►
From YouTube: Jupyter/IPython Dev Meeting, September 27, 2016
Description
Meeting of the Jupyter/IPython development team, September 27, 2016.
Meeting notes: https://jupyter.hackpad.com/September-2016-ncdE6E6XQmc
A
Hello,
everyone:
this
is
the
Jupiter
developer
meeting,
it
is
Tuesday
September
27th
and
a
number
of
us
are
travelin.
Some
people
are
gone
today.
We're
a
couple
of
us
are
in
New
York
City
for
strata
which
starts
I.
Guess
it
actually
technically
started
yesterday.
But
our
talk
is
tomorrow
and
looks
like
we've
got
a
lot
on
the.
D
A
F
Everybody
I
don't
have
a
whole
lot
other
than
we've
been
working
hard
on
Jupiter
hub
0.7
and
getting
some
things
documented
and
tested
and
getting
a
process.
F
And
I
think
that
you
know
Brian
I've
got
the
get
sub
tree
proof-of-concept
going
and
also
another
theme
that
we
might
want
to
consider
migrating
to,
because
it's
very
similar
to
read
the
docs,
but
a
lot
more
visually
appealing
and
conscious
using
it.
So
that
is
up
and
the
only
other
question
I
had
is.
F
D
Ok,
so
to
me,
sometimes
to
unmute,
because
my
computer
was
so
so
I
finally
posted
a
blog
post
about
ipi,
which,
at
the
creativity's
four
of
those
who
are
not
involved.
Last
May
Brian
discover
the
bug
in
Hawaii
widgets
that
would
trigger
a
trusting,
trusting
internal
book
on
seven
node,
which
was
a
giant
security
issue,
because
it
would
allow
arbitrary
code
execution.
Basically,
you
send
someone
a
lot
book
and
Wednesday
loads
notebook
if
they
had
the
wrong
version
of
a
pib
jet.
D
We
still
made
the
disclosure
about
the
security
issue,
and
so
for
everyone
listening
to
us,
sorry
for
taking
that
long
and
to
publish
the
blog
post
and
please
update
IDI,
we
chat
as
soon
as
you
can.
If
you
are
a
user
of
conduct,
I
remember
correctly,
and
you
didn't
install
a
TI
widgets
using
pic,
you
shouldn't
be
able
to
get
a
wrong
version
of
IPI
widget.
So
if
you
have
never
even
started
going
jet
roaring,
oh
girl,
you're,
probably
probably
safe.
D
That
was
it.
If
you
have
any
questions,
feel
free
to
email,
us
even
privately
or
use
a
security
meaningless
security,
a
type
I
cannot
org
to
ask
more
question.
I
also
got
a
seminar
at
UC
Merced
last
friday,
which,
when
great
it
was
just
a
small
caramel,
45-minute
dog
and
people
are
excited
still
easy.
D
D
Instead
of
having
tools,
detecting
ipython
to
do
a
back,
porch
of
tr's
mean
have
separated
at
in
a
separate
people,
and
we
now
recommend
to
use
the
specific
project
to
have
backport
clear
and
everything,
and
some
people
from
at
lucky
bar
actually
really
watching
closely
what
we
are
doing,
because
as
soon
as
we
change
the
doc
say,
you
wrote
to
me
a
chat
and
say:
hey,
that's
so
great.
Can
we
use
it,
but
we
have
some
a
change
that
we
would
like
to
do.
D
So
it
seems
that
some
some
projects
across
the
side
by
community
you
want
to
uniform
eyes
the
work
flow
across
project
to
make
users
and
developers
having
less
context
switching
when
they
move
from
one
one
project
to
another
end
and
stupefied.
You
need
from
that
is
a
process,
and
that's
about
it
for
me,
from
Alec.
A
And
then
yesterday,
sylvan
Matt,
Rockland
and
I
were
sprinting
at
Bloomberg
and
Sylvan
was
working
on
some
stuff
I'll.
