►
From YouTube: Hyperledger Sweden Study Circle - May 23, 2020
Description
This is a recording of the weekly Hyperledger Study Circle hosted by Hyperledger Sweden where community members go through online training resources to prepare for Hyperledger certification exams. This presentation included Arun S. M. presenting "Hyperledger Sawtooth - Smart Contract Development".
More information about this event is at:
https://www.meetup.com/Hyperledger-Sweden/events/ztszrrybchbfc/
If you are interested in attending future Study Circle calls, please feel free to join. Details are at:
https://www.meetup.com/Hyperledger-Sweden/
----
Hyperledger – Open Source Blockchain Technologies
Hyperledger is an open source community focused on developing a suite of stable frameworks, tools and libraries for enterprise-grade blockchain deployments.
It serves as a neutral home for various distributed ledger frameworks including Hyperledger Besu, Fabric, Sawtooth, Indy, as well as tools like Hyperledger Avalon, Cactus and libraries like Hyperledger Aries, Ursa. Learn more about Hyperledger projects: https://www.hyperledger.org/use
Case Studies: https://www.hyperledger.org/learn/case-studies
Training & Certification: https://www.hyperledger.org/learn/training
Tutorials: https://www.hyperledger.org/use/tutorials
Webinars: https://www.hyperledger.org/learn/webinars
Events: https://www.hyperledger.org/events
Vendor Directory: https://www.hyperledger.org/use/vendor-directory
Subscribe to the Hyperledger Newsletter: https://www.hyperledger.org/newsletter
Follow-us on Twitter @Hyperledger
Learn about Hyperledger Membership: https://www.hyperledger.org/about/join
#Hyperledger #HyperledgerMeetups #Blockchain
A
And
Kent
requested
that
I
go
through
this
slide
deck,
which
we
initially
prepared
for
hyper
nature,
Global
Forum,
which
happened
in
March
and
Phoenix
and
I'm
really
happy
to
go
through
this
slide
deck
and
answer
any
of
your
questions
on
hyper.
Laser
sought
to
find
transact
in
general
or
if
you
have
any
other
questions
and
feel
free
to
shoot,
shoot
I
mean
now.
You
may
also
reach
out
to
me
later.
A
A
B
A
Sure,
when
I
initially
started
on
started
working
on
sawtooth,
it
was
more
of
getting
to
know
how
Appalachia
sawtooth
works
versus
other
frameworks
I'm
getting
introduced
to
some
of
the
then
the
under
development
activities.
The
first
version
of
hyper
ledger
Sawtooth
had
the
consensus
algorithm
integrated
alongside
with
it
and
then
with
later
releases
or
not
about
five
consensus,
engine
was
moved
out
of
the
core
component.
A
Validator
and
B
does
have
we
and
after
that,
we
now
we
now
we
have
multiple
consensus
engines
along
with
point,
so
that's
rails,
that's
the
time
when
I
started
getting
involved
into
the
project
and
one
way
in
which
I
would
suggest
most
of
the
beginners
to
get
involved.
When,
when
you
say,
if
you
want
to
get
involved
into
project,
there
are
multiple
aspects
of
it.
One
could
be
that
your
enterprises
want
to
build
a
solution
on
this
platform.
A
If
that
is
the
case,
then
the
bit
based
approach
would
be
to
understand
the
architecture,
diagram
or
understand
the
documentation
through
one
example.
Application
I
would
strongly
suggest
you
to
try
out
an
existing
application
before
you
develop
your
own
application
and
then
start.
So
writing
your
own
sample
application
on
Sawtooth
is
it's
much
simpler
than
what
you
think
it
may
look
clumsy
initially,
but
once
you
understand
the
concepts
then
you
you'll
be
like.
Oh,
it
is
so
easy
to
write
an
application
on
so
to
that's.
B
A
Us
to
get
involved,
and
if
you
want
to
get
involved
in
two
core
contributions
itself,
then
best
way
to
get
involved,
would
be
to
be
active
on
rocket
shot
channels
and
subscribe
to
the
PR
requests.
Any
put
requests
on
any
of
the
sort
of
repositories
just
make
sure
that
you
are
in
loop
and
you
understand
how
the
functionality
works.
B
A
A
Let's
get
started
on
understanding
so
to
this
this
light
deck
is
very
comprehensive.
It
was
originally
planned
for
a
complete
three
hour
session
and
I'll
try
to
get
this
as
well.
I'll
try
to
get
to
so
far.
They
in
the
next
one
are
in
the
next
50
minutes
and
I'll
try
to
keep
it
interactive
and
any
slide.
You
any
slide
like
you
find
in
interesting
God,
you
find
it's
not
I'm,
not
able
to
understand,
follow
through
it.
Please
feel
free
to
ask
questions.
A
A
A
So
I'm,
sorry
for
the
disturbance.
There
is
no
curfew
here
in
India
from
7:00
to
7:00,
so
you
may
be
hearing
some
disturbances,
so
yeah
I
mean
why
we
initially
why?
Why
do
we
even
need
blockchain,
when
we
already
have
enterprise
application
solving
some
business
problems
right?
That's
where
that's
where
the
question
starts
and
in
normal
transactions.
What
we
have
is.
We
have
a
trusted
third
party.
B
A
Awesome
thanks
guys
so
yeah.
When
two
party
wants
to
perform
any
transaction,
they
always
relied
on
third
party.
For
these
reasons,
they
want
to
monitor,
what's
happening
between
them,
so
that
there
is
trust
established
with
the
third
party
and
this
third
party,
you
could
be
in
government
agency
or
a
trusted
bank,
but
so
general
transaction
happens
in
presence
of
retrospect
third
party
and
talking
about
the
trusted
third
party.
A
If
I
have
to
map
it
in
digital
world,
you
can
imagine
it
to
be
something
like
a
centralized
database
where,
if
I
am
transacting
with
somebody
else,
then
I
will
put
all
the
log
of
transactions
or
log
of
records
whatever
happening.
Whatever
is
happening
between
me
and
another
person
into
that
ledger
and
that
gets
accumulated
low
and
over
again,
and
if
we
have
some
disputes
we
go
to
that
trusted
third
party
and
we
say
hey.
This
is
what
I
did
two
days
ago,
and
this
person
is
not
believing
me.
B
A
So
yeah,
now
that
we
have
seen
how
blockchain
concept
has
to
be
seen
in
general,
the
if
I
have
to
put
it
in
terms
of
definition
or
core
properties,
then
these
are
the
things
which
I,
which
comes
out
or
I,
stopped.
Five
properties
of
any
blockchain
network.
I
would
expect
my
network
to
replicate
the
data
among
all
the
participants,
and
it
is
of
course
distributed.
It
can
be
anywhere
in
the
cloud
and
I
expect
it
to
be
decentralized
when
I
say
decentralized.
What
I
seek
for
it
is.
No
one
party
has
authority
over
the
Ledger's.
A
That
is
because
of
core
property,
which
blocks
and
provides
me
I
am
very
sure
that,
because
of
these
properties,
which
blockchain
provides,
we
will
in
sometimes
see
how
we
made
abilities
is
achieved
in
Sawtooth,
but
because
of
these
properties
is
how
I
can
be
sure
that
whatever
log
he
at
the
party
is
showing
to
me
is
valid
and
I
can
again
verify
it
against
my
own
transaction
record
and,
of
course,
blocked
blocked.
Some
of
the
blockchain
provides
us
secure
access
to
perform
transactions
also
to
read
the
data,
and
it
also
goes
with
sawtooth.
A
A
Let's
quickly
see
what
sort
it
is
so
sawtooth
is
a
blocks
in
architecture.
It
provides
all
the
features
which
we
discussed
so
far
and
it's
highly
modular.
You
will,
if
you
see
one
of
the
darker
compost
files
in
enough.
The
repository
which
hyper
laser
sawtooth,
as
you
will
be
amazed
at
how
modular
it
is,
we
can
literally
plug
and
play
with
multiple
components.
We
cannot,
like
I,
can
say,
remove
in
a
docker
compose
file,
a
poet
consensus
engine
and
then
plug
it
plug
in
pbft
and
then
configure
my
network
to
use
pbft.
A
That's
all
I
would
do
to
start
using
PFD
instead
of
poet.
So
similarly
with
others,
there
is
one
for
developers
as
well.
We
generally
use
dev
mode
consensus
engine.
That's
that's
usually
for
single
node
on
a
development
mission.
I
used
that
to
build
and
to
perform
any
sort
of
quick
tests
which
I
want
to
do
so
provides
permission,
permission,
less
booth
kind
of
architecture.
Then
one
of
the
core
feature
which
it
also
has
this
parallel
transaction
execution.
A
So
if
you
can
design
your
application
in
a
way
that
you
can
utilize
this
feature,
then
you
your
throughput
will
be
really
high.
This
is
one
of
the
feature
which
is
not
available
in
other
blockchain
architectures
and
a
hyper
ledger,
and
it
truly
depends
on
how
we
design
our
architecture
or
our
application
solution,
and
you
know
also.
A
When
I
say
permissionless
sawtooth,
if
you
setup
multiple
instances
of
validator
notes
by
the
way
when
I
say
validator
node,
it's
a
node
which
which
does
all
the
backend
infrastructure
thing,
which
I
earlier
mentioned,
imagine
it
to
be
like
a
micro
service
running
somewhere
in
the
cloud.
So
when
I
have
my
micro
service
and
then
I
say
hey
here
is
another.
This
is
another
IP
of
another
micro
service
instance.
My
validator
instance-
and
here
is
the
or
rather
I
can
say
here-
is
the
list
of
validators
there.
A
I
if
I
specify
the
IP
address
by
default,
they
connect
to
each
other
and
in
sawtooth.
If
I
have
to
permission
them,
then
there
is
option
4
for
me
to
configure
it.
For
example,
as
I
said
by
default,
it
allows
any
invalidate
or
note
to
join
the
network
if
I
want
to
restrict
access
to
specific
notes.
Maybe
Kent
is
running
a
node.
They
company,
XYZ
and
I
am
running
a
node.
A
A
These
are
the
public
keys
of
validator
nodes,
for
which
you
are
supposed
to
accept
connection
requests
from.
If
you
request,
if
you
receive
a
request
from
any
other
validator
node,
then
do
not
accept
the
connection
request,
just
reject
them.
That's
how
the
permissioning
is
happening
like
the
validator
level,
and
then
there
is
second
set
of
permissioning,
which
I
will
like
again
covered
later
in
detail.
That
is
on
the
transactions
itself.
So
the
first
level
of
permissioning
is
at
the
validator
notes.
The
second
level
of
permissioning
is
at
the
transaction
level.
A
I
can
say
these
are
the
exact
prime
public
keys
which
can
perform
transactions
in
this
network
or
then
I
can
also
specify,
so
in
sort
of
there
is
also
concept
of
transaction
in
batches.
It's
better
if
I
cover
that
and
then
explain
this
permissioning
so
that
it's
easily
analyzed.
So
there
is
two
set
of
permissioning
two
categories
of
permissioning
liked.
I
love
it.
Let's,
let's
keep
it
to
that
transaction
level,
permissioning
and
then
validator
level,
permissioning,
so
yeah.
It
quickly
covered
the
respects.
A
It
has
sawtooth
saber,
which
I
mean
there
is
something
called
a
sawtooth
saber
the
advantages
of
a
charge.
You
deploy
your
smart
contract
from
any
of
your
node
and
the
same
copy
is
run.
It's
like
a
stored
procedure
in
your
database
applications.
We
will
cover
each
of
this.
This
light
deck
has
comprehensive
details
on
each
of
these
topics
and
yeah
you
will
be
able
to
you
have
ability
to
develop
your
application
in
multiple
languages.
There
are
stickers
available.
A
A
Few
things
which
I
have
highlighted
here
are
when
we
say
validated,
it
is
actually
even
connecting
to
three
different
databases
at
its
end.
Right
now,
the
support
is
for
LM
DB,
but
there
is
consideration
for
putting
in
multiple
other
databases
as
well,
and
there
is
consideration
for
putting
in
a
remote
instance
of
a
database.
