►
From YouTube: Clojure visual-tools meeting 2
Description
In this work meeting of the Clojure visual-tools group, we discussed potential compatibility/composability across tools
Background: https://clojureverse.org/t/visual-tools-meeting-2-compatibility-across-tools/
Moderator: Kira McLean
Zulip topic thread: https://clojurians.zulipchat.com/#narrow/stream/313390-visual-tools/topic/meeting.202
A
All
righty,
okay,
so
hello!
This
is
yeah
just
a
little
meeting
this.
The
closure
data
science
community
we're
here
to
talk
about
some
sort
of
compatibility
between
the
tools,
mostly
today,
and
also
just
hear
from
some
other
library
and
tool
authors
about
where
the
ecosystem
is
at
and
what
kind
of
problems
we're
experiencing
and
what
kind
of
direction
to
take.
Next,
I
guess
mostly
so
I'm
kira,
just
a
closure
developer
sort
of
a
data
science
noob,
but
interested
in
all
this
kind
of
stuff.
A
A
B
F
G
E
Is
the
eternal
question
just
so
yeah
I'm
kind
of
working
on
this
right
now,
so
this
is
the
project
I'm
working
on
and
it's
in
this
kind
of
tools
for
foot
space
and
it's
this
kind
of
visual
closurely,
small
talky
kind
of
thing.
So
that's
one
thing:
I'm
working
well,
that's
the
main
thing:
I'm
working
on
helping
out
to
maintain
this
library,
which
is
the
bridge
to
wolfram
language
and
helping
to
organize
this
conference.
So
so
that's
me
very
sort
of
quickly.
E
G
This
is
the
logo,
and
so
I'm
super
interested
in
how
hotel
can
can
support
all
these
and
callaway
is,
of
course,
a
visual
tool
as
well,
but
any
anyway,
it's
more
about
me
curious
about
how
ocala
can
can
support
yeah
this
space,
the
best
and
I'm
sorry
for
for
being
late,
and
thank
you
lucas
for
sending
that
reminder
to
me.
That
was
awesome.
H
Dave
arm,
I'm
here
kind
of
80
lurking,
I'm
very
interested
in
native
user
interfaces
for
closure.
I
have
a
project
that
I
will
be
releasing
officially
and
maybe
even
today,
certainly
in
the
coming
weeks
on
github
that
allows
you
to
do
swt
user
interfaces
really
easily
in
closure.
So
that's
that
mauricio.
I
I
I
I
So
in
fact
I
did
the
example
rendering
the
chess
board
all
the
time,
so
it
was
like,
like
kind
of
dog
food
in
I
mean
I'm
using
my
own
tool
to
develop
things
that
I
feel
like
they're
comfortable,
and
I
am
also
suffering
the
same.
Problem
of
lots
of
people
are
doing
great
visualizations
in
the
closure
space,
but
mine
have
to
be
implemented
almost
from
scratch,
so
both
on.
I
Clover,
I
mean
everything
from
scratch,
so
I'm
kind
of
interested
to
see
if
I
can
reuse
the
tooling
of
other
people
and
just
have
a
richer
set
of
libraries
and
I'm
also
starting
something
that
I
don't
know
if
I
will
continue
that's
trying
to
revive
all
the
autumn,
editor
with
clojurescript,
I'm
not
sure
if
I
will
keep
doing
that.
It's
simply
too
much
work,
but
well
we
never
know.
F
J
K
Regarding
the
problem
of
composibility,
I
just
want
to
share
a
small
thought.
It
reminds
me
of
like
this
like
the
problem
is
we
have
a
lot
of
visual
tools
and
they
usually
have
to
coexist
with
the
ides,
and
there
is
also
like
a
lot
of
ids
and
like
this
feels
like
m
by
and
problem
that
a
language
server
protocol
solved
for
programming
languages
and
ids.
C
C
I'm
not
you
know,
I'm
not
sure
exactly
what
we
can
share,
but
it'd
be
really
cool
if
if,
if
things
were
interoperable-
and
I
guess
even
if
we
can't
do
the
code
level,
I
wonder
if
we
could
share
ux,
so
that
would
feel
more
familiar
to
users,
so
they
could
jump
around.
So
I
don't
know
but
yeah.
That's
me.
A
I
think
that's
everyone
yeah,
so
I
had
to
step
aside
there
for
a
second,
but
I
think
that's
all
yeah.
So
maybe
sorry,
I
have
some
dogs
here,
so
I'm
running
in
and
out
with
this
puppy
yeah,
but
I
guess
maybe
it
makes
sense
next
week
is,
if
you
want
to
go
ahead
with
what
you're
saying
there
start
start,
the
conversation
about
just
issues
you're
having
or
integrating
or
whatever
just
yeah
the
goal
of
the
whole.