Let
him
update
you
as
long
as
he
can
get
blue
jeans
working,
but
then
Matt
Rockland
showed
me
a
lab
extension
or
a
plug-in
that
Luke
from
continuum
had
written
that
allows
a
user
to
open
up
bouquet
visualizations
that
are
plugged
in
to
desk,
and
so
you
can
get
these
really
nice
visualizations.
A
A
Something
like
that,
it
we're
really
quick
and
I.
Did
it
by
hand
I
didn't
use
the
cookie
cutter
that
Dan
Alan's
been
working
on
I
think
once
we
get
that
cookie
cutter
working
that
would
go
down
significantly
and
so
I
think
that's
it
other
than
the
global
issues
up
above
I
wanted
to
bring
up
and
looks
like
Peters
next.
G
I
can
hear
me:
I
took
I,
think
I
put
a
minus,
so
I.
You
know
nothing
other
than
what
I
wrote
there
to
report.
I
guess
one
thing
since
I'm
talking
about
the
mb
format
issue.
If
you're
curious
about
are
interested
in
tracking,
you
know:
environments
or
libraries,
con
environments,
virtualized
environments
lock,
remember
to
whatever
you
want
to
call
them
alongside
notebooks.
There's
an
issue
there
where
there's
a
discussion,
I'm
going
so
tribe
in
before
I,
go.
E
A
E
E
H
Just
a
couple
eyes,
blue,
but
yeah
most
of
my
stuff
would
just
be
following
up
them
last
week.
If
anyone
had
any
for
this
loss,
templates
and
custom
exporters
and
stuff
like
that
and
other
kinds
of
customization
to
the
end
users
process
happy
to
talk
about
it,
but
otherwise
really
still
doing
research
on
what
the
best
way
to
pursue
a
couple
things
it
so.
A
A
D
A
H
Because
yeah.
A
E
C
The
Python
36
support
I'll,
probably
cut
trail
it's
with
with
that
fix
tomorrow,
so
that
we
can
start
testing
other
things,
the
only
the
other
things
to
bring
people's
attention
to
our
there's,
a
current
PR
to
finally
get
container
configurable
container
traits
to
be
accessible
on
the
command
line.
So
you
can
do
our
part
style
like
dash
dash
food
thing
that
shush
food
thing
to,
and
then
you
get
those
items
into
a
list.
C
The
only
shoe
with
that
is
that
the
technically
works,
but
largely
undocumented
or
may
be
entirely
undocumented
way
of
setting
a
list
traits
and
things
on
the
command
line.
Look.
It
will
probably
be
hard
to
make
that
still
work.
I'm,
not
so
concerned
about
that,
since
that
has
always
been
really
horrible
and
undocumented
I
think
Jess,
as
one
person
I
know,
who
had
used
it
and
I
her
reply
was
basically
that
she
had
given
up
because
it
was
so
horrible
which
makes
me
feel
good
about
removing
it.
C
But
if
anybody
has
wants
to
bring
up
keeping
that
a
old
way
working
hop
on
the
issue,
I
liked.
The
other
thing
is
that
there's
a
proposal
for
kind
of
defining
them
metadata
for
things
like
title
and
authors
in
MV
format.
So
if
you're
interested
in
that
stuff-
and
I
pinged
a
few
people
like
Brando
about
that
one
up
on
that
issue-
yeah,
that's
all
I
have
to
bring
to
every
attention.
E
Great
thanks
so
I
have
a
note
in
the
chat
from
guest
6
that
says
sorry
was
responding
to
fires.
E
I
Hey
sorry,
Jake,
honey,
Kish
communication.
These
are
really
bad
because
of
blue
jeans,
crushing
so
I
hope
you
can
hear
me.
I
just
wanted
to
give
a
heads
up.
We
had,
as
we
were
doing
some
demo
driven
development
with
a
Brian.
Yesterday
we
got
some
really
nice
green
cast
of
integration
of
I
polyfit
and
we
keep
got
into
Jupiter
lab
so
yeah.
I
If
you
wanna
see
something
nice,
you
can
see
these
in
the
pull
request
of
these
projects
correspond
to
the
integration
with
the
ipad,
and
there
is
the
other
thing
that
I
just
put
in
the
back
burner
because
of
the
demo
tomorrow,
which
is
the
work
and
Jupiter
server.
