►
From YouTube: JupyterLab Team Meeting - October 5, 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/152
A
Okay,
hello
and
welcome
to
the
October
5th
Jupiter
lab
weekly
call.
Today
we
have
18
people
on
the
call
pretty
good
and
we
have
oh
look
at
that:
a
fairly
full
agenda-
that's
nice
too
great!
So
maybe.
A
I
just
wanted
to
see
if
there's
no
there's
I
was
gonna
say
we
just
I
think
have
finished
the
Jupiter
lab
software
steering
console
Representative
election,
but
I,
don't
think
I,
don't
think
the
person
who
administered
it
is
here
so
we'll
leave
that
for
an
announcement
later
it
might
even
still
be
running
because
it's
not
midnight
I'm,
not
sure
but
anyway,
for
now,
why
don't
we
get
started?
The
first
person
on
the
agenda
today
is
Jeremy,
but
he
says
he's
not
joining
the
call
and
he's
just
posting
his
updates.
A
B
Posted
a
nice,
my
binder
demo
of
Jupiter
notebook
7,
which
I've
been
playing
around
with
for
a
few
minutes,
and
it's
really
nice
but
I
noticed
maybe
a
little
bug
and
I'm
just
curious
where
to
post
that
book,
there's
a
link
to
a
pull
request,
but
that's
for
something
completely
different,
even
though
it's
right
there
in
that
list.
That's
for
the
oh.
A
A
Cool
okay,
so
yeah,
so
Jeremy's
posts
are
largely
about
the
raft
of
releases
that
landed
in
Notebook
7
for
a
pre-release
that
includes
all
of
the
recent
changes
you've
seen
in
the
Jupiter
lab
code
base
with
code
mirror
6
RTC
improvements,
the
virtual
notebook
rendering
it
also
includes
all
of
the
lumino2
stuff
we've
been
talking
about
and
because
notebook
7
is
composed
of
Jupiter
lab
components.
It
is
by
necessity.
A
Third,
in
the
order
of
releases
whenever
say
something
like
lumino
changes.
First
luminal
has
to
publish,
then
lab
has
to
publish,
then
notebook
can
publish.
So
this
is
the
first
notebook
7
pre-release
that
has
a
lot
of
new
things
in
it
and
is
distinctly
different
from
retro
Labs.
So
please
play
around
with
it.
A
If
you
are
interested
in
the
stuff
at
all,
and
also
it
looks
like
there's
a
bunch
of
different
places
to
talk
about
it,
there's
the
discourse
topic
about
it
and
there's,
of
course,
the
notebook
repo
in
case
you
find
anything
wrong
and
if
you
don't
want
to
install
it,
you
just
want
to
play
with
it.
Just
click
on
that
binder,
so
yeah
any
questions
or
comments
about
that.
A
Okay,
cool!
Well,
you
know
what
then,
before
we
get
fully
started
with
other
stuff.
Why
don't
we
it's
been
a
while
since
we've
done
this,
but
I
think
we
actually
have
some
new
people
on
the
call
and
how
about
we
pause
for
a
second
and
do
some
introductions
because
I'm
not
sure
everybody
knows
everybody.
A
So
with
that
being
said,
if
you
would
like
I
will
be
silent
and
cause
an
awkward
silence
and
you
can
jump
into
the
silence
and
introduce
yourself,
but
you
don't
have
to.
A
I
think
anyone
who
feels
like
other
people
may
not
know
them
and
wants
to
should
introduce
themselves
because
no
one's
new
to
me,
oh,
no,
really
the
difference
on
where
you
feel
you
want
to
what
whether
you
feel
you
want
to
do
an
intro
I
would
say
and
I
just
basically
don't
want
to
point
to
someone
and
say:
do
it:
okay,.
B
D
I'm
Isabella
sheeher
I
am
formerly
a
ux
UI
designer
I
care.
A
lot
about
accessibility,
just.
D
F
Hey
I
wanted
to
say
hi
again
to
everyone.
My
name
is
Andre
I
do
full
Stacks,
so
I
work
once
it
was
Jupiter
lab
buzz
buzz
with
all
parts
was
a
stack
and
yeah
hi
again.
