this
schedule,
scheduling,
service
or
or
doing
or
writing
actual
code
to
to
enable
this
scheduling
process,
they
can
just
do
it
by
drag
and
drop
and
also
in
Apache
during
schedule.
There
is
centralized
Resource
Center.
They
could
manage
all
the
resources
through
this
resource
center
and
basically,
actually
Apache
Dolphins
schedule
is
scattering
tool,
but
it
has
a
lot
of
extensible
task
plugins.
F
So
if
you
want
to
add
more
tasks
for
other
services,
or
you
want
to
do
some
kind
of
customizations-
you
can
the
development
it
very
easily,
and
last
but
not
least,
average
YouTuber
schedule
has
a
kind
of
good
Integrations
with
cloud
services,
so
in
production,
environment
yeah
when
it
comes
to
some
kind
of
Enterprise
using
you
can
deploy
it
with
some
cloud
provider
services
and
make
good
use
of
this
cloud
services.
So
here
is
some
kind
of
solution.
F
We
can
use
domain
scheduler
with
distributor
to
to
build
a
studio,
to
develop
code
and
schedule
those
tasks
in
production.
We
can.
We
can
deploy
this
in
a
kind
of
Studio
cluster
and
we
can
use
distributor,
Task
Launcher,
to
launch
to
launch
those
Jupiter
tasks
in
in
remote
clusters.
F
For
example,
if
you
want
to
do
your
computations
in
Hadoop
cluster,
you
can
you
can
encapsulate
these
some
kind
of
paper
milk
amount
to
execute
a
Jupiter
nodes
in
young
containers
or,
if
you
want
to
deploy
your
domain
schedule
service
in
kubernetes,
you
can
kind
of
use
domain,
scheduler,
kubernetes
task,
plugin
to
to
pull
the
image
to
pull
the
container
from
the
from
the
remote
repo
and
and
run
your
Jupiter
Jupiter
notes
in
the
containers.
F
Also
for
this
kind
of
metadata,
you
can
store
it
in
distributed
database
system
or
for
those
logs
and
resources.
You
can
store
it
in
the
remote
object,
storage,
for
example,
at
3,
for
AWS
or
OSS
for
Alibaba
cloud
and
for
those
kind
of
metrics
of
the
whole
system.
We
can
use
parameters
and
gravana
solutions.
F
Okay,
when
it
comes
to
the
Integrations
with
distributor.
As
we
know
for
the
most
of
the
cases,
we
need
python
environment,
virtual
environment
to
run
our
python
code,
so
so
as
the
Jupiter
task.
So
we
have
passing
environment,
managed
service
of.
A
F
Intervention
schedule:
we
have
three
options
for
the
first
options.
We
can
upload
and
conduct
Anaconda
environment
chart
to
domain
scale
the
resource
center,
and
when
you
try
to
execute,
when
the
worker
try
to
execute
the
nodes,
it
will
pull
the
Anaconda
environment
to
the
worker
and
On
Target
and
then
run
run
the
run.
F
The
python
job
in
the
virtual
environment
and
the
the
second
option
is
to
upload
the
requirement
text
to
the
resource
center
and
when
you,
when
the
worker,
trying
to
execute
the
Jupiter
note,
it
will
pivot
install
the
those
third-party
libraries
from
these
requirements
text
and
then
create
a
virtual
environment.
Then
run
the
actual
jobs
in
the
environment
and
the
option
three
is
to
install
kunda
environment
with
share
tasks
as
a
pre-node
as
we
as
we
know
opposite
domain
scale,
is
a
workflow
tool.
So
you
can.
We
can
have
a
pre-task,
for
example.
F
Shell
task
to
to
install
this
virtual
environment
then
run
the
python
pattern
code
in
the
virtual
environment.
So
any
questions
so
far.
G
How
do
you,
how
do
you
represent
the
the
environment
that's
being
passed
around
so
like
if
you
have
a
Docker
image,
for
example,
is
it
tied
in
the
metadata
of
the
Jupiter
Taft
launcher,
or
is
that
abstraction
different
for
each
dolphin
scheduler
instance.
G
G
You
know
setting
a
Docker
container
you
might,
it
might
only
support.
