►
From YouTube: 2015-12-01 IPython/Jupyter Dev Meeting
Description
Meeting of the IPython/Jupyter development team.
B
Awesome
so
most
of
my
work
over
the
past
week
has
been
documentation.
Type
stuff,
I
spent
a
ton
of
time,
cleaning
up
our
ops
handbook,
which,
if
you
have
access
to
that
internally,
you
can
check
it
out
and
look
how
it
looks.
I've
always
wanted
marking
our
game.
B
B
If
you
want
to
check
it
out,
I
also
opened
up
an
issue
like
early
last
week
regarding
sort
of
standardizing
a
little
bit
and
approach
to
tagging
issues
or
just
making
sure
that
we
have
each
issue
tagged
across
all
our
repositories,
to
make
it
easier
for
new
contributors
and
also
want
to
want
to
check
it
out
and
give
their
feedback.
B
My
side,
project
of
building
a
racket
kernel
and
then
also
documenting
the
process,
has
been
going
pretty
well.
I've
got
a
pretty
lengthy,
blog
post
going
about
building
kernels.
If
you
want
to
kind
of
check
it
out,
while
it's
in
progress
I'll
be
glad
to
figure
out
how
to
share
some
tumblr
and
get
that
out
to
you
also
been
working
on.
Some
random
could
really
at
PRS
here
and
there
across
the
different,
the
different
repos
and
finally,
I
opened
up
the
call
for
proposals
for
Jupiter
day
Chicago
and
I
tweeted
it
out.
B
A
Thanks
I
think,
and
it's
like
a
Peter
and
Gino
next
sure.
D
Hi
we
haven't
been
in
a
couple
weeks,
so
this
is
kind
of
a
brain
dump.
Things
have
been
going
on
just
make
sure
we're
writing
things
down,
but
I
think
most
important.
We
started
doing
pi
PI
releases
of
the
incubator
projects,
sort
of
making
them
independent
of
the
docker
image
that
we
put
up
there
earlier
just
to
get
people
to
try
it
so
there
now
installable,
we're
actually
gonna
do
a
release.
D
Sometime
this
week,
maybe
today,
where
we've
moved
the
notebook
related
extensions
up
to
Jupiter
for
da
no
compatibility
that
were
back
on
32,
so
the
port
was
actually
pretty
straight
for
I.
Think
we
open
a
couple
bugs
against
notebook
and
I
pipe
widgets,
and
so
on
related
that
so
it
was
good
exercise
go
through
like
we
talked
about
in
new
york
with
brian
and
others.
D
On
the
dashboard
side
of
things,
we've
been
starting
to
focus
more
on
the
back
end
pieces,
because,
ultimately
we
know
the
shift
is
going
to
be
more
towards
using
phosphor
in
the
fraud
dashboarding
with
that.
So
did
a
analysis
of
the
security
of
deploying
dashboards
externally
from
Jupiter
and
looking
at
how
that
might
work.
You
know
what
what's
missing
today
with
respect
to
like
API
keys
and.
E
D
Gateway
and
things
like
that
and
that's
posted
to
the
wiki
on
the
github
repo
for
dashboard,
so
anyone
that
wants
to
take
a
look
at
that
I
appreciate
the
feedback,
be
great
to
get
some
feedback
on
that
along
the
way.
We're
also
at
least
looking
at.
You
know
what
it
means
to
move
the
dashboard
work
that
we've
got
using
you
actually
using
Jupiter
j/s
services
and
Jupiter
j/s
output
area.
You
know
the
new
goods,
we
have
a
branch
with
that,
actually
mostly
working,
but
we're
kind
of
dependent
on
ipad
widgets,
reaching.
D
They're,
actually
starting
to
use
Jupiter
jes
services
to
I
forget
the
exact
details.
I
wasn't
the
one
working
on
that,
but
basically
we're
putting
that
in
the
shelf
and
we're
willing
to
help
out
with
the
I
widgets.
If
there's
some
work
there
that
we
can,
you
know,
help
contribute
to
get
that
that
rolling
too,
but
I
think
somebody
pointed
out.
You
know
five
dot
over
I
PI
widgets
is
probably
one
that'll
land,
so
we
show
that
for
the
time
being,
but
again
it
was
an
exercise
to
see.
D
If
we
can,
you
know
continue
to
roll
forward.
We
can
on
the
colonel
gateway
side.
Again
we
did
a
release,
it
turns
out
I.
Think
Kyle.
You
pointed
out
it's
actually
being
used
in
the
atom
notebook
now,
so
I
was
a
nice
surprise,
someone's
actually
taking
advantage.
So
that's
good
and
we
cut
a
stable
release.
I
think
it
was
a
couple
weeks
ago
and
we're
kind
of
experimenting
with
what
other
types
of
api's
and
protocols
we
can
expose
from
the
colonel
gateway
to
the
outside
world,
while
keeping
zeromq
on
the
back.