Everyone.
A
I'll
introduce
myself
hi
I'm
Darian
I
work
primarily
on
the
front-end
side
of
things
in
Jupiter
lab,
but
I
guess
I
have
written
some
python.
It's
not
my
thing,
though,.
G
H
I'm
vedar
I
I
work
on
a
bunch
of
different
Jupiter
projects.
Open
source
I
also
help
run
the
Jupiter
deploy
internally
the
entire
company.
Mostly
lately,
it's
been
iPad
widgets
and
Jupiter
service,
stuff,
yeah
London
based.
G
J
G
J
I,
can
you
hear
me
yeah
I'm,
David,
I'm
I've
been
working
on
Jupiter
lab
recently,
especially
in
the
in
in
rtcy
collaborative
mode
and
I.
Also
work
at
one
stack.
L
Okay,
hi
me
too
I'm
working
at
one
stack
and
I
also
work
on
Jupiter
lab
and
the
Jupiter
Arabic
station
and
I'm
Nicholas.
M
I
Guess
that's
my
cue
hi
I'm
Steve
working
on
duper
a
long
time
as
well,
currently
at
mongodb,
doing
Jupiter,
stuff
kind
of
part-time.
C
I'm
Jason
Weil
I
work
at
AWS
and
I
do
work
on
Jupiter
lab,
currently
working
on
a
new
extension
and
I
work
with
Andrea,
piyush
and
David,
who
you'll
see
on
other
calls
as
well.
I
also
often
run
the
triage
call
for
Jupiter
lab
and
notebook
and
anyone
else
who
wants
to
triage
new
issues
in
their
projects,
which
is
Thursdays.
M
There
we
go
I'm
Alex,
I
work
at
IBM,
I
work,
mostly
in
the
front
end
on
Jupiter
lab
and
I
spent
my
time
between
working
in
core
Jupiter
lab
and
working
on
The
alira
Suite
of
extensions
that
adds
pipeline
support
to
Jupiter
lab.
N
E
Hi
everyone
I'm
Jessica
I,
am
a
student
at
Cal
Poly,
currently
working
on
some
Jupiter
stuff.
As
my
side,
project
I
want
to
say
hi
to
everyone.
P
I'll
go
I,
guess
I'll
go
since
everybody
else
was
so
gracious
I'm
Detroit,
my
pronouns.
Are
she
her
I?
Am
a
tech
newbie.
So
to
be
honest,
a
lot
of
the
things
that
y'all
said:
I
have
no
idea
what
it
means,
but
I'll
have
a
lot
of
questions,
I'm,
sure
and
just
very
happy
to
be
here
and
to
learn
more
I.
Guess
that's
it
for
now.
P
Oh
sorry,
I
do
want
to
say
that
I
have
learned
some
full
stack,
stuff
and
but
I
think
I
prefer
front
end
specifically
on
the
accessibility
side,
so
I'm
very
encouraged
to
hear
so
many
other
people
have
that
interest.
P
But
thank
you
very
much
for
your
time
and
I
hope.
You'll
have
a
good
one.
M
Yeah,
just
a
quick,
Dev
question:
the
answer
might
be
nobody's
tried.
This
and
I
have
to
keep
banging
my
head
against
it,
but
I
was
wondering
if
anybody
has
ever
tried
supporting
taking
a
file
from
the
file
browser
in
Jupiter
lab
and
essentially
drag
and
dropping
it
into
a
normal
HTML
file,
upload
like
form
input,
and
if
anybody
has
if
they
could
either
help
me
or
point
me
in
the
direction
of
their
code,
because
I'm
trying
to
do
that-
and
it
is
so
far
not
easy
for
me.
M
A
So
in
the
drag
events
there's
a
DOT
files
attribute
does.
Is
that
something
you
can
pass
along?
Oh
wait!
No!
That's
only
for
the
native
one
I
guess!
The
question
is
in
the
non-native
one
that
we've
implemented
within
lab.
Can
we
create
a
synthetic
event
that
has
that
files
attribute,
but
I?
Imagine
you've
probably
been
attempting
this
exact
thing
for
a
while.
