►
From YouTube: JupyterLab Dev Meeting, April 13, 2017
Description
Meeting of the JupyterLab development team, April 13, 2017
Meeting Notes: https://paper.dropbox.com/doc/JupyterLabNotebook-Weekly-Meetings-rIhXeFWYRgCiCFKiz2gv4
Meeting Note Archive: https://jupyter.hackpad.com/JupyterLabNotebook-Weekly-Meetings-UUJ3gIQ3iBS
A
A
B
Yeah,
so
we
are
introducing
the
single
document
mode
and
the
multiple
document,
tolerable,
basically
from
the
doc
panel,
so
phosphor
had
this
added
a
few
weeks
ago
and
now
we're
introducing
it
into
Jupiter,
lab
and
I
guess.
The
best
thing
to
do
still
really
sure
screen
share
right
now,
as
the
PR
is
still
outstanding.
So
you
can't
check
it
out
until
it's
merged,
but
this
is
the
current
state
of
it
and
then
we
can
probably
iterate
from
there.
Oh.
B
B
B
B
Okay
right
so,
oh,
it
gets
better,
save
it,
but
anyway,
so
you
can
see
the
tab
bar
just
disappeared
on
top
and
there's
this
new
side
bar
tab
called
tabs.
That
shows
the
tags
that
were
open
so
that
you
could
switch
those
to
be
the
single
document
that
you're
interested
in.
So
where
this
is
most
compelling
is
say,
you
have
a
notebook
and.
B
Everything's
collapsed.
This
is
a
much
more
familiar
view
to
users
of
the
classic
notebook,
because
there's
no
tab,
Chrome
or
any
of
the
dock
panel
layout
that
might
be
confusing
for
users,
there's
still
the
left-hand
sidebar
because
well,
first
off
there's
no
way
out
without
it,
and,
secondly,
because
we
do
support
all
the
things
it's
it's
just
that
they're,
not
all
showing.
So
you
know
this
is
fairly
straightforward.
For
now
it's
pretty
simple.
We
we
might
want
to
make
some
changes
to
have
that
interaction
with
the
tab.
B
B
A
So
yeah,
okay,
would
you
expect?
And
now,
if
you'd
like
I
can
output,
you
know
to
do
something
inside
the
notebook
and
so
that
you
can
drag
an
output,
okay,
just
type
one
or
two
or
whatever.
What
plus
one
yeah
and
now
drag
that
output
right?
Remember
we
can
I
thought
we
could
drag
output,
yeah,
okay
and
then
we
use
this
and
ruined
you
yeah.
Well
so
I
mean
now
you
got
your
dashboard
right.
A
A
A
B
Actually,
the
money,
so
I
have
a
problem
because
that
keyboard
shortcut
is
one
that
I
can't
override.
So
I
need
to
recompile
the
app
and
the
keyboard
shortcuts
that
I
have
right.
Now
that
works
is
Excel
shift,
enter
or
command
shift
enter
that
works
but
enters
maybe
non-intuitive
for
people.
We
thought
about
command
shift
T
for
the
word
toggle
that
the
trouble
with
that
is
that
in
Chrome
that
doesn't
seem
to
be
a
shortcut
you
can
override.
So
we
might
want
to
think
about
what's
a
good
shortcut
when.
A
A
A
A
D
That's
something
we're
going
to
have
to
address
handle
on
your
left
side
because
you
can
add
new
stuff,
because
it's
in
the
back
of
the
boat
and
the
question,
the
other
should
yeah
the
way
that
it's
the
way
that
single
document
mode
is
implemented
internally
in
the
dock
panel.
Is
it
just
takes
all
the
widgets
that
are
in
the
dock
panel
and
restores
a
layout
that
is
nothing
but
a
single
pads
area
and
then
hides
that
pan
bar
and
then
anything
that
you
continue
to
add.
