►
From YouTube: JupyterLab Weekly Dev Meeting, May 26, 2017
Description
Meeting of the JupyterLab & Notebook development team, May 26, 2017
Meeting Notes: https://paper.dropbox.com/doc/JupyterLabNotebook-Weekly-Meetings-rIhXeFWYRgCiCFKiz2gv4
A
Now
those
in
Florida
those
discrete
as
well
as
design
pass
that
binds
working
on
to
include
breaking
out
the
CSS
in
such
a
way
that,
because
now
support
theming
your
we
have
a
working
electron,
app
demo
at
least
the
start
of
one,
so
the
it
loads.
The
application
on
the
pages
using
some
manual
configurations
to
start
and
Darian
is
pretty
close
to
having
working
implementation
of
the
setting
system
right
now,
using
local
storage
in
the
browser,
but
follow-on
to
that
will
be
actually
using
files
on
disk
to
back
that
up.
A
A
B
You
guys
hear
me
good
morning:
alright,
I'm
just
waking
up
still
so
one
as
I'm
doing
the
cleanup,
the
ui/ux
review
and
finding
so
I.
Have
this
other
PR
open
where
I
started
to
create
a
standardized
set
of
UI
controls,
button
dropped
off,
shut
down,
etc.
B
I'm,
finding
myself
spending
a
lot
of
time,
sort
of
micro
tweaking
all
the
different
buttons
and
dropdowns
we
have
styled
so
far,
so
we
have
some.
We
have
this,
this
styling
namespace
that
does
some
of
this,
but
still
pretty
painful.
B
Our
CSS
is
is
it
covers
the
case,
certain
cases
well,
but
there's
a
lot
of
fine
tweaks,
I'm
having
to
do
and
and
so
I
don't
I
didn't
I'm
I'm
wondering
if
we
want
to
go
ahead
and
try
to
do
the
just
a
few,
the
UI
controls
for
the
beta
and,
if
that's
going,
to
help
the
process
to
standardize
that
stuff
quickly
or
if
it's
going
to
help
to
not
do
it
I'm
sort
of
on
the
fence
originally
I
was
thinking.
B
Let's
definitely
not
do
it,
for
the
beta
and
live
is
what
we
have,
but
that
you
know
him
styling
and
styling.
A
drop
down
is
pretty
painful
and
slows
down
the
the
process.
Quite
a
bit,
I,
don't
know
I
again,
I'm,
not
I
was
wanting
to
avoid
doing
that
for
the
beta
that
I'm
after
spending
way
too
long
doing
that
last
night
on
one
I'm
having
guiding
thoughts.
So
I
don't
know
any
any
thoughts
on
that.
B
B
We
tend
to
have
no
borders
on
the
dropdowns,
although
in
this
particular
case
that
it
says
the
limiter
doth
drop
down,
and
so
the
text
and
the
drop
down
is
really
minimal
and
it
looked
weird
without
a
border
I
don't
know
I,
it's
just
a
matter
of
sort
of
capturing
the
different
cases
we
wouldn't
want
to
handle
capturing
size
assumptions
and
then
I
think
part
of
it.
I
think
are
the
the
code
base
for
things
like
the
toolbar
will
be
cleaned
up
dramatically.
B
If
we
make
the
toolbar
itself,
one
of
these,
the
beat
on
blender
objects
that
you
know
we
can
literally
put
all
the
logic
in
a
single
model
and
put
all
the
logic
in
the
rhetoric.
Logic
in
a
single
render
method
where
it
basically
just
calls
all
factory
functions
that
create
V,
Dom
nodes
and
I
think
that
that
type
of
thing
is
going
to
be
simplified
greatly,
but
again
I'm
trying
to
avoid
not
doing
the
API
cleanup
I
was
more
stumbling
on
the
just
visual
design,
consistency,
questions
and
I
know.
B
B
C
B
B
B
I
will
try
not
to
dive
into
that
much
so
a
few
other
things
that
I'm
running
into
on
the
ui/ux
review.