D
So
one
we
had
talked
about
earlier
was,
you
know,
just
limiting
channel
access
so
that
you
could
have
a
publish
only
current
óleo
something
use
server,
sent
events,
that's
on
the
radar
we're
also
looking
at
this
crazy
idea.
I
had
just
as
an
experiment
of
game.
We
turn
a
notebook
with
annotations
into
it
into
a
sort
of
a
micro
service,
a
rest
api
in
the
outside.
D
So
you
could
actually
reuse
the
colonel
running
that
notebook
but
expose
a
REST
API
on
the
outside
to
certain
cells
or
certain
functions
in
it
instead
of
the
full
Jupiter
protocol.
So
again,
you
know
Colonel
gateway,
being
incubator,
we're
sort
of
experimenting
with
ideas.
They're
definitely
welcome
some
feedback
on
that
as
well.
But,
ultimately,
if
we
want
to
mature
this
out
of
there,
we
could
refactor
and
split
out.
You
know
crazy
features
from
you
know
things
that
should
be
core
features.
F
A
Various
mundane
things
related
to
merging
many
PRS
on
many
repos
and
bug
fixes
all
over
the
place,
cut,
I,
think
up
by
Colonel
42
last
week
with
some
fixes
for
opening
new
comms
that
will
help
some
widgets,
tough
and
then
doing
participating
in
the
lengthy
conversation
about
multi
selection.
You
I
that
I
think
is
coming
to
resolution
all
over
the
primary
contributors
on
that
Brian
Matthias
are
not
here
today.
A
It
also
been
spending
some
time
on
a
futures
but
I
rewriting
some
of
the
client-side
stuff
I
python
perl
using
futures
that
should
clean
up
some
of
the
temporary
state
and
get
a
executor
api
out
of
it.
It
should
be
nice
that
some
people
have
been
asking
for
and
then
thomas-
and
I
will
be
heading
to
sir
next
week
for
a
few
days
for
pork,
shoppe
talking
about
jupiter,
hub
and
I,
Python
parallel
and
a
root
kernel
and
various
other
Jupiter
at
CERN
type
topics.
H
Okay,
guess
I'm
next
I've
built
out
to
pay
some
examples
on
phosphoryl
plugins
to
show
how
you
can
use
the
plugins
from
either
system
j
s
or
steel
jeaious.
Look
at
this
options
there,
which
everyone
ends
up
working
better.
We
can
choose
to
go
with
that.
One
also
made
a
plug-in
for
phosphide
that
allows
you
to
load
and
unload
CSS
from
the
page
as
part
of
a
plug-in.
I
And
so
it's
got
like
a
premier
editor
in
there
right
now
and
a
bunch
of
brocade
pots,
because
these
are
the
sorts
of
things
that
are
going
to
come
out.
But
it's
a
sort
of
meant
to
just
show:
that's
not
going
to
be
a
major
hassle
going
forward.
There's
still
a
bunch
stuff.
I
need
to
do
to
clean
it
up
a
bit
but
I.
Think
after
that's
done,
it'll
be
kind
of
one
things.
We
could
stick
right
in
the
center
of
of
an
ID.
Basically,
so
that's
that's!
J
All
right,
the
gum
x,
don't
spend
the
last
week,
basically
bringing
in
the
phosphide
and
due
to
repost
back
up
to
speed
with
the
for
the
demo,
so
just
got
to
make
demo
working
again.
I
haven't
pushed
any
of
the
changes
there,
one
folks
just
to
make
sure
that
everything's
kind
of
sensible
before
we
before
we
push
them,
but
I
hope
you
have
the
the
main
app
working
again
that
people
can
play
around
with
today
or
tomorrow,
and
then
next
chris
is
just
a
few
minutes
ago
pushed
an
update
to
the
command
staff.
K
K
It's
got
one
major
missing
feature,
and
that
is
that
you
cannot
start
a
drag
and
drop
operation,
programmatically,
presumably
the
guys
as
security,
because
you
don't
want
programmatic
access
to
the
clipboard
and
that's
all
well
and
good,
but
it
basically
means
we
can't
do
anything
that
we
can't
start
any
kind
of
drag
operation.
Well,
we're
also
handling
mouseevent.
So
if
you
wanted
some
nice
interactions,
where
you're
moving
tabs
around
in
a
tempo
or
and
then
want
to
tear
one
of
them
off,
you
cannot
ride
on
top
of
the
drag-and-drop
API.
K
For
that,
because
you're
already
handling
Mouse
events,
which
means
you
kill
the
drag
and
drop
events.
So
so,
basically,
we've
got
this
big,
gaping
hole
of
functionality
running
on
the
native
API
and
then,
in
addition
to
that,
IE
doesn't
implement
custom
support
for
custom
I'm
types
in
the
drag
and
drop
API.
