►
From YouTube: RATS Architecture Design Team Meeting, 2020-09-01
Description
RATS Architecture Design Team Meeting, 2020-09-01
A
I
guess
I
can
share
my
screen.
I
guess
you
guys
got
through
a
pr
or
two
last
week
without.
B
C
C
Did
that
it
references
in
my
comments
in
the
initial
comment,
this
pr
is
one
two
three,
I
actually
don't
know
what
the
new
number
is
one
for
five.
Yes,.
C
A
Okay,
just
keeps
track
of
what
we're
eating.
So,
let's
just
start
with
this
one,
then,
if
that's
okay
or
you
want
to
do
the
144.
B
B
So
hank
has
a
comment
on
this
one,
and
this
is
one
of
the
main
questions
we
had
for
people
and
the
only
reason
why
reviewing
the
other
one
before
this
one
might
make
sense
just
because
they
are
interrelated
previously
in
the
document
there
were
these
two
rows
which
you
can
see
in
red
up
at
the
top
5253
value
generation
and
test
your
awareness.
B
Currently,
the
tester
awareness
was
not
used,
that's
what
would
be
used
in
hank's
per
request,
but
this
grows
in
the
table
here.
One
of
the
observations
that
we
had
was
that
every
example
not
just
the
one
that
hanks
hank
is
adding,
but
every
example
there
can
be
two
possibilities
for
any
particular
claim
right.
One
is
that
the
value
was
created
before
evidence
was
done
and
the
tester
can
actually
get
what
time
that
was
created
so
take
a
case
where
you
have
a
gps.
B
Even
the
you
know
the
the
vg,
maybe
top
even
before
the
boot
time
of
the
main
processor,
then
you
could
also
have
a
case
where
you
can't
tell,
and
so,
when
you
boot
up,
you
don't
have
any
idea
how
long
that
value
has
been
that
way
before,
and
so
as
a
workaround.
You
use
the
tester
awareness,
which
is
the
time
that
you're
creating
a
claim
and
those
two
possibilities
apply
to
every
example,
and
so
how
do
we
do
that
without
invalidating
all
of
the
pictures,
and
so
we
or
without
causing
lots
of
duplication?
B
So
one
approach
that
we
discussed
was
to
say:
well,
let's
combine
the
events
and
put
it
in
the
explanation
of
events
saying
it
could
be
either
way.
Hink
has
a
comment
at
the
at
the
bottom
of
your
screen
now,
which
says
that
it's
not
distinguishable
by
the
receiver
and
one
of
the
questions
that
we
had
that
I
wanted
to
do
that
hank
and
I
had
discussed
that.
I
wanted
to
think
about
some
more
because
I
said
what
I
wonder
is
if
there
is
a
case
if
it
actually
should
be
different
times
in
community.
B
Another
way
to
do
it
is
to
say
you
put
in
a
time,
and
then
you
have
a
separate
piece
of
metadata
that
talks
about
your
confidence
in
another
time
resolution,
and
that
would
be
more
like
what
lauren,
I
think,
florence
in
the
eat
document
you
have
like
what
type
of
bus
it
was
received
across,
was
the
bus
trusted
or
was
it
inside
or
outside,
and
so
that's
like
an
orthogonal
piece
of
metadata
and
so
exactly
how
you
would
encode
the
difference.
B
A
So
so,
if,
if
I
can
repeat
back
to
you
what
I
think
you
said,
on
the
one
hand,
you
suggested
that
that
the
this
time,
whether
it's
value
created
or
value
generated
essentially,
might
be
encodable
with
a
an
error
amount,
meaning
I
think
it's
around
now,
but
it
could
be
a
little
bit
up
as
much
as
this
much
before
another
and
the
other
part
not
mutually
exclusive.
Is
that
the
claim
would
say
well,
it
came
across
this
kind
of
a
bus,
and
so
that
implicitly
means
that
it
must
have
at
least
taken.
B
The
lighter
part
is
already
true,
meaning
I
think
lawrence
had
some
attribute
in
the
somebody
had
an
attribute
in
the
eat
document.
