►
From YouTube: Jupyter Community Call - April 30th, 2019
Description
NOTE: Unfortunately, we had some technical difficulties with recording the call this month. We missed recording the first 20 minutes of the call. We apologize especially to the members from the University of Edinburgh, since we failed to record half of their presentation. Please check out their slides below!
Slides to the presentation from the University of Edinburgh:
https://edin.ac/2PBJc1T
The notes from this call can be found here:
https://jupyter.readthedocs.io/en/latest/community/community-call-notes/2019-april.html
Read more about these calls in Discourse:
https://discourse.jupyter.org/t/all-jupyter-community-calls/668
A
Okay
and
I'll
just
stay,
so
this
is
really
near
close
to
the
end
anyway.
Actually,
but
so
this
is
our
V
le.
So
this
is
the
virtual
learning
environment
that
all
of
our
students
and
our
staff
get
access
to.
This
is
obviously
a
test
course,
and
this
is
the
the
notable
service
so
when
their
users
click
on
this
they're
taken
out
to
our
version
of
Jupiter
hub
and
they
can
select
from
a
variety
of
notebooks
that
we've
got
set
up
for
people
already.
A
So
these
are
sort
of
pre-built
environments
that
are
docker
containers
from
my
kubernetes,
so
I
think
we're
saying
that
a
lot
of
the
zero
to
Jupiter
have
provided
the
basis
for
what
ended
up
becoming
the
notall
service,
and
so
each
of
these
is
a
container,
has
a
an
amount
of
dependencies
that
someone
used
for
a
specific
class,
and
so
we've
got
a
haskell.
Notebook
is
relatively
new
and
LSST,
which
is
a
very
specific
set
up
for
the
Royal
Observatory
here,
which
is
I,
can't
remember
the
what
the
acronym
stands
for
base
something
to.
B
A
Obviously,
and
so
I
mean
once
you
get
through,
it
looks
exactly
like
a
normal
Jupiter.
We
have
added
in
a
few
of
our
own
little
elements,
so
there's
the
form
green
but
form
greater
tab
at
the
top,
which
is
where
we
look
at
the
assignments
and
assessments.
This
is
a
test
course,
so
I
have
a
variety
of
oddly
named
examples
here,
and
but
this
looks
for
the
course
ID
that
they've
had
coming
through.
So
it
knows
what
course
you're
accessing.
A
So
that's
part
of
what
we
want
to
do
is
we've
done
a
lot
of
this
work
now
to
get
it
working
in
our
own
institution,
but
we're
putting
the
effort
in
to
push
it
back
and
make
sure
that
it's
open
for
everybody
else
to
use
as
well.
So
that's
the
the
end
of
my
talk,
I'm
happy
to
take
any
questions
that
anybody's
got
and
I'm
happy
to
pass
any
questions
on
to
bear
as
soon
as
I
get
ever
so
slightly
technical.
B
D
This
is,
this
is
run
on
our
in
our
own
data
centers.
We
have
two
data
centers
and
we
have
a
kubernetes
cluster
that
spans
across
both
yeah.
That's
that's!
That's
how
we
run
it
might
not
stay
there
forever.
It
sort
of
depends
on
how
much
scale
we're
going
to
need
in
the
future,
but
for
now
it's
a
it's
more
than
enough
where
we're
provisioning
about
half
a
terabyte
of
RAM.
At
this
moment,.
B
All
right
and
also
a
related
question,
so
imagine
another
university
would
like
to
continue
with
your
experience
and,
for
example,
provide
centralized
Jupiter
hub
of
a
similar
of
a
similar
type.
Using
your
work
as
a
starting
point.
Do
you
know
roughly
how
whether
this
would
be
like
a
lot
of
work
or
whether
it
could
be
a
project
for
one
person
for
a
few
weeks
or
like?
What's
there
and
I
understand.
D
B
D
That
depends
on
the
amount
of
customization
you
want
to
offer
to
your
students
like
if
you
want
to
get
get
something
up
and
running.
This
is
a
one-person
team.
I
think
that's
pretty
doable,
I
think
where
things
become
more
difficult.
