►
From YouTube: Datahike and Datalog - Konrad Kühne - Scicloj meeting 8
Description
Konrad Kühne told about Datahike, and the Datalog ecosystem, with a live demo of using Datalog to work with data.
https://github.com/replikativ/datahike
https://lambdaforge.io/2019/09/20/datahike-release-0.2.0
slides:
https://lambdaforge.io/assets/slides/2019_10_14_scicloj/datahike.html
Followup comment by Konrad Kühne:
After watching the Q&A again, I noticed that I understood a question about recursion not completely. I thought the question was about unbounded recursion like in Prolog which datahike doesn't support. But simple recursions can be expressed like in the following example: https://gist.github.com/kordano/39398882d99b46ebd61155fcb566872a
text chat:
https://tinyurl.com/yywaeqoq
previous and next meetings:
https://scicloj.github.io/pages/web_meetings/
A
Welcome
everybody
to
sigh
closure.
Sigh
closure
is
a
community
that
discusses
science,
data
science,
machine
learning,
tools
and
techniques.
I'm
Sammie
Karolina
I'll
be
moderating
this
session
today
and
and
try
to
help
things
run
as
smooth
as
possible,
and
it's
good
that
everybody
knows
and
are
aware
that
this
these
meetings
will
be
on
YouTube
in
the
future,
and
so
so
you
can
go
back
and
back
and
look
at
them,
but
also
just
to
be
aware
that
this
is
not
a
private
session.
A
Just
one
word
about
myself:
I
have
a
company
here
in
in
in
Helsinki
collate
the
chip.
We
work
a
lot
with
data
data
applications
of
different
sorts,
b2c
type
of
services,
but
the
main
point
of
today
is
that
we
have
a
Conrad
here
called
a
goon
he's
he's
from
Heidelberg
Germany
and
is
a
C
CEO
and
data
engineer
at
lambda.
Lambda
forge
consultancy
that
specializes
in
information
system
engineering,
but
today
he's
here,
I'm
happy
to
have
it
having
you
had
the
chance
to
meet
him
in
the
closure
track.
A
A
couple
of
weeks
ago,
a
very
nice
guy
and
I'm
super
super
curious
about
this
presentation.
His
content
is
going
to
talk
about
data
hike
and
data
log
in
general.
Go
as
if
I
understood
correctly.
Conrad
is
going
to
first
talk
about
the
background
and
then
jump
into
live
coding
and
see
how
things
actually
work.
A
A
You
can
ask
whenever
you
want
in
the
middle
of
the
presentation
you
can
also,
if
you
don't,
have
a
microphone
or
feel
a
bit
shy.
You
can
put
questions
in
there
in
the
chat
and
I.
Try
to
monitor
that
and
ask
hundred,
if
there's
anything
through
that
channel,
but
really
be
brave
and
and
discuss
directly.
It's
we
have
a
lot
of
time.
It's
not
a
formal,
formal
gathering.
A
It's
it's
meant
to
be
a
discussion,
and
then
also
we
like
to
just
note
that
if,
if
you're
also,
if
you
notice
I,
we
would
encourage
to
keep
your
camera
on
while
listening,
it's
nicer
for
the
presenter
to
get
the
sense
of
the
people
in
the
other
end
and
also
like
normal
sort
of
how
this
works.
It's
good.
A
If
you
put
your
microphone
on
mute
while
you're
listening
so
background,
noise
is
dogs,
barking
family
live
going
about
their
lives
or
offices
depending
on
what
part
of
the
world
you
are
currently,
so
those
noises
don't
come
come
into
the
stream.
So
please
please
go
on
mute
but
of
course
feel
free
also
to
jump
in
with
with
questions
like
us
and
after.
If
we
have
time,
we
probably
would
like
to
have
a
discussion
about
the
community
challenges
issues
within
the
community.
A
After
Conrad's
presentation
we
have
about
two
hours,
so
I
want
I,
won't
say
their
exact
times,
because
it
will
be
different
in
in
in
all,
depending
where
you
are
but
it
from
from
about.
In
about
two
hours,
we
will
usually
stop
yeah,
okay
enough
about
the
general
things
I'm
super
stoked
and
happy
to
invite
Brad
and
please
the
mic
is
yours.
All.
B
Right,
thank
you
so
much
for
the
introduction.
All
right
I
will
share.
First,
my
screen
all
right
all
right.
So
basically,
I
did
a
similar
talk
a
couple
of
weeks
ago
in
Stockholm
just
before
the
so
some
some
of
the
slides
will
be
will
be
the
same,
but
I
cut
it
a
little
bit
empty.
The
light
coating
will
be
will
be
different
so
for
in
the
first
part,
I
will
just
have
a
smaller
presentation.
B
B
Data
log
itself
is
pretty
popular
in
academic
research,
so
my
co-founder
Christian
is
a
researcher
he's
doing
his
PhD
and
in
Canada
right
now
he's
he
is
he's
pretty
well-known
or
he's
interested
in
in
programming.
