►
From YouTube: JupyterLab Team Meeting - June 22, 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
Hi
everyone
welcome
to
the
june
22nd
weekly
jupiter
lab
call.
Today
we
have
16
people
and
a
light
agenda.
So
let's
see
how
much
we
actually
have
to
discuss.
B
I'm
happy
I
wrote
the
bullet
point,
okay
and
I
was
going
to
ask
darian
to
discuss
them.
Here's
my
introduction.
A
Thanks
jason,
I
asked
sylvan
to
talk
about
them,
but
I
don't
see
him
here.
So
that's
cool
okay!
So,
as
we
are
approaching
jupiter
lab
4
and
looking
to
release
rtc
in
a
production
ready
way,
we
have.
A
A
The
new
api
is
in
a
refactor
of
rtc
and
there
has
been
a
lot
of
discussion
in
that
pr
over
the
approach
and
there
is
at
its
core
some
disagreement
over
where
collaborative
data
structures
that
model
concepts
in
jupiter
lab
ought
to
reside.
A
We
eventually
ship
with
rtc
and
as
a
way
to
help
figure
out
if
we
can
arrive
a
consensus
between
two
different
ideas
of
how
to
implement
and
organize
these
packages.
Brian
wrote
two
issues.
One
of
them
has
to
do
with
this
problem.
I've
been
describing
of
where
should
collaborative
models
reside?
Where
should
the
cell
model
be?
Should
it
be
in
the
sales
package
or
somewhere
else?
Where
should
the
notebook
model
be?
Should
it
be
the
notebooks
packages
notebook?
A
So
that's
one
issue
and
the
other
issue
is
also
brought
about
by
the
real-time
collaboration
conversation
that
is
about
the
metadata
top
level
attribute
that
exists
in
the
notebook
format
and
whether
it
ought
to
be
seen
from
the
point
of
view
of
the
data
structure
that
resolves
multiple
rights
to
it.
In
the
real-time
collaboration.
A
Functionality,
whether
it
should
treat
metadata
as
an
opaque,
json
blob,
where,
if
you
are
an
extension
author,
for
example,
you
are
writing
an
extension
that
allows
you
to
comment
on
individual
cells
and
the
data
store.
For
your
comments
is
notebook
metadata
or
you
are
an
extension
that
allows
for.
A
A
And
so
those
are
the
two
issues
being
called
for
vote
now.
Hopefully,
what
that
means
is,
if
we
arrive
at
a
definitive
answer
to
those
questions,
it
gives
us
a
better
frame
for
reviewing
carlos's
pr
versus
alternate
models
of
implementing
the
same
functionality.
A
That's
the
level
of
detail
that
I'm
comfortable
going
into
this,
because
I
am
not
completely
immersed
in
this.
The
way
other
people
are,
but
but
yeah,
that's
that's
the
sort
of
top-level
context
of
those
two
issues.
B
A
We
did
this
once
already
fred
about
the
ui
toolkit.
Although
we
didn't
follow
the
letter
of
the
law
there,
we
actually
kept
that
vote
open
longer
than
a
week.
These
two
issues,
brian
wrote
the
issues
inside
the
jupiter
lab
repo,
and
I
wrote
their
counterpart
votes
in
the
jupiter
lab
team
compass
and
I
wrote
them
monday
night.
A
A
A
C
Yeah
brian
is
out
on
vacation
for
much
of
this
week
he's
in
transit.
He
told
our
team
here
at
aws
that
he
might
be
available
to
like
occasionally
chat
via
slack,
but
he's
going
to
be
somewhat
incommunicado.
This
week.
B
A
I
think
that's
probably
sound,
I'm
not
sure
I
guess
it
also
depends
on
so
suppose.
It's
next
tuesday
and.
A
B
Yes,
I
think
it
ought
to
extend
the
voting
phase.
I
think
a
week
is
kind
of
a
short
voting
phase
for
an
involved
architecture
decision
for
people
to
dive
in
if
they
haven't
been
paying
attention
and
if
they
are
able
to
dive
in
in
the
first
week
and
start
discussing
things,
then
I
think
that
would
letting
just
the
discussion
proceed
would
be
a
wise
thing.
A
Let
me
first
address
jason's
question
and
then
see
if
I
understood
tony's
and
try
that
too.