One
is
the
FAQ
and
be
about
page.
Both
are
out
of
date
right
now,
my
so
for
those
of
us
who
hasn't
looked
at
for
a
while
and
they
cues
a
bunch
of
texts
that
basically
answers
questions
about
your
lab.
The
about
page
is
more
of
a
visual
tour.
B
The
both
haven't
been
updated
in
about
a
year
and
I'm
my
fault
with
the
about
pages.
That
will
be
because
it's
all
visual,
it
would
be
quite
a
bit
of
work
to
redo
and
some
I
might
not
right
now
is
to
combine
get
rid
of
missing
about
page
and
then
update
the
FAQ
so
that
it
has
the
content
in
some
appropriate
form.
So
have
the
version
there
have
information
about
where
people
should
file
bugs.
B
If
they
have
questions,
they
want
to
write
an
extension
where's
our
documentation,
basically,
everything
you
need
to
know
about
it,
otherwise
I
think
we're
going
to
have
to
do
a
full
and
time-consuming
design
pass
on
the
about
page
which
I
I.
Just
don't
think
it's
worth
our
time
on
that
right
now,
I
think
we're
about
off
spent
on
polishing
the
actual
user
interface
or
the
different
packages.
So
any
thoughts
on
that
I.
B
Okay,
any
other
thoughts
on
that
one
and
then
the
the
other
is
the
launch
or
landing
experience
and
could
be.
Our
long-term
plan
is
to
combine
the
log
launcher
and
landing
experience
so
that
there's
some
plus
button
that
the
user
clicks
that
opens
a
panel
in
the
dock
panel
that
has
the
different
activities
and
then
wanting
you
to
click
on
an
activity
that
panel
gets
replaced
by
that
actual
activity.
So
the
user
click
on
new
notebook
in
the
panel
that
panel
turns
into
that
notebook.
B
So
they
can
move
the
panel
around
where
they
want
the
notebook
to
be
right
up,
and
so
we
have
different
pieces
of
this.
But
we're
not
wanting
to
do
the
full
refactor
for
the
beta.
So
we
have
an
actual
launcher,
API
the
launcher.
Currently,
those
in
the
size
annal.
We
have
the
landing
page,
that's
more
the
what
people
first
see
and
wondering
if
we
can,
we
clean
this
just
a
little
closer
to
where
we're
going
to
end
up
and
the
main
question.
B
B
Where
would
we
put
a
plus
button
that
opens
the
landing
page,
the
banner
part
of
it,
and
it
means
that
one
option
is
to
just
do
the
plus
button
up
in
the
capital
file
browser
where
our
existing
plus
button
is
we'd
need
to
pull
out
the
new
folder
button
there,
because
it's
not
something
that
moves
in
the
doc
panel
or
we
guy
was
talking
with
Steve
about
you,
know,
work
where
elsewhere
else
in
the
UI.
Could
we
put
such
a
plus?
B
B
B
B
D
That
is
the
like
when
I
the
thing
that
I
create
new
of
every
day,
the
most
up
or
browser
tabs
and
text
files
since
abandoned
and
in
sublime
text,
there's
a
file
new
menu
item,
but
I
never
use
it.
I
always
use
the
keyboard
shortcuts
in
Chrome
I'm,
probably
50/50,
between
using
the
plus
button
next
to
tabs
and
general
t3
grabs,
but
Chrome's
are
restricting
Eli
where
you've
always
got
this
tap
border
already
visible.
D
There's
no
single
document,
though
there's
no
other
like
it
makes
sense
for
them
to
have
a
plus
thing
there
you
could
have.
If
there's
nothing
in
the
dot
panel,
you
can
just
have
a
big
Center
nicely
styled
plus
button,
but
just
a
create
a
new
thing
in
this
empty
area,
and
then,
once
you
have
stuff
in
the
dot
panel,
we
can
add
maybe
a
plus
icon
to
as
a
custom
tab
bar
thing
so
that
you
can
create
the
new
item
in
any
of
the
tab
areas.
