►
From YouTube: JupyterLab Team Meeting - 28 June 2023
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/170
A
We've
got
how
many
people
on
the
call
17
people
participating
on
the
call
right
now
and
we've
got
an
agenda
on
hack
MD
and
it
looks
like
Frederick
is
first
up
on
the
agenda.
Go
ahead,
fritter.
B
Hello,
everybody
I
first
want
to
congratulate
Eric
Gentry
for
officially
becoming
a
Jupiter
content
member.
So
thank
you
welcome
to
the
to
the
team
and
thanks
a
lot
for
all
the
contribution
you
have
done
so
far
and
we'll
do
it
in
the
future.
I'm
sure.
Thank.
B
But
then
I
just
copy
paste.
The
list
of
repository
we
I
briefly
spoke
about
last
week
for
cleaning
up
the
Jupiter
app
keep
up
organization
I
think
we
can
delay
that
for
like
additional
discussion
and
stuff
like
that
and
go
to
the
other
point,
and
maybe
finish
by
that
I.
Don't
think
what
do
you
think
Jason.
A
Sorry
I
was
trying
to
figure
out
how
to
unmute
myself
here,
I'm,
okay,
with
like
proceeding
with
the
archiving,
because,
as
someone
noted,
if
we
archive
something
in
Arrow,
we
can
always
like
recover
it
later
on
right.
D
A
A
two-way
door
is
people
at
my
company
like
to
say,
and
that's
what
Paul
evenoff
said
as
well.
I
mean
I
meant
to
cite
him
for
that.
B
The
question
also,
it's
probably
about
the
message
we
should
set
on
those
repository
so
like
the
the
good
practice
is
to
add
on
the
ring
me
that
this
represents
this
repository
is
going
to
be
archived.
Should
we
just
say
that
or
should
we
like
give
a
reason
for
archiving
them?
What
can
be
different
from
one
to.
A
The
other
yeah
I've,
seen
some
repositories
become
archived
in
the
past
and
what
I've
noticed-
and
this
isn't
just
a
Jupiter
thing-
is
to
say
hey.
This
is
no
longer
maintained
if
you're
interested
in
this
project,
please
go
here
as
a
kind
of
alternative,
so
I
like
the
idea
of
just
saying,
like
we've,
we've
archived
this
repository
because
it's
no
longer
being
maintained
if
you're
interested
in
this
technology,
please
follow
up
in
like
Jupiter
lab
or
on
team
Compass,
or
something
like
that.
It
may
vary
from
from
repo
to
repo
I'm,
not
sure
I'm
I'm.
B
B
It
was
an
external
extension
for
Jupiter
rap
2
and
because
we
officially
support
two
major
version:
It's,
that's
probably
the
reason.
The
the
repository
was
not
archived
but
know
that
Jupiter
4
is
out,
then
that
one
is
an
easy,
an
easy
one,
I
think
and
for
the
others
there
are.
A
couple
of
them
are
experiments
like
xkdc
X,.
A
B
And
the
commanding
one
that
Explorer
and
stuff
like
that,
so
probably
what
I
can
propose
to
do
is
maybe
open
an
issue
on
all
of
them
and
giving
a
period
like
something
like
two
weeks
period:
And.
If
there
is
no
nobody
bumping
on
or
reacting
on
that
issue,
then
updating
the
ring
me
saying:
okay,
we
are
archiving
this
because
there
is
no
no
activity
and
probably
linked
to
the
team
Compass.
If
people
wants
to
to
reopen
them
and
don't
have
rights
for
doing
so,
yeah.
A
I,
like
the
idea
of
linking
back
to
team
Compass
issue
200,
so
that
the
message
comes
across
as
saying
we
decided
to
do
this,
not
I,
Frederick
and
unilaterally
doing
this,
like
we
had
a
discussion,
we
agreed
to
do
this
I
I
sense
consensus
to
move
forward
with
the
archiving.
B
Okay,
so
I
I
write
that
down
and
I
take
the
action
to.
A
Okay,
so
those
are
yeah,
so
Frederick
I'll
open
the
issue,
so
you
can
take
that
as
the
action
item.
