►
From YouTube: JupyterLab Team Meeting - March 23, 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,
great
hello,
welcome
to
the
jupiter
lab
weekly
meeting
for
march
23rd.
Today
we
have
about
yeah.
We
have
18
people
on
the
call.
Now,
hopefully,
a
few
more
will
come
and
we
have
an
agenda
in
the
chat.
Please
add
your
name
and
any
subject
you'd
like
to
discuss
to
the
bullet
points
on
there.
If
you
have
something
to
discuss,
that's
going
to
take
longer
than
five
minutes,
please
add
it
to
the
bottom
section
for
additional
discussions
and
with
that
let's
get
started.
The
first
person
on
the
agenda
today
is
federic.
B
B
There
has
been
a
pr
opened
by
new
contributors
for
passing
by
the
current
working
directory
from
the
launcher
to
the
terminal.
So
when
you
open
the
terminal
from
the
launcher
instead
of
being
at
the
what's,
I
don't
think
I
think
it's
the
current
working
directory
of
the
server
you
will
be
on
the
on
the
folder
of
the
launcher.
That's
me,
not
the
one
of
the
five
browser
and
the
user
asked
if
it
was
possible
to
bring
it
to
33
x.
B
So
usually
the
answer
is
no
because
it's
changing
the
api,
like
it
will
change
the
api
of
the
of
the
terminal
creation,
just
adding
new
arguments.
So
the
question
is
like:
is
it
acceptable
for
some
exception,
or
should
we
start
thinking
of
doing
a
3.4
yeah?
That's
that's
the
question
I
get
for
that
one.
So
this
is.
C
A
C
C
C
B
A
A
So
there's
there's
enough
blocking
4.0
that
we
can
probably
do
a
four,
because
there
are
features
ready
to
roll
right
now.
E
Is
it
possible
to
make
it
backward
compatible,
then
this
function
with
the
default
parameter?
Maybe.
B
I
mean
it's
fully
backward
compatible
the
option.
Everything
is
it's
possible.
B
B
Okay,
thanks
and
the
other
point
on
friday
at
the
jupiter
governor
school,
we
discussed
the
possibility
to
submit
a
letter
of
intent
for
czi
grant
in
light
of
the
coming
notebook.
Seven,
the
content
of
the
of
that
grant
will
be
addressing
documentation
and
maintenance
mainly
related
to
migration
of
extension,
so
I
have
opened
and
this
link
in
the
note
I
have
opened
a
pro
request
with
the
draft
of
the
letter
of
intent
on
the
team
compass
so
feel
free
to
to
look
at
that
and
comment.
B
The
gold
is
ready
to
try
to
have
the
content
of
that
letter
of
intent
to
align
as
much
as
possible
to
what
the
community
wants
with
pr
high
priority
for
the
notebook,
7
migration
and,
of
course
it's.
It
will
be
mainly
related
to
jupiter
f7
through
jupiter
4.
As
the
the
code
base
is
the
same,
that's
why
I
bring
it
er2.
B
B
So
there's
the
accessibility
meeting,
15
minutes
after
this
one
for
people
that
are
interested
in
it.
It's
really
a
great
subject,
and
there
is
also
another
quote
coming,
but
that
will
be
next
next
tuesday
yeah
next
tuesday,
it
will
be
the
jupiter
community.
Call
so
don't
forget
to
add
it
to
your
calendar.
That's
usually
a
good
school
for
saying
nice
demo.
D
I
think
I'm
the
only
jason
on
the
call
this
week,
because
mr
grout
is
currently
moving
house,
so
the
cell
toolbar,
that
pull
request
has
been
in
work
in
progress
for
a
long
time,
but
last
week
I
removed
the
whip
tag,
so
I'm
now
considering
it
a
candidate
to
merge
into
the
four
point
x
series.
I
know
I've
got
some
feedback,
but
I
figured
I'd
use
this
time
to
ping
the
group
for
for
some
more.
I
was
also
talking
with
folks
on
my
team.
D
There
are
folks
within
my
organization
that
are
trying
to
upgrade
our
commercial
offerings
from
jupiter
lab
one
to
three
and
whenever
I
tell
them,
oh
by
the
way,
we're
working
on
four
point
x,
they
kind
of
react
like
I
just
kicked
their
dog,
and
so
I'm
wondering
whether
the
cell
toolbar
should
get
back
bordered.