M
Yeah
I've
only
been
working
on
this
for
about
a
couple
days,
so
I
haven't
like
actually
coded
those
edge
cases.
It's
been
more
of
a
brainstorm
in
that
sense,
but
yes,
I
did
look
at
those
apis
I
guess
you
call
them
like
the
different
fields
in
that
drag
event,
and
as
far
as
I
can
tell
there's
nothing
in
there
that
I
could
directly
just
pass
on
to
a
file
upload
that
would
enable
it.
M
A
So
what
I
would
try
is
use
the
Dom
API
to
create
an
event.
That's
a
drag
event
that
has
a
files
attribute
in
it
with
just
like
a
file
name
and
let's
see
if
you
can
just
create
an
empty
file.
That
does
that,
if
you
can,
then
maybe
you
pursue
all
right.
How
do
I
populate
the
content
of
that
file?
But
if
you
can't,
you
might
be
stuck
having
to
figure
out
a
way
to
work
around
instead
of
relying
on
the
native
apis,
but
I
don't
know.
Does
anyone
have
experience
with
this
stuff.
M
Yeah
I'm
I'm
trying
to
support
being
able
to
drag
and
drop
a
file
from
the
Jupiter
file
browser
into
a
standard
HTML
file
upload
like
a
file
input
that
I
don't
have
control
over
it's
in
a
web.
App.
That's
in
a
knife,
I
see
like.
Q
I
I
haven't
done
anything
like
that,
actually
yeah.
So
sorry,
another
well.
M
A
H
I,
don't
think
you
have
the
option
of
doing
like
a
deferred
thing,
so
I
delete
that
something
you
would
be
able
to
do
and
you
can
search
to
see
if
it's
possible.
But
you
know
if
it's
not
possible,
you
have
to
get
the
entire
file
contents
from
the
contents
manager.
When
the
drag
starts.
That's
a
bit
of
one
potentially
an
overhead
for
people
who
just
want
to
open
the
file
in
in
well
I!
Guess
it's
cached!
Oh
no!
We
don't
do
caching,
but
that's
a
ticket.
H
M
Yeah
I
actually
am
pretty
solid
on
the
dragon
drop
part
because
we
already
actually
have
that
implemented
in
elyra
because
you
can
drag
files
into
our
pipelines
directly
from
the
file
browser.
The
caveat
there
is
the
pipeline.
M
Editor
only
cares
about
the
path
because
it
is
eventually
run
on
the
same
server
as
the
file
exists
in
Jupiter,
so
it
doesn't
know
the
file
contents,
whereas
this
upload
needs
to
essentially
get
the
file
contents
and
that's
where
I
kind
of
got
stuck
because,
like
I
couldn't
figure
out
how
to
do
that
without
essentially
like
I
think
somebody
said
like
including
the
whole
contents
of
the
file
as
part
of
the
drag
and
drop
where
you
like,
actually
like,
essentially
buffer
the
whole
contents
of
the
file
in
I.
M
Guess
the
browser
but
I
have
a
few
ideas
that
I've
gotten
from
the
various
things
that
people
have
said.
So,
oh
there's
a
drag
and
drop
attachment.
Thank
you.
I'll.
Look
at
that.
Well,
that's
in
the
cells,
yes
I
believe
I
actually
already
did
look
at
that,
but
yeah
thanks
not
to
drag
the
meeting
on
too
long,
but
thanks
for
all
the
help.
A
Cool
I
think
just
the
hearing.
The
different
points
of
view
on
this
I
I
think
that,
given
the
arbitrary
size
of
files,
given
the
fact
that
you
might
need
them
when
the
drag
starts
and
all
it
to
me,
points
to
a
server-side
solution
which
just
looks
like
a
drag
and
drop
to
the
client.
But
actually
it's
the
server
doing
the
work.
But
you
might
discover
that
there's
a
clever
workaround,
so
I'd
be
curious
to
see
what
you
do.
H
Yeah
I
do
see
that
you
can
Define
URI
mime
types,
so
you
might
be
able
to
do
some
clever
things
with
that
by
having
a
URI
to
the
contents,
API
or
the
static
file
API
or
whatever.