Language
research
and
their
data
locus
is
used
widely,
so
they
are
using
it,
for
instance,
for
static
code
analysis.
So
it's
not
it's
not
a
debt,
it's
not
a
dead
language,
but
the
problem
right
now
is
in
the
closure
world.
We
have
only
few
implementations
of
data
and
and
only
few
open
source
solutions,
so
we're
very
data
lock
used.
B
It
is
used
in
in
triple
stores,
so
we
have
stores
that
have
three-part
segments
which
are
entity
attribute
and
value,
and
these
segments
are
stored
on
the
database
and
then
queried
with
data
lock,
and
each
element
can
be
either.
Each
value
can
be
either
primitive
data
type
like
integer
or
long
or
string,
or
it
could
be
a
reference
to
another
data
set
and
the
most
famous
database
in
the
close
of
course,
Taconic
I,
which
made
it
popular.
B
So
it's
transactional
immutable
distributed
database
with
commercial
licensing,
so
you're
paying
for
license
per
year,
and
then
you
can
use
it.
So
there's
a
database
of
effects.
You
know,
so
you
only
append
effects
in
there.
It
is
developed
by
Richie
key
and
it
was
the
motivation
for
closure
in
the
corridors,
several
interdependent
components.
So
it
is
a
separate
client,
server,
storage,
potential,
trans
actor
and
this
console-
and
it
plans
implements
a
flavor
of
data
log
with
a
with
a
certain
syntax
and
since
the
atomic
dozens
or
isn't
interested
in
a
specific
storage
you
can
have.
B
B
B
So
now
now
take
a
look.
How
does
it
look
like
or
first
what?
What
is
it?
It's
decorative
logic,
programming,
language,
it's
a
subset
of
rock
for
for
you
of
who
heard
of
that.
It
uses
first-order
logic
and
it's
not
locked
or
incomplete,
so
it
definitely
terminates
and
it's
in
the
end.
It's
just
a
list
of
predicates
your
your
defining
and
it
tries
to
join
all
together
and
then
you
get
you
get
the
data.
So
how
does
it
imply?
B
B
You
have
you,
have
this
vector
syntax
where
we
have
filed
and
then
variable
you're
looking
for
and
then
the
clause
you
want,
you
want
to
have
so
the
idea
we
want
to
find
the
ID
where
we
have
the
attribute
name
and
the
value
others
and
yeah
this
finds
finds
you
are.
You
did
and
I
will
show
you
some
some
simple
queries
you
can
use
in
there,
but
first
I
will
define
the
effects
that
are
stored
in
our
database.
So
we
have.
B
We
can
make
unions.
So
let's
find
all
the
people
that
have
a
certain
age.
We
can
also
find
anything,
that's
also
possible
and
also
you
can
have
input
in
there.
So
you
can
say:
okay
find
find
the
age
for
a
certain
input.
In
this
case,
we
have
the
input
LS
and
then
we
have
just
the
unification
over
the
name
and
age,
and
then
we
find
we
find
the
value
we
want
to
add.
B
B
The
next
interesting
syntax
is
the
pooled
syntax
in
there.
So
this
this
is
used
to
select
certain
information
from
your
entity.
For
instance,
if
we
have
name
age
and
account
in
our
data,
we
can
say:
okay
either
give
us
everything
or
give
us
the
name
or
the
age
or
if
the
car
is
a
reference,
give
us
everything
that
reference
and
so
on,
and
you
can
you
can
embed
it
in
within
your
queries.
Okay,
so
you
can
say:
okay,
first,
give
me
all
all
entities
we
have
and
then
just
give
me
the
name
of
the
age.
B
B
So
we
try
to
combine
that
so
we
could
have
a
data
lock
engine
or
a
data
log
database
that
can
be
can
be
stored
and
queried
and
yeah.
We
started
two
years
ago
with
that
over
the
years
last
year.
We
didn't
do
that
much,
but
this
year,
with
support
from
from
the
Swedish
government,
we
could
improve
it.
So
we
added
schema
capabilities
and
time
travel
and
we
could
clean
it.
Skin
is
really
up
to
make
it
production
ready
and
within
the
development
process.
We
thought
about.
Ok,
what
what
should
it
be
in
the
future?
B
B
B
B
We
have
different
storage
solutions
in
place
right
now,
so
either
in
memory
or
file
or
LDP
or
post
press.
So
this
piece,
these
things
are
right
now
defined
we
have
protocol,
so
you
can
extend
it
if
you
like,
so
some
people
already
wrote
a
layer
for
first
we.
So
this
is
pretty
easy.
Also,
data,
I,
guess
right
now
auditable.
So
you
have
time
travel
and
the
history
of
all
facts
that
were
added
or
removed
and,
of
course,
with
GPR
compliancy.
