►
From YouTube: CBOR WG Interim Meeting, 2021-04-07
Description
CBOR WG Interim Meeting, 2021-04-07
A
But
you're
recorded
and
that's
the
not
well
applied,
especially
with
that
to
each
other.
I've
posted
a
link
to
the
minutes
and
to
the
work
in
progress
minute
in
the
chat.
Would
someone
volunteer
to
add
a
few
things
in
there,
given
that
barry
is
the
notification
and
don't
want
to
join
me
in
sharing
this
today.
A
A
A
A
Otherwise,
we
put
I've
had
to
leave
out
a
casting
of
carson's
last
slot
during
the
physical.
I
guess
meeting
the
virtual
idf
meeting
that
we
have
so
I'd
like
you
to
ask
your
carson
to
basically
continue
where
we
left
you
out,
and
I
think
I
can
show
this
live
if
you
wanted,
but
it
won't
take
me
a
moment
to
have
them
online.
A
B
A
A
B
A
A
B
A
B
B
So
one
one
document
that
we
generated
at
the
time
when,
when
cd
cda
was
being
completed,
was
the
freezer
document
which
contains
various
components
that
didn't
make
it
into
8610,
but
that
potentially
could
become
useful
in
in
the
next
round
of
the
working
group
and
obviously
we
are
in
the
next
round
so
yeah.
So
this
document
hasn't
really
changed
very
much,
except
that
the
the
computed,
literals
and
dot
abnf
part
now
is
its
own
document.
The
dash
control,
a
document
cdl-control
so
that
that
is
in
good
shape.
B
So
it's
something
that
that
does
come
up
again
and
again,
but
right
now,
I'm
not
seeing
people
having
a
good
use
case,
and
I
think
it's
best
to
define
these
controls
at
a
time
when
you
actually
have
the
use
case.
So
I
think
these
are
living
well
in
the
freezer
document
and
we
can
pull
them
out
when
whenever
it
turns
out,
somebody
has
a
use
case
and,
and
we
can
turn
them
into
a
document.
B
B
B
You
don't
have
to
to
write
code
to
to
combine
the
rules,
but
on
the
other
hand,
of
course
it's
it's
harder
to
go
back
from
there
to
the
original
city
source.
So
I
think
both
representations
are
useful,
but
the
cooked
form
is
is
maybe
something
that
that's
more
useful
in
a
tool
that,
for
instance,
wants
to
convert
into
a
different
language
that
doesn't
have
the
the
slash
equals
or
slash
slash
equals
form.
B
B
B
One
is
a
brief
discussion
of
a
module
superstructure
and
I
think
most
people
who
have
been
working
on
on
larger
sets
of
ctdl,
in
particular
in
the
rats
context,
would
agree
that
we
need
to
do
something
very
soon.
That's
why
this
is
in
purple,
and
then
we
have
co-occurrence
constraints
and
literal
syntaxes
and
literal
syntaxes
is
maybe
the
the
least
needed
of
these.
So
I
think
we
are
mostly
waiting
for
strong
use
cases
there
co-occurrence
constraints
actually
in
in
review.
I
just
did
I
already
forgot:
what
was
it
acme.
C
B
B
But
I
think
more
more
examples
like
these
are
going
to
to
come
up
so
saying
something
like
here's
a
field
and
that
field
needs
to
be
a
multiple
of
that
other
field,
and
so
on.
So
yeah
again.
If,
if,
if
you
are
writing
cddl
and
find
out,
you
are
missing
something
here.
You
would
like
to
be
able
to
express
a
constraint,
but
don't
have
the
the
means
in
the
language.
B
B
C
B
B
B
There's
nothing
can
be
added
to
the
prelude
because
it
would
possibly
conflict
with
namespaces
in
existing
documents
and
that
that's
also
something
where
the
module
structure
could
help,
because
you
could
just
include
the
the
prelude
component
that
comes
from
a
particular
standard
and
to
have
that
available
in
your
city.
At
specification.
B
Yeah,
I
think
it
needs
to
be
a
mix
and
match
thing.
So
you
you
cannot
really
I
mean
you
could
do
a
city
at
2.0
and
that
has
just
a
different
prelude
and
and
if
you
bought
something
from
cdl
one
zero
to
two
zero.
