►
From YouTube: JupyterLab Dev Meeting, January 20, 2017
Description
Meeting of the JupyterLab development team, January 20, 2017.
Meeting Notes: https://jupyter.hackpad.com/JupyterLabNotebook-Weekly-Meetings-UUJ3gIQ3iBS
A
C
I
went
through
the
notebook
and
widget
hierarchies
and
audited
anything
that
was
a
subcomponent
that
includes
child
widgets,
like
cells
now
put
areas
the
child
models,
which
includes
the
cell
models
and
output
area
models
and
made
them
all
configurable
from
the
top
levels.
So
you
can
inject
your
flavor
of
output
area
from
the
top
level
without
subclassing
the
entire
tree.
C
It
gets
in
this
thing
you
need
so
this
will
help
folks
that
want
to
customize
those
components
moving
forward
and
the
reason
to
do
it
now,
head
of
data
is
trying
to
get
to
minimize
the
amount
of
changes
we
make
after
beta,
as
people
start
adding
things
off
to
tutor
lab,
and
then,
with
that
in
mind,
we
I
mean
on
Monday
to
discuss
the
need,
compatibility
of
being
built
or
whatever
the
artifact
is,
that
comes
out
of
a
juvenile
extension
how
that
gets
rectified
by
the
server
and
then
on
the
client
and
and
what
steps
we
need
to
make
that
work.
C
We
had
naively
made
some
assumptions
about
being
able
to
use
the
highest
matching
December
range,
but
it
turns
out.
We
need
a
better
constraint
than
that.
So
what
we're?
Looking
at
now
is
it
is,
it
looks
like
we're
going
to
emulate
the
no
be
duped
script,
which
is
just
it
hoists
requirements
up
a
tree
and
and
tries
to
minimize
the
duplication.
C
By
doing
so,
I've
looked
at
the
code
that
does
that
I've
talked
to
some
optimization
experts
and
I
think
that's
a
feasible
platform
so
that
special,
we're
exploring
right
now
and
we've
got
the
the
no
December
the
core
of
that
ported
in
Python
and
MIT
library
available
to
us.
So
we
don't
have
to
do
that
part.
It's
just
the
horse
thing
algorithm
of
bringing
bringing
requirements
of
a
tree
at
the
same
manner
that
we
have
to
emulate.
C
That
looks
like
that's,
feasible,
not
a
showstopper
released
version
13
this
week
yesterday,
I
think
so
that
comes
with
some
of
the
state
database.
Improvements
is
really
the
main
user
facing
chains
there.
A
lot
of
its
just
the
backend
works
other
than
that
part
gas.
Part
I've
been
working
on
most
like
that.
The
backend
work
and
adjusting
bugs,
but
during
state
restoration
work
is
hopefully
done.
That's
why
we
put
out
there
see
if
anyone
finds
any
any
bugs
with
that
and
the
so.
C
The
refactoring
in
the
book
beta
addressing
issues
are
really
what
I'm
working
on
outside
of
looking
at
the
lab
extensions
that
abilities
so
trying
to
hit
those
issues
hard
this
week,
the
data
issues
like
being
able
to
deed
encode
the
way
that
the
classic
notebook
was
done.
So
I
just
took
the
coat
mirror
directive
that
command
from
the
classic
notebook
ported
to
type
scripting
with
what
we
went
to
things
like
that
before
so.
A
C
D
A
D
A
Thank
you
a
lot
better
awesome,
ok,
I,
just
love,
it's
clear
that
it
was
both
models
and
use
and
said
cat.
That's
really
really
great.
I
I
know
that's
something
we
had
talked
about
previously
and
we
weren't
quite
sure
how
to
balance
that,
with
our
top
level,
dependency
injection,
that's
great
enemy.
E
A
A
F
Yeah
so
I
think
I
mean.
Obviously
this
is
sort
of
a
blanket
statement
and
almost
certainly
false,
but
everything
I
could
imagine
breaking
state
restoration
is
something
I've
tried
to
address
and
also
wrote
you
could
test
for
it,
and
I
am
hoping
that
the
worst
possible
thing
that
happens
to
somebody
in
say,
restoration
is
a
thing
that
can
reproduce
previous
state.
Like
you
restarted
your
server,
so
calm
folded,
it
exists.
