►
From YouTube: sqlotus preview
Description
Uploading this mini version here because one can speed up on youtube. DrRobinson - let's trash this when the better versions come.
A
Yeah
cool
next
up,
we
have
peter
reba
talking
about
sequel,
lotus,
I'm
giving
us
a
preview.
This
is
what
will
be
the
back
end
of
insta
lotus,
which
maybe
you
can
provide
a
bit
more
context
on,
but
also
I
recommend,
reading
the
proposal
for
it,
because
it's
really
good
hey,
we
will
go
off.
B
Everybody
can
see
a
terminal
right,
yep,
perfect,
okay,
cool,
so
this
journey
actually
started
like
way
back
in
december,
when
I
needed
some
sort
of
very
up-to-date,
essentially
a
real-time
sentinel,
so
to
speak,
and
we
didn't
have
this
at
the
time.
So
I
started
digging
into
things,
and
this
is
a
combination
of
about
like
one
month
of
work.
You
know
spread
out
over
time,
so
I
will
just
walk
through
what
you
see
on
the
screen
and
it
will
become
obvious
like
what
this
is
all
about
until
later.
B
So
what
you
see
right
now
is
a
lot
of
demon
actually
running.
You
know
it
is
sync
to
chain
it's
like
five,
five,
two,
nine
eight.
If
you
go
to
fuel
fox,
you
can
see
that
it
is
all
well,
it's
297
because
you
know
fuel
fox
itself
is
lagging
behind
a
little
bit,
but
you
know
everything
works
correctly
and
it
will
at
right
now
it's
30
second
mark.
It
goes
to
99
so
and
so
forth.
Now
what
is
interesting
here?
This
is
an
entire
full
node.
B
But
if
you
look
at
our
free
space,
we
basically
take
only
7.9
gigs
and
no
tricks.
They're,
like
no
no
disks
or
anything
like
this
attached
to
this
thing,
in
fact,
if
we
go
and
look
in
the
lotus
folder
and
look
in
the
data
store,
there
is
no
chain
directory
at
all.
This
entire
thing
works
fully
under
network,
taking
a
completely
remote
block
store
and
using
it.
You
know,
as
we
go.
B
One
of
the
advantages
of
not
having
a
badger
in
store.
This
is
a
full
node
that
you
know
that
takes
like
no
amount
of
ram
at
all
it
just
you
know,
sits
there
and
runs
and
does
its
thing
and
on
the
server
side
to
which
I
just
switched.
B
We
have
basically
just
post
wrestling
there
and
it
uses
also,
essentially
no
memory
at
all
now
made
the
the
demo
gods
be
kind.
What
I
have
here
on
this
box
is
another
lotus
which
once
again,
is
the
same
database,
but
with
its
own
state
and
everything,
so
it
shares
the
entire
block
store
with
the
one
that
is
running
here.
B
That
is,
you
know,
continues
to
operate,
but
it
has
been
stopped
quite
some
time
ago,
so
we
are
now
at
555.302
and
I
shut
down
like
a
214.,
so
I'm
going
to
put
it
back
up
and
it's
going
to
start
working
and
you
may
notice
it
starts
instantly.
There
is
no,
like
you
know,
time
to
open
badge
or
anything
like
this.
It's
like
there
right
away
and
right
now
it
starts
from.
I
did
not
resize
something
here.
B
One
moment
we
want
to
see
just
the
top
oops,
just
the
top
two
yeah,
so
it's
starting
from
you
know
five
five,
one
three,
where
I
pretty
much
shut
it
down
and
what
we're
going
to
do
is
while
it's
been
down,
the
other
node
has
been
constantly
populating
the
block
store,
so
we're
just
going
to
jump
forward
without
you
know
having
to
do
anything
else
and
let's
see
if
this
works,
let's
see
if
you,
where
are
we
now?
B
Let's
take
this
too
guys.
So,
ideally,
if
I
do
this-
and
if
I
do
this.
B
It
should-
and
it
just
took
it
so
we're
now
four
blocks
away
from
being
synced,
and
if
we
do
this
in
about
in
about
like
this,
current
epoch
is
weird,
I
remember
we
actually
need
to
restart
photos
after
we
do
set
head
because
it
will
not
interrupt
the
thing
that
is
going
right
now.
So
now,
if
we
start
it,
it
is
going
to
be
at
the
right
place.
Yep
there.
