►
From YouTube: Jupyter/IPython Dev Meeting, December 6, 2016
Description
Meeting of the IPython/Jupyter development team, December 6, 2016
Meeting Notes: https://jupyter.hackpad.com/December-2016-weekly-meetings-EqtMljhwq6X
A
Morning,
everyone:
this
is
the
tuesday
december
six,
the
Jupiter
ipython
that
meeting
we
have
a
smaller
than
usual
crowd,
because
there's
a
Jupiter
dev
Jupiter
days
going
on
in
Paris
today
and
a
bunch
of
people
are
there
so
yeah?
Why
don't
we
go
ahead
and
Jamie?
Do
you
want
to
keep
us
on
track?
You're.
B
B
Ladies,
so
the
PI
ladies
event,
a
Bloomberg
on
December
13th
and
a
number
of
Jupiter
folks
here
in
Berkeley
are
gonna,
attend
so
Fernando
and
just
Hamrick
and
Carol,
and
not
sure
who
else
but
you're
welcome
to
come.
I
can
post
the
link
in
the
chat
when
I'm
finished.
With
my
update,
let's
see
I've
just
been
going
through
some
old
hack
pad
notes,
I
really
liked
what
Carole's
doing
with
the
herd
Jupiter
hub
highlights
for
the
week.
B
So
I
posted
a
link
here
in
the
december
half
pad
to
circle
back,
and
I
encourage
people
to
take
a
look
at
this.
Especially
team
leads
who
are
trying
to
figure
out
ways
to
to
combine
all
the
you
know
different
work,
that's
going
on
with
with
your
sub-projects.
You
know
you
can
drop
it
in
github
and
maybe
if
we
have
all
have
a
standardized
anodized
way
of
doing
this,
we
can
actually
figure
out
a
way
to
automate
the
the
feed
into
a
single
document
each
week.
I
don't
know
just
an
idea
Carol.
A
Just
wanted
to
reiterate
my
thanks
to
the
whole
team
for
the
for
the
help,
with
the
grant
reports
right,
that
same
material
can
be
used
by
Brian,
given
the
structure
that
we
used
will
also
be
used
by
Brian
for
the
help
sleep
foundation.
So
basically,
we
were
able
to
kind
of
knock
out
with
one
single
effort
and
small
amounts
of
fine-tuning
our
reporting
for
our
three
funding
agencies,
which
is
great
canoes.
It
was
really
much
better
to
have
everyone's
help.
A
One
thing,
but
I
would
like
to
do
all
probably
do
it
in
the
next
few
days
is
I'll
basically
hollow
out
the
content,
but
keep
the
key
outline
of
the
of
the
report.
We
worked
on
and
make
a
copy
of
it,
and
what
I'd
like
to
see
is
that,
as
time
goes
on,
we
basically
brought
material.
Anyone
can
drop
material
into
it.
Oh
as
time
goes
body,
and
if
you
use
I've
done
that
in
the
past,
where
I
keep
a
little
file
and
as
I
see
things
that
are
kind
of
relevant
to
the
project.
A
We
can
go
back
to
those
wings,
polish,
the
writing
and
then
we're
basically,
eighty
or
ninety
percent
of
the
way
there
so
I'd
like
to
do
that.
It
was
really
really
great
to
have
more
team
support
on
this
I'm
and
so
I
think
we
can.
We
can
even
stream
line
that
further
for
the
next
one,
which
will
be
do
I
mean
the
next
big
one
which
will
be
doing
about
a
year.
B
C
D
We
released
11.0
point
11.2
the
main
feature.
There
was
a
more
handling
of
the
session
state,
so
if
you
refresh
the
page
most
of
what
was
there
is
still
there
and
the
the
activity
you
were
currently
on,
we
should
read
the
currently
highlighted
activity
as
well,
so
Darien's
put
a
lot
of
work
into
that.
It
seems
to
be
working
with
a
will.
The
abstract
editor
has
been
a
long-standing
request
from
the
our
brain
folks.
D
A
For
the
state
state
management
work,
Steve
I,
seeing
it
because
I've
been
learning
off
of
master,
so
I've
been
already
incrementally,
seeing
the
benefits
of
that
as
it
landed,
and
it
really
makes
a
big
difference
if
I
was
really
pleasantly
surprised
when
when
I
went
when
I
first
saw
it
happen
that
way.
Oh,
this
is
so
much
nicer,
so
yeah.
A
E
We're
really
happy
to
say
that
last
week,
Jupiter
hub
0.7
went
out
there's
a
link
to
the
change
log
in
the
hack
pad
and
also
messages
went
out
to
the
different
mailing
lists.
So
far
we
have
it's
been
pretty
quiet,
so
people
are
either
probably
still
getting
their
deployments
upgraded
whatnot
and
we
are
working
on
upgrading
the
reference
deployments
and
as
well
as
the
tutorial,
so
probably
in
January,
when
we
do
the
thing
for
the
team.
