►
From YouTube: Jupyter/IPython Dev Meeting, August 30, 2016
Description
Meeting of the IPython/Jupyter development team on August 30, 2016
Meeting Notes: https://jupyter.hackpad.com/August-2016-NVKlJmYxWK2
A
B
Let's
see
I
spent
a
great
week
a
couple
days
down
in
San
Luis
Obispo
last
week,
working
with
Carol
and
on
an
intern,
hey
Brian,
learning
more
about
the
documentation
process,
and
you
know
just
the
nuts
and
bolts
of
how
we
build
docs
incredibly
useful
session.
So
thanks
so
much
Carol
and
thanks
so
much
Anna
for
your
hospitality,
hi
file,
and
so
this
week
we'll
just
be
spent
and
following
up
on
some
of
the
things
we
talked
about
here,
kind
of
building,
more
muscle
memory
and
doing
you
know
working
with
github,
etc.
A
Maybe
you
can
give
us,
give
us
some
tips
and
pointers
on
that,
because,
from
from
what
I
heard,
that
session
was
very
useful
and
I
and
I
actually
suspect
that
quite
a
few
of
us
are
not
completely
up
to
speed
on
all
of
the
work
that
has
happened
in
that
front
and
I
was
sort
of
wondering
how
how
we
can
make
sure
that
that
knowledge
can
percolates
better
across
the
team,
but
I
think
up.
Next,
we
have
jamie
is
that
it.
A
C
C
I
can
post
them
later,
but
that
combined
with
there's
a
documentation
section
in
the
Jupiter
/
Jupiter
docks
that
sort
of
walks
people
through
how
the
docks
are
built,
how
you
would,
if
you
were
starting
from
a
brand
new
repo,
how
you
would
set
up
docks
and
Jamie
kind
of
took
a
list
of
things
that
we
might
want
to
add
in
her
notes
to
that,
but
it
I
think
covered
it
actually.
Quite
well,
so
you
know.
Basically
those
would
be
the
two
best
resources.
A
Is
that
the
link
so
I
just
pasted
a
link
on
the
hack
pad
to
the
in
the
contributor
guides
the
documentation
guide?
I
presume,
that's
the
one
that
you
were
referring
to
correct,
ok
and,
and
so
yeah,
maybe
later.
If
somebody
has
the
links
to
the
two,
those
posters,
because
I
did
hear
from
Jamie
that
those
posters
were
very
useful
and
I
have
to
admit
that
I
actually
haven't
seen
them.
So
I
I
think
it
would
be
kind
of
useful
stuff
to
share
more
wife
with
with
folks
at
large
sure.
C
D
C
He's
been
involved
with
some
discussions
recently
on
the
services
API
with
men
and
others.
So
but
I
will
keep
my
ears
open
and
what
we
might
do
is
actually
remix
some
of
the
workshop
content
into
a
notebook
and
then
post
that,
but
that
depends
on
whether
or
not
I
have
time
to
do
it.
But
it
might
just
make
sense
to
send
out
a
link
again
to
the
repo
so
that
you
know
maybe
early
September.
So
people
know
it
exists.
Just.
A
Just
kind
of
as
a
reminder
yeah
and
actually
actually
we
probably
should
move
it.
Github
is
pretty
good
about
redirecting
automatically,
but
we
should
probably
move
it
to
the
resources
org,
since
it
is
now
that
we
have
the
resources
org.
This
is
exactly
those
kinds
of
kind
of
training,
materials
talks,
tutorials,
etc,
yeah,
but,
but
even
if
we
move
it
to
the
github
reader
is
very
good
at
redirecting
automatically
moved
repo,
so
no
no
links
will
break.
A
C
E
C
A
C
A
Just
written
there's
a
recording
of
the
entire
meeting,
so
it
ended
up
in
to
youtube
video.
So
it's
to
roughly
one-and-a-half
our
videos
and
there's
and
the
repo
has
the
slides
from
everyone.
So
the
repo
has
a
bunch
of
pdfs
and
then
there's
links
to
the
to
youtube.
The
big,
the
two
big
YouTube
set
of
videos
back-to-back
you.
F
Think
there
might
be
an
advantage
to
having
say
a
five
minute,
video
walking
through
just
getting
deduct
up
and
running
because
since
that
can
be
faster
for
offices,
then
actually
like
going
through
the
processes
like
if
they
had
a
screenshot
sort
of
way
of
approaching.
