►
From YouTube: IPython/Jupyter Dev Meeting
Description
Meeting of the IPython/Jupyter development team.
A
Welcome
Internet:
this
is
the
July
21
2015
ipython
Jupiter
def
meeting
and
we'll
get
going
with
our
note,
taking
I
guess
since
I'm
Tinson
speaking
I'll
just
go
first
with
a
quick
report,
mostly
communications,
a
little
bit
of
the
usual
background
work
that
always
keeps
me
busy
on,
but
I
did
get.
I
guess
that
I
did
actually
put
some
productive
time
into
the
project
with
jupiter
day
last
friday.
We
all
worked
on
that,
but
we
had
a.
A
We
basically
had
a
all
day
event
at
lbl,
a
jupiter
day
which
was
kind
of
a
dry
run
off
I.
Guess
what
will
become
Jupiter
cons
in
the
future?
It
was
a
sequence
of
short
technical
talks
on
the
various
aspects
of
the
project.
It
was
very
useful.
We
gave
a
series
of
presentations,
we
have.
A
repository,
will
put
the
link
up
on
the
hack
pad
to
the
repo
I
think
mean
I,
don't
know
if
your
slides
are
up.
I
haven't
put
them
in
there.
A
A
Cool
and
I
need
to
check
in
the
agenda
because
that'll
serve
output,
the
agenda
into
the
readme
that
also
kind
of
as
a
guide
as
a
guide
for
what
we
did
through
the
day.
But
it
was
very
useful.
We
presented
the
various
aspects
of
of
the
project
as
a
sequence
of
short
talks,
and
then
people
from
LVL
presented
short
lightning
talks
and
what
they
are
doing
with
the
project,
which
was
very
useful
feedback.
A
Their
lightning
cops
are
checked
into
the
repo,
so
people
can
see
what
they
presented
and
then
we
had
a
short
section
of
discussion
which
most
of
the
questions
centered
around
Jupiter
hub
and
customized
colonel
deployments,
and
so
we
ended
up
having
about
a
30
minute
demo
of
basically
just
working
on
how
she
deployed
she
did
her
complicated
to
get
her
hub.
The
point
it
was,
it
was
a
very
productive
day
and
I
think
I'll
leave
it
at
that
to
keep
us
moving
and
anything
else.
F
E
So
the
I
pine,
the
metadata
stuff
is
in
so
that's
great
and
I
still
do
a
lot
of
what
it
means
to
do,
but
it
will
fully
round
trip
all
the
metadata
and
try
to
strip
out
metadata
that
people
don't
probably
want
to
version
or
have
other
people
working
on
in
github
and
cetera,
et
cetera.
The
next
thing
we're
going
to
be
adding
are
some
little
killed.
E
My
other
stuff
are
some
little
shortcuts
that
let
you
have
kind
of
like
scotch
of
cones
thing
where
you
could
say
like
this
next
chunk
of
markdown
as
a
slide,
but
then
work
with
the
rest
of
the
reveal
things
for
it's
a
slide
or
it's
a
it's
a
fragment
or
whatever
and
been
talking
with
Damian
on
Rhys
packaging.
So
right
now
it's
kind
of
kind
of
have
to
clone
the
repo
or
use
the
bookmarklet
or
some
like
that,
and
so
we're
working
on
figuring
the
story.
E
There
I
think
somebody
else
probably
going
to
talk
a
little
bit
about
this,
but
we
are
going
to
be
needin
some
additional
help
that
will
help
Conda
be
probably
the
optimal
way
to
install
it,
because
it
will
be
able
to
activate
the
MV
extension
after
it's
installed.
I've
been
working
with
knut,
on
which
I
love
using
in
the
notebook
to
make
sure
that
it's
the
build
going
forward
is
all
require
JS
functional.
E
It's
if
you
haven't
used
mermaid
before
it
lets
you
draw
nice
sequence,
diagrams,
Gantt,
charts,
flowcharts
a
couple.
Other
things
would
start
a
little
mark
down
like
language.
It's
a
really
nice
way
to
get
struck
diagrams
in
and
manageable
into
a
notebook,
and
then
I
finished
up
my
first
round
of
work
on
I
pilot
coder,
which
is
just
a
little
thing
lets
you
write
just
just
the
update
method
from
a
widget
and
automatically
rerun
that
every
time
you
change
the
code
or
you
change
the
treatments
that
drive
it.