If
you
want
to
provide
storage
for
every
student
for
and
you
want
them
to
access
all
their
notebooks
across
the
hall,
their
university
career,
etc,
then
things
start
to
come
more
difficult,
but
I
think
the
sort
of
vanilla
flavor
you
can
probably
get
up
and
running.
In
a
few
weeks
time,
Thanks.
A
There
was
a
lot
of
extra
work
for
further
developers
as
well
in
putting
together
the
LTI
integration,
which
was
a
big
thing
for
us.
It
was
one
of
the
things
that
all
the
other
services
were
missing
as
well,
so
that
kind
of
handles
the
roles
of
all
of
the
students
and
maps
it,
and
they
said
they
have
an
easier
access
point
for
them
to
be
passed
across
to
their
their
area
and
that's
what
we
used
as
a
basis
for
then
go
in
and
adding
them
be
greater.
A
So
I
think
like
Bert
was
saying
initially
it's
a
small
amount,
but
then,
as
soon
as
you
start,
adding
all
the
things
that
a
university,
or
at
least
our
university,
really
wanted.
It's
a
lot
of
extra
work.
So
we're
still
in
a
pilot
phase
at
the
moment
until
the
end
of
this
semester
and
then
we'll
launches
a
full
service,
and
that
is
still
extra
work.
Because
then
we
need
to.
A
E
A
B
F
Just
saw
the
on
the
demo
the
long
list
of
profiles
you
have
for
a
different
kind
of
images
that
you
may
need
to
run
so
I'm,
just
wondering
if
you
checked
binder
how
about?
Maybe
because
it
can
create
different
kinds
of
images
from
small
or
instructions
files
and
that's
maybe
a
way
to
get
customization
of
damages.
A
Yeah
I,
so
we
we
were
aware
of
binder
hope
and
but
one
of
the
things
that
we
needed
was
we
needed
to
have
a
set
of
files
that
they
can
go
on
and
they
can
work
on
and
they
can
save
the
state
so
that
when
they
go
back
to
it,
they
can
carry
on
working
from
where
they've
previously
left
off
and
it'll
be
a
portfolio
that
builds
up
across
their
entire
sort
of
university
lifetime
and
I.
Think
we've
also
put
a
lot
of
work
into
creating
Colonel
sort
of
images.
A
Docker
images
for
specific
kernels,
so
I
think
that
made
it
a
bit
easier
for
us
to
be
able
to
add
in
extra
things
a
lot
of
time.
They
were
very
nice
for
a
teaching
case,
and
so
we
kind
of
needed
to
make
sure
that
it
was
something
that
we
could
configure
and
we
can
make
changes
to
as
and
when
we
needed
to.
But
we
do
yeah.
A
C
A
For
the
moment,
I'll
I'll
put
a
few
links
in
so
one
of
them
is
the
blog,
which
is
sort
of
a
good
way
to
follow
up
on
what
we're
doing.
I'll
put
the
contact
details
of
myself,
because
I
think
I
have
a
bit
of
an
overview
of
it
from
a
nice
technical
side
and
then
I
can
handle
queries
and
pass
them
on
and
bio
holes
are
the
details
in
the
agenda.
So
you
don't
have
to
look
very
far
great.
That's.
C
B
Hi
everyone,
my
name,
is
santanico
Knopf
I'm,
a
physicist
from
Delft
University
of
Technology
in
the
Netherlands
and
I've,
been
following
the
project
Jupiter
or
various
various
aspects:
Jupiter
Hobbit
cetera,
but
now
I'm
going
to
discuss
specifically
the
Jupiter
as
an
altering
environment
and
in
particular,
I
wanted
to
show
some
of
my
experiences.
Let
me
figure
out
how
sharing
works.
B
B
No
just
a
moment:
yeah,
alright
I
think
that's
good,
so,
like
ever
encountered
a
bunch
of
situations
when
I
need
to
show
computed
output
that
that
needs
to
be
updated
and
that
many
people
need
to
edit
and,
in
addition,
it
needs
to
be
in
a
nice
context
to
make
an
specific
example.
These
are
the
lecture
notes
from
my
course,
and
they
have
plots
that
are
computed.