That
already
had
that.
I
was
giving
an
analogy
that
you
could
have
yet
another
piece
of
death
of
metadata
yeah
similar
to
that
one,
but
not
what
type
of
post
it
came
across,
but
whether
it
came
from
some
place
that
has
its
own
time
stamp
or
not.
B
A
B
Those
are
just
an
example.
We
already
have
multiple
pieces
of
orthogonal
metadata
about
the
same
claim
set.
This
would
this
it
could
be
done
as
another
one
of
those,
and
I
again
I
did
not
propose
any
text
to
do
with
that
fact,
but
that-
and
that
is
what
hank
is
highlighting
here
is-
that
is
a
point
that
we
had
discussed,
but
there's
nothing
in
the
text
that
talks
about
this,
and
how
should
we
deal
with?
This
is
kind
of
an
open
question.
A
Okay,
so
I
got
you
so
we
don't
know
whether
or
not
we
need
to
put
this
other
claim
in
about
the
time
having
originated
from
another
device
and.
B
Right
there's
at
least
two
ways
we
could
do.
A
Claim
a
specific
claim
may
want
to
detail
this
and
it
might
want
to
standardize
the
way
in
which
that
is
detailed,
but
I'm
not
clear
that
the
architecture
this
sounds
to
me
like
another
policy
for
appraisal,
rather
than
a
specific.
A
C
B
In
here
is,
if
you
combine
the
rows,
it
implies
that
there
is
some
other
way
to
tell.
If
you
don't
combine
the
rows,
then
you
have
to
have
text
that
says
in
this
example,
we're
going
to
use
vg
in
this
example
we're
going
to
use
a
a
in
this
example.
We're
going
to
use
vg
and
we've
got
to
make
sure
that
across
the
examples
we
arbitrarily
pick
one
or
the
other
and
explain
that
that
could
have
been
the
other
one.
B
Sorry
other
way
around,
if
you
have
fewer,
if
you
have
more
labels,
it
requires
a
bunch
of
text
in
various
places.
If
you
have
fewer
labels,
you
only
have
text
in
one
place.
B
A
Somewhere
else,
someone
has
to
tell
you
whether
or
not
this
was
generated
by
the
in
your
example,
by
the
gps
or
by
the
tester.
B
If
we
go
with
the
combined
place,
then
each
example
does
not.
You
just
have
to
say
here
that
every
example
could
be
either
one.
If
you
have
two
different
rows,
then
every
example
has
to
say
we're
using
you
know
vg,
but
it
could
have
been
a
a
or
something
like
that,
because
otherwise
you'd
think
that
it
was
inherent
to
that
example,
which
it's
not.
D
I
kind
of
like
them
separate
the
reason
it
was
a
tester
awareness.
I
think
it
goes
back
to
the
discussion
from
a
while
ago
for
handles
what
happens
if
you
generate
a
not
some
nonce
or
some
value
centrally,
but
you
don't
get
it
until
later
now
I
know
the
point
you're
talking
about
is:
can
you
really
tell
the
difference.
B
B
Because
in
if
you're
receiving
a
handle,
aa
and
vg
are
both
long
before
that
point
and
you
might
even
know
the
delta
between
vg
or
aaa
and
the
hender
received,
and
so
that
will
show
up
in
hank's
pr.
So
this
is
orthogonal
to
that.
D
C
So
more
and
more
I
would
this
is
hank's
opinion
I
could.
I
can
see
how
this
could
be
collapsed
and
then
specialized,
for
example,
in
each
that
then
deals
with
specific
claims
about
that.
C
We'll
flesh
that
out
again,
I
don't
think
that
the
architect
has
to
be
overly
expressive.
We
have
less
growing
pains
if
we
keep
the
acronym,
that's
why
entry
is
generated,
but
that's
only
only
a
super
micro
net.
I
think
it's
it's
especially
easy
not
to
edit
all
the
text
again
and
then
arbitrarily
choose
a
a
and
v
whatever
at
some
point.