E
So
it's
a
really
fun
little
way
to
introduce
somebody
to
hacking
on
real
widgets
I
want
to
get
to
where
that
generates
old
package
from
just
the
widget
definition
and
I.
Think
that's
pretty
possible!
So
I
guess
that's
about
that
there
and
going
to
be
talking
a
little
bit
more
about
my
little
tool
to
been
working
on
for
the
narrative,
driven,
storytelling
and
modeling
stuff
and
still
working
on
getting
that
open
sourced
and
get
in
our
IPO
there
in
Buddhism
result.
G
E
E
We
have
so
I
got
some
guys
really
excited
about
the
last
thing.
I
did
with
these
telescopes
and
all
and
all
that
stuff,
though
we're
looking
at
a
trapping
an
existing
rest
framework
into
widgets,
so
that
you,
you
know
you're
using
this
rest
brain
work
that
we've
exposed,
but
you
know
type
it
on
stuff
there.
So
that's
God's
people
really
excited
about.
You
know
especially
the
stuff
in
new
Wi-Fi
widgets,
where
you
can
add
the
trade
'let's
and
stuff
at
runtime.
So
that's
that's
pretty
cool,
but
then.
G
A
A
J
G
Whole
day
was
what
else
so
I
I
didn't
get
a
talk
in
last
week's
meeting.
So,
of
course
a
bunch
of
things
happened
at
Syfy.
You
know
normal
hacking
on
things,
but
in
specific
I
want
to
talk
about
a
project
that
Kyle
and
I
started
working
on
called,
transform,
I'm
and
I.
Think
yeah
I
think
men
may
have
mentioned
it.
In
last
meeting
I
forget
I
watched
the
recording
somebody
mentioned.
G
So
Kyle,
of
course
he
has
a
sidecar
application,
which
is
just
something
that
that
renders
outputs
only
outputs
of
an
existing
ipython
session
and
right
now
in
his
side,
Cara
application
he's
kind
of
re-implemented
the
output
area
from
scratch.
So
that's
a
little
frustrating
having
that
in
two
places,
so
we
decided
that
it
would
be
a
good
idea
to
start
trying
to
take
the
output
area
out
and
make
it
something
that's
reusable
so
first
we
start
it
with
web
components
and
we
realized,
as
we
were
working
on
it,
that
there's
really
a
more
generic
piece.
G
I
K
G
Is
a
project
that
just
takes
mimetypes
or
mime
type
bundles
and
create
HTML
elements,
so
I've
been
working
on
that
with
him
we're
trying
to
get
some
demos
together,
the
show
everybody
I
think
he's
actually
using
it
inside
sidecar
now
successfully.
Mom
is
N
and
I
think
I
think
we'll
who's
on
on
the
chat
today,
I
think
will
was
going
to
investigate
using
it
to
that's
right,
yeah
and
the
two
more
things
I'm
going
to
PI
data.
So
I'm
working
on
a
talk.
G
I'm
gonna
talk
about
ipython
and
Jupiter
past
present
in
future
and
kind
of
the
greater
Jupiter
org.
So
it's
very
similar
to
a
talk
that
Fernando
gave
recently.
In
fact,
men
sent
me
the
slides
to
it.
Oh
I'll
be
plucking
plucking
information
from
that
and
then
lastly,
I
wanted
to
mention.
There's
a
problem
with
the
widgets
right
now,
I've
been
kind
of
looking
at
it
as
I'm.
L
That
was
actually
I
was
gonna,
suggest
that
so
I've
been
so
sorry
if
Stephen
Stephen
Dave
are
out
for
various
reasons
this
week,
but
they've
been
basically
continuing
the
work
that
they
did
from
last
week,
Steve
still
working
with
Matthias
and
Jason
grouts
doing
the
support
of
the
front
end
notebook
the
that
to
tech
script
and
you're
working
on
starting
to
tease,
tease
out
what
the
core
interfaces
of
that
should
be
and
how
the
whole
new
front
end
should
be
built
up.
Dave
is
still
working
on
the
ipython
cutie
like
console
for
the
browser.
L
L
Console
that
works
for
me,
okay,
w
on
my
end,
I,
had
a
conversation,
late
Friday
night
with
Brian
and
his
students
about
the
phosphor
in
general
and
what
we
can
do
to
kind
of
have
something
that
provides
kind
of
like
a
standard
hierarchy
that
people
can
use
to
build.
You
know:
custom
Dom
things
that
have
lifecycle.
Events
like
attached,
detached
show/hide
events
and
the
kinds
of
things
that
propagate
correctly
throughout
the
entire
hierarchy,
which
you
don't
get
with
backbone.