It
doesn't
really
help
us
with
single
document
mode.
C
Single
button
does,
maybe
item
is
fine.
The
nice
thing
about
having
a
plus
bar,
but
in
each
tab
bar,
even
though
it
may
be
visually
lettering,
and
they
do
probably
turn
down
the
opacity
or
something
it
gives
you
the
locality
that
you
need.
So
if
you
have
four
different
panels,
yeah,
it's
very
easy
to
create
a
new
thing
and
particular
location,
and
it's
it's
pretty
clear
where
that
new
thing
should
go
or
idea
to
go
at
the
very
end
of
that
tab.
Panel,
I
suppose
at
that
particular
location,.
D
D
Right
now,
the
tab
bar
itself
is
mostly
invisible
and
only
the
attacks
have
color
to
them.
But
if,
if
you
put
a
button
there,
it's
probably
going
to
end
up
for
various
reasons
at
the
right
edge
of
the
the
ten
horns,
they
probably
want
some
sort
of
background
color
on
the
top
bars
to
distinguish
the
top
bar.
D
For
that
tap
button
that
we
want
to
put
the
logic
that
the
knowledge
of
a
plus
button
that's
far
down
into
the
stack,
and
so
the
alternative
would
be
to
offset
media
and
move
that
not
have
that
plus
button
in
the
same
content
known
as
the
taps,
and
so
the
most
logical
place
word
at.
That
point
is
to
the
right
edge
of
the
tab
bar
at
which
point
a
ddos,
our
background
purchased.
So
this
is
an
aside.
D
This
is,
if
you,
if
you
remember
the
original
way,
that
I
had
styled
the
tabs
and
the
tab
bar
in
the
examples
that
I
put
out
there
was
a
background
color
from
the
tab
bar
that
offset
it
from
the
rest
of
the
content.
That
is
why,
because
I
was
expecting
you're
going
to
have
a
plus
button
on
the
top
bar
point,
I
mean.
D
Generally
delayed
actions
like
that,
where
it's
like
I've
got
to
go
here
and
then
wait
for
something
else
to
happen
in
half
again:
I,
don't
like
those
types
of
interfaces.
One
thing
that
just
pops
into
my
mind,
though,
is
that
maybe,
instead
of
button,
what
about
just
a
right-click
menu
on
the
tab
bar
new
item
here
like
it's,
not
as
discoverable,
but
once
you
don't
know
about
it,
it's
they've
got
less
visual,
clutter,
yeah.
B
I
mean
that
it
is
the
probability
that
I
mean
that
yeah,
like
that.
One
of
the
most
common
thing
users
will
do
is
how
do
I
create
something
new
and
I
want
that
to
be
extremely
obvious
and
I
mean
that,
like
from
a
from
a
visual
design
perspective,
the
clean
the
Queen
in
the
cleanest,
most
discoverable
and
sort
of
simple
way
of
doing,
this
would
be
to
do
a
circular.
B
B
B
B
Pull
that
out
of
the
separate
icon,
okay,
Cameron
already
did
the
designs,
a
nice
new
new
new.
We
call
it
new
folder
by
open
folder,
with
a
little
overlaying
plus
on
it.
C
D
I
was
going
to
say,
I
mean
that's
not
that
could
still
be
done.
This
launcher
pad
widget
thing,
whatever
we
we
decide,
couldn't
take,
but
an
option
to
be
work.
The
director
create
new
items,
and
so,
if
you
open
this,
if
you
watch
this
thing
from
a
file
browser,
then
the
file
browser
is
going
to
you
know
took
that
option
to
whatever
it's
working
current
directory
is,
but
maybe,
if
you
watch
this
tab
from
the
menu
bar,
then
it's
going
to
use
the
server
root,
festa
directory
and.
B
I
think
that
that
was
our
original
idea
about
how
to
combine
the
launcher
idea
with
the
the
current
working
directory
notion
is
to
run
either
rather
than
have
the
file
browser
expose
of
public
current
working
directory
for
this
purpose.