E
It'll
have
some
of
the
new
API
stuff
and
services
in
that
tutorial
as
well,
and
then
the
other
big
thing
that
we've
been
working
on
this
week
was
supporting
getting
and
bead
I'm
to
beta,
and
it's
actually
looking
really
good.
So
if
people
want
to
try
it
out
for
dipping
and
merging
and
and
and
play
around
with
it,
the
instructions
are
in
the
documentation
and
in
the
repo
and
I
think
the
first
release
will
be
a
minute.
That's
all
I
am.
A
Congrats
on
on
the
release
Carol,
this
is
this
is
wonderful
and
I
would
imagine
I'm
not
too
surprised
with
the
with
the
more
the
quieter
uptake
of
this,
because,
given
that
hot,
this
kind
of
it's
a
deployed
thing
more
than
a
personal
single
user
thing,
there's
always
a
slower
cycle
for
people
test
and
adopted
a
professor
test
instance
somewhere.
So
I'm,
not
too
surprised.
But
it's
great
to
see
this
I
have
a
quick
question
on
envied
I'm.
A
The
have
you
done
anything
on
envied
I'm
kind
of
the
exposing
the
experience
of
envied
I'm
within
hub
or
you
for
now,
just
letting
them
be
completely
independent
and
think
about
that
for
later
kind
of
in
the
I
was
thinking
using
the
services
machinery
how
to
best
bring
that
into
a
smooth
experience
for
the
users.
But
if
you
haven't
thought
about
that,
yet
that
time
is
it
just
kind
of
a
spaghetti
at
the
wall,
question
yeah.
E
E
B
G
A
Nick
quick
question:
where
is
that
is,
is
the
is
your
up?
Is
the
Jupiter
kernels
weep
over?
We
should
point
people
to
this.
There
were
some
folks,
some
folks
in
the
past.
We
had
expressed
interest
and
perhaps
helping
with
this,
and
so
now
that
this
is
beginning
to
take
shape.
I
wanted
to
point
them
to
is:
should
I
point
them
to
the
Dubrow
/
kernels
reef,
oh
yeah,.
B
H
So
lots
of
work
on
alki
widgets,
pushing
towards
a
six-point
x,
release
tons
of
little
things
track
down,
which
is
overhauled,
beta
releases,
etc.
You
can
look
at
the
pulse
on
the
project
to
see
you
know
a
dozen
or
so
pull
requests,
etc.
I
I
think
we're
looking
a
lot
closer
to
I-5
widget
6.
Someone
did
a
lot
of
work
with
getting
witches
to
work
in
documentation,
etc.
I've
been
tracking
down
minor
issues
here
and
there
so
progress
towards
6x
can.
A
H
Thinks
it's
just
stinks
extension
that
I
think
cash
is
the
state
to
the
client
side
and
then
reconstructs,
the
widget
on
the
client
side,
and
so
things
like
the
jas
the
JS
link,
which
it'll
let
you
hook,
which
is
up
on
the
client
side
I
but
I,
but
it
doesn't
involve
a
kernel
state
that
is
equity,
mint.
No.
H
If
they
were
made
in
advance
or
or
not,
my
understanding
is
what
it
does
is
it
takes
the
current
state
and
saves
that
and
then
reconstructs
the
widget
from
the
from
the
save
state.
It
doesn't
cash
like
if
you
do
in
interact
and
you
have
several
different
images
that
are
that
are
generated.
It
doesn't
cash
results
of
computations,
just
the
current
state?
Okay,
that's
my
understanding.
It's
not
the
memorizing
stuff
that
we've
discussed
in
the
past.
That's
right!
That's
correct!.
H
B
I
H
I
Enough,
okay,
sorry
so
Sylvan
open
an
issue
to
represent
the
widget
views
and
states
as
custom
mime
types
in
at
least
the
HTML
exporter
for
NB
convert.
What
is
supposed
to
be
accomplished
with
that,
like
I
tried
to
meet
the
spec
that
he
had
opened
the
issue
with,
but
it
would
help
me
to
know
like
what
the
like
intended.
I
H
I
He's
busy
with
Paris
yesterday
so
I
had
put
in
extra
effort
yesterday
because
it
sounded
like
it
was
important
for
you
guys
to
get
your
like.
Full
testing
like
workflow
done
for
all
of
the
like.
Also
if
you
wanted
I
know
that
this
was
actively
being
worked
on
and
if
you
wanted
to
change
the
API
or
something
like
that,
I
wanted
to
give
you
feedback
immediately.
I
appreciate
that.
H
I
Alright
and
then
I'll
just
continue,
then
be
convert
stuff.