A
B
Well,
these
two
maps
have
these
new
things
in
common
and
green
and
red,
and
whatever
it's
really
nice
and
like
there
are
a
lot
of
use-
and
I
I
don't
know-
clark
has
like
a
few
different
views
and
reveal
has
a
there's
a
lot
of
stuff
in
there.
That's
very,
very
incompatible
because
it's
closure
and
built
on
top
of
closure
fx.
B
The
other
one
was
that
keyboard
shortcuts
and
interactions,
and
so
on
that
was
similar
to
chris's
point
where
it's
like.
Even
if
we
can't
share
a
code,
we
might
be
able
to
have
very
similar
ux
designs
where
the
user
knows.
Okay,
if
he
hits
the,
I
don't
know
the
g
button
and
something
gets
grabbed
or
whatever
or
in
review
in
portal.
When
you
hit
the
e
button,
something
expands
and
so
on,
we
could
come
up
with
similar
key
bindings.
B
You
could
hit
the
similar
buttons
and
so
on,
although
I
think
it
would
make
sense
to
be
able
to
also
configure
them
just
have
similar
defaults,
maybe
yeah,
then
the
next
point
was
like
just
reusable
components
beneath
it.
I
think
we
had
to
talk
about
it
last
time
after
the
recording
was
stopped
about
a
lot
of
the
tooling
having
like
infrastructure
that
could
be
pulled
out
of
the
tooling,
where.
B
B
So
my
omnitrace
library
has
now
a
few
configurations
that
get
set
inside
kauba
and
I
could
do
the
same
portal
could
be
doing
the
same
stuff
where
it's
like,
oh
yeah.
Okay,
now
you've
got
this
snippet.
That
gets
called
that
the
user
can
just
call
and
he
doesn't
need
to
know
how
to
start
portal
or
something
or
how
to
tap
stuff
into
it
or
whatever,
and
I
saw
I
don't
remember
his
name
peter.
B
J
J
I
Yes,
okay,
just
a
little
step
for
evaluation,
most
of
the
stuff
in
chlorine
and
clover,
my
plugins
for
autumn
and
vs
code.
They
are
completely
separated
from
the
from
their
ui
stuff
from
the
editor
from
internal
apis.
So
they
live
in
a
different
library.
It's
a
closure
script
library!
We
can
reuse
it.
If
you
want.
I
B
Especially
with
closure
scripts
we've
had
to
talk
multiple
times
now,
where
it's
like.
Okay,
it's
hard
to
pull
in
the
closure
script
source,
because
then
you
have
to
build
it
and
you
have
to
know
which
npm
packages
also
to
pull
if
it
depends
on
something
and
so
on
so
enclosure
script,
kind
of
makes
sense
to
have
different
different
boundaries
than
if
you're
wearing
closure,
where
you
can
just
say
well
I'll,
just
pull
in
the
culture
code
and
be
done
with
it.
I
J
D
I
think
I
think
it
is
kind
of
related
to
your
four
points,
but
I
think
the
way
I
would
love
to
state
it
is
just
being
able
to
write
things
on
the
user
side.
That
would
work
with
all
tools
and,
of
course,
that
is
stated
in
a
kind
of
provocative,
too
ambitious
way,
but
I
think
that
matters,
and
if
there
is
time
I
would
love
to
share
something
for
a
few
minutes
to
kind
of
make
it
concrete.
What
we
could
hope
for
is
it?
Is
it
good
if
I
shut
the
screen
for
a
moment.
K
D
Yeah
thanks
so
and
cher.
Can
you
hear
me?
Okay,
yeah,
so
here
is
calva
peter.
Thank
you
so
much
for
calva.
Can
you
see
calver
now?
No,
maybe
not,
maybe
it
will
appear
in
a
moment.
I
guess
right
so
here's
kalpa
and
what
we
have
here
is
a
little
project
that
I'll
share
later
with
portal-
and
you
know
portal
can
be
embedded
inside
caller,
and
so
I
just
connected
it
to
the
nrepel
protocol
so
that
we
can.
D
So
I
think
you
in
a
moment-
probably
now
you
see
the
browser,
so
in
node
space
we
had
three
ways
to
express
that
this
thing
needs
to
to
be
rendered.
This
way,
one
was
adding
metadata
to
the
code.
So
the
code
is
this
expression
of
code.
We
are
evaluating.
It
gets
this
additional
metadata
that
marks
it
to
be
rendered
some
way.
Another
way
was
adding
metadata
to
the
value,
so
we
compute
some
value
and
then
we
attach
some
metadata
to
it
by
some
function.
D
Saying
this
thing
is
of
this
kind
and
another
way
was
through
protocols.
So
we
have
this
protocol.
