►
From YouTube: Jupyter/IPython Dev Meeting, March 7, 2017
Description
Jupyter/IPython development team, March 7, 2017
Meeting Notes: https://paper.dropbox.com/doc/March-2017-Jupyter-Weekly-Meetings-rLiXjbl77MPLNNpNB1hdu
A
Hi
the
morning
internet
today
is
Tuesday
marks
7th.
This
is
the
Jupiter
weekly
def
meeting,
we're
over
in
the
march
issue
of
our
meeting
notes
in
dropbox
paper
the
link,
if
you're,
watching
this
on
youtube.
The
link
is
in
the
notes
section
there
in
YouTube.
So
let's
see
last
week,
so
I
put
a
link
out
to
a
februari
notes
if
folks
want
to
go
back
and
look
at
our
archives
and
the
video
should
be
in
our
notes
also
for
this
week.
A
A
A
B
B
A
B
C
So
I
was
actually
when
I
was
at
I
am
last
week
his
or
the
plan
for
a
session
organised
by
Randy,
and
his
co-author
gave
catches
him
on
that
book.
Was
there
and
we
chatted
a
little
bit
about
things,
not
not
something.
We
didn't
make
a
lot
of
progress
but
I'm
aware
of
this
project.
If
she
does
that
by
this
week,
runs.
A
Great
cool
all
right
and
finally,
from
last
week,
so
big
announcement
for
today
the
Jupiter
con
CFP.
It
has
another
week,
we've
scented
the
deadline
by
one
Weesa
March
14,
so
that
gives
us
another
week
to
reach
out
to
our
networks
and
folks
for
watching
this
reach
out
to
your
networks
and
encourage
them
to
submit
a
talk
to
Jupiter
con.
That
also
includes
tutorials,
so
folks
want
to
give
a
tutorial
on
a
Jupiter,
related
projects
or
approach.
A
A
Okay,
so
recap
of
some
of
the
things
I
worked
on
last
week
and
when
I'm
working
on
this
week
already
mentioned,
the
CSP
deadline,
that's
been
extended,
Fernando
and
the
team
and
I
submitted
the
foundation
report.
Last
Wednesday.
We
did
a
lot
of
really
good
work,
get
updating
our
report
and
sending
it
off.
So
thanks
to
the
Berkeley
team.
For
all
your
help
so
sunday
we
met
with
the
binder
crew
and
I,
looked
at
ways
to
continue
that
forward.
A
So
I'll
bring
back
any
lessons
learned
from
that
and
then
this
week
of
course,
I'll
be
working
with
men
and
Carol
on
on
different
things,
including
the
release
schedule
in
a
release,
history
and
possibly
other
other
metrics.
If
there
is
time
all
right,
so
I
Python
Matthias,
are
you
ready
to
get
your
update
or
do
you
want
me
to
skip
if
I
see
y'all
I'm
taking.
D
D
It's
not
a
good
idea
that
we
could
just
be
a
completely
different
package,
the
first
level
2i
by
all
that
there
is
a
issue
being
in
the
release
out
if
you
have
to
use-
or
if
you
don't
have
to
use,
please
go,
reduce
you
and
weight
in
on
whether
it
can
I
things
on
up
and
keep
the
proposed
solution
would
be
enough
for
you
otherwise
this
week,
we're
about
to
stop
debating
into
the
background,
which
is
here
in
Berkeley,
also
on
everywhere.
Online
ten
times
were
focusing
on
my
kingdom
and
licking
the
competition
better.
D
D
Looking
at
making
the
interactive
experience
to
meet
I
think
an
elite
with
Python
3
kicks
and
about
better.
In
servers
not
have
to
manually
make
that
out
of
christine's
and
passionately,
but
it's
just
a
complex
if
you're
interested
in
have
ideas
about
that
just
says:
new
ideas,
another
buddy
policy.
B
They
actually
had
a
couple
of
questions
about
the
notebook
stuff,
so
tots
are
so
Paul
and
I
last
week,
sat
down
and
started
looking
through
some
of
the
testing
procedures
that
we
have
for
the
notebook
and
all
of
them
right
now
are
essentially
any
like
they're,
using
a
browser
emulator
to
create
a
DOM
and
like
investigate
that
directly.