Thank
you
also
Paul,
who
I
don't
think
is
on
the
call
right
now
for
your
comment
on
that
as
well.
A
Okay,
so
Frederick
will
take
the
action
item
on
that.
Fernando
has
another
item
on
the
agenda
about
a
file.
Corruption
bug
go
ahead,
Fernando.
E
Yeah
I
I
filed
it
and
thanks
so
I
saw
you
300
Jason,
but
I
kind
of
wanted
to
flag
it,
because
I
think
that
data
corruption,
bugs
should
be
escalated.
Pretty
rapidly
I
mean
it's
I,
I,
don't
know.
If
I
can
reproduce
it,
it
happened
immediately.
I
posted
the
steps
to
reproduce,
but
the
fact
that
the
editor
is
literally
like
corrupting
a
file
as
you
edit.
E
E
It
literally
started
a
brand
new
session.
I
had
rebooted
my
computer
and
then
all
of
a
sudden
I
make
a
brand
new
markdown
file
and
the
the
name
on
the
tab
is
the
name
of
the
last
markdown
file.
I
had
been
editing
before
I
rebooted,
my
computer,
so
there
was
some
weird
bit
of
stale
State
I,
don't
know
that
that
one
can
lead
to
data
loss,
but
it's
just
that
anything
that
has
to
do
with
data
inconsistencies.
I
think
needs
to
be
like
escalated.
E
We
can't
be
losing
people's
data
and
we
have
a
long
history
of
destroying
people's
data
with
with
the
RTC
Machinery.
I
would
like
to
kind
of
get
away
from
that
reputation
and
the
one
with
the
editor
it
was
really
bad.
Like
I
was
literally
I
was
working
on
my
taxes.
Late
I
was
trying
to
add
it
to
CSV
file
and
I'm
like
what's
going
on,
am
I
like
fingering.
E
This
and
I
miss
typing
on
I
was
backspacing
over
these
dollar
signs,
so
a
column
on
a
CSV
would
be
like
numbers
to
load
it
into
pandas
and
then
all
of
a
sudden
like
wait
a
minute.
This
is
wrong
and
it's
literally
corrupting
like
as
I
edit
the
file
it
corrupts
it.
It's
just
plain
out:
mised
the
data
I,
don't
know
how
the
editor
can
do
that,
but
it's
really
really
bad,
like
the
fact
that
an
editor
like
literally
corrupts
a
file
as
you
edit.
It
should
just
that
should
not
be
possible
in
Jupiter
lab.
A
So
yesterday
morning,
in
triage
we
saw
a
bug
come
in
and
I
I
just
pasted
it
into
the
into
the
chat
14752,
which
is
that
the
user
on
windows
with
chrome
is
reporting
that
markdown
code
blocks
are
getting
corrupted
when
they're
pasting
this.
This
is
likely
not
the
same
issue
Fernando,
but
it
is
similar
in.
A
Is
that
you
can
see?
Unfortunately,
the
video
is
two
minutes
long.
It
really
could
have
used
some
editing,
so
it's
a
little
hard
to
follow.
But
long
story
short.
Is
that
pasting
in
some
of
this
information
causes
like
the
markdown
to
render
without
some
of
the
data
which
is
a
little
concerning
I,
don't
think
that
they
are
using
RTC
because
they're
using
Jupiter
lab
4-0,
and
for
that
version
they
need
to
actually
install
a
separate
package
to
use
RTC.
E
Yeah
I
I
couldn't
replicate
the
problem
without
RTC
on,
though
I
didn't
do
a
huge
amount
of
testing
it
would
be,
and
I
I
was
busy
and
I
was
traveling,
so
I
didn't
I,
also
didn't
test
with
multiple
browsers,
I,
don't
know
it's
browser.
Specific
I
was
on
Firefox
when
it
happened,
but
it
did
happen
like
three
times.
In
fact,
when
it
first
happened,
I
thought
it
was
me
and
then
later
when
I
realized
what
was
going
on
I
made
a
separate
test
case.
I
made
a
copy
of
the
files
I
anonymized.
