►
From YouTube: JupyterLab Team Meeting - June 8, 2022
Description
A meeting to share and discuss features, ideas, issues, and pull requests in JupyterLab and other Jupyter frontends. This meeting is open to anyone and everyone.
Join future calls via the Jupyter community calendar: https://docs.jupyter.org/en/latest/community/content-community.html#jupyter-community-meetings
Notes for upcoming meetings can be found on the agenda: https://hackmd.io/Y7fBMQPSQ1C08SDGI-fwtg
Past notes can be found on the JupyterLab team compass: https://github.com/jupyterlab/team-compass/issues?q=is%3Aissue+label%3A%22Dev+Meeting+Minutes%22
A
Okay,
welcome
everyone
to
the
june
8th,
weekly
jupiter
lab
call.
Today
there
are
21
of
people
on
the
call
and,
if
you
haven't
already,
please
click
on
the
link
in
the
chat
for
the
agenda
and
add
your
name
in
the
attendees
list.
A
If
you
would
like
and
add
a
bullet
point
in
the
agenda
section,
if
you'd
like
to
discuss
something
for
roughly
on
the
order
of
five
minutes,
if
there's
a
longer
discussion,
you'd
like
to
have
add
it
at
the
bottom
to
the
additional
discussions
and
it's
alex
on
the
call
okay,
he
is
thanks
for
hosting
last
week.
I
appreciate
it:
okay
cool,
so
why
don't
we
get
started?
The
first
person
on
the
agenda
today
is
federic.
D
F
E
Sorry,
so
I
just
want
to
point
out
that
the
virtual
windowing
notebook
pr
is
ready
for
review.
Everything
works
in
terms
of
ci
testing,
so
feel
free
for
having
a
look.
If
you
feel
lucky,
then.
G
E
E
Come
on,
allow
me
to
switch
tabs
yeah,
so
the
idea
of
that
pr
was
to
to
remove
some
like
I
don't
know
if
I
can
call
them
fixes,
but
some
changes
in
the
in
the
base
style
that
were
due
to
due
to
come
on
due
to
a
group
rent
and
mainly
it
was
some
some
kind
of
onset-
and
I
know
like
this
one,
those
box,
sizing
and
set
f
like
are,
can
be
very
annoying
for
downstream
extension.
So
the
idea
was
to
remove
those,
and
I
have
one
questions,
one
questions.
E
Let
me
go
back
to
the
link,
so
this
removes
this
removes
some
rules
for
the
focus
come
on
too
many
people
here.
E
Come
on
and
the
thing
is,
it
seems
that
on
on
chrome,
it
renders
not
nicely
like
the
file
browser.
We
get
a
focus
on
the
on
the
text
and
that
make
even
the
text
difficult
to
to
to
read.
So
I
just
want
to
like
to
have
some
feedback
on
that
and
the
question.
C
And
we
move
them
only
to
the
file
browser
so
that,
if
the
not
having
the
focus
is
a
problem
in
extensions,
at
least
five
brothers
looks
right
on
them.
If,
on
many
of
my
whatever
I
sorry,
I
my
understanding
is
that
you
removed
the
rules
because
they
were
annoying
somewhere
else,
so
just
maybe
make
them
more
specific
to
what.
E
As
as
jason
said,
there
are
other
cases,
and
the
thing
is
that
the
focus
as
we
don't
have
tests
specifically
for
focus
it
may
show
that
even
in
tests
that
we
don't
are
in
places
that
we
don't
test
for,
because
we
we
don't
don't
have
explicit
tests
but
as
I'm
fighting
to
get
one
example,
but
I
know
that
on
the
menu
there
are
some
examples
and
also
I
cannot
find
them
for
another,
but
there
are
those
that
odd
line
is
appearing
on
the
menus
of
lumino
and
stuff.
E
Like
that,
so
that's
mainly
the
question
like.
Should
we
say
we
keep
it
like
global,
so
we
that
mean
we
keep
the
behavior
that
we
we
used
to
have
for
focus,
or
should
we
say,
okay,
we'll
remove
it
globally
and
every
time
we
see
some
glitches
for
for
o.