So
the
actual
implementation
of
the
rtc
features
is
quite
a
large
pr
and
intimidating,
and
probably
an
unreasonable
thing
to
ask
everyone
in
the
jupiter
lab
council
to
read
and
understand
these
two
issues
are
designed
to
be
more
bite-sized
and
easy
to
reason
about
on
their
own.
In
the
absence
of
an
rtc
conversation
at
all,
because
it
just
didn't
seem
likely
that
we
would
be
able
to
get
you
know
whatever.
A
A
I
think
that
I
think
that
the
hope
is
these
issues
don't
require
deep
dives.
I
think
it's
more
like
if
you
were
designing
an
api
and
you
want
to
ship
an
api,
and
you
knew
this
about
the
history
of
the
project.
What
would
your
preference
be?
Would
the
api
look
like
this
or
look
like
that,
and
I
don't
know
if
it
is
still
too
big
a
climb,
because
it
still
might
be
a
question
that,
even
with
the
description
there
doesn't
have
enough
context
to
decide
on?
Maybe
that's
true.
A
D
Well,
you
know
I
feel,
like
the
pool
doesn't
have
enough
waves
in
it
today.
So
part
of
me
thinks
that
maybe
it's
time
to
reevaluate
the
notebook
format
like
what
does
v5
look
like,
because
we've
done
a
lot
of
weird
stuff
in
v4,
and
sometimes
I'm
wondering
if,
like
the
current
document
format
that
we
have
is
amenable
to
the
future
that
you
all
are
talking
about
like,
I
can
see
it
complying
with
version
four.
D
But
what
does
the
jupiter
universe
look
like
in
version
five,
where
we
can
use
schema
across
the
applications
when
we
can
refer
de-reference
schema?
When
we
have
you
know,
we
have
a
lot
of
sk
we're
talking
a
lot
of
schema
now
and
I'm
wondering
if
there's
a
bigger
leap
like
I
mean
I'm
talking
a
year
leap
of
discussion,
but
like
is
there
a
version?
Five,
that's
more
amenable
to
the
future
that
y'all
are
dreaming
of.
C
B
Yeah
code
architecture,
decision
that
doesn't
touch
the
actual
format.
However,
I
think
a
separate
conversation
about
the
feature
of
the
notebook
format
is
is
very
interesting.
E
I'm
just
curious
about
the
question.
There
are
both
technically
possible.
A
E
A
A
Probably
either
definition
of
metadata
beyond
what
we
currently
have
in
md
format
or
some
other
thing.
So
one
level,
I
think,
is
achievable
because
we
know
it's
a
json
blob
and
we
know
json
blocks
are
made
of
keys
right
keys
with
arbitrary
values.
So,
if
you're
just
saying
that
you,
the
top
level
keys
of
this
json
blob,
are
addressable,
that's
achievable
immediately
beyond
that,
I
don't
think
we
can
apply
more
structure
without
making
fundamental
changes,
but
really.
E
I
was
just
wondering
about
this
thing
that
it
says
that
it,
this
decision
would
rule
us
out
us
making
decisions
now
that
would
have
make
it
difficult
or
impossible
to
move
in
this
direction
over
time.
So
is
that
the
way
you
just
described
is
that
compatible
with
future
things
to
make
it
more
fine-grained.
A
A
Okay,
I'll
remind
you
guys
again,
if
there's
something
you'd
like
to
add
to
the
agenda,
please
do,
but
for
now
let's
see
some
shiny
stuff
from
alex.
F
Sure,
let
me
pad
out
the
agenda
with
a
quick
demo
I'll
share
my
screen.
F
So
I
don't
know
how
many
people
here
know
what
streamlit
is
it's
a
pseudo
competitor
with
voila?
It
uses
python
as
a
base
instead
of
notebooks,
but
it
spins
up
its
own
web
apps
per
file
to
run,
and
so
I
took
advantage
of
the
iframe
widget
to
allow
me
to
open
up
files.
So
if
you
can
see
here,
I
just
open
up
a
python
file
and
I
can
just
start
a
streamlit
app
based
on
that
file
and
it
will
spin
it
up
in
a
new
tab.
F
Sometimes
it
takes
a
second
and
it
just
inserts
extremely
app
into
the
tab.