I
think
it's
it's
literally
easier.
C
The
way
dave
is
proposing
it
right
now
and
as
long
as
it's
clear
that
there
is
some
attester
awareness
there,
and
it
says
that
literally
I'm
fine
and
then
each
can
maybe
deal
with
something
that
is
appropriate
for
really
differentiating
a
and
b
something.
B
F
B
F
C
B
Strongly
on
this
one,
including
me
right,
it's
just
I
I
propose
combining
them,
because
I
think
it's
actually
less
text
overall
in
the
architecture
document,
if
you
combine
them,
but
again,
none
of
us
feel
strongly.
It
sounds
like.
A
Okay,
so
are
you
happy
about
I'm?
Yes,
I'm.
I
would
go
for
combining
them
at
this
point
then,
so
and
and
hank's
suggestion
that
we
call
it
value
generated.
C
Yeah,
it's
consistent
with
evidence,
generated
or
end
of
a
generation,
but
like
just
consistency
with
terms,
that's
all
basically.
B
That's
true
yeah
because
I
think
it
is
used
in
another
document
or
two
so
yeah,
and
I
don't
know
if
you
want
to
do
this
now.
Michael
er
can
do
it
offline
in
a
because
there's
a
there's
like
10
of
them.
A
Okay,
I
won't
do
it
now,
all
right
okay,
so
we
got
this
yeah.
Probably
just
you
know.
B
This
is
the
other
notational
question,
the
one
that
I
said
if
you,
if,
if
we
go
with
this
style,
it
will
affect
hank's
pr
where
he
didn't
use
the
stocks.
We
were
doing
this
in
parallel
right.
So
the
observation
here
is
that
it
would
not
be
relevant
if
somebody
were
to
say
take
time,
eg
and
time
ns
and
look
at
the
delta
between
those
two
right.
B
B
B
B
You
can
see
ns
prime
in
here
whether
we
keep
the
prime
or
not.
I
think
when
we
look
at
hanks,
we
may
want
to
argue
to
keep
the
prime,
but
you
can
see.
Nsr
is
different
from
nsv
up
in
1245,
because
they're
different
nonces,
so
maybe
that
should
be
ns
prime
underscore
r.
But
again
and
you'll
see
that's
the
approach
that
hank
took
because
there's
different
handles
in
his
equivalent
diagram.
D
E
E
Breaking
the
convention
of
identifying
the
clock,
it
should
be
rp
or
something
an
srp.
B
E
B
Is
missing
a
prime?
That's
all
I'm
pointing
out
right
now,
and
we
may
want
to
add
that
back
in
and
I
think
when
you
look
at
hexpr
you'll
see.
Maybe
why.
B
Correct,
that's
the
that's
the
r
suffix
in
1254,
it
didn't
know.
A
B
Them
then
it
makes
you
wonder:
oh
is
that
doing
this
incomparable
clock
comparisons
and
no?
No,
no,
I
checked
they're,
all
all
the
deltas
are
between
things
in
the
same
clock
and
then
the
last
nomenclature
difference
you'll,
see
between
mine
and
hank's
as
you'll
see
in
the
document.
Right
now
you
can
see
all
the
deltas
are
expressed,
as
you
know,
time
a
minus
time
b
and
in
hank's
pr
he
has.
B
It
is
a
delta
parentheses,
a
comma
b
and
my
comment
is:
we
should
just
pick
one
and
use
it
whether
it's
a
time
a
minus
time
b
or
delta,
a
minus
or
a
comma
b.
B
B
I'm
okay
with
either
one.
My
question
is,
which
one
people
think
is
actually
clear
to
understand.
C
I
I
was
using
younger
time
minus
older
time,
which
results
in
a
positive
duration,
yeah,
and
so
that
could
be
something
we
can
also
try.
A
Yeah
yeah,
I
think.
A
B
B
That
one
are
you
using
the
latest
one
or
is
it
entered
on
farther
down?