E
The
data
was
like
my
banking
data
and,
and
it
happened
again,
so
it
was
pretty
easy
and
it's
like
a
it
can
make
it
happen
with
a
three-line
file,
so
I
don't
know
how
to
debug
it,
but
I
was
just
hoping
to
flag
it
for
the
team,
because
I
think
data
corruption
is
is
really
not
something
we
should
be
having
out
in
the
open
and
sorry
Darian
about
it.
E
E
Yeah,
let
me
try
well
I,
got
it
with
Mamba,
but
but
or
or
maybe
it
was
with
Pip
yeah
I
will
give
it
a
try
and-
and
you
guys
can
carry
on
with
other
things.
It'll
take
me
like
a
minute
or
two
yeah.
E
Me,
let
me
try
I'll,
do
that,
while
we're
talking.
A
Okay,
any
other
questions
comments
about
this
one
I
I
think
like
the
inconsistencies
and
file
names,
we've
seen
some
like
file,
name
issue
bugs
and
I
agree
that,
like
you
know,
an
editor
should
fundamentally,
you
know,
preserve
data
and
save
files
with
a
particular
name.
These
are
pretty
fundamental
things,
so
anybody
with
the
availability
and
knowledge
should
please
take
a
look
at
these
as
soon
as
possible.
F
You
will
definitely
put
people
on
this
right
away.
Thanks.
A
G
I
gotta
find
an
issue
for
this,
so
I've
been
spending
all
of
my
time
in
notebook
and
one
of
like
I
work
for
hours
a
day
in
a
notebook
right,
so
I'm
a
daily
notebook
user,
and
there
have
been
some
major
inconsistencies
with
the
editing
experience
between
I
I,
it's
in
Jupiter
lab
also
and
notebook
version.
Seven.
So
I!
G
Imagine
it's
with
the
new
editor
changes,
but
if
folks
are
going
to
be
upgrading
from
MB
classic
and
then
using
the
new
notebook
I
think
there
should
be
some
more
consistent,
editing
experience,
I've
started
just
a
little
running
list
of
them
and
I'm
gonna.
Add
it
to
the
notes,
but
I
think
it
would
be
really
helpful.
Anybody
who
is
a
notebook
user
spends
time
in
notebooks
and
has
some
sets
of
expectations.
G
It
would
be
good
to
get
y'all's
feedback
on
the
experience
that
you're
having
and
where
some
of
those
discomforts
are
happening
like
in
the
last
notebook
meeting.
Some
of
the
things
I
mentioned,
Eric
related
to
but
I
think
some
other
people
might
find
some
stuff
and
it
will
probably
be
really
good
to
get
eyes
on
the
editing
experience.
G
Cool
has
anybody
else
run
across
any
inconsistencies
that
they've
seen
with
the
keyboard
or
is
am
I
the
just
making
things
up.
A
We've
gotten
quite
a
few
bugs
that
have
come
in
since
Jupiter
lab
4
came
out
about
keyboard,
shortcuts,
changing
or
no
longer
being
available.
A
One
key
thing
to
keep
in
mind,
given
the
international
nature
of
both
this
call
and
our
user
base
is
try
to
avoid
using
shortcuts
that
depend
on
symbols,
because
we
just
had
a
bug.
For
example,
I
didn't
know
this
that
in
the
Latin
American
keyboard,
a
forward,
slash,
doesn't
have
a
key
by
itself.
It's
shift
7
on
a
Latin
American
keyboard,
so
anything
dependent
on
that
is
problematic,
I
think
on
a
German
keyboard.
A
A
But
if
you
use
letters
a
to
z,
those
are
generally
going
to
be
okay,
but
we
we
had
a
bug
come
in
from
somebody
who
had
one
of
those
trendy,
60
keyboards
like
like
one
of
these,
but
but
cooler,
because
it
has
like
a
rainbow
LED
on
it
that
they
lost
the
ability
to
use
a
key,
a
keyboard
shortcut
because
they
were
you
it's
now
a
key
that
is
not
present.
A
It's
one
of
the
40
keyboard
keys
that
are
absent
on
there,
so
we
we
have
seen
a
few
of
those
I,
usually
tag
them
with
accessibility.
So
if
you
click
on
the
tag
accessibility,
you
should
see
that.
G
So
I
think
that
these
are
ux
things
right,
because
keyboard
shortcuts
aren't
in
the
scope
of
accessibility.