Basically,
have
it
pack
like
if
it
is
the
one
that
is
triggering
the
launcher
to
open,
have
it
just
past
the
current
working
directory
into
that
launcher?
At
that
time,.
A
B
Another
way
to
thinking
about
this
is,
for
example,
on
the
Mac.
There
are
applications
are
represented
as
files
and
so
to
start
the
neck
tivity
on
the
Mac,
even
one
that
knows
nothing
about
the
file
system.
I
go
to
a
particular
folder
and
click
on
Spotify,
or
you
know
something
like
that,
and
so
there's
there
are
places
and
other
operating
systems
where
non
document
related
things
are
still
treated
and
started
in
the
same
way,
the
document
related
things
are
but
like
I,
think
yeah.
It's
a
question
of
how
do
you.
C
A
C
C
D
Yeah
I
like
that,
just
because
it's
going
to
be
familiar
for
everyone
that
uses
the
browser
ever.
It
would
be
helpful
in
this
case.
If
I
don't
know
what
how
much
time
Cameron
has
available,
but
if
we
can
think
about
what
we
want
that
to
be
in
kind
of
toss
around
the
pros
and
cons
the
various
ways
that
we
could
make
that
with
respect
to
whether
I
need
to
do
stuff
and
phosphor
just
support
it.
The
sooner
we
can
do
that,
the
better
okay.
B
What
would
folks
be?
Okay
with
that
button?
Only
appearing
1d
when
that
panel
is
active
I'm,
just
the
only
doubt,
I
mean
I
that
that
approach
has
a
lot
of
pros
that
I
really
like
you
know
it
has
the
locality,
information,
it's
very
accessible,
it
I,
don't
know
it's
a
model
that
people
are
used
to
from
other
applications.
B
The
only
problem
is,
if
you
have
a
lot
of
panels,
there's
just
a
lot
of
plus
buttons
to
look
at
and
the
the
more
discoverable
we
make
them
the
more
knowing
it
will
be
that
there's
many
of
them,
that's
a
problem,
and
so
some
some
way
to
subdue
or
I
mean.
Maybe
we
we,
we
I,
don't
know
that
that
that's
my
only
concern
about
that
approach,
even
though
I
may
be.
You.
C
Can
make
it
so
I
noticed
that
chrome,
for
example,
does
not
have
a
plus
sign.
It
has
a
has
a
shape,
that's
visually
similar
to
the
other
tab
shapes,
and
so
it
doesn't
stand
out
nearly
as
much
yeah.
So
there's
other
ways
to
make
it
more
subtle.
To
me.
I
think
it
would
be
annoying
if
I'm
trying
to
set
up
tabs
and
different
tab
panels
to
have
to
click
on
that
tab
panel
to
activate
something
in
that
tab
panel
before
it's
adding
clicking
up
less.
A
C
B
Yeah,
that
might
be
just
simple
enough
and
through
the
job,
okay,
yeah,
let
me
I
mean
I'll
go
to
driving
into
campus
here
in
a
few
minutes
and
Cameron
I
think
is
working.
I
can
have
him
walk
up
some
of
this.
What
I
think
that
camera
gonna
be
there
next
week?
No
he's
still
got
a
class.
B
Yeah
and
then
one
one
other
question
right
now.
The
help
extension
can
open
iframes
in
the
doc
panel
I'm
wondering
if
it
makes
sense
for
us
to
have
an
actual
iframe
extension
where
you
can
pass
at
a
URL
and
it
will
open
it
and
the
other
thing
I'm
thinking
of
there
is
that
would
enable
us
to
have
a
a
local
file
handler
for
HTML
file,
editing
to
open
them
in
an
iframe
and
render
the
HTML
like
web
development.
A
B
Actually,
I
think
the
way
I
would
do
it
is
have
the
low-level
iframe
extension
just
pass
the
URL
as
an
argument
to
the
command
and
then
have
other
parts
of
the
UI
either
like
the
help
system
is
going
to
already
know
the
URLs.