You
have
to
to
clean
your
namespaces,
but
I
think
the
the
the
inverse
direction
is
probably
the
best
one
where
we
start
actually
treating
the
prelude
as
another
namespace
and
have
that
implicitly
imported.
B
So
we
we
don't
have
a
big
difference
between
the
prelude
and
other
modules
that
that
are
being
imported.
So
the
import
interface,
for
instance,
would
allow
you
to
rename
something
that
is
in
the
prelude
or
yeah,
something
like
that.
So
yeah.
Maybe
we
need
to
to
spend
a
little
time
actually
making
some
some
strawman
proposals
and
and
discuss
this
in
in
an
interim
meeting,
maybe
maybe
in
four
four
weeks
or
something
like
that.
A
Sounds
good
to
me
now
from
from
the
group,
do
you
do
you
do?
Do
you
have
documents
you've
worked
on
where
you
where
a
particular
set
of
statements
would
come.
You
took
to
mind
right
that
you
would
like
to
introduce
introduce
for
more
general
purpose
consumption
where
you
could
try
out
what
is
what
is
being.
A
So
that's
that's,
definitely
something
where
we
would
need
to
see
a
bit
more
except
we
if
this
is
to
go
on
the
working
group.
I
think
we'll
have
we'll
have
to
have
some
examples,
but
I'm
confident
that
as
we
go
along,
we
can
find
some,
especially
with
the
documents
like,
like
you
mentioned,
of
a
tank
where
this
would
be
useful.
B
C
B
C
C
A
C
B
B
B
B
Then
there
was
a
question
I
already
forgot,
who
who
put
it
in,
but
the
the
question
was:
how
do
we
handle
128-bit
entities
so
when
we
define
sibo
in
in
2013,
we
said
64-bit
is
enough
for
everyone.
No
that's,
of
course
not
what
we
said,
but
we
said
that
that
this
is
really
the
the
basic
data
model
can
stop
at
64-bit
and
if
we
need
something
larger
we
can
put
it
in
and
actually
for
integers.
B
B
B
B
The
the
other
way
to
address
this,
of
course,
is
to
open
pandora's
box
and
say:
okay,
we
we
extend
cba
to
allow
128-bit
data
in
in
the
argument.
Part
not
not
in
the
content.
Part
like
like
tag
two
three
do
but
in
the
argument
part,
but
I
think
the
the
only
data
type
that
actually
can
use
this
at
this
point
in
time
would
be
float
and
we
may
be
better
off
with
spending
another
byte
or
two
and
using
a
tag.
B
C
D
D
D
D
C
A
Would
it
would
a
tag
really
help
in
your
use
case,
given
that
you
that
you
don't
know
what,
given
that
you
are
in
a
situation
where
it's
configured,
what
the
other
party
are
supposed
to?
What
not
I
mean
you
currently
consider
the
case
where
you're
converting
to
someone
who
doesn't
support
the
array
tag
or
the
tag
for
aries,
but
if
that
that
receiving
party
then
doesn't
support,
attack
or
slope
for
128
bit
floats
either.
B
A
B
A
Would
there
be
much
more
to
to
to
this
endeavor
of
including
those
batteries
than
defining
those
two
or
those
one
or
two
tags
with
their
with
their
semantics
and
be
done
with
it
in
this?
If,
if
it's
just
about
defining
the
text,
it
may
not
even
require
any
standard
action,
but
just
someone
writing
it
up
and
then
custom
pointing
it
out
in
the
individual
text.
D
B
B
B
A
On
on
angular
net
and
on
one
of
the
earlier
slides,
my
impression
from
the
from
the
discussion
was
in
the
direction
that
yeah
I
mean
one
could
do
that,
but
interoperability
would
be
so
it
would
be
non-interoperable,
so
there's
only
so
much
you
can
do
without
breaking
the
ecosystem.
Is
there
much
more
to
that
discussion?
Is
that
something
that
we
expect
to
continue?
Is
there
other
voices
here
that
one
that
haven't
been
troubled,
maybe.
A
B
Towards
well
explaining
that
there's
very
little
text
in
in
89
49,
giving
the
rational
for
this
and,
of
course
the
the
rational
is
that
most
of
us
have
experienced
both
with
standards
that
actually
defined
a
single
endiness
and
standards
that
allowed
both
and
those
usually
are
pretty
icky.