You
know
running
something
at
runtime,
whereas
the
kubernetes
would
support
a
custom
docker.
F
Okay,
so
that's
a
good
question,
for
example,
if
we
are
not
in
the
kubernetes
environment,
if
we
are
in
some
kind
of
Hadoop
system,
we
can
have
our
Resource
Center
connect
to
connected
to
the
hdfs
system
and
when
we
are,
for
example,
if
we
want
to
use
some
kind
of
kunda
environment,
we
can
use
condom
unpack
to
generate
the
tar
tar
file
and
then
upload
it
to
the
hdfs
file
system.
So
there
will
be
a
distributed
system
to
manage
these
virtual
environments
for
the
cable
for
the
kubernetes
situation.
F
We
can
build
our
image
with
the
virtual
environment
inside
of
it
and
we
can
store
the
image
in
some
kind
of
remote
image
repository
and
when
we
execute
the
task,
we
will
pull
the
image
and
run
the
python
code
directly
in
inside
the
image.
So
you
can
use
the
pre-build
a
virtual
environment.
C
F
Okay,
so
that's
a
good
idea
to
to
abstract
this,
but
currently
we
we
for
different
different
I
mean
for
different
tasks,
or
instance,
or
workflow
instance.
We
just
use,
maybe
different
environments
to
currently.
We
don't
have
a
abstraction
for
those
yeah.
G
That's
I
remember,
like
your
system
understands
that
there's
there's
different
systems
under
the
hood,
so
that's
I
was
curious.
How
you
do
it
so
that's
neat
to
know.
C
F
So
may
I
go
to
the
demo
part.
So
I
can
give
us
your
video
on
it.
So
you
can.
Maybe
I
can
show
you
the
better
Okay.
C
F
So
first
I'm
going
to
demonstrate
the
workflow
part
and
okay,
just
a
minute
now
to
make
things
simple.
I
will
just
use
shell
task
to
represent
different
kind
of
tasking
machine
learning
process
such
as
process
data
and
the
training
model,
and
the
prediction
and
evaluation
just
just
for
demo
purpose:
okay,
you
can
see
we
can
create
from
here
and
then
we
can
drag
and
drop
a
share
task
here
now,
the
first
first
one
we
are
simulates
some
kind
of
process
data,
but
actually
we
are
going
to
in
production
environment.
F
We
are
going
to
have
some
kind
of
python
script
here,
or
something
or
Jupiter's
RGB
node
here
to
do
this,
but
just
to
mix
in
simple
use,
I'm
using
Echo
here
to
to
save
my
to
show
my
first.
So
this
task
menu
is
for
pre-processing
data.
F
And
the
second
note
is
for
Penny
model
again
I'm,
just
using
a
shell
Echo
to
simulate
this
process.
F
And
third
part:
we
are
going
to
do
some
kind
of
evaluation.
A
F
F
Cool
then
we
are
going
to
online
this.
This
workflow
instance
definition.
F
Then
we
started
and
you
can
see
in
the
workflow
instance
page.
There
is
a
for
instance.
It's
it
had,
it
had
been
complete
completed
and
there
are
those
task
instance
we
connected
before.
Okay.
This
is
just
a
just.
We
trigger
the
workflow
manually,
then
we
can
schedule
it
so
so
that
can
be
executed
in
period.
For
example,
if
you
have
a
python,
if
you
have
a
machine
learning
task,
you
need
to
do
it
every
maybe
every
day
sometime,
then
we
can
use
this.
It's
scheduled
to
schedule
it.
F
As
you
can
see,
you
don't
need
to
write
any
extra
code,
you're
just
doing
everything
on
the
web.
Ui.
F
So
here
is
a
timing.
Tab,
and
this
is
there
is
a
chrome
expression,
but
actually
you
don't
need
to
to
learn
this
chrome
Chrome
expression,
because
there
will
be
some
kind
of
drop
down
menu
for
you
to
to
select.
F
This
is
how
we,
how
we
create
a
workflow
and
schedule
it
in
domain
schedule
and
here's
a
second
demo
about
how
to
integrate
the
schedule.
Distributor
here
is
a
jupyter
note.
