►
From YouTube: RATS Architecture Design Team Meeting, 2020-09-08
Description
RATS Architecture Design Team Meeting, 2020-09-08
A
A
C
A
A
E
A
Yeah
all
right,
so,
let's
start
so
we
have
three
open
pull
requests
plus
this
one,
where
it's
not
really
me
just
restructuring
the
yaml
to
make
the
contributors
list
of
contributors
structured.
So
if
I
get
that
working,
I
think
it
was
an
xml
to
rfc
issue.
Then
I
hope
there'll
be
no
objection
to
doing
changing
that
around.
A
I
want
to
go
to
the
one
that
is
the
oldest,
which
is
this
one
description,
more
descriptions
of
endorsements.
Where,
aside
from
a
little
bit
of
editing,
we
appear
to
be
having
a
bit
of
a
period
war.
I
think
it's
in
that
one
or
that's
a
96
94.
That's
a
94!
We'll
come
back
to
that.
So
I
don't
know
who
has
seen
this
one.
A
B
C
B
C
B
B
A
A
A
A
Yeah
yeah,
so
I
think
that
we
dealt
with
part
of
of
a
pr
before
that
may
have
moved
some
text
around,
but
that's
about
it-
and
I
also
kind
of
wish
lawrence
had
acted
on
some
of
these
suggestions
that
people
made
like
he
can
accept
these
suggestions
and
and
solve
the
problem.
I
guess
right
the
friendly
amendment,
which
is
accepted
right.
B
D
Here
and
I
think
they
have
to
be
done
thoroughly-
I
mean
this
is
exemplary.
What
I
did
I
did
not
try
to
fix
everything,
only
that
that
stand
out
was
standing
out
and
and
now
I
think
we
have
yeah
it's
hard
to
deal
with
this
pr,
because
I
would
rewrite
it
and
I'm
at
a
loss
so
to
fix
it
with
changes
in
text
that
are
not
only
editorial.
D
C
D
B
B
A
A
D
That
is
true.
We
all
don't
know
how
this
works.
If
the
ietf
wants
to
finish
a
sentence
and
not
a
sentence
with
a
period
which
is
broken,
it's
fine,
but
I
don't
agree
with
this
on
a
semantic
language
level.
Sentences
have
to
be
sentences.
If
it's
not
a
sentence,
it
doesn't
get
a
period.
It's
not
rewarded
without
one
well.
B
A
B
Yes
and
what
you'll
notice
that,
if
you
don't
put
a
period
before
the
word
compare,
then
it
looks
like
it's
a
run-on
sentence,
because
it
has
a
period
at
the
end
where
it
would
be
grammatically
incorrect.
So
you
have
to
have
a
period
before
the
beginning
of
a
new
sentence
right.
So
if
you
have
because
conceptually
in
a
dictionary
it
is
a
blah
is
and
so
on,
and
so
the
colon
is
like
the
termination
of
the
is
of
the
when
you
read
it
out
loud.
C
A
B
D
D
B
A
A
B
B
A
B
B
B
In
that
phrase,
I
would
prefer
how
to
appraise
regarding
it
makes
it
sound
like
you
can
do
it.
However,
you
want
I'm
just
going
to
give
you
some
information
that
may
be
relevant,
whereas
the
eg
is
actually
an
instruction
on
how
to
do
so,
and
so
it
doesn't
quite
match
the
regarding
the
appraisal,
it's
actually
how
to
appraise
in
your
eg.
B
Now
you
can
leave
in
forms
of
verifier,
that's
fine,
instead
of
regarding
the
appraisal,
change
regarding
the
appraisal
to
how
to
appraise
or
perhaps
yeah
that's,
okay,
how
to
appraise
or
okay.
So
here's
what's
interesting.
If
you
look
at
the
red
text
and
then
look
at
140.,
you
can
see
the
nice
symmetry
almost
of
an
evaluates
versus
users
is
not
136
and
140
used
to
be
a
lot
more
symmetric
and
now
they're
not.
D
B
A
A
B
A
B
A
B
B
Because
it's
defined
as
a
set
of
rules
and
the
example
is
an
example
of
one
rule
that
that
I
don't
have
a
negative.
It's
like
I
said
you'd.
I
would
expect
it
to
always
be
such
that.
Please
check
the
signature
against
the
trust
anchor
list,
the
signing
key
against
the
trust
anchor
list
to
make
sure
that
it
changed
to
it
right.
Otherwise,
you.