A
This
all
will
be
features
of
two
dot
x:
release,
I'm,
not
sure
which
exact
release
it
will
come
up
in
so
yeah
quickly,
databases
we
have
something
like
global
state
and
then
there
is
this
global
state
is
a
unique
concept
which
I
will
explain
in
next
slide.
Then
there
is
change
told.
Whenever
we
talk
about
blockchain,
it's
it's
a
chain
of
different
blocks,
joining
together
to
form
the
immutability
concept
at
a
high
level.
That's
what
is
stored
in
change
tour
and
then
we
also
have
transaction
receipts.
A
This
is
a
record
of
all
your
transactions,
so
there
is
distinction
between
global
State
and
transaction
receipt,
which
we
will
understand
once
we
understand
what
global
State
is
and
then
validator
itself
won't
be
sufficient.
It
performs
all
the
core
operations,
it's
like
a
brain
of
like
a
central
unit,
central
it
processes,
many
things
and
it
connects
to
other
components
as
well.
Do
you
also
see
my
cursor
if
I
use
it.
A
The
validator
itself
is
has
multiple
other
components
such
as
like
from
the
time
you
receive.
There
is
something
called
as
completed
or
the
journal
module,
which
I
say
like,
for
example,
if
transaction
processor
or
another
validator
is
connecting
to
this
validator,
then
the
flow
of
that
transaction
will
be
through
multiple
layers
within
validator
mode,
so
that
I
have
not
shown
all
those
components
within
the
validator
here.
I
have
rather
only
shown
some
of
the
other
aspects
which
which
so
that
I
can
so.
A
This
is
for
the
beginners,
then
what
I
mean
is
when
the
messages
come
in
it
enters
something
called
as
journal
layer
of
validator
and
from
within
there.
There
is
something
said
there
is
a
module
which
waits
for
that
message
to
completely
arrive,
and
there
are
modules
within
validator
which
checks
for
block
existence
with
respect
to
block
ID
locks
existence
with
respect
to
transaction
ID,
and
then
there
are
modules
which
I
mean
there
are
multiple
cues
within
validator
one
to
store
the
incoming
messages
from
each
of
these
interfaces,
which
you
see
through
CMD
connectors.
A
A
It
has
core
components
is
like
a
brain
for
network,
so
it
has
processing
ability,
task
queues.
It
has
these
data
base
connection
abilities,
it
provides
different
components,
ways
to
operate
so
each
validator
node
is
equal
in
sawtooth,
very
high
level,
but
how
do
they
get
into
consensus
with
each
other?
That's
where
the
consensus
engine
comes
into
picture,
depending
on
which
consensus
algorithm
we
choose.
If
we
choose
raft
consensus
algorithm,
then
among
the
chosen
validators,
one
of
them
will
acting
as
leader
and
rest
of
them
will
be
kind
of
followers.
Leader
will
produce
the
block.
A
Rest
of
them
will
accept
the
blocks.
If
we
follow
ppfd
consensus
algorithm,
then
periodically,
each
validator
will
take
a
chance
to
produce
the
new
block
and
whichever
node
produces
the
block
that
becomes
leader
for
that
duration
of
time.
Every
other
validator
in
the
network
will
be
in
the
mode
of
accepting
the
blocks
and
if
we
use
point
as
a
consensus
algorithm,
then
it's
like
at
each
iteration,
there
can
be
any
any
any
node
can
be
a
leader,
so
it's
purely
random
defined
again
within
poet.
A
A
No
node
can
cheat
with
each
other
provided
by
the
intent
for
tolerance,
so
yeah.
We
will
comprehend
this
answers
in
multiple
slides
sure,
thanks
for
asking
that
question
so
quickly,
it's
talking
the
validator
itself.
When
we
write
our
application,
the
validator
itself
may
not
be
sufficient
for
us
right.
We
may
need
we
will.
What
we
do
is
in
any
blockchain
architecture.
A
We
have
core
network,
and
on
top
of
that
we
attach
interfacing
layer
which
the
components
which
directly
talks
to
blockchain
network,
and
on
top
of
that
we
write
our
own
application
layers
or
application
micro
services.
It's
the
same
concept
in
sawtooth
as
well.
We
can
attach
something
called
as
event
listener
which
listens
to
all
the
changes
which
is
happening
within
validator
note.
A
It
specifically
is
like
targeting
transaction
receipt
in
any
event
which
is
captured
in
blockchain
is
in
the
transaction
receipt,
so
validated
node
takes
care
of
red
to
or
how
to
put
that
data
to
event
listener,
which
is
an
external
component
and
it
gets
connected.
It
can
be
connected
to
validated
through
zmq
connector,
and
then
there
is
REST
API.
This
is
available
out
of
the
box.
A
It's
the
Sawtooth
rest
api,
which
you
see
generally
in
any
application
in
instead
of
using
this
version
of
rest
api,
you
can
use
your
own
custom,
REST
API,
to
provide
read
protection.
So
this
is
one
of
the
question
which
I
keep
hearing
from
many
people
who
you
saw
tooth.
They
asked
me:
hey
anybody.
Can
anybody
read
they
time
sort
of
how
it
provides
anybody
to
read
data
through
REST
API
and
how
is
it
secure
so
answer
for
this?
A
Is
anybody
can
read
data
through
default
available
REST
API,
but
it
also
makes
sure's
that
access
to
the
REST
API
itself
is
protected
if
I
as
an
organization.
I'm
part
of
a
network
I
already
have
access
to
all
the
data
which
I
am
supposed
to
have.
Then
my
REST
API
has
access
to
data
which
my
validator
has
and
it's
my
responsibility
to
protect
the
REST
API.
If
I
don't
want
to
reuse,
this
version
of
API
I
can
replace
it
with
my
own
version
of
API,
with
probably
SSO
authentication
or
sam'l
or
another.
A
Some
some
other
SSO
solutions
to
protect
or
in
I
can
also
integrate
with
ad
solutions
to
have
that
access
management
done
for
reading,
and
then
there
are
something
called
as
transaction
processors.
This
is
the
core
engine
or
sorry,
I
already
told
core
engine
is
a
validator
node.
So
this
is
the
actual
processor.
Is
it
is
a
smart
contract,
the
core
engine
of
the
business
part
of
your
application?
That's
where
that's
what
we
have
in
transaction
processors,
and
then
there
are
consensus
engines,
it's
a
plug-and-play
architecture.
We
have
multiple
consensus
engine.
A
We
can
attach
any
one
to
the
validator
node
at
a
high
level
if
I
have
a
validator
network
or
a
sawtooth
network
for
each
application,
I
would
have
one
such
network.
In
my
deployment,
it's
like
one
use
case
we'll
have
one
validator
node
network.
If
I
have
another
use
case
which
requires,
if
I
have
like
two
use
cases,
one
of
which
requires
three
part
is
one
of
it
requires
just
two
parties,
then
I
would
ideally
have
two
such
networks.
A
I
have
a
slide
on
that
to
make
it
more
clear,
I'll,
probably
move
ahead
because
more
of
much
of
these
I'm
again
going
to
discuss
in
detail
in
later
slides
and
since
I'm
on
the
full
screen
and
in
Kent
or
somebody.
Let
me
know
if,
when
I
have
to
stop,
keep
me
check
on
the
time.
Please
because
I,
don't
say
time
on
the
screen:
I'm
on
full
screen
mode
I'm
going
to
next
going
to
next
slide,
we'll
quickly
talk
about
global,
you.
A
Yeah,
so
thanks,
so
the
next
component,
which
we
are
going
to
talk
about,
is
the
global
State.
Here
I
told
you
that
we
will
discuss
in
detail
and
let's,
let's
take
that
up
global
state
and
addressing
in
general
what
why
it
mattered
since
out
to
top
locations
at
the
at
the
least.
If
you
can
see
the
left
bottom
section
there
is.
This
is
how
we
visualize
any
storage
in
sawtooth.
A
There
is
some
address
where
we
store
some
value
and
when
I,
if
I,
have
to
write
my
application
on
blockchain
I
store
something
in
blockchain
and
somebody
else
performs
a
transactions
which
reads
some
data
in
the
blockchain,
and
it
asserts
whether
that
user
has
ability
to
do
that
transaction
or
not.
Do
the
transaction
or
make
some
decisions
transition
this
value
to
another
value.
These
are
some
possible.
This
would
be
what
we
generally
writing
applications
for
in
so
on
in
any
blockchain.
So
at
least
there
is
some
address.
A
We
store
some
value
in
it,
as
I
said
it
can
be
transformed,
it
can
be
just
read,
it
can
be
used
for
making
decisions
and
for
anything
and
the
way
we
organized
this
in
sawtooth
is
we
use
some
concept
called
as
a
radix
Merkle
tree
or
it's
a
property.
It's
like
two
different
concepts
put
together
in
single
tree.
A
What
that
means
is
every
you
can
imagine
it
as
a
tree
with
a
root
node
branching
out
to
multiple
child
nodes
and
at
each
level
there
could
be
multiple
children
when
I
say
that
what
I
mean
by
that
is,
let's
say
we
have
an
address
and
a
value.
This
always
will
be
a
child
node
in
my
root
or
sorry
in
my
tree
in
my,
it
will
definitely
always
be
a
child.
Whenever
we
say
there
is
an
address
and
I
store
a
value
in
sawtooth.
First
thing
we
should
come
to
our
mind.
A
Is
it's
a
some
leaf
node
in
my
tree,
and
these
leaf
nodes
are
connected
to
a
parent,
and
the
way
they
are
connected
is
what
makes
it
sort
of
interesting
and
as
well
as
what
makes
it
immutable
in
general.
How
we
store
is
at
a
each
each.
Let's
say
each
node
will
have
two
bytes
in
two
bytes
of
information
to
represent
its
uniqueness
in
the
path
what
I
mean
by
that
is.
A
If
we
travels
from
root
node
to
any
of
the
leaf
node
at
each
level,
a
node
will
uniquely
have
two
byte
information,
which
differs,
which
makes
it
different
from
its
parent
and
again
same
way.
Its
child
will
have
tune
bytes
in
addition
to
its
parent
node,
and
we
generally
put
we
call
it
as
address
of
that
node
in
a
tree.
For
example,
if
we
just
had
two
leaf
nodes
in
a
tree,
then
there
is
a
parent,
a
parent
will
have
address
like
0
0,
then
child
will
have
2
bytes
in
addition
to
its
parent.
A
So
if
parent
has
0
0
has
is
address,
child
will
have
0
0,
followed
by
0
1.
Another
child
will
can
have
address
0
0,
followed
by
zero.
So
this
is
a
unique
I
mean.
Can
somebody
quickly
tell
me
how
many
children
we
will
have
like
this?
If,
if
two
bytes
are
used
for
addressing
each
child,
not
much
long,
maybe
I
think.
A
A
Yeah,
so
each
you
can
imagine
how
many
addresses
we
have
in
sort
of
in
general
like
each
it
with
just
one
parent.
It
has
2
bytes
of
information
where
each
address
can
be
so.
Each
child
can
be
addressed
with
a
unique
address
of
ranging
from
0
0
to
F
F.
So
FF
is
the
maximum
number
of
nodes
which
each
parent
can
have,
and
similarly
we
will
have
multiple
such
parent
nodes
in
a
tree
and
in
total
what
we
have
is.
A
A
The
address
is
this:
has
35
bytes
in
it
any
leaf
node
address?
Is
this
clear,
I
hope
I
was
able
to
explain
it?
Yeah,
that's
good
awesome.
So
one
more
thing
which
I
will
quickly
point
out
is
we
generally
talk?
You
know
we
generally
use
the
term
called
namespace
in
sawtooth,
and
we
it's
not
so
complicated
concept.
So
once
we
know
that
this
is
how
the
addressing
works
in
sawtooth,
we
know
leaf.
A
B
A
If,
hypothetically,
if
this
is
my
node,
which
I
want
to
represent
in
my
Merkle
tree,
then
it
has
address
0,
0,
0,
0,
and
then
its
child
will
have
0
0
2
FF
number
of
children.
This
particular
node
will
have
so
many
times,
then
the
names
if
I
want,
if
I,
want
some
weight
for
grant
access
to
my
smart
contract
access
to
all
these
addresses,
then
I
just
specify
the
parent
nodes
address.