It
is
to
I
use
it
to
do
the
weather
forecast.
F
As
in
I'm
acquiring
some
data
from
the
from
a
remote
API,
then
I
do
some
analysis
and
draw
some
pictures.
For
example,
I
assume
that
I
have
already
developed
this
task
in
Jupiter.
Interactively
then
I
want
to
schedule
it
no
well.
We
will
show
how
to
schedule
it
easily
in
doing
schedule.
F
Those
texts
here
is
about
parameterizations
because
in
Dublin
schedule
we
use
paper
mail
to
execute
the
Jupiter
node.
So
the
phone
talking
schedule.
We
will
pass
the
parameters
in
runtime
dynamically
and
you
will
go
to
this,
the
the
cell
and
we
can
see.
F
Yeah
we
create
a
project
in
the
project.
We
have
have
multiple
workflow
definitions.
If
you
are
working
in
a
team
you
can.
Oh
sorry
here
is
how
we
about
the
python
environment,
part
hallway,
track
and
share
the
python
or
virtual
environment.
F
So
just
just
a
minute,
so
we
go
to
the
resource
center
and
here's
the
upload
files
and
we
can
upload
the
files
yeah.
F
This
is
this:
is
a
Torah
file
packed
by
conduct
pack
and
the
resource
center.
Actually,
we
can
integrate
with
remote
storage
such
as
AWS,
S3
or
or
you
can
do
it
with
hdfs.
If
you
are
using
a
Hadoop
cluster.
F
Okay,
we
we
can
also
upload
the
requirements
with
the
only
dependencies
inside
you
can
see.
F
Yeah
we
give
it
a
name
like
with
a
forecast
or
with
the
Shanghai,
because
I'm
going
to
pass
the
Shanghai
City
parameters
into
into
the
from
the
paper
mill
to
execute
the
node.
Here's
the
path
of
my
Jupiter
note.
It's
a
template.
F
Actually,
you
can
see
those
parameters
are
very
exactly
those
parameters
are
passing
into
the
paper
mill
such
as
kernel
engine
and
execution,
timeout,
timeout,
or
something
like
that.
But
we
are
just
using
the
default
values
here:
okay,
I'm
choosing
the
environment,
virtual
environment,
I
just
uploaded.
F
Jupiter
task
plugin
will
detect
the
the
kind
of
methods
you
are
using
for
for
managing
the
environment.
If
it
is
tar
here,
it
will
On
Target
and
use
it
as
a
virtual
environment.
If
it
is
a
text
text
file,
it
will
treat
it
as
a
requirements
and
people
in
store
it.
People
install
the
environment
from
it
dynamically
and
then
run
the
actual
job.
A
Eric
sorry
to
interrupt,
but
I
want
to
make
sure
we
get
to
another
agenda
item,
so
I
just
want
to
give
you
a
few
minutes
to
keep.
Thank
you.
If
you
can't
sorry
about
that,
oh.
F
That's
totally
fine,
thank
you
for
the
reminder,
so
we
are
just
going
going
quickly
because
we
have
already
showed
the
the
process
of
triggering
the
workflow
instance
and
we
don't
spend
much
time
to
winning
for
waiting
for
it
to
get
executed.
So
you
can
say
you
can
see
here.
There
is
a
very
nice
bar
provided
by
the
paper
mill
if
you
are
familiar
with
paper
mill,
so.
F
Yeah,
oh
well,
I
reload
it
you
can
see.
There
is
a
new
Jupiter
nose
here
which
is
generally
generated
from
the
template.
It
is
a
weather
forecast
for
the
Shanghai
City
and
here's
a
nice
report,
maybe
not
so
nice,
but
yeah
I
report
for
you,
okay,
that
is
my
demonstration.
So
any
questions.
G
Anything
that
you
found
needed
that
you
would
have
preferred
to
have
with
the
the
integration
or
with
the
libraries
you're
interfacing
with
that
would
have
made
things
easier
or
that
you
noticed
were
inconsistent.
F
F
You
you
mean
whether
I
can
make
things
easier
with
the
integration
by
putting
these
two
things
together.
Oh.
G
No
I
was
just
asking
if
there
were
things
that
you
found
in
the
open
source
tooling,
that
could
be
improved
in
in
the
process
of
going
through
this
integration.