B
Notebooks,
specifically,
for
that
are
a
relatively
bad
format,
because
markdown
is
is
very
limited
and
at
the
very
least
you
want
to
extend
markdown
at
the
same
time,
I
realized
what
I
think
is
an
extremely
important
property
and
okay,
it's
it's
not
only
my
realization,
I
think
this
is
something
that's
emerging
across
the
community,
but
Jupiter
protocol
has
a
very
useful
specification
of
how
to
treat
outputs
of
arbitrary
code
being
evaluated.
So,
if
I
look
at
the
source
of
this
file,.
B
If
I
look
at
the
source
of
all
of
these
files,
these
will
contain
your
Python
code.
This
is
this
is
using
the
wonderful
Jupe
text.
Extension
that
also
allows
to
run
and
develop
this
file
in
a
notebook,
but
it's
a
combination
of
Python
code,
where
all
the
outputs
are
then
processed
by
by
by
first
Jupiter
and
then
inserted
back
in
the
output
and
in
particular
this
becomes
valuable
when,
when
developing
documentation,
so
for
for
any
tutorials,
I've
seen
a
lot
of
packages
across
the
community
that
develop
their
tutorials
in
the
notebooks.
B
But
then,
for
example,
they
are
not
able
to
use
any
of
the
advanced
Sphinx
features,
for
example,
by
link
across
different
files,
etc.
So
to
incorporate
these
features
of
processing
the
output
in
the
output
of
code
execution
in
a
Sphynx
documentation,
together
with
a
colleague,
Joe
Weston
and
vast
I
hold
these
are.
These
are
all
my
colleagues
from
Delft
we
wrote
a
minimal
extension
or
as
a
rather
simple-minded
extension
that
allow
us
to
to
take
to
send
an
arbitrary
code
to
Jupiter
and
use
the
other
result
of
its
output
in
in
the
documentation.
B
B
So
here
we
show
a
bunch
of
code.
We
also
show
an
output,
a
resulting
output.
This
is
this
is
an
eye
by
widget
heavily
customized.
This
is
a
bouquet
plot,
with
all
of
its
controls
still
enabled
so
and,
and
importantly,
the
source
of
this
file
does
not
contain
a
lot
more
information
than
then
just
what
you
saw
before.
Essentially,
every
code
block
is
encoding
embedded
in
the
Jupiter
execute
environment,
and
if
we
want
to
hide
it
from
the
readers,
we
use
an
extra
command.
There
are
also
a
couple
other
commands
that
are
not
extremely
important.
B
The
rest
is
just
a
code
as
if
you
would
write
it
in
the
notebook,
so
we'd
like
to
invite
everyone
to
give
a
to
take
a
look
at
this.
The
link
is
now
already
in
the
in
the
minute
and
as
a
final
remark,
I
also
wanted
to
bring
up
something
else.
So
so
now
here
we
are
using
a
feature
of
Jupiter
that
that
specifically
focuses
on
being
able
to
execute
arbitrary
code
and
process
it
its
outputs.
So
this
could
work
with
with
Giulia
our
assembly,
anything
for
which
there
is
a
jupiter
kernel.
B
At
the
same
time,
the
markdown
here
is
is
completely
alien
to
jupiter,
so
while
I
can
produce
the
outputs
rather
nicely,
I
cannot
convert
these
these
this
documentation
into
notebooks,
because
the
documentation
is
in
restructure
text.
The
notebooks
are
in
markdown
and
there
is
no
reasonable
way
to
convert
one
into
another.
So
there
is
now
an
a
very
slow
and
ongoing
debate
in
the
Jupiter
community
about
specifying
the
markdown
spec.
B
This
is
sorry
Jupiter
and
before
Matt,
yes
issue.
Eighty,
this
is
also
linked
in
the
minutes.
So
there
is
a
proposal
to
make
a
formal
specification
of
the
markdown
format,
and
this
this
immediately
has
has
problems
because
already
across
between
classic
the
classic
notebook
and
Jupiter
lab,
the
formats
are
actually
different.