D
It
just
ignores
whatever
you
specify
and
add
into
that
same
temper,
and
so
in
that
case,
if
you're
wanting
to
restore
what
you
try
to
restore
some
kind
of
layout,
that
was
there
beforehand
before
you
put
it
into
single
document
boat.
You
would
that
life
out
and
then,
when
you
try
to
toggle
it
back,
you
figure
out
what
widget
you
have
in
one's
reviewing,
what
one's
a
bit
removed
and
put
the
new
ones
someplace
reasonable.
A
B
B
I
can
imagine
basically
here
where
the
command
is
implemented,
the
set
mode
command
that
actually
changes.
The
mode
I
could
imagine
here
having
some
private
collection
that
saves
the
layout
from
before
keeps
track
of
the
newly
added
things
and
basically
what
Chris
described
and
if
that
exists
it
does
it.
If
not,
it
just
runs
the
command,
as
is
right
now,
but
that's
not
currently
there,
no
cool,
I
really.
A
A
B
Because
I
would
really
like
for
some
some
design
that
goes
to
be
put
on
the
interaction
with
regard
to
the
sidebar.
The
tab
was
just
basically
because
it's
not
a
it
could
be
subtle,
there's
a
lot
of
ways
we
could
handle
it.
So
I
would
like
some
people
to
think
about
it,
and
the
easiest
way
is
just
put
it
out
there:
cool
thanks.
C
Okay,
I
guess
I'm.
Next
look
a
couple
days
ago,
we
finally
decided
to
make
things
as
easy
as
possible
on
the
extension
mechanism,
concerns
of
using
standard,
mpm,
workflow
standard
mpm
tools
not
trying
to
hack
around
anything
to
make
it
work.
So
what
we
have
today
in
the
current
PR
is
you're
using
an
MPN
pack,
MPM
install
NPM
the
link
standard,
standard
command
line
tools
from
p.m.
C
So
we
we
preserve
the
existing
commands
of
Jupiter
lab
Jupiter,
lab
extension
installed,
jable
of
extension,
uninstall
google
I've
extension
list
enabling
and
disabling
is
now.
There
is
no.
That
notion
goes
away
the
way
you
would
disable.
You
can
disable
individual
plugins
bye-bye
right
now
manually,
adding
them
to
the
ignored
plugins
page
config,
so
those
will
get
ignored
by
token
name
we
can
from
within
the
app
told
you
I,
that
with
all
the
available
tokens
and
whether
they're
explicit
ignored
and
that's
where
you
could
enable
disable.
C
So
that's
a
that's
where
we,
without
without
rebuilding
anything
we
can
have
the
disable
enable
feature
right
now.
Install
or
uninstall
also
builds
the
application
and
later
we
could
make
it
so
that
that
they're,
not
necessarily
linked
but
you'd,
have
to
if
its
sale
you'd
have
to
the
server
extension
for
lab
would
have
to
see
that
build
on
the
fly,
inform
the
user
that
it's
building
show
spinning
bar
on
the
page
and
then
load
itself,
but
we
deferred
that
to
just
say
just
do
it
all
at
once.
C
What
is
not
yet
automated
is
the
building
against
your
burlap
master,
so
that
that
link
workflow
is
only
for
building
against
a
release.
Super
lab
at
this
point,
so
it
has
to
be.
You
have
to
be
on
a
tagged
release
to
build
again
super
lab.
We
there
it
is
technically
feasible
to
build
against
at
the
master.
Just
we
haven't
implemented
that
part.
Yet
so
I
did
prove
that
out
manually,
just
we
don't
have
a
command
for
yet
I
think
I've
covered
it
all
what's
missing.
C
So
from
an
extension
authors
standpoint,
the
biggest
changes
are
an
extension
is
just
an
NPM
package
that
you
hand
over,
and
it
has
to
be
the
if
you
run
NPM
pack
on
it.
That
should
contain
all
the
artifacts
that
it
needs.
