►
From YouTube: Jupyter/IPython Weekly Dev Meeting, April 4, 2017
Description
Meeting of the Jupyter/IPython development team, April 4, 2017
Meeting Notes: https://paper.dropbox.com/doc/April-2017-Jupyter-Weekly-Meetings-kML3Uu32Y4HNAlpJkHRXX
A
Good
morning,
everyone,
today's
April.
Fourth,
this
is
the
Jupiter
weekly
def
meeting.
We
have.
We
have
a
good
crew.
Today
we
have
about
11
folks
here
on
the
call
and
they'll
give
you
their
updates
as
we
go
through
our
meeting
notes
in
dropbox
paper.
The
link
to
the
meeting
notes
are
in
the
notes,
section
and
youtube
if
you'd
like
to
access
them,
while
you're
watching
the
video.
Let's
just
go
to
the
top
of
the
list
where
we
keep
our
action
items.
A
We
had
one
action
item
from
last
week
and
that
was
grant
was
looking
for
folks
to
test
notebook,
five-point,
oh
release,
candidate
to
and
I
think
we
released
I
don't
want
to
steal
the
surprise,
the
announcement,
but
that's
checked
off
in
our
notes.
I,
don't
feel
grants
thunder.
Next
we
have
some
other
action
items
from
earlier
in
March
and
also
in
febuary.
So
folks,
who
are
on
the
call
who
created
those
if
you
could
check
off
the
ones
that
are
done,
looks
like
Mateus
is
still
looking
for
feedback
on
this
async
rebel
implementation.
A
I
know
Fernando
made
a
couple
comments
on
that
and
in
the
conversation,
looks
like
it's
still
going:
Matthias
soap.
That
was
just
there.
We
go
we're
getting
stuff
checked
off
as
I
speak,
so
thanks
so
much
great.
So
it
looks
like
the
only
outstanding
one.
Then
there
is
the
async
rebel
implementation
so
head
on
over
to
our
github
repo
I
pais
on
my
side
paisa.
So
it's
a
python,
org
ipython
repo
and
then
it's
pull
10,000
390
cool
thanks
for
putting
the
cursor
right.
There
awesome
all
right.
A
So
then
we
link
our
weekly
news
by
organization.
If
you
click
on
those
subprojects,
those
words
you
can
find
updates
about,
what's
been
going
on
in
those
those
those
org,
so
I,
Python,
Jupiter,
Jupiter,
lad,
Jupiter,
widgets
and
Jupiter
hub.
That
I'll
give
you
the
number
of
commits
the
number
of
issues
that
have
been
closed:
lots
of
good
metrics
there.
A
So
hopefully,
I'll
have
an
update
in
the
next
couple
of
weeks,
just
looking
for
themes
in
ways
that
we
can
cluster
some
of
our
work,
see
if
anything
bubbles
to
the
top
and
hopefully
I
can
dub
tail
that
with
the
work
that
Brian's
doing
the
work
with
friends
doing
with
the
metrics
on
Jupiter
lab
another
work
that
we're
doing
on
the
project
management
side.
I
also
updated,
are
so
when
we
we
had
our
project
management
tutorial
back
in
February
now
and
one
of
my
action
items
from
that
was
to
post.
A
Basically
anonymizer
summarize
the
survey
results
there
was
a
survey
distributed
to
the
team
after
the
week-long
def
meeting
in
spring
of
2016.
So
that's
about
a
year
ago
now
so
I
went
through
that
survey.
I
summarized
the
responses,
so
the
responses
aren't
identifiable
to
individuals
and
I've
posted
them
on
the
project
management
wiki.
So
you
guys
can
take
a
look
at
that
feel
free
to
make
any
comments,
but
much
I.
Don't
think
you
can
actually
comment
in
that
space.
A
But
actually,
if
you
want
to
make
a
section
at
the
end,
for
comments
feel
free
to
do
that.
So
you
can
take
a
look
at
that
and
you
know
we're
still
using
that
to
kind
of
guide
some
of
the
leadership
and
the
work
that
we're
doing
on
the
team-
and
this
is
in
preparation
for
our
staff
meeting
in
May
Seth
all
I
have
from
the
project
management
side.
A
A
B
Five
point
0
is
is
finally
at
long
last
out,
so
so
that's
quite
exciting
and
I
know
that
we
already
have
a
number
of
things
queued
up
that
we
have
been
waiting
until
the
book
price
point.
0
is
out
so
that
we
can
start
landing,
5.1
PRS,
including
one
that
got
landed
and
then
and
then
forced
to
take
off
again
and
say.
Yes,
we
we
can
start
start
bringing
those
in
now.
A
C
Live
so
yeah
I
mean
blog
post,
you
know
just
read
that
make
sure
I
mean
I
know
some
people
reviewed
it
already,
but
just
you
know
give
it
a
read:
make
sure
that
everything
looks
good
but
other
than
that
I
think
I.
Think
I
think
it's
like
celebration
time.
You
know.