We
are
implementing
that
says
this
thing
was
is
of
this
kind,
and
we
decide
that
a
certain
type
of
thing
would
implement
this
protocol.
So
we
had
different
ways
and
we
actually
needed
all
of
them
to
handle
different
situations
on
the
user
side
where
we
just
wanted
things
to.
D
F
Daniel
that
is
exactly
what
I
was
thinking
about.
So
if
we
go
back
to
the
web
example
that
I
mentioned
all
that
metadata
does
is
a
collection
of
functions
that
have
a
flow
that
attach
ideas
or
parameters
or
formats
as
data
moves
up
and
then
with
interceptors
up
and
down
and
up
the
stack
right.
So
the
idea
would
be
that,
as
a
user,
think
of
a
data
frame,
I
package
my
data
in
a
data
frame.
I
use
those
functions
to
put
my
data
in
a
specific
format.
F
F
Yes
think
about
when
you're
handling,
when
you,
when
you're
handling
a
web
request,
you
know
so
you
you
get
it
and
you
look
at
the
headers
and
then
you
need
to
add
something.
Maybe
there's
a
routing
component
that
needs
to
go
and
then
just
you
add
functions
as
you
process
down
you.
You
processing
your
response.
So
the
response
that
you
have
is
the
application
of
functions.
So
your
request
need
needs
to
be
authorized.
Then
you
have
an
authorization
function.
Your
request
needs
to
do
something
with
a
route.
Then
you
route
it.
F
L
F
K
F
F
F
There
should
be,
ideally
there
should
be
no
separation
between
data
processing
and
visualization
creation,
and
the
only
difference
when
you
do
the
whatever
different
aspects
of
of
processing
visually,
it's
the
renderer.
So
if
you've
looked
at
the
glamorous
toolkit,
you
have
data
that
can
be
shown
in
different
ways.
F
J
I
know
that
I
watched
there's
some
long
sessions
that
are
really
interesting,
that
gene
kim
and
eric
norman
did
where
they
kind
of
yeah.
I
think
there's
a
set
of
the
three
videos
I
can
put
in
the
chat,
but
it's
interesting
because
they
bring
the
closure
perspective
and
then
they
work
with
the.
I
think
it's
the
creator
of
climate
race
toolkit
and
which
is
kind
of
a
small
talk
environment
which
has
a
lot
of
these
features
that
we're
interested
in
and
they
work
through.
It.
J
They
also,
while
getting
feedback
about
how
they're
using
it.
So
it's
it's
really
interesting.
I
think
one
of
the
things-
that's
that
I
noticed
is
that
for
them
they
did
the
advice,
like
both
the
coding
environment,
the
program
that
you're
working
on
all
the
visualizations,
those
are
all
in
process
and
they're
all
connected.
J
It
is
similar
to
closure
in
the
sense
that,
like
the
your
program
is
all
running
through
the
replica,
as
well
as
your
kind
of
dev
tools,
the
only
difference
is
that
for
a
lot
of
closure
environments,
the
the
programming
environment,
the
ide,
isn't
necessarily
running
in
the
same
process
as
the
as
your
program
and
your
dev
tools.
So
that
is
a
one
thing
that
I
noticed.
B
Yeah,
I
would
find
it
like
to
me
as
a
user.
The
reveal
and
portal
are
pretty
similar
right.
I
mean
underneath
they're,
very
different,
rebuilds
and
so
on,
but
in
the
end
they're
more
or
less
like
the
idea
is
to
replace
the
rebel
output
and
they
do
the
showing
of
stuff
of
the
print
in
rebel,
and
it
would
be
interesting
to
be
able
to
like
write
viewers
and
so
on
the
ones
and
be
able
to
use
them
on
both,
and
I
realized
that's
pretty
difficult
because
they
use
very
different
visualization
models.
B
If
we
can
get
to
some
point
where
they
can
interact,
we
can
pretty
easily
take
note
space
and
cleric
and
whatever,
because
because
they
are
basically
already
doing
the
same
thing.
We
just
need
to
come
up
with,
like
the
correct
names
inside
the
hash
maps
that
they
add
the
user's
vector
as
views
and
then
they're
be
pretty
easily
compatible.
B
J
I
will
say
so
all
the
visualizations
that
I've
made
so
far
are
using
my
own
ui
library
called
membrane,
which
I'm
not
trying
to
get
everybody
to
use,
but
the
they
all
run.
They're
renders
for
canvas
2d
and
the
browser,
html
javafx
swing
and
there's
also
an
ios
implementation.
So
you
can.
These
visualizations
can
run
in
all
these
different
environments,
and
the
key
is
that
the
key
is
really
just
building
your
visualizations
out
of
primitives
like
shapes
text
and
images.