I
know
that
Matias
for
some
concerns
about
it,
I
think
we
should
take
this
offline
and
have
a
more
thorough
discussion
with
brian
and
everyone
who
wants
to
be
involved
so
yeah.
That's
it
for
me.
E
Wait
excellent
on
Nick,
nothing.
Okay,
I'll,
give
a
short
update
for
honor,
so
we're.
E
We
have
a
location
for
the
fall
depth
meeting,
it's
called
the
sillim
are
so
we
have
a
great
space
there.
However,
they
couldn't
accommodate
us
during
the
week
that
we
wanted
so
honest,
sent
out
an
email,
I
believe
yesterday
checking
with
people
to
see
if
those
days
will
work.
So
I
think
that
the
days
are
the
11th
through
basically
was
up
like
a
friday
to
a
tuesday.
So
please
respond
to
her
on
her
survey,
her
inquiry,
so
we
can
get
a
headcount
for
that.
It's
a
beautiful
space.
E
Yes,
I,
think
that's
it.
I
think
in
the
next
couple
weeks,
we'll
we'll
look
more
at
github
projects
and
how
we
can
use
it
on
some
of
the
larger,
with
some
of
the
larger
teams
so
stay
tuned
on
them
and
it
looks
like
Daniel
and
just
joined
hi
Dan
all
right.
How
are
you
good
good?
Do
you
have
anything
you
want
to
share
with
the
group
or
just
I'm
reaching
the
end
of
the
meeting?
Oh
great.
H
B
A
Yeah
there
was
a
couple
global
topics:
I
wanted
to
bring
up
the
one
of
them.
The
question
of
the
Jupiter
server
I
think
we
probably
want
to
have
a
separate
meeting
for
that,
because
I
think
it's
going
to
be
a
longer
discussion
and
I'm
wondering
Jamie.
If
you
could
help
us
schedule
that
at
some
point,
I
think
it's
beyond
the
scope
of
this
meeting
and.
A
A
We
would
have
a
custom,
mime,
type
or
widgets
or
bouquet
for
all.
Tear
for
geo,
JSON
and
I
think
we.
We
need
to
standardize
what
we
call
those
mimetypes.
Even
if
we
don't
so
I,
don't
think
it's
quite
at
the
level
of
a
message.
Spec
change,
but
we
basically
need
different
front
ends,
need
to
agree
on
what
we,
how
we
spell
the
different
one
types
up.
So
if
it's
do
you
Jason?
How
do
we?
What
do
we
call
that
in
classic
notebook,
Jupiter
lab,
interact
and
other
front
ends?
A
H
A
Tricky
I.
C
It
seems
likely
that
they'll
end
up
at
the
display
data
message
in
in
those
docs,
and
so
unless
it
doesn't
feel
quite
right
to
me
to
put
it
in
the
protocol
documentation,
since
it
isn't
really
really
like
the
HTTP
documentation.
If
you
look
up
HTTP,
it
doesn't
describe
all
the
mime
types
for
the
files.
Your
HTTP
server
can
send
those
are
defined
elsewhere,
but
if
we
don't
have
a
better
place
than
next
to
the
display
data
message,.
H
H
C
I
D
Those
are
zara
guideline
why
you
need
to
use
the
vendor
prefix
when
you
create,
am
I
in
fact,
which
is
application
or
application
specific
and
github
uses
that
and
tributes
mean
everything.
Is
that
when
you
have
the
v,
ng
prefix
dot
and
then
something
I
lender?
Okay,
that's
why
that's
why
you
see
vnd
everywhere
in
keystone,
mime
type,
because
they're
made-
and
there
is
there-
is
a
guideline:
it's
not
a
bit
something
enforcement,
but
there
is
a
guideline
if
you
make
something
which
is
application,
specific
use,
these
specific
prefix.
C
Oh
so
on,
in
addition
to
kind
of
all
the
standard
mimetypes,
the
things
like
the
widgets
mime
type
that
is
currently
used
by
the
widgets
there's
going
to
be
somewhere
for
these.