It's
not
like
a
priority,
but
just
as
just
an
alternative
means
of
conveying
the
information
for
someone
who
is
used
to
reading
documentation
up.
A
A
C
A
E
Yep
the
main
thing
I
like
to
draw
people's
attention
to
is
Jupiter,
client
and
trail.
It's
both
have
just
a
couple,
tiny
PRS
in
between
them
and
Jupiter
clients
and
the
release
and
trail
it's
Anna
and
the
beta
they
should
be
able
to
get
out
this
week.
Otherwise,
I'm
working
on
Jupiter
hub
service
api's.
Finally,
and
that's
what
I
plan
to
spend
most
of
this
week
on,
should
have
fairly
complete
implementation
of
that
emerged.
I
think,
hopefully,
at
the
end
of
this
week,.
A
Right,
my
anything
you
want
to
highlight
for
the
team.
A
G
Excellent,
no
I
mean
just
what's
in
the
hack
bed.
There,
someone
contributed
tensorflow
notebook,
docker
image,
I
thought,
I
was
kind
of
cool,
was
going
to
work,
John
Gurda
on
getting
I.
Think
someone
client-side
package
is
set
up
for
distributed
moda
to
a
calmness
words
so,
but
that
was
cool
other
than
that.
You
know
just
what
it
says
there:
nothing,
nothing
critical
great.
A
Yeah
and
one
thing
actually
about
Conda
Forge
that
we
recently
realized
have
been
added
to
Conda,
but
if
people
aren't
familiar
with
that,
because
it's
poorly
documented,
even
Peter
wank
wasn't
aware
of
the
feature-
is
that
now
in
con
that
you
can
actually
add
a
channel
for
more
experimental
stuff
like
Conda,
forge
in
within
a
specific
content,
you
have
to
put
the
con
dar
see
file
in
the
condign
rather
than
at
the
top
level.
But
that
means
that
you
can
have.
A
You
can,
for
example,
activate
Conda
Forge
only
in
a
specific
environment
rather
than
having
it
applied
globally
and
that's
actually
not
easily.
It's
not
easy
to
find
it
in
the
in
the
docks.
But
that
means
that
adding
things
like
this,
the
specific
end
is
viable
without
having
the
condo
for
channel
affect
your
stable,
repose
and.
A
Pulling
in
yeah
it's
not
documented
anywhere,
in
fact,
Peter
Wang
and
I
were
asking
for
the
future
and
someone
from
continuing
said
no.
Actually,
we
did
add
that
recently,
but
it's
wow,
if
you
just
put
the
Conda,
is
it
Conda,
RC
in
the
root
of
the
environment?
I
think
that's
the
case.
There
is
a
place
I
make.
I
can
try
to
find
the
issue
or
I
actually
begged
them,
to
add
it
to
the
docs
actually
filed
a
bug
about
adding
that
to
the
dogs,
but
I
think
that's
it.
A
If
I
remember
correctly,
I
think
I
put
the
doc
on
our
see
other
way
to
the
end.
Ok,
oh
yeah!
Well,
we
can
dig
out
the
link.
It's
super
useful
on
because
channels
like
Conda
Forge,
our
kind
of
dangerous
to
activate
globally
because
they
can
pull
in
new
versions
of
stuff
and
though,
oh
right,
so
they
can
easily
wreck
your
normal
environments.
If
you
do
a
condo
update
but
being
able
to
add
them
in
isolation
is
really
really
fantastic.
Yeah.
All
oh.
A
G
A
G
I
Yes,
I
just
posted
the
link
to
the
the
heated
discussion
about
third-party
extensions
for
Jupiter
lab
I
tried
to
link
to
the
actual
roll-up
comment,
but
hack
pad
is
not
being
nice
and
letting
me
link
directly
to
an
issue
comment,
but
it's
the
one
with
three
major
bullet
signs,
probably
just
move
it
to
the
top
on
at
this
point,
but
where
I
think
converging
on
a
solution
that
will
allow
us
to
not
have
to
have
a
build
chain
on
the
user's
machine,
they
won't
need
to
have
note
installed
to
add
new
packages
but
at
the
same
time
not
have
the
same
JavaScript
loaded
twice
on
the
page.
I
So
it
will
have
a
duping.
You
would
have
gotten
from
running
the
build,
but
have
that
happened
in
the
client
so
make
things
a
lot
easier
for
the
user.
It
just
involves
us
making
a
couple.
Plugins
up
front
looks
like
we're
going
to
make
plugin
for
web
pack,
which
I'm
working
on
right
now
and
then
one
from
requirejs
to
make
this
all
happen
seamlessly
behind
the
scenes.