J
K
J
Yeah
I
built
some
revealed
plugins
and
the
api
is
really
great
as
a
visualization
creator,
because
one
as
if
I
was
saying
you
can
use
tooled
apps,
so
you
can
actually
load
it
at
runtime
and
just
say
like.
Oh,
I
like
this
visualization,
I
didn't
load
it
because
I
didn't
put
it
on
my
class
path,
but
even
while
my
rebel's
already
running,
I
can
just
add
that
dependency
and
then
I
think
it's
there's
just
some
protocols
that
you
implement.
J
If
I'm
not,
I
think
I
haven't
used
any
of
the
ones
that
run
in
the
browser.
Yet
I
don't
know
it
sounds
like
that's
a
more
difficult
problem.
Just
because
look,
there's
no
tool
steps
where
you
can
or
maybe
there
is.
I
don't
know
if
there's
a
tool,
steps
that
lets
you
just
dynamically
load
a
library
and
then
implement
some
protocols
and
then
have
it
part
of
your
dev
environment.
C
C
B
Yeah,
I
think
it's
a
little
bit
of
both
like
the
ux
is
the
easier
part.
Maybe
because
then
we
can
say
well,
it's
just
a
closure
map
with
like
these
keys
and
like
the
first
key
is,
like
I
don't
know,
the
name
of
the
viewer
and
so
on,
and
maybe
there's
like
a
tiny
sci
function
in
there
somewhere.
That
can
also
do
stuff
whatever,
and
I
think
that
could
be
used
basically
by
any
tool.
B
That's
enclosure
anyway,
because
everybody
can
run
a
ci
code
like
somehow,
probably
the
harder
part
would
be
to
like
use
the
whole
underneath
stuff
like
it
would
be
interesting.
I
haven't
you
looked
deeply
into
membrane
yet
or
like
basically
not
more
than
the
two
minutes
of
looking
at
the
github
but
like,
if,
like
the
tool,
could
really
be
like
okay
and
anybody
can
write
a
membrane.
E
B
And
instead
of
writing
the
portal
views
in
like
the
react
reagent.
Whatever
way
we
have
right
now,
we
would
write
them
a
membrane
and
the
same
thing
for
reveal
and
so
on,
and
it
could
spit
out.
Then
the
react
part
or
the
closure
fx
part
or
whatever,
and
then
we
it
would
be
way
easier
for
all
of
us
to
just
take
one
like
build
the
same
membrane
library
where,
like
there's
a
bajillion
viewers
and
people
can
just
use
what
they
like.
C
Yeah,
I
would
say
like
as
an
implementer
the
main
thing
I
feel
like
I
get
out
of
something
like
css
layout.
That's
probably
the
big
thing
so
like
specify
things
in
terms
of
shapes,
I
think
works
until
I
need
to
start
organizing
things,
and
I
don't
want
to
have
to
organize
things
I
just
want
to.
Like
I
wanna,
you
know
some
algorithm
to
organize
things
for
me.
C
B
My
code
can
that
I
just
pull
in
and
like
if
I
pull
reveal
in
or
portal
or
whatever
it's
running
in
the
same
process
as
my
code,
and
it
can
do
a
lot
more
stuff
than
kalva
can
do,
because
color
will
then
help
have
to
run
to
the
rebel
and
tell
it
well.
Please
do
this
and
then
you
get
some
information
back,
but
if
the
information
is
like
some
lazy,
forever
sequence,
you're
now
exploding
and
now
you
need
to
handle
that
and
so
on
and
if
you're
in
the
same
process
it's
way
easier.
B
B
That's
super
annoying,
it's
way
easier
if
I
just
hover
over
some
line
hit
this
hotkey
and
that
gets
tapped
over
automatically
like
as
if
I
was
like
the
ide
itself,
but
I
think
that
comes
back
to
the
word
to
the
configuration
integration
I
was
talking
about
initially
and
for
at
least
for
cover,
it's
not
possible
that
you
can.
Your
tool
can
set
hotkeys
inside
cover
and
run
arbitrary
code
inside
the
actual
repo
that
you
then
do
so
that
your
user
can
basically
hit
a
hotkey.
B
It
gets
spit
into
a
rebel
or
reveal
or
whatever,
and
they
feel
as
if
your
tool
was
actually
like
the
ide.
You
know
it's
still
in
a
different
window,
but
it
doesn't
really
matter.
Do
we
actually
kind
of
want
it
into
in
a
different
window,
especially
on
emacs,
where
there's
like
no
like
really
deep,
the
display
of
stuff,
I
think.
I
I
The
problematic
part,
for
my
part
at
least,
is
I
also
support
non-traditional
clergy
environments,
for
example,
clojure,
language
and
lumo
and
blank,
and
these
do
not
have
any
visual
tool
written
from
them.