That
kind
of
thing
like
the
widget
line,
type
to
go
like
these
are
a
couple,
a
couple
things
that
are
defined
pacifically
for
use
in
this
protocol.
C
H
I'd
like
it,
if
we
could
at
least
document
all
the
silver
custom
formatting
on
the
output,
so
I
mean
I
can
just
go
through
and
add
that,
but,
like
anything
that
we
use,
it
would
seem
like
actively
in
the
code
base.
To
have
a
special
meaning
would
seem
to
be
a
worthwhile
thing
to
have
like
at
least
included,
to
the
extent
that
we're
using.
C
A
D
If
library
wants
to
be
rendered
and
give
your,
I
guess
most
library
we
want
you,
because
it's
a
metal
is
the
main
way
of
sharing
book,
usually
and
say:
hey
here,
make
a
pull
request
and
if
you're
that
make
your
clicking
available
to
require
jess
to
whatever
I
don't
care
and
it
would
register
the
pen
type
with
and
the
convert
and
then
the
viewer
as
a
Python
package.
This
entry
point
we
can
chat
later
and
then
it
will
give
an
incentive
to
people
to
register
Thurman
types
there
and
the
whip.
D
I'm
types
exists,
and
when
we
really
plugin
behavior,
we
can
have
a
custom
page
that
lists
all
the
mime
type
that
end
with
your
dozen
ever
yeah
or
does
not
enable-
and
we
know
about
and
say,
hey
here
is
an
up
to
date
list,
because
I
bigger
is
something
dynamic.
Is
that
show
something
to
users?
I
think
that
would
be
nice
great
is
to
contact
so.
I
C
A
A
B
C
A
Great
so
I'll
put
the
comment
on
that
issue
about
that.
I
think
that
is
everyone.
Okay
with
that
solution,
as
good
as
any
collection
me,
okay
and
then
one
other
issue
that
came
up
with
is
that
basically
how
payloads
are
rendered
in
friends?
And
the
history
here
is
that
the
payload
message
traditionally
had
been
measured
in
the
thing
that
we
call
the
pager,
but
the
ipython
kernel
for
a
while
has
had
a
flag
that,
on
the
colonel
side,
routed
?
do
output
rather
than
payload.
A
So
it
was
a
colonel
side.
Configuration
I've
been
running
the
notebook
with
a
guy
I
present
Colonel
bending
hell,
basically,
the
pager
output
to
the
actual
output
for
a
long
time
and
Kyle
said
that
interact
is
doing
this
by
default
in
the
front
end,
and
it's
something
we've
talked
about
for
a
long
time
and
so
I
think
at
this
point,
Kyle
and
I
are
both
in
favor
of
informally
they
were.
Men
should
probably
typically
display
payload
output
or
payloads
as
regular
output
and
the
main
issue.
A
The
main
reason
for
that
is
that
you
typically
get
a
payload
by
writing
code
in
a
in
a
cell.
Like
no
I'm
pedal
ended
up,
rent
?
and
by
putting
the
output
in
the
axils
output
them
sell.
You
make
that
notebook
more
reproducible
right
now.
If
we
don't
do
this,
those
notebooks
on
github
and
then
be
viewer.
If
they
have
a
question
mark
cell,
you
don't
see
the
output
because
it
was
sent
to
this
weird
pager
and
location
and
I
just
wanted
to
get
feedback
about
that.
A
A
C
C
But
some
people
do
like
it
and
the
main
reason
to
like
it
is
that
you
can
it
lets
you
have
that
information
documentation?
What
have
you
that
you
can
see
as
you
scroll
around
right
right
and
that's
the
main
thing
that
regular
output
doesn't
do?
Is
it
stays
where
you
put
it?
But
if
you,
if
you
make
it
easy
kind
of,
pin
or
float
an
output
which
I
believe
interacts
does
and
Jupiter
lab
at
least
can
do,
then
I
think
that
significantly
reduces
the
value
of
the
pager
as
a
dedicated
thing.