M
It
is
yeah
that
that
was
one
of
the
ideas
that
I
had,
but
I
had
not
yet
attempted,
so
that
is
actually
probably
going
to
be.
The
first
idea:
I
try.
A
Q
M
I
did
not
have
any
control
over
the
destination
Source,
no
matter
how
much
I
wish
I
did
I.
M
It
there's
an
actual
up
file
upload
field
like
an
input
field
inside
the
iframe.
Q
Yeah,
but
you
shouldn't
be
allowed
to
do
it.
Programmatically
from
the
security
perspective.
M
I
see
what
you're
saying
yes,
that
could
be
a
problem.
Yeah
I
do
have
a
fallback
in
my
head,
which
is
when
they
try
to
drag
and
drop.
It
gives
them
a
nice
little
dialogue
that
says:
hey
download
this
and
upload
this,
which
is
not
the
best
ux,
but
it
is
a
fallback
to
prevent
confusion
of
people
trying
to
drag
and
drop
files
and
and
lumino
instead,
just
opening
that
file
in
a
new
section
of
your
UI.
G
D
Hello
yep,
it's
Isabel
again
I'm
here
with
my
usual
announcement
that
we
will
have
the
Jupiter
lab
accessibility
call,
even
though
really
it's
kind
of
all
of
Jupiter
accessibility.
Conversations
at
this
point
happening
after
15
minutes
of
this.
The
end
of
this
scheduled
call
that
was
a
lot
of
fumbling,
but
basically
it's
on
the
community
calendar.
It's
accurate
hope
to
see
some
of
y'all
there
linked
agenda
if
you're
interested
everybody's
welcome
to
that
so
yeah.
Thank
you.
O
Do
notebook
I
I,
updated
a
beer
I'm,
leaving
the
same
difficult
that
we
had
before,
where
we
are
not
using
collaboration,
we
have
the
collaborative
flag
to
then
I'm
gonna
win
they're
not
going
to
be
empty
and
actually
I'm
also
working
on
another
another.
G
Yeah,
it's
just
one
thing
that
we
would
like
to
add
it
so
from
an
end
user
perspective.
In
both
cases
they
should
not
really
change
their
experience,
because
I
could
have
already
go.
The
sale
is
going
to
be
added
to
the
back
end
in
that
priority
code.
You
retain
your
current
behavior.
G
However,
in
the
collaborative
mode
you
know
you
have
some
latency
right,
because
the
back
end
is
the
one,
the
cell,
so
it's
not
done.
One
transaction
and
therefore
we
actually
need
to.
You
may
actually
need
to
change
some
more
code
than
we
thought
mentioning
to
to
make
that
work,
and
we
wanted
to
sort
of
make
sure
that
this
is
going
to
be
like.
This
is
an
approach
that
people
like
before
putting
too
much
work
into
this.
G
G
So
that
so
the
issue
that
we
have
maybe
I
see
some
people
running
is
that
if
someone
deleted
a
CR
before
anyone
and
like
it
was
the
last
year,
everybody
every
collaboratory
would
add
the
new
Services,
they
would
find
themselves
with
an
empty
notebook
and
me
it's
a
corner
case
that
actually
occurs
but
of
and
who
yeah
that's.
The
fix
that
we
proposed.
N
A
So,
regardless
of
what
user
experience
we
end
up
going
with,
this
has
to
be
resolved
because
otherwise,
you
might
put
everyone
in
a
collaborative
notebook
in
a
broken
state.
Is
that
right.
A
B
K
A
A
All
right,
Vidar,
you
are
up
next.
H
Yeah,
so
for
me
it's
just
I
found
a
specific
issue
or
Corner
case
Performance
Corner
case
where,
if
you
have
a
very
large
source
file-
and
you
have
this-
this
contextual
help
the
inspector
open.
H
It
will
freeze
the
UI
for
a
long
time
as
it
loads
that
in
it's,
not
the
network
thing
it's
the
rendering
of
the
text,
my
model,
because
it
has
all
these
NC
color
codes
in
it.