There's
it's
more
that
you
should
be
able
to
navigate
with
the
primary
Keys
but
yeah.
The
trouble
is
so
yeah
I,
don't
know
how
to
I.
Don't
think
these
are
I
think
these
are
ux
issues
they're,
not
specifically,
accessibility
issues,
so
that
would
probably
be
worth
labeling
there.
G
So
yeah,
I,
guess,
yeah
you're
right
I've
noticed
that
some
keyboard
shortcuts,
like
Ctrl
D,
doesn't
work
some
common
ones
that
become
normal
sort
of
things
in
the
editor
experience
that
you
don't
even
remember,
you're
doing
so,
I
guess
it's
a
matter
of
reporting,
keyboard
shortcuts
and
then
some
of
the
inconsistencies
with
the
cursor
but
yeah.
If
anybody
can
put
eyes
on
that
or
collate
some
of
those
issues,
maybe
I
don't
know
how
to
organize
that.
A
I
also
want
to
give
a
shout
out
to
the
people
on
The
Notebook
team
for
spinning
out
some
of
the
issues
in
6800
into
separate
issues.
We
saw
those
at
triage
yesterday
and
that's
really
useful
to
identify
like
high
priority
High
Velocity
issues,
especially
ones
that
you
know
more
complexity,.
C
Yeah
I
think
that's
been
mostly
Italian's
work,
so
kudos
to
Tony
on
that.
G
Yeah
well,
let's
talk
about
that
then
cool.
So
there
is
a
mega
issue
and
it's
really
neat
because
we've
actually
been
chipping
away
at
these
things.
One
of
these
became
a
good
first
issue
and
we
got
some
contributors
like
in
a
short
period
of
time,
but
Andre
converted
this
larger
issue
into
a
milestone,
and
these
are
a
thing
family
and
man
yeah.
G
If
we
get
these
things
done,
then
I
like
to
repeat
this
quote:
notebook
7
is
better
than
Jupiter
lab
and
what
should
be
recommended
to
persons
with
disabilities
and
just
needs
a
few
small
fixes
to
be
fully
supporting
of
low
vision,
colorblind
and
ambulatory
users.
So
this
we
won't
have.
We
won't
be
able
to
support
yes,
everybody,
but
this
will
be
a
good
start
here.
So
if
folks
can
take
a
look
at
this,
some
of
these
I
think
some
people
have
some
better
experience
than
others.
G
Then,
if
you
can
like
take
a
look
at
some
of
these
things
and
add
any
knowledge
that
you
might
have
I'm
not
asking
people
to
solve
it,
I'm
just
asking
people
to
put
input
in
places
that
they
know
stuff.
This
issue
is
filed
in
Notebook,
but
a
lot
of
these
things
are
in
Jupiter
Labs.
Some
of
them
might
be
in
lumino,
so
any
more
information
that
we
can
get
into
solving
these
issues
and
triaging
these
issues
would
be
super
duper.
Great
I
did
want
to
talk
about
two
issues:
yeah.
G
Well,
let's
talk
about,
there's.
Definitely
one
yes,
I
wanted
to
talk
about
two
issues.
One
of
them
is
not
in
this
actually.
So
the
first
issue
that
I
wanted
to
talk
about
is
over.
In
Notebook
it
says
that
cell
actions
are
not
announced
to
the
screen
reader.
Okay,
so
in
the
Aria
standard,
there's
Arya
live,
which
is
a
updating,
an
alert
system
for
a
screen
reader.
G
So
with
Aria
live.
You
can
set
certain
kinds
of
settings
that
when
the
announcement
should
be
made,
but
ultimately
what
we
need
is
we
need
a
visually
hidden
output
somewhere
in
the
cell,
so
that
when
a
cell
completes
or
when
a
sales
executed
will
say,
sellers
sell
whatever
is
executing
and
then,
when
it
completes
it
says,
sell
whatever
has
finished
completing
or
there
was
an
error
so
like
it's.
G
Basically,
we
basically
need
to
be
able
to
high
visually
hide
these
outputs
and
then
they'll
get
announced
to
the
screen
reader,
and
this
is
a
thing
that
nobody
else
is
necessarily
doing,
but
one
of
the
questions
I
had
is:
where
should
this
go
yeah?