B
So
I
think
it
would
be
nice,
and
this
is
something
that
I
would
like
to
invite
everyone
to
think
or
to
consider
to
make
notebooks
doubly
multilingual,
where
both
the
they
form,
the
language
that
exhibit
runs
the
code
is,
is
arbitrary
and
the
markdown
format.
So,
just
like
there's
a
kernel
for
for
rendering
or
for
for
executing
the
code.
B
There
is
also
there
would
also
be
a
kernel
for
rendering
the
markdown
of
the
notebook
I
think
this
is
all
I
wanted
to
say
our
perhaps
since
there's
an
opportunity,
I'd
like
to
make
a
shout
out
to
a
small
but
absolutely
glorious
library,
that's
that
was
started,
I,
think
by
menurkey
and
now
there
are
several
others
like
it
that
allow
us
to
click
a
button
and
change
all
the
that
Steve
lab
change.
All
the
parameters
of
the
code
and
run
it
using
by
connecting
it
to
binder
all
right
thanks,
Aaron.
C
Thank
You
Anton
I
was
awesome,
so
we'll
open
up
the
floor
for
maybe
a
couple
of
questions.
So
if
anybody
has
a
question,
go
ahead
and
shout
out
I
know
the
T
is
typing
one
right
now:
I,
don't
know
you
can't
speak
so
I
can
read
off
his
point
but
basically
said
in
love
to
merge
this
into
I-5
phone's
built-in
syncs
extension.
That's
and.
C
B
I
see
this,
this
sounds.
This
sounds
really
good,
so
so
right
now
we
are
still.
We
still
did
not
actually
release
the
new
version.
This
was
this
was
inherited
from
the
Jupiter
widget
jupiter
widgets
repository
with
the
same
project
with
the
same
goal,
but
a
narrower
scope.
There
is
a
milestone,
v-point
v
0.2,
but
I
really
like
I
really
like
the
idea,
so
so,
let's
that
sounds
real
good
and
let's
discuss
this
later.
E
So
spent
a
night
like
I've
wrestled
with
similar
issues
in
the
past,
as
you
try
and
figure
out
what
the
the
most
authoritative
version
of
your
content
might
be
when
you're
preparing
these
course
notes.
What's
yours
when
you
edit
the
file
to
deliver
this
course,
are
you
adding
a
restructure
text
file?
Oh.
B
E
B
The
other
context
in
in
the
documentation-
it's
obviously
an
rst
file,
but
in
many
cases
there's
a
so
so
with
with
the
markdown
thanks
to
Zhu
Patek,
we
are
able
to
edit
the
whole
thing
in
the
notebook
except.
The
markdown
is
not
always
correctly
rendered,
because
notebook
does
not
understand
that
it
uses
a
bunch
of
extensions.
Okay,.
E
E
E
B
C
Have
you
seen,
have
you
checked
out
in
Jupiter
book
I've.
B
Seen
I
have
seen
Jupiter
book
indeed,
and
so
so
so
it
has
the
same
fundamental
problem
that
essentially
every
project,
and
that
is
I,
think
there
is
no
one
fundamental
markup
mark
a
markup
language
to
rule
them
all
like
as
soon
as
you.
As
soon
as
you
have
an
implementation.
There
will
be
a
limit
to
the
functionality,
so
Jupiter
book
has
has
a
bibliography,
but
it
might
not
have
admonitions
or
footnotes
or
something
and-
and
the
details
really
strongly
depend
on
the
context
so
for
for
for
packaged
documentation.
B
Resolving
resolving
names
from
your
package
is
something
mark.
No
markdown
parser
will
enable
so
I
in
the
end
gave
up
on
the
hope
to
find
one
best,
markup
language
and
now
I
believe
everything
depends
on
the
circumstances.
But
yes,
Jupiter
book
is
indeed
one
project,
which
is
also
closely
related,
and
they
also
are
producing
now
markdown.
That
does
not
work
with
Jupiter
notebooks,
so
you
cannot
edit
Jupiter
notebook
files
with
with
Jupiter
lab,
for
example,
or
you
can
edit,
but
you
cannot
see
how
they
look.
G
C
Know
we're
seeing
yeah
I
think
we're
seeing
four.