So
it
creates
a
lot
of
span
elements
and
then,
when
it's
trying
to
create
these
links,
fix
the
any
automatic
links
to
URLs
in
that
text.
H
It
just
takes
forever
it's
not
off
tomorrow.
So
you
know
that,
hopefully
that's
something
I
can
fix
separately,
but
there
is
also
a
more
General
issue
there
that
if
a
specific
you
know
asynchronous
task
takes
a
long
time
since
the
browser
is
still
single
threaded,
it
will
still
freeze
the
UI.
H
H
H
J
C
H
That
that
that's,
then
you
know
specific
solution
to
this
problem
which,
but
then
the
generic
part
would
be
a
clever
or
easier
way
to
use
web
workers
from
lab.
If
there
is
anything
needed
there.
A
H
And
what
it
does
is
that
it
it
takes
the
document
and
it
converts
any
ANSI
color
codes
to
span
elements
with
with
this
the
correct
color,
and
then
it
uses
the
HTML
sanitizer
to
remove
any
HTML
that
it
only
allows
span
elements
and
then
it
passes
through
it.
To
looking
for
links
to
cover
to
to
HTML
link
objects
is.
H
Still
looking
for
links,
but
as
far
as
able
to
tell
just
it
is
just
if
you
have
a
lot
of
large
documents,
a
lot
of
color
codes,
there's
like
a
while
loop
that
iterates
over
tens
of
thousands
of
elements
right.
G
A
G
H
Anyway,
so
this
feature
was
added
in
the
previous
pull
requests,
so
I
don't
want
to
revert
that,
but
we
can
have
that
discussion,
but
the
main
and
I
I,
no
I
agree
that
there's
many
ways
to
fix
that
issue.
That
isn't
a
problem.
I
just
wanted
to
know
if
there's
another
more
clever
way,
because
there
is
an
issue
with
JavaScript
that
everything
is
happening
on
the
main
thread,
and
you
know
if
we
were
writing
something
that
we
were
comfortable
with
having
the
software
phone
home.
H
The
Telemetry
you
would
be
interested
in
is
like
every
time
you
have
a
long
running
task,
send
a
message,
a
home
saying,
like
this
user,
did
this
this
and
this,
and
now
they
have
the
UI
was
stuck
for
two
seconds
so,
but
we
don't
do
that.
So
we
don't
really
have
a
good
way
of
capturing
that
those
kind
of
things.
B
Projects,
notebook
virtualization,
I
mean
that
seems
relevant
to
this,
because
then
you
wouldn't
you
wouldn't
look
at
the
cells.
These
tens
of
thousands
of
Links
at
all.
H
A
A
A
H
Yeah,
so
that's
a
generic
ux
design
right
that
the
main
thread
is
at
some
point
only
for
rendering
and
then
all
other
work
should
be
happening
in
separate
threads
right.
That's
for
for
desktop
UI
applications.
There's
been
a
pattern
for
a
long
time,
but
you
have
a
separate
Loop
thread
running
the
message
Loop
for
just
for
for
rendering
yeah
it's
a
little
harder
in
the
browser
so
and
look
a
little
bit
more
in
web
workers
to
see.
R
For
for
web
workers,
there
is
an
extension
which
would
benefit,
and
that's
the
spell
checker.
So
there
is
one
spell
checker
package
and
npm,
which
is
very
good,
but
we
have
not
adopted
that
because
of
the
potential
overhead
of
her
purpose.
If
there
was
a
luminal
package
which
made
communication
easier
or
took
some
of
the.
R
Data
taxation
developments
need
to
know
in
addition
to
the
standards
web
workers.
Api.
Sorry,
I,
don't
know
whether
this
is
something
that
a
small
benefits
than
six
languages
and
for
the
very
specific
issue
that
you
mentioned,
of
the
rendering
that
could
potentially
just
be
split
up
into
many
promises,
so
that
would
still
be
on
the
same.
H
E
H
B
H
R
H
H
H
Yeah
as
far
as
I'm
able
to
tell
from
what
I'm
reading
online,
at
least
so
it's
only
like
things
like
the
the
set
timeout
and
and
request
animation
frame,
and
these
things
like
this
splits
it
up
into
a
macro
task.