F
Okay,
I
I
think
I,
actually
I
read
a
article
posed
by
you
or
it
was
the
integration
between
dagister
and
so
the
The
Notebook
I
think
is
very
cool
and
fantastic
to
make
some
kind
of
tables
or
some
kind
of
nice
charts
or
some
some
very
cold
pictures
and
I
think
maybe
I'm
concerning
about
putting
it
as
a
task
plugin
of
domicile
so
that
we
can
schedule
some
kind
of
notable
tasks.
We
can
develop
code
it
there
and
then
schedule
it.
G
So
the
the
notable
is
running
Jupiter
notebooks
as
well,
and
when
it
runs
paper
mail.
The
way
it
works
is
you're
just
setting
that
engine
field
in
the
and
you
have
to
have
the
the
open
source
paper.
Mail,
origami,
Library,
installed
in
Python
and.
C
G
The
actually
I
think
the
interface
you
have
it
have
like.
If
you
want
Taylor,
so
it
defaulted
that
engine
type
you
could,
but
it
basically
just
uses
a
different
engine
to
support
it
with
paper
mill.
C
G
How
do
you
do
like
API
token
secrets
in
the
execution
environment.
G
For
example
like
when
you're
trying
to
connect
to
a
an
API-
and
you
want
to
have
like
the
user
needs
to
be
able
to
access
it
like
either
a
data
source
or,
like
you
know,
you
know
an
example
there
like,
if
you're
talking
notable
or
some
other
system,
to
identify
the
caller
to
the
API.
C
G
F
F
So,
for
example,
if
we
are,
we
are
deployed
this
in
in
some
kind
of
you
know
some
kind
of
stock
provided
by
the
cloud
providers
such
as
Alibaba,
EMR
or
some
kind
of
AWS
stuff.
Then
we
can
use
the
IAM
or
Ram
to
to
to
to
do
this.
We
can
do
the
assume,
Rule
and
give
some
kind
of
STS
token.
Then
we
use
the
token
to
to
to
to
to
making
requests
to
those
API
foreign.
F
Of
course
you
can
use
other
kind
of
things
such
as
access,
key
or
stuff
like
that,
but
yeah,
you
know
it's
not
very
secure.
Yeah.
G
Yeah
I
was
basically
curious
about
that,
because
I
think
one
of
the
things
that
comes
up
a
lot
when
you
get
the
Integrations
for
folks
is
how
to
get
like
the
secure
tokens
for
arbitrary
systems
and
and,
for
example,
like
either
using
you
know
something
like
Vault
to
store
it
locally
or
one
of
the
AWS
services
to
a
Secret
store
to
kind
of
capture
things
on
behalf
of
the
user,
so
that
things
will
get
accidentally.
F
Yeah
and
also
we
are
doing
some
kind
of
integration
with
oauth
if,
if
two
different
Services
is
both
support
all
so,
we
can
use
the
OS
to
to
do
the
single,
sign-on
or
stuff
like
that
again.
Yeah.
G
F
Don't
don't
forget
no
okay,
so
I
know
there
are
a
lot
of
machine
learning.
Scientists
or
data
and
Analysis
are
using
Jupiter.
So
we,
our
doping,
scheduler
Community,
also
provides
those
kind
of
ml,
Ops
tasks
or
stuff
like
that,
so
wondering
whether
it
is
possible
for
this
two
Community
to
do
some
fantastic,
fantastic
stuff
together,
for
example,
do
some
kind
of
code,
Meetup
or
stuff
like
that?
Is
that
possible.
A
I,
do
definitely
think
it's
possible
I,
don't
know.
Are
you
looking
to
connect
with
specific
like
right?
We
have
Matt
here
with
paper
mill
and
among
other
things,
is
that,
like
you,
want
to
be
closer
to
put
those
two
together
further
or
are
there
other
people
you're
trying
to
contact.
F
Okay,
we
want
to
do
the,
maybe
we
can
do
a
committa
with
the
Jupiter
Community
and
the
komita
with
with
Matt.