We
are
also
able
to
remove
data
completely
and
we
want
to
be
community
friendly.
B
So
we
wanted
to
be
modular
and
with
many
extensible
components
all
right.
So,
let's,
let's
have
a
short
link
and
another
hood.
So
what
are
the
internals
when
I
started?
It
was
like
magic
because
I
didn't
knew
the
code
base
and
it
was.
It
was
not
that
easy
for
me
because
at
that
time,
I
didn't
do
that
much
of
clothing,
but
for
customers
other
stuff.
So
I
dig
into
that
and
I
put
some
things
apart
and-
and
now
it
looks
like
that,
so
we
have
a
simple
API
level
with
connection
from
the
web.
B
B
So
there
are
the
protocols
for
for
all
your
storage
layers
and,
of
course
we
have
the
query
engine
which
has
a
data
log
parser,
which
is
also
in
our
separate
library,
which
joins
the
predicates
from
in
the
in
the
query
and
excuses
just
the
database
search
capabilities
and
the
database
itself
is
just
a
closure
record
which
has
the
different
indices
in
it
and
some
functions
to
search
it
to
store
it
and
so
on,
and
on
the
other
hand
we
have.
We
have
different
indices.
B
So
basically,
as
I
said,
we
have
a
triple
store
and
everything
is
start
as
entity
attribute
value
and
we
have
different
indices
for
which
are
differently
sorted.
So
the
first
is
the
EA
B
and
the
tears
from
transaction
EA
B
is
sorta
like
entity
first
entity
then
attribute
and
value.
The
AEV
T
is
sorted
like
attribute
and
then
entity,
and
then
value
and
the
last
one
is
attribute
value
entity.
This
is
for
four
different
ferries
and
four
different
schema,
and
also
we
have
some
some
temporal
indices
for
just
for
the
storage
of
historical
data.
B
Really
works,
so
if
you
transact
new
data
in
there
like
this
map,
we
have
the
name
Ellis
at
the
age
42,
and
you
just
translate
this
this
vector.
Basically,
what
what
happens
internally
is.
It
is
split
into
the
two
datums,
because
we
have
two
different
attributes
and
then
it
is
stored
as
as
these
couple
into
into
the
store,
and
vice
versa,.
B
B
B
A
B
Where
we
build
some
some
stuff
with
the
public
transportation
system,
and
we
had
some,
we
had
access
to
that
and
it's
also
opened
items.
So
we
between
the
data
and
we
directed
with
it
and
for
this
or
for
my
for
my
example-
I
just
took
one
of
the
data
and
cleaned
it
for
me
to
work
with,
and
they
I
will
show
you
how
our
data,
I
or
data
log
works.
B
B
B
Open
data,
so
this
is
just
data
from
open
data
from
from
from
the
city,
like
parks
or
museums
and
and
stuff
like
that
and
yeah
and
I
used
I
use
just
that
I
use
the
subsets
from
from
that
and
okay,
that's
how
the
data
looks
like
right
now,
so
you
can
just
ignore
the
first
parts
here
they
are
just
for
refer
for
the
queue
data
in
there.
So
basically
we
have.
We
have
just
a
few
attributes.
We
want
to
use
in
there
just
the
name
of
the
location.
The
category
like
is
it.
B
Health
care
is
the
safety,
it's
a
day
care
and
so
on
so
forth,
and
also
some
some
location
data.
Of
course,
it's
name
and
I
just
cleaned
all
that
up
because
open
always
they
taste
a
little
bit
messy
in
there
were
duplicates
and
and
I
did
that
hour
beforehand.
So
right
now
we
have
some.
Let's
see
some
1800
data
points
in
there
and
let's
just
try
to
work
to
work
with
that.
So
basically
what
I?
B
What
I
did
is
I
just
played
around
the
last
days
and
just
thought:
well:
okay,
let's,
let's
do
that
together
and
on
that
way,
I
will
I
will
show
you
how
how
to
interact
with
it
using
using
data
log.
So
here
we
are
so
here's
the
data
and
for
data
itself.
We
need
as
a
configuration,
because
just
using
your
on
in
this
case
I
said:
okay,
I'm
using
a
storage,
the
local
file
system
and
in
the
file
system,
just
in
in
the
temp
folder
and.
B
B
Let's
see
what
what
we
do,
the
name
so
basically,
that's
that's
the
way
I'm
starting
to
work
with
with
new
data,
I
will
just
say:
okay,
how
does
how
does
the
data
look
like
and
what
what
time
does
it
have
and
so
on
so
forth?
So
for
for
data
like
we
need
to
define
for
each
for
each
attribute
three
three
things
for
once:
we
need
an
identifier.
How
does
what
what's
the
name
of
that
attribute?
We
need.
B
B
B
B
B
B
B
B
B
So,
let's
see
so
this
looks
like
that.