So
if
anyone
else
wants
to
add
their
voice
to
the
ferbin
discussion
at
the
post
there,
that's
it
cool.
A
Really
it's
a
tricky
problem
with
lots
of
sides
to
it
and
and
lots
of
kind
of
delicate
counterbalancing,
prop
compromises,
and
it
seems
like
you
guys,
are
working
well
on
it.
But
I
do
have
a
question.
The
client-side
dee
doop.
Does
that
happen
at
runtime,
or
is
that
a
client-side
installation
step
that
the
user
will
have
to
do?
I
just
haven't
followed
all
the
details.
I
On
it
turned
out
to
be
a
custom
required,
yes
loader,
so
it
will
get
the
prefix
whatever
we
call
it
!,
which
will
tell
it
to
load
using
our
loader.
So
we
just
provide
a
load
function
that
amongst
to
see
based
on
the
summer,
stringing
that
load
path
string
if
we
the
best
model
the
hand
back
based
on
the
ones
that
are
loaded
on
the
page
already.
So
it's
just
an
indirection
to
allow
us
to
only
give
that
one
thing
as
well:
ok,.
A
I
H
A
A
J
The
balance
there
is
that
somebody
that
wants
to
stand
up
a
server
and
host
particular
instances
of
Jupiter
lab
applications
can
use
node
to
creed
ii.
Do
whatever
bundle
they're
going
to
serve
your
clients
that
would
that
clearly
require
note,
but
for
the
average
user,
that's
just
running
on
local
host
of
all
there
yeah.
D
A
Be
great
for
these
long
and
complicated
issues.
Often
it's
a
good
idea
to
then
at
the
end,
go
back
and
edit
edit
the
issue
description
at
the
top
and
and
just
sort
of
add
whatever
whatever
came
out
as
as
as
resolution,
just
slap
it
at
the
bottom
of
the
issue
description
and
that
it
makes
it
easy
to
find
once
they
get
closed
and
sorted
out.
A
A
K
A
I
am
sorry
I
was.
I
had
scrawled
too
far
down
on
the
on
the
hack
pad
thanks
for
catching
me,
which
means
that
we've
completely
messed
up
the
order.
I
guess
we
should
just
go
back
to
the
top
of
the
page
and
then
keep
running
through
a
keep
running
through
the
people
that
we
that
we
skipped
I
think
you
were
first
on
the
pen
on
the
actual
page,
yeah.
L
A
B
B
B
B
B
L
Okay,
so
my
only
other
update
would
be
on
the
dev
meeting.
I
am
working
on
venue
research
this
week
and
I
should
have
more
info
to
share
next
week
so
that
we
can
hopefully,
in
the
coming
weeks
book
our
flights
just
as
soon
as
it
makes
sense
and
get
a
final
head
count
on
who
will
be
attending
the
meeting.
Oh
also
I
have
I
am
working
on
the
final
details
of
announcing
Jupiter
day
Hawaii.
We
have
a
potential
date.
We
have
a
potential
event
venue.
A
A
O
Just
released
a
version
0.5
or
of
the
trade
types
psychiatry
types
and
I'm
working
on
the
dock
now,
so,
if
anyone
is
interested
in
this
specialized
trace
type
library,
I'm
going
to
use
it
in
big
blocks,
so
we
are
stroking
it
out
from
baku
plot,
and
I
guess
that
people
like
the
authors
of
Alger,
for
example,
could
be
interested
in
it.
So
we're
trying
to
make
it
as
per
level
as
possible.
This
Haslam
I've
been
working
on
the
new
and
the
extensions
mechanism
and
there
has
been
some
meetings
with
the
open
dream.
O
He
talks
for
at
the
schedule.
A
sparse-
and
we
spoke
mostly
spoke
about
the
packaging,
and
you
have
done
some
work
on
downstream
widget
libraries
to
and
by
the
way,
I
also
wanted
to
quickly,
maybe
introduce
a
ball
Allison,
who
I
think
he's
listening
on
the
call
and
we'll
studied
in
an
epic
refactoring
of
by
Fugees
JavaScript
hi.
A
Excellent
well,
hey
welcome.
Welcome
to
the
team.
Absolutely
delighted
to
have
you
on
board
and
yeah
by
three
Jas
is
something
that
I
haven't
used
it
as
probably
as
much
as
I
would
as
I'd
like
to,
but
every
every
time
I've
demoed
it
people
are
people,
love
it,
and
so
and
that's
that's
the
kind
of
functionality
that
we
should
probably
highlight.