So
I
kinda
want
to
have
the
visualizations
working
in
these
environments,
but
well
I
can't
inject
a
closure
code
in
them
because
they
are
not
claudia
percy.
They
are
like
airline
with
a
closure
thing
inside
it.
It's
hard.
K
H
Differences-
I
just
excuse
me,
I
just
threw
up
I've
been,
I
started,
drawing
a
diagramming
escape
just
trying
to
keep
track
of
all
the
different
concerns
that
we're
discussing.
I
threw
that
up
and
if
people
like
it,
I
can
keep
it
up.
If
people
would
rather
see
each
other's
faces
more,
then
I
can
take
it
back
down
or
if
we
want
to
do
something,
a
mixture
of
that
I'm
cool,
but
I
just
this
is
my
way
of
contributing
right
now,
all
right
thoughts.
J
C
Yeah
the
like
one
experience
I
have
with
him:
what
is
it
called
pagination?
I
guess
like
shipping,
less
data
than
the
user
has.
Is
that
you
don't
ever
get
the
the
number
right?
So
that's
the
guy.
I
I
used
to
do
that
within
a
portal
and
be
like
you
only
get
like
10
000
or
something,
but
someone
was
like.
I
have
11
000
and
I'm
like.
C
How
because
like
if
you're
doing
a
plot-
and
you
just
have
like
10
000
points
you
want
to
ship
that
over.
But
if
you
have
like
a
vector
of
10
000
maps,
then
it's
like,
maybe
you
only
care
about
one
of
them
individually,
as
opposed
to
all
of
them
together
and
like
aggregate
so
but
yeah.
This
is
a
challenge
I
experienced.
C
But
I
do
think
that
vladis
brings
up
an
interesting
point
which
is
like,
instead
of
shipping,
the
data.
What
if
we
shipped
like
a
specification
for
the
ui?
That
is
data
because
that's
a
that's
a
much
smaller
subset.
Instead
of
like
trying
to
support
all
data,
we
support
a
subset
of
the
data
which
is
like
here's
a
visual
language,
whether
that's
something
like
hiccup
or
I'm.
Not,
I'm
not
sure
what
that
would
be,
but.
H
That
makes
that
was
kind
of
what
I
was
driving
at
with
okay.
Where
is
that
again?
I
took
it
down
because
didn't
want
to
like
dominate
everything
with
calling
these
a
binding
layer.
I
mean
we
saw
a
kind
and
its
way
of
doing
that
earlier,
where
there
was
both
metadata
at
one
time
there
was
a
protocol,
and
I
forget
what
the
third
one
was
did.
I
remember.
H
Right,
I
think
it
was
daniel.
You
showed
this
kind,
yeah,
yeah
and
then
yeah.
Somebody
else
showed
something
that
was
like
hiccup
and
I
think,
if
we're
going
ipc,
we
need
something
kind
of
like
hiccup
or
data,
some
data
way
of
describing
this
or
abstract
way
of
doing
this,
or
maybe
the
abstraction
we
got
the
abstraction
and
the
ways
of
attaching
it,
which
is
like
the
concern
kind
addresses.
C
B
K
L
B
C
Yeah
it
is
like
transit.
I
had
issues
with
fixing
a
version
on
the
class
path
and
I
couldn't
can't
do
that
so,
but
it
also
does
other
things
like
holding
on
to
runtime
objects,
so
you
can
inspect
them
later
on.
So
there's
it's
like
doing
a
bunch
of
things
in
memory
to
hold
on
to
references
which
is
like
dangerous.
If
you
do
that
too
much,
then
you
don't
have
any
more
memory.
D
And
by
the
way
we
have
half
an
hour
till
the
official
time,
and
maybe
maybe
we
could
think
of
concrete
steps
in
a
moment
kira.
What
do
you
think?
Is
it
a
good
time
to
kind
of
transition
to
thinking
of
how
we
should
continue
like,
for
example,
lucas
suggested
a
concrete
thing
we
could
try,
which
is
writing
something
that
works
in
both
reveal
and
portal.
D
B
I
B
I
B
C
D
B
Sorry
from
my
side,
I
think
it
would
be
interesting
to
then
be
able
to
share
viewer
implementations
because
currently
I'm
like
okay
yeah.
No,
I
really
like
the
diff
view
in
portal,
but
I
wanna
see
it
inside
like
a
sticker
or
I
want
to
use
it
in
a
clerk
notebook
or
there's
some
cool
thing
that
clerk
does
with
like.
B
I
don't
know
how
to
describe
it,
but
whatever
it's
just
a
few,
and
if
I
could
use
that
in
a
portal
that
would
be
really
nice
and
so
on,
because,
like
everybody
has
like
a
few
views
that
are
unique
and
if
we
all
used
a
very
similar
language
to
build
the
views,
we
could
then
use
one
in
the
other
and
it
wouldn't
be
like
well.