Something
like
that?
So
it's
basically
like.
Where
would
a
good
place
for
these
things
to
go?
I'm,
not
sure
the
prompt
area
is.
Is
it
but
like
is
this
something
that
we
could
maybe
add
to
the
output
group?
G
Is
it
something
we
could
put
in
the
footer?
But
if
anybody
has
any
input
in
this
either
speaker
piece
now
or
throw
it
in
the
issue,
that
would
be
really
great.
D
Is
there
a
known
way
of
giving
the
markups
that
has
this
announcement
enough
metadata
so
that
we
can
put
it
anywhere
on
the
page
because,
for
example,
we
have
a
status
bar
and
a
status
bar
allows
you
to
do
stuff
with
it?
Is
there
a
way
of
saying
this
particular
type
of
text?
Added
to
the
status
bar
is
actually
the
kind
of
notification
you're.
Looking
for.
G
So
like,
if
we
had
something
in
the
status
bar
that
was
like
this,
many
percent
of
running
cells
are
completed,
then
that
would
make
sense
right
because
that
would
be
like
a
dialogue
or
a
progress
bar
I
think
that,
right
now,
what
we're
asking
for
is
that
the
cell
resolution-
so
it
might
be
a
little
bit
too
granular
for
the
status
bar
but
I,
could
see
an
aggregate
happening
on
the
status
bar.
G
You
know
this
is
more
like
if
you
hit
the
keyboard
and
you
don't
have
any
haptics
or
audio,
how
do
you
know
that
or
yeah
or
sound
with
the
keyboard?
How
do
you
know
that
the
cell
was
submitted?
So
you
know
cell
two
submitted
cell
cell
two
completed.
You
know
like
something
like
that:
they're
really
short
messages
that
you
typically
send
to
these
things.
So
they're
not
really
they're,
not
post
notifications,
there's
something
a
little
bit
extra.
E
I
was
just
leaving
background
background
on
on
this
one
that
unfortunately,
I
can't
seem
to
replicate
it
right
now
and
I'm
testing
it
actually
with
RTC
on
and
multiple
views,
and
it's
not
happening
on
the
exact
same
setup,
I
hadn't,
updated
my
environment
or
anything
so
I
don't
know
what's
going
on,
but
I
do
worry,
like
the
fact.
If
you
see
these
things
once
you
you'll
see
them
again.
So
at
least
I
think
we
should.
We
should
keep
an
eye
on
it.
So.
F
A
recently
that
character
appear
to
disappear
and
they're
they're
weren't
quite
sure
that
it
did
happen
or
whether
they
mistyped
or
something,
and
so
they
they
throw
that
to
us.
So
this
may
be
related
because
it
seems
to
be
either
the
characters
preceding
the
place
where
things
are
being
edited,
but
same
thing.
We
weren't
able
to
reproduce
yeah.
So
it
looks
real
yeah.
E
And
if,
if
I
happen
to,
if
it
happens
to
me
again,
I'll
try
to
also
then
make
not
just
screenshots
but
I'll
try
to
make
a
screencast
so
that
the
act,
the
specific
keyboard
actions
are
the
same
though
I
am
pretty
sure.
I
was
doing
the
same
thing.
I
was
backspacing
over
the
characters.
I
was
not
editing
with
delete.
I
was
editing
with
backspace
I
tried
to
do
the
exact
same
thing.
That
happened
a
few
days
ago
when
I
filed
the
bug,
and
it's
just
not
happening
now.
F
G
Some
of
the
I've
known
they're,
there
are
some
situations
that
I
haven't
used
the
recent
RTC
because
of
spending
a
long
time
in
RTC
and
then
getting
these
weird
highs
and
bucks.
G
So
this
is
I,
think
an
ongoing
challenge
and
good
luck,
but
yeah
I,
don't
know
how
that
you
go
and
Report
these
weird
States
because,
like
I,
think
Fernando
has
the
best
idea,
which
is
the
best
you
can
record,
is
like
the
state
you
got
to
but
like
getting.
There
is
a
real
challenge,
sometimes
right.