L
So
you
could
use
these
kinds
of
widgets
as
a
base
hierarchy
to
do
things,
whether
to
make
basically
make
reusable
components
whether
they're
used
in
CSS
layout
for
another
regular
standard,
phosphor,
absolute
Lance.
So
that's
kind
of
the
idea
that
I'm
playing
with
right
now,
it's
a
very
tricky
problem
to
get
that
elegantly
working
together,
but
it's
so
far
it's
going
good.
I'm
right
now,
I'm
optimistic
that
it
can
be
done,
but
I'm
kind
of
in
the
middle
of
it.
L
So
it's
about
where
I
but
kind
of
maybe
I,
don't
know
how
well
that
ties
into
the
stuff
that
that
John
and
Kyle
are
doing
right
now
it
sounds
like
it
may
be.
Tangentially
related,
but
I
think
it
will
certainly
help
the
guys
at
Cal
Poly
in
the
immediate
term
at
maybe
also
the
court.
Would
you
print
work
in
the
longer
term,
Oh.
A
Chris,
if
I
may
ask
you,
I
was
directly
trying
to
take
notes,
but
between
the
fact
that
you
talk
really
fast
in
the
fact
that
the
ACT
pad
has
horrific
lag.
I
I
just
wasn't
able
to
so
may
when
the
call
wraps
up,
if
I
could
ask
you
just
to
take
two
minutes
to
summarize
your
own
notes,
just
so
that
the
hack
pad
has
some
notes,
or
maybe
right
now
what
somebody
else
is
speaking
if
you
could
just
type
up
a
quick
summary
of
what
you
just
said,
so
that
we
absolutely
basically
a.
A
C
L
I
A
I
Had
a
good
discussion
and
thanks
everyone
that
participated
in
that
we
we
had
some
direction
going
forward,
so
they're
working
on
implementing
those
changes.
Also
Chris,
culvert
and
I
were
asked
to
give
a
strata
Hadoop
talk
in
September
here
in
New,
York,
so
an
hour
and
a
half
on
ipython
notebook,
the
Jupiter
notebook,
something
of
maybe
a
little
tutorial
as
well
as
some
more
advanced
stuff.
So
we
might
be
taking
some
of
the
stuff
you
guys
are
done
or
doing
our
own
stuff
as
well
by.
A
A
Ben
Ben
I,
I
spoke
with
bender
rika,
yesterday's
I,
just
I
saw
him
in
the
city
and-
and
he
was
asking
me
about
who's,
going
to
be
there
to
make
sure
that
we
have
enough
presence
at
the
open
source
table
since
they're
going
to
have
open
source
boots.
This,
for
the
first
time
at
strata,
they're
going
to
have
kind
of
a
community
community
projects
area,
so
it'll
be
it'll,
be
good
for
us
to
be
well
represented
in
the
open
source
boots
area.
So
I'm
glad
to
hear
that
enough
of
us
are
gonna.
E
J
A
A
Right,
okay!
Well
at
least
it
seems
like
we
have
already
for
people,
so
it's
not
only
Brian,
so
I
don't
want
the
load,
the
load
to
be
only
only
on
Brian
at
strada.
It
would
just
over
overload
him
too
much,
but
but
if
Kai,
if
Kyle
and
Jason
and
Chris
or
there,
that
that
allows
that
allows
for
good
coverage
so
that
Brian
isn't
completely
overwhelmed
good.
I
No
problem,
so
just
can't,
like
Chris
mentioned
just
continue.
Work
on
the
phosphor
notebook
in
particular
I
mean
right
now,
we're
just
concentrating
on
lower
level
stuff
sort
of
sessions
and
had
notebook
objects
and
things
like
that.
Nothing
higher
level
just
yet
and
then
I
just
want
to
mention
that
typescript
1.5
has
been
released
so
we're
official
now
using
1.5
and
we've
upgraded.
The
foster
notebook
and
chris
has
already
upgraded
to
the
phosphor
package
itself
is
well
to
type
script.
1.5.
C
D
C
Know,
but
so
as
of
today,
I
have
permission
to
live
in
Norway,
so
I
will
actually
be
doing
that
at
the
end
of
the
summer.
Over
the
past
week,
other
than
preparing
for
the
the
Jupiter
day,
I've
been
working
on
for
quenelle
release
stuff.
We've
now
released
almost
every
package
for
four-point-oh
it,
the
ones
that
are
left
our
ipython
itself,
the
notebook
package,
the
widgets
package
and
then
the
Connecticut
trivial
meta
package.