D
D
I'll
do
it
if
there's
a
demand
to
do
it.
I
this
is
something
that
I
can
probably
prioritize
with
my
team,
I'm
open
to
doing
it.
I
just
wanted
to
kind
of
take
the
pulse
of
the
organization
to
see
if
this
is
something
that
we
should
do
or
if
there's
any
objection
to
it.
If
there's
no
objection
and
this
general
favor
for
doing
it,
then
I
can
proceed
with
it.
I'd
probably
own
it
anyway,.
A
A
I
I
kind
of
I
kind
of
don't
want
to
put
a
thumb
on
the
scale
and
say
what
I
think
before
asking
people
to
vote,
but
I
do
have
a
thought,
and
that
is
you
choosing
how
to
spend
the
time
that
you
have
working
on
stuff
probably
is
more
important
than
most
of
the
other
considerations,
because
this
feature
is
coming
out
anyway.
So
it's
sort
of
low
stakes
and
we're
talking
a
matter
of
a
few
months
but
yeah.
A
D
Also
see
a
green
yes
and
a
red,
no
reaction
under
that
menu
as
well.
A
A
D
For
sure
I
mean
that
this
is
something
that,
like
we
are
trying
to
deal
with
with
our
within
our
own
organization,
is
like
getting
people
to
upgrade,
because
we
we
occasionally
get
complaints
that,
like
xyz
feature,
doesn't
work
in
jupiter
lab
one,
and
our
reaction
is
invariably
well.
You
should
upgrade
then,
and
that
touches
off
some
awkward
conversations
I
see
mike,
has
the
handout.
E
I'm
very
happy
for
that
to
be
backported
to
3.4.
The
only
concern
would
be.
Can
we
make
sure
that
it's
as
easy
to
disable
as
possible,
possibly
without
going
to
the
terminal
and
saving
the
extension,
and
that
does
not
overlay
the
contents
of
the
cell
on
the
smaller
screens.
So
these
are
two
small
unresolved
issues
which
were
postponed
and
it
would
have
been
fine
to
postpone
them
with
the
original
plan
of
working
conduct
in
the
4.0
timeline,
which
is
possibly
longer.
D
D
So
otherwise,
if
we
backport
it,
we're
gonna
have
to
port
all
the
fixes
as
well
so
yeah.
I
I've
already
opened
an
issue
for
some
of
the
like
cell
toolbar
v2,
feature
requests
such
as
the
ability
to
like
collapse
and
expand
the
cell
toolbar
right
now
it
either
appears,
or
it
doesn't,
which
folks
are,
are
not
universally
in
love
with.
I
respect
that.
D
Okay,
well
I'll
continue
to
pursue
this
within
my
own
team.
I
appreciate
the
team's
time
I
did
want
to
talk
about
a
couple
of
other
issues,
though
one
is-
and
I
think
brian
is
on
the
call
here-
we've
had
a
request
again
coming
from
internal
to
my
company
is
about
customizing
the
can
start
kernel
option
in
the
notebook
extension.
Currently
that's
hard-coded
to
true,
and
we
want
the
ability
to
turn
that
off.
D
I
opened
an
issue
and
I
have
a
pull
request
about.
I
also
had
a
thing
in
jupiter
lab
server
that
was
approved
overnight
to
add
this
in,
and
I
just
wanted
to
make
the
broader
team
aware
of
that.
I
don't
know
if
there's
any
like
questions
or
if
people
are
curious
about
that.
D
That's
a
good
question
I
know
brian
is
on
the
call.
I
haven't
heard
anything
from
him,
yet
the
first,
the
first
whack
I've
made
at
it-
was
actually
to
do
it
on
the
front
end,
but
it
was
requested
to
put
it
on
the
back
end
and
put
it
in
page
config
instead,
so
I'm
I'm
open
to
doing
it
in
either
way.
I
see
brian
is
just
unmuting.
G
Yeah
hi
everyone
right
now.