Is
there
a
reason
why
we've
taken
that
approach
as
a
but
I
know?
Historically,
that
was
easier
to
do
at
the
time,
because
testing
wiccan
javascript
per
se
wasn't
always
going
to
be
as
reliable,
given
JavaScript.
B
I,
don't
think
the
idea
which
what
it
doesn't
do
suffering
something
on
the
page,
but
more
that
so
this
is.
This
came
up
in
the
context
of
testing,
latech
and
mathjax
rendering,
and
if
you
try
to
actually
test
what
mathjax
is
rendering
on
the
page.
First
of
all,
if
your
test
is
too
closely
linked
to
what
they're
laying
out
then
every
time
they
changed
something,
then
it
now
failed,
but
also
it's
a
really
comfortable
like
for
a
sufficiently
complicated
input
like
it
is
an
incredibly
complicated
output,
and
so
there
are
simplifications
that
we
can
make.
B
But
it
would
be
helpful
to
know
like
for
just
some
of
the
string
parsing
stuff,
which
was
there
was
the
cause
of
some
of
the
bugs
that
we've
been
having
with
text
parsing
for
a
long
time.
Now.
Is
that
fine
to
do
just
in
the
pure
JavaScript?
It
can
still
be
done
inside
catcher,
but
just
without
actually
rendering
a
DOM
and
querying
the
Dom.
E
That
is
they
just
as
a
data
point.
We
use
karma
with
firefox
to
test
on
Travis
and
headless
mode,
and
we
touch
the
rendered
output
for
lots
of
things,
and
it
works
quite
well
about
not
that's
not
saying
that
testing
the
rendered
la
tech
output
in
EG-
that's
extremely
hard,
but
it's
the
concern
is
about
running
headless
and
not
having
a
dom
that
actually
renders
in
terms
of
geometries.
You
can
solve
that
with
karma
and
firefox
on
Travis
yeah.
C
You
definitely
can't
just
you
can
test
directly.
There
are
some
tests
already
that
just
take
a
javascript
function,
run
it
and
then
test
as
outputs
and
don't
actually
interact
with
the
page.
It's
just
it
because
we
already
are
using
test.
Runner
stuff
happens
to
instantiate
a
notebook
page,
that's
where
it
runs,
but
it
doesn't
matter
that
that's
where
I
wrote.
Okay,.
C
Just
happens
to
be
the
machinery
that
we
have
for
getting
a
context
in
which
to
run
so
Java
Script
at
this
point,
I
I
I,
don't
think
anybody's
super
happy
with
the
Casper
stuff,
but
given
the
current,
given
the
status
of
the
notebook,
rebo
and
most
people
thought
that
it's
not
it's
not
worth
like
migrating
all
those
tests
over
to
karma.
Since
that
whole
repo
is
set
to
be
replaced
by
Duke
11,
a
not-too-distant
time
frame
that
reading
test
for
the
existing
framework,
which
she
is
fine.
C
F
B
True,
we
don't
really
have
much
in
or
the
documentation
that
we
had
for
writing.
Test
is
woefully
out
of
date,
so
I'm
not
sure
if
I'll
be
able
to
get
to
that
this
week,
but
but
something
that
I'll.
If
the
other
thing
was
Thomas
had
the
sort
of
opinion
on
a
couple
of
issues.
But
we
just
shouldn't
bother
with
tests
in
the
notebook
or
at
least
with
new
one
and
I
think
Paul
objected
and
I
just
wanted
to
know.
Should
we
continue
to
add
test,
or
should
we
sort
of
keep
that,
like
at
spacious,
yeah.
C
No
I
think
it's
appropriate
to
to
add
I
think
even
given
the
turkey
nature
of
the
test,
wait
for
the
notebook
repo
I,
don't
think
we
want
to
require
new
contributors
to
have
tests
for
their
new
code,
but
I
mean
we
should
write
that
for
people.
