►
From YouTube: Jupyter/IPython Dev Meeting, September 20, 2016
Description
Meeting of the Jupyter/IPython development team, September 20, 2016
Meeting Notes: https://jupyter.hackpad.com/September-2016-ncdE6E6XQmc
A
Hello,
everyone
today
is
Tuesday
September,
20th,
Jupiter
or
i5
honey,
and
we
have
a
big
crowd
of
air
in
New
York
part
of
the
team,
that's
holding
the
Jupiter
lab
Daphne
and
before
we
get
going,
we
want
to
make
kind
of
the
meeting
version
of
the
the
intro
that
that's
Jen
you
sent
to
the
dead
list.
We
have
Ian
rose
who's
joining
the
team.
Just
just
got
started
this
week
with
Austin
and
got
started
in
New
York
as
a
postdoc
at
UC,
Berkeley
and
poly,
when
I
lose
an
old
face
from
the
project
and
who's.
A
B
A
One
has
anything
that
they
want
to
bring
up.
So
at
the
very
least,
it
would
be
good
to
hear
from
the
folks
on
in
New
York
on
what
you
are
all
up
to
how
things
are
going
etc.
I,
can
you
hear
me?
Yes,
we
can.
The
audio
is
pretty
pretty
good.
Ok,
I've,
not
type
stuff
into
the
back
pad.
We
got
started
yesterday
so
we're
having
a
do.
A
Pa'dar
lab
sprint
here
in
New,
York,
City
and
there's
a
dish,
maybe
nine
well
again,
eight
people
today,
dan
allen
was
here
in
person
yesterday
he
at
his
normal
job,
I
think
today,
but
then
sylvan
joined
and
we
spent
yesterday
doing
sort
of
getting
organized
for
the
week.
And
then
we
did
a
bunch
of
issue
triage
and
we've
created
some
new
labels
to
help
us
do
that,
and
it
wasn't
just
sort
of
playing
tree
in
terms
of
dealing
with
issues
we
have
already
open,
but
also
looking
at
the
different
components
of
jitter
lab
and
identifying.
A
What
issues
are
missing
from
the
that?
Currently,
that
would
get
us
all
the
way
to
10
and
so
we're
basically
trying
to
get
a
full
catalogue
of
issues
required
for
us
to
close
for
each
component
of
Jupiter
lab
4,
10
and
0.9
along
the
way
and
for
the
notebook
component
that
took
us
half
the
day.
Maybe
it
was
a
lot.
A
significant
fraction
of
the
issue
is
open
right
now
or
on
the
notebook
we're
going
to
continue
to
pick
away
at
that.
A
A
lot
of
the
other
components
are
a
lot
simpler
and
then
today
we
started
by
having
Steven
run
through
and
do
a
I
wouldn't
say
super
detailed,
but
a
technical
introduction
to
the
new
lab
extension
mechanisms
that
he's
developed
over
the
last
many
months.
At
this
point-
and
so
we
just
finished
that
and
we're
going
to
try
this
coming
week,
we're
getting
a
talk
at
strada
and
our
goal
is
to
demo
a
couple
different
lab
extensions
to
show
sort
of
third-party
code
working
with
it.
A
The
one
part
of
that
that's
not
done
is
right
now
we're
using
the
the
same
logic
as
envie
extensions
in
terms
of
config
files,
installation
and
activation,
and
we
know
that
we
want
to
migrate
that
to
a
comedy
style
approach
that
won't
require
package
or
package
systems
to
edit
files,
but
rather
than
just
drop,
be
able
to
drop
files
in
comedy
style
directory
to
activate
an
extension
so
packing
us
basically
is.
That
was
not
the
only
point.
A
Last
word
package
installation
would
require
codexis
sort
of
arbitrary
code
execution
so
that
they
become
and
package
managers
kind
of
want
want
to
not
have
to
execute
code
and
only
put
files
on
the
path.
Does
that
that's
a
lot
of
piece
of
that
yeah?
That's
that
as
I
understand
it.
That
is
the
the
last
part.
A
It's
long
been
on
my
list,
but
was
prompted
by
a
comment
by
Hadley
Wickham
last
week
on
Twitter
about
how
nasty
our
table
styling
was
and
I
encourage
people
to
look
at
that.
There's.