A
If
I
specify
this
address
in
my
smart
contract,
then
my
smart
contract
literally
has
access
to
each
of
its
child,
or
even
if
this
node
has
for
that
child,
then
access
will
be
granted
to
each
of
the
leaf
node
coming
under
this
particular
node,
and
the
address
of
this
node
is
nothing
but
a
namespace.
It's
the
name
given
which
it's
a
name,
which
we
generally
use.
Many.
We
write
applications
in
sawtooth
now
that
we
understood
what
namespaces
and
how
addressing
works
in
Merkle
tree
now
there
is
another
interesting
concept
here
which
I
did
not
discuss.
A
B
A
Correct
that's
correct,
so
the
reason
is
if
multiple
nodes
are
running
the
same
contract,
they
all
run
the
same
code.
They
read
the
same
data
they've,
so
in
general,
at
very
high
level.
Smart
contact
accept
some
input
processes.
The
messages
coming
in
that
input
output,
some
information
and
how
do
they
come
in?
How
do
they
come
to
consensus
is
by
making
sure
that
each
of
the
output
in
different
nodes
is
same.
The
output
values
are
compared.
A
One
way
to
compare
output
values
is
literally
by
throwing
all
the
output
packaging
it
in
some
form,
maybe
find
a
hash
of
it
or
package
it.
As
a
jason,
then
compare
this
value
with
another
nodes
output,
literally
like
a
string
comparison
or
a
byte
comparison.
That
is
one
way
of
comparison,
but
this
particular
property
of
merkel.
Radix
tree
allows
for
different
way
of
comparing
the
elements.
A
This
at
the
end,
I
just
need
to
compare
my
state
root,
hash
against
another
node
state
root
hash
and
if
they
differ,
there
is
definitely
some
mismatch
of
any
of
these
values
at
the
leaf
node
level.
So
that
sounds
out
uses
the
concept
to
to
make
sure
that
nodes
are
in
consensus.
State
route
hash
is
another
term
which
we
generally
use
in
in
sort
of
development.
We
should
be
familiar
with
it,
so
it's
nothing
but
a
root
nodes,
hash
value
which
has
in
in
simple
terms
it
has
hash
of
each
of
the
values
underneath
it.
A
A
When
I,
when
I
explain,
block
I
will
I
will
see,
I
mean
I
will
show
how
this
is
related
to
block
and
how
do
we
guarantee
mutability,
but
at
high
level
for
nodes
to
get
into
consensus,
as
I
told
the
values
of
the
leaf,
node
is
hashed
together
and
put
as
value
of
its
parent
node.
So
if
this
is
my
parent,
my
value
would
be
all
the
I
mean
hash
of
all
the
values
which
my
children
node
stored
in
it.
Then
my
parents
will
be
at
this
level.
Whatever
are
the
nodes?
B
A
A
Yep,
so
we
understood
what
State
Route
ash
is,
so
it's
a
hash
at
very
high
level
at
the
topmost
level,
so
that
basically
identifies
if
there
is
any
mismatch
between
two
nodes
in
the
values
which
we
store
in
the
leaf
nodes.
Then
my
state
root
hashes
will
be
different
in
in
both
those
nodes
because
they
are
connected
through
a
hashing
property
and
the
way
to
visualize
block
block
chain
or
blocks.
Cherien
sawtooth
is
like
this
any.
It
starts
with
a
Genesis
block.
A
The
reason
I
marked
it
in
some
other
color
is
because
it
has
similar
property,
like
any
other
block,
just
that
this
block
does
not
involve
in
consensus
process.
The
reason
because
we
use
this
Genesis
block
to
configure
things
in
the
network.
We
have
multiple
unchained
permissioning,
where
I,
don't
I,
don't
store
explicitly
along
with
validator,
saying
that
these
are
the
public
keys
who
should
have
access
to
the
network
rather
I
should
or
that
information
inside
the
blockchain
itself.
So
that's
the
reason
we
eat.
A
It
has
the
configuration
information
and
it
since
it's
a
very
first
block.
We
don't
need
agreement
between
participants.
We
don't
actually
put
our
business
logic
related
information,
it's
more
of
a
network
bootstrapping
information
which
is
stored
in
it,
so
it
also
has
its
own
State,
Route
hash
and
when
I
process
my
transactions
from
my
application,
my
application
would
write
some
information
in
some
leaf
node
in
my
Merkel
tree,
and
thus
my
state
root
I
should
will
change
so
in
a
block
which
I
produce
at
that
level.
A
I
store
the
state
root
hash
for
which
every
participant
agreed
upon
at
that
at
that
level
at
that
block
level.
So
similarly,
we
keep
adding
a
new
chain
head
to
existing
chain
and
one
more
property
which
is
not
available
in
other.
So
this
is.
This
is
one
important
property
which
is
not
available
in
other
projects
in
hyper
ledger
that
we
don't
store
the
State
Route
hashes
at
at
this.
No,
there
are
actually
two
things
to
benefits
of
State
Route
hash.
One
benefit
is,
of
course,
to
get
consensus
from
different
validators.
A
The
second
benefit
is
even
let's
say
somehow.
Somebody
hacked
this
particular
block
or
they've
tried
to
modify.
They've
tried
to
spoof
with
high
computation
power,
came
to
came
to
same
state
route,
harsh
with
different
version
of
values
at
the
leaf.
They
cannot
produce
same
output
at
next
level
because
their
values,
if
they
have
modified
the
values
at
leaf.
So
that's
that's
how
they
are
connected
in
a
in
a
way:
it's
not
only
connected
to
other
validators.
It's
also
connected
to
next
block
in
the
same
block
chain
as
well
and
along
with
State
Route
hash.
A
B
A
These
previous
block,
hash
touring
previous
block
hash,
is
same
concept
in
most
of
the
blocks
and
networks,
but
in
networks
like
sore
tooth
or
aetherium
or
in
any
any
blocks
in
architecture,
we
chooses
Merkle
tree
concept
that
that
has
this
property
of
state
rotation
in
in
other
blockchain
architectures.
What
we
generally
see
ec
is
they
compute
hash
of
transactions
happening
in
that
particular
block?
A
They
don't
so
the
hash
does
not
in
a
way
connect
to
the
previous
blocks
or,
if
I
mean
in
a
way
they
connect,
because
the
outputs
will
differ,
but
they
do
not
provide
this
guarantee
through,
like
as
I
explained,
if
even
if
I
had
this
particular
block,
the
value
I
don't
need
to.
I
cannot
have
the
same
state
root
high
for
the
next
because
of
the
hashing
property.
B
B
A
A
Do
next
slide,
then?
Okay,
so
this
is
another
core
concept
of
sawtooth
where
it
provides
blog
and
like
consensus
mechanism.
What
we
do
is,
as
I
said,
in
a
roof
block
or
my
first
Genesis
block
I,
put
some
configuration
information,
and
that
also
has,
along
with
the
configuration
information
we
generally
put
in
consensus,
related
information.
A
I
can
say
when
now,
when
I
bootstrap,
my
network
head
I'm,
going
to
use
for
it
in
my
network
or
I'm,
going
to
use
pbft
in
my
network
or
I'm,
going
to
use
raft
in
my
network
or
I
can
also
use
demo'd
consensus
engine
in
my
network
or
any
other
new
consensus
which
are
being
developed
and
I
know.
There
are
some
more
some
more
consensus
being
developed
for
sawtooth
I
skips,
seeing
them
from
independent
parties
in
rocket
jet
channels,
so
yeah.
It
provides
that
SDK
in
Python
rust.
A
Currently
it
supports
two
languages
for
writing
consensus
engines.
If
you
have
your
own
consensus,
you
can
write
that
there
are
a
sticky
methods
which
you
need
to
implement
in
your
consensus
engine.
What
we
do
is
we
say,
register
mine
consensus
engine
and
validator
node
will
keep
track
of
whichever
is
activated
at
that
instance.
This
activate
is
in
fact
introduced
in
1.2
version
of
sawtooth.
A
If
you
go
to
our
later
versions,
you
did
not
have
some
activate
message,
so
if
you
are
trying
for
a
CH
si
examination,
probably
this
it
was
this-
it
was
not
like
this.
It
it
improved
over
a
period
to
come
to
1.2
version,
maybe
in
105
version
which
the
CHCH
si
examination
is
based
on
it
did.
It
does
not
have
this
register
and
activate
concept
in
it.
It
has
different
mechanism,
it's
very
similar.
A
So
yeah
we
connect
distinguish
different
consensus
engine
through
different
conventions.
One
is
CFD
versus
PFD
or
we
can
say
for
king
versus
fast
finality.
This
is
one
important
thing
which
I
would
which
I'll
like
quickly
talk
about,
because
CFD
and
BFT
we
generally
here
in
blockchain
word,
the
forking
and
frosh
finality
is
comes
in,
comes
in
to
the
fact
where
some
of
the
algorithms
like
in
public
blockchain
networks,
there
are
for
king
processes
and
poet,
is
also
one
of
such
forking
algorithm.
What
I
mean
by
Fork
is
each
validate.
A
The
node
has
equal
power
in
producing
the
block
and
they
agree
to
somebody
else
block
based
on
predetermined
conditions.
In
case
of
poet,
it
is
the
timer
value
which
they
see
in
other
validators
to
come
to
conclusion,
which
block
to
use
and
some
other
consensus.
Maybe
it
is
a
proof-of-work
in
in
case
of
poet,
as
it
is
proof
of
elapsed
time,
a
timer
value
which
we
compare
in
some
other.
It
could
be
proof
of
work
it
some
other.
It
could
be
proof
of
stake
or
anything
else.
So
these
algorithms
are
forking.
A
Why
I
say
forking
is
because
a
block
which
is
already
added
to
blockchain
may
get
replaced
by
another
block
based
on
how
we
write
the
consensus
algorithm
why
it
gets
replaced?
Maybe
because
my
network
is
so
large,
it
has
hundreds
of
validators
right
now.
I
saw
only
50
of
my
validated
nodes.
Rest
of
the
50
nodes
are
still
maybe
mine.
A
I
did
not
receive
a
message
from
one
of
them
which
actually
had
lowest
waiting,
timer
value
and
I
agreed
upon
somebody
else
block,
but
later
I
realized
that
oh
no,
this
is
not
the
right
block
everybody
else's
on
some
other
block.
Let
me
go
and
attach
to
that
block.
So
that's
what
happens
in
point.
So
it's
a
fork
in
Calgary
thumb
in
first
finality,
once
a
block
is
committed,
it
can
never
be
replaced.
A
One
carrot
for
forking
algorithms
when
designing
applications
is,
if
you
are
relying
on
event.
Listeners
event
is
emitted
whenever
a
transaction
is
committed
or
whenever
a
block
is
produced,
and
if
you
are
listening
to
your
bands,
when
a
block
is
committed
in
poet
or
in
any
forking
algorithms,
you
may
experience
that
block
Mary
get
replaced
by
another
block.
So
you
should
your
event.
Listener
also
should
be
intelligent
enough
to
handle
that
any
other
questions
on
this
slide.
B
A
None
of
the
consensus
algorithms
we
weave
in
in
done
of
the
consensus
algorithm,
we
will
lose
the
data,
so
all
of
them
guarantee
when
I
say
one
block
gets
replaced
with
another
block.
Its
just
committed
transactions
are
different,
but
the
block
which
is
removed
all
those
transactions
will
be
part
of
some
other
block.
Eventually,
it's
it's
not
like
it's
it's
getting
lost,
but
rather
it
will
be
put
into
some
other
block
and.
A
Right,
we
don't
yeah
block,
store
gains,
replaced
in
fast
finality,
based
once
any,
because
raft
is
algorithm
by
nature
of
the
algorithm
itself.
There
is
one
master
and
rest
of
them
are
followers.
That's
the
rough
concept
in
at
very
high
level.
Whoever
is
the
master
node
that
node
produces
the
block
every
other
followers.
They
commit
the
block
they
validate
the
block,
and
then
they
commit
the
block
pbft.
B
B
A
A
So
this
is
the
database
which
it
uses
at
its
level
and
there
are
plans
to
integrate
external
database
and
that
will
be
coming
in
to
dot
or
like
further
releases.