So
the
Q
is
for
query
and
then
you
just
define
D,
take
a
look
very
and
then,
of
course,
you
we
have
to
use
the
database
at
that
point
in
time,
which
is
done
really
the
DRF
on
the
country,
and
here
we
can
see.
Okay,
we
have
we
have
the
bomb
for
that.
But
let's
see
okay,
we
have
different
categories.
So
let's
see
how
many
categories
we
have
in
here,
let's
see
category.
Oh,
that's!
That's
wrong!
That's.
B
We
want
to
see
the
categories,
so
we
have
to
count
the
category,
so
you
have
seven
categories
in
there.
Alright.
What
else
can
we
do
we
can
we
can
see
how
the
how
the
distribution
of
the
categories,
for
instance?
So
let's
see
okay,
let's
get
all
the
categories
out
there
that's
wrong.
So
you
have
to
do
it
like
that.
B
Alright,
so,
but
we
we
got,
we
got
only
this
human
costly.
The
return
of
the
of
the
query
is
always
always
set,
so
we
don't
have
duplicates
that's.
Why
I?
Why
I
have
to
add
the
the
entity
ID
here,
so
we
can.
We
get
the
yeah
can
each
entry.
So
now
we
have
no.
We
have
a
list
of
all
the
categories,
all
entries
in
its
category
and
let's,
let's
just
work
with
that
data
and
say:
okay,
category
and
the
frequencies,
and
we
can
see
okay,
how
many
different
categories
are
there
in
our
in
our
database?
B
It's
it's
wrong
second,
and
then,
and
we
can
see
okay,
the
most
part
is
steak
here,
because
in
the
the
hackathon
we
selected,
we
selected
the
data
that
way
we
have
some
some
culture,
some
environment
which
are
like
Gardens
or
something
like
that
with
education,
for
schools,
into
universities,
health
for
doctors
and
so
on,
and
safety
for
I,
don't
know
what
that
was,
and
traffic
which
are
just
terminals
and
turbulence
and
stuff
like
that
yeah.
So
this
is.
This
is
how
you
can
get
some
very
simple
data
from
that.
B
What
we
can
we
can
do
more.
Let's
see,
maybe
maybe
you
can
see
yeah,
let's,
let's,
let's
use
the
qh
I
created
in
there.
So
basically
what
what
the
gear,
which
is
it
creates
in
a
certain
area,
one
one
hash
for
all
the
electrical
authority
to
which
is
in
there
and
I'm,
using
that
to
find
the
surroundings
of
my
of
my
point
on
in
a
map.
So
if
we
have
a
look
at.
B
B
Different
areas
and
each
area
has
the
same
hash
value,
and
then
you
can
say:
okay
if
every
entry
in
this
area
has
that
same
hash
value.
So
this
is
nearby
and
you
can
use
that
to
find
things
in
your
database
and
yeah
I
did
turn
here.
So
that's
what
what
I
did
with
these
all
these
functions
up
here
and
as
a
first
point,
I
just
copied
the
from
the
Helsinki
train
station
in
here
and
wanted
to
work
with
that.
B
So
this
is
just
latitude
and
longitude,
and
here
I
wanted
to
see
okay,
what
what
hashes
are
nearby
so
this
this
gives
me
a
certain
amount
of
fashion's.
So
look
at
that.
Okay,
these
are
these.
Are
the
hazard
just
nearby
and
we
can
query
that
and
see
what's
what's
in
there
and
let's,
let's
do
that.
So,
let's
say
okay
find
find
me
all
entities
with.
B
B
B
Yeah
yeah
train
station.
That's
it
yeah,
that's
a
list
so
we
have
to.
We
can
also
as
input
we
come
out,
so
we
can
have
single
values
like
like
this
one,
but
also
we
can
ever
a
whole
list
which
can
be
which
can
be
used
which
can
be
used
to
join
values,
and
in
this
case,
let's
see
what
we
can
find
here.
Okay,
we
can,
we
can
find
some
okay,
I
should
I
should
use
the
English
name,
probably.
B
B
Let's
say
we
only
care
for
for
a
certain,
so
we
can
just
add
another
value
in
there.
Let's
say:
okay,
we
want
to
have.
We
want
to
search
only
for
a
certain
category
and
then
we
just
edit
it
there.
So
we
say:
okay
now,
I
want
to
find
all
nearby
health
care
facilities,
and,
oh
okay,
there
are
none.
Let's
see
the
equation.
B
B
Yeah,
so
that's
that's,
okay.
What
we
are
working
on,
okay,
yeah,
so
now
I
will
show
you
a
little
bit
the
pool
syntax
I
mentioned
before
so
you
can
select
any
data
you
like,
so
you
can
just
say
well
and
we
want
to
have
maybe
everything
that
this
entity
and
you
can
see:
okay,
the
ID,
the
category,
the
t,
where's,
the
location
and
so
on
and
so
forth.
And
that's
that's
basically
yeah!
That's
that's!
Basically
we
do
it.