You
can
also
right-click
a
file
and
run
in
streamlight,
and
it
will
also
open
it
up.
You
can
have
as
many
of
these
as
you
want
as
long
as
your
processor
doesn't
struggle
too
much,
and
you
can
also
right
click
and
run
and
streamline.
F
So
it's
a
cool
little
demo.
This
is
just
taking
again
advantage
of
the
iframe,
so
this
is
a
back-end
server
extension
that
is
called
by
an
api
to
spin
up
an
instance
for
the
given
file,
and
then
it
just
returns
the
url
that
that
instance
span
up
and
then
it's
inserted
into
an
iframe
using
the
iframe
widget.
F
So
this
is
specific
to
streamlight
currently,
but
this
methodology
can
be
used
for
pretty
much
any
web
app.
That
could
be
spun
up
on
the
back
end
and
my
co-worker
alan
is
also
working
on
leveraging
the
jupiter
server
proxy
extension
to
proxy
these
applications
directly
into
the
the
lab
path.
F
Frame
got
it.
Yes,
though,
I
believe
yeah
that
should
still
be
possible
even
after
we
do
the
proxy
it'll
just
be
that
the
url
at
the
top
is
the
lab
path
instead
of
localhost.
A
F
My
biggest
worry
is
because
the
iframe
widget
actually
turns
on
all
of
the
what's
it
called
all.
It
turns
on
all
the
security
by
default,
the
I
lost
the
name
container
like
they
can
it
blocks
every
thing
from
coming
in
and
out
of
the
iframe,
which
is
not
default.
If
you
just
like
write
up
an
iframe
in
html,
you
have
to
add
those
things.
F
In
order
for
this
to
work,
you
have
to
actually
disable
all
that
security
so
because
any
web
app
will
need
to
have
the
ability
to
run
a
javascript
code
and
access
the
javascript
libraries
that
are
located
in
your
web
browser,
so
the
what's
in
the
iframe
can
run
javascript
in
jupiter
can
access
jupyter
lab.
So
in
this
case
I
confirmed
that
streamlet
itself
has
no
code
that
could
do
that
and
that
streamlit
itself
actually
encapsulates
any
user
inputted
code
and
prevents
it
from
being
run
outside
of
its
own
iframes.
F
So
in
this
case
the
only
security
constraint
here
would
be.
If
the
actual
streamlet
web
app
code
itself
was
compromised,
you
don't
have
to
worry
about
user
inputs.
F
That's
a
good
point.
I
did
run
into
an
issue
yesterday
if
you
notice-
let's
just
back
all
these
out,
also
when
I
click
when
I
close
that
tab.
It
also
sets
an
api
call
to
end
that
process.
So
it's
not
like
kernels,
where
they
stay
up
and
running
once
you've
closed
them,
okay,
cool!
We
did
consider
that
and
then
we
thought
about
it
real
hard
and
we're
like
that's
not
worth
the
effort.
A
F
This,
if
you
notice
this,
is
all
blank
and
white.
I
did
consider
finding
a
way
to
put
some
sort
of
loading
text
in
there.
Essentially,
if
the
because
that
time,
where
you
saw
it
was
blank
and
white
is
actually
the
url
inside
the
iframe
is
just
not
set.
I
wait
for
the
api
call
to
return
before
I
set
that
url.
A
F
Yeah,
it's
just
a
matter
of
tying
that
into
the
no
url
being
set,
and
that
would
be
something
that
would
potentially
work
inside
the
iframe
class
itself
rather
than
needing
to
implement
on
a
per
usage
basis.
F
A
B
A
That
particular
use
case
you're,
better
off
using
main
area
widget
as
the
container
that
you
populate
with
your
loading
or
whatever
you
want
to,
but
the
place
where
I
wonder
more,
if
you
want
customizability,
is
in
setting
iframe
options.
F
So
I
actually
think
the
iframe
security
stuff
is
actually
pretty
well
done
already.
It
allows
you
to
fully
customize
it
it's
it's
that
it's
turned
on
by
default.
So
I
forget:
let's
just
inspect
this
real
quick.
I
remember
correctly.