That
then
kind
of
resets.
R
I
think
I
think
I
run
into
a
similar
issue
before
if
there
is
an
issue
where
I
could
comment
on
GitHub,
if
I
find
an
answer
to
that
are
going
to
see
the
link.
A
Leader,
this
micro
task
business
I,
ran
into
it
with
that
lumino
PR
about
this
exact
thing
in
messaging
and
yes,
that
is
why
a
promise
based
async
solution
for
the
messaging
in
lumino
works,
but
a
promise
based
solution
for
polling
doesn't
because
polls
are
promise
based,
and
so
you
can't
basically
cut
ahead
in
the
way
you
can.
If
you
rely
on
promises
for
the
asynctness,
then
you
can
have
the
callbacks
also
be
relying
on
promises
is
what
I
discovered.
K
Then
the
other
announcement
is
that
there
has
been
a
new
patch
release
for
3.4.
The
change
lock
is
linked.
In
the
note.
The
main
highlight
is
that
we
have
reverted
the
default
tab
switch
method,
so
we've
got
four.
We
have
introduced
what
we,
what
I
expected
to
be
a
more
efficient
way
of
switching
tabs.
It
works
pretty
well
in
Firefox,
but
actually
in
Chromebase.
So
that
means
Chrome
chromium
Edge.
K
As
soon
as
one
of
the
tab
has
an
iframe
or
is
a
PDF
viewer,
I,
don't
know
if
it's
on
the
way
from
too
something
that's
of
that
nature,
then
the
the
UI
is
broken
and
the
user
cannot
switch
again
and
the
UI
is
free
Frozen.
So,
as
the
modification
was
not
enforced,
but
it
was
only
settings
with
a
new
default
value
where
it
was
reverted
to
use
the
old
way
of
doing
things
so,
like
Firefox
user
can
still
benefit
of
that.
K
If
they
want
and
we'll
see
in
the
future,
maybe
with
windowing,
we
can
reach
a
better
execution
time
for
time.
Switching
for
tab,
switching,
let's
see
and
as
a
follow-up
of
that
I
have
created
a
branch
for
3.5.0,
almost
all
PRS
that
are
for
that
new
version
are
merged.
The
only
one
that's
missing
is
the
bump
of
the
lumino
package,
so
a
new
version
of
liminal
one
has
been
released.
K
The
only
difference
is
the
addition
of
deprecation
tag
in
the
documentation
to
notify
user
that
some
API
are
going
to
change
in
nimino2,
and
so
that's
the
only
PR
that's
missing
in
to
be
merged
and
then
I'll
do
a
better
release
for
3.5.0
should
be
tomorrow
and
because
it's
a
new
minor
version
will
go
for
a
one
week
time
before
moving
to
release
candidate
and
another
one
week
for
doing
the
final.
If
everything
is
is.
Q
K
And
so
that
was
it
for
the
announcement,
then
I
have
to
question.
There
are
still
the
notification,
PR
that's
open
and
if
somebody
has
time
to
review
it,
that
I
will
be
very
grateful,
especially
not
like
DUI.
It
will
be
nice,
but
especially
the
new
API
that
has
been
set
up.
It
will
be
very
nice
to
to
review
that
part
and
and
I
get
the
question
about.
Let
me
know
two:
is
it
time
to
move
to
release
beta.
R
A
About
your
notifications,
PR
I
will
take
a
look
on
Friday
and
Andrew
is
on
the
call.
If
he
has
time,
maybe
he
might
want
to
take
a
look
as
well,
because
this
is
stuff
that
he's
worked
on
as
far
as
lumino
2
I
wanted
to
ask.
If
is
piyush
on
the
call
today,.
A
Off
today,
okay
Mike,
do
you
have
any
idea
where
the
what
the
current
state
of
affairs
is
with
the
menu
bug
fix
the
performance.
R
Yes,
I'm
pushing
kpr
today
night
or
plural,
and
that's
actually
my
next
point,
I,
don't
know
if
you
want
to
move
to
that
foreign.
R
So
I
I
created
this
wonderful
question,
Jupiter
lab,
which
removed
six
Styles
and
made
many
existence
of
those.