Actually
a
Twitter
poll
out
right
now:
I'm,
not
sure
if
that's
a
good
idea
or
a
bad
idea
to
ask
the
internet
what
they
think
about
design.
But
it's
a
it's
a
very
constrained
question,
and
so,
if
you
want
to
have
a
look
at
that
issue
on
on
github,
we're
planning
I
saw
that
one
and
it
looks
great.
A
We
keep
sort
of
visual
consistency
and
second,
because
if
this
is
something
that
is
grading
on
people,
there's
still
going
to
be
a
period
of
usage
of
the
classic
notebook,
and
so
we
probably
want
that
one
back
board
it
and
it
seems
a
small
and
simple
enough
once
once
the
decisions
are
made
once
kind
of
a
dub
settle.
The
actual
amount
of
code
is
not
that
much
yeah
I
was
thinking
the
same
thing.
A
I
think
that's
a
really
good
idea
and
even
pushing
that
out
to
like
and
be
viewer
yeah
would
be
well
NBD
viewer
and
github
styling
as
well
one
just
one
other
comment,
I
mean
I,
know,
I,
know,
I,
know:
we've
moved
to
this
new
system
at
the
dev
meetings
of
using
a
plus.
They
give
people
an
option
to
say
things
rather
than
sort
of
strongly
encouraging
them
to
and
I
I
personally
think
the
default
should
be
maximal
disclosure
as
possible
within
reason.
A
A
Different
kind
of
expression
points
come
out
so
I,
that's
my
own
ten
cents
on
the
how
we
run
this
meeting,
yeah
I'm
open
to
that
I.
Actually
wasn't
wasn't
present
when
or
when
the
idea
for
the
pluses
came
up
and
I
think
we
can
I
know
what
what
do
play
a
pool
of
people
in
the
room.
What
do
people
think
was
everybody
kind
of
annoyed
with
with
everybody
having
to
talk
or
was
it
was
the
issue
that
people
felt
that
that
they
were
just.
B
Reading
from
their
notes,
I
think
it
was
more
an
indication.
C
D
A
D
D
A
A
Don't
provide
kind
of
a
short
summary
and,
as
always,
you
don't
have
to
reread
everything.
So
it
really
is.
It
really
is
kind
of
just
let
us
know
what
you've
been
up
to
I
do.
Think
I
think
I
think
that
Brian
does
have
a
valid
point
in
that
leaving
leaving
things
just
in
text
from
the
notes.
I
probably
leads
the
West
engagement
rather
than
hey.
Oh,
this
is.
F
A
G
Have
an
outstanding
PR
in
Jupiter
lab
for
the
completer
widget
and
because
this
is
something
that
everyone
who
creates
know
folks
are
using.
The
council
will
end
up
inevitably
interacting
with
and
because
it's
something
that
people
have
developed
a
muscle
memory
for
and
have
ideas
about,
I'd
encourage
anyone
who
is
curious
to
play
with
the
PR
and
see
what's
there
because
some
of
the
behaviors
we
know
what
we
want
and
that's
what's
currently
in
there.
Some
things
were
just
bug
fixes.
Of
course
you
know
if
you
notice
anything
broken.
G
If
you
don't
want
to
actually
build
the
pr
yourself,
you
could
at
least
check
out
the
issue
and
see
what
you
think
I've
link
to
it
in
the
hack
pad,
because
right
now,
it's
mostly
been
me,
Brian
Matthias,
talking
about
it
and
then
I
talked
about
it
a
while
ago
before
none
though,
but
nobody
else
has
really
expressed
strong
opinions
about
it,
but
I
think
it's
the
kind
of
thing
that
will
annoy
you.
G
A
And
obviously,
you're
right
that
complete
our
complete
our
behavior
is
one
of
those
things
that
is,
that
is
very
persnickety
in
that
everyone
has
a
lot
of
muscle
memory
on,
but
so
it's
something
that
requires
very
bit
of
discussion.
So
this
is
the
kind
of
thing.
That's
probably
you
folks
well,
while
you're
there
this
week,
you'll
have
a
good
chance
of
brainstorming
face-to-face.
Quite
a
bit
on
I
would
imagine
yeah.
A
All
right
anything
else.
From
from
your
side,
your
collective
site.