B
C
B
B
C
D
B
B
C
A
A
D
B
That
is
part
of
her,
that
is
part
of
lawrence's
point
and
he
actually
had
a
request,
or
maybe
it
was
his
email
to
the
list
that
he
was
actually
suggesting
actually
adding
that,
and
I
would
not
object
to
that
if
we
want
to
go
in
that
direction,
but
I
think
yeah.
I
think
that
was
a
message
that
lauren
sent
to
the
list
last
week.
B
B
F
C
C
B
B
Right
now,
there's
two
different
things:
right,
there's
the
statement
that
you're
vouching
for
the
integrity
of
a
tester,
and
so
that's
the
one
that
I
had
just
rephrased
for
purposes
of
conversation
here
that
it's,
the
endorser's
key
is
vouching
for
the
integrity
of
the
attester's
key
right.
That's
the
one
part,
okay,
there's
another
part
that
says
what
are
the
reference
values?
What
I'm
saying
is
that
you
can
get
either
of
those
two
things
without
the
other
one
meaning
you
can
have
two
different
sources.
If
you
change
the
definitions
that
may
include.
B
That
implies
that
all
providers
of
reference
values
also
provide
information
about
the
validity
of
the
tester
key,
and
I'm
saying
that
part
is
not
true.
You
can
get
reference
values
from
a
source
that
is
not
voting
for
your
key.
It's
just
vouching
for
the
reference
values.
That's
why
it's
a
separate
conceptual
message,
or
it
can
be
a
conceptual
measures-
can
be
conveyed
in
the
same.
B
A
A
F
Well,
just
saying
I
agree
with
that
point
of
view.
The
endorsement
is
a
different
idea.
In
fact,
I
think
it's
wrong
to
think
about
a
requirement
of
reference
values
period,
because
not
all
appraisals
will
require
a
comparison
of
reference
values.
There's
other
kinds
of
things
that
you
can
include
as
in
claims
that
might
need
to
be
verified
that
don't
require
you
to
check
against
some
known
set
of
good.
B
Yeah,
I
agree
with
peter
peter.
Would
you
be
okay,
I'm
trying
to
because
it
sounds
like
you
and
I
might
be
a
line
of
these
things.
What
if
we
were
to
add
another
definition,
that
was
reference
values?
Basically
another
conceptual
message
and
to
your
point
that
conceptual
message
might
not
always
exist.
Okay,
but
it's
separate
from
these
other
conceptual
messages.
Would
you
and
I
be
aligned
on
that.
E
F
A
F
B
B
F
F
F
D
C
They
got
there
that
was
the
conveyance
mechanism
right
and
so
we're
not
trying
to
prescribe
a
particular
protocol
for
conveyance.
But
the
point
is:
is
that
the
point
that
the
architecture
wants
to
focus
on
is
the
entity?
That's
most
authoritatively,
you
know
making
the
reference
claims.
So
if
it's,
the
manufacturer
is
the
one
who's
most
authoritative
about
what
it's
supposed
to
be
as
a
reference
claim,
then
that's
how
we've
defined
endorser.
B
E
Yeah
sure,
hey
everyone,
okay!
What's
that
it's
all
right!
I
think
I
may
know
some
of
you.
It's
sorry,
I'm
I'm
at
microsoft.
I
work
on
azure
sphere.
Security,
azure
sphere
is
one
of
the
iot
things
and
we
do
at
the
station
all
over
the
place.
So
dave-
and
I
have
been
talking
about
this
so
here
I
am-
and
I
started
reviewing
the
id,
but
I'm
not
entirely
done
yet.
So
thanks
dave
for
inviting
me,
this
is
very
useful.
So
far,.
B
Thanks
for
being
here
and
would
love
to
have
your
input,
if
you
have
opinions
on
these
things
sure,
so
what
I
was
saying
is
the
way
that
we
have
defined
endorsement.
This
is
from
memories.
Maybe
you
can
bring
it
up.
The
way
that
we've
defined
endorsement
is
sorry.
The
way
that
we've
defined
endorser
and
endorsement
is
an
endorser
is
just
somebody
who
sends
an
endorsement.
So
an
endorser
is
somebody
typically
a
manufacturer,
I
think,
is
the
phrase
typically
a
manufacturer
that
sends
it
out.
B
C
B
Okay,
thank
you.
So
the
point
is
the
endorser
is
not
defined
as
a
manufacturer,
it's
defined
as
the
entity
that
sends
the
endorsement,
and
that
is
typically
a
manufacturer.
We
can
define
reference
values
or
whatever
the
message
name
that
we
want
to
use
for
reference
values
as
being
reference.
Values
could
be
something
like
you
know,
expected
values
of
claims.
B
Sorry,
there
are
a
sender
of
reference
values.
I'm
trying
to
answer
your
question,
who
sends
reference
values.