After
to
not
oh,
and
if
you
are,
writing
applications
its
we.
What
we
generally
do
is,
we
add
event
listener
and
impact
event.
Listener
will
write
data
to
another
database
from
where
we
perform
read
operations
generally
for
applications
or
analytics
purpose.
So
it
can
be
anything
event.
Listener
can
eventually
write
to
no
sequel
databases
or
any
of
them.
A
A
Yeah
I
had
a
slide
on
different
messages
exchanged
so
at
a
high
level
consensus
engine
gets
involved
in
two
verticals
are
two
different
processes
and
in
fact,
in
sawtooth
consensus
engine
is
the
is
like
a
driver
for
creating
the
blocks.
Validator
on
its
own
will
not
do
anything
without
instructions
from
the
consensus
engine.
So
what
we
do
is
at
the
block
creation
phase,
depending
on
which
consensus
algorithm
we
choose.
If
it
is
raft,
it
is
the
leader
node
or
a
master
node,
which
runs
this
process
if
it
is
PBF,
be
attacked
level.
A
Whoever
is
the
leader
that
processes
this
step.
If
it
is
poet,
then
every
node
processes,
this
type
and
the
step
is
blocked.
Creation,
consensus
will
say
whenever
it
gets
activated
from
a
validator,
it
says
depend
on
census.
It
is
so
it
says,
initialize
a
block
based
on
it
could
be
time
or
value.
It
could
be
some
some
things
which
we
set
different
consensus:
algorithms,
our
different
ways
to
know
when
to
produce
a
block
and
then
validator
will
start
initializing
constructing
the
block.
It's
like
you
can
imagine
it
to
be.
A
Like
a
box
empty
box,
it
starts
putting
transactions
into
that
box
and,
when
consensus
engine
says,
stop
putting
transactions
into
block,
get
me
that
created
block,
so
it
provides
that
created
block
back
to
consensus
engine.
Then
consensus
adds
consensus,
related
message
to
the
block,
so
consensus
block
any
block
at
high
level.
You
can
imagine,
has
transaction
information
and
it
also
has
consensus
information
in
it
once
we
add
that
we
say.
Ok
here
is
your
block.
Finalize
it
and
start
broadcasting
it
to
other
validators.
A
The
flow
is
transactions
are
arriving
to
validator
node,
randomly
from
multiple
ends.
Whenever
validator
receives
the
transaction,
it
broadcasts
to
each
other
validators.
So
everybody
has
transaction
data
with
them
in
the
network.
In
that
particular
network,
whoever
is
participating
in
the
network
and
when
consensus
tells
to
one
of
the
validator
in
the
node
or
whichever,
if
validated
node
it
tells.
A
Generally,
we
attach
one
consensus:
I
mean
consensus
along
with
a
validator,
so
it's
like
a
driver
for
validator,
it
says
create
my
block
block
is
validator,
starts,
creating
the
block
based
on
time
or
value
or
something
we
say,
stop
creating
the
block.
Summarize
it
to
me
once
we
get
the
block,
we
add
consensus,
information
in
the
block
and
we
say
broadcast
it.
So
first
broadcast
was
for
transactions.
Second
broadcast
is
for
block
and
once
blocks
are
broadcasted
to
each
of
the
validators
block.
A
Validation
face
will
come
into
picture
again
that
block
validation,
phase
validator
will
perform
its
own
validations
when
I
say
its
own
validation.
It
is
the
transaction
execution,
validation
and
then
it
says,
hey
I
got
a
new
block.
Would
you
like
to
validate
that
consensus?
Engine
says
yes,
I
would
like
to
validate
your
block.
It
extracts
the
consensus
engines,
parameters
from
the
block
validates
that
other
block
which
is
arriving.
He
is
according
to
the
consensus
algorithm,
which
we
agreed
upon
and
says:
hey
I
checked
your
block,
and
what
do
you
say?
Is
it
valid
or
invalid?
A
So
shall
I
come
go
ahead
with
it
or
not?
That
is
what
check
block
says,
or
it
can
also
say
consensus
process
fails.
So
you
do
not
have
to
proceed
with
this.
Ignore
this
block,
which
you
which
came
to
you,
then
validator
will
say.
Ok,
you
asked
me
to
check
the
block.
Aye
I
checked
it
thoroughly
when
you
were
checking
for
validation
of
your
consensus
parameters.
So
here
is
the
validity
information.
It's
a
valid
block,
my
transaction.
All
the
transactions
are
valid
or
it's
invalid.
A
Somebody
is
trying
to
play
with
the
network,
something
like
that
and
consensus
will
finally
say:
ok,
I
validated
you
also
validated
everything
is
done.
Let
us
commit
this
block
or
it
can
also
say,
since
you
are
saying
it
is
invalid,
let's
just
fail
it.
We
will
not
proceed
with
this
block,
so
these
are
the
different
communication
exchange
happening
between
consensus,
engine
and
validator
service.
So
a
broad
category,
there
are
two
things
happening
on
two
major
functionalities
of
consensus
in
patience
here.
B
B
A
B
A
20
node
network,
it
has
been
tested
up
to
80
TPS
and
it
has.
It
has
been
successful
in
that
in
a
20,
node
Network,
and,
having
said
this,
there
are
no
benchmarking
done
and
results
published
on
Sawtooth
on
different
crops,
so
it
is
I
truly,
don't
know
the
numbers.
The
numbers
which
I
told
you
again
are
just
a
part
of
release:
criteria
for
any
new
release.
Community
does
some
activity,
so
this
results
are
from
that
result,
a
community
activity
for
just
before
the
release
we
check
for
stability.
A
B
A
A
B
B
B
A
Thanks,
okay,
let
me
see
if
I
can
cover
it
as
much
as
possible
in
today's
session,
so
transaction
processors
is,
as
I
said,
it
runs
the
smart
contract
logic
in
our
applications,
and
this
is
also
another
component
which
connects
to
my
validator,
something
like
consensus
engine.
We
have
multiple
transaction
processors:
okay,
I,
don't
yeah.
This
is
showing
that
diagram.
I
have
my
validator
instance
and
my
smart
contract
will
connect
to
my
validator
and
see.
We
will
see
each
of
this
message
later,
but
at
eye
level,
it's
connecting
through
some
port.
A
Let's
say
validated
is
exposing
a
port
where
I'm
connecting
to
it
and
I
can
write
my
transaction
processor
in
multiple
languages.
There
are
SDKs
available
in
I
guess
some
of
the
languages
are
missed
here,
but
yeah
you,
you
name
a
language
and
it's
available.
In
fact,
cha
was
listed,
whiten,
okay,
yeah
there
are
multiple
languages,
I,
don't
remember
all
of
them.
A
A
Transaction
processor
is
also
very
similar.
It
is
not
it
so
it's
providing
you
an
ability
to
write
your
smart
contract
logic
so
that
you
can
communicate
with
the
validator.
You
can
say
intent
of
your
transaction
to
validated
what
happens
is
whenever,
at
the
time
of
block
creation
or
at
the
time
of
block
validation,
when
transactions
are
getting
added
to
block
or
when
transactions
are
getting
validated
from
the
block,
validator
will
send
a
message
to
your
transaction
processor.
It
will
say:
hey,
can
you
apply
this
transaction,
then
your
transaction
processor
will
run
the
apply
method.
A
So
you
write
your
logic.
You
consume
the
transaction,
you
process
it.
Do
some
Gators
write
based
on
what
actions
you
need
to
take
and
then
finally
say:
hey
validator.
This
is
a
valid
transaction,
a
validator.
This
is
a
invalid
transaction
and
one
cool
thing
is
you
don't
need
tools
again
say
this
is
what
you
need
to
set
as
output,
or
this
is
what
you
need
to
set
as
input,
because
validator
gives
you,
along
with
transactions.
It
also
passes
something
called
as
context
because
parallely
there
could
be
multiple
things.
A
Opening
in
a
validator
one
could
be
like
creating
the
block
or
validating
the
block
whenever
I
validate
a
block,
I
need
to
validate
it
with
respect
to
previous
block
or
current.
Whatever
is
the
immediate
previous
block,
so
validator
sends
a
context
information
and
we
get
our
set
value
based
on
that
context,
it's
it's
very
simple.
It's
like
context,
dot,
get
context
not
set
as
simple
as
that
you
receive
something
from
validator.
A
B
B
A
So
I
will
ask
you:
question
I
will
ask
question
back
to
the
community
here.
Can
you
quickly
tell
me?
Okay,
let's,
let's
cover
the
next
slide
and
then
I
will
come
to
your
answer
and
answering
to
your
question.
Okay,
this
is
a
slide
which
I
had
like
stuck
the
next
upcoming.
So
when
I
say
transaction
processor,
the
name
processor
is
attached
here
because
it's
a
processor,
it
processes
something
for
us.
It
is
processing
messages
from
the
validator
node.
So
this
transaction
processor
is
processing
messages
from
validator
node.
It
receives
this
method
apply
method.
A
Each
transaction
processor
can
have
multiple
transaction
handlers
between
itself
and
we
can
attach
multiple
transaction
processors
to
validator,
so
each
transaction
processor
can
have
multiple
handlers
and
we
can
attach
not
just
one
processor
to
validator.
In
fact,
multiple
processors
to
so
I
can
have
four
different
logics
written
in
same
processor,
attached
to
validator
or
I
can
have
four
different
processors
itself
written
attached
to
validator.
This
comes
with
the
main
program
and
within
it,
I
can
have
four
threads
in
other
terms
or
I
can
have
one
thread
running
in
it.
A
Each
processor,
as
simple
as
that
to
visualize
it
I'm
each
transaction
handler
it's
this
core
logic
or
score
business
logic,
which
we
are
going
to
write
and
the
way
we
identify.
Each
thing
is
by
having
a
name,
a
version
for
that
logic
and
a
namespace
for
that
logic,
and
then
the
core
business
logic
apply
method
it
which
accepts
context
and
transaction
as
inputs.
A
A
Right
so
when
I
have
multiple
versions
of
same
business
logic,
let's
say:
I'm
writing
a
business
logic
for
for
like
just
storing
something
or
can
somebody
give
me
an
example.
I
ran
out
of
examples,
let's
say:
I'm,
just
storing
a
key
value
pair
in
my
blockchain
testing,
key
value
pair,
so
when
I
store
some
key
value
pair,
initially
I
support
integer
an
integer
as
value,
maybe
in
my
second
version,
I
support,
both
integer
and
string
as
values.
A
So
when
I
upgrade,
my
TP
I
need
to
give
it
a
new
version
so
that
when
transactions
are
coming
in
from
the
client,
validator
can
know
where
to
distribute
that
transaction
to
whether
it
is
the
first
version
of
handler
or
second
version
of
handlers.
So
answered
for
your
question
is
in
fact
dependent
on
the
client.
I
can
attach
multiple
versions
of
same
transaction
handlers.
A
We
are
transactional
with
same
name
but
with
multiple
versions.
I
can
attach
it
to
same
validator,
node
and
the
client.
When
sends
the
request,
it
is
actually
requesting
against
a
particular
name
and
particular
version.
That's
so
validator
knows
where
to
distribute
the
transaction
to
that's,
how
it
is
used
and
in
fact
we
can
say
we
can
restrict.
Permissioning
is
available.
A
A
B
A
This
particular
handler
has
access
to
all
the
addresses
under
this
under
this
node
in
my
market
tree
so
by
the
I
will
make
a
quick
note
here.
The
other
thing
is
by
default.
Validator
does
not
make
use
of
this
namespace
parameter
unless
we
enable
that
permissioning
check
so
many
times.
People
have
asked
discretion,
also
in
Community
Challenge.
So
just
thought
I'll
point
this
out.
My
default
permissioning
is
not
enabled
is
we
need
to
enable
it.
A
B
A
So
quickly
talking
about
client,
we
discussed
about
validator,
we
discussed
about
rest.
We
discussed
about
TPS
in
general.