D
A
E
So
we,
a
couple
weeks
back
I,
pushed
out
a
bunch
of
modifications,
the
templates
so
that
we
can
have
global
filtering
using
a
bunch
of
trail.
It's
and
then
a
new
command
line
argument
that
gives
people
what
they
wanted
to
a
very
long
time,
which
is
the
ability
to
just
came
to.
Let
all
prompts
initially
that
had
looked
pretty
ugly
and
now
it
looks
pretty.
E
E
I'm,
mostly
working
right
now,
and
it
changed
the
download
menu
in
the
notebook
for
how
it
interacts
with
NB
convert,
got
lots
of
help
from
various
people
on
this,
and
it's
been
taking
a
lot
but
I
think
and
there's
a
lot
even
once
this
is
done.
There
will
be
a
lot
more.
They
do
but
I'm
excited
for
what
it's
looking
like,
because
I
think
it's
going
to
massively
improve
our
publications
sort
of
story,
and
then
I
was
just
going
through
and
adding
more
things.
E
I
know
that
there's
been
a
lot
of
bugs
that
were
picked
up
by
damien
and
many
others
as
well,
but
I
need
the
time
to
actually
summarize
those
before
summarizing
them
here,
but
but
in
terms
of
the
things
that
could
use
attention
from
others,
the
pr
that's
waiting.
If
anyone
wants
to
review
that
and
give
it
a
thumbs
up,
that'd
be
great
and
I'm
a
bug.
People
more
about
the
about
the
configurable
uploads.
A
Okay,
thanks
Mike
sure,
okay,
all
right
Pete,
don't
think
Pete's
here
today.
So
then
you
impede
usually
share
the
Services
section.
Did
you
have
any
updates
on
that
now.
D
F
So
yeah
it's
been
a
week
of
going
through
just
the
remaining
issues
toward
a
beta
release.
We
have
this
new
concept
on
the
client
side
called
the
client
session,
which
is
distinct
from
the
server
idea
of
a
session.
It's
actually
I
think
that
wraps
metadata
around
the
kernel,
and
so
things
like
notebooks
and
consoles
and
inspectors,
and
all
those
things
that
need
to
talk
to
a
kernel
now
just
take
a
client
session.
They
don't
they
don't
manage
their
own.
F
They
don't
manage
their
own
relationship
with
the
API
the
way
they
used
to
so,
for
example,
a
console
used
to
not
need
all
the
same
things
that
a
notebook
needs,
but
now
that
there's
one
uniform
thing
is
actually
sort
of
simplifies
things
and
another
thing
that
we're
trying
to
simplify
in
the
same
way
is
the
way
foul
browsers
are
created
and
the
way
they
were
internally.
So
we
want
the
document
manager
to
be
a
singleton
that
eventually
it
probably
gets
renamed
the
file
manager
and
is
effectively
analogous
to
a
file
system.
F
So,
irrespective
of
how
many
servers
you're
connected
to
they're
just
different
mounts
inside
the
file
manager-
and
that
way
you
don't
have
to
manage
your
own,
you
don't
have
to
manage
the
way
you
create
a
phone
and
a
file
browser.
Suppose
you
want
it
on
the
left
sidebar
to
have
your
local
file
browser
and
on
the
right
sidebar.
You
wanted
some
server
and
you
even
want
to
be
able
to
drag
and
drop
between
them
and
those
sorts
of
interactions.
They
become
really
difficult
unless
they're
actually
sitting
on
top
of
a
file
manager.
F
That
basically
knows
how
to
mediate
between
a
different
mount
points.
So
right
now,
the
the
the
file
browser
extension
offers
a
factory
that
generates
a
file
browser
for
you
with
all
of
the
application
level
static
classes
for
state
management
for
connecting
the
services
and
for
connecting
to
the
document
manager.
F
Basically,
we
have
exhausted
all
of
the
logical
possibilities
of
building
an
extension
on
an
author's
machine
with
oh
sorry,
installing
an
extension
on
a
client's
machine
without
having
access
to
the
node
ecosystem.
And
it's
just
it's
not
a
tractable
problem
because
of
the
non-standard
way
that
libraries
can
different.
Javascript
libraries
can
offer
chunks
of
functionality.
So,
for
example,
code
mirrored.
F
There's
no
single
file
in
the
code
mirror
package
that
you
can
say
bundle
this
and
every
single
thing
that
comes
with
premier
comes
with
it
because
you
could
actually
say
require
code,
mirror
/,
add-on,
/
food
and
so
there's
no
guaranteed
way.
Even
if
you
have
the
same
exact
version
of
code
mirror
on
Jupiter
lab
as
you
do
in
some
extension.
F
There's
no
way
to
one
hundred
percent
guarantee
that
you
have
all
the
bits
that
some
extension
author
might
be
using
and
that
problem
is
just
not
solvable
without
basically
rewriting
parts
of
NPM
and
parts
of
web
pack
ourselves
in
Python.
So
our
operating
model
right
now
is
we
want
to
offer
Jupiter
lab
with
maybe
more
default
extension,
so
that
the
normal
use
case
is
just
to
install
Jupiter,
lab
or
or
comments
all
Jupiter
lab.