F
F
That's
default
behavior,
but
if
you
set
sandbox
equals
empty
string
that
turns
on
sandboxing,
because
these
are
white
listing
of
disabling
specific
things
and
specifically,
when
you
allow
same
origin
and
allow
scripts,
it's
essentially
the
same
as
not
having
sandbox,
because
allowing
scripts
lets
you
run
javascript
and
allowing
same
origin
allows.
Well.
That's
that's
a
cost
thing,
but
it's.
This
is
key,
because
if
you
don't
have
allow
same
origin
with
iframes,
the
iframes
can't
access
standard,
javascript
libraries
located
in
your
browser
like
because
they
can't
access
the
sources
and
the
storage.
F
F
F
F
It
might
be
that
I
never
played
with
that
security.
I
I
just
know
this
from
what
the
documentation
of
iframe
sandbox
says.
Iframe
spanbox
specifically
says
when
you
turn
these
two
scripts
on,
you
might
as
well
not
be
sandboxing
that
in
the
mozilla
documentation
for
iframe
sandbox.
E
Yeah,
so
that's
as
far
as
I
understand
it,
that's
in
with
regards
to
the
server
as
long
as
you
have
a
loss
emerging
involved
scripts,
any
scripture
in
there
can
hit
the
server
with
the
same
permissions
as
your
outer
frame
can,
but
that
doesn't
mean
it
can
break
out
of
the
iframe
into
the
javascript
context.
F
That
that
is
my
my
best
guess
that
is
an
interpretation
of
that
text
in
the
documentation.
I
read.
That
would
also
make
sense.
Okay,
so
this
is.
I
wanted
to
share
this,
because
this
is
what
I've
been
doing
kind
of
in
my
spare
moments
for
the
past
couple
weeks
for
one
of
our
customers
and
this
we
haven't
decided
how
we
want
to
open
source
this.
F
Yet
it's
currently
just
sitting
on
my
own
repo
just,
but
it
will
probably
go
into
some
open
source
org,
if
not
lyra
code
or
if
there's
interest
elsewhere,
and
I
still
need
to
finalize
the
licensing,
because
I
made
the
season
cookie
cutter
and
it's
currently
still
just
cookie
cutter
licensed,
but
this
is
available
for
playing
with
if
people
want
to
and
it's
a
good
example
of
how
to
embed
a
web
app
in
an
iframe.
F
We
have
a
couple
other
examples
floating
out
and
about
there
that
I
also
looked
at,
but
it's
always
nice
to
have
a
couple
more.
A
F
In
theory,
I
would
be
able
to
do
that.
Yes,
I
don't
that's,
not
a
csv
file,
though
let's
see
is
there
a
way
to
just
show
this?
No
there's
not.
G
F
Yeah,
I
just
use
the
iframe,
as
is
so
any
behavior
is
okay.
Coming
from
that,
naturally,
I
am
intrigued
to
see
what
this
403
was,
though,.
F
Apparently,
there's
a
permissions
issue
there,
that's
good
to
note
yeah.
As
I
said,
this
is
still
in
progress,
just
tinkering
with
it,
but
also
it
does
handle
opening
non-streamless
supported
files.
It'll
just
throw
errors
at
you.
F
Web
app
for
everybody
to
see
so,
but
this
methodology
can
be
extended
to
pretty
much
any
web
app
that
you
want
to
run
within
jupiter
lab.
F
Yeah,
thanks
for
letting
me
fill
time.
A
Yeah,
no,
so,
okay,
two
things
three
things:
one
is
six
items
need
triage,
but
I
would
propose
putting
that
at
the
end,
what
with
a
dedicated
triage
call
that
happens
tomorrow,
two
is
it
looks
like
mike
would
like
to
do
a
demo
and
three
is
that
sylvan
wasn't
here
before?
A
A
Stem
okay
cool
so
mike
it
is
your
turn.
H
Unfortunately,
the
api
for
that
is
only
available
for
google
chrome
and
we
can
see
whether
it
works
open.
Notebooks
example.
H
Yeah,
it
does
have
some
problems
with
recognition,
sometimes,
but
restarting
that
process
should
usually
fix
that
enter
edit
mode.
H
H
So
what
it
does
it
basically
allows
you
to
execute
any
command
that
is
in
the
command
palette,
so
it
it
tries
to
match
those
and
using
some
edit
distance
measure
will
suggest
if
the
voice
recognition
didn't
match
exactly
the
command
that
you
that
you
set.