Yes,
what
I
had
said
is
his
comment.
B
I'm
checking
yeah
here
right
hold
on.
B
Okay,
there
you
go
so
now,
let's
scroll
down
a
little
bit
more,
so
you
can
see
the
suggestion
as
well.
Okay,
so
william
was
pointing
out
that
there
is
arguably
an
error
in
this
inequality
which
was
already
there
before.
Okay,
because
you
can
see
the
the
I
would
say
equation,
but
inequality
in
the
red
and
the
green
is
the
same
and
he's
pointing
out
that
there
should
be
this
other
factor.
That's
in
there,
that's
missing
and
so
yeah.
So
that's!
B
B
My
original
comment
was
well.
I
didn't
change
that
before
you're
right,
but
maybe
that's
a
different
pr
and
I
said
it's
only
one
line:
let's
just
do
it
here
so.
B
B
A
C
Thompson's
I
just
caught
this.
Unfortunately,
in
my
pull
request,
but
the
lip
updated
and
now
it
says
main
and
if
it
doesn't
say
main,
it
flows
a
warning.
So
this
is
not
me.
This
is
thompson's
lib.
So
this
is
an
artifact
that
is
just
logistics.
A
C
C
Yeah
also,
I
caught
a
reference
knit.
A
Yeah
I
fixed
those
two
in
the
main
as
well.
So
that's
why
I
asked
you
rebate
so
ignore
that
I
put
that
into
the
master
already
and
if
you
do
your
rebase
and
push
go
away
from
that
yeah.
I
can
do
this
afterwards.
C
So
also
I
have
this-
I
mean
it's
especially
lawrence
created
these
two
references
because
they
look
very
fight
to
ask
lawrence.
Please
check
if
you
think
of
heights
before
you
commit,
but
that's
okay,
because
I
was
like.
Why
is
this
all
compiling?
It
took
me
like
three
minutes
to
find
out
that
the
offer
was
missing
august,
missing
sort
of.
A
Org
was
missing.
That
may
be
me
yeah.
That
might
be
me
because
I'm
stupid
that
way,
but
I
thought
I
compiled
it,
but
anyway,
okay.
C
So
that
was
that
yeah.
So
again,
as
they've
highlighted,
we
had
a
small
offline
discussion
and
we
came
to
the
resolution
that
handles
is
probably
a
third
type
that
does
not
really
fit
nons
and
time
stamps.
So
first
paragraph
here
is
about
context:
what's
a
handle
basically.
C
And
so
they've
made
some
comments
here,
a
change
actually
yeah.
Thank
you.
English
is
better
kept
with
native
speakers.
C
Yeah,
so,
okay,
that's
always
welcome,
of
course,
so
yeah
then
I'm
very,
very
easy.
The
handles
are
handles
are
because
why
are
they
different?
You
cannot
just
claim
it.
You
have
to
highlight
why
this
is
the
case.
It's
just
me
being
a
little
bit
furrow
moving
down
a
little
bit,
please
and
you
can
parse
that
one.
C
And
there's
the
event,
so
the
handler
received
outdates
old
one
and
supercedes
by
a
new
one.
A
C
C
Okay
yeah,
it's
that's
just
me
not
knowing
the
right
terms
in
english
language
here
so,
but
thank
you.
I
like
not
only
invalidate
but
also
outdates.
That's
why
I
wrote
out
dates
and
therefore
in
validates
these
are,
but
that's
again
only
me
if
it's
clear
enough
it
probably
it's
fine.
C
A
B
Where
oh
mismatched
parentheses,
there
is
more
opens
than
closes.
C
B
But
but
the
better
question
this
is
what
I
was
talking
about.
You
can
see
delta
with
a
comma
and
if
you
replace
the
delta
with
time
e
g,
if
you
put
a
minus,
then
you
can
delete
the
time
the
delta
open
parenthesis.
C
Yeah,
that
is
absolutely
correct.
I
would
agree
that.
B
I
would
replace
validity
duration
with
the
word
threshold
with
a
capital
t,
just
because
that's
how
the
other
examples
did
it.