I
need
to
now
use
all
five
at
once
and
I
need
a
new
macbook
with
more
rum,
so.
J
J
And
that
would
be
a
good
step
because
it
already
works
and
reveal-
and
I
mean
that's
kind
of
my
dream-
is
like
I
make
the
visualization
and
I
just
have
implementations
of
the
protocols
that
make
all
the
different
tools
happy
or
potentially
even
one
protocol
that
is
shared
across
all
of
them.
And
then
I
can
just
create
visualizations
and
then
I
can
use
whichever
tool
feels
right
and
I
can
just
make
them
available
to
whoever
wants
to
use
them,
and
it
makes
it
easier
for
me
to
to
do
that.
C
C
I
B
I
don't
know
if
it's
been
there
always,
but
you
posted
this-
I
don't
know
maybe
a
month
ago
or
something
where
you
can
click
on
something
in
reveal
and
it
posts
back
into
that
into
your
actual
running
code
portal
can
do
the
same
thing.
That's
always
been
able
to
do
it
like
in
the
commands
thing,
and
now
you
can
like
also
like
manipulate
some
atoms
or
whatever
so
the
communication.
B
There
are
more
multiple
different
ways
to
communicate
back
and
forth
and
in
chlorine,
like
you
said,
you
can
always
also
do
it
because,
like
it's
built
on
the
interaction
side
of
things
and
just
sending,
I
mean
just
generating
some
jpeg
and
showing
it
in
some
tools,
not
super
interesting.
I
think
like,
like
you
said
chris.
The
interesting
part
is
to
be
able
to
have
some
language
where
we
can
say
well.
This
is
but
when
you
click
this
also
called
this
in
some
other
code.
C
B
I
think
it
would
make
sense
first
for
plug-in
authors
and
then
for
users,
because,
as
plug-in
authors
are
usually
like,
take
more
complicated
stuff
and
are
okay
with
at
first
and
once
we
find
the
kinks
and
so
on.
We
can
make
it
one
level
higher,
so
it's
actually
usable
by
people
who
don't
want
to
dig
into
the
source.
B
I
For
example,
I
don't
know
vega
light
viga
high
line
high
charts,
anything
and
just
plug
in
inside
the
chlorine,
so
you
can
render
a
high
charts
or
a
bigger
light
inside
the
chlorine
without
too
much
work,
and
it's
like
user
space,
not
plugin
space,
but
not
on
vs
code
and
that's
one
of
the
sad
things
about
atom
dying,
because
I
wanted
to
support
that.
Also
on
the
escort
side,
too,.
D
So
that
is
one
thing
just
being
able
to
write
ui
code,
that
is
shareable
between
tools
and
another
proof
of
concept,
is
being
able
to
write
some
user
code
that
renders
in
a
few
tools,
maybe
a
name
space
that
you
would
like
to
just
be
able
to
see
in
reveal
portal
note,
space
clerk
or
whatever
right.
Is
it
a
good
way
to
look
into
it?
I
Yeah
you
you
mentioned
like
a
namespace,
something
that
we
can
import
inside.
I
don't
know
the
rapport
and
something
like
this
yeah
from
my
party.
This
could
not
work
like
I,
as
I
said,
I
do
support,
for
example,
clojure
that's
under
alongside
and
there
it's
something
that
I
would
like
to
not
depend
on
external
toolings.
I
It's
just
something
that's
running
on
the
plugin
side,
if
I
could
say
like
that,
so,
for
example,
if
I
evaluate
something
in
the
example
of
the
chess
board,
for
example,
the
chess
board
was
a
hiccup
data
structure
and
I
was
just
like
evaluating
something
returning
a
vector
that
represents
the
the
hiccup
data
structure,
and
then
I
was
rendering
this
structure
and
that's
all
so,
there's
no
like
specific
name
space.
Nothing,
that's
like
specific
from
the
from
the
visualization
that
I
want.
I
So,
if
I
had
to,
I
don't
know
import
a
library
inside
the
clojure
repo
or
in
in
repo
or
whatever
I'm
using
to
support
that
this
would
break
some
assumptions
that
I
have
in
my
system,
like
it
doesn't
matter
what
you're
running.
This
is
something
that
chlorine
do.
It
simply
says.
Okay,
are
you
running
in
a
socket
repo?
Yes,
okay,
so
I
support
your
user
case,
and
this
is
something
like
complicated
for
multiple
of
things.
First,
that
all
closure
implementations
have
different
small
but
annoying
chains,
things
that
are
different
and
then
not
support.
I
Well,
for
example,
in
closure
closure
along.