Http
can
have
multiple
handlers
and
we
can
have
multiple
such
teepees
attached
to
validator
in
most
of
the
applications
which
you
see,
we
have
just
one
transaction
handler
in
one
TP
attached
to
validator,
so
client
again
can
be
written
in
any
of
the
languages
for
which
SDK
is
provided,
or
you
can
write
your
own
SDK
as
well.
It
is,
in
fact
writing
your
client
SDK
is
much
simpler
because
you
don't
need.
A
All
you
need
to
do
is
just
a
way
to
package
the
protobuf
messages.
So
if
you
have
a
new
language
coming
in,
you
can
write
those
wrappers
to
package.
Your
client
transactions
into
those
protobuf
messages,
send
it
to
validator
it
could
you
can
send
through
REST
API,
or
you
can
also
send
through
zm2
connection
most
of
the
people.
They
try
to
use
the
REST
API
interface,
so
ie
represented
that
in
this
slide
it
can
be
web
app,
it
can
be
mobile
app.
A
It
can
be
a
console
app
anything
depending
on
which
language
you
choose
and
again.
Sdks
are
available
in
multiple
languages.
You
can
write
your
own
SDK
as
well
thing,
one
thing
which
it
is
here.
So
if
this
is
clear,
maybe
I
will
go
deep
into
client
and
talk
few
important
aspects
in
it.
Any
other
questions
here,
I.
B
Have
question
because
I
understood
in
Sawtooth
we
have
transactions
as
well
as
batch
formation
right.
So
in
such
a
case
is
like
in
these
pictures.
You
can
bench.
You
mentioned
that
batcher
say,
for
example,
I
am
doing
your
transactions
on
their
transactions
through
mobile
apps
and
the
batch
formations
will
be
done
at
some
point
of
time.
I
don't
know
where,
like
maybe
here
is
another
one
client
who
will
be
you
know,
accumulating
collect
transactions
of
the
same
category,
inform
the
batches
right.
A
That's
correct
I
mean
that's
a
valid
question
and
it's
in
fact
a
good
question.
That's
what
I
was
about
to
talk
in
next
slide.
Okay,
I'll
cover
that
in
the
next
slide.
So
what
the
question
was
is
in
sawtooth
terms
till
now
I
was
talking.
I
was
using
the
term
called
as
transaction
when
I
spoke
about
TPS,
because
at
its
core
TP
is
receiving
transactions
through
transactions,
but
when
it
comes
to
client
there
is
I
wrote
something
called
as
batcher
in
this
diagram.
A
What
that
means
is
I
actually
batch
together,
a
set
of
transactions
before
sending
to
validator,
and
even
if
I
have
just
one
transaction.
The
way
current
validator
supports
is
by
batching.
It
I
need
to
still
create
a
batch,
for
it
then
send
the
batches
to
validated
a
list
of
batch
to
validator.
As
a
transaction
again,
validator
will
open
up
the
batch.
It
will
process
individual
transactions,
but
that
has
its
own
advantages
and
concept
in
more
included
with
it
benefits
included
with
it.
We
will
cover
that
in
this
slide.
A
As
I
said,
we
have
validated
connect
that
is
connected
with
multiple
transaction
processor.
It
can
have
multiple
handlers
or
just
single
handler
and
then
at
the
client
end.
What
we
do
is
we
compose
a
transaction
as
as
I'm
sorry
I
did
not
get
your
name.
What
was
your
name,
as
somebody
pointed
out
on
the
call
she's
Ruben.
A
So
assets
have
been
pointed
out
on
the
call
when
I
write
a
mobile
application.
There
is
an
example:
mobile
application,
client
available
in
Java
SDK.
If
you
are
interested,
is
it
a
job?
I
think
it
is
Java
SDK
where
it
is
put
up.
If
you're
writing
such
applications,
you
can
add
batcher
itself
in
your
mobile
device
or
you
can
say
my
mobile
is
only
composing
this
transaction
and
there
is
somebody
else
who
is
creating
the
batch
for
me.
A
It's
so
that's
that's
a
concept
which
is
which
sort
of
provides
if
which
means,
if
you
want
dependency
between
transactions,
that
if
this
transaction
is
successful,
only
then
you
want
to
process
other
transactions
or
you
want
to
say.
If
any
of
these
fails,
I
don't
want
to
process
any
of
these
transactions
getting
committed
to
my
network,
so
you
can
batch
them
together.
That
is
one
advantage.
The
other
advantage
is
with
parallel
transaction
or
with
I
may
not
be
remembering
all
the
exact
same
thing.
Sorry
for
that
I
should
have
realized
a
bit.
A
A
Speaking
in
any
blocking
application,
you
don't
necessarily
need
your
client
to
sub
even
compose
the
transaction.
You
can
in
fact,
have
another
level
of
abstraction.
You
can
add
your
own
micro
service,
which
creates
that
which
generates
the
transaction.
Your
micro
service
itself
can
be
acting
as
a
transaction
composer.
There
could
be
another
queue
after
which
there
is
another
micro
service
which
is
composing
the
batch
together,
and
then
that
is
finally
submitting
transaction
to
validator.
It
depends
on
which
use
case
we
are
implementing
and
how
you
want
to
design
your
application
and
okay.
A
I
forgot
to
answer
your
question
so
with
respect
to
permissioning
or
to
know
who
is
allowed
to
perform
transactions
or
not
validator
actually
allows
you
to
configure
in
a
way.
You
can
say
only
these
public
keys
can
be
allowed
to
perform
transactions,
or
you
can
also
say
only
this.
Public
key
is
a
low
platform
all
allowed
to
send
batches.
So
you
can
permission
it
at
transaction
level
as
well
as
batch
level.
A
Both
if
batch
itself
is
not
accepted,
then
it
will
not
go
to
even
transactions
or
if
batch
is
accepted.
Maybe
within
that
you
can
have
transaction
level
checks,
signer
checks
happening
and,
in
fact,
since
you
are
getting
transaction
itself
to
your
smart
contract,
you
can
also
do
something
nice
with
the
public
key
arriving
as
part
of
the
transaction.
A
Your
smart
contracts
won't
be
saying
batches.
Your
smart
contract
will
be
seeing
transactions
when
I,
say
transaction.
You
will
also
receive
two
signer
information.
You
will
receive
a
public
key
of
signer
or
you
will
receive
hash
of
the
transaction.
I
have
not
pointed
out
this
client
structure
in
this
diagram.
A
Some
message
in
the
slightest
kiss
may
be
a
dream.
Yes
slide.
Deck
is
available
in
my
github
account,
as
well
as
in
Piper,
later
Global
Forum
site
yep
that
sterling
from
my
github
thanks
for
posting.
That
okay,
one
thing
which
I
have
not
pointed
out
here
is
the
weight
lines.
Sorry,
the
weight
transaction,
the
different
parameters,
which
transaction
and
batches
have
that
becomes
too
much
of
information.
That's
why
I
did
not
put
it
in
this
slide,
so
there
are
ways
in
which
you
can
control
hey
when
you
hae
transaction
processor,
when
you
receive
a
transaction.
A
A
So
let's
say
you
are
writing
your
mobile
application
to
read
values
from
some
particular
sub
part
of
your
Mulk
Merkle
tree.
Then
you
can
compose
your
transaction
in
a
way
to
a
low
transaction
processor.
Read
only
that
part
and
no
writes
only
read.
You
can
also
do
in
a
way
that
you
can
only
allow
write
access
without
read.
A
All
such
permutation
combinations
are
possible,
and
this
has
not
to
be
confused
with
the
namespace
permissioning,
which
I
told
earlier
or
namespace
restrictions
when
I
spoke
about
transaction
processor,
so
at
high
level
you
have
validated
permissioning
transaction,
permissioning
and
again
within
each
transaction.
We
have
a
client
level,
read
and
write
permission
in
Jax,
which
is
by
default,
enabled
no
matter
what
it
will
happen
for
sure.
Only
the
transaction
addresses
are
loading
that
your
newer
transactions
are
allowed
to
be
read
and
write
written
for
the
structure.
A
Please
refer
to
the
documentation
because
that's
elaborate
information
again,
it
has
multiple
parameters
in
it.
It's
a
different
topic
to
discuss
altogether
and
here
the
namespace
restriction
is
a
different
thing.
Namespace
restriction
is
at
very
high
level.
You
are
telling
you
can
say
you
know
we
have
four
different
handlers
right.
I
can
say
this
handler
has
access
to
these
two
namespaces
or
this
handler
has
access
to
these
three
namespaces.
So
this
is
itis
level.
So
there
is
another
level
of
restriction
which
is
possible.
A
A
Okay,
so
this
is
one
other
equation
which
I
keep
hearing
from
the
community
or
receiving
in
different
channels.
It
could
be
so
stackoverflow
rocket
shot
or
any
other
forums
how
this
privacy
guaranteed
in
sawtooth,
as
I
said.
Initially,
if
you
have
two
different
network
requirements
where
you
as
an
organization
want
to
have
a
network
with
these
three
three
node
network,
or
you
just
want
a
two
node
network,
you
need
one
validator
per
network
in
current
version
of
sawtooth.
A
A
The
reason
we
do
this
is
because
each
validator
has
its
own
global
state
and
we
can
provide
read
restrictions.
Like
note,
11
I
can
say
this.
Node
information
shall
not
be
available
to
some
set
of
groups
of
users,
or
this
node
information
is
not
available
to
this.
Node
I
can
also
exceed
here
note
network
to
node
from
network
to
go
and
talk
to
node
from
network
1
and
read
something
for
me,
so
that
would
be
an
event
listener
attached
to
node,
Network
2
talking
to
node
network
1
or
sorry
event
listener
attached.
A
A
Oracle
service,
ok,
yeah!
So
that's
that's
a
whole
research
topic
or
total
different
topic.
If
we
have
to
discuss
at
very
least
as
we
know
if
we
have
to
write
anything
or
read
anything
in
the
blockchain,
we
do
that
using
the
gate
and
set
methods
exposed
here,
no
matter
which
blockchain
we
use.
There
are
some
put
methods.
There
are
some
read
methods
available
for
us.
A
The
reason
we
do
this
is
because
we
want
to
know
the
context
in
which
it
something
is
written
so
that
we
agree
upon
some
value,
but
there
could
be
scenarios
where
we
need
application
or
my
smart
contract
to
read
the
data
available
outside
the
network.
Maybe
I
want
to
rely
on
the
information
available
in
Internet
to
make
some
decisions
by
the
way.
It's
very
bad
idea
to
make
your
application
rely
on
data,
which
is
not
in
your
net
worth.
A
The
very
simplest
reason
is
because
no
matter
when
you
want
to
execute
your
transaction,
you
always
expect
same
output
from
your
smart
contract.
If
you
execute
your
smart
contract
at
10:00
a.m.
today
or
5:00
p.m.
or
it
could
be
after
a
year,
you
always
expect
same
output
coming
in
from
your
node.
That's
why
we
use
blockchain.
We
guarantee
that
transaction
always
produces
at
with
respect
to
that
context
or
that
block
it
always
produces
same
output.
A
If
I
have
block
one
processed
XYZ
transactions
I
get
block,
it
should
be
same
even
if
I
do
that
after
one
year,
if
you
rely
on
information
available
outside
the
network
outside
the
global
state
or
your
state
store,
there
could
be
possibility
that
somebody
modifies
that
data
of
outside
your
network.
If
you
are
relying
on
internet,
maybe
somebody
edited
the
wiki
page
or
somebody
changed
that
value
outside
the
network
and
your
smart
contract
logic
is
relying
on
that
data
which
is
outside
your
network.
A
B
B
A
You
so
yeah
we
can
consider
a
block
to
network
as
itself
as
a
trusted
Oracle
service,
because
we
can
trust
the
network.
We
nobody
can
modify
the
data
once
it
is
written.
At
least
I
can
trusted
data
coming
in
from
a
blockchain
network
at
that
particular
block,
ID
I
can
say
with
block
ID.
This
was
the
data
which
I
read.
So
if
my
transaction
is
reading
data
at
block
ID,
no
matter
when
I
read,
maybe
today
or
after
a
year
or
maybe
after
10
years,
I
still
read
the
same
Delta
so
that
trust
is
established.
A
Okay,
so
heading
towards
saw
2
dot
o
as
I
said
right
now.