We
can
expect
people
to
put
in
the
effort
to
deal
with
the
other
tools
today.
G
G
Don't
understand
why
me
oh
ok,
but
I
was
going
to
say,
there's,
there's,
there's
a
huge
number
of
tests
that
were
running
into
lab
and
seasoned
Darian
have
both
worked
quite
a
bit
on
testing
I,
don't
know
what
the
coverage
is.
That's
where
I
was
a
theater
referred
to
with
the
coverage
of
the
true
key
lab
notebook
component
is
in
testing
I'm.
B
Okay,
what
would
the
idea
then
be
four
new
contributors?
They
submit
a
PR,
we
merge
it,
and
then
one
of
us
and
takes
on
immediately
creates
a
new
issues
lego
this
prh
test
if
it
introduces
new
features
and
then
making
a
PR
and
that
way
so
that
there
is
some
period
of
time
during
which
the
new
features
are
added,
but
none
tested
since
or
do
we
just
require
that
all
contributors
allow
us
to
push
to
their
rethought
like
the
check
the
Edit
like
from
computers,
button
thing,
I.
C
Don't
think
we
need
to
be
as
formal
as
that,
but
hey
when
roughly
yeah.
So
you
know
I
don't
want
to
contribute
something
and
you
you
wanna,
make
sure
their
success
for
us
and
because
a
merge
in
that
a
test
in
a
second
PR,
even
open.
You
can
even
open
that
second
PR
before
the
first
ones,
March
it
and
just
bigger.
I
J
I
just
come
all
the
way
out:
yeah
I
just
remember
local
test
Jason.
If
you're
curious
for
just
a
notebook
component,
we
have
about
seventy
two
percent
coverage
connectify
to
keep
wipe
or
component
yeah
I'm
the
Jupiter
lab
notebook
component.
Now,
that's
seventy
two
percent
of
lines,
but
I
think
all
the
public
API
functions
like
all
the
surface
areas
hit,
but
not
like
every
branch,
probably.
C
D
A
A
L
So
I'm
not
sure
if
Thomas
is
in
the
meeting
right
now,
but
we
are
close
to
doing
another
beta
release.
I
have
one
question
for
men
regarding
the
tests
of
cell
tags:
toolbar
PR,
yes,
I!
Is
that
ready
to
merge.
L
Okay,
look
at
the
severing
to
ask
so
I'll
get
that
merged
and
then
Thomas
started
the
show
server
root
directory
and
within
the
move.
Dialogue
so
I'll
finish
that
up
for
him
today
and
get
both
of
those
merged.
Do
a
new
release,
hopefully
by
end
of
day
and
then
I
think
it's
just
a
matter
of
you
know
we'll
test
it
for
a
couple
days
and
then
I
think
we
can
escalate
to
a
full-on
public
release.
L
G
B
H
L
C
B
B
Otherwise
it
seems
strange
to
rely
on
Travis
and
I
think
it
has
something
to
do
with
the
way
it's
building
good,
the
directory
structure
and
what
it
thinks
are:
it's
available,
fonts
and
how
tech
lives
to
standardly
install
fonts,
because
the
litec
should
be
able
to
find
them
wherever
they
are
in
their
systems.
So
but
I've
gotten
at
least
the
Unicode
coverage
in
verbatim
math
and
and
Roman
text
and
italicize,
and
all
that
the
problem
is
that
combining
characters
are
still
acting
very
strangely.
B
Sometimes
they
are
combining
seemingly
wrong
direction.
It
may
have
been
a
change
in
Unicode
nine-point-oh
that
I
need
to
look
up
because
it
is
producing
this
problem
more
in
some
cases
and
others
so
yeah,
that's
just
something.
I
need
to
do,
but
there's
still
a
complication
around
even
finding
a
font
that
has
all
the
symbols
that
we
are
supposed
to
be
able
to
include
according
to
ipython
there's
a
set
of
symbols
that
we
can
have
as
valid
identifier,
names
and
I
have
yet
to
find
a
font,
even
a
web
font
available.