Then
we
add
a
new
rules
for
covering
that
case.
F
I
really
don't
like
it
if
it's
like,
what's
appearing
on
text
cells,.
E
B
Right
right,
different
browsers,
have
different
outlines,
yeah
speaking
from
vaguely
accessibility
recommendations.
I've
read,
I
think
the
preference
I've
heard
is
usually
to
provide
your
own
focus
options,
because
usually
people
are
not
satisfied
with
the
browser
ones,
but
my
main
thought
is:
we
don't
have
like
a.
I
mean
we
kind
of
do,
but
it's
not
always
consistently
applied
at
the
moment
right.
So
I
guess
my
question
is,
and
you
may
have
said
this
and
I
got
confused
but
like
is
this?
Isn't
a
solution
for
a
global
focus
right?
A
E
That
one
you,
for
example,
in
input,
that's
the
kind
of
blue
grower,
glow
things
you
are
thinking
like
like.
If
you
put
some
some
text
in
the
filters
of
the
fire
browser,
you
get
a
blue
blue
glow
around
it,
those
ones
will
be
kept
because
they
are
specific.
B
E
C
C
Yeah,
so
what
like,
if
this
fool,
was
added
like
two
months
ago,
then
you
could
discuss
yeah,
some
blue
glow
might
disappears
because
it's
really
there
and
we're
not
removing
it,
but
if
it
has
been
there
for
a
really
really
long
time,
it
means
that
most
elements
that
have
focus
do
have
a
visual
element
that
shows
them.
So
I
would
be
in
favor
of
keeping
it
as
well.
H
Thanks
the
other
big
difference,
I've
been
scrolling
through
these
screenshots
is
that
there's
a
it
seems
like
a
line
height
difference
that
a
lot
of
the
screenshots
are
becoming.
H
E
H
It
also
subtly
shifts
like
in
the
fine
box
the
number
of
matches
it
suddenly
shifts
it.
So
now
it
might
not
match
the
line.
A
E
C
H
F
F
G
It
was
something
that
I
discussed
at
length
with
brian,
but
the
reason
for
that
is
to
indicate
visually
that
the
cell
toolbar
will
work
on
what
you're
seeing
because
otherwise
it
would
just
be
floating
in
space.
D
I
D
Was
we
we
found
that
in
certain
notebooks,
the
the
association
between
the
cell,
toolbar
and
markdown
content
wasn't
clear?
I
realize
it
there's
other
dimensions
that
I'm
not
thrilled
about
it,
but
it
was
really
hard
to
to
sort
of
associate
the
the
cell
to
a
bar
with
the
the
cell.
It
was
attached
to.
G
So
so
one
thing
you
need
to
know,
though,
is
that
there
are
actually
two
two
borders
on
every
cell
now
as
a
result
of
the
cell
toolbar,
one
that
you
can
see
and
then
one
one
pixel
transparent
border,
because
if
the,
if
that
second
border
weren't
there,
things
would
be
jumping
around
by
a
few
pixels,
which
is
mildly
infuriating.
A
Folks,
I
think
we
should
move
on
to
another
subject,
just
in
the
interest
of
getting
through
the
agenda
fred,
you
had
two
more
points.
E
Just
just
an
announcement,
so
there
is
the
performance
meeting
after
this
meeting.
So
if
you
are
interested
stay
after
this
one
and
personally
I
will
go
a
bit
more
in
details
on
the
notebook,
windowing
implementation
and
also
I
will
present
some
prayer
menu
plan
for
adding
memory
leak,
benchmark
on
jupyter
lab,
and
that
was
all
for
me.
A
Thank
you,
okay,
jeremy
url.
J
Yeah,
just
a
quick
one
to
let
you
know
that
there
is
a
new
patch
release
for
3.4
out
and
yeah.
You
can
install
your
tip
or
try
the
binder.
There
were
a
couple
of
fixes
that
were
pending
for
the
past
month
and
yeah.