You
can
also
just
use
the
basic
pool
syntax
for
that.
B
B
Name:
let's:
let's:
let's
do
that,
the
intrinsic,
so
you
can
update
your
schema
as
as
you
like
concern,
okay,
that's
her
name
and
should
be
also
the
should
be
unique.
That's
very
differently!
That's
since,
since
we
have
to
create
a
new
index,
it
doesn't
do
it
right
now
in
the
in
the
background,
when
you
just
change
the
schema,
let's
create
the
database
from
from
you
just
to
do
that.
So
that's
that's.
B
It's
the
easiest
way
right
now
what
we
have
to
so
to
do
that
or
to
be
able
to
do
that
in
the
future.
So
now,
I
will
I
will
just
this
unique
identifier
for
the
name.
So
we
can.
We
can
index
that
by
a
by
value
in
the
future.
So
let's,
let's
do
that
again:
let's
transact
all
the
data
again
and
then
we
can.
We
can
pull
the
data
just
using
just
using
it's
time,
for
instance,.
B
Now,
right
now,
it's
possible
okay,
so
can
see
here.
The
EB
Botanic
Garden
can
just
reference
it
bites
by
its
name
it
not
not
by
just
by
some
some
ID.
So
if
we
want
to
have
it
like
an
eye
like
an
IDE
or
directly,
we
can
also
do
that.
Something
like
that.
This
is
also
possible
if
you,
if
this
doesn't
have
an
unique
index
all
right,
let's
see
what
what
else
we
can
do.
B
B
B
B
B
Okay,
now
we
try
to
pull
it
yeah.
So
it's
it's
not
found
in
the
current
index
anymore,
so
the
entity
is,
is
gone,
let's
see
if
we
could
find
it
in
the
history.
So
since,
since
the
last
release
we
have,
we
have
a
history
which
can
be
which
can
be
queried.
The
pullback
that
puts
on
tax
isn't
working
there,
but.
B
B
Let's
see
yeah,
let's
see
is
still
in
there
and
yeah
in
the
history.
You
can
find
it
and
let's
see
how
this
looks
like
if
we
just
carry
that
in
the
in
the
normal
I,
don't
know
in
the
in
the
current
yeah
and
it's
gone
so
now
we
have
Planck
element.
You
can
also
query
that
at
a
certain
point
in
time,
if
you
like-
let's
say
we
have
the
current
date,
which
is
just
a
few
to
date:
okay,
let's
retreat
something
else.
Let's
say
that's.
B
History,
it
should
be
and
the
current
database
it
should
be
gone
yes
in
the
history
you
would
find
you
would
find
it,
but
you
can
also
see
that
you
would
see
the
whole
history
of
everything,
so
you
would
see
a
current,
a
current
entry
and
also
the
past,
but
if
you
like
to
see
the
database
at
a
certain
point
in
time,
you
will
use
the
as
off
database
with
the
day
you
want
to.
You
want
to
see
it,
and
here
you.
B
A
C
B
B
Let's
see,
let's,
let's
it,
you
can,
let's
see
how
our
reference,
so,
let's,
let's
transect
something
like
okay,
let's,
let's
have,
let's
have
a
an
entity.
That's
that's
that's
a
visit
at
a
certain
date
and
let's
edit
as
a
reference.
Let's
at
one
of
the
location
we
found
nearby
and
as
a
and
then
we
can,
we
can
see
how
how
the,
for
instance,
this
nested
put
syntax
would
work
like
so
how
our
references
in
queries
would
work
to
other
to
other
entities.
So
it
would
look
like
that.
So,
let's,
let's
have
it
like
something
like.
B
Instant,
so
at
the
moment
we
we
only
support
a
subset
of
of
the
schema
of
atomic,
so
missing
our
right
now,
tuples
and
some
white
stuff,
but
internally
you
should
be
able,
in
the
future,
to
define
any
closure
spec
you
like
to
and
define
just
new
new
shapes
of
your
data.
This
is
not.
This
is
also
something
we're
working
on
right
now.
This
is
not
not
finished
yet
and.
B
Yeah,
that's
all
its
have
many
notes
in
there.
Yeah
that's
okay
and
let's
see
how
hard
it'd
be
many
reference
looks
like
and
then
we
have,
of
course,
the
location.
So
this
is
you
see
it's
a
typo.
Well,
you
type
the
eat
of
time,
and
this
is
a
reference.
So
here
we
will
reference
the
each
each
location
we
have
in
there
and.
B
B
B
B
We
have
the
location
as
a
reference,
and
you
can
see
here
the
references,
it's
a
DVID
and
with
a
standard
pool
you
only
get
the
reference,
and
if
you
want
to
have
the
data
behind
the
reference,
you
will
just
say:
okay
visit
location
and
there
you
can
have
again
the
pool
synthetics
you
want
to
have
what
you
want
to
have,
and
in
this
time
we
get
the
nested
data.