I
did
have
to
do
some
hacking,
because
some
of
the
elements
don't
don't
support
what
I
expect
so
yeah
importing
a
name
space
for
visualization
would
break,
for
example,
the
visualizations
for
closure
or
closure
on
their
lung
or
for
lumo,
for
example,
and
yeah.
It
will
not
work
on
my
side.
B
I
think
it
wouldn't
make
much
of
a
difference
for
the
rest
of
us
to
not
say
well,
we
don't
want
to
be
able
to
render
namespace.
We
want
to
render
like
a
map
that
has
stuff
in
it
I
mean
rendering
namespace
is
basically
the
same
thing
here
are
a
few
deaths
that
have
some
stuff
whatever
is
in
those
deaths
or
difference
or
whatever,
but
in
the
end
it's
just
some
bars.
If,
instead
of
saying
well
here's
a
few
bars,
we
say
well,
here's
one
map
with
a
few
names.
B
D
B
I
B
So
actually,
if
we
said
well,
we
need
to
ship
coder
on
because
something
needs
to
be
interactive.
So
you
actually
need
code
unless
we
want
to
write
basically
our
own
way
of
interpreting
stuff
and
then
we're
actually
writing
our
own
programming
language,
which
doesn't
really
make
sense
and
then
saying
well,
we
just
do
it
with
sei
and
then
everybody
is
happy
right.
B
Then
I
mean
the
other
thing.
Is
that
if
somebody
says
well,
no
sei
sucks,
I
don't
want
it
and,
like
I
don't
know,
maybe
vlad
will
reveal
because
he's
got
access
to
actual
closure.
Well
then,
it's
still
running
sdi
right
because,
like
closure
can
just
run
the
sci
code
without
having
sci
it's
just
gonna
say:
well,
that's
just
closure
code,
I
don't
care
and
it's
just
gonna
run
inside
all
the
other
stuff
right.
K
B
Plus
you
don't
really
need
it
right.
You
can
just
eva
closure
code.
That's
like
the
problem
in
the
browser
where
we
like.
Well,
we
want
to
evolve
some
user
code
that
he
gave
us
and
then
it's
like
yeah,
but
we
can't
because
we
can't
evolve
the
closure
script
or
in
the
well.
We
could
in
a
browser
with
like
importing
the
whole,
the
self-hosted
culture
script,
but
that's
a
whole
different
set
of
problems,
and
you
can
just
say
well:
it's
social
code,
I'll
just
evolve
it
in
the
jvm.
I
H
I
mean
at
the
end
of
the
day.
Sci
is
just
a
run
time
right
and
we
could
you
know
if
we
are
agreed
that
we'll
support
the
subset
of
closure
that
runs
in
sci
that
doesn't.
I
don't
think
that
precludes
us
from
all
also
running
in
in
say,
self-hosted
closure
script
or
inside
of
closure
proper
right
or
am
I
missing
something.
B
No
you're
right,
we
could
also
just
run
it
in
something
that
actually
has
more
capabilities.
I
mean
there's
not
that
much
that's
missing
from
sci
these
days,
but
you
can
also
just
run
it
in
self.
Also,
the
close
script
and
the
user
wouldn't
notice
the
difference,
and
I
think,
like
what
blood
is
saying,
makes
sense.
We
should
be
as
descriptive
as
possible
and
have
as
little
code
going
over
the
wire
as
possible,
but
well
at
some
point.
B
You
need
to
have
to
be
able
to
say
well,
I
want
to
run
an
edition
or
something
and
yeah
sure
we
could
basically
say
well,
then
tell
me
you
need
inc
or
whatever,
but
in
the
end,
then
we're
just
defining
a
new
language
right
at
some
point,
when
you
want
interactivity,
you
need
to
drop
some
kind
of
code
and,
like
I
don't
think,
there's
a
way
around.
It.
H
To
that
point,
another
source
of
possible
source
of
inspiration
is
the
eclipse
modeling
tools.
A
lot
of
people
know
eclipse
as
the
ide,
but
the
modeling
tools
is
like
the
eclipse.
Modeling
framework
is
like
small
talk
reinvented
inside
of
java,
where
you
just
have
everything
is
just
data,
and
then,
on
top
of
that
you
have
meta
data,
a
meta
model
that
can
describe
user
interfaces
in
the
abstract
and
then
on
top
of
that
and
there's
they've
just
done
a
tremendous
amount
of
work
in
making
java
work.
H
D
And
by
the
way
we
have
nine
minutes
to
the
official
time,
and
maybe
it
would
be
good
to
try
to
conclude
and
afterwards
a
few
of
us
might
wish
to
stay
after
we
stopped
recording.