It
makes
an
API
request
that
gets
an
error
and
just
deletes
itself
in
the
database,
but
nothing
nothing
noticeable
to
the
user
happen.
F
Nothing
bad
happens.
Basically.
So
if
you
notice
anything
worse
than
that
happening
and
that's
just
not
preventable,
then
you
know
keep
me
posted
and
I've
been
working
on
the
tooltip,
widget
and
actually
I
kind
of
have
the
opposite
of
the
comments
that
I
had
last
week.
Now
that
I've
been
working
on
it
the
way
I'm
implementing
it
is
basically
with
a
standalone
plugin
that
registers
a
command
that
to
command
sorry
one
command
into
team,
app
shortcuts
one
on
the
console
and
one
on
the
milk
book
and
based
on
that
command.
F
It
just
instantiates
a
tooltip
widget.
It
gets
all
of
the
things
it
needs
from
just
the
editor
and
the
colonel
and
the
render
mine
that
both
notebooks
and
consoles
have
so
those
are
already
public
attributes
and
it
can
totally
independently
of
those
classes
just
run
as
a
standalone,
plugin,
and
so
I
was
talking
this
over
with
Steve
and
Chris
and
and
my
intuition,
and
those
guys
also
agree.
But
we
want
to
be
cautious.
F
So
that's
the
future.
That's
the
path
forward,
because
I'm
not
done
with
the
tool
tip
which
is
but
I
think
if
it
comes
out
as
clean
as
it
looks,
and
it's
not,
it
doesn't
have
any
weird
sort
of
interaction
that
I
haven't
yet
foreseen.
I
think
that
that's
a
good
model
for
the
completer
I,
don't
think
anything
else
is
quite
analogous
in
the
same
way.
So
the
other
thing
you
can
imagine
maybe
wanting
to
do
something
like
this
book
is
the
inspector,
but
the
inspector
is
actually
passive.
F
It's
this
thing
where
you
open
up
and
if
it
happens
to
be
open
as
you're
typing
things
happen
in
there.
So
it's
not
really
analogous
to
all
we'll
just
add
a
keyboard
shortcut
in
the
command,
so
I
think
an
inspector
will
probably
remain
the
way
it
is,
but
I
think.
If
this
works
out,
we
can
have
tool
tips
and
we
can
have
completers
be
standalone
things
and
we
can
remove
something
from
the
abstract
editor.
F
So,
whereas
last
week
I
was
saying,
oh
maybe
we
need
something
for
tooltips
in
the
abstract,
editor
now
I'm
actually
saying
no,
maybe
we
can
pare
it
down
even
more
than
it
is
so
that's
where
I
am
right
now,
hopefully,
next
week,
we'll
have
the
tool
to
put
it
merged
in,
and
people
can
play
with
it
and
see
what
they
think
of
it.
I'm
still
not
convinced
that
everyone
will
love
it,
because
when
we
had
something
like
it
earlier,
people
didn't
love
it,
but
but
we'll
see,
and
that's
at
community
well.
F
B
F
A
A
G
But
I
would
say
it's
in
a
usable
state
at
this
point,
there's
an
unfortunate
amount
of
communication
with
the
server
in
order
to
get
it
to
work,
because
all
the
content
manager
functions
are
based
on
the
path
on
the
server,
which
is
not
how
Google
Drive
represents
a
which
is
not
how
it
represents
the
locations
and
file
the
represents
them
as
unique
file
hashes.
So
you
have
to
do
a
paramedic
painting
back
and
forth
in
order
to
in
order
to
actually
get
a
handle
on
the
right
file.
A
G
So
that
would
be
one
way
of
speeding
this
up
and
reducing
the
amount
of
server
communication
would
be.
That
would
be
to
catch
that
information
somewhere
right
now,
I'm,
not
doing
that,
so
it's
just
there
reconstructing
that
information
every
time
you
need
this,
so
that
that
would
be
one
obvious
way
of
trying
to
cut
down
on
the
amount
of
communication
yeah.
G
D
A
And
then
myself,
I
am
spending
most
of
my
time
teaching
right
now.
I,
it's
also
fun
in
the
sense
that
I
get
to
be
just
a
regular
user
of
all
of
our
tools.
The
students
right
now
are
using
the
classic
notebook
with
jupiter
hub
in
the
main
barrier.