The
approach
is,
you
need
to
write.
You
need
to
have
two
different
nodes
if
you
want
to
participate
in
two
different
networks,
but
in
next
version
of
sawtooth,
it's
possible
for
us
to
run
just
one
instance
of
validator
and
still
participate
in
two
networks
and
get
the
privacy
it's
expected
privacy.
A
Those
course
features
of
sawtooth
will
be
running
as
service
between
a
validator,
node
or
validator
will
be
I'll
prefer
not
spending
more
time
on
this
slide,
because
it's
not
finalized,
which
external
component
we
are
trying
to
use
it's
most
likely
this
printer,
but
I'm,
not
sure
which
one
it
is.
So
this
is
the
approach
which
it
is
tending
towards.
We
will
use
services
between
a
node
to
create
this
virtual
private
networks.
A
B
A
Right
so,
in
current
version,
if
you
have
to
like,
if
you
have
to
network,
for
example,
you
would
need
to
validate
on
instances
I'm
in
maybe
in
later
versions.
I
don't
know
if
it
is
too
dot,
or
maybe
it
is
too
dot
or
dot
something
or
two
dot
something
itself
well
there.
We
can
have
something
like
this.
What
just
I
need
to
bootstrap
one
node
within
that
I
can
have
multiple
services
of
sawtooth.
B
A
Again
so
fuke
comparing
two
different
works
is
not
to
be
done
here,
because
it
has
multiple
things.
If
we
have
to
compare
it,
for
example,
in
sawtooth
we
don't
have
something
like
ordering
service,
because
each
validator
has
both
validating
and
creating
power
in
sawtooth,
whereas
in
fabric
there
is
something
called
as
committing
peer.
There
is
a
validating
peer,
I'm,
not
sure
if
the
same
exact
convention,
which
is
still
followed,
but
something
like
that,
then
there
are
ordering
service
notes.
So
it's
a
different
comparison
and
I
would
not
prefer
to
compare
okay,
sorry.
B
A
System
consideration
as
per
the
rockatuer
conversations,
and
it's
set
to
be
determined
or
accepted
by
the
complete
community.
You
can
imagine
it
to
be
a
like
a
networking
layer
implemented
or
it's
a
library
implemented
for
providing
networking
layer.
So
if
you
see
current
sawtooth,
most
of
it
is
being
converted
from
Python
to
rust
and
the
course
or
truth
is
once
as
in
it
is
as
and
when
it
is
motorized
the
core
sorted
functionality
itself
completely
available
in
rust.
You
can
attach
any
other
networking
layer
to
it.
A
A
A
A
A
Okay,
any
anytime,
we
design
our
application.
We
need
to
make
sure
that
our
smart
contract
is
deterministic
in
nature.
The
word
determinism
here
means
no
matter
when
I
execute
my
transaction
should
always
it
produce
the
same
output.
For
me,
I
can
give
you
one
another
example,
apart
from
relying
on
data
coming
in
from
outside
the
network,
the
simplest
example
could
be
timestamp
itself.
Most
often,
what
we
do
is
when
we
write
our
applications.
We
generate
a
system
timestamp
or
we
take
reference
of
system.
Timestamp
try
to
do
something.
A
For
example,
I
received
a
transaction
I
wrote
some
value:
X
Y
Z
equal
to
1,
to
3
in
some
address
and
I
want
to
associate
with
the
timestamp
and
if
we
are
making
a
design
choice
where
I
generate
a
timestamp
in
my
smart
contract,
probably
that's
a
bad
idea,
the
reason,
because
going
back
to
how
and
when
the
transaction
is
executed,
it
is
actually
executed
in
two
different
cases.
One
is
when
block
is
created
other
instances
when
block
is
X
validated.
A
So
if
you
are
creating
a
timestamp,
you
are
in
fact
creating
like
two
different
timestamps
at
two
different
times
and
if
the
same
transaction
is
executed
in
another
validator,
it
could
be
different
timestamp
on
that,
so
your
smart
contract
times
generated
timestamp
may
not
be
same
in
and
your
nodes
may
never
come
to
agreement
on
the
data
returned
to
blockchain.
So
that's
what
I
mean
by
determinism.
Your
output
should
always
be
deterministic
in
nature.
A
This
is
also
true
for
any
other
block
chains
as
well,
and
you
would
do
something
called
as
serialization
and
encoding
and
addressing
techniques.
One
thing
you
would
do,
no
matter
which
application
you
write
on
so
tooth
is.
First
to
you,
I
mean
this
is
just
a
practice
which
I
suggest
would
be
great.
You
can
say
at
a
high
level.
This
is
what
my
smart
contract
is
going
to
do,
if
possible,
try
to
represent
it
in
the
state
diagram.
A
You
can
say
my
smart
contract
logic
will
move
from
this
state
to
this
state
when
this
input
comes
or
all
those
the
fundamental
diagrams
which
we
generally
write
for
state
transition.
Things
like
when
we
do
some
language
checks
at
very
high
at
a
very
low
level
in
the
compilers.
Those
things
generate
the
state
formations
right,
so
this
would
be
like
and
paper
and
pen
practice,
which
we
generally
do
in
writing.
Applications
install
smart
con.
When
we
write
smart
contracts,
it
will
be
easy
for
us
to
process
from
there.
A
So
once
we
write
the
state
transition
diagrams,
we
know
exactly
what
addresses
we
rely
on
to
make
the
decisions
or
to
make
the
transitions.
So
we
need
to
compute
the
addressing
techniques.
I
need
to
logically
divided
my
Merkel
tree
in
a
way
that
any
address
which
I
store
under
this
node.
If
I
can,
if
I
can
say
this,
is
storing
this
critical
information.
A
I
can
restrict
access
to
it
from
the
client
when
somebody
else
is
sending
transactions
or
some
some
design
decisions
which
you
make
initially
so
that
is
about
addressing
techniques
which
you
need
to
worry
about.
So
I
would
prefer
that
writing
a
state
transition
diagram
from
there
going
to
next
step
would
be
writing.
A
So
this
would
be
a
process
in
writing
business,
the
production
grade
application
and
when
it
comes
to
encoding,
we
need
to
make
sure
that
whatever
we
write
to
leaf
node
in
a
Merkel
tree
is,
is
encoded
or
I
mean
I,
cannot
try
and
ameliorate
some
binary
data
to
that
location
because
that
is
transmitted
over
the
network.
I
need
to
represent
it
in
some
format.
Easy
way
to
do.
That
would
be
maybe
64
encode
my
binary
data,
so
that
it
does
not
exceed
the
range
of
that
base.
A
A
Then
serialization,
as
I
said
in
sawtooth
when
we
store
something
in
the
leaf
node,
we
are
storing
something
some
value
at
an
address
and
that
value
should
be
all
should
always
be.
It's
just
a
value.
It's
not
a
structure
or
it's
not
a
fancy
storage
mechanism
widgets
which
is
providing
us.
So
if
we
have
to
store
something
fancy
like
a
structure
data
or
something
like
an
object
into
that
location,
we
need
to
serialize
it
into
string
format
or
byte
format
and
store
it
at
location.
A
One
advice
here
is
that
we
need
to
be
aware
of
which
serialization
technique
we
are
following,
so
that
we
can
use
the
same
serialization
technique
when
decoding
the
data
as
well.
So,
let's
just
practice
and
again
all
this
information
is
distributed
in
documentation
in
multiple
places.
This
document
is
just
aggregating
them
from
multiple
sources
or
providing
some
means
to
logically
visualize
it.
A
Then
we
need
to
make
some
choose
write
datatypes
when
I.
What
I
mean
by
this
is,
as
I
said,
when
I
serialize
the
data
and
deserialize
I,
expect
the
same
structure
back
if
I
am
storing
map
as
a
data,
if
I
am
using
map
as
a
data
structure
in
Java,
it
is
not
actually
providing
me
the
ordering
guarantee
it
can
write
parameters
in
any
order
and
read
parameters
in
any
order.
A
We
need
to
be
careful
about
that
serialization
or
choosing
which
data
type
we
want
to
use
may
be
linked.
Is
it
hash
map
or
anything
I,
don't
know
tree
map
is
a
better
example
than
just
a
map.
Tree
map
provides
me
that
ordering
world
not
remap,
sorry
yeah
I'm
running
photo
for
it,
so
we
some
some
structured
way
of
writing
and
serializing
and
deserializing
data
should
be
followed
here.
That's
what
I
intend
to
say.
A
B
Have
one
question
on
deterministic,
because
what
I
understood
from
certain
document,
it
is
also
the
language
features
like
Java
or
any
other
language
support,
Sartain
of
chips
or
functions
which
are
non
deterministic
in
nature.
To
me,
like,
is
there
any
repository
or
something
like
that
way
like?
What
are
the
language
features
are
deterministic
in
nature,
I
mean
as
a
reference.
Otherwise
it
is
very
deeply.
A
A
B
A
Yeah,
the
general
advice
is
as
a
programmer.
We
need
to
be
aware
of
what
is
possible.
What
is,
if
I,
am
writing
in
some
order,
then
I
should
be
able
to
read
it
in
the
same
order.
That's
what
I
need
to
worry
about
it
from
programming
aspect.
I,
guess
that's
what,
in
other
words,
is
also
intended
in
the
documentation
as
well.
We
need
to
know
the
language
properties
to
know
to
choose
between
right,
datatypes,
Thanks
and
I
saw
some
judge
happening
so
the
see
and
I'm
okay
quickly,
checking
which
lights
I
have.
A
Next
week
we
have,
and
you
went
from
India
chapter
blockchain
stories
coming
in
and
I
would
suggest
that
most
of
you,
if
possible,
to
join
that
as
well.
I'll,
see
I
cannot
guarantee
that
time,
because
most
of
this
time,
like
most
of
this
public
presentation,
I
also
need
to
get
a
corporate
communications
team
approval,
as
per
organizations
policy.
B
A
Project
and
thank
you
also
I
will
try
to
keep
it
short
and
maybe
take
more
questions
so
that
I
understand
what
exactly
you
people
are
looking
for,
then
explaining
everything.
So
the
important
concept
which
I
told
initially
was
parallel
transaction
execution
at
a
high
level.
What
this
means
is
when
I
compose
a
transaction
in
my
client
I
can
set
dependency
on
other
and
another
transaction
I
can
say,
execute
transaction
one.
A
Only
when
you
execute
transaction,
because
my
for
me
to
process
this
I
need
output
from
transaction
two,
we
can
say
this
dependencies
as
a
plane
parameter.
If
you
see
the
client
structure
the
way
it
composes
the
message
there
is
a
parameter
for
light
to
put
dependency
on
this
and
documentation
is
a
nice
way
to
put
that,
because
that
would
become
too
much
of
programming
aspects
if
I
try
to
put
it
here,
but
conceptually
I
can
do
that
say.
B
A
What
I
intend
to
show
here,
and
if
we
are
putting
transaction
dependency,
then
we
need
to
be
clear
that
I
cannot
execute
them
in
parallel
order,
because
this
depends
on
output
of
this.
So
I
need
to
wait
for
a
result
of
transaction
to
to
process
transaction
one,
whereas
most
of
the
times
it
would
be.
This
may
not
be
the
case.
I
would
be
need
needing
to
make
some
decisions.
Maybe
I
am
executing
multiple
transaction
processors
attached
to
same
validator,
so
I
would
want
to
execute
them
in
parallel
order.
A
I
don't
want
to
restrict
them
from
the
validator
itself
when
moving
to
transaction
processor,
if
I
configure
this,
then
it
is
automatically
taken
care
is
what
I
wanted
to
point
out.
If
there
is
a
dependency
that
dependency
validator
will
take
care
of
it?
The
dependency
we
set
is
by
using
input
and
output
addresses
we
field
to
be
very
specific
in
terms
of
programming.
When
you
compose
the
client
transaction,
there
is
something
called
as
client
input
address,
client
output
address,
which
is
saying
this
transaction
has
read.
A
Access
to
these
input
addresses
this
transaction
has
write
access
to
these
output
addresses
when
this
transaction
is
executed.
I
can
input.
I
can
take
input
from
these
address.
In
my
global
State
or
I
can
output
to
these
addresses
in
my
global
State.