B
That
covers
all
of
those
characters.
If
you
currently
look
at,
if
you
just
load
that
page
or
load
that
file
into
a
current
notebook,
you
will
see
that
many
of
the
characters
simply
are
not
rendered,
or
rather
they
have
their
whatever
the
particular
formats
hey.
This
is
a
non
non
legitimate
character
shape
instead,
so
working
on.
All
of
that
that's
been
fun,
and
then
there
was
discussion
after
last
week's
meeting
about
generating
JSON
schema
or
trail.
B
It's
so
that
you
could
do
automatic
UI
for
configuration
in
Jupiter
lab
I
in
talking
with
other
people
this
week
and
looking
forward
I'm.
Looking
into
that
I
think
my
first
plan
of
attack,
there
is
going
to
be
just
getting
some
nice
JSON
export
of
the
configuration
and
just
better
documenting
how
that
is
supposed
to
work,
because
that
documentation
doesn't
exist
and
taking
that
as
a
first
step,
because
once
we
have
that
generated,
then
the
possibility
of
like
building
the
schema
that
it
can
check
against.
The
things
that
we
are
to
use
becomes
more
plausible.
M
A
M
Fish,
they
have
a
colonel
gateway.
There's
a
couple
of
issues
open
just
following
through
them,
one
at
a
time
then
helping
to
just
like
a
compatible
window
for
five
to
one
of
those
with
the
board.
I
just
mentioned
that
son,
sex
or
other
is
deprecating
the
activity
api
put
in
the
colonel
dale.
A
while
ago
I
wrote
them
on
a
notebook
that
could
be
very
simple
change.
It
is
a
breaking
a
TI
change,
so
we'll
call
the
next
round
babe
I
released
to
god.
M
Oh
and
the
third
one
I
had
open
for
this
release
would
refuse
a
request
a
while
ago
do
match
internal
gateway,
the
token
team
that
the
book
was
not
even
which
I
think
we
can't
in
spirit.
So
not
you
know
using
the
code,
that's
fair,
because
the
notebooks
could
still
relies
on
one
time
token
and
then
such
a
cookie,
which
doesn't
really
help
programmatic
access,
which
I
believe
is
the
coral
gables
name,
is
worthless.
M
M
D
A
It
really
is:
we
have
a
comments
on
the
mailing
list
asking
about
binder
and
they're,
finding
that
the
incubator
showcase
is
down
on
so
I,
don't
I
think
it's
probably
on
the
binder
in
and
then
failing,
but
I,
don't
know.
If
you
have
time
to
look
at
the
code
to
just
see
if
there's
anything
in
there,
that
would
cause
it
to
fail.
He.
N
F
M
A
So
maybe,
if
you
take
a
look
at
it
and
there's
not
worth
investing
a
lot
of
time
in
now,
maybe
in
the
future
we
can,
we
can
update
it.
Maybe
could
make
a
comment.
You
can
just
comment
on
that
mailing
list.
Note
to
let
those
user
know
that
there's
other
ways
to
view
the
information
yeah
be
cool,
all
right
right.
Thanks
all
right,
Jupiter
lab.
H
Okay,
we
released
zero
point
17
recently
I
cant
rember
the
day
and
there's
a
link
to
the
the
release
notes.
Basically,
it
comes
with
the
new
command
palette.
I
came
to
free
when
we
switched
over
to
phosphor
upgrade
as
well
as
some
changes
to
render
mine
haunted
by
some
some
API
changes
at
Grant
wanted
for
his
advanced
rendering
with
alternator
widgets
and
his
other
crop
of
vendors.
H
Additionally,
we
experimented
with
lerner,
which
is
a
tool
to
lift.
You
have
multiple
packages.
Npm
packages
in
the
camry
co
Interactive's
are
as
using
it,
and
so
we
we
modeled
after
the
same
pattern
and
that
each
Pat
each
sub
package
has
an
independent
version,
but
for
the
developer,
they're
all
symlink
symbolic
link
together.