F
So
every
you
know,
crdt
patch
is
actually
recorded
in
a
white
store,
so
we
literally
have
a
database
of
all
of
the
patches
during
a
session.
So
we
presumably
we
should
be
able
to
replay
this
and
actually
also
to
use
this
for
generating
and
stuff
like
that,
eventually
as
a
feature,
but
first
there's
a
debugging
tool
for
for
yjsn
scholar,
version
in
Gypsy
lab.
E
Yeah
and
for
me,
Tony
I,
totally
hear
you
because
that's
kind
of
my
concern
right
that
I
keep
using
it
but
I'm
trying
to
use
it
first
before
I,
put
in
front
of
a
lot
of
students
right
and
I.
Just
keep
running
into
these
weird
things
that
I'm
like
all
right.
This
is
still
a
little
dangerous
right
and
and
I
want
to
put
it
in
front
of
students
at
Berkeley.
Many
many
instructors
have
said
this
is
exactly
what
we
want
to
use
for
taing
for
support
for
teamwork,
Etc
and
I.
E
D
H
If
we
have
these
logs,
even
if
we
can't
replay
them
if
there
is,
if
they
can
be
like
manually
inspected
or
they
can
be
inspected
later
once
that
will
arrives
if
there
was
like
a
button
or
some
other
tool
that
could
like
extract
those
logs
for
submission,
I,
don't
know
what
format
they're
in
whether
so
people
can,
you
know,
inspect
them
for
previously
concerns
before
they
submit
them
or
not.
But
that
would.
E
Because
it's
that
sqli
the,
why
store
sqlite
database?
That
has
everything
that's
been
happening-
we're
definitely
not
going
to
want
to
send
that
as
of
whole
upload,
there's
people's
entire
document.
History
may
be
there.
They
can
also
be
huge
if
they
haven't
cleared
it
in
a
while,
but
the
idea
it
would
be
great
to
be
able
to
say
quickly.
Okay,
here's
how
to
view
a
portion
of
that
log
corresponding
to
this
document.
E
Please
copy
and
paste
this
in
a
way
that
we
can
use
it
or
cut
it
out
and
and
Export
it
in
a
way.
That
makes
it
easy
for
the
team
to
to
maybe
just
replay
that
section.
The
history
for
that
document,
rather
than
saying
oh,
go
find
this
like
20
50,
megabyte
of
giant
sqlite
database
in
your
home
directory
and
try
to
sanitize
it,
which
obviously
is
just
too
much
to
ask
for
regular
users.
E
A
All
right
well,
thank
you,
everyone
for
your
contributions
on
this
anything
else.
To
add
on
the
editor
Behavior
topic.
A
Okay,
oh
sorry,
go
ahead,
the.
H
H
You
can
capture
keyboard
maps
for
users
with
non-us
layouts
are.
A
H
That
was
a
wild
wiggle.
Oh.
A
Thank
you,
yeah
lumino,
the
the
pull
request,
okay
board
event:
code
yeah.
The
fact
that
we're
using
scan
codes
makes
this
very
painful,
like
you.
H
Does
it
does
rely
on
having
keyboard
Maps
but
yeah
so
produce
that
as
your
own
offline,
okay.
A
A
Thank
you
for
sharing
that
yeah.
This
is
something
that
I
was
discussing
after
time.
Zones
keyboards
are
among
the
most
difficult
things
for
me
as
a
front-end
developer,
to
diagnose.
G
Yeah,
this
I
have
one
more
issue
that
I
wanted
to
bring
up.
That
is
a
big
blocker
for
V7
I.
G
Think
there
is
a
issue
open,
hasn't
seen
any
kind
of
hits
in
like
a
month
and
a
half
or
so,
but
it's
on
tap
traps
in
Jupiter
lab,
but
once
that
one
gets
sorted,
this
isn't
really
tracked
anywhere,
I'm,
not
sure
if
it's
like
put
as
a
Blocker
on
notebook
or
anything,
but
it
really
should
be,
and
I
was
wondering
if
you
know
folks
could
put
some
eyes
on
this
and
maybe
start
driving
this
one
forward.
C
It's
it's
cracked
in
my
head,
but
the
I
did
have
a
discussion
last
week
with
some
folks.