So
we
have
the
location
again
and
human,
the
location
we
get
all
the
data
from
the
locations,
so
they
keep
gee.
B
B
B
B
B
B
B
B
Okay,
so
now
we,
if
we
go
up
and
and
have
a
look
at
the
environment,
yeah,
obviously
it's
gone
and
also,
let's
copy
that
down
and
fact
and
try
to
find
this
name
in
the
in
the
history
and
it's
also
gone
so
the
data
is
completely
removed
from
from
from
the
history
and
from
the
current
view
and
yeah,
and
here
GPR
officer
would
be
happy,
yeah,
I
think
what's
the
time
yeah
are
there
more
questions?
What
what
do
you
want
to
see
what
I
will
do
with
the
data
cuz?
A
B
B
About
it
stores
its
data
in
without
a
schema
in
it
in
a
document
store,
so
it's
like
a
document
store
and
not
a
triple
store
like
like
us.
It
also
has
an
index
in
their
search,
but
it's
it's
not
a
triple
store,
yeah
and
also
I'm.
We
were
we're
trying
to
work
with
them
so
to
have
some
sunlight
booster.
That
would
work
on
data
high-end
Crocs
so,
for
instance,
working
working
with
the
data
log
engine
or
how
to
manipulate
the
data
loggers.
D
B
Yeah
closure
script
up
until
the
last
release.
The
script
was
supported,
but
with
also
with
update
from
from
some
libraries,
we
were
using
the
support
isn't
anymore,
so
we
we
can't.
We
can't
respect
right
now
and
we
have
to
fix
that
so
I.
But
right
now
we
have.
We
have
an
internal
discussion
and
whether
we
want
to
support
closure
script
or
not,
because
we're
moving
data
to
the
server
and
not
to
the
browser.
So
and
then
it's
the
question
why?
B
B
E
Hi
yeah,
so
my
my
question
is
actually
an
apology,
because
I
open
an
issue
on
conserve
the
consular
website
just
before
the
the
meet
up,
because
I
was
looking
at
the
code,
I
was
wondering
how
conserved
was
creating
store,
backed
by
a
hitchhiker
tree
and
I
realized
that
it's
something
that
is
not
exposed
in
the
API
of
conserved
from
a
user
perspective.
So
if
you
look
at
the
concert
with
me,
you
will
see
how
to
create
file
stores
in
memory
five
days.
E
B
B
D
B
B
F
A
B
B
But
I
started
I
started
writing
a
blog
post
about
that
with
with
the
numbers
with
the
first
numbers
and
there
you
can
see
how
how
the
different
starch
layer
bags
on
different
and
also
the
different
indices,
because
you
can,
if
you
so.
Basically,
we
have
the
each
hiker
three
as
an
index
and
also
the
persistent
sorted
set
from
data
script
and
also
I
I,
compared
how
they,
how
they
behave
on
different
on
different,
with
different
in
silence.
I.
E
You
know
hard
link
with
the
implementation,
so
the
question
I
have
is
is
yeah,
so
the
question
is,
is
the
only
thing
is
that
is
this?
The
maximum
obstruction
is
it
it
is
abstracted
to
where
in
the
map
in
the
object
it
the
most
obviously
in
the
maximal
way,
or
is
it
possible
to
be
beyond
the
past
or
maybe
there
is
some
logic
that
can
be
abstracted
away
from
the
underlying
implementation
or
is
or
not
like.
It
is
this.
E
B
B
E
B
B
Look
in
here
the
data
just
just
matches
for
the
for
the
triple
entries
in
the
database.
Of
course
we're
also
it's
not
just
triple
they're,
also
transaction,
and
in
atomic
there
is
some
some
added
entity,
so
you're
trying
you're,
storing
five
and
five
entries
in
there,
but
the
data
not
just
matches
for
that.
So,
but
what
internally
happens
is
we
just?
B
B
So
basically
you
can.
You
can
write
any
efficient
query.
So
let's
say:
if
we
have
such
a
query,
you
can
write
it
in
a
way
where
you
can,
where
you
join
pretty
late,
where
you
filter
data
pretty
late,
and
this
can
be
optimized.
So
you
can.
You
can
see
you
once
you
have
statistics
from
the
data.
You
can
see.
Okay,
if
you
draw
an
hour
and
data
or
a
small
data
set
early
on
it's
it's
better.
So
you
can
write
a
planner
for
that.
B
A
D
And
yeah,
so
if
that
is
okay,
then
maybe
I
I
could
ask
Conrad,
since
you
are
using
it
your
company,
and
do
you
feel
that
you
need
some
layer
above
data
log
in
yo
and
I'm
asking
it
also
because
Danielle
earlier
told
us
that
he
has
built
a
certain
layer
above
database
kind
of
an
OEM?
Maybe,
and
and
do
you
feel
that
anything
is
necessary
as
another
layer
used
together
with
the
data
log
in
general.