If
you
have
some
kind
of
community
or
you
have
a
company,
maybe
we
can
do
something
like
I,
don't
know
whether
you
are
building
the
community
of
notebook
or
something
like
that.
G
G
If
we
want
to
follow
up
on,
like
you
know
how
using
Jupiter
is
the
medium
to
communicate
between
between
different
proprietary
tools,
but
then,
like
the
other
side
of
things,
is
how
you
want
to,
for
example,
if
we
want
to
be
like
adding
things
to
like
the
documentation
for
how
to
use
different
tools
in
the
wild,
like.
C
B
G
That
for
like
here's,
how
you
find
the
parameter
cell
in
each
type
of
interface
or
here's,
how
you
would
Register
paper
million
different
tools,
and
so
one
thing
we
could
also
talk
through
is
you
know
if
they're
a
commonplace,
updated
docs
like
how
to
use
different
Jupiter
tools
with
other
things
in
the
open
source?
Docs
I
think
that's!
That's
really
helpful
for
people
because
a
lot
of
times
people
come
to
the
docs
where
they've
got
hey
I've
heard
I
can
use
x
with
Y
I've
got
a
Jupiter
notebook.
How
do
I
do
it.
F
Okay,
let's
go
may
I
ask,
is:
is
that
possible
for
us
to
do
some
kind
of
online
meeting
together?
Maybe
some
kind
of
online
Meetup
of
two
community.
G
Oh
I
see
what
you're
asking
yeah
I
think
we
haven't
really
had
many
meetups
on
the
paper
mill
headless
orchestration
side
of
things
for
a
bit.
G
I,
don't
know
Isabella
if
we've
been
doing
any
like
like
walk-throughs
for
for
people
that
want
to
join
on
like
how
to
use
like
Jupiter
lab
features
or
other
things
in
the
OSS
side.
So.
A
It's
possible
there's
none
that
I'm
currently
aware
of,
but
thinking
of
this
I
may
also
connect
with
the
Jupiter
Community
Building
committee
might
know
if
there's
been
something
there
as
well,
because
I
know
that
they
have
some
workshops.
They've
been
running
that
I'm
not
staying
on
top
of
I,
don't
think
I
know
what
is
running
right
now,
so
that
might
be
the
next
people.
I
can
connect
workshops.
D
Go
ahead,
one
of
the
things
that
will
be
important
with
moving
forward
with
this
is
bringing
people
to
the
table
who
also
help
run
these
events,
because
this
community
work
has
been
severely
under
resourced
and
supported
so
I
think
in
order
to
make
this
happen,
we'll
probably
need
people
who
are
interested
in
organizing
these
things
from
other
from
the
other
side,.
C
H
C
F
G
I
think
we've
done
some
of
these
types
of
things
too
in
the
past,
at
some
of
the
pi
data
conferences,
where,
like
there'll
be
a
workshop,
so
like
Matthias
is
running
some
of
these
before
and
Paul's
run
these
before,
where
they'll
they'll
walk
through,
like
you
know
how
to
set
up
a
condom
and
do
something
interesting
to
the
combination
of
tools
around
Jupiter
and
like
do,
that
could
be
another
place
to
get
people
engaged
or
interested.
But
yeah
I'd
be
kind
of
there's
like
some
different
options
to
look
into.
H
We
I
think
we
can
just
make
up
the
email
group
to
discuss
this
and
we
like
to
organize
the
workshop,
but
online
or
offline
I
think
online
is
better
because
we
were
in
China
and
some
people
cannot
go
abroad
and
if,
if
you
need
some
support
on
organize
the
meeting,
we
have
the
resource
and
we
we
have
many
volunteers
who
want
to
to
hold
the
meet
up.
So
we
can
make
an
email
and
discuss
this
in
the
future.
I
think.
A
H
Yeah
and
I
view
my
email
right
here
and
I
think
our
weekend
and
the
community
are
very
interested
about
how
to
combine
the
two
stuff
together
and
let
people
easier
to
use.
A
Yeah,
thank
you
for
following
up
on
that.
It
was
really
good
to
see
this
demo
after
talking
about
it
a
little
bit.
The
last
call.