So
I
remember
the
x
window
system
where
essentially,
the
the
two
sides
at
the
start,
try
to
find
out
whether
they
are
having
they're
using
the
same
endian
this
and
then
switch
to
that.
B
E
The
all
of
our
mars
robots
are
big
indian
and
if,
if
I
were
to
imagine
a
long-term
use
of
of
seabor,
you
know
interplanetary
craft
is
kind
of
one
of
the
places.
I
would
think
it
would
be
you
if
you
were
inventing
it
now.
You
would
do
this
and
you
would
care
about
not
having
to
swap
bites
and
as
far
as
I
can
tell
they're
all
big.
B
A
B
A
Yes,
but
I
think
the
expectation
there
was
that,
if
you,
if
you
build
your
code
or
regular,
say
regular
platform,
then
the
compiler
would
just
emit
the
byte
12
instruction.
If
it
sees
you
read
from
there
and
right
from
there
and
then
right
in
the
other
direction,
and
that's
not
an
optional
drug
assembly
sounds
a
lot
like
premature
optimization.
But
if
we
have
anything
to
point
at
in
terms
of
efficiency
that
might
help
say
yeah.
Yes,
it
does
make
a
difference,
but
not
one.
Let's
do
that,
which
I
think
is
the
line.
A
A
A
A
Basically,
with
the
words
around
is
that
it's
not
something
that
needs
to
concern
the
list,
but
at
some
point
there
will
be
at
least
an
expert
need
to
be
an
expert
opinion
on
this
in
terms
of
allocating
tags.
Would
you
like
to
say
a
bit
about
that
or
just
pointing
at
the
document
and
asking
people
to
read.
A
I
see
a
bit
in
in
terms
of
allocating
several
truth
points.
I
see
a
bit
of
overlap
with
compression
in
the
sense
that
this
enumeration
is
might
could
be
equivalent
to
a
compressed
form
of
a
longer
explanation
that
this
is
a
data
item
that
has
type
this
and
value
that
do
you
see
that
overlap
as
well,
or
is
that
just
me
pushing
too
many
things
together.
A
A
B
Okay,
so
so,
given
that
this
is
being
used
from
a
strongly
typed
environment,
the
compiler
actually
knows
what
structure
any
any
alternative
has.
So
the
the
compiler
implicitly
has
a
schema,
even
though
that
is
never
expressed
as
a
schema
in.
In
that
sense,
it's
just
part
of
the
the
original
program.
B
A
A
D
First
of
all
that
proposal
I
can
see
that
being
very
useful
to
encode
optional
values
where
one
alternative
is
null
and
the
other
is
whatever
value
or
structure
or
also
for
encoding.
What
we
call
variance
in
c,
plus,
plus
or
other
languages,
might
call
it
a
tagged
union
where
it's
a
possible
limited
set
of
data
types
that
occupies
the
same
space
in
memory,
so
those
would
be
use
cases
in
c,
plus,
plus
or
any
other
language
that
supports
optionals
or
variants.
A
But
you
you
point
out
one
one
case.
That
also
came
to
my
mind
that
that's
optionality
is
that
all
the
cases
we
would
use
this,
or
is
this
more
a
case
where
we
would
like
on
many
written
out
protocols,
there's
the
option
to
have
a
null
somewhere
or
some
particularly
type
value
for
many
for
many
nullables
of
for
many
cases
where
an
application
would
have
an
option.
A
B
A
B
B
The
draft
not
only
defines
the
sibo
tags,
but
it
also
defines
a
way
to
specify
these
things
in
cdl,
so
we
should
think
about
whether
that's
the
way
we
want
it
and
in
the
discussion
on
the
mailing
list.
I
also
pointed
to
an
older
male
about
a
year
old,
that
kind
of
opened
up
the
whole
space
on
the
representation
side.
B
So
the
the
discussion
of
that
is
really
old,
so
so
the
first
applications
of
cdl
excuse
me
of
sibo
for
for
management
in
the
constraint
space
ran
into
that
before
we
we
finally
decided.
We,
we
just
wanted
to
have
a
young
representation
in
in
sibo,
so
we
we
kind
of
lost
that
fell
out
of
our
collective
consciousness.