►
From YouTube: IPython/Jupyter Dev Meeting, May 17th 2016
Description
Meeting of the IPython/Jupyter development team.
A
A
So
if
you
require
a
courier
and
you
load
the
mud
with
courier
or
an
extension
or
something
else,
it
cannot
affect
the
instance
of
the
module
which
is
used
by
cut
cells
and
everything,
otherwise,
something
prevent
camel
phone
rings
which,
if
your
camel
is
broken
with
net,
if
you
have
many
environment
and
one
environment
is
disappeared,
you
literally
cannot
get
a
notebook
that
work
together.
I
think
it
would
be
really
important
to
fix
that
and
make
a
four-point,
2.1
or
4.3
relatively
soon.
B
C
B
D
A
A
E
Here
so
I
just
a
quick
thing,
which
is
that
the
Jupiter
education
survey
is
closed
and
I've
released.
The
data
set
on
github
I'm
still
working
on
doing
some
additional
analysis
on
it.
But
if
people
are
interested,
you
can
go
ahead
and
take
a
look
at
it
and
when
I
have
at
least
some
analysis
done
on
it,
I'll
send
an
email
out
to
all
the
mailing
lists,
oh,
and
also
that
the
repository
that
it's
in
on
github
is
like.
We
created
it
specifically
for
Jupiter
datasets.
E
A
C
We're
unmuted
now
I
am
here.
Last
week,
I
gave
a
talk
at
minnesota's
user
group,
which
is
kind
of
a
nice
cross
section
between
academics,
dev
ops
ii,
people
and
developers,
and
talking
about
how
to
use
stupider
Holub
in
their
user
group
meetings
and
so
I'd
link
to
the
presentation
there
and
this
week
is
going
to
be
pretty
much
working
with
the
interns
and
some
jupiter
has
documentation.
F
Let's
see
most
noteworthy
is
that
on
the
cloud
of
widgets
we
started
some
experimentation
on
building
queries
for
data
frames.
Declaratively
I
put
a
link
there,
so
you
guys
can
go
and
check
out
us
screenshots
that
we've
taken
screen
little
animated
gifts.
What
I
think
is
interesting
is
that
without
any
code
Colonel
side
code,
we
are
actually
able
to
to
do
template
ice,
a
query
to
a
data
frame
and
then
build
user
interface.
On
top
of
that
that
template
ice
query
and
that
works
on
on
Python
pandas
spark
data
frames.
F
Scala
data
frames
are
data
frames
in
our
spark
data
frames,
so
one
declarative
notation
for
all
which
is
kind
of
cool,
so
just
go
to
that
PR
and
check
it
out.
If
you
have
guys
have
comments,
let
me
know
because
it's
still
work
in
progress,
but
I
think
it's
pretty
cool
and
then
the
other
thing
is
there's
this
book
on
ipad
widgets
that
it's
pretty
pretty
bad
for
declarative
with
it'sit's
number,
5,
54
and
I
think
I
mentioned
it
last
time.
I
talked
I
would
like
to
help
fix
it,
but
I
don't.
F
A
A
Who's
next
to
me,
enemy
is
not
not
here
because
its
first
of
Judah,
you
know
in
Norway
and
is
asking
if
we
should
publish
Jupiter
ipython
obstacles.
Next
we're
going
to
write
at
some
point
on
the
journal
of
open
source
software,
who
has
announced
not
not
long
ago
and
I,
saw
that
too
and
I
was
thinking
the
same
thing.
So
for
me
its
a
surplus
one
and
I
don't
know
if
some
of
you
also
have
an
event
years
and
and
next
it's
darian
are
you
there.
D
G
So,
for
the
past,
a
month
or
so,
I've
been
working
on
porting,
the
Qt
Council
functionality
to
Jupiter,
lab
and
I.
At
this
point,
there's
more
usability,
so
I
would
encourage
anyone
who
actually
has
any
interest
in
it
to
check
out
Jupiter,
j/s
notebook
and
build
the
example
and
play
with
it.
The
feature
most
recently
that
that
is
is
currently
merged
is
the
tool
tip
that
you
get
and
if
you
compare
when,
when
you're
typing
as
you're
typing,
you
get
a
tooltip.
G
If
you
compare
the
behavior
of
the
Jupiter
lab
version
to
the
Qt
console
they're
very
different,
the
console
explicitly
has
logic
that
says:
if
the
character
that
you
just
typed
is
the
open,
/
ends
only
show
tool,
tips
for
that
and
the
Jupiter
lab
version
doesn't
do
that.
The
Jupiter
lab
just
displays
anything.
The
colonel
gives
you,
if
there's
a
tooltip
to
show,
and
the
reason
for
that
is
because,
as
I
was
writing,
it
occurred
to
me
that
we
support
a
lot
of
different
languages
and
different
languages
have
radically
different
grammars
from
each
other.
G
An
open,
paren
Xand
scheme
is
very
different
than
the
nope
and
friends
in
Python
and
every
single
character
in
a
terse
array.
Programming
language,
like
Jay,
really
does
matter.
You
might
want
to
see
tooltips
and
so,
as
you're
playing.
You
might
find
that
there
are
too
many
tooltips
and
it's
annoying
because
you're
writing
Python,
but
you
might
consider
how
we
should
deal
with
this
may
be.
A
colonel
should
send
back
some
sort
of
guidance
like
a
regular
expression
that
the
client
uses.
G
Maybe
the
colonel,
can
send
back
some
sort
of
metadata
about
how
important
the
information
is.
So
the
user
can
pick
what
sort
of
tool
tips
I
want
to
see,
but
basically
just
play
with
it
and
see
if
it
if
it
inspires
you
to
come
up
with
some
sort
of
solution
or
if
you
don't
think
it's
a
problem,
it
might
not
be
a
problem
for
most
users.