C
We've
also
really
things
not
really
24
point
on
which
or
two
per
of
an
env
crater,
which
released
attacked
by
primary
stuff
between
us
and
four
point:
oh,
are
the
docs
in
the
new
kind
of
the
new
central
Jupiter
docs,
that's
kind
of
the
landing
page
with
install
information
and
and
overview
stuff.
That,
honestly,
is
the
primary
thing
in
between
us
and
releasing
four
point.
Oh
there's
very
little
code
left
to
right.
C
The
code
left
to
right
that
does
exist
is
really
dealing
with
some
of
the
install
and
build
stuff
on
the
notebook
repo
and
the
widgets
repo
yeah,
and
something
I
have
left
to
do
is
enter
clear
in
this
meeting
is
I
need
to
put
my
stuff
in
the
Jupiter
day,
repo
that
I'll
do
after
this
meeting
yeah,
but
we're
we're
very
close
to
four
point.
Oh
very
close,
yeah.
M
A
K
K
So,
for
the
last
few
days,
I've
been
working
with
Cameron
a
lot
on
the
redesign
of
the
Jupiter
website
and
we've
been
sort
of
bouncing
back
and
forth
with
me,
developing
content
him
trying
to
consume
that
content
and
then
sort
of
kicking
it
back
to
me
saying
hey
that
that's
that
sentence
is
way
too
long
or
it's
sort
of
there's
a
feedback
loop
that
we
have
set
up
between
the
visual
design
and
the
content,
and
it's
coming
along
very
nicely.
I'm
really
really
excited
I.
K
Think
some
of
you
have
seen
some
earlier
prototypes
and
it's
moved
forward
considerably
since
then,
and
next
I've
been
working
on
the
the
top
level
documentation
that
min
mentioned.
I
took
another
pass
at
it
and
still
still
playing
with
different
aspects
of
the
organization
and
I'm
going
to
try
to
get
done
before
I
leave
for
pi
data
on
midday
thursday,
and
so
hopefully
that
will
stop
being
a
bottle.
Next
to
the
release.
D
K
Lots
of
conversations
with
various
salts
about
Jupiter
spark
integration
and,
in
particular,
trying
to
to
spawn
conversations
amongst
various
stakeholders
that
are
interested
in
Jupiter's,
then
spark
and
getting
Oberon
plugged
into
that.
Also.
Having
conversations
both
on
some
issues
in
our
repo
and
phone
calls,
and
so
on
about,
we
can
use
docker
more
effectively
and
one
of
the
questions
is:
what
do
we
do
about
the
growing
problem
that
we
want?
We
want
that.
K
You
can
imagine
like
a
data
science
staff
that
has
Python
our
Julius
Kayla,
spark
that
people
can
literally
just
say,
go
and
get
the
notebook
with
those
kernels
and
then
I've
also
been
working
on
the
organizational
aspects
of
the
phosphor
notebook
fernand
and
we
did
a
doodle
poll
over
the
last
few
days
and
so
that
the
meeting
for
that
is
today
from
noon
to
three.
If
this
meeting
here
goes
a
little
bit
late,
we
could
definitely
postpone
that
till
twelve
thirty
or
something
to
let
people
get
lunch
on
the
west
coast.
D
K
I
H
So,
like
it
says
in
the
dock,
then
I've
mostly
been
working
on
talking
to
people
about
the
proposed
completion,
spec
and
getting
feedback
on
that.
A
lot
of
people
have
had
like
really
good
suggestions
for
things
to
add
or
things
to
change.
I
think
that
with
some
additions
like
including
documentation
and
docs
links,
and
things
like
that,
then
this
spec,
roughly
in
its
current
form,
will
probably
work.
H
F
J
J
F
So,
as
Brian
mentioned,
I've
been
looking
at
some
of
the
more
Jupiter
in
spark
integration.
I've
had
some
conversations
with
men
and
a
couple
other
stakeholders
outside
of
the
Jupiter
organization,
we're
starting
to
look
into
what
kind
of
work
we
want
you
on
our
side
and
what
kind
of
work
we
want
to
do
by
going
and
making
requests
to
the
actual
pie.
F
Spark
developers
I've
also
started
looking
into
deploying
ipython
parallel
on
yarn
and
what
I
started
to
experiment
with
is
a
project
called
Apache
slider
that
would
allow
us
just
to
kind
of
package
it
and
not
have
to
change
much
of
the
ipython
parallel
machinery
get
it
running
on
yarn,
mom.