So
you
have
a
coherent
view
of
all
your
packages
while
you're
developing,
and
then
they
get
released
as
a
unit
to
learn
a
lot
up.
Automates
all
that
for
you,
so
we're
not
the
right.
The
scripts
to
learn
has
already
done
it.
H
So
we
did
the
experiment
and
then
converted
phosphor,
to
use
that
instead
of
using
the
typescript
compiler
to
work
around
the
issue
alone,
it
does
the
sim
linking
for
us
and
then
so
we're
going
to
follow
that
same
pattern,
and
you
put
a
lot
more
progress
right
now.
Just
cleaning
up
where
the
folders
are,
what
the
file
package
names
will
be
on
and
then
I'll
switch
over
and
the
end
result
is.
H
The
individual
packages
will
be
consumable
as
opposed
to
need
to
require
the
entirety
of
super
lab
as
a
single
unit,
so
for
third-party
use
they
can
just
pull
them
apart.
They
need
and
then
for
extension
office
as
well.
That
is
pulling
the
parts
of
Jupiter
lab
busy.
They
want,
for
extension
purposes
and
additionally,
during
battle
the
hard
battle
of
figuring
out
how
to
get
a
hover
box
to
coherently
behave
in
the
face
of
multiple
methods
scrolling
and
where
should
draw
itself
on
the
page.
H
So
this
is
used
by
our
tool
tip
as
well
as
computer
as
widgets
that
floats
above
a
point
of
interest
in
an
intelligent
way
and
goes
away
when
it's
no
longer
relevant.
There's
a
lot
of
education
that
he'd
to
work
out
with
that.
So
I
don't
envy
that
and
then
that's
about
it.
Oh
and
then
chris
is
working
on
single
document
mode
for
phosphorus
that
allows
the
the
dot
panel
to
enter
into
a
a
sort
of
maximizers
gen
view,
so
that
you're
working
on
just
a
single
document
and
making
up
talk
between
the
two
views.
H
N
A
G
Absolutely
the
the
announcement
of
the
year
for
us
is,
it
really
expect
I
would
did
69,
oh
yeah
things
to
be
seeing
seem
to
be
going
fairly
smoothly.
There's
a
comment
here
and
there
main
hiccup
seemed
to
be
a
lack
of
an
output
widget
on
web
contact
so
that
it's
the
interacts,
don't
show
up
correctly
and
I
like
the
sinks,
documentation
itself,
and
so,
when
I
think
is
going
to
work
on
that,
and
also
there
are
some
questions
about
that
pot.
Lid
plotting.
G
Now
they
interact,
you
need
to
do
PL,
pho
explicitly,
but
there's
some
conversations
were
having
ipython
repo
about
making
it
automatic
again,
but
other
than
that
is,
it
seems
like
it's
been
fairly
smooth
we're
moving
ahead.
I'm
moving
the
widgets
over
to
the
new
foster
framework
and
updating
it
to
work
in
the
new
jupiter
lab,
which
is
a
continual
Rackham
bowl
game,
but
with
good
reason
and
jiggle
abbas
off.
But
we
all
know
that
and
anything
else.
You
want
to
add
a
look
yeah.
F
G
Yeah
and
if
you
want
to
see
it
a
cool
examples
of
widgets,
you
should
go
check
out
the
IP
volume
package.
I
py
volume
package.
Their
documentation
has
some
live
widgets
with
some
cool
volumetric,
rendering
that
even
works
in
stereo
and
even
works.
If
you
put
it
on
your
phone
and
you
have
one
of
those
little
cardboard-
viewers
even
works
in
like
3d
virtual
reality
type
mode
really
pretty
cool
has
been
amazing
package.
Real.
F
G
Another
is
we're
hoping
to
submit
a
high
widgets
tutorial
first
night
I
and
probably
for
Jupiter
con
as
well.
It's
basically
think
of
the
widgets
as
an
ecosystem.
Here's
a
lot
of
the
school
stuff
is
going
on
and
here's
how
you
can
write
your
own
widget
and
so
so
we're
getting
up
together
and
hopefully
have
submitted
this
week.