Our
use
case
for
this
is
more
at
the
deployment
level,
so
an
entire
deployment
of
jupiter
lab
the
admin
would
want
to
set
this
and
it
would
be
done
in
coordination
with
other
things
like
swapping
out
kernel
managers
or
terminal
managers,
and
so
for
that
use
case.
I
was
thinking
more
of
a
command-line
config
file
setting
we're
open,
though
you
know,
if
it
does
make
sense
to
be
in
the
jupiter
lab
settings,
that's
also
a
possibility
and
we're
aware
that
the
admin
can
set
those
up
as
well.
B
Thanks
brian
yeah,
I
I
I
don't
have
any
strong
opinion
on
the
best
position
it
was
just
like.
I
was
surprised
as
usually
we
don't
put
much
settings
on
the
server
side
and
it
was
just
to
to
try
to
get
the
rationale
behind
it
so
yeah
if
it
makes
more
sense,
because
it's
a
related
to
some
deployment
and
nobody
objects
that
it
should
be
frontend
to.
Then
I'm
fine
with
what
was
proposed
differently.
G
I
would
love
to
get
jason
grout's
input
on
this.
I
know
he
spent
a
lot
of
time
on
the
kernel
logic,
so
I
I
don't
think
he's
here
today.
D
D
D
So
right
now
you
can
only
navigate
around
using
the
ui
controls,
but
matthias
hauser
suggested
navigating
around
using
something
similar
to
file
browsers
on
on
linux,
uis
windows,
mac,
where
you
can
kind
of
jump
to
a
path,
and
I
wanted
to
jump
down.
D
Josh
gayden
suggested
doing
this
instead
of
using
the
standard
jupiter
lab
dialogue
instead
putting
in
this
kind
of
inline
text
box
to
be
able
to
navigate
by
clicking
or
double
clicking
on
the
breadcrumbs
to
make
this
editable
type
something
in
hit
enter,
and
then
that
changes
the
directory,
and
he
had
some
some
questions
about
that.
What
do
folks
think
about
this
approach?
About
modifying
the
file
browser
to
have
this
kind
of
jump
functionality.
D
Yeah,
if
you're
trying
to
copy
paste
something
from
like
any
other
context
in
your
file
system,
it's
not
going
to
work.
D
It
looks
like
what
josh
is
requesting
is
that
it
normally
looks
the
same
and
then,
if
you
double
click,
you
get
this
this
text
box,
that
that
replaces
it
and
then,
when
you
enter
it,
reverts
to
this
assuming
you've
entered
something
valid.
Of
course,
if
you
enter
something,
that's
invalid
you'd
get
either
some
kind
of
error
or
we'd
ignore
it.
But
the
idea
is
that
if
you
don't
click
on
anything,
it
would
look
like
it
does.
Today.
A
D
A
Well:
okay,
there's
a
few
things.
I
think
this
definitely
needs.
Some
accessibility
thought
mike
shares
a
really
good
link
about
breadcrumbs
nick
points
out.
What
I
was
thinking
too
about
autocomplete
is
something
you'd
really
want
in
a
feature
like
this.
A
I
think
a
jump
to
directory
that
is
accessible
by
mouse
and
accessible
by
keyboard,
shortcut
and
accessible
by
file
menu
and
accessible
by
command
palette
is
probably
an
attractive
thing
to
offer.
The
question
of
how
that
manifests
in
the
file
browser
ui
itself
is
less
obvious
to
me.
There's
there's
a
lot
you
give
up
by
changing
user's
expectations
of
what
those
breadcrumbs
mean
and
doubleclick
itself
is
a
specific
like
hard
to
discover
thing
in
a
web
ui.
Although
we're
trying
really
hard,
we
have
other
doubleclicks
in
the
system.
B
G
It
sounds
like
the
ux
and
accessibility
ambiguity
at
this
point
is
significant
enough
that
we
would
probably
need
someone
to
really
dive
into
it.
Explore
those
dimensions
before
we
could
really
commit
to
a
particular
concept.
I
think
it's
an
interesting
question.
I
just
don't.
I
can't
see
through
all
the
ambiguity,
to
a
yes
or
no,
yet.