Now
they
are
out
and
that's
it.
A
You,
okay,
so
we
were
talking
about
this
in
the
notebook
call
as
well.
I've
linked
to
a
pr
that
I
am
working
on.
It
is
in
draft
mode
and
I'll
confess
up
front
that
it
definitely
is
two
things
that
should
have
been
two
pr's,
but
I
ended
up
doing
them
in
the
same
place
because
it
was
the
same
file.
So
I
don't
have
good
hygiene.
A
The
first
one
is
uncontroversial.
It
is
adding
a
service,
a
manager
for
the
events
endpoints
that
have
been
recently
added
to
jupiter
server.
I
mean
it
might
be
controversial
in
the
apis
that
I
pick,
but
it's
not
controversial
that
we
need
this
thing
to
support.
What
the
back
end
supports.
The
second
one
actually
might
be
controversial,
and
it
is
this.
A
It
came
out
of
a
question
that
fred
asked
a
couple
weeks
ago,
saying
if
I
want
this
new
events
manager
that
has
some
sort
of
signal
that
emits
every
time
I
hear
something
on
the
websocket.
Do
I
have
to
thread
down
app.servicemanager.events.
A
So
then
I
talked
about
it,
some
with
steve,
and
we
thought.
Maybe,
if
you
have
two
functions,
a
get
instance
that
either
returns
null
if
no
one's
ever
set
one
or
returns
the
existing
one,
and
also
you
have
a
set
instance
where
you
explicitly
say
no
matter
what
that
singleton
is.
I
want
my
service
manager
in
there
instead
of
the
one
that's
already
there,
then
maybe
at
least
that
ambiguity
goes
away,
but
the
next
question
that
arose
was
what
happens
to
app.service
manager.
D
A
So
I
made
the
service
manager
be
an
observable
disposable
so
that
you
can
listen
to
its
disposed
signal
and
react
to
it.
So
now
the
jupiter
front
end
can
listen
to
the
disposed
signal
of
a
service
manager.
If
it's
disposed,
then
it
can
say
get
instance
of
the
new
service
manager.
If
it
happens
to
not
have
been
set,
then
it'll
set
instance
and
do
create
its
own.
A
I
don't
know
I
mean
this.
Isn't
I
kind
of
like
it
because
that
means,
if
you
have
a
component
like
six
levels
down,
because
it's
a
react
component
within
another
react
component
within
another
et
cetera,
it
means
you
could
just
say
from
jupiter
lab
services,
import
service
manager,
and
then
you
could
say
my
server
and
I
care
about.
Let's
say
events
right.
You
can
say,
events
equals
service
manager,
dot
get
instance,
dot
events
and
then
you're
done.
You
have
an
event
manager,
it's
the
one!
A
A
So
for
the
for
the
current
situation,
the
way
so
jupiter
light
wants
its
own
service
manager
and
jupiter
light
has
to
in
creation
of
its
application.
It
has
to
set
a
service
manager
there,
but
that
service
manager
has
been
accessible
to
every
extension.
This
whole
time
and
they've
all
been
able
to
dispose.
It
actually
not
replace
it,
but
make
it
broken
right.
A
So
this
takes
that
status
quo
and
adds
a
new
element
to
it,
where
an
extension
can
come
in
and
say,
I'm
going
to
swap
out
the
global
service
manager
and
then
there's
a
risk
that
somebody
who
cached
the
last
one
is
holding
a
bad
reference,
and
there
is
definitely
like
some
ambiguity
that
might
arise
here,
but
you're
not
actually
you're,
not
actually
introducing
more
dysfunction,
because
anytime,
someone
hits
dispose
on
the
current
one,
they're
breaking
the
app
right.
This
actually
allows
a
way
out
of
that.
I
A
I
See
a
hand
oh
robert
hi,
there
I'll
have
to
formally
introduce
myself
at
some
point.
Can
you
hear
me?
Can
you
hear
me
okay?
So
I
have
a
special
interest
in
replacing
the
service
manager
for
my
use
case.