Using
Jupiter
lab
is
that
we
haven't
finished
getting
Altair
vega,
make
a
life
working
in
jupiter
lab
and
grant.
Neptr
has
been
working
on
that
last
week
and
I
think
we're
pretty
close.
Actually
like
there's
a
couple.
A
Different
labels
were
doing
work
in,
and
so
once
that's
done.
I
probably
will
just
update
to
the
latest
near
about
13
Jupiter
lab
and
have
the
students
start
to
use
it
and
I
think
that
the
usability
information
will
get
will
be
useful.
I.
Don't
I
haven't
quite
figured
out
how
to
hoard
that
useful
way
other
than
me
walking
around
class.
What
watching
the
students
use
it,
and
so
I
don't
know
I
may
I
may
have
Cameron
come
to
classroom
and
just
do
some
observations.
A
Carol
also
has
come
to
cross
bones,
and
so
we
may
try
to
do
that.
Part
of
what
I'm
interested
in
is
to
see
I'm
not
going
to
mandate
the
day
use
Jupiter
lab
both
will
be
installed
and
I
want
to
sort
of
see
like
given
the
choice.
So
they've
learned
the
classic
notebook
for
a
couple
weeks
and
given
the
choice
to
use
classic
or
Japan
lab,
what
do
they
do
that?
A
That
I
think
will
sort
of
approximate
our
current
users
sort
of
experiences?
Really
they
already
know.
How
do
you
say
and
want
to
give
him
the
option
to
use
at
the
new
lab
stuff?
Be
what
will
they
will
they
do
that
and
if
not,
what
what's,
the
pain
of
what
was
when
is
preventing
them
from
doing
that,
and
so
grant
grant
has
done
a
great
job
of
going
through
different
custom
mime
types
and
getting
them
working
on
both
notebook
and
lab.
A
So
we
now
have,
obviously
we
had
to
do
Jason
working
and
he
stated
that
we
have
just
raw
application.
Json
we've
got
a
very
nice
bone
for
that.
Their
hearts
plot
me
working
I,
Vega
I'll
tear
a
big
a
light
and
then
he's
been
working
with
various
folks
on
a
new
table.
Json
schema
as
well,
so
it
looks
like
this
I.
A
A
A
A
C
Yes,
we
do
have
the
ability
to
launch
a
notebook
without
a
colonel,
and
then
the
notebook
widget
itself
has
no
notion
of
a
kernel
that
doesn't
come
until
the
no
per
panel
level.
But
we've
we've
yes
carve
that
out
as
well
as
possible.
I
think
an
MP
time
is
using
that
the
notebook
widget
for
cube
okay.
H
C
A
If
we
just
have
a
sort
of
a
native
way
of
rendering
the
notebook
on
interviewer
that
looks
the
same
way,
Jupiter
lab,
and
then
you
know
we
could
do
something
like
have
a
some
sort
of
configuration
file
and
then
be
viewer
that
allows
people
to
configure
which
NPM
packages
and
custom
mime
renders
far
as
round
on
the
given
instance,
and
then
we
could
make
decisions
ourselves
about
which
my
members
would
want
to
hold
by
default.
On
the
you.
A
B
B
notoriously
notorious
for
taking
management
of
the
time
I
am
going
to
move
with
cameron
afterwards,
after
this
meeting
about
him
doing
a
design
run-through
for
the
widgets,
which
I
think
will
help
a
lot
in
some
of
the
polishing
issues
that
we're
dealing
with,
but
we're
hitting
it
pretty
hard
and
things
are
shaking
out
if
we
can
get
if
we
can
get
a
sort
of
stable
beta
in
conjunction
with
the
Jupiter
lab
data.
I
think
that
would
be.
That
would
be
fantastic.
So,
a
month,
okay,.
B
I'm
going
versioning
work
yeah,
the
Jupiter
lab
extension
is
probably
separate
from
the
which
is
separate
from
the
javascript
etc.
So
the
Jupiter
lab
extension
right
now
is
in
oak
point,
whatever
it's
not
even
beta
its
just
oh
point
whatever,
and
it's
understood
it.
It
depends
on
a
alpha
level,
piece
of
software
and
it
has
no
illusions
about
backwards,
compatibility
or
anything,
its
attendant
still
in
the
Oak
Point
stage.