B
It
is
and
now
to
basically
say
that
it
needs
to
do
four
and
it
will
be
ready
to
go,
and
that
is
almost
always
the
insta
lotus
proposal
that
I
that
I
wrote,
although
I
still
have
like
stuff
on
disk
and
so
on
and
so
forth.
B
So
it's
not
fully
fully
in
place,
but
that
is
the
part
that
basically
would
allow
somebody
to
have
a
replica
of
you
know
of
a
database
like
this
and
just
stream
it
to
something
like
will's
graphql
interface,
which
here
we
literally
don't
run
anything
aside
from
a
postgres,
a
replica
that
is
streaming.
B
You
know
all
the
changes
from
this
main
note
in
and
just
here,
stateq
server,
nothing
else,
and
this
thing
will
tell
us
that
we're
now
at
300,
where
is
which
is
where
we
left
this
node
at
and
if
we
refresh
now,
we
should
see
once
it
gets
to
302.
The
graphql
is
one
block
behind
always
so
once
this
goes
to
302
the
graphql
will,
you
know,
do
things
for
us
as
expected
and
yeah.
B
So
this
entire,
like
contraption,
is
basically
to
allow
places
like
euro,
for
example,
to
run
a
bunch
of
nodes
out
of
the
same
data
store
which,
because
this
is
like
a
grown
up
database,
so
to
speak.
It
is
designed
for
multiple
reads
over
few
rights
and
it
is
extremely
optimized
for
huge
data
stores.
So
we
can
like
throw
all
our
three
terabytes
of
data
into
this
thing
and
it
doesn't
bad
at
night.
It
just
continues
continuous
working
and
yeah.
Now
one
more
thing
that
I
want
to
like
go
over
here,
real
quick.
B
If
you
see
here,
we
scroll
that
it
took
about
nine
or
seven
seconds
to
validate
the
tip
set.
The
reason
this
happens
is
because,
in
addition
to
keeping
track
of
all
the
blocks
that
go
in,
we
also
keep
track
of
the
chain.
So
we
actually
keep
track
of
orphans
and
stuff
like
that
within
this
very
same
store,
so
that
you
can
stream
out
the
replica,
and
somebody
by
just
looking
at
the
database
knows
exactly
where
the
head
is
what
orphans
are
like.
B
You
know
what
the
chain
looks
like
from
the
perspective
of
this
particular
instance
that
has
gone
into
right
into
this
database
and
each
individual
instance,
like
both
this
instance
and
this
instance
arrived
to
separate
places
in
the
database.
They
have
their
own
instance
name
spaces
so
that
you
potentially
could
run.
B
I
don't
know,
like
20
notes
that
all
right
to
the
same
place-
and
you
will
see
all
of
them-
how
are
they
seeing
orphans,
how
the
chain
evolves
from
their
point
of
view,
aggregate
in
the
same
place,
and
the
other
thing
that
we
do
here
is
that
every
individual
block
that
is
sent
into
this
block
store
is
actually
parsed
on
the
fly
and
its
entire
ancestry
is
being
recorded
in
the
database
as
well.
B
So
you
can
actually
do
graph
walks
through
the
chain
or
through
individual
blocks
or
whatever,
within
false
graphs,
with
just
recursive
queries
and
nothing
else,
and
this
actually
has
implications
also
like
if
we
decide
to
reuse
this,
for
something
like
ipfs.
B
There
is
a
way
to
retrofit
this
for
garbage
collection,
because
then
you
have
a
reverse
index
of
I'm
a
blog
and
who
actually
is
mentioning
me
and
if
nobody
is
mentioning
me,
I'm
just
going
to
delete
myself
and
then
you
do
the
scan
again
and
you
basically
go
level
down
a
little
down
level
down,
at
which
point
you
basically
will
have
garbage
collected
here
in
the
star
store
without
any
locks
without
any
without
anything
and
yeah.
All
of
this
is
already
on
github.
B
I
need
to
grab
somebody
from
lotus
to
basically
figure
out
how
how
to
massage
some
of
this
stuff
in
will
has
already
written
a
front
end
for
this,
so
that
tommy
is
trying
to
like
get
sentinel
to
run
off
of
this
up-to-date
data
store
and
yeah.
That's
it
any
questions.
A
I
think
questions
in
chat
have
already
been
answered.
This
is
fantastic
work.
How
many
of
our
project
partners
are
already
aware
of
this,
because
a
few
of
them
will
like
this
eh.
B
A
C
A
All
right
last
call
for
questions.
I
don't.
I
hope
I
didn't
miss
it
just
on
me
yell,
if
you
have
any
questions.