K
O
O
Mockup
side
of
making
refining
them
day
/
day,
and
so
I
should
have
those
done
either
by
the
end
of
this
week,
or
maybe
the
early
next
week
have
those
up
and
we
could
start
putting
that
together
through
an
actual
website,
and
that
was
pretty
much
it
I'm,
just
focusing
on
the
website
manly
right
now.
So.
H
I
did
a
coke
review
with
Brian
over
the
weekend
and
I
think
I
resolved
where
I'm
using
a
new
workflow
right
now
and
it's
making
the
refactoring
go
a
lot
faster.
I
have
some
small
scripts
that
I'm
just
using
to
do
some
of
that.
For
me
and
yeah,
it's
going
a
lot
smoother
now.
Oh
yeah.
P
I
K
Hey
so
I'm
having
a
bad
network
connection,
so
if
I
freeze
up
feel
free
to
jump
pass
me
because
the
whole
meeting
keeps
reasoner
I'm
outdoors
in
Seattle
yeah,
so
I've
been
working
on
the
transfer
amount
of
stuff
with
John
I
didn't
actually
get
to
here
with
John
covered,
but,
like
my
goal
has
been
to
build
up
towards
components
that
like
Deeb
needs
for
like
the
o'reilly
stuff,
but
also
toured
stuff.
That
can
be
shared
across
a
bunch
it
over
the
context.
K
B
A
B
J
A
Good
to
have
enough
notes,
remove
it
from
everybody.
Okay,
I
think
we've
managed
to
hit
everything,
and
we
have
a
few
minutes
left
in
the
window
of
time
in
case
there's
any
kind
of
top-level
discussion
or
questions
or
I
mean
I.
Think
it's
a
good
idea
to
try
to
cover
the
reporting
from
everybody
yeah
in
time
and
then,
if
we
have
a
window
of
time
left,
then
we
can
have
kind
of
open,
open
discussion
in
case
there's
some
brainstorming
that
people
would
like
to
cover
I
think.
I
N
It
so
so
to
me,
I
mean
TV.
Signals
are
really
different
from
signal
signal
is
really
different
from
you
know,
events
in
Kovacs
and
that
there
is
sort
of
static,
typing
new
signature,
so
in
that
they
are
very
similar
to
tread
let's,
and
so
the
basic
idea
of
this
tiara
is
that
all
the
communication
between
widgets
happens
on
the
Java
subside,
but
you
wire
them
from
the
tip
I
side
with
a
cutie
star
API
and
the
signature
of
the
core
bags
and
signals
is
given
via
a
trade
type
mm-hmm.
L
A
N
In
the
description
of
the
TR,
actually
explain
its
morph
protocol
to
as
a
front-end
to
create
a
new
signal
connectors
in
your
emitter
signal,
so
besides
backbone,
update
message
and
the
other
message
type
that
we
had
already,
which
are
display
and
custom
I,
think
now
we
have
connect
disconnect
M.
It
invokes
so
that
any
back-end
can
actually.
You
know,
speak
that
first
of
all
and
implement
signaling
for
ipad
languages.
N
So
called
silence
now
so
yeah
and
by
the
way,
actually.
So
this
is
only
for
the
front
and
side,
but
there
there's
another
subject
and
credits
which
I
think
we
could
start
discussing
is
to
have
also
a
Python,
only
signaling
library,
because
there
are
many
things
which
I
mean.
Trade
has
really
come
very,
very
well
in
the
ocean
of
state
and
how
to
synchronize
state
between
different
models
and
different
hazards
instances.
But
we
don't
have
such
a
high
level.
33
types
invalidated
API
for
events
at
the
moment.
C
N
N
As
I
no
Gerta
is
not
dependent
signals,
okay,
sing-off
dependent
red
lights,
so
every
signal
is
a
descriptor
that
ears
from
the
base.
The
scripture
that
designing
the
trellis
library,
so
it
can
has
resistance,
can
hold
trade
types
but
also
signals
and
slot
and
and
every
signal
holds
a
trade
type
corresponding
to
each
signature
so
to
the
signature
of
the
core
back
set
much
that
can
be
registered
to
this
to
the
signal.
N
N
B
N
N
C
C
A
E
D
K
D
K
I
K
K
A
Party
anything
else,
we
aren't
I'm
we're
not
over
already
hey
okay,
goodbye
internet
and
see
you
all
later,
buttocks.