F
But
then,
if
you
want
to
install
a
an
extension-
and
you
do
not
want
to
use
Conda
which
can
have
known
in
it,
then
you
will.
You
will
have
a
requirement
of
having
no
vending
machine,
at
least
at
least
for
beta,
but
it
may
well
turn
out,
but
that's
just
what
we
need
to
do
and
on
the
phosphor
front,
chris
is
continuing
to
work
on
the
grid.
F
Yesterday
he
added
a
small
feature:
that's
actually
a
reason
for
a
minor
bump
on
all
the
packages,
because
it
is
effect
signal
and
the
messaging
system
where
basically,
we
didn't
have
a
good
way.
A
signal
connection
had
thrown
an
error
because,
in
order
to
make
sure
the
rest
of
the
app
doesn't
fall
apart,
what
was
happening
was
that
all
errors
were
basically
being
caught
in
a
try-catch
block
and
just
counsel
warned
out,
which
is
fun
in
the
browser,
but
not
so
great
for
writing
tests.
F
A
F
Think
at
the
moment,
what
we
really
would
like
to
do
is
once
once
we
have
the
extension
piece,
those
that
part
of
the
extension
beside
the
scribe.
Rewritten
would
really
like
people
to
write
extensions
and
not
not
just
the
ones
that
we
have
now,
but
even
like
totally
extensions
or
thinks
the
demo,
or
something
like
that,
and
to
get
back
to
us
how
good
or
bad
that
war
slower
experiences.
Because
clearly
that
is
one
of
the
main
demographics
we
want
to
target
and
we
want
to
accommodate.
F
So
we
know
what
the
technical
limitations
are
and
we
you
know
where
we
can't.
We
can't
do
what
is
not
tractable,
but
if
there
are
ways
of
making
things
easier
that
that
that
is
often
only
discoverable
by
tons
of
users
and
not
just
a
you
know,
deep
thought
so
so
that
that
is.
But
that's
not
that's
not
something
right
now,
I
would
say
you
know
once
once
we
get
very
close
to
beta,
but
before
beta.
A
G
H
We're
continuing
we're
continuing
to
overhaul
the
protocol
and
clean
it
up.
I
think
there's
basically
one
last
major
part
that
sylvan
and
I
are
just
still
discussing,
and
then
we
plan
on
fixing
a
few
things
that
have
cropped
up
since
six
point
0
and
then
releasing
a
seven-point,
oh
pretty
quickly.
I
don't
want
to
commit
to
a
timeline,
but
not
nine
months,
like
six
point
I'll
take
commit
to
that.
So
so
we
hope
to
get
them
point
out
out
the
door
pretty
quickly.
H
A
A
H
I
D
A
I'm
staying
somewhere
else
today
shows
the
Alpha
my
weekend
trips
and
those
are
the
mugs
specific
that
wouldna
the
kitchen.
So
thanks
man,
alright,
what
else
do
we
have
going
on?
So
let's
see
conferences
and
outreach.
It
looks
like
for
the
first
time
in
a
couple
weeks
or
no
new
conferences
that
we're
adding
to
our
list,
and
they
have
to
take
that
back
in
a
minute,
as
we
are
updating
our
leg
notes
and
then
we
have
a
whole
bunch
of
stuff
coming
up
in
the
spring
and
so
I.
B
A
A
A
A
I
A
I
can't
wait
for
that.
Let's
see
craft
2017
and
Budapest
April
25th,
228,
I'm,
thinking
that
that
silver
and
silver,
giving
a
tutorial
and
I
widgets
pi
day
to
London
five
to
seven
ma,
Taiwan
psych
on
Taiwan
jun,
9
to
11,
hi,
Paris
june,
twelfth
or
thirteenth
and
science
Gateway,
Community,
Institute,
webinar,
July,
eighteenth,
o
July,
eighteenth
or
20th
and
Carols
presenting
that
sounds
pretty
cool.
We
also
have
Python
in
Portland
coming
up
that
needs
to
be
added
to
that
list.
So
that's
in
may
right.
A
A
This
Lance
they'd,
like
others,
to
kick
the
tires
by
writing
extensions,
even
toy
ones,
just
to
play
around
with
the
workflows
and
then
releases
this
week
are
on
notebook
five-point,
oh
hi,
Kurt,
I,
pi,
Colonel
4.6
and
the
Jupiter
client
5.0
point
one
and
that's
the
end
of
the
note:
does
anybody
have
anything
else
that
they'd
like
to
add,
never
share
with
the
community
while
we're
still
on?
Oh.
E
One
of
the
bugs
that
was
fixed
is
related
to
defining
dotted
string
default
values
inside
trait
lips,
which
apparently
will
trigger
an
import
in
a
way
that
dynamic
default
values.
Won't
that
is
so
just
know
that
those
are
not
actually
equivalent,
even
if
you
weren't
ever
to
change
the
dynamic
value
that
is
good
to
know,
we're
not
sure
why
state
trigger
and
import
right.
Thank
you.