B
Layers,
because
you
have
to
write
so
many
things
again
and
again,
and
the
even
the
put
some
taxes
and
often
that's-
that's-
that's
really
fun,
and
if
you,
if
you
want
to
have
more
features
like
do
you
have
stuff
you
can
you
can
you
can
add
that
data
lock
itself,
so
you
can
have
in
the
future.
If
you
it's
not
the
area,
but
you
could
have,
let's
see,
that's
it,
so
you
can
have
a
function
in
here
that
that
finds
that
finds
nearby
data,
for
instance.
So
this
is
possible.
I.
B
E
E
Data
hike
actually
is
a
much
is
a
production
already
system.
It
has
many.
Many
more
features
than
the
combat
Campbell
is
a
very,
very
small
thing.
That
only
does
this
case
scheme
on
right
thing
and
it's
not
it's
not.
It's
like
a
one
person's
project
for
my
needs.
I
wouldn't
so
that
is
very
nice.
It's
a
it.
Has
it's
a
so
it's
a
complete
system.
E
It
has
many
features,
actually
yeah
yeah,
so
yeah
I
mean
it's
great,
that
we
have
these
dis
options,
because
when
when
commit
act
announced
the
atomic
they
made
an
excellent
campaign
around
it
and
people
really
wanted
it.
Of
course
it
was
close
source
and-
and
it
was
for
pay-
and
it
was
be
difficult
for
some
people
and
in
some
context,
so
now
we
have,
we
have
more
well,
you
know
have
more
options
and
and-
and
it
looks
like
it's
only
expanding-
because
this
is
two
new
databases
in
one
year
and
with
Crocs,
and
we.
D
B
B
E
E
E
B
A
B
G
B
B
B
A
Bill
who
asked
the
original
question
just
said
he
had
to
leave
for
another
appointment.
Okay,
thank
you.
I
think
it
was
very,
very
clear
while
other
people
are
are
thinking
about
questions.
Let
me
ask
a
little
bit
of
a
more
general
question.
When
we
met
I,
think
Daniela
was
asking
sort
of
a
question
well
how
way
what
what
our
personal
wishes
and
visions
are
for
the
community
and
the
development
of
of
things
and
I.
E
A
B
C
E
I
think
data
log
is
a
is
a
fascinating
piece
of
idea
because
it's
it
goes
back
to
prologue
the
logic
programming
paradigm.
So
that's
where
the
roots
of
data
log
is
in
Prolog
and
in
logic,
programming
and
logic,
programming
itself,
as
it
rules
in
artificial
intelligence
and
goes
back
to
the
beginning
of
the
computers,
computer
research
into.
A
E
F
E
E
E
E
Programming,
language
and
yeah
I
mean
there
are
many
resources,
but
that's
that's
basically
why
I
am
interested
personally
in
that
a
lot
because
of
this
heritage
and
this
legacy
and
I'm
happy
to
see
that
it
still
lives
on
in
these
kind
of
small
small
things.
So
like
data
log
and
whole
logic,
American
remnant
stuff
like
that.
F
D
Maybe
if
I
can
ask
a
small
question
and
so
Connor
Connor,
you
had
some
examples
showing
how
you
use
data
log
for
actual
analysis
of
data,
which
is
something
that
many
people
like
to
do
with
SQL
and
and
I.
Think
you
also
shot
some
little
examples
that
how
extensible
it
is,
how
sometimes
you
actually
use
functions
inside
your
data
data,
log
and
I,
wonder.
Well,
it
stops.
B
Basically,
I
have
to
see
I
have
to
see
where,
where
the
the
end
is
so,
we
haven't
started
yet
to
to
use
any
any
analysis
functions
in
data
itself,
so
I
just
extended
data
log
by
some
functions
like
I,
don't
know,
take
functions
and
stuff
like
that.
I
haven't
I,
haven't
reached
that
yet
so
I
don't
know,
I
think
it's.
It's
I
think
the
the
limits
are
what
what
the
JVM
can
do.
G
B
So
I
just
extended
data
itself
with
so
basically
we
have
some
built-in
functions.
So
if
you,
if
you
have
stuff
like
stuff
like
less
something
like
that,
these
these
functions
are
built
in
right
now,
so
they
were
in
data
script,
so
they
they
had
a
map.
I
can
I
can
just
show
you
with
the
code
if
you
like,
yes,
Chris
so
so,
basically
they
are
just.
B
B
Yeah,
so
here
we
have
some
some
functions
that
that
are
in
in
a
data
lock
right
now,
so
here
we
have
get
else,
get
some
function,
of
course
the
the
something
like
missing.
So
if
we
we
want
to
find
missing
stuff
and
I
edit
a
lesser
and
greater
functions
in
there,
so
you
can.
You
can
use
that
in
in
data
box.