H
A
E
E
But
basically,
a
component
is
a
location
that
you
can
point
to
a
file
or
directory
or
where
something
is,
and
then
the
cat
is
a
category
for
the
kind
of
work
that
it
is
now
so
that
make
across
multiple
things,
whether
that's
documentation
and
in
particular,
11
new
category.
That
we've
added
that
I
wanted
to
highlight
is
accessibility.
E
A
If,
if,
if
she
or
anyone
else,
has
expertise
on
this,
I
realized
that
good
accessibility
is
probably
a
long
road
and
there's
going
to
be
a
lot
of
work
to
be
done
on
it,
but
once
one
thing
that
I
imagine
can
be
done
now
is
even
if
even
if
not
everything
is
needed,
at
least
as
Jupiter
live
evolves,
making
sure
that
we
don't.
They
paint
ourselves
into
a
corner
inadvertently,
especially
right,
so
that
certain
design
decisions
that
certain
tooling
decisions
are
made
to
make
it
possible
to
do
that
work
later
on.
A
So
even
if
she
doesn't
have
the
bandwidth
to
do
everything
I'm
having
her
or
anyone
else
with
good
expertise
on
this
point
out,
what
we
should
be
doing
to
make
that
job
easier
in
the
future
would
be
super
super
useful
on,
because
this
is
certainly
an
area
where
we
want
to.
We
want
to
improve
and
delighted
to
hear
to
hear
that.
G
H
A
I
Yeah
I
mean
PI
data.
Carolina's
was
really
good.
There
was
a
lot
of
hpc
interest
as
well
as
interest
in
jupiter
lab,
and
we
didn't
impromptu,
Jupiter
hub
presentation,
similar
to
what
we
did
it
at
Jupiter
days
Atlanta,
and
at
the
end
of
that
we
had.
How
do
you
spin
up
Jupiter
lab
as
well,
so
folks
had
that
as
a
going-away
present,
men
and
I
are
going
to
work
on
0.7,
Jupiter,
hub
and
upgrading
things
and
testing
upgrading
databases.
Things
like
that
this
week
so
and
he
should
be
back
online
tomorrow.
Okay,.
A
Excellent
and
actually
quick
question,
do
you
remember
and
I?
Don't
know
if
you
can
add
to
the
to
the
Hat
fight,
or
maybe
you
should
me
an
email
about
the
the
HPC
crowd
that
you
interacted
with
says
at
the
lab.
That's
kind
of
that's
kind
of
one
of
one
of
my
hats
at
lbl
is
kind
of
interacting
with
the
HPC
people
would
be
good
to
know
what
context
you
mean
in
that
kind
of
that
direction
of
people,
people
who
were
there
from
the
HTC
yeah.
I
Sure,
I,
probably
what
I'll
do
is
I'll,
send
you
an
email
later
with
some
of
the
links
and
stuff
great.
A
Great
awesome
and
and
if
you
and
men
obviously
are
interested
in
doing
I
mean
that's
that
meaning
it
works.
We
spotted
informally
because
somebody
was
stopping
by
and
he
said
somebody
from
Harvard
or
something,
but
instead
I've
been
working
on
Jupiter,
hobbin
and
Wes
to
talk
to
you
guys,
but
it's
kind
of
bump
into
other
people
from
from
Berkeley
and
an
lbl
participating
as
well.
A
So
if
you
and
or
mean
I
mean
obvious
or
anyone
elses
but
specifically,
but
the
two
of
you
guys
want
to
want
to
join
in
as
tomorrow,
and
there
will
be
a
blue
jeans
link,
I
actually
had
a
conflict.
I
want
I'm
not
going
to
make
it
but
Jamie.
It
might
actually
be
a
good
idea
to
record
that
meeting.
That's
might
prove
useful.
If
people
above
mine
you
can
check,
and
if
people
don't
mind
just
recording
that
meeting
is,
it
might
actually
be
a
useful
discussion.
A
You
got
it
all
right,
so
Mike
go
Carol
is
that
that
seems
like
you
were
wrapping
up
cool
thanks.
Mike.
Are
there
things
that
you'd
like
to
bring
out
for
the
group?
Quite
a
few.
D
Hold
my
finger
on
space,
so
I
can
look
at
that
yeah,
okay,
so
the
first
thing
I
wanted
to
talk
about
is
there's
sort
of
been
some
back
and
forth
about
how
to
handle
a
custom
script
exporter
and
whether
we
want
to
define
that
globally
for
a
notebook
or
globally
or
on
a
per
notebook
basis.