D
It's
it's
very
much
a
power
user
type
feature
in
my
opinion
like
on
the
mac,
you
can
jump
inside
a
file
picker,
but
you
have
to
hit
command
shift
g,
at
least
in
english
language
versions
of
mac
os,
which
is
not
exactly
something
like
someone
showed
me
that
must
have
been
10
years
ago
and
I've
I've
memorized
it.
But
there's
nothing
that
kind
of
cues
you
to
do
that
on
windows.
You
can
like
click
in
the
address
bar,
which
sort
of
works
like
a
like
a
web
browser
circa
20
years
ago.
A
D
A
A
If
you
begin
typing
a
path,
it'll
navigate
to
a
path,
so
the
command
palette
will
also
be
a
path
navigator,
modal.
So,
if
you're,
if
the
proposal
you
showed
us
is
specifically
about
the
file
browser,
that's
one
thing:
if
it's
just
about
a
quick
navigation,
that's
a
different
thing
and
they
might
have
different
answers.
D
D
But
I
I
we
may
have
to
like
negotiate
this
all
right.
Well,
I'm
gonna
add
some
comments
to
that
particular
item,
one
triple
two
one.
I
added
it
to
the
notes.
So
if
some,
if
other
folks
want
to
talk
about
it,
I
decided
to
bring
it
up
just
because
the
person
who
responded
to
it
came
up
with
some
really
interesting
concepts
that
I
figured
the
the
team
would
have
some
opinions
about
and
we
could
talk
about
that
for
hours.
H
Hey
guys,
so
I
have
updated
the
the
wrapper
for
the
database
history,
so
I
moved
the
updates
I
had
to
the
original
data
explorer
repository
and
I
reorganized
the
code,
so
it's
just
a
pre-built
extension
now
and
there's
a
binder
link
in
the
pr
as
well.
If
somebody
wants
to
try,
I
have
copied
vdar
and
max
on
that,
but
if
anyone
else
is
interested,
you
know
please,
please
feel
free
to
ask
some
feedback
here.
H
A
H
Posted
it,
oh.
B
I
Hi
everyone-
I
just
have
a
quick
update
on
jupiter
desktop.
I
released
the
version
332-1
and
this
upgrades
the
bundle
jupiter
lab
version
232,
and
this
version
also
comes
with
some
fixes
to
condy
environment
activation
problems.
A
F
I
Yeah
yeah.
Currently
we
are
providing
notification.
I
If
there
is
an
update,
available
and
user
still
has
to
go
to
github
releases
page
and
download
it
from
there,
but
I
am
working
on
some
changes
to
allow
us
to
do
automatic
updates
and
once
that
is
done,
what
will
happen
is
the
update
will
be
downloaded
in
the
background
and
you
will
need
to
click
a
button
to
like
restart
to
activate.
You
can
see
this
on
vs
code,
for
example,
so.
A
It
looks
like
I'm
max
and
I've
just
written
a
single
word.
That's
not
really
helpful.
Let
me
let
me
post
the
link
to
the
repo
that
I'm
discussing
so
last
summer,
one
of
the
teams
at
cal
poly,
who
worked
on
different
jupiter
lab
extensions.
A
They
did
an
extension
for
notifications
inside
jupiter,
lab
and
andrey
who
came
to
this
call
a
couple
weeks
ago,
he's
still
working
on
it,
and
this
touches
a
lot
of
parts
of
our
stack.
There
is
a
server
extension
and
there's
a
front-end
extension
for
jupyter
lab
to
display
notifications
and
some
pieces
of
this
functionality
belong
in
different
packages
if
they're
ever
to
be
in
core.
A
So,
for
example,
an
event
bus
very
much
like
this
one,
if
not
identical,
maybe
would
be
slightly
different
because
there's
already
prior
art
in
a
package
that
was
originally
thought
of
as
a
telemetry
package,
but
might
actually
be
more
fit
for
purpose.
As
the
general
purpose
event,
bus
exists
and
so
pieces
of
this
probably
should
be
in
jupiter
server
and
definitely
some
aspect
of
a
notifications.