B
Oh
and
here
you
have
just
a
big
map
of
functions,
you
can
use
so
UK
here
you
can
see
okay
equals
as
the
equals
functions
and
so
on
and
so
forth
and
the
lesser
function,
that
is
this
lesser
function,
the
bigger
function
at
the
greater
function.
Is
this
one
and
so
on
and
so
forth,
and
we
want
to
extend
this.
This
has
some
fashion
above
function,
so
you
can
say:
okay,
I
want
any
functions.
Any
side
effect
free
functioning
there
that
that
that's
possible.
A
There
is
theater
is
posting,
a
question
that
at
the
chat
can
can
you
convert
a
see
it
because
it
will
be
maybe
a
little
bit
difficult
to
read
out
I
start
in
any
case,
so
I'm,
quoting
it.
We
were
talking
a
little
about
prologue
in
prologue.
You
can
express
a
personal
system
relation
by
expressing
it
as
a
recursive
broom
involving
a
person
parent
relation.
Can
this
be
done
with
data
log.
A
A
D
I'm
not
sure
if
it
is
a
good
question,
because
maybe
the
answer
is
too
long
for
the
time
we
have
but
Conrad
would
you
like,
maybe
to
tell
a
little
more
about
the
implementation
and
how
the
implementation,
the
data
structures
are
maybe
different
than
other
implementations.
Is
it
a
good
question
for
this
time.
B
As
I
said,
it's
a
fork
of
a
tough
script,
and
so
many
many
things
from
from
data
script
code
is
still
in
there.
I
just
pulled
it
apart
and
edit
stuff
and
I
had
a
header
just
some
some
things
in
there.
So
basically,
a
lot
of
the
internal
data
structures
are
still
like
in
like
a
data
script.
So
if
you
look
at
the
database
records,
they
are
pretty
much
the
same
I
just
added
some
more
in
eces
and
some
more
attributes
there
and
the
whole
interaction
with
with
the
query.
B
Engine
I
didn't
touch
that
because
the
query
was
working
and
I
hadn't
the
time
this
year
to
to
go
into
that
I
I
have
I
will
have
an
understanding
how
it
works,
but
I
haven't
written
that
much
into
that,
so
yeah,
basically
the
primary
Krotz.
We
worked
on
this
as
a
set.
Okay,
it's
it's
the
it's
the
schema
stuff.
B
It's
the
temporal
index
I
worked
on
edit,
see
it's
really
putting
it
apart,
making
making
as
many
packages
as
as
possible
and
having
as
many
abstractions
above
the
different
components
as
possible,
and
the
other
part
was
of
course,
the
storage
implementation.
But
it's
at
that
point.
Some
some
guys
helped
us
with
the
with
the
protocols
above
the
stores.
So
this
was
just
haven't
written
that
it
was
just
a
pull
request
and-
and
we
had
the
protocols
with
the
four
functions
that
that
are
necessary
for
storing
it
in
on
different
time.
Different
packets.
A
D
B
B
Could
you
could
you,
and
in
there,
and
also
the
the
other
thing,
is
right
now
associate
we're
working
on
the
client-server
stuff
and
right
now
we
were
in
the
in
the
design
phase
how
to
how
to
do
that?
How
the
clients
are
they
how
they
should
look
like,
but
we
have
right
now
only
some
some
ideas,
how
the
communication
works
between
that
and
what
the
best
practices,
which
would
people
that
and
how
the
security
works,
I
think
I
would
I
would
like
to
have
some
some
open
discussion
on
how
to
how
to
do
that.
A
B
Zulu
or
just
a
kid
hub
issue
or
in
this
in
the
select
random
trust,
just
write
us
and
yeah,
and
the
the
discussion
itself
would
be
I.
Think
some
I
will
post
our
ideas
about
the
client-server
model
on
github
in
a
markdown
file,
and
you
can
just
just
have
a
pull
request
with
your
own
ideas
on
that
and
or
your
own
comments.
And
then
we
can
discuss
that.
There.
A
Yes,
if,
if
not
feel
free
to
jump
in
still,
if
you
think
of
something,
but
if
not
and
I'm
before
we
may,
we
wrap
this
up.
I
should
I
should
announce
the
LA.
The
next
meeting.
It's
pretty
soon
there's
going
to
be
a
meeting
the
28th
of
October
this
month
in
a
couple
of
weeks
and
that
time
we
will
have
Tommy
Raymond
from
from
Finland
as
well.
He
will
talk
about
manly.
A
Plain
data
schemas
in
closure
and
closure
script.
So
so
please
keep
an
eye
on
Eaton
and
I'm
joined,
join
again,
it's
of
course
related
to
to
the
subject
matter
of
this
talk
and
and
the
the
guy
Thomas
gang
are
doing
interesting
stuff.
So
it
could
be
interesting
to
talk
to
him
as
well.
So
please
put
that
in
your
agenda,
your
calendars
and
and
join
us
again
then,
but
yeah.
Maybe
we
could
wrap
this
up.
It's
many
many
thanks
Conrad.