So
so
that's
completely
separate
from
the
rest
of
religious.
B
Thanks
everyone
else,
I'm
really
excited
about
seeing
the
work
that
you
did
and
the
work
that
Darien
you've
been
doing,
I
sort
of
stuff
that,
as
you
could
allow
for
a
little
bit
to
concentrate
a
lot
on
our
pilots.
But
it
as
things
are
settling
down
there.
I'm
pushing
back
into
Jupiter
lab
at
least
make
sure
widgets
are
working
well
and
stuff
and
I'm
in
kind
of
to
see
some
of
the
stuff.
You
guys
have
been
doing
and
Ian.
You
too,
excited
to
see
that
as
soon
as
it
comes
out,
yeah.
A
E
Do
have
a
question
about
some
of
the
work
that
I'm
doing
in
github
project
so
currently
I'm
keeping
like
a
master
calendar
at
the
org
level
in
jupiter,
org,
and
so
just
a
keep
things
simple
for
now,
Jupiter
lab
stuff
will
be
on
there.
I
have
some
contact
at
github
that
I'm
going
to
follow
up
with
to
see
what
the
future
looks
like
for
aggregating
orgs,
because
we
really
do
need
something
that
sits
on
top
of
all
of
our
org.
E
E
It's
late,
isn't
it
you
well.
E
You
see
right
so
you
know
once
we
kind
of
get
like
a
good
cadence
going
I'd
like
to
start
looking
at
how
we
can
start
pulling
in
those
to
github
issues
into
some
sort
of
a
board.
So
it's
easy
to
see
like
what's
on
deck
I
know
we
can
do
it
with
the
filtering
and
github,
but
I
think
visually.
It's
nice
to
see,
like
you
know,
to
do
in
progress
whatever
the
workspace
are,
that
we
want
to
implement.
Just
have
a
good
visible
for
folks.
A
A
Having
the
org
wide
and
so
I'm
just
looking
at
now
so,
but
there's
a
notion
of
a
okay,
so
there's
a
we
have
a
project
management
repo.
Now
that
we're
opening
issues
on
help
track
some
of
the
reserves,
the
project
management
related
things
and
then
zones
are
showing
up
on
the
boards.
Okay,
I'm
just
trying
to
sort
of
understand
and
that's
great
yeah
I-
think
it's
so
like
we
want
to
put
the
IP
I
widgets
60,
no
release
on
that
on.
E
The
project
I
would
do
it
on
the
or
level
the
Jupiter
org
yeah.
So
that's
a
good
question
I,
so
I
one
of
the
to
do
so.
The
project
management
repo
is,
for
you
know,
like
the
meta
meta
work,
that
I
do
as
a
project
manager,
and
you
know
some
of
the
high-level
things
that
we're
trying
to
get
done,
but
doesn't
really
make
sense
to
track
as
an
issue
in
like
a
code
riso.
E
So
that's
what
slipping
in
there
so
because
the
repo
is
there
to
bring
visibility
to
the
project
management
work
and
to
kind
of
go
it
just
a
step
higher
and
granularity
to
get
bigger
picture
concepts
with
the
work
across
the
body
of
the
work,
so
not
just
the
dev
work,
but
trademarks
and
community
development.
You
know
everything
that's
kind
of
on
my
plate
and
then
I'm
monitoring
at
you
know
at
some
level
what
I
would
like
to
do
is
for
the
development
work
have
boards
where
we're.
E
You
know
this
is
you
know
to
help
soak
on
the
team
here
see
what's
going
on
with
the
project
that
also
helped
a
larger
community,
see
where
they
can
step
in
you
know
we
want
to
develop
that
like
contributor
pipeline
or
funnel
or
whatever
we
want
to
call
it
I'm,
just
going
to
throw
them
a
bone
and
see
like
here's
all
the
stuff
that
we're
working
on.
These
are
the
things
that
you
can.
You
know
help
us
out
with.
A
Great
so
yeah
so
yeah,
just
just
let
us
know
so
as
you
explore
this
and
want
to
try
things
out
with
us
with
the
Jupiter
live
team
like
I
would
love
to
help
in
any
way
possible.
I
think
it's
really
really
helpful,
and
especially
for
games
like
this,
that
manage
a
number
of
repose
board
sort
of
a
central
goal.