In
fact,
this
would
prevent
me
from
having
to
rebuild
the
entire
jupiter
notebook
base.
Just
so
I
can
tweak
something
for
one
of
my
extensions,
so
I'm
really
excited
to
have
you
bring
up
this
topic?
I
I
think
it's
a
reasonable
assumption
from
my
perspective
to
say
that
the
service
manager
is
going
to
be
installed
once
at
the
beginning
of
the
notebook
and
then
never
get
replaced
for
those
people
who
are
really
ripping
out
the
guts
of
jupiter.
Notebook
like
I
am
they're
they're
going
to
know
like
this
is.
This
is
what
I
want
to
do
so,
those
those
are
my
two
cents.
Thanks
cool.
A
J
Since
we
are
talking
about
swapping
things
for
other
things,
I
was
wondering
if
you
were
so
considered
like
making
the
service
manager
plugin
based.
That's
something
we
that
we
do
actually
in
jupiter
light
like
on
the
server
part.
So
all
of
the
server
is
implemented
using
the
aluminum
operating
system,
and
that
means
that
someone
can
swap
how
the
kernels
are
managed
in
the
browser,
for
example,
by
something
else,
because
it's
just
a
plugin.
So
I
was
wondering
like.
J
Maybe
if
we
want
to
go
like
even
further
with
this
idea,
maybe
we
should
start
exposing.
You
know
tokens
like
the
services,
the
other,
the
dependency
injection
system,
so
that
if
someone
wants
to
swap
just
the
kernel
service,
the
kernel
manager,
part
of
the
service
manager,
they
could
get
an
extension.
A
A
Well,
that's
overstating
it,
but
so
many
extensions
are
going
to
talk
to
services
that
it's
a
little
bit
like
the
other
top
level,
things
that
we
have
like
a
command
registry,
or
something
and
like
you
know,
just
imagine
how
many
cert,
how
many
extensions
want
to
grab
their
own
settings,
for
example,
if
all
of
them
have
to
bring
in
one
additional
extension,
it's
not
the
end
of
the
world,
but
it
is
something
that
we
were
seeking
to
avoid
and
also
the
app
itself
might
want
a
handle
to
the
services
so
like.
A
So,
like
you
could
imagine
a
world
where
the
way
you
do
it
is,
you
still
have
like
a
get
instance,
but
you
don't
have
a
set
instance
and
the
plug-in
that's
going
to
share
it,
grabs
it
via
get
instance
and
also
so
does
the
application.
If
the
application
needs
to
communicate
with
the
server
or
something
but
yeah,
I
don't
know,
I
think
that
is
certainly
like
a
sound
model
to
put
it
in,
just
like
any
other
plug-in.
It's
just
that
it's
so
ubiquitous.
H
Ask
one
more
question
about
your
current
model.
Sure
so
currently,
what
you
do
is
essentially
a
dynamic
dependency
injection
right.
You
get
an
active
object
at
runtime
and
that's
your
service
manager
with
a
singleton
pattern.
You're
importing
something
from
the
from
the
services
package.
Is
that
right
and
using
the
get
instance
on
something
that
you're
importing.
So
is
this,
so
this
is
introducing
a
static
dependency
on
a
specific
package
rather
than
a
dynamic.
A
A
Okay,
so
maybe
if
you
have
an
interest
in
this
subject,
please
comment
on
the
pr
it's
a
draft
and
it's
a
draft
partly
because
we
have
to
publish
jupiter
server
for
events
and
again
I
apologize
for
conflating
these
two,
but
I've
done
it.
Sorry,
but
but
yeah.
I
think
jeremy
already
had
some
interesting
thoughts
and
we
have
some
conversation
there
and
I
will
not
belabor
this
point,
because
we
could
talk
a
long
time
about
stuff
like
this
cool.
Thank
you
so
matthias.
You
are
up
next.
C
Thank
you
so
first,
I
want
to
thank
jason
that
with
him
yesterday,
and
he
told
me
to
come
here
so
everything
which
is
great
about
what
I'm
going
to
show
is
thanks
to
jason
grout.