C
The
the
main
Blocker
on
this
is
the
as
a
design
question
around
the
focus
State
around
cells,
and
so
there
are
some
ideas
that
I
have
to
and
I'm
gonna
I'm
gonna
put
those
out
to
I
put
those
on
on
the
pr
to
like
test
them
out
and
see
if,
if
that
can,
if
those
can
make
it
past
review
right
now,
the
current
issue
is
that
the
the
focus
cell
looks
a
bit
heavy
to
some
to
some
reviewers.
C
G
F
So
I
just
wanted
to
announce
real
quick
that
we
have
received
some
funding
to
work
on
accessibility
with,
and
so
they
wanted
us
to
scoop
this,
because
it's
only
for
eight
weeks
of
developments
of
development,
and
so
we
we
decided
to
scoop
it
to
keyboard
accessibility
and
to
actually
follow
the
thorough
review
of
The
Notebook
7
keyboard,
accessibility,
audit
that
Isabella
did
maybe
a
few
months
ago,
and
so
we're
aware
of
the
pr
of
Gabriel,
and
we
think
that
this
is
definitely
clean
scope
and
so,
depending
on
whether
this
is
merged.
Already.
F
G
G
C
I
do
think
it's
possible
to
not
add
the
focus
styling
I
didn't
want
to
create.
One
I
didn't
want
to
solve
one
or
CAG
issue
and
then
create
another
one
at
the
same
time,
but
maybe
for
the
purposes
of
just
getting
it
through
review.
That
might
be
the
best
route
forward.
F
If
this
is
basically
ready,
maybe
this
is
worth
having
folks
on
the
team.
You
know
on
this
contract.
You
know,
do
as
much
review
as
possible
with
this
PR
and
see
whether
we
can
get
it
through
within
the
temp
scale.
That
was
discussed
earlier
for
the
book.
Seven
I
I,
don't
know
about
that.
So
I
I
think
it's
not
really
interesting.
My
decision
or
anything
I.
G
A
Okay,
yeah
thanks
Sylvan,
so
unfortunately,
my
teammate
piyush
had
to
drop
for
another
meeting,
but
I
can
present
his
agenda
item
on
his
behalf
on
our
team's
behalf,
so
our
team's
been
working
on
Jupiter
AI.
A
A
We
are
putting
together
a
road
map
for
issues
that
we're
working
on
for
the
next
quarter,
which
starts
next
week.
So
we
have
a
list
of
Milestones
which
is
linked
from
the
agenda,
but
I
can
also
drop
it
in
the
chat.
080
is
the
most
recent
release
of
Jupiter
AI
and
we've
been
planning
out
090
and
any
future
releases.
A
So
if
you've
been
looking
at
it,
we've
been
getting
some
feedback
from
elsewhere
in
the
community,
but
if
you're
at
all
interested
in
generative,
AI
we'd
love,
if
you
could
try
it
out
on,
you
can
try
it
with
many
popular
providers,
including
AWS,
but
not
limited
to
it,
and
if
you
have
particular
suggestions
or
bugs
that
you
found
we
can
get
those
prioritized.
F
Yeah,
maybe
I
just
wanted
to
get
something
on
your
radar,
so
we've
studied
discuss
discussing
having
a
chat
UI
as
part
of
the
collaboration
package,
so
that
would
not
be
necessarily
bound
to
the
Jupiter
AI.
Although
you
already
have
one
in
the
Jupiter
AI
repository.
Yes,.
A
F
A
This
so
Nikola
briche
opened
issue
205,
which
I
just
dropped
into
the
chat,
which
is
to
effectively
refactor
the
chat
interface
into
a
separate
extensible
part
of
Jupiter
Labs.
So
exactly
as
Sylvan
said,
you
wouldn't
need
to
install
Jupiter.
Ai
just
have
chat
I,
that's
this
is
something
we've
talked
about
for.
A
Quite
some
time
is
the
idea
that,
like
the
the
chat
is
not
just
a
two-party
chat
you
and
Jupiter,
not
the
the
chat
bot
that
we
have
for
Jupiter
AI,
but
instead
we
could
have
crosstalk
and
have
multiple
people
chatting
and
then
only
tag
in
Jupiter.