It
will
just
suggest
several
options
and
you
can
choose
from
from
those,
and
mostly
it
allows
to
operate
to
the
lab
without
using.
H
A
A
F
D
G
B
Chinese
thanks
for
thinking
of
a
much
more
innocuous
example
silvan.
This
is
really
cool.
H
H
A
H
Yes,
so
this
is
really
going
to
the
google
servers,
probably
because
the
implementation
of
the
speech
recognition,
api
is
browser-dependent
and
mozilla
has
made
some
efforts
to
implement
this,
but
it
was
originally
proposed
by
google
and
in
chrome.
It
is
said
that
it
was
said
that
it
will
go
to
their
servers
for
recognition
by
the
neural
networks,
but
the
details
of
that
I,
I
guess,
might
have
changed
since
then.
H
H
H
So
what
you
didn't
hear
was
that
when,
when
it
was
asking
me
whether
did
I
mean
some
comment,
I
can
also
have
speech
synthesis
enabled
and
it
was
actually
speaking
to
my
headphones.
So
I
wasn't
just
saying
yes
into
the
voids.
It
was
more
like
a
dialogue
because
of
the
zoom
options.
I
didn't
know
how
to
share
that.
H
So
on
the
side
of
voice
recognition,
so
I
guess
that
works
very
well,
because
google
does
them
very
well
on
the
level
of
the
extension,
because
we
have
language
backs,
it
will
match
against
the
labels
in
the
language
that
is
currently
being
used
by
the
user.
There
is
still
some
work
needed
to
be
done,
because
currently,
the
translator
api
does
not
expose
which
language
is
currently
set,
so
it
cannot
tweak
the
voice
recognition
to
listen
in
any
specific
language,
but
I
guess
this
is
something
that
there
is
already
an
issue
open
for.
H
H
So
that
literally
took
10
hours
of
coding,
so
it
was
done
in
24
hours
after
after
my
community
yeah.
So
it's
it's
the
beauty
of
the
template
of
the
cookie
cutter
that
allows
to
set
up
everything
so
quickly.
It's
really
amazing
and-
and
I
think
that's
the
ability
to
create
an
extension
so
quickly
is
something
which
is
really
a
fruit
of
work
of
many
of
you.
Thank
you.
A
A
To
be
needed,
I
mean
look,
you
know,
cities
need,
cities
need
sewer
systems,
but
not
everyone
wants
to
do
that.
Job.
C
Well,
I'm
proud
to
be
the
sewage
engineer
for
our
our
issues:
queue
notwithstanding
how
darian
feels
about
our
our
brave
issue
filers.
So
we
have
eight
eight
minutes
and
six
open
issues.
C
So
let's
take
a
look
at
at
least
a
few
of
them,
the
the
main
purpose
of
triage
for
folks
who
haven't
been
to
a
meeting
or
discussed
it
on
this
call
is
we
are
going
to
accept
an
issue
by
removing
the
needs
tree
hashtag
if
it's
something
that
someone
developing
in
jupiter
lab,
would
conceivably
fix
and
then
be
accepted
by
the
community,
and
if
we
need
more
info,
then
we'll
tag
you
with
needs
info.
C
So
this
issue
came
in
nine
days
ago.
It's
asking
that
saying
all
files
and
directories
are
gone
from
the
file
explorer
in
jupiter
lab.
So
last
week
we
asked
what
jupiter
lab
version
are
they
running?
Are
there
any
errors
and
we
have
not
gotten
any
info
back,
so
I
know
there
was
another
issue
that
mike
had
helped
out
with
that.
Hidden
files
were
just
had
disappeared,
but
that
was
by
design,
but
I
don't
know
of
any
other
things
that
might
be
related
to
this.
Any
thoughts
on
this.
C
All
right,
in
the
absence
of
any
additional
guidance,
we'll
just
move
on
users
getting
signal
15,
stopping
where
a
long
running
jupiter
lab
instance
crashes.
This
person
is
running
seven
kernels,
letting
it
sit
active
and
then
jupiter
lab
gets
disconnected
and
they're
running
on
aws
eks,
kubernetes
cluster.
C
A
This
made
me
realize
I
should
probably
stop
recording
at
this
point.
Yeah.