But
maybe
if
anybody
would
like
to
kind
of
think
of
how
you
would
like
to
conclude-
and
maybe
you
could
also
write
later,
your
bottom
lines
in
the
chat
so
that
we
can
have
some
notes
of
this
meeting.
D
Yeah
so
maybe,
for
example,
I'll
suggest
something.
What
I
hope
suggests
is
that,
from
my
conclusion,
is
that
two
or
three
of
us
will
need
to
write
a
proof
of
concept
and
then
share
with
the
group
so
that
we
have
an
idea
of
what
we're
trying
to
do
and
then,
if
anybody
wishes
to
join
those
two
or
three,
then
please
tell
me
and
we'll
set
a
small
meeting
and
write
something
so
and
maybe
you
others
would
also
like
to
kind
of
offer
some
idea
of
how
we
proceed.
J
Yeah,
so
for
me,
if
so,
I've
made
a
reveal
plug-in
if
there's
other
tools
that
I
can
point
to
either
even
internal
apis
for
creating
visualizations
or
I'm.
Basically,
I'm
happy
to
guinea
pig
any
kind
of
way
to
make
a
visualization
available
to
other
tools
and
yeah,
and
I'm
happy
to
do
it
for
any
of
the
environments
that
are
available,
whether
it
be
web
or
on
the
jvm,
etc.
C
F
K
D
D
B
I
think
the
I
don't
know
if
you're
still
here,
oh
peter
is
still
here
right.
I
think
we
should
take
the
opportunity
and
since
multiple
of
us
were
saying
well,
it
would
be
nice
if
we
could
integrate
deeper
into
the
ids
with
our
tools
and
take
the
opportunity
to
like.
I
don't
know,
hit
him
over
the
head
with
some
ideas
that
we
or
some
features
that
we
would
like
to
see
inside
something
like
alvar
or
like.
B
We
don't
have
the
mr
cider
guy
right
and
no,
we
don't,
but
at
least
we've
got
50
of
the
no.
I
think
there's
still
neobim
or
something,
but
maybe
I
don't
know
if
chris
or
vlad
or
whoever
is
like
building
the
actual
tools,
I'm
just
a
user.
Basically,
if
you
guys
have
something
that
you
would
like
and
maybe
like
check
it
out
together
or
something
I
don't
want
to
put
you
in
a
spot.
If
you
don't
want
to.
C
When
I
was
working
on
the
vs
code
plugin,
it
would
have
been
nice
to
have
access
to
the
raffle.
Let's
add
like
my
own
like
or
to
just
add
my
own
vs
code,
commands
and
then
interact
with
closure
repl
so
that
I
don't
have
to
do
that
wiring.
But
it
sounds
like
that.
You
solve
that
problem
lucas
by
just
adding
some
config
to
your
jar.
So
I
don't
know
if
I
don't
know
if
that's
really
necessary
anymore.
G
C
So
I
was,
I
was
going
to
add,
like
a
few
commands
to
the
command
palette
in
vs
code
and
then,
when
the
user
interacts
with
that,
I
would
invoke
some
api
methods
that
I
had
and
that
would
be
available
in
the
runtime,
with
some
configuration
to
do
all
the
wiring
for
them.
But
it
sounds
like
if
I
understand
correctly,
lucas
is
saying
that
you
can
provide
stuff
in
your
jar
that
callvo
will
pick
up.
That
will
be
commands
that
have.
B
What
I
was
looking
for
so
would
there
be
like
there?
Would
there
be
more
that
you
would
want
to
do
with
the
rebel
connection,
because
I've
been
also
talking
to
peter
or
we're
like
well,
maybe
we
would
want
to
expose
like
the
just
in
some
way,
the
socket
that
and
rebel
exists
on,
and
then
you
can
just
have
your
own
and
rebel
client
and
fire
directly
against
it.
I
don't
know
if
it's
helpful,
but
then
we
like
sometimes
it's
probably
going
to
be
annoying
to
go
through
the
cowboy
replace
whatever
stuff
and.
C
I
would
rather
it
be
like
a
function.
Call
like
here's.
Here's
the
code
evaluated
community
result
the
reason
being
that
and
ripple's
only
one
protocol
there's
also
socket
replica
and
all
the
other
protocols.
So
if
they
could
just
be
like
call
a
function
and
then
you
know,
do
the
eval
stuff
and
then
maybe
you
get
a
promise
back,
I'm
not
sure.
K
Is
it
sort
of
like
replay
repo
commands
in
cursive,
like
you
can
define
shortcuts,
so
you
can
define
a
short
snippets
where
you
can
select
where
you
like,
insert
at
the
time
of
notification
like
current
current
top
level
form
that
is
selected
in
the
editor
or
current
form
or
name
of
the
file
yeah.
I
think
this
thing
is
is
very
useful
for
me.