The
registry
is
going
to
hardly
know
the
URLs,
but
if
someone
else
needs
to
have
you
pasted
over,
you
can
do
that.
So,
basically,
those
other
parts
of
the
UI
to
offer
things
like
that
all
right,
Aysen
any
II,
want
to
cover
anything
on
my
PI
widgets,
steady.
C
Think
it's
just
a
bunch
of
cleanup
stuff,
a
few
tweaks
here
and
there
is
concentrating
on
testing
right
now.
The
three
TRS
that
are
in
in
progress
are
testing
the
widget
page
code
and
making
it
an
output
legit
for
the
embedded,
widget
manager,
which
you
mean
that
it
can
be
converted
and
the
Jupiter
sinks
can
vendor
the
interact
command
and
as
well
as
acidity.
I
am
a.
C
C
Hey
guys,
so
no
I
don't
have
any
updates
on
my
end,
I've
just
been
caught
up
as
another
thing,
the
past
two
weeks,
so,
okay
I'm,
looking
getting
back
into
it
this
coming
week,
yeah.
B
New
view
there
at
the
dabney
definitely
okay,
yeah
one.
One
thing
we
should
discuss
at
the
dev
meeting
and
figure
out
is
now
that
if
the
extension
stuff
is
settled
down,
what's
the
best
way
for
us
to
organize
these
other
extensions
that
you've
been
maintaining
it
even
hey,
I,
don't
know
what
we
want
to
do
it,
but
it
even
occurred
to
me
that
now
that
the
lab
expansions
are
just
NPM
packages,
managing
them
all
together
and
like
a
get
a
single
learner,
man
repo
would
help,
but
I
would
really
help
me.
C
Just
managing
them
in
the
future
yeah
I've
been
I
haven't,
had
it
like
a
dedicated
couple
days
to
work
on
it,
but
I've
been
thinking
about
it
and
yeah
you're
right,
like
the
fact
that
tutor
lab
is
managing
extensions
differently
now,
as
NPM
packages
changes
things,
and
it
makes
it
much
simpler
to
just
manage
Jupiter
lab
extensions,
but
since
the
nipple
book
still
needs
to
don't
need
to
be
in
a
Python
package,
form
yeah,
it's
just
it's
just
kind
of
like
we
got
to
make
some
design
decisions
and
we
should
just.
C
A
C
B
And
then
there's
also
the
the
other
question
of
are
these
ones
that
were
maintaining
common
enough
or
are
some
subset
of
them
coming
up
that
we
literally
must
adjust
bundle
them
in
this
lab
and
classic
notebook.
That's
something
like
that.
Kyle
brought
up
and
that
I
think
again,
especially
something
like
the
JSON
mine
render.
That's
it's
difficult
to
imagine
people
not
wanting
that
installed.
C
C
We
have
a
poly
one,
that's
kind
of
kind
of
like
a
toy,
render
extension
that
we
could
totally
point
people
to,
but
I
grew
through
and
I
think
the
JSON
one
should
be
bundled
it's
just
a
better
way
to
it's
better
way
to
view
the
data
and
do
JSON
seems
like
that's
probably
common
enough
as
well.
Let
me
keep
the
Geo
JSON
one
out,
I
mean
it's
so
simple,
like
it
only
handles
points
and
the
Geo
JSON
file,
etc.
C
It's
so
simple
that
I
think
if
we
were
going
to
fund
the
one
we
should
make
get
handle
a
lot
more
to
enjoys
on
stretchers,
so
polygons
and
whatnot.
It
helps
the
ads
is
I.
Think
it's
more
of
an
example
starting
from
for
I/o,
and
that
means
I
get
I
see.
Okay,
so
are
you
saying
that
the
leaflet
library,
just
out
of
the
box,
doesn't
support
things
like
that?
Oh
no
leaflet
supports
a
great
that
geo.
Json
plugin
is
written
in
like
two
hours
and
I
think
the
price
date
the
only
thing
ever.