So
Thomas
in
the
notes
had
asked
for
a
five
point:
O
release
I'm
interested
in
doing
that.
However,
some
of
the
things
that
we
have
in
the
pipeline
right
now,
our
notebook
breaking
like
the
changing
in
the
way
that
we
handle
header
IDs,
especially
automated
Hetar
ids,
is,
if
we're
going
to
do
it
properly
like
or
properly
as
in
the
parts
of
the
panda
approach
to
it
that
are
more
or
less
just
ensuring
that
you
have
a
valid
ID.
I
Then
that's
going
to
break
some
people's
links.
Currently,
there
cross
references,
we
also
aren't
doing
a
bunch
of
other
things
and
I
was
kind
of
thinking
of
holding
off
on
a
major
release.
Since
major
releases
are
when
you
we
as
a
project
issue,
breaking
things
until
that
was
all
settled
he's
there
open,
because
that's
probably
too.
I
Issues
open,
okay,
okay,
there's
actually
considering
making
this
into
a
github
project
of
its
own,
like
as
a
different
way
to
approach
as
a
different
way
to
approach
the
github
projects
as
something
other
than
just
release
managers,
but
actually
feature
managers
and
trying
to
put
together
all
of
the
separate
issues
that
are
arising
having
to
do
with
these
header
links
and,
more
generally,
the
ePub
exporter
and
whatnot,
which
isn't
a
release
per
se.
But
is
a
nice
way
to
organize
everything
and
shouldn't
be
given
its
own,
like
label?.
A
Okay,
it
was
like
there's
a
bunch
of
kind
of
detailed
stuff
that
is
two
in
the
weeds
to
discuss
here
in
the
bigger
meeting.
But
but
we
can,
we
can
try
to
work
on
it
on
those
issues.
Maybe
if
you
had,
if
you
could
maybe
add
to
the
env,
convert
section,
a
link,
just
the
numbers
of
the
most
relevant
issues
there,
so
we
have
in
for
reference
and
sure
try
I.
I
Will
make
a
note
for
myself
to
do
that
since
right
now,
I
need
since
right
now
I'm
going
to
continue
to
talk
about
the
other
things,
so
one
of
the
other
questions
about
release
stuff
is
how
good
we
want
be
like
this.
Do
we
want
to
release
a
good
epub
exporter,
or
do
we
want
to
release
an
epub
exporter
that
matches
our
current
DDF
capabilities,
in
which
case
that's
actually
pretty
easy
because
of
things
that
are
missing
are
the
same
things
that
are
missing
in
PDF
I.
A
Would
say
that
I
would
say.
That's
enough.
Just
put
a
note
note
in
the
release
notes
stating
that
fact,
because
we
don't
really
know
how
widely
used
and
popular
or
the
ePub
is
going
to
be.
If
it
turns
out
that
we
got
a
lot
of
feedback
and
there's
a
lot
of
interest
in
it,
then
it's
worth
putting
extra
cycles
into
finishing
and
polishing
it.
If
we
get
the
sound
of
crickets
because
it
turns
out
that
W
here
is
a
body
found,
then
we
don't
need
to
learn
a
lot
of
cycles
into
it.
A
So
I
would
I
would
if
it
matches
what
we
do
on
PDF.
That's
already
useful
and
just
just
add
a
note
in
the
release
notes
saying
these
are
the
current
limitations
and
that's
it
keep
it
keep
an
MVP.
Remember
just
release
early
release,
often
and
and
and
let
let
let
the
user
community
give
us
feedback
on
how
much
work
is
really
warranted
there.
Okay,.
I
A
Follow
semantic
versioning
major
release.
Numbers
are
really
about
about
change,
breaking
about
API
breaking
changes,
so
it
really
it
it's
really
much
more
driven
by
that.
You
could
imagine
there
being
a
fairly
long
time
between
them
or
if
we
realize
that
we
do
have
an
absolutely
burning
need
to
introduce
one
such
change
that
could
come
more
quickly.
Obviously
you
don't
want
to
be
just
for
the
sake
of
a
community.
A
You
don't
want
to
be
introducing
major
number
changes
from
the
semper
perspective
every
three
weeks,
because
that
will
just
drive
everybody
of
the
wall
in
terms
of
instability,
but
but
it's
it
really
should
be
driven
by
the
sin.
Ver
kind
of
mindset,
rather
than
buy,
buy,
buy
buy
time,
I
think
but
photos
of
use.
Others
may
disagree.
I
The
reason
why
in
this
case
is
because
we
could
do
if
I
point
no
release
like
today,
we
could
incorporate
the
epub
into
it
or
we
could
wait
until
all
of
these
issues
are
finished
that
are
going
to
be
breaking
because
we're
going
to
want
to
do
six
point
o
with
that.
If
we
were
going
to
yeah,
that's
more
stray
Lisa
50.
Let's.