If
you
don't
like
something,
it's
my
fault,
so,
as
you
may
or
may
not
know.
For
the
past
almost
a
year,
I've
been
working
on
trying
to
improve
the
documentation,
experience
in
global
way
across
python,
with
a
project
that
I
call
papyri.
You
have
a
cci
grant
and
there
is
a
lab
extension.
C
It's
trying
to
push
further
than
some
of
the
folks
have
done
with
duck
wrapper
to
allow
like
having
access
to
complete
documentation,
not
only
dot
strings,
but
everything
which
is
narrative
documentation
in
jupiter
lab,
and
so
I'm
going
to
show
the
really
rough
draft
of
jupiter
extension
I
have
so
here.
I
have
a
jupiter
lab
open
with
one
function
from
from
scipy,
and
here
is
how
it's
look
like
right
now
and
with
my
draft
extension,
you
have
all
of
the
narrative
documentation,
plus
plots,
plus
everything,
and
you
should
even
be.
C
So
for
every
library
across
libraries
you
should
be
able
to
navigate
and
access
like
a
real
documentation
with
all
the
features
that
html
documentation,
like
things,
have
plus
some
more
fancy
things.
So
here
I'm
just
switching
to
a
flask
server,
because
I'm
not
good
at
typescript,
and
I
wasn't
able
to
implement
that
in
jupyter
lab.
C
I
don't
find
it.
I
will
find
it
yep
if
you
want
to
find,
for
example,
all
the
mats
in
all
of
numpy.
You
get
all
the
maths
of
numpy
or
the
math
equations,
and
so
my
goal
is
to
bring
some
of
that
which
is
like
right
now
in
the
fast
server,
because
I
know
how
to
write
python
into
typescript
and
into
jupiter
lab.
So
I'm
going
to
show
some
of
that
at
scipy
and
I'd
love
insights
and
help
into
tying
up
things
into
the
lab.
C
For
example,
I'd
love
the
question
mark
operator
to
directly
open
the
right
documentation
on
the
side
panel,
which
currently
I
have
difficulties
doing
and
I'm
going
to
keep
it
short
at
that
because
I
know
other
folks
have
things
to
say.
Maybe
one
of
the
questions,
if
you
have
questions-
and
otherwise
I'm
happy
to
talk
with
you
later-
do
not
take
too
much
time.
C
No,
no,
it's
render
from
the
extension
so
the
server
there
is
a
server
extension
that
sends
json
to
the
front
end
and
it's
rendered
using
react
right
now
and
it
does
follow
jupiter
lab
theme.
Maybe
I
can
retry
that
I
haven't
tried
it
in
a
long
time,
but
if
I
go
to
theme
and
I
go
to
jupiter
dark,
this
is
this:
is
there
and
all
of
that?
The
navigation
like
if
I
want
to
go
to
the
narrative?
Well,
not
everything
is
hooked
is
triple.
C
As
much
as
as
we
can,
you
see
like
the
highlighting
is
not
great
right
now,
but
it's
it's
supposed
to
be
native
jupiter
component
and
one
of
the
reason
is
security.
There
are
similar
extensions
that
basically
use
pre-built
html.
The
problem
with
probably
html
is
that
you
may
get
injections
or
things
like
that.
You
would
have
to
sanitize
it
or,
and
also
it's
hard
to
hook
into
them.
Click
events,
like
you
know,
if
I
click
on
caser,
I
can
go
to
paser.
C
For
that.
I
need
to
have
a
custom
component
that
renders
everything
my
goal
is
to
also
have
things
like
the
signature
right
now.
I
don't
have
that,
but
I
would
have
the
type
annotation
and
have
settings
like,
for
example,
users
can
select,
do
you
want
type
annotation
or
not,
so
you
could
hide
and
show
or
other
things,
and
in
particular
you
know.
I
need
to
to
do
fancy
things
like
when
I
showed
you
the
back
reference.