That's
what
in
simplest
term
it
means
and
if
the
dependency,
if
two
transactions
are
having
same
address,
which
means
they
are
dependent
to
each
other
in
other
terms,
I
can
explicitly
set
the
dependency
or
I
can
also
do
it.
A
Do
this
dependency
setting
using
that
input
and
output
address
parameters,
validator
will
understand
if
it
is
just
about
reading
and
I'm
not
going
to
write
in
the
same
address.
Maybe
it
is
okay
to
execute
them
in
parallel,
I
can
run
to
transaction
processor
or
same
transaction
processors
to
different
transactions
in
parallel,
so
it
is
taken
care
by
the
validator.
Is
what
I
wanted
to
point
out?
It
may
not
be
required,
but
it
as
application.
Developer.
A
It's
a
it's.
An
array
of
addresses
in
input
and
output
advantage
with
complete
address
is
validator
will
be
flexible
to
run
them
in
transaction.
In
parallel,
if
I
put
namespace
itself,
which
means
some
parent
nodes
address,
if
I
put
in
input
and
output,
then
validator
will
be
confused,
whether
you
are
using
that,
particularly
if
node
or
not,
if
these
two
transactions
are
using
same
leaf,
node
or
not,
so
it
may
serialize
them
at
its
end.
So
it's
always
better
to
use
the
complete
address
in
the
client
composed
transaction.
A
B
A
A
So
I
can
set
dependency
such
that
when
a
transaction
comes
in
I
want
my
managers
transaction
to
happen.
First,
maybe
write
some
reference
data
on
the
global
state
or
state
store
and
then
CityWalk
and
perform
the
transaction
only
if
manager
has
approved
it.
Some
reference
is
returned
to
state
I.
Read
that
state
store,
I
write
something
else
into
that
same
location,
and
then
it
goes
to
CEO.
They
read
that
data
they
put
so
this
is.
This
is
one
way
we
set.
Dependency
in
transactions
may
be
the
example
I
gave
you
is
bad,
but
yeah.
A
It
depends
on
use
case
how
we
want
to
set
dependency
between
the
transactions
and
for
parallel
transaction
execution
by
default.
If
it's
a
configuration
parameter
in
the
validator
I
can
set
run
all
the
transactions
in
parallel
and
validator
will
make
sure
if
we
have
dependent
input
and
output
addresses.
Only
then,
even
though
we
put
it
as
parallel,
it
will
try
to
serialize
them,
otherwise
it
will
run
them
in
parallel.
If,
for
example,
in
my
first,
let's
forget
this
transaction
dependency
here,
let's
just
say
these
are
two
independent
transactions.
A
In
that
case,
if
I
say
transaction
one
has
address
input
address.
This
output
address
this
transaction
two.
Has
this
input
this
output,
then
they
both
can
be
run
parallel,
because
there
is
no
common
child
node
between
those
transactions,
so
validator
will
run
them
in
parallel.
If
we
have
a
same
address
in
both
the
transactions,
then
validator
will
serialize
them.
B
A
A
Ok,
so
this
is
just
a
good
practice
which
we
follow,
because
most
of
the
permissioning
in
sawtooth
also
happens
wire
on
chain
permissioning.
The
permission
as
I
said,
the
validator.
Sorry
in
both
transaction
permissioning
invalidated
permissioning
restore
set
of
public
keys
in
the
blockchain
itself,
so
there
are
identity
teepee.
There
are
some
transaction
processors
readily
available
in
open
source,
few
of
which
are
settings
TP
identity,
TP.
We
can
reuse
these
to
manage
identity
and
access
problems
in
our
application.
We
don't
to
rewrite
them.
Is
what
I
wanted
to
point
out
in
this
diagram.
A
Any
questions
this
becomes
clear
when
you
write
your
own
application,
but
at
very
high
level
settings
TP
is
a
key
value
pair.
You
can
store
any
key
against
any
value.
Settings.
Tp
will
take
care
of
storing
it
in
some
location.
Retrieving
it
back
to
you
when
your
application
wants
it.
Identity
TP
provides
you
some
access,
the
access
restrictions,
so
this
is
where
this
is
what
we
generally
use
for.
Permissioning
things.
If
you
are
saying
trans
actor
permissioning,
you
said
something
in
identity
namespace
in
some
address.
You,
like
identity
transaction
processor,
handle
that,
for
you.
A
A
Yes,
you
can
use
Indy,
you
can
use
external
I
mean
you
would
use
Indy
for
its
own
benefit.
So
the
difference
between
using
that-
and
this
is
this-
is
for
Unchained
permissioning
checks
and
you
would
attach
Indy
with
sawtooth
to
solve
some
other
business
use
case
itself.
It's
a
business
use
case
level,
problem
against
on
chain
level,
permissioning
checks
problem.
A
When
I
say
this
Unchained
permissioning
check
problem.
In
fact,
you
can
write
your
own
transaction
processor,
your
own
smart
contract
logic
to
read
some
state
settings.
Tp
address.
Take
another
action
based
on
that.
You
don't
need
to
rewrite
that
store.
This
key
get
this
key
value
that
logic
you
can
reuse
from
settings
TP.
A
B
A
Different
use
cases
in
public,
so
recently
an
project
on
sawtooth
was
announced
and
and
by
AirAsia
I
guess:
I
forgot
the
name.
It's
called
Freight
chain,
I,
don't
know
any
other
white
papers,
but
that
is
one
in
production
which
I
am
unaware
of,
and
then
there
is
Tel
Aviv
Stock
Exchange
use
case,
which
I'm
aware
of.
A
There
there
are
few
more
use
cases
which
generally
I
come
to
know
through
the
community
channels
when
they
ask
questions
on
rocket
chart
and
then
they
say,
they're
working
and
so
on
so
use
cases.
I've
seen
use
cases
on
tokenization
have
seen
use
cases
on
bridging
sawtooth
width
or
bridging
hyper
laser
with
titanium
smart
contracts
seen
use
cases
on
banking
solutions,
supply
chain
use
cases,
but.
A
A
This
is
using
this
identity
and
access
management
solutions
or
the
identity
TP,
which
is
our
label
out
of
the
box.
To
enable
this
thing
again,
this
is
not
pointing
out
exact
example.
I
have
examples
written
in
another
repository
in
this
repository
points.
To
each
of
those
examples
are
there
are
open
source
versions
of
examples
available
by
the
core
community
in
Soto
documentation
that
can
be
referred
as
well.
A
So
I
was
expecting
this
question,
but
nobody
asked
this
question
when
I
said
we
cannot
add
timestamp
in
my
transactions
or
by
smart
contract,
I
expected
equation
on
this
asking.
Then,
how
do
we
add
timestamp
information,
sorry,
yeah,
here,
I'm,
sorry,
okay,
I
will
come
to
that
equation,
how
to
hide
timestamp,
it's
still
possible
for
me
to
identify
when
the
block
was
created.
That
is
using
another
concept
which
we'll
cover
quickly
here.
A
A
So
any
read
and
write
on
blockchain
is
expensive
if
it
is
for
analytics
purpose,
we
better
store
it
in
external
database
and
read
it
from
there.
Okay
event
handling
is
again.
These
are
so.
This
material
was
actually
for
a
workshop
with
hands-on
exercise.
That's
why
you
keep
seeing
this
slice
with
time
zone,
pointing
to
a
repository
for
examples.
A
A
A
A
Sorry
for
the
disturbance,
so
yeah
one
way
to
do
this
is
we
can
continuously
create
a
block
on
blockchain
network
in
my
blockchain
network,
maybe
at
five-second
interval
or
like
it
is
configurable
depending
on
which
consensus
algorithm
will
follow.
I
mean
there
is
a
concept
called
as
block
info
injector
in
sawtooth,
which
allows
us
to
add.
Timestamp
information
and
I
can
always
rely
the
data
which
is
available
on
chain.
So
what
I
would
do
is
every
face.
A
B
A
A
There
could
be
as
simple
as
if,
if,
let's
say,
I'm
accept
transactions
only
when
CTO
allows
it
so
CT
was
transaction
could
be
used
to
verify
and
we
can
set
dependency
on
that
transaction
in
somebody
else
transaction.
If
I
am
a
manager
and
I
am
allowed
to
perform
transaction
only
when
my
CTO
allows
it.
A
So
my
transaction
has
a
dependency
on
CTOs
transaction
I,
read
latest
latest
block
information
or
latest
state
information
from
data
from
my
state
from
my
blockchain
put
that
as
a
dependency,
or
put
this
as
a
parameter
some
way
to
identify
that
I'm
allowed
to
perform
transaction
and
make
that
decision
in
my
application.
That's
another
way
depending
on
which
use
case
it
is
we
can
set
in
multiple
ways.
We
can
use
design
it
in
multiple
ways
and
I'm.
A
A
A
Ownership
pattern
there
can
say:
I
can
I
can
set
on
chain.
If
this
value
has
to
be
modified,
then
their
public
key
should
be
only
these
public
keys
should
be
allowed
to
modify
this
value.
This
is
again
very
similar
to
kind
of
permissioning.
We
can
do
it
in
our
mod
contract
as
well.
I
received
something
this
is
the
owner
who
signed
this
transaction
I
verify
whether
this
owner
has
access
to
this
value.
A
The
reason
I'm
pointing
out
these
different
use
cases
or
different
patterns
is
because
we
could
be
familiar.
Writing
a
database
dependent
applications
any
web
application.
We
depend
on
databases.
This
is
a
new
way
of
writing
application
when,
when
it
comes
to
blockchain,
so
that's
the
only
reason
why
I
am
pointing
out
these
things.
A
So
if
I
want
to
secure
some
asset,
I
can
do
that.
I
can
transfer
this
file
off
the
chain,
whereas
on
chain
I
can
store
hash
of
this
document
and
then,
when
the
other
party
receives
that
actual
data,
they
can
verify
whether
whatever
they
received
is
valid
or
not
using
blockchain
by
verifying
the
hash
on
chain.
This
is
another
pattern.
B
A
Can
do
that,
as
I
said
we
can.
We
can
rely
on
previous
block
time
information.
We
do
assess
current
time.
Information
and
one
pattern
is
to
generate
a
block
every
five
seconds,
for
example,
so
my
accuracy
would
be
for
five
seconds
when
this
lady
receives
the
data
and
she
wants
to
say
or
like
she
wants
to
verify
whether
the
data
is
same.
She
will
compute
the
hash
perform
this
transaction,
the
transaction
checks
whether
this
data
already
exists
in
this
address.
A
A
B
A
Transactions
will
never
be
agreed
upon
because
they
should
produce
same
output
for
any
input.
They
can
have
additional
checks,
but
they
should
ideally
produce
same
output
for
transaction
to
be
accepted,
so
each
participant
can
have
their
own
transaction
processor
attached
in
the
network
to
their
own
validators.
If
the
output
is
same,
transaction
is
agreed
upon,
it
is
accepted,
but
what
if
somebody
wants
to
have
same
version
of
smart
contract
in
all
the
nodes?
That
is
when
we
use
something
called
as
sabers
sawtooth
saber?
So
it's
a
transaction
processor,
it
is
attached
to
validator
node.
A
What
happens
is
the
business
logic
which
we
write?
This
is
outside
the
transaction.
Processor
I
write
this
contract
I,
compile
it
into
Bassem.
It's
webassembly
compilation,
which
I
do
currently
type
script
and
rest
are
supported
for
writing.
These
contracts
say
in
Sabre
when
I
compute,
the
vasimr
I,
can
submit
this
by
some
to
Sabre
transaction
processor,
and
this
takes
care
of
executing
all
my
Sable
transactions.
A
Instead
of
writing
the
TP
itself,
I
write
a
vassal
contract
or
Sabre
smart
contract
and
then
submit
it
to
sable
I
mean
this
will
be
executed
via
sable
transaction
processor.
This
is
submitted
and
stored
on
chain
in
the
global
State.
The
reason
we
store
it
in
global
state
is
because
global
State
is
the
replicated
database
in
all
the
nodes.
So
if
I
am
storing
it
in
this
node,
the
same
data
is
available
in
all
the
other
nodes
as
well.