B
1304
the
inequality
should
be
something
like
the
delta
on
the
right.
Minus
the
delta
on
the
left
is
less
than
threshold.
B
So
it
should
be
like
this:
the
delta,
that's
on
the
right,
so
the
hd
hd
prime
delta
minus
the
hdeg
delta.
Now
you
have
to
reorder
the
phrases
because
the
one
on
the
right
is
bigger
and
you
do
the
subtraction
you
put
the
bigger
one.
First.
B
C
Also,
it
has
to
be
as
a
hd
prime,
which
also
should
be
h,
r,
I'm
very
sorry
about
that.
This
old
text
doesn't
really
make
sense
to
me.
B
B
Just
for
consistency,
michael
with
where
the
spaces
go,
when
we're
doing
the
first
delta
delete
the
space
before
it.
In
other
words,
and
the
other
examples,
we
don't
have
a
space
around
the
equal
sign
when
we're
just
computing,
a
delta
and
so
the
same
thing.
At
the
end
of
the
line,
there
is
yeah
that
one.
B
C
Come
before
you
commit
this,
could
you
scroll
up
this
as
a
suggestion?
Could
you
scroll
up
again,
please,
because
I
am
in
line
sorry
to
watch
so
far
so
far
below
and
line
1301,
I'm
defining
validity
duration,
which
is
time
h,
r,
prime
plus,
offset
minus
time
h,
r?
That's
why
I'm
using
the.
B
B
B
B
E
C
I
will
make
a
pass
after
this
and
read
the
merge
master
in
the
end
main
story
and
then
yeah.
So
basically,
this
is
all
set
if
nobody
disagrees
right
now.
The
diagram
is
ugly.
Sorry,
but
I
did
not.
B
I
didn't
finish
reviewing
it,
so
I
can't
guarantee
that
I
won't
find
some
other
bug
in
the
diagram,
since
I
kind
of
got
down
to
about
line
13
10
before
the
call
so
we're.
B
A
And
so
someone
will
do
that
consistently
across
the
right.
F
A
E
A
A
Okay,
I'll
because
I
just
fixed
the
base
I'll
I'll
fix
this
okay.
So
what
next.
B
A
A
B
So
what
I
want
to
make
sure
that
markdown
does
right.
So
if
you
go
into
the
rich
disk
view
button
up
top,
I
want
to
make
sure
the
underscores
render
right
in
github
yeah
you
had
yeah
there.
You
go
so
example
two
and
four
okay
yeah.
They
do.
Okay,.
B
B
I
just
want
to
make
sure
I
didn't
need
to
backslash
escape
them
or
something
so.
B
B
One
I
have
not
looked
at
any
of
lawrence's
new
ones.
I
see
lauren's
open
new
ones.
Yesterday
I
reviewed
hanks
but
didn't
have
time
to
get
through
lawrence's.
C
Fresh
ones
are
relatively
concise,
like
four
two
and
four.
C
A
B
I
think
we
did
that's
why
lawrence
is
capitalizing
claim.
B
B
B
B
E
B
Let's
see
about
the
eighth
word
or
something
is
places
yeah
yeah,
sorry,
you
know
you
had
the
right
michael,
the
word
no
limit
to
or
requirement
of
places
right
before
places,
nope,
not
the
word
of
advance.
One
word
later
yep
that
one
is
there
a
better
preposition,
there's
no
requirement
on
places
or
four
places.
B
Wrong
place,
move
it
upward
leftward
there
you
go.
Yes,.
B
B
B
B
Oh
okay,
I
think
that
lawrence's
change
is
still
consistent
with
what
the
diagram
shows.
If
you
go
to
the
first
diagram
of
the
whole
document,.
B
B
Not
that
one
just
go
to
the
full
editors
view.
E
B
So,
let's
actually
compare
it
with
the
sentence.