So
if
it's
depending
on
a
private
package,
you
can
you
have
to
bundle
that
thing
or
you
can
empty,
you
could
install
a
live
extension
from
anywhere.
That
is
a
valid
NPM
install
target,
so
you
can
install
from
a
get
URL
from
NPM
registry
itself.
C
What
I
just
said
has
to
be
written
down
I'm.
Some
of
it
is
in
the
top
level
comment,
but
yeah
I'd
like
to
get
documentation
in
there,
because
we're
throwing
out
a
bunch
of
documentation
that
was
an
extension
builder,
all
right.
The
four
thousand
lines,
whatever
foot
that
we're
moving
in
this
PR,
so
yeah
I'm
done
making
changes
to
the
workflow
with
the
only
thing
we
haven't
thrown
out
any
capabilities
and
that
we
have
install
uninstall.
We
have
a
way
to
enable
and
disable,
and
we
have
a
way
to
link
the
only.
C
C
So
both
the
building
against
released
artifacts
and
against
master
Richard,
the
lab
goj
song
and
I-
had
to
change
it
to
not
you,
because
it
was
using
a
private
sibling
package
tribes
a
little
bit
to
make
that
work.
Otherwise,
nothing
major
had
to
change
their
it.
All.
The
other
part
is
where
it
was
using
a
web
pack
loader
to
translate
for
me
as
six
using
Babel.
Instead,
the
build
step
now
produces
es
5.com
KS,
so
the
extension
must
provide
an
NPM
package
that
is
common
Jas.
C
Well,
is
is
consumable
by
web
pack
and
not
es6,
because
neither
support
that
and
not
and
only
include
the
file
types
that
we
have
loaders
for
and
we
can
and
that
list
will
be
part
of
the
documentation,
so
CSS
certain
image
files.
If
you
don't,
if
something
is
not
in
that
format,
you
have
to
either
exclude
it
or
put
it
into
that
format,
or
also
the
final.
C
What
Peck
told
won't
go
through
because
there
is
no
such
thing
as
a
as
far
as
I
could
tell
a
fallback
webpack
plug-in
where
if
it
doesn't
recognize
a
file
it'll
there,
you
could
use
the
file
rotor
or
something
I'll.
Look
again
on
that,
because
that
would
be
nice
to
say.
You
know:
I
recognize
the
style
type
I'm
just
going
to
load
as
a
file,
but
I
don't
yet
see
a
way
to
do
that,
because
that
would
be
a
nice
thing
to
have.
A
A
On
widgets
I
think
we're
halfway
done.
We've
got
21
close
issues
and
21
open
issues
with
for
fi,
which
is
seven
so
we're
almost.
There
got
30-some
odd,
full
requests
and,
as
for
i586
7,
the
main
things
of
note
words
that
we
fixed
the
long-standing
syncing
issue
that
we
had
basically
up
to
this
point.
If
we
couldn't
think
widget
state
right
away,
we
would
save
the
widget
state
to
sink
when
if
we
throttle
messages,
if
there
was
already
a
message
in
flight,
we
would
not
sink.
A
Another
message
would
not
extend
another
sink
message:
instead,
we'd
hold
on
to
it
until
the
last
ink
message
had
been
processed,
but
the
problem
was
is
that
we
didn't
take
a
snapshot
of
the
widget
state.
At
that
point
we
just
said
in
the
future.
We
will
ask
you
again
what
your
values
are
when
we
actually
send
the
sink
message.
A
Think
I
just
finished
it
just
like
two
seconds
before
the
call
we
finally
change
widgets
in
the
classic
notebook
to
be
in
outputs
instead
of
widget
areas.
So
so
what
this
does
is
it
means
that
people,
even
in
the
classic
notebook,
will
be
used
to
the
workflow
of
having
widgets
in
the
output
areas,
hopefully
by
the
time,
to
put
a
lot
rolls
around.