In
that
network,
what
happens
is
when
a
transaction
is
sent.
A
Sable
transaction
processor
will
read
this
stored
contract
or
the
Sabre
smart
contract
extract
it
get
it
to
run
time
instance.
It
will
internally
attach
to
the
turn
time
instance
of
contract.
If
a
transaction
comes
to
this
contract,
it
will
validator
will
pass
that
to
saber.
Saber
will
pass
that
to
that
contract,
so
this
becomes
a
distributor
for
running
different
contracts.
A
B
A
B
A
B
A
A
Sabre
transaction
processor
is
available
out
of
the
box
as
a
community
provided
community
provided
processor
similar
to
any
other
settings,
TP
or
identity.
P
saber
transaction
processor
is
also
a
TP
available,
but
specialized
special
feature
of
this
is
it
allows
you
to
write
via
some
smart
contracts
and
execute
through
this.
So
in
fact,
sabre
transaction
processor
has
its
own
SDK
written.
So
if
you
have
to
write
the
smart
contracts
you
will
instead
of
using
sawtooth
SDK,
you
will
use
saber
SDK.
It
has
similar.
A
A
We
need
to
have
a
specification
that
depends
on
use
case
for
sure
and
the
payload
format
and
the
payload
like
the
weight
of
payloads.
That's
for
sure
and
I
don't
have
a
pre-written
answer
for
that
question
because
it
purely
depends
on
which
use
case
it
is
generally
advisable.
Is
we
need
like
at
least
two
GB
of
space
and
sorry
1
GB
of
space
in
RAM?
It
purely
depends
on
how
many
transactions
you're
handling
at
run
time
because
of
LM
DB
property.
You
may
see
that
your
computer
memory
or
Ram
is
completely
occupied.
A
B
Just
a
question:
it
is
not
directly
related
to
sortal,
but
in
terms
of
application
development.
What
are
the
clock
may
be
features
like
we
want
to
take
care
about.
You
know
that
everything
is
getting
deployed
in
cloud
and
we
need
great
stat
based
of
the
behavior
of
the
cloud.
So
what
are
the
things
we
should
learn
or
it
is
necessary
to
deploy
or
design
and
applications
so
that
it
will
be
run
effectively.
Cloud.
B
B
A
We
in
production,
the
precautions
we
generally
take,
are
permissioning
requirements.
We
need
to
be
very
vigilant
on
that
and
since
anybody
can
read
the
data
through
a
REST
API
interface,
we
ideally
also
want
to
add
a
rest
proxy
layer
or
write
our
own
version
of
list
API
for
reading
the
data
and
provide
ss4
based
mechanism
for
secure,
read
of
the
data.
A
What
else
you
have
some
most
of
the
other
cloud
related
things
which
we
follow,
such
as
container
izing
things
sort
of
provides
as
option
it?
It's
available
container
images
are
available
for
us
docker
images.
We
can
run
them
in
cuban,
it
is
clustered
we
can
run
them.
As
on
just
a
native
Linux
machine,
that's
possible.
A
It
depends
on
the
platform
choice
which
we
make
sorry.
The
infrastructure
choice
which
we
make
it
can
be
in
cloud.
Everything
can
be
different,
can
be
on
different
loads
can
be
on
same
cloud,
it
can
be
in-house,
and
cloud
I
can
just
expose
my
validated
port,
validated
to
validator
connecting
port
everything
else.
I
can
restrict
access
from
within
my
VPN.
All
such
possibilities
are
is
is
available.
It's
it's
a
deployment
choice
which
we
need
to
make
based
on
what
we
want.
B
A
Definitely
it
it
depends
on
what
again
the
choice
of
deployment
as
I
said.
If,
if
you
are
talking
about
some
terraform
scripts,
then
I'm
sure
if
one
is
available
in
open-source,
but
I
have
seen
few
people
writing
their
own
terraform
scripts
and
it's
in
their
own
private
repositories
orally
and
under
their
name
in
their
github
repositories.
I
have
seen
some
examples
and
I've
seen
also
I
guess
Cuban.
It
is
example,
is
available
out
of
the
box
and
yeah
terraform
is
one
thing
which
I
haven't
seen
in
open
source.
A
A
So
when
we
test
smart
contract,
so
I
guess
this
is
the
last
light
or
upgrade
okay,
we
did
Nene
our
mechanism
to
test
our
smart
contracts.
One
is
at
the
development
time
which,
for
which
we
can
use
demoed
consensus
algorithm.
Just
a
single
instance
of
validator
node
perform
different
automated
tests.
There
is
in
fact,
a
library
available
in
Python
which
we
can
make
use
to
simulate
the
validator
itself.
A
So
my
test
case
has
to
be
written
in
Python
I
set
up
a
virtual
network
of
sawtooth
I
perform
as
if
I
am
running
on
the
actual
validator
node.
We
can
validate
my
smart
contracts
in
automated
test
cases.
Example
is
available
in
one
of
the
Python
SDK.
If
you
open,
Python,
SDK
examples,
it's
available
the
integration
test,
then
the
other
is
like
testing
the
network
itself.
We
may
need
to
perform
for
our
use
case.
What
are
the
network
characteristics
or
the
requirements?
The
system
memory
requirement?
A
I
cannot
have
infinite
number
of
transactions
sent
and
only
finite
number
of
transactions
output,
so
so
better
practice
to
add,
make
sure
that
whatever
client,
which
is
sending
finally
to
the
blockchain,
it
retries
the
transactions
if
they
are
not
executed
or
if
they
are
not
accepted
by
the
validator.
If
it
is
accepted
by
the
validator,
we
get
200
response.
A
I
am
all
good,
but
if
I
am
grading
for
twenty
nine,
which
says
hey
block
well,
block
production
rate
is
slower
than
what
is
your
input
rate,
so
maybe
I
will
use
a
queuing
mechanism
or
a
Kafka
and
say:
okay,
I
will
retry
this
after
some
time.
You
process
all
the
transactions
which
you
have
now
all
this
are
to
be
done
before
we
put
something
to
production,
and
we
also
generally
run
in
our
application
for
like
7
days
or
14
days,
to
make
sure
that
long
running
setup
is
working
fine.
So.
A
Yeah,
if
we
have
set
permissioning
with
respect
to
transaction
processor
itself,
it
is
possible
for
us
to
upgrade
still
and
any
client.
As
I
said,
the
transaction
itself
is
carrying
name
and
version
parameters
where
to
send
the
transaction
to.
If,
in
a
case,
we
want
first
version
to
completely
deprecated,
it
is
still
possible
in
case
of
permissioning
I
can
send
a
transaction
which
says:
hey,
don't
accept
any
more
transactions
from
version
1,
except
only
for
version
2
and.
A
It's
very
important
to
archive
current
version
of
TP.
Now
this
is
not
a
problem.
If
we
are
using
Sabre
TP,
because
Sabre
has
on
chain
or
it's
a
stored
procedure
thing,
our
smart
contract
is
in
fact
stored
inside
the
blockchain,
so
we
don't
need
to
worry
about
maintaining
older
versions.
This
is
important
only
if
we
are
using
the
vanilla
version
of
sawtooth,
where
we
use
teepees
any
questions
here.
I.
B
A
No,
there
is
nothing
called
as
come
archive
command
in
validator.
It's
just
a
process
which
we
do
archiving
would
be
I.
Maybe
I
store
this
version
of
TP
source
code
in
some
location
so
that
when
somebody
adds
a
new
node
I
provide
them
this
to
them,
but
maybe
my
node,
crashed
and
I
want
to
add
a
new
node
in
place
of
this.
So
I
have
attached
this
so.
A
A
Our
batch
injection
I
guess
it's
the
last
slide.
Yes,
I
told
you
one
use
case
of
the
batch
transaction
or
batching
for
injector
if
we
want
to
add
timestamp
into
my
I
block
or
if
I
want
to
rely
on
the
time
information.
This
is
one
way
to
do
that.
This
is
a
facial
special
thing
where
I
have
attached
a
TP
to
validator,
but
I
am
not
writing
any
clients
for
this
TP.
In
fact,
validator
itself
is
acting
as
a
client.
A
So
the
feature
is
when
a
block
is
produced.
Validator
knows
that
it
is
composing
a
block
and
if
it
knows
it
is
producing
the
block
and
I
want
something
else
to
be
done.
When
a
block
is
produced,
maybe
I
will
say
when
you
create
a
block.
Let
me
know
I
want
to
add
some
information
to
the
block,
so
validator
itself
acts
as
a
client.
It
creates
that
transaction
which
you
want.
A
It
attaches
it
to
the
block
which
is
created,
so
it
injects
those
transactions
right
now
it
allows
you
to
inject
transactions
of
block
info
injector
type,
so
batches
injected
will
be
along
with
batches
added
from
the
transactions
coming
in.
From
client,
then
you
finally
create
a
block.
That's
why
you
can
rely
on
the
data
which
is
produced
at
runtime
now
right
now
there
is
support
of
block
info
injector
and
if
you
want
to
add
additional
t-piece
to
it,
you
would
need
to
rebuild
the
validator
source
code,
because
validator
should
act
as
a
client
to
it.
A
So
you
need
to
write
client
code
in
validator
similar
to
how
block
info
injector
is
written
and
recompile.
This
use
that
version
of
validator
for
your
application,
I'm
I,
know
one
of
the
person
in
the
community
is
doing
this
and
that
person
keeps
asking
so
many
questions
on
racket
chart
regarding
this.
B
A
B
A
So
yeah
so
the
way
we
can
add
timestamp
to
block.
We
cannot
definitely
hide
it
from
the
smart
contract
because,
as
I
said,
it
can
be
executed
at
different
times,
and
we
cannot
also
rely
on
time
coming
in
from
the
client,
because
client
Kansas's
give
wrong
information.
A
client
is
composing,
the
transaction,
so
client
cannot
any
time
say,
use
a
future
time
or
can
use
a
past
time,
so
injecting
timestamp
will,
if
it
is
done
from
validator
itself.
A
Validator
is
a
trusted
entity
in
the
network.
So
since
in
case
of
injecting
transactions
batch
injection
validator
is
acting
as
a
client.
We
are
injecting
time
stamp
from
the
validator,
it's
not
composed
by
the
client
outside
the
validator.
It
is
composed
by
validator
itself,
so
we
can
rely
on
the
data.
A
B
B
A
So
this
was
written
for
like
documentation
purpose.
We
wanted
to
put
everything
together
in
here.
It
has,
in
fact
just
docker
compose
files
in
different
folders,
for
example.
If
you
want
to
try
hands-on
on
how
event
listener
works,
you
can
use
this
docker
compose
file.
If
you
bring
the
bring
this
up
as
per
the
instructions
written
here,
you
will
be
able
to
test
the
feature
of
event
handling
and
similarly
for
block
info
injector.
If
you
bring
this
up,
then
you
will
be
able
to.
A
A
Similarly,
for
permissioning
this
same
way,
we
have
written
instructions
here
on
how
to
use
these
docker
compose
files.
So
it
in
fact,
is
setting
up
five
different
notes
on
your
mansion
and
then
using
a
shell
to
permission
in
and
then
send
transactions
and
make
sure
that
those
transactions
are
not
accepted
by
date.
That's
the
instructions
which
is
put
up
here
in
the
just
below
that
as
a
readme
file.
A
Similarly,
for
saber
there
is
an
example
written
here
and
if
you
want
the
source
code
itself,
the
one
which
I'm
making
use
in
this,
it's
it's
it's
a
custom,
custom,
TP
and
custom
smart
contract
which
I
have
written
is
called
produce
consume.
Let
me
see
if
I
can
quickly
show
that
all
it
does
is
it
just
produce
I
mean
it
has
a
logic
which
writes
you
can
say,
I'm
producing
this
and
I
am
consuming
this.
A
A
This
is
what
is
packaged
as
docker
images
and
those
docker
images
are
used
in
this
in
these
stalker
compose
files
for
hands-on
exercise,
and
these
packages
are
available
in
github
itself.
Sorry,
in
the
other
repository
yeah,
it
will
pull
automatically.
If
you
have
two
factors,
authentication
enabled
with
github
we'll
be
able
to
use
those
directly.