A
Api
belongs
in
jupiter
lab,
so
I
just
wanted
to
re-introduce
folks
to
this
set
of
functionality
in
its
current
state
and
to
say
that
this
is
a
piece
that
I'd
like
to
work
on
this
year
and
I
think
it's
probably
like
it's
not
a
hard
requirement
of
shipping,
an
app
with
real-time
collaboration,
but
it
seems
like
pretty
useful
for
any
app
that
has
lots
of
people
doing
stuff
to
have
a
coherent
notification
strategy
instead
of
each
individual
plug-in
deciding.
How
am
I
going
to
tell
you?
A
I
updated
that
kind
of
thing,
so
I
guess
I
want
to
take
the
temperature
of
the
room.
I
asked
in
the
jupiter
server
call
last
week
and
they
were
on
board
with
having
server
endpoints
for
publishing
events
for
deleting
events
and
a
websocket
for
publishing
notifications.
A
But
even
if
that
exists,
we
could
not
ship
a
javascript
api
inside
jupyter
lab
with
that.
That
could
be
a
thing
that
gets
added
later,
but
I
wanted
to
ask
if
people
think
this
is
a
worthwhile
thing
to
bring
into
core
for
jupiter
labs,
for
if
the
answer
is
yes,
I'm
happy
to
work
on
it.
If
the
answer
is
no
I'll,
look
for
something
else,.
A
But
if
you
have
an
event,
bus
and
they're
all
subscribed
to
a
websocket,
they
would
all
receive
that
and
maybe
choose
to
update
on
themselves,
but
it
also
raises
the
question
of
if
I
have
four
notebooks
open
and
the
tree
page
open,
do
I
want
five
different
websocket
connections?
I
don't
I
don't
know
the
answer
to
that.
We
can.
We
can
solve
that
later.
A
Cool
thanks.
That's
it
for
my
update
mike
you're
up
next.
E
So,
regarding
the
discussion
about
editable
paths
in
the
file
browser,
I
noted
that
we
have
status
needs
discussion
label,
but
we
do
not
have
status
needs
design,
even
though
we
have
status
design
complete,
and
I
was
wondering
whether
that
would
enhance
our
triage
process
a
little
bit.
If
we
had
such
a
label
comments,
and
the
other
thing
is
that
I
opened
an
issue
to
discuss
how
we
could
bring
mark
down
into
the
new
settings-
editor
user
interface
and
three
options
listed,
and
maybe
you
have
other
ideas.
It
was
open
for.
E
A
E
A
Okay
cool,
so
we
have
one
additional
discussion:
point
from
federic
about
web
components
in
jupiter
and
along
with
a
demo
ordinarily,
we
haven't
figured
out
whether
we're
recording
these
or
not,
but
this
seems
kind
of
cool.
It's
going
to
have
a
demo
so
frederick,
unless
you
object,
I
propose
we
keep
recording
the
call
for
your
demo.
B
A
Cool
one
thing,
though,
before
we
launch
into
it:
if
there's
anyone
who
wanted
to
raise
something
in
the
agenda
and
didn't
now's
a
good
time,
because
if
not
we're
done
with
the
agenda,
we're
gonna
do
additional
discussions.
B
So
let
me
share
my
screen,
pick
up
the
right
one,
this
one
so
I've
presented
in
the
past
already
that
I've
been
working
on
a
toolkit
of
component
based
on
standard.
That's
what
component
or
custom
component,
so
the
main
difference
compared
to
react.
It's
like
it's
an
official
standard
and
what
happens
is
you
are
creating
also
component
like
react
view
and
stuff
like
that,
but
it
gets.
B
B
Everything
is
prefixed
in
this
toolkit
by
gp
for
jupiter
and
those
components
are
using
what's
called
shadow
dub,
so
the
real
description
of
every
component
is
actually
a
subtree
of
elements.
So,
for
example,
if
I
take
the
search
to
search
it,
there
is
a
label
within
the
search,
and
then
there
there
is
the
multiple
div
that
wraps
the
input.
The
input
is
there
and
there
is
also
a
clear
button.
That's
in
here,
so
everything
is
in
the
shadow
dump.
B
B
It's
just
locally
specified
the
only
thing
that
can
go
for
from
styling
point
of
view
from
the
outside
words
inside
the
shadow,
dom
are
css
properties
and
that's
cool,
because
that's
what
we
are
using
for
teaming,
so
that
helps
so
that's
the
structure
of
it
and
on
the
javascript
so
to
bring
like
what's
called
behaviors
on
the
on
the
component,
you,
you
will
define
a
class
that
define
the
behavior
of
this.