C
You
don't
want
to
rebuild
everything
when
you
get
a
new
library
installed,
and
this
is
smart
enough
to
build
the
cross
reference
in
a
way
which
is
optimized.
So
you
only
update
the
pages
that
are
targeted
by
new
references
if,
if
that
makes
sense,
it's
hard
to
condense
in
a
few
minutes,
but
we
can,
we
can
discuss
it.
E
C
Yeah,
so
so
json
is
expecting
a
bit
how
it
works.
So,
instead
of
building
html,
you
build
an
intermediate
representation
for
each
packages
where
the
references
are
not
complete,
but
when
it's
inside
the
package,
the
reference
are
complete
between
the
across
packages,
it's
a
default
references,
and
so
when
you
install
a
new
version
of
numpy,
you
know
that
the
different
default
references
need
to
be
updated
and
you
can
update
the
back
reference
in
in
the
original
package.
C
Those
are
the
new
new
links
and
you
can
do
a
lot
of
fancy
things
that
I
won't
get
into
right
now
and
I
I'd
love
to
get
some
help
from
the
projector
lab
to
hook
some
of
the
things
and
also
beta
tester
at
some
point
to
have
better
documentation
for
users,
and
I
want
to
also
thanks
being
on
accessibility
and
design.
L
Hey
folks,
I
know
last
week
some
of
y'all
were
talking
about
going
to
sci-fi
and
so
we're
building
off
of
some
work.
It
all
started
at
project
jupiter,
basically
with
the
jupiter
accessibility
work,
trying
to
get
we're
wondering.
How
do
you
lower
the
barrier
to
entry,
get
rid
of
get
out
of
contributions
and
you
know
get
people
into
a
project,
so
we've
been
successful
at
doing
that
by
adding
alt
text
to
images.
L
Isabella's
really
pushed
this
work
forward,
and
it's
kind
of
grown
past
our
reach
at
this
point,
and
it's
shown
up
in
some
different
projects,
so
at
scipy
we're
going
to
try
and
run
these
in
person.
We've
only
run
them
did
virtually
for
an
hour
and
we've
been
able
to
get
multiple
people
in
revision
histories
in
numpy
jupiter
anyway.
L
L
It's
gonna
have
some
placeholders,
where
we
need
some
alt
text
and
some
images
and
some
documentation,
and
once
we
have
that,
that's
all
I'm
gonna
really
need
help
with.
So
the
idea
is
gonna,
be
there's
gonna,
be
a
scavenger
hunt
across
big
projects,
we're
looking
at
projects
like
scipy,
perhaps
some
numpy,
perhaps
some
desk,
hopefully
jupiter.
So
if
anybody's.
At
this
event,
you
know
we'll
basically
have
some
content
available
for
you
all
to
at
sprints
buffs,
something
like
that.
L
Just
some
kind
of
engagement,
some
sort
of
onboarding
things
along
those
lines.
We're
gonna
have
one
of
our
co-workers
at
kwonsite,
marsley
she's,
going
to
be
there
kind
of
ushering
this
around
and
she's
gonna
be
premiering
a
zine
about
disability
awareness
and
open
source
science.
All
of
this
stuff
goes
together,
so
whoever
is
at
scipy
we'd
love
your
help
in
participating
in
this
alt
tech,
scavenger
hunt.
It
goes
along
with
a
much
bigger
production.
L
K
K
So
I'm
just
gonna
show
how
we
got
our
got
the
github
usernames
next
to
the
cursor.
So
what
I'm
talking
about
is
this
so
there's
two
github
users
and
it's
in
collaborative
mode.
K
So
the
way
we
did
it
is.
This
is
just
an
ordinary
jupiter
hub
configuration
file
here
and
what
we
do
is
there's
this
post
off
hook
that
you
can
set
here
and
the
documentation
indicates
that
it's
okay
to
do
this,
and
so
we
all
we
do
is
when
the
person
authenticates
we
just
set
a
cookie
with
their
name.
K
A
Why
did
you
need
to
cast
as
any
to
get
the
awareness
attribute?