The
first
sentence,
the
appraisal
policy
for
evidence-
might
be
obtained
from
an
endorser
or
might
be
obtained
via
some
other
mechanism,
such
as
being
configured
in
the
verifier
by
the
verifier
owner.
Maybe
that's
why
it
was
an
example
used
for
parity
with
that
sentence
which
says.
B
E
F
I,
like
the
is
because
it's
I
think
it's
sharper
and
I
don't
like
the
idea
of
sort
of
important
characteristics
or
data
items
or
configurations
coming
sort
of
from
ambient,
or
you
know
not
specified
places.
I
mean
that
the
the
arrows
in
that
diagram
are
there
for
a
reason
and
they
really
are
showing
something
and
if,
if
something
important
is
coming
from
somewhere
else,
there
should
be
an
error
for
it.
E
C
E
That's
fine
too,
but
I
think,
but
either
way
I
think
it
can
be
interpreted
correctly
by
just
using
the
word
is.
B
E
A
B
A
Okay,
well,
I
don't
see
I'm
not
hearing
a
resounding;
it
must
be
changed.
I'm
not
hearing
that
a
preference
for
going
back
to,
for
example,
but
typically
is
typically
maybe
that
would
satisfy
the
the
not-white
agreement
here.
F
E
It's
like
saying
in
my
security
kernel:
I
have
a
security
policy,
that's
hardwired
and
it
was
written
by
the
manufacturer
and
then
you
could
say
another
part
of
the.
D
E
At
the
end
of
the
day,
both
policies
affect
the
security
behavior
of
the
system.
One
was
you
know,
written
by
one
entity,
the
other
by
another
entity,
but
you
can't
really.
You
know.
At
the
end
of
the
day,
the
behavior
is
one
behavior
and
we
we
presume
that
the
whatever
whatever
work
was
done
by
the
manufacturer,
who
is
okay
by
the
owner.
E
You
know
for
practical
reasons,
the
owner
can't
write.
You
know
some
of
the
code,
that's
in
the
system,
yet
he
agrees
with
it.
So.
F
So
then,
let's
say
that
the
appraisal
policy
might
be
pre-configured
into
the
relying
party
code
or
it
might
be
con
configured
by.
A
So,
do
you
want
to
say,
do
you
want
to
say,
do
you
want
to
insert
something
here
about
it
being
pre-configured
or
do
you
want
to
put
an
or.
A
Yeah
and
then
you
want
to
put
something
before
the
is
or
do
you
want
to
say
is
configured.
A
E
Or
is
configured
yeah?
That's
fine!.
A
I
can
put
or
both,
but
I
want
inclusive
or
rather
than
exclusive
or
right,
I
think,
is
the
point.
E
B
Okay,
let
me
rephrase
in
what
case
would
it
be
inclusive?
Are
you
saying
one
overwrites
the
other,
or
do
you
mean
that
it
is
a
combination?
That's.
E
B
A
A
E
B
B
A
B
B
So
on
that
one
hank
actually
asked
a
question
that
would
be
useful
for
us
to
see
whether
we
have
agreement
on.
If
you
just
go
back
one
thing
in
the
browser
in
his
comment
on
145
I
mean
you
can
see:
no,
no,
not
not
not
not
in
the
text.
In
the
pure
request
description,
he
says
also
a
background
check
version
might
be
perceived
as
missing.
Do
we
need
one
question?
B
C
B
Well,
that's
a
question:
there
are
six
theoretical
ones.
Do
we
actually
need
to
add
the
sixth
one
or
do
we
think
that
adding
the
sixth
one
would
not
add
any
other
information
I
mean?
Should
we
add
a
tiny
one
that
says
oh
and
this
one
is
the
same.
That's
the
open
question
there.
B
A
Exactly
yeah,
let's
ask
the
working
group
and
that
I
think
that's
a
good
way
to
engage
them.
C
Before
we
ask
someone,
do
we
want
to
push
a
update.
B
B
C
A
He
underscores
it
can't
do
it.
I
can
do
that
today
and
and
get
review
it,
because
that
will
be
probably
time
zone
wise.