Then
our
screen
not
that.
G
Okay,
I
won't
know
there.
We
go
no
we're
okay,
so
I
wouldn't
talk
a
little
bit
about
this
to
see
our
Chi
project
that
we
have
and
how
it
interacts
with
Jupiter
before
I.
Do
that
that
one
give
you
a
little
bit
of
rundown
of
who
we
are
so
I'm,
actually
part
of
a
an
institute
called
the
molecular
sciences,
software
Institute's,
so
we're
an
NSF
funded,
Institute's
kind
of
like
one
of
the
first
of
its
kind
in
the
u.s..
G
So
any
kind
of
like
biomolecular
simulation,
any
kind
of
drug
design
falls
within
our
regime
and
there's
a
something
called
quantum
chemistry.
So
we
learned
about
hamiltonians
and
that
notebook
last
presentation,
so
we
can
go
out
to
molecular
hamiltonians
as
well,
and
also
it
kind
of
goes
up
to
quantum
computing
and
have
fun
on
computing
applies
to
molecules
so
anything
in
the
molecular
regime,
we're
kind
of
a
part
of
and
we're
building
out
some
rapport.
G
So
there's
eight
different
universities
that
really
represents
our
board
of
directors.
We
have
larger
science
advisory
boards
about
thirty
folks,
but
then
the
real
part
of
it
is.
We
have
about
40
software
engineers
and
graduate
students
directly
involved
to
the
Institute.
So
we
actually
have
a
reasonable
amount
of
people
to
start
to
make
a
dent
in
this
field,
because
it's
really
large
and
then
sorry
I
got
a
well
okay.
G
Apologies
I
have
my
own
screen
has
to
start
blocking
it
and
what
we
really
do
is
we
do
all
kinds
of
things
as
we
do
software
infrastructure.
So
we
do
a
lot
in
terms
of
different
projects
that
support
our
community.
We
do
a
lot
of
teaching
and
best
practices,
so
I'm
a
software
carpentry
instructors.
So
all
of
our
education
derives
from
software
carpentry
and
Jupiter
notebooks.
So
we
do
a
ton
there
with
that
kind
of
Education
and
the
other
thing
we
do
is
finally
interoperability
standards,
as
as
one
would
expect.
G
G
So
the
real
thing
that
were
after
is
this
project
called
QC
archive,
which
falls
underneath
our
kind
of
infrastructure
build
outs.
So
one
of
the
things
that
we
hear
a
lot
when
we
talk
to
people,
we
do
these
hundreds
of
interviews
trying
to
figure
out
what
the
community
actually
wants
and
they
say
that
what
we
want
is
we
want
to
be
able
to
do
a
high
throughput
computing
easily
and
so
that's
challenging,
because
we
can't
use
any
kind
of
commercial
resources.
G
We
have
to
use
academic,
supercomputers
and
clusters
which
make
things
kind
of
rough,
but
they
want
to
be
able
to
get
these
computations
done
in
an
efficient
manner
which
some
sort
of
easy
interface
on
top.
So
whenever
you
something
like
this
out,
you
almost
always
come
out
with
this
kind
of
architecture,
where
you
have
some
sort
of
portal
fronting
that
you
interact
with.
That,
goes
to
central
servers,
you're
able
to
submit
computations
and
pull
computations
back,
and
then
these
kinds
of
things
can
go
out
to
many
different
backends.
G
Depending
on
what
kind
of
computer
resources
you
have
so
I
think.
Actually,
one
of
our
parallel
distributors
is
actually
the
the
ipython
core
distributor,
so
we
can
actually
use
that
on
supercomputers
to
partial
out
many
different
quantum
chemistry
computations
in
Lisp
interpreter
says
the
overall
infrastructure.
What
it
kind
of
looks
like
is,
we
can
actually
do
these
really
complex
workflows.
This
is
kind
of
an
example
of
the
torsion
drive.
G
You
know,
pass
these
to
different
people,
so
they
want
to
be
able
to
contribute
to
the
central
repository
they
want
to
be
able
to
access
other
people's
data,
and
this-
and
so
this
is
where
and
Jupiter
comes
in
as
the
front
end
is
really
on
this
front.