Six
CSS
rules
meant
that
the
performance
of
operating
over
the
menu
was
great
detriment.
R
It
was
quite
environmental
to
Performance
in
Chrome
browsers,
and
this
is
because
browsers
have
different
engines
for
the
calculation
of
styles
and
validation
of
styles,
and
the
chromium
is
not
the
best
one.
So
the
problem
is
the
screen
which
is
an
older
plug,
which
is
slower
than
Firefox
system,
and
when
the
browser
matches
CSS
role,
it
actually
goes
the
other
way.
R
It
will
invalidate
and
go
over
every
single
element,
which
is
what
is
causing
performance
issues
and
for
menu
and
some
other
components
of
typical
app,
so
part
of
that
is
fixed,
but
there
are
other
styles,
so
so
the
problematic
styles
are
removed
when
it
comes
to
the
new
hovering.
But
there
are
other
styles,
which
has
specific
problems,
but
they're
really
addicted
to
that.
I
will
be
fixing
some
other
issues
with
the
menu
not
trying
to
do
styles
and
a
generic
solution.
R
R
I
know
that,
with
web
components
exploration,
it
was
probably
someone
looked
at
it,
and
my
question
is:
how
do
we
feel
about
adding
Shadow
Dom
to
aluminum
widgets
so
that
we
can
later
explore
explore
switching
people
for
specific
components
into
the
lab
and
I
already
did
an
experimentation?
And
it's
like
six
lines
of
codes.
It's
just
allowing
the
coupling
the
widget
node
from
the
widget
attachment
nodes
in
the
aluminum
widgets,
and
then
we
can
switch
any
single
nominal
widget
Dom
via
transitive.
K
Yeah
my
ID
with
the
Jupiter
toolkit,
was
to
move
in
that
direction
in
lumino,
I
wasn't
sure.
If,
in
the
future,
we
will
like
keep
widget
and
create
something
that
is
like
widget,
but
we've
Shadow
them
like
to
a
little
abuse
move
for
transistor
transition
like
that's,
probably
requires
some
some
more
testing
but
I.
Think
it's
a
it's
a
good
move
to
do.
K
R
So
I
actually
made
a
simple
look,
which
runs
every
second
and
looks
up
the
Styles
style
sheets
attached
to
the
packs
and
copies
them
to
the
shuttle
which
is
problematic
because
we
are
copying
their
style
sheets
one
by
one,
but
it
keeps
them
in
sync.
So
that's
that's
a
very
dirty
solution.
What
I
bought
would
do
instead
is
to
expose
an
API
to
ads
Styles,
so
that
would
be
a
new
addition
to
aluminum
the.
K
K
Actually
yeah
and
the
other,
and
the
other
trick
is
to
be
clever
a
bit
slot.
Sorry,
it's
getting
quite
technical,
but
so
slot
in
Shadow
Dom
is
a
way
for
user
to
insert
triggering
in
placeholder
that
the
we
left
inside
a
shadow
Dom
tree
and
those
are
also
started
from
external
styling
rules.
A
To
answer
Fred's
question
about
going
to
beta,
then:
do
you
expect
the
Contours
of
the
solution
to
modify
the
apis
in
any
way
or
are
they
internal
like?
Should
we
release
a
beta
or
not.
R
So
all
the
API
candles
is
literally
I
think
a
new
data
for
so
so
when
there
is
a
note
which
is
an
HTML
element,
there
will
be
another
node
which
is
attachments
built,
which
is
also
an
HTML
element
by
default.
R
But
if
the
widget
is
confused
to
shuttle
from,
then
its
shadow
will
work
instead,
and
so
the
world
is
a
subset
of
HTML
elements,
so
that
advice
would
be
to
anyone
who
uses
lumino
and
attaches
widgets
to
the
website
to
switch
from
using
widget.notes
to
widgets
Docs
attach
themselves
so
that
we
can
attach
the
shadow
World
instead
of
the
the
note
itself,
so
that
it
is
a
kind
of
a
public
API
change,
but
not
in
breaking
one
is
more
like
recommendation.