We
could
use
something
that
is
typically
a
manufacturer
and
it's
sent
by
the
entity
that
sends
entity
that
sends
reference
values.
We
just
need
a
term
for
it.
You
know
reference
value
provider
or
something
like
that
would
be
an
entity
typically,
a
manufacturer
that
provides
reference
values
used
in
appraisal.
F
C
F
B
C
B
I
think
it
confuses
the
verb
endorse
if
we
were
to
do
that
right
now.
Endorse
is
used
to
be
about
endorsing
the
attester's
signing
capability
and
I
think
what
you're
asking
is.
Could
we
use
endorse
with
a
different
meaning
could
endorse
be
to
endorse
reference
values
without
endorsing
the
a
tester
signing
capability?
I'm
saying
I
would
find
that
slightly
more
confusing.
I
agree
it
would
be
internally
consistent.
I
think
it
is
a
less
desirable
state,
but
could
I
live
with
it?
Yes
do
I
like
it?
No.
F
F
The
same
thing,
but
you
know
whoever's
providing
those
reference
values,
whether
it's
beforehand
or
actively
as
part
of
an
attestation.
You
could
imagine
that
they
would
need
to
have
an
endorsement
with
them
to
say
that
they're
trustworthy,
that
they
were
signed
by
somebody,
and
that
you
believe
that
signature.
B
B
D
C
D
A
So
I
I
want
to
suggest
that
maybe
we
could
go
on
to
other
parts
of
this
request
and
and
and
then
come
back
circle
around
to
this
question.
What
I
heard
is
that
there
was
there
was
a
soft
objection
to
this
sentence
that
you
would
rather
do
it
some
other
way,
but
that
it's
better
than
what
was
there
before
and
that
dave.
If
you
were
going
to
write
that
pr,
you
might
well,
I
might
commit
this
one,
but
then
you
might
then
delete
it
and
move
it
somewhere
else.
C
B
C
A
B
A
B
A
B
That
would
be
my
ideal
line,
136.
or
sorry
one
dirt.
Yet
your
ideal,
I'm
sorry
the
red
text
with
two
changes:
yeah
you're,
basically
reverting
a
bunch
of
the
changes
that
were
made
by
laurence
and
stuff.
There
are
two
things
that
we've
talked
about
with
that.
I
think
the
red
text
could
be
fixed
in
the
first.
One
is
the
word
in
forums.
If
you
change
that
to
direct
so
then
it
matches
line
140.
B
C
B
C
A
A
A
A
A
B
D
E
A
B
F
B
So
let
me
give
you
an
example
of
where
I
say
it's
not
required
because
I
mentioned
this
with
lawrence,
but
since
there's
perhaps
other
people
in
the
call
now
the
endorsement
is
a
case
where
an
entity,
but
typically
the
manufacturer,
is
vouching
for
the
signing
key
of
the
tester
and
so
when
you're
using
endorsements.
That
means
that
your
verifier,
when
the
evidence
comes
in
it,
says
hey,
is
it
signed
by
a
key
that
I
trust
what
I'm
looking
at
my
trust
anchor
for
is
I
use
the
the
key.
B
That's
in
the
evidence,
and
I
use
the
endorsement
to
say
is
this-
is
the
key
in
the
evidence
signed
by
the
manufacturer's
key
and
then
is
the
manufacturer's
key
in
my
trust,
anchor
store?
If
so,
then
the
signature
chain
matches
right
a
far
less
scalable
one.
If
I've
only
got
three
devices
that
are
in
the
entire,
you
know
chain
of
stuff
that
I'm
ever
going
to
use.
B
Then
I
can
skip
the
middle
step
and
just
put
all
three
of
the
testers
signing
keys
in
my
trust,
anchor
score
store,
and
if
I
do
that,
I
don't
need
the
endorsement.
It's
just
not
scalable,
because
I
can't
scale
it
out
and
I
can't
manage
those
keys
if
they
get
revved
or
has
devices
add,
and
so
the
endorsements
give
you
the
scalability.
It
doesn't
give
you
the
correctness
right
right.
B
So
that's
what
I
mean
by
it's
not
required.
It's
just
typically
used
because
everybody
wants
stuff
to
scale
right,
just
just
add
a
layer
of
direction.
But
if
you
have
a
deployment,
that's
only
ever
going
to
have
three
devices,
you
could
just
stick
those
three
keys
once
you've
invented
the
device.
Do
some
out-of-band
mechanism,
your
verifier
might
just
stick
those
three
keys
in
its
trust
anchor
store,
and
you
don't
need
endorsement
because
you'll
need
that
if
the
manufacturer's
key
is
what's
in
your
trust,
anchor
store
you've
kind
of
front
loaded,
the
endorsement
of
that
installation.