So
basically,
we
can't
use
the
native
DRAG&DROP
API,
all
of
that
unwinds
to
mean
that
we
had
to.
Basically
we
implement
the
drag
and
drop
API
from
scratch
in
Java
scripts
to
support
custom
I'm
types
and
also
allow
starting
drag
and
drop
operations.
Programmatically.
K
Of
course,
this
is
only
limited
to
dragon
drops
in
app
because
we
don't
have
programmatic
access
to
the
clipboard
or
the
native
drag
drop
you
guys,
but
anyway
we
now
have
an
API
that
does
what
we
want
and
is
clean
and
be
used
to
drag
arbitrary
stuff
around.
Within
your
application.
The
dock
panel
has
been
updated
to
use
that
tab
bars
have
been
updated
to
use
that
to
all
that's
working
now
all
do
releases
or
cuts
the
website.
K
K
In
addition
to
that,
I
updated
phosphor
commands
after
some
back
and
forth
well
decided
that
it's
better
to
have
commands
big
arguments,
because
that'll
ease
a
lot
of
pain
with
some
of
the
things
that
we
want
to
do
in
the
context
of
an
IDE
such
as
programmatically,
selecting
a
tab
in
the
dock
area,
programmatically
opening
particular
sidebar,
widgets,
etc.
So
updated
foster,
commands
to
an
athlete
arguments,
so
that'll
start
being
a
starting
start
to
integrate
that
into
the
actual
phosphide
here
shortly
and
then
just
some
other
various
tweaks
to
pause
for
plugins.
Those
related.
A
E
K
K
L
L
So
I'm
pretty
much
gonna
finish
that
up
in
he,
starting
on
Jupiter
mock-ups,
get
back
on
the
workbench
and
take
a
second
pass
of
the
hope
and
dream
kid
logo
for
Matthias,
and
that
group
and
I
don't
know
just
ping
me
on
get
out
of
here.
You
have
a
problem.
That's
it
right!
It's
a
bird
couple,
intent.
A
Okay,
do
we
have
Tamia.
M
Trying
to
package
things
in
the
current
state
of
the
notebook
and
AE
is
to
provide
people
with
some
documentation,
template
and
kind
of
a
partial
solution
of
potassium
and
the
extension
and
the
server
extensions
in
a
quick
and
easy
way
using
ganda,
but
also
later
a
person
Philadelphia
mainly
using
on
and
solve
the
problem.
So
we
are
working
on
on
that
right
now.
E
What
okay
go?
First,
let
me
turn
back
on
my
camera,
so
I
was
out
for
the
holidays
for
the
last
week,
but
I
gave
a
talk
at
BYU
and
the
Jupiter
notebook
they
all
liked
it
and
worked
a
lot
at
this
age
days
in
San,
Francisco
and
then
later,
with
Matthias
and
Fernando
at
Bloomberg
in
real
time
collaboration.
So
a
lot
of
design,
discussions
and
I
think
these
have
been
summarized
and
posted
up
and
then
worked
a
lot
on
the
Jupiter
jf
input
area
which
probably
will
be
merged
today
or
tomorrow.
E
N
N
O
With
me,
ok,
hey
guys,
can
you
hear
me
yeah?
O
Ok,
I
got
new
headphones,
so
it's
figuring
how
they
work
so
yeah
I've,
been
mostly
doing
Thanksgiving,
which
was
great
I,
was
disconnected
for
a
couple
isms
really
nice,
but
in
that
I
did
get
out
a
and
be
present
and
envy
browser
least
that
I
feel
like
our
are
ready
to
be
looked
at
other.
You
know
still
dev
quality
and
still
a
lot
of
stuff
going
on
there.
But
it's
been
a
lot
of
interesting
work
there.
O
That's
part
of
some
of
the
stuff
working
with
Damien
and
in
some
related
conversations,
have
been
talking
about
what
a
Jupiter
extension,
yeoman
generator
would
look
like
with
the
different
kinds
of
features
in
different
places
and
handling
all
the
config
stuff,
so
probably
raise
that
to
the
list.
That's
about
it.
G
Hey
everyone,
sorry
I
was
late,
so
I
didn't
have
Thanksgiving
here
in
Europe,
because
we
don't
know
what
it
is.
I
work,
mostly
on
multi-select
I
finished
in
to
select
if
people
want
to
try
mean
tried
and
already
gave.
Some
feedback
is
mostly
happy
with
that,
except
for
duplicate
variables,
and
that
would
be
one
of
the
last
huge
things
for
4.1
and
I
also
spend
sometimes
pure
coding,
with
a
new
contributor
I'm
teaching
him
again,
the
in
terminal
ipython,
and
we
are
trying
to
take
care
of
a
few
bugs.