So
this
came
up
in
the
context
of
an
are
exporter
that
was
supposed
to
be
nicer.
D
J
Do
that
we
have
a
script
exporter
when
she
paints
to
converge
on
a
plane
y,
la
scala
by
whatever,
whatever
language
plane,
powered
vacuum
interpreter
that
passcode,
very
basic
because
of
things
like
human
languages
have
different
duties.
To
my
comments
with
the
script,
exports
doesn't
throw
out,
raise
the
pictures
right,
cons.
J
H
Monument,
your
audience
I'm
extremely
robotic
I,
don't
know
that
info
comics
all
actually
get
the
content
that
the
comments
prepared
or
justice
on
orange.
C
C
The
other
possibility
would
be
when
we
install
a
specific
and
this
specific
exporter
using
a
python
package
for
the
specific
language
to
register
a
map
from
this
language.
When
you
have,
when
you
have
that
book
with
this
language,
use
this
exporter,
and
so
there
are
challenges,
drawbacks
and
advantages
to
both
approaches,
and
we
would
like
people
to
come
by
look
at
the
issue.
Look
at
the
PR
and
give
their
salt
on
that.
H
Perfect,
so
you
guys
have
yeah
if
I
have
provided
that
the
numbers
in
the
hot
pads
it
may
be
useful.
There
are
specific
people.
You
want
to
have
take
a
look
at
these.
It's.
C
Mostly,
why
design
discussion?
So
anybody
who
is
concerned
about
maintenance
since
five
years
is
welcome
to
come
on
the
issue,
because
it
thinks
it
seems
that
if
we
change
will
stick
around
affect
a
lot
of
moving
pieces
will
become
de
facto
stood
out
if
we
implement
them.
One
of
these
are
more
indication.
D
H
H
A
And
Mike
Mike,
you
might
want
to
just
sing
the
list
with
things
like
this.
That
basically
require
where
you
want
to
flat,
for
others
to
engage,
just
drop,
a
quick
email
summarizing
the
highlight
that,
just
as
the
two
or
three
high-level
points,
the
mailing
list
and
encouraging
people
to
then
pitch
into
the
issue
so
that
at
least
at
least
people
are
aware-
and
the
actual
technical
discussion
will
probably
happen
on
a
combination
of
the
mailing
list
and
the
specific
issues.
A
But
for
things
like
this,
it's
a
good
idea
just
to
flag
the
list
and
say
hey.
If
anybody
cares
about
this
topic,
important
is
happening
over
here.
Please
come
on
over
and
and
then
a
few
of
us
will
have
a
look
because
here
into.
A
D
That
makes
sense,
I
just
figured
that,
especially
since
we're
focusing,
especially
in
labs,
on
lots
of
external,
like
lots
of
extensibility
and
stuff
like
that,
and
that
this
was
focusing
about
that
people
might
have
a
specific
thought
as
to
how
to
do
it.
Just
talk
about
their
head,
but
I'll
put
it
on
the
list.
Some
of
the
other
ones
might
be
a
bit
more
technical,
too
I've
got
to
go
through.
C
D
Also
have
a
function
under
NV
convert,
called
export
and
a
module
called
export,
and
so,
depending
on
how
exactly
you
import
things
you
could
get
one
or
the
other,
which
then
we'll
make
it
extremely
confusing
as
to
what
exactly
you're
doing,
and
so
one
of
the
things
that
were
going
to
decided,
you
is
to
deprecate
the
export
module
in
favor
of
the
functions
and
have
another
module
that
holds
that
function
and
also
it
definitely
doesn't
help
that
the
export
function
right
now
is
defined
within
the
export
module.
So
expert,
export
and
export
are
ya.
D
C
The
else
we've
also
we
design
or
what
also
an
issue
we
would
like
people
to
have
a
look
at
and
give
their
salt
on
it
because
it
might,
it
might
break
API
a
little
bit,
but
we
like
to
smooth
out
how
it
goes
for
users
and
it's
quite
complex.
You
have
to
have
the
same
name
and
we
built
a
long
time
to
convict
and.
D
Idea
is
actually
to
break
the
API
say:
come
Sixpoint,
oh
we're
just
deprecating
it
for
now.