A
Okay,
maybe
not
yeah
I
was
checking
because
I
was
being
messaged
on
the
side
and
I
wanted
to
make
sure
I
didn't
miss
anyone
by
accident.
A
E
You
very
much
okay,
hi
everybody.
This
is
picture.
It's
from
Jew
picture,
which
is
from
deathbeds
picture,
is
a
embedded
diagramming
language.
It's
very
plain
text,
very
plain
language
focused,
which
is
always
a
thing
that
I
like
I'm,
Nick,
Tony
and
I
are
deathbeds
and
you
can
see
all
of
our
stuff
in
when
I
get
my
back
and
you
can
see
that
Death
Beds
few
picture
and
you
picture
are
you
after
you
know
you
can
go
play
with
this
stuff
yourself,
all
right,
let's
hit
it
so
Jupiter.
E
The
first
thing
I
wanted
was
to
be
able
to
write
these
little
things
inside
of
diagrams,
and
so
this
is
what
some
looks
like
inside
of
a
a
notebook-
and
this
is
what
the
text
looks
like
for
that.
So
we
boil
it
down
to
just
a
particular
thing:
Arrow
right,
200,
markdown
to
source.
Now
we've
got
a
markdown
Source
arrow,
and
then
we
can
have
a
box,
it
points
in
a
box
and
it's
got
a
radius
of
10
PX
and
you
know
whatever
this
is
it's
a
very
opinionated
language
and
I?
E
Don't
have
a
syntax
highlighter
for
it
yet,
but
whatever
you
know,
sometimes
you
can
do
stuff
that
breaks.
It
tells
you
if
it's
broken,
it's
cool,
it's
stupid
fast,
because
it's
just
C
compiled
the
webs
and
and
it's
it
doesn't
do
anything.
It
doesn't
need
anything
it's
insane.
Okay,
so
that's
the
markdown
flavor.
E
They
have
to
determined
a
mime
type
I
guess
when
you're,
the
guy
that
wrote
sqlite
you
get
to
do
that.
So
we
just
got
this
text
X
picture.
We
can
build
off
of
that.
This
is
some
crazy
thing.
I
pulled
off
of
there,
but
it
had
attribution,
which
is
why
I
felt
okay
with
it,
and
so
you
can
use
the
normal
IPython
display
stuff
to
get
it
there
I'm
talking
about
how
this
is
selectable
text
and
I
picked
one
that
doesn't
have
any
text.
E
So
that's
silly,
whatever,
let's
see
if
I
can
do
live,
got
some
stuff
copy
and
pasted
here.
Yeah,
oh,
come
on:
let's
try
that
again
yeah!
E
E
And
so
there's
the
inevitable
widget
setup
page
works
in
Jupiter
light,
which
is
why
all
this
content
so
I'm
going
to
need
some
widgets
I'm
gonna
need
some
picture.
Here's
some!
You
know
your
standard,
sandbox
stuff.
Nice
I
already
ran
it.
So
you
get
to
see
the
candy
first,
but
here's
some
stuff.
So
as
you're
typing,
you
know
you
can
see
what
mistakes
you're
making
and
just
keep
making
boxes
boxes
all
day
boxes
everywhere.
All
the
boxes
anyhow
that
one's
better
I,
don't
know
what
happens
if
you
make
another
one.
This
makes.
C
E
You
can
build
little
playgrounds,
I
love,
playground,
let's
see
here,
I
think
that's
it!
That's
it
that's
the
four
minute.
Oh,
it
works
with
all
the
other
stuff
too
right.
So
it
works
with
you
know.
If
you
go,
this
is
all
their
homepage.
Stuff
ends
up
in
here.
So
you
know
if
we
open
this
up
with
the
markdown
preview
tastes
great
in
here
same
old
stuff.
A
lot
of
these
links
don't
work,
but
you
know
here's
some
swim,
Lanes
yeah!
You
want
to
go
all
out.
If
you
want
to
do
it.
E
F
E
Annoying
yeah
and
then
you
open
it
wrong,
Nick
open
with
markdown
preview
yeah.
So
we
can
do
crazy
things.
It
knows
about
north
and
south
and
all
that
kind
of.
E
To
be
really
fun
to
write
the
grammar
for
this
I
tell
you
what,
but
it
can
do
like
constrained
boxes
around
stuff,
which
is
insane,
but
really.