Not
if
we
need
the
AI
to
answer
a
question
we've
been
playing
around
with
this
like
on
our
own,
like
not
in
the
context
of
Jupiter
lab,
but
almost
like
as
a
UI
concept
with
our
ux
designer.
We
don't
really
have
a
whole
lot
to
share
with
that.
A
Yet,
but
that's
something
that
we
might
want
to
discuss
among
our
team
next
time
we
meet
I,
see
vdar
has
a
hand
up
as
well.
H
Yeah,
that
was
for
another
all
right:
I,
don't
I
haven't
gone
through
dishes.
I,
don't
know
if
this
is
a
duplicate
or
not,
but
some
models
are
don't
really
understand,
English
or
natural
language.
Very
much,
but
understand
code
right,
so
they're
more
like
code
completion
stuff,
they
will
so
like
they
take
your
current
cell
and
they
will
try
to
expand
upon
it
right.
Yep,
yeah,.
A
Even
like
some
of
them,
like
the
the
flan
models
that
are
available
on
hugging,
face,
don't
really
answer
questions,
but
it's
more
like
if
you
say,
if
you
say
something
like
what
is
the
capital
of
Norway,
it
wouldn't
answer
it.
But
if
you
say
the
capital
of
Norway
is
space
and
then
it
will
auto
fill
the
you
know.
What
it
thinks
is
the
correct
answer.
At
least.
H
Yeah
so
the
main
point
is
they're
dressing.
Currently
a
ux
or
or
a
way
of
of
doing
those
I
think.
A
H
Little
bit
tricky
at
least
because
if
you
use
the
current
Magics,
for
example,
if
you
say
like,
oh,
the
format
should
be
code
because
coding
code
out
right.
It
adds
a
bunch
of
English
to
The
Prompt,
whatever
yeah.
A
The
the
admittedly
naive
intention
with
the
prompts
was
that
the
same
prompt
could
be
used
with
any
model
and
what
we're
seeing
is
exactly,
as
you
describe
vdar,
that
the
same
prompt
given
to
10
different
models
might
work
as
intended
in
like
two
of
them
and
the
others.
It
you
don't
get
something
sensible
back
so
I
think
like
prompt
templates
as
a
function
of
the
model
might
be
something
useful
to
add.
I,
don't
know
whether
we
have
that
in
there.
A
A
H
A
Yeah
I
think
improving,
like
the
prompt,
engineering
and
kind
of
tweaking
things.
What
we're
also
seeing
is
that
you
know
right
now,
especially
because
we're
tied
to
Lang
chain
we're
bias
towards
text
to
text.
As
our
format,
you
give
you
give
the
model
text,
it
gives
you
text
but
text
to
image
text
to
video
text
to
audio
image
to
text.
These
are
all
popular
use,
cases
for
generative
Ai
and
we're
not
really
well
set
up
for
those.
A
So
we
we
invite
everyone
in
the
community,
no
matter
how
familiar
you
are
with
either
generative
AI
or
Jupiter
AI
see
some
good
ideas
coming
through
in
the
chat,
but
the
chat's
gonna
disappear
when
this
meeting
does
so.
If
you
want
to
preserve
that
in
our
our
issue
list,
please
open
that
up.
A
We've
got
a
small
team,
but
we're
also
willing
to
you
know,
accept
a
lot
of
good
feedback
from
elsewhere
in
the
community,
Sylvan
says:
hugging
Chet
does
respond
to
questions
yeah
hugging
the
hugging
chat
hugging
face
stuff
we're
following
pretty
closely
because
hug
and
face
is
like
a
partner
of
AWS
I.
Don't
know
the
exact
business
relationship,
but
that's
one
of
our
larger
providers
is
the
hug
and
face
Hub
because
hug
and
face
is
an
independent
company,
I
believe
they're,
a
French
company
in
fact,
and
they
work
with
you
know
we.
A
Believe
we've
reached
the
end
of
the
agenda
because
Sylvan
is
next,
but
we
we
did
the
accessibility
funding
discussion
before
this
one,
so
I
think
we're
good.
Unless
you
have
anything
to
add
there.