B
So
that's
that's
the
technology.
Let's
say
behind
it.
One
of
the
the
limitations
is
known
or
risk
it's.
You
cannot
register
multiple
times
the
same
new
new
tags
because
the
browser
won't
like
it.
B
So
the
the
work
I
have
done
lately
was
to
try
okay,
different
version
of
my
toolkit
to
see
what's
happening
when
I'm
doing
that
and
to
do
that.
What
I
did
with
some
colleagues
is
that
we
have
work
on
creating
some
some
extension.
So,
for
instance,
here
we
have
created
an
extension.
That's
a
bit
like
in
vs
code
is
a
search
through
the
tree
of
files
and
looking
for
matches,
so
that
extension
is
using
those
those
web
components,
but
they
are
using
a
different
version
of
the
library
and
that
can
be
seen
easily.
B
If
you
look
at
the
panel
here,
so
I
changed
the
panel.
This
is
the
latest
version.
The
accordion
panel
is
using
like
the
arrow
to
match
what
we
are
using
in
jupiter
most
of
the
time,
and
if
you
look
here,
we
can
see
that's
using
a
previous
version,
because
I
was
not
changing
the
default
button
that
were
like
this
plus
and
minus
thing.
B
So
the
good
news
is
that
it's
not
crashing
everything
works,
and
the
only
thing
is
that
if
you
have
multiple
extension
that
brings
their
own
version,
you
don't
know
at
the
end,
which
version
is
gonna
be
used.
So
that's
the
that's
the
conclusion
of
that
has
discussed
previously
if
this
made
it
to
core.
So
if
jupiter
app
itself
is
serving
this
library,
the
advantage
is
that
you
can
specify
to
webpack
that
we
want
to
use
only
one
version
in
the
federation,
and
so
we
will
ensure
that
we
have
a
strict
control
of
it.
B
Then
the
other
point
was
how
to
deal
with
reaction
like
how
to
deal
with
the
update.
So
because
it's
not
reacting,
it's
not
luminal
things,
it's
yet
another
thing,
but
we
need
to
be
honest
about
it.
So
the
idea
was
to
test
some
some
custom
element
and
to
see
how
we
can
use
both
mechanism
so
luminous
signals
and
what
they
are
using
in
the
library
that
I'm
used
using
here.
So
the
fast
library
from
microsoft,
so
in
the
fast
library
of
microsoft,
they've
defined
what
they
call
observable.
B
What's
an
observable
it's
so
this
is
the
code
I'm
showing
for
so
what
I'm
showing
here
in
the
the
demo
is
just
a
button
and
a
a
span
that
display
the
level
of
a
contour
and
every
time
you
click
it
will
increase
the
the
contour
by
one.
So
that's
quite
quite
easy,
and
so
you
can
either
do
an
observable
model
using
an
observable
from
fast.
So
that
means
you
use
a
decorator
and
you
say:
okay,
your
attribute
is
observable
and
by
doing
that,
every
time
you
use
it
within
the
template.
B
What's
happening
is
behind
the
scene,
the
library
it
will
keep
reference
and
it
will
say
okay.
This
is
an
observable,
and
so,
if
the
observable
in
the
in
the
html
template
is
changed,
then
it
knows
that
this
part
of
the
templates
need
to
be
updated.
That's
how
fast
is
working,
so
that's
kind
of
magic,
because
the
only
thing
you
you
need
to
do
is
to
to
decorate
your
attribute
and
you
get
it's
like
for
taking
a
parallel
with
luminol.
It's
like
you,
define
another
formula.
B
B
So
that's
what
I
have
done
here.
I've
done
one
model,
that's
using
fast
observable
and
I
have
done
another
another
model.
That's
using
the
inspiration
of
this
virtual
dumb
model
that
you
are
using
in
jupyter
core.
So,
basically,
when
you
inherit
from
this
this
model,
you
get
for
free
signals
that
state
change,
and
so
you
can
link
a
view
and
so
every
time
you
do
a
change
in
that
model.
You
can
trigger
that
signal
and
that
should
update
the
view.