End
client,
so
I
won't
spend
too
much
in
the
architecture,
but
just
I'm,
showing
that
the
this
fronting
client
that
we're
working
with
is
really
layered.
G
G
G
So
the
way
we
do
this
is
we
we
have
a
couple
different
bolts.
So,
first
of
all,
we
have
a
general
website
and
so
on.
This
website
will
put
kind
of
like
eventually
like
a
some
sort
of
gateway,
so
a
very
user
facing
thing
where
you
can
just
kind
of
click
around
and
some
scripts
to
to
access
to
access
various
compute,
but
the
the
one
that
we're
really
interested
in
is
these
Jupiter
notebooks.
G
So
what
we
have
here
is
the
ability
inside
of
a
jupiter
notebook,
to
connect
to
our
central
repository
of
data,
and
so
because
one
of
the
easiest
interfaces
that
we
found
is
mr
notebooks
everything
is
here,
is
to
pure
native,
and
so
we
connect
the
central
posit
ori.
I'm
able
to
look
at
all
the
possible
collections
in
here,
so
these
collections
are
kind
of
like
these
high-level
references
to
different
sets
of
data
that
might
be
grouped
in
various
ways
and
then
they
can
start
pulling
down
these
data
sets.
Everything
is
backed
by
candice
as
well.
G
So
it's
easy
to
look
at.
We
can
kind
of.
We
can
look
at
what's
been
computed.
We
can
query
new
data
into
these
super,
their
notebooks.
We
have
visualization
through
plotly
we're
working
with
the
bio
folks.
Do
some
local
visualization
as
well.
So
that's
all
coming
and
we
can
do
all
kinds
of
interesting
diagnostics
and
statistics
right
there
in
these
notebooks
and
so
from
these
notebooks.
You
can
also
spawn
you
compute
as
well.
So
within
a
couple
lines
of
code,
you
can
kind
of
like
look
at
a
molecule
and
say
hey.
G
I
want
to
run
this
molecule
and
it
will
go
out
to
the
network
and
be
distributed,
and
we
have
a
variety
of
these
different
notebooks
here,
but
the
overall
synopsis
is
that
we're
really
looking
into
jupiter
notebooks
as
a
way
to
distribute
computation
and
life
sciences
data.
This
involves
not
only
these
notebooks,
but
also,
how
do
we
get
journals
involved?
So
we're
talking
to
our
journal
editors,
who
are
actually
interested
in
saying,
like
hey,
hey
this
notebook
actually
via
supplemental
information.
G
So
if
someone
wants
to
pull
down
this
data,
investigate
this
data,
you
know,
can
this?
Can
this
kind
of
scheme
work
for
that
as
well?
So
this
project
is
fairly
new.
It's
about
three
months
old
at
this
point,
so
I
just
wanted
to
kind
of
give
like
a
quick
rundown
of
what
we're
doing
and
really
looking
for.
G
B
G
B
C
C
Awesome
well
Daniel.
Thank
you
so
much
for
presenting
awesome
stuff
going
on
there
and
and
stayed
stay
connected.
Sir
we'd
love
to
work
yeah.
We
appreciate
the
contributions
back
as
well.
This
is
it
was
cool
to
see
that
stuff.
So
we're
we're
over
time
right
now,
so
Nick
I
see
that
you
put
a
last
little
point
here.
Did
you?
Is
it
just
a
quick
announcement
about
this
accessibility,
octagon
well,
I,
think
Nick's
mighta
left
so
there's
an
accessibility
hackathon
on
Jupiter
lab.
C
You
can
check
that
out
on
May,
13th
and
14th,
so
there's
a
link
in
the
agenda.
Thank
you
to
everyone
who
came
our
I
just
want
to
remind
everybody
that
there's
a
there's.
A
discourse
thread
opened
up
about
these.
All
Jupiter
community
calls
we're
going
to
try
to
do
this
monthly.
This
is
exactly
the
format
I
was
expecting.
So
this
is
great.
Thank
you
to
all
the
presenters
with
a
lot
of
fun.
It
was
good
to
see
you
all
yeah.