Is
there
something
missing
from
the
api?
No.
K
That's
the
time!
That's
that's
just
no!
That's
that's
just
me
not
finishing
off
the
code,
so
yeah
that
needs
to
be
typed
out
properly.
This
is
just
something
I
got
working
a
little
while
ago,
but
yeah
that
there's
nothing
wrong
with
that.
I'm
sure
I
can
import
that
and
get
it
to
work
right.
A
K
K
That's
what
all
this
stuff
is
down
here,
but
this
is
just
an
ordinary
github
authenticator
set
up
here.
The
only
thing
I
added
is
just
this
this
bit
here
and
then
again
over
here.
It's
just
a
simple
extension:
where
it
every
time
a
notebook
gets
launched,
it
sets
the
hub
username.
So
when
the
person
uses
the
share
link
extension
to
share
their
link,
this
runs
and
sets
that
cookie,
whatever
whatever
their,
whatever
the
cookie.
Whatever.
However,
the
cookie
got
set
when
they
authenticated
it
just
sets
that
cookie
as
their
hub
username.
K
K
Oh
yeah,
I
don't
I
don't
know
that
I'm
actually,
I
don't
actually
know
about
the
user
api,
and
I
don't
know
that
this
is
the
right
way
to
do
this.
I'm
not
saying
that
it
is.
It
works
pretty.
Well,
though,
one
of
the
issues
I
ran
into
so
there's
lots
of
ways
to
do
this
right.
K
One
of
them
was
to
you
know,
pull
the
username
from
the
backend
or
something
like
that.
But
what
I
discovered
you
know
when
you
do
that
share
link
thing.
There
isn't
really
anything.
That's
indicating
who
you
are.
You
know,
as
far
as
I
know,
it's
like
you're
the
person
who
who
who's,
who
shared
the
link
when
you
get
into
it.
When
I
looked
at
all
the
cookies,
I
didn't
see
anything
differentiating
that
user
once
they
get
into
that
person's
shared
notebook.
So,
and
maybe
there
is,
there
might
be
something
there
that
I
just
missed.
K
E
A
So,
in
addition
to
this
in
jupiter
hub,
there's
also
a
new
identity,
api
being
added
to
jupiter's
server
directly,
and
that
means
that,
once
that
version
of
server
is
released,
even
if
there's
no
identity
information
in
the
url
you're,
hitting
by
virtue
of
being
a
server
user
like
an
authenticated
user
of
this
jupyter
server,
it'll
know
who
you
are
and
it'll
have
an
end
point
you
can
hit
just
like
that
user.
One
for
hub,
but
it'll,
be
even
pure
jupiter
lab
without
hub
will
have
that.
A
So
I
think
that
the
functionality
you're
building
basically,
is
because
some
of
the
building
blocks
you
needed
weren't.
Yet
all
there
right,
but
they're,
sort
of
coming
in
as
different
pieces
of
rtc
land.
So
you
at
some
point
soon,
instead
of
having
to
populate
a
cookie
and
read
your
own
cookie
can
hit
the
rest
endpoint.
That
just
says
who,
who
who
am
I
and
I
will
tell
you
who
you
are?
If
you
use
that
information,
your
authentication
mechanism
will
still
remain
whatever
you
were
using
before.
A
So,
in
your
case,
it'll
still
remain
authenticating
with
github,
but
your
identification
can
come
straight
from
the
server
without
you
having
to
intervene
to
do
it
yourself
once
that
api
is
in
to
server,
but
as
fred
says,
if
you
are
going
to
use
hub
anyway,
hub
does
come
with
that
endpoint.
You
could
hit
to
get
your
user
identity
as
well.
D
I
I
would
talk
to
zach
saylor
at
the
jupiter
server
meeting.
I
think
he
has
a
demo
of
all
this
working
for
just
jupiter
hub.
I'm
sorry,
just
jupiter
live
outside
of
jupiter
hub
and
I
think
min
has
demos
of
it
all
working
inside
of
jupiter
hub.