The
reason
it
exists
is
for
it
doesn't
do
anything,
that's
great.
They
have
some
stuff
where
they
use
it
for
like
syntax
stuff.
If
I
look
for
this
real
quick.
E
D
So
that
pairs
well
with
Jupiter
lab
markup.
E
E
Yep
this
is
Google,
I've
missed
installed,
so
yeah.
If
you
like
mermaid,
you
can
do
mermaid
in
there.
If
you
like,
SVG
OB,
you
can
do
SVG
up.
So
it's
you
know
what
should
these
things
when
you
push
into
the
disadvantage?
Is
it
only
works
in
fossil
outside
of
here,
whereas
mermaid
Works
in
GitHub?
So
you
know
that's
kind
of
a
bummer.
What.
E
What
okay,
so
this
is
Jupiter
lab.
C
E
It's
kind
of
a
reboot
of
a
lot
of
stuff
that
I
built
in
the
past
in
the
present
whatnot.
It
can
move
things
around.
You
know
you
can
fart
around.
A
A
E
For
this
yeah,
it's
all
made
out
of
you
know
you
know
slideshow
whatever
you
know
whatever
just
I
mean
they
can't
do
everything
with
that
right,
like
it's
got
some
I
got.
Some
I
got
some
crazy
stuff
that
I
do
in
here
that
you
know
you
don't
want
to
know
about,
but
why
it
worked,
that
works,
I,
don't
know
I,
guess
flexbox,
Works,
weird,
but
yeah
yeah.
This
is
powered
by
Jupiter,
lab
fonts
and
maybe
I've
broken
it
entirely.
At
this.
C
E
A
Thank
you
so
much
for
that
share
sorry
to
put
you
on
the
spot,
but
it
is
always
fun
to
see
you
perform
in
a
different
constraints.
Nick.
It's.
E
Always
I
always
enjoy
performing,
and
part
of
this
is
pursuant
I'm
gonna
drop
a
link
in
the
chat,
so
we
got
a
Jupiter
Delight
Workshop
coming
up
and
we're
going
to
be.
You
know
putting
out
some
stuff
about
that,
and
one
of
the
things
that
I'm
super
interested
in
is
Jupiter
light
because
of
how
you
can
deploy.
It
could
actually
be
the
front
end
for
GitHub
or
git
lab
or
fossil
or
gitty
or
dogs,
or
any
of
those,
as
long
as
they
can
deploy
static
websites.
E
So
it's
really
got
me
thinking
about
fossil
because,
like
I'm
on
fossil
in
the
browser,
you
can't
run
gitlab
in
the
browser.
You
can't
run
GitHub
like
the
whole
thing
in
the
browser,
but
you
can
run
fossil
in
the
browser,
and
so
as
with
that
as
sort
of
the
Bastion
Point
of
distributed
Version
Control
in
the
browser
light
can
do
all
of
these
things.
It
can
talk
to
apis,
it
can
run
clients.
E
So
if
you
know
possible
is
just
kind
of
the
beginning
of
me
getting
my
feet,
wet
on
or
sorry
the
picture
stuff.
E
Can
we
really
do
what
it
does,
but
so
there's
some
food
for
thought
stuff.
There.
I
E
There
is
no
bleedinger
I
mean
you
know,
we
gotta,
you
gotta
check
your
blood
type.
When
you
go
in,
you
know,
we
might
need
transfusions.
I
mean
it's
crazy,
it's
crazy,
but
just
no
Docker,
no
Docker
just
keep
the
docker
away.
Freaking
docker.
A
Okay,
yeah
no
blood
and
no
whales,
okay,
yeah
I
am
trying
to
sorry
anyone
to
please
leave
if
you
need
to
I
will
stop
recording.
Oh
no,
that
was
perfect,
I'm
glad
I
didn't
stop
recording
for
the
cat.
Thank
you
all
for
being
here.
I
really
appreciate
it.
So
thanks
for
bearing
with
us,
it
was
awesome
to
have
so
many
great
demos
today
hope
to
see
some
of
y'all
again
and
have
a
good
rest
of
your
day.