Scientific
all.
A
K
Des
groupes
here
so
the
founder
tutorials
you
can
have
submitted
from
our
team
as
well
as
the
community,
the
better
the
old
workshop
tutorial
experience
will
be
so
get
your
toxins,
but
also
get
some
tutorials
in.
G
G
K
B
K
I've
been
traveling
as
well:
I
did
the
whole
Midwest
College
circuit,
University
circuit
people
love
to
her
lab.
They
are
really
excited
about
using
24
hubs
and
education
and
computing
so
and
the
widgets.
Thank
you
sullen
for
all
that
stuff
that
you
gave
me
before
the
talks,
because
they
went
really
well
and
people
just
love
it.
So
thanks
everybody
yeah.
B
A
C
A
A
E
A
So
what
I'm,
showing
you
right
now
is
this
month
march
going
forward,
and
I
wanted
to
point
out
that
we
have
we're
starting
to
get
into
conference
season
and
you
know
we're
starting
to
add
different
conferences
to
each
of
the
month.
So
if
you
are
attending
the
conference
a
conference,
the
easiest
way
to
add
a
card
to
this
board,
which
is
what
I
would
love
the
team
to
do-
is
to
go
up
here
to
where
this
plus
sign
is
and
click.
It
and
says
enter
a
note.
A
A
He'll
ask
you
a
I'd
it'll:
ask
you
which
repo
you
want
to
add
the
issue
too,
and
you
can
right
now,
we'll
just
add
it
to
project
management,
okay
and
then
in
the
body.
It's
really
useful.
If
you
add
a
link
to
the
conference
website,
but
you
know,
if
you
don't,
that
can
be
done
later,
so
you
click
convert
to
issue
and
that
will
turn
it
into
an
issue
on
the
board.
One
final
step
is
actually
clicking
on
it
to
open
the
issue
and
assigning
yourself.
So
we
know
who
is
actually
sewing
that
event.
A
So
you
know
with
the
other
events
that
we
have
going
on
that
are
coming
up.
Hi
data
Amsterdam
is
on
here
already
on
sunshine:
I've
been
adding
events
as
they
come
up
in
our
meeting,
but
if
you
see
an
event
on
here
that
you're
going
to
that's,
not
a
sign,
go
ahead
and
add
yourself,
and
one
quick
look
at
may
shows
that
may
is
full
of
conferences.
A
A
A
So
one
thing
that
I'll
do
for
folks
is
I
might
schedule
a
viewing
party
because,
like
Jeff
Hamrick
couldn't
come
either,
and
it
would
be
great
to
get
a
few
folks
together
here
on
campus
anyway
and
online
somewhere
able
to
make
it.
We
can
watch
it
together
and
ask
and
answer
questions
and
then
we'll
also
provide
a
link
to
the
YouTube
video
online,
with
the
notes
so
hope
to
do
that.
This
week,
I
am
traveling
a
little
bit
this
week,
so
you
know
next
week
at
the
latest.
A
A
Thank
you
all
right,
right,
cuz
you're,
the
action
items
back
all
right
and
then
right
documentation
as
a
satellite
participant
at
the
dock
Yvonne,
and
let
me
just
say
that
the
dock
athan
has
an
amazing
turnout.
The
the
whole
did
auditorium
or
the
room
is
full
on
and
everybody
is
really
into
it.
So
it
looks
like
it's
a
great
success
this
week
so
way
to
go
Fortius
and
all
the
other
docs
on
folks
to
put
time
into
that
releases
this
week,
nope
notebook
five-point,
oh,
we
know
that's
the
end
of
my
part.
B
Wanted
to
point
out
that
I
did
link
to
a
new
document
that
is
specifically
at
the
end
of
the
conference's
section.
If
anyone
wants
to
suggest
and
I
pre
populated
it
with
a
template
that
people
can
copy
and
paste
or
other
suggested
things
and
included
the
one
that
carol
was
just
suggesting
a
jupiter
lab
like
instruction
with
steve,
potentially
as
the
lead
on
that
and
I've
listened
it
great.