I
think
it
I
think
it
helps
to
have
a
higher
level
of
sort
of
the
metal
layer
to
help
manage
what's
going
on
and
get
a
sense,
both
are
selves
and
also
oddly.
E
So
whether
it's
a
you
know,
category,
you
know
type
of
issue
or
a
category
of
work.
You
know
we
could
have
boards
for
that.
But
right
now
it's
fairly
manual-
and
you
know
github
building
this
out
more
and
more,
and
you
know
I'm
trying
to
kind
of
be
on
the
edge
of
their
development
and
also
be
willing
to
be
a
guinea
pig
for
them
trying
to
get
stuff.
So
the.
E
E
E
E
H
Still
pushing
forward
on
the
repackaging
a
phosphor
I've
done
in
the
last
sub
package
now,
which
is
the
widget
package.
Anybody
following
along
when
the
feature
sub
modules
slightly
miss
named
branch
to
be
featured
some
packages
or
whatever,
but
you're
following
along
with
that
branch.
The
last
bit
to
do
is
move
widgets
over
that's
going
well.
H
The
one
I
just
finished
up,
which
is
probably
the
biggest
change
in
terms
of
API
consumers,
is
virtual
Dom
module
which
I
broke
out
doing
this
work,
I've
been
able
strict
null
checks
in
the
tetra
compiler,
so
which
is
actually
a
really
big
boon
to
correctness
and
productivity.
So
now
you
don't
have
to
guess
whether
some
of
these
national
parameter
or
not
the
compiler
guarantees
that
they
don't
unless
you
actually
want
them
to
and
then
forces
you
to
check
if
it's
know,
if
you
allow
them
to
pass
another
argument
in,
but
in
the
process.
H
Turning
that
in
turning
that
on
for
virtual
Dom
stuff,
it
had
me
going
and
correcting
a
lot
of
things
which
had
me
go
back
and
look
at
some
of
the
type
definitions
that
pulled
in
based
on
the
IDL
spec.
For
various
HTML
Dom
nodes
and
in
the
time
frame
since
I,
originally
wrote
the
virtual
Dom
and
now
that
stuck
has
changed,
and
so
some
of
the
types
that
I
had
written
are
now
incorrect.
H
With
respect
to
like,
if
you
create
a
and
with
respect
to
what
kind
of
attributes
you
can
pass
when
you're
creating
a
Dom
element
of
a
particular
type.
So
before,
like
if
you're
just
create
like
a
div
element,
the
attributes
that
you
pass
it
would
be
tight
check
based
on
what
the
loud
for
a
div.
Now.
H
So
it
basically
creates
it
makes
the
problem
in
terms
of
making
sure
we're
following
respect
a
linear
problem
rather
than
a
quadratic
problem
in
terms
of
to
resolve
attributes
against
the
type
of
elements
that
are
valid
for
and
that
kind
of
stuff.
So
there's
a
bit
of
an
API
change
there,
but
it
is
removed
like
1300
lines
of
code
from
virtual
realms.
So
that's
much
less
maintenance
on
our
end
and
if
people
discover
we're
missing
something
and
the
spec
has
changed
or
anything
like
that,
it's
much
easier
now
correct.
H
You
just
add
a
removed
from
a
list
of
defined
attribute
names
rather
than
have
to
track
down.
You
know:
12
different
interfaces,
plus
a
load
table
on
the
renderer
and
all
sorts
of
other
stuff.
So
it's
simpler
to
maintain
simpler
to
drop,
have
the
cost
of
a
bit
left
bit
less
by
KP,
but
I
felt
it
was
a
pretty
good
trade-off.
Mom
so
made
that
trade
off
again,
that's
in
the
feature
sub
module
branch.
H
If
anybody
wants
to
know
just
exactly
what
I'm
talking
about
and
talking
about
a
further
offline,
we
can
do
that,
but
other
than
that
I
expect
next
week
to
have
one
dunno
versions
of
phosphor.
Out
and
again
these
these
new
packages
are
going
to
be
under
the
Act
sauce
for
namespace,
so
adopt
them
as
you
will.
The
old
software
packages
are
still
airing
in
p.m.
so
it's
not
going
to
magically
break
your
build.
If
you
update
or
anything
like
that,
you'll
have
to
opt
into
movie
stuff.
Okay,.