G
I
kind
of
personally
want
to
see
a
little
bit
less,
even
though
it's
not
in
my
way,
I
still
think
I
would
want
to
see
less,
but
other
people
might
have
a
different
impulse.
Anyway.
That's
what
I've
been
working
on
and
I
encourage
you
guys
to
check
it
out.
It's
in
the
Jupiter
j/s
know
people
and
when
you
build
examples,
there
are
two
examples:
there's
a
notebook
example
and
a
console
example
you
just
you
can
navigate
to
the
page
when
you
run
the
main
dot
pie
file,
so
I
wouldn't.
A
Worry
about
replicating
exactly
a
view
of
the
console,
because,
if
you
see
console
is
quite
old,
so
these
are
changing.
Behavior
yeah!
I
think
that
the
console
is
roughly
a
notebook
with
less
feature
or
you
cannot
edit
produced
cell
when
you
just
append
cells
and
have
less.
Why
and-
and
this
is
a
something
we
have
already
like
actor
sometimes
ago-
and
it's
something
which
is
not
only
for
the
Potala,
which
is
across
all
all
front-end,
like
even
Adam,
had
because
of
work.
A
D
So
Steve
and
I
have
been
working
a
ton
on
the
document
architecture,
favorite
movie
notebook,
so
no
blockers,
no
questions
leave
with
a
group
I
think
at
this
point
town,
but
by
this
time
next
week
probably
will
have
something
for
the
group
to
play
with
them
and
put
it
back
up.
Well,
let's
see
address.
D
B
Yeah
so
so
I'm,
just
keeping
on
keeping
on
with
the
phosphor
repo
verge,
tidying
up
commands
right
now.
At
the
moment,
one
question
I
did
want
to
bring
up
to
the
group
is
we're
doing
this,
the
big
squeeze
or
whatever
we're
calling
it
to
reduce
the
number
of
repose
we
have
in
phosphor,
because
we
were
at
like
25,
and
that
was
way
too
many
for
us
to
maintain
just
with
with
the
CI
systems
and
the
testing
in
the
documentation.
B
Given
that
we're
not
having
like
some
top-level
index
file
that
re-exports
everything
and
we're
already
doing
path
based
import,
so
we're
actually
like
importing
from
like
phosphor
core
/
lib
/
patterns,
whatever
so
that
web
pack
doesn't,
you
know,
bundle
up
everything
in
the
entire
phosphor,
repo
and
you're.
Only
actually
still
getting
what
you're
using
in
the
web
pack
bundle.
Is
it
even
necessary
for
us
to
have
three
repose
at
this
point,
even
with
three
we're
still
running
into
some
version.
B
Syncing
issues
like
if
I
decide,
I
want
to
add
a
feature
in
cost
for
you,
I,
which
requires
some
addition
to
phosphor
core
I
have
to
first
go
update,
philosopher
core.
You
know
it
haven't
fully
tested
release
it
pump
out
a
version
before
I
can
ever
pick
it
up
in
phosphor.
You
I
to
even
test
it
to
see
if
it
works
out
and
does
what
I
want
to
do
in
the
experimental.
B
You
know,
working
through
the
experimental
new
feature
in
the
UI
repository,
so
I'm
wondering
it
would
definitely
be
easier
from
still
from
a
development
perspective
to
just
have
one
repo
that
has
everything
we
need
in
it.
That's
just
basically
well
factored
in
terms
of
folder
structure,
and
then
you
can
just
import
from
path
whatever
you
need.
Let
pack
it
up
and
you're
going
to
get.
Lady
is
so
I
kind
of
wanted
to
get
what
people's
opinions
were
on
that
or
how
they
feel
about.
B
D
Think
my
biggest
concern
is
we're
trying
we
want
to
move
our
stuff
to
use
the
new
phosphorus
stuff,
so
another
major
delay
in
getting
one
point
out
is
is
impacting
our
work.
How
much
time
do
you
think
it
is
to
consolidate
the
3.3
route,
goes
into
one
repo,
probably
40,
minutes?
Okay,
then
that's
not
a
big
deal.
If
it's
going
to
take
couple
days
and
you're
reworking
things
and
everything
then
maybe
maybe
I'll,
let
me
sing,
don't
you
get
using
your
three
repose
for
now
and
then
and
then
consolidating
the
future?
D
It
make
more
sense,
but
if
it's
just
like
an
hour
project
or
something,
then
then
I
think
it
makes
sense
that
have
one
repo.
The
I
guess
would
also
make
sense
to
have
phosphide
as
a
separate
repo
since
that
sort
of
built
on
top
that
I
mean
I,
don't
know
you
could
draw
the
line,
a
lot
of
places,
I
guess.
B
H
Jason
linked
to
the
issue,
the
PR
on
puter
de
COI,
where
we
implemented
the
base
the
base
level
part
of
that
right
now,
I'm
working
on
cooperating
those
changes
in
the
notebook
cleaning
up
the
notebook
models
and
views
arm
and
we'll
just
keep
marching
forward
on
that.
And
then
that
should
get
us
to
a
point
where
we
have
a
pretty
stable
architecture
for
Jupiter
lab
and
people
can
start
doing
extensions
on
it.
I
Nothing
in
particular,
just
probably
check
out
20
sky
interesting.
Another
react
based
notebook
JavaScript,
only
also
working
on
some
of
the
increasingly
increasing
calls
for
help
on
documentation
and
tools
for
doing
extension,
testing
and
all
that
stuff,
so
yeah,
nothing
to
show
you
out
there,
but
I
got
some.
That's
a
good
stuff
coming.