If
goal
because
having
this
around,
even
if
we
deprecate
it,
it's
still
going
to
be
an
issue
like
the
because
we're
shipping
it
it's
still
exports.
The
module
is
still
in
there,
we're
just
going
to
be
discouraging
people
from
using
it.
D
F
D
So
that
their
sequel
database
can
pull
in
requires
Python,
so
that
we
can
actually
only
upgrade
people
to
python
3
only
upgrade
people
to
the
is
it
ipod
on
610
or
notebook
six-point.
Oh
that's
types
of
three.
Only
I
always
forget
I
advise
on
okay.
So
then
I
Python
six-point.
Oh!
When
that
comes
out,
it
will
actually
be
possible
to
stop
people
from
upgrading
if
they
only
work
on
iphone
to
its.
D
One
of
the
things
is,
I
discovered
that
we
have
never
handled
chinese,
japanese
or
korean
exported
to
PDF,
even
though
we
represented
in
the
browser
it's
partially
because
cjk
languages
are
sort
of
a
whole
bag
of
a
can
of
worms
or
a
bag
of
worms
for
that
matter,
in
unicode
and
getting
support
for
them
at
all,
let
alone
in
litec
is
a
bit
of
a
mess.
I
have
an
open
PR
that
works
locally.
It
fails
on
Travis.
If
anyone
has
any
idea.
D
But
if
anyone
wants
to
look
at
that,
oh
the
reason
I
bring
it
up
is
that
there
are
a
couple
of
different
ways
of
trying
to
actually
implement
that,
some
of
which
involved
building
out
libraries
like
an
external
library.
One
of
them
involves
building
out
a
bunch
of
templates.
Another
one
is
just
a
preprocessor
that
sort
of
handles
foreign
languages,
and
one
of
the
things
I
wanted
to
know
is
just
stupid
or
lower.
Has
anyone
really
tried
to
deal
with
any
I
know?
D
We've
only
gone
on
to
Zoila
tech
for
a
little
bit
now,
but
has
anyone
dealt
with
foreign
language
support
because
I've
seen
stuff
about
stopping
foreign
language
support
in
certain
cases,
and
if
that
is
the
case
that
are
general,
like
biases
toward
reducing
a
foreign
language
like
characters,
support
should
I
not
be
pursuing
her.
No.
A
I,
not
quite
the
opposite,
I
mean
we
would
like
to
have
a
good
foreign
language
support
in
as
much
as
we
have
the
expertise
to
do
it.
Right,
though,
and
so
I
think
one
of
the
fear
is
simply
that
each
of
those
topics
tends
to
be
huge
and
very
complex,
and
unless
you
really
really
have
a
lot
of
expertise,
I'm
certainly
not
going
to
be
the
one
making
any
decision
on
CJ
state
support,
because
I
don't
have
the
foggiest
idea.
A
What
the
right
decision
is
on
CJ
k,
the
languages
that
I
that
I
speak
for
all
Western
languages,
and
so
once
you
have
enough
unicode
support
to
get
accents.
That's
it
I
mean
the
difference
between
supporting
Spanish
in
English
on
an
export
is
trivial.
Once
you
have
proper
unicode
support,
because
from
from
an
encoding
perspective,
the
only
differences
are
just
worth
capturing
things
like
accents
correctly
right,
and
so
it's
really
languages
that
exist
in
other
scripts
like
Cyrillic
and
Hebrew,
and
Arabic
and
CJ
k
and
things
of
that
nature.
A
And
for
that
we
really
mean
unless
we
have
someone
very
similar
to
the
topic
of
accessibility.
Unless
we
have
the
person
with
the
right
expertise,
we
can't
really
dig
into
it.
So
I
would,
I
would
say,
do
no
harm,
but
don't
don't
get
sucked
into
that
into
that
rabbit
hole.
Unless
we
have
someone
who
actually
knows
a
knows
what
what
needs
to
be
done,
and
so
otherwise
you
just
try
not
to
break
things,
but
but
but
we
can't
really
be
building
that
right
now:
okay,
it
it's
really
a
matter
of
prudence.
D
A
Okay,
Paul
I'm
skipping
the
minuses,
Paul
already
yeah.
You
think
that
you
want
to
bring
up.
In
addition,
nope.