More
and
Sylvan
is
sage
using
pi
to
bjs
in
any
in
any
way,
because
they
have
a
lot
of
3d
funny
and
say.
O
A
A
Come
to
the
team
able,
let
us
know
if
you
have
any
questions,
if
we
can
help
you
in
any
way
great
great
to
see
new
faces,
great
thanks,
dimion,
not
sure
if
you,
if
you
had
the
plus
or
not,
if
you
meant
to
have
the
boss
or
not,
so
I
don't
know
if
you
wanted
to
bring
anything
to
the
group's
attention
or
not.
I
think.
A
B
A
Yeah
I
have
just
a
couple
things
that
are
in
the
notes,
but
the
main
the
main
thing
that
I
want
to
flag.
It
is
that
I
just
got
news
that
we
got
awarded
a
small
grant
at
LPL
for
Jupiter
usage
in
HTC,
workflows
in
kind
of
large-scale
super
computing
environments,
and
so
we'll
be
looking
for
a
postdoc
I
just
got
the
news
recently
and
will
be.
There
will
be
a
postdoc
opening
for
someone
who's
interested
in
that.
A
That's
awesome,
yeah,
I'm,
quite
happy
about
it
and
it'll,
be
it's
with
shreya
Stroh
Leah,
who
some
of
you
folks
have
met
at
the
lab
and
as
for
the
rest,
it's
basically
sort
of
background
stuff.
That
is
in
the
notes,
and
we
can.
We
can
continue
Brian
anything
that
you
want
to
flag
with
the
team
yeah.
A
D
E
Last
week,
ok.
A
A
K
Yes,
sorry
has
difficulty
to
a
to
Anu
to
my
computer.
Is
super
slow
just
having
a
sending
quite
some
time
on
a
side
project
working
with,
took
it
all
out
and
feel
really
really
great?
It's
almost
as
a
normal
in
that
book,
and
sometimes
even
didn't
realize
that
I
was
actually
on
the
puter
lab
almost
full
screen
personal
google
TV,
really
nice
star
a
few,
a
few
side
tight
things
that
a
hint
the
joint
of
install
app
one
of
the
thing
that
I
discovered
is
when
we
start
to
grow
code.
K
K
The
gsi
take
over
and
just
desert
rough
inspection
of
the
token
with
a
cell
and
allow
you
to
at
least
complete
when
you
create
a
function,
local
variable,
you
have
already
declared
or
missiles
on
a
class
you're
actually
building,
and
it
was
one
of
the
things
that
app,
which
is
still
slowing
me
down
on
this
side.
Project
but
I
only
discovered
it
once
website
project
grew
quite
a
lot
and
I
had
a
huge
chunk
of
code
executed,
I.
A
A
N
Like
this
is
some
well,
we
decided
that
this
was
a
one
point,
0
issue
and
not
a
0.9
issue.
Basically
we're
not
doing
any
client
side
tokenizing
and
were
aware
of
it
if
it
becomes
a
thing
that
we
want
to
eat,
prioritize
and
set
it
to
a
different
house
phone
I'm
happy
to
do
it.
I
think
we
just
said
it
to
work
on
your
milestone,
but
you
know,
whichever
way
we
decide,
we
know
it's
there.
So
we
can.
We
can
disconnect.
K
On
friday-
and
I
was
from
your-
I
mean-
I
was
the
same
mind
that
you
are
that
it's
it's
one
point
of
pain,
but
after
doing
this
iphone
check
on
that
book
for
four
or
five
days,
it's
starting
to
be
a
thing
that
is
impending
me
into
moving
fast
it
to
writing
this
project,
because
I
make
a
lot
of
typos
and
I
keep
missing
tab
and
it
doesn't
complete.
But
otherwise
experience
is
close
to
see
less
and
I
think
for
many
people.
A
A
Thanks
materia
and
I
think
we're
kind
of
coming
towards
the
end.
There
is
a
note
from
the
Cal
Poly
students
will
have
no
internet,
so
people
can
rift,
read
it
on
the
on
the
notes
and
then
very
useful
upcoming
release
activity,
so
things
that
are
kind
of
coming
coming
off
the
oven
in
this
week
to
predict
line
for
for
trade
plus
4
30
and
trade
types
on
on
incubator
and
anyone
who
has
anything
that
we
missed
the
end
of
the
last
call
ass
before
we
stop
the
recording.