B
B
A
B
B
C
A
So
so
what
I'm
gonna
do
is
I've
made
a
note?
I'm
gonna,
I
I'm
gonna,
say
because
we
closed
a
couple
of
his,
which
are
our
last
ones
really
and
say
this.
Is
this
disposition
of
these
and
I'm
going
to
say
that
we
we
acknowledge
that
the
there's
a
better
need
for
reference,
the
word
reference
values
in
places
and
that
that
dave
had
some
idea
of
a
back
up
here
about
what
to
do
well.
Dave.
B
C
B
F
Yeah,
I
I
kind
of
believe
that
you
really
don't
need
to
add
a
kind
of
a
formal
role.
It's
just
state,
as
a
matter
of
fact
that
these
things
need
to
happen.
I
mean,
I
think,
you've
already
got
comments
back
about.
You
know,
making
something
overly
complicated
where
they
don't
think
it
needs
to
be
and
making
it
more
complicated
is
not
going
to
help
that.
B
So
I
think
one
of
the
things
that
lawrence
in
particular
was
looking
for
is
when
we
looked
at
that
figure
number
one
which
is
the
one
with
the
roles
and
the
conceptual
messages
labeled
as
a
reference
architecture,
whatever
he
was
looking
for
somebody
to
answer
sorry
for
some
place
in
the
text
answer
and
wearing
this
with
the
you
know,
which
line
or
whatever
or
which
role
or
whatever
is
reference
values
in
here.
I
think
that
was
his
original
question.
Yeah.
C
B
Remember
right
so,
should
we
have
another
line
or
should
we
overload
this
line,
given
that
at
least
peter-
and
I
agree
that
endorsements
and
reference
values
are
two
different
things
that
you
could
have
as
reference
values
that
are
provided
without
an
endorsement
in
terms
of
endorsing
the
signing
key,
or
should
it
be
combined
into
that?
I
think,
is
the
question
right.
D
C
B
D
I
may
have
a
strong
objection
with
saying
the
other
box
is
something
else
in
an
endorsement
because
from
semantically
only
because
it
really
endorser
and
endorsements
unfortunately
have
the
same
syntactic
notation
but
endorser.
Most
certainly
semantically
includes
the
producer
of
reference
values.
It's
just
another
endorsement.
You
said.
E
B
All
right
so,
michael,
if
you
want
to
keep
going,
I
don't
think,
there's
anything
else
in
there.
I'm
going
to
go
ahead
and
generate
a
pull
request.
It
might
not
be
today,
but
it'll,
be
before
the
next
time
that
we
meet,
and
I
don't
want
to
say
whether
it's
next
week
or
two
weeks
or
whatever,
and
I
think
peter
and
I
will
be
aligned
on
it
and
you
can
beat
it
up.
Then.
A
A
B
F
B
F
B
A
A
A
F
B
C
E
C
F
C
F
D
D
Has
the
semantics
the
container
is
the
evidence
or
the
reference
values
container
whatever
that
is,
and
and
and
and
both
contain
claims
to
compare
the
reference
value
claims
with
the
evidence
claims
claims
is
not
has
not
semantics
in
the
context
here.
It's
just
a
assertion
made
by
an
endorser
used
for
the
reference
values
or
an
assertion
created
by
the
tester
used
for
its
evidence
that
that's
the
the
the
common
semantics
of
claim
it's
very
very
abroad.
I
think.
C
C
F
Way,
I
would
look
at
you
know
what
hank
just
said
is
that
that
changes
appraisal
from
this
idea
of
verifying
claims
to
adjudication
of
of
balancing
competing
claims
to
see
which
is
correct
and
that's,
okay
right
all
right,
I
I
don't
think
that's
what
attestation
really
is.
You
know
the
verifier
is
looking
at
evidence
presented
in
my
terminology.
It
would
be
a
claim
or
a
set
of
claims
and
making
a
decision
that
says
yes,
I've
evaluated
that
evidence
and
it
is
either
okay
or
not.
D
E
D
B
A
All
right,
thank
you
all,
so
the
the
conclusion
is
that
we're
not
going
to
do
anything
with
this
one
right
now
finish.
It
dave
is
going
to
propose
some
things
and
I'm
going
to
follow
up
to
the
list
and
lawrence
to
say
that
we
understand
and
we're
thinking
of
changing
the
diagram,
maybe
like
this,
and
that
we're
resuming
in
two
weeks.
Yep
sounds
good
all
right.
Wonderful,
thank.