B
So
that's
the
other
implementation
I
have
done
just
to
see
if
we
can
do
both
and
okay,
I
successfully
did
both
so
that's
cool,
because
that
means
we
could
also
introduce
those
component
and
link
them
to
some
part
of
the
code.
That's
using
signal
it's
possible,
I'm
not
saying
it's
easy,
but
it's
possible.
So
so,
basically,
the
what
needs
to
be
done
is
you
need
at
some
points
to
do
explicitly
what
this
decorator
is
doing.
Implicitly.
B
So,
when
you
are
doing
this
decoration
of
attribute
what's
behind
the
scene
is,
is
the
equivalent
of
saying
okay.
When
you
get
the
model,
you
are
telling
the
fast
system
that
you
are
tracking
the
modification
of
a
certain
properties.
So,
by
adding
this
line
and
when
you
are
setting
the
model,
you
need
to
call
a
notification
saying:
okay,
you
are
gonna,
notify
that
specific
property
has
changed
and
that's
how
we
can
hook
the
behavior,
like
the
reaction
of
luminous
signal
change
to
a
fast
change,
and
so
the
the
template
is
updated.
B
So
that's
pretty
much
it
for
the
the
demo.
As
you
can
see,
the
syntax
looks
close
enough
to
what
you
would
expect
like
using
view
or
react.
So
you
directly
put
your
html
when
you
want
to
add
a
listener.
You
have
to
use
the
hat
and
then
the
the
name
of
the
event
you
are
listening
and
then
you
you
place
the
callback
there.
There
are
specific
things
if
you,
for
example,
want
to
display
a
list
of
elements
in
an
array
but
yeah
that's
specific
to
the
technology,
but
everything
is
quite
well
organized.
B
There
are
also
something
a
bit
similar
to
react.
The
capability
of
saying
I
render
this
part
only
when
the
condition
is
true,
so
that's
also
available
in
that
technology,
and
one
thing
also:
it's
really
easy
to
hook
that
up
in
a
widget,
because,
as
those
elements
are
html
elements,
if
I,
if
I
open
the
index
file
where
I'm
instantiating
things.
B
That's
here,
so
what
we
are
just
doing
saying
is
like
we
are
creating
the
the
view
and
the
view
we
directly
say.
That's
a
widget
and
the
note
is
that
view,
because
it's
directly
in
html
element,
so
we
can
directly
create
a
widget
like
this.
So
that's
really
easy
to
place
it
inside
the
widget
for
integration
in
jupyter.
A
B
It
will
endure
in
this.
The
travel
is
what
the
standard
is
agree
upon.
Is
this
way
of
registering
new
tags
as
an
html
element,
but
you
don't
get
the
goodies
like
this
syntax
to
easily
add
the
listener
to
event
and
stuff
like
that.
If
you
want
to
do
that,
you
can
you
can
with
like
the
standard,
but
you
need
to
go
back
to
the
way
the
vania
gs
way
like
you're,
saying,
okay,
I
I
take
this
element
and
then
I
call
I
add
a
listener
too.
B
So
you,
you
are
the
main
thing:
if
we
don't,
if
we
think
we
will
drop
it
later,
is
that
you
you
will
be
dropping.
Those
helpers
is,
for
example,
the
reason
also.
So,
of
course,
the
microsoft
is
a
version.
Google
has
another
one,
that's
called
lit
and
it
does
the
same
thing.
It's
bringing
those
html
templating
things
that
has
lots
of
goodies
to
help
the
user
getting
more
endless
characters
in
the
javascript
side.
A
And
can
you
explain
a
little
bit
what
the
full
landscape
looks
like
between
some
luminal
widgets
that
are
pure
html
and
javascript
lumino
widgets,
some
that
are
react
and
some
that
are
fast
elements
like.
What's
the
what
is
a
good
state
of
the
code
base
in
terms
of
ui,
say
I
don't
know
two
or
three
years
from
now.
B
So
my
opinion
is
the
widget
are
too
complex
to
manipulate
compared
to
to
to
write,
correct
element
or
stuff
like
that.
I
think
the
the
life
cycle
of
them
is
tough
and
also
it's
very
hard
to
integrate
them
inside
other
technology
and
that
web
component
standard.