So
it
we're.
B
A
Cool
thanks
for
showing
the
demo
adam.
So
we
do
have
one
additional
discussion
point
and
we
have
nine
minutes
so
I'll.
Just
eight
minutes
is
there
anyone
who
wanted
to
show
any
last
thing.
Otherwise,
let's
spend
the
rest
of
the
call
talking
about
this.
H
I
don't
think
we
need
to
discuss.
I
was
just
copying
something
isabella
put
in
the
chat
into
the
notes.
Okay,.
A
H
Unless
isabella
wants
to
say
something
about,
it
is
the
the
the
workshop
planning
notes
that
you
put
in
the
net.
B
What
did
I
do?
Oh
scrolling,
I'm
looking.
Oh,
this
is
the
wrong
hack.
Md
one
moment.
Sorry
too
many
things
open.
Apparently
I.
B
Oh
okay,
yes,
just
to
be
clear,
I'm
not
running
anything
to
do
with
the
sci-fi
things,
so
it
was
just
more
last
week.
One
of
the
things
you
asked
for
jason
was
like.
Oh
these
workshops.
I
don't
know,
and
so
I
was
like
oh
you're
right.
I
should
have
kind
of
an
index
of
things.
We've
done!
So
that's
what
it
is.
I
am
happy
to
talk
more
about
it
in
general,
because
I
think
of
it
kind
of
like
traveling
circusy.
If
you
will,
like
the
event,
can
be
brought
anywhere.
B
So
it's
pretty
cool
one
day
I'll
drag
some
of
you
other
people
into
an
event
and
you'll
see
why
it's
cool,
but
yeah
the
records
are
cool.
The
ones
I
have
labeled
as
collaborating
prs,
are
also
really
awesome,
because
you
see
the
discussion
people
have
because
the
collaborating
ones
is
just
where
everybody's
like.
Oh,
should
I
do
this
for
this
alt
text
or
this,
whereas
the
contributing
tend
to
be
a
lot
cleaner.
The
maintainers
just
they're,
either
like
wow,
thanks
or
they're
like
what
are
you
doing
and
that's
kind
of
how
it
goes
so.
K
H
Okay,
I
did
just.
H
Point:
okay,
a
question,
an
open
question:
we're
coming
up:
I
it's
less
than
a
month
away:
yeah!
Maybe
it's
exactly
a
month
away
just
over
a
month
away,
and
there
is
at
least
one
tutorial
about
building
jupiter
lab
extensions
and
there's
typical
plenary
tools
talk
about
the
state
of
the
jupiter
ecosystem.
A
L
A
What
what
what
are
other
people
thinking,
I.
E
I
think
we
could
move
to
a
better
cycle
for
sci-fi,
but
rc
seems
a
bit
premature,
especially
because,
with
all
the
apa
changes
there
are
a
couple
of
extensions
that
should
probably
try
to
migrate
and
so
yeah.
E
A
A
I
think
I
think
it's
not
unreasonable
to
try
to
push
some
betas
out
before
then.
It
seems
like
an
rc's
unlikely
unless
it's
the
kind
of
rc
that's
followed
by
six
more
rc's.
So
maybe
we
shouldn't
do
that
so
that
we
don't
dilute
the
definition
of
an
rc
but
yeah.
It
doesn't
really
seem
likelier
than
a
beta
phase.
By
that
point,.
K
E
H
I've
heard
from
somebody
you
always
say
q4,
because
it's
always
not
this
quarter,
but
then
q4
everyone's
on
holiday.
Nobody
cares,
and
so
you
could
slip
without
without
ruffling,
too
many
feathers.
Unless
it's
christmas
present
to
somebody
but
yeah,
okay,
I
think
that's
a
decent
message.
If
we
can
get
a
beta
out
that
we're
at
beta
and
we're
moving
forward
to
rc.
G
H
I
think
sends
a
signal
that
we're
sort
of
wrapping
up
the
feature
part
of
this
and
and
we're
moving
towards
stabilization
and
finalization.