That's
one
of
the
advantages
for
free.
You
can
integrate
them
in
view
and
spelty
or
whatever
you
want.
The
only
trouble
for
now
is
react.
B
B
Yeah
mike
it's
the
you
can
wrap
them.
I
didn't
spoke
about
it,
but
you
can
wrap
them
in
inside
track
component.
So
if
I'm
showing
back
the
demo
so
on
the
the
difference
between
the
two
panel,
you
think
it
looks
the
same.
Actually,
it
should
look
the
same
because
that's
the
difference
here
on
the
on
the
top
is
directly
web
component
and
on
the
bottom,
the
those
are
the
same
component
but
rat
inside
react
element
and
so
basically
for
the
browser.
B
Those
are
the
same,
but
you
use
what
you
prefer
as
a
technology.
There
are
some
bugs,
like
the
more
obvious
one,
is
the
slider
the
position
of
the
level
and
yeah
so
that
that's,
I
think,
that's
why
I
really
like
this
this
technology,
it's
it's
because
it's
it's
as
it's
a
standard.
You
can
use
the
same
thing
and
and
apply
the
framework
you
prefer
and
so
for
the
long
term.
I
think
it's
the
best
and
also
because
they
come
with
encapsulated
styling.
B
I
think
it's.
I
know
I've
seen
some
questions
on
the
chat.
Some
people
say
that's
mean
you
need
to
yet
learn
another
thing,
but
I
think
you
could.
B
Reduce
the
amount
of
things
you
need
to
learn
in
the
sense
that
the
styling
comes
for
free,
like
your.
If
you
reuse,
the
the
toolkit,
the
teaming
will
be
for
free,
you
don't
need
to
figure
out
what
are
the
the
css
variable
for
most
of
the
the
style
will
be
positioning
and
layout.
So
that's
make
think
easier
like
for
the
extension,
for
example,
we've
been
working
on
the
the
quantity
of
code
you
write
within
the
extension
is
a
bit
less.
B
A
I
have
a
question,
so
would
you
want
to
bring
in
like
how
would
you
socialize
this?
Would
you
go
into
some
components
that
exist
already
inside
jupiter
lab
and
replace
them
with
fast
elements,
or
what
are
you
talking
about
an
extension
ui
toolkit
like
who?
What
which
user
interfaces
should
be
built
with
this
toolkit.
B
So
my
plan
in
mid-term,
that
means
couple
of
months,
would
be
to
open
a
pr,
that's
replacing
or
multiple
prs
that
start
replacing
some
of
the
ui
that's
existing
by
just
the
the
same
from
the
toolkit.
B
One
of
the
reasons
behind
is
they
come
with
accessibility,
behavior
right
in
and
also
they
will
allow
a
more
consistent
look
and
feel
throughout
the
the
the
core
application,
especially
because
if
people
want
to
change,
that's
something
that
for
now
it's
complex
like
if
you,
if
you
think
a
button,
the
change
in
the
color
when
you
are
doing
over
or
focus
on
the
button
is
not
good
enough.
If
you
want
to
improve
that,
you
will
only
need
to
change
that
in
the
toolkit
and
it
will.
B
B
J
I
mean,
I
think,
rather
than
going
after
everything
across
the
whole
ui
concentrating
on
going
after
the
settings
editor
to
where
we
had
a
a
you
know,
react
json,
schema
form
that
was
pluggable
that
was
overrideable.
J
B
J
Okay,
so
those
are
already
off
in
some
other
crazy
town.
You
know
doing
some
other
thing
that
has
nothing
to
do
with
anything
else
that
we
have
they're.
Just
you
know,
react
dom
baseline
stuff.
So
that's
that's
great!
That's
not
gonna!
Those
were
a
new
thing
that
was
added
because
they've
provided
value
at
this
top
level
thing.
So
if
you
offered
to
an
extension
author
extension
author,
here's
a
form
give
me
a
json
schema,
maybe
mark
it
up
a
little
bit
and
now
you
get
an
observable
model
off
the
end
of
it.
J
That's
that's
getting
the
job
done
for
sure,
so
yeah
that
that's
good
carrot.
I
mean
I
I
echo.