►
From YouTube: IETF-CBOR-20230111-1500
Description
CBOR meeting session at IETF
2023/01/11 1500
https://datatracker.ietf.org/meeting//proceedings/
A
A
For
the
screen,
I
would
I
would
grant
that
as
soon
as
we
are
done
with
the
startup
stuff,
which,
given
that
it's
four
o'clock
at
least
in
my
time
zone,
and
that
it's
a
new
year
in
which
we
can
do
away
with
old
habits,
we
might
just
as
well
start
right
on
the
top
of
the
hour
now
hello
and
welcome
everybody
to
this
first
interim
of
the
sibo
working
group
in
this
new
year.
My
name
is
Kristen
Andrews,
you're,
probably
all
familiar
with
how
this
works.
B
A
That's
how
we
work
as
usual.
The
agenda
is
not
really
on
the
slides
or
in
the
agenda
field,
but
in
the
notes.
So
I
would
like
to
ask
you
to
just
switch
your
screens
over
to
the
to
the
to
the
minute
taking
tool
and
invite
you
to
contribute
to
the
minutes
for
anything
that
Marco
will
not
catch
or
just
to
make
sure
that
this
is
all
a
collaborative
thing
with
that.
C
A
There
are
no
new
updates
on
on
PSL
receiver.
I've
had
a
few
words
with
me
with
Martin.
Previously,
this
is
mainly
blocked
on
the
general
need
to
to
do
stuff
on
the
DNS
side.
So
there
is
no
current
input
that
this
needs
from
from
the
working
group,
either
on
cdtl
2.0
custom.
Do
you
want
to
to
summarize
what.
C
You've
wrote
so
we
have
a
document,
we
don't
have
an
implementation
I'm,
actually
writing
the
implementation
or
an
implementation.
That
already
does
interesting
things,
unfortunately,
not
exactly
the
things
I
wanted
to
do
so
it
will
probably
take
another
week
or
so
and
until
we
can
start
playing
with
it
and
starting
building
an
opinion
on
whether
this
is
the
right
direction.
A
I'm
looking
forward
to
the
playing
stuff,
thank
you
and
then
the
large
agenda
point
that
we
will
actually
talk
about
today
is
the
zebra
object,
type
extension
that
unders
is
proposing.
There
is
one
more
point
on
the
agenda,
but
as
I
understand
the
slides
that
Carson
proposed,
they
are
basically
his
comments
on
cotx
or.
C
A
A
A
A
D
B
B
D
B
D
C
D
A
A
A
C
C
C
So
in
2013,
when
when
we
we
started
to
work
on
sibo,
we
thought
that
we
would
need
to
do
two
main
steps
to
improve
on
Json.
One
was
embracing
binary
in
various
Ways
by
by
having
byte
strings
by
having
binary,
encodings
and
so
on,
so
that
that
was
a
big
step,
of
course,
away
from
the
text-based
Json
protocol
and
the
other
one
was
that
we
really
wanted
to
embrace
extensibility
or
evolution,
and
that
is
the
domain
of
the
semantic
tagging.
C
So
that's
why
why
we
actually
put
that
in
so
this
has
been
a
great
concern
in
the
Civil
Community
from
the
start
and
let's
dive
into
that
so
c.
Bar
tags
are
essentially
the
main
extension
point
for
sibo.
There
is
another
one,
the
the
simple
data
items,
but
this
extension
Point
has
been
used
very
sparingly,
so
people
usually
use
tags
as
the
extension
point
and
we
have
a
basic
generic
data
model
which
is
essentially
Json
plus
binary.
C
So
how
does
this
tag
mechanism
work?
First
of
all
to
to
get
a
new
tag.
We
decided
we
want
to
go
through
an
Ina
registry,
and
we
wanted
to
do
this
because
we
wanted
to
have
a
simple
unsigned
number,
a
big,
potentially
big
number.
So
we
we
can
accommodate
a
64-bit
numbers,
but
numbers
are
relatively
easy
to
process
are
pretty
efficient
in
the
representation,
so
this
is
useful
for
the
the
main
target
application.
The
concise
binary
object,
representation
where
we
care
about
size
and
efficiency
of
ingest
and
and
generation.
C
So
this
this
tag,
namespace
is
global.
That's
the
other
big
decision.
So
if
you
compare
this
with
A's
and
ones
tag,
spaces
Ace
and
one
also
has
tags,
but
these
are
not
64-bit
numbers.
These
are
essentially
five
plus
two
bit
numbers,
so
you're
a
bit
constrained
in
what
you
can
do
there
anyway,
Acer
one
actually
has
Universal
tags
which
actually
are
quite
similar
to
sibo
tanks,
except
that
in
sn1.
Everything
has
a
tag.
C
A7
also
has
application
tags
where
the
the
number
space
is
local
to
the
asn1
module
that
you
are
looking
at.
So
you,
you
really
need
to
have
the
the
module
at
hand
to
understand
this
tag.
We
didn't
want
to
do
that.
We
wanted
to
make
sure
that
sibo
data
items
can
be
interpreted
even
in
the
absence
of
of
a
schema
document,
and
then
there
are
private
tags
which
I
haven't
seen
in
in
actual
use,
because,
probably
because
they
are
a
bad
idea,
so
we
didn't
copy
them.
C
C
D
D
No,
do
you
hear
me
now?
Yes,
yes,
okay,
good,
okay,
it's
a
little
bit
complex
here
with
mobile
phone.
Well,
what
you
said
is,
of
course,
completely
right.
There
is
no
objections
to
that
and
I
understand
the
tag
concept
pretty
well,
since
I
have
done
an
Seabury
implementation
that
uses
it
as
well,
but
this
is
sort
of
a
we
could
say,
a
provision
for
people
who
are
working
with
high
level
systems,
where
you
have
lots
of
objects
that
are
constantly
revised.
D
That
means
that
you
will
have
many
identifiers,
and
then
it
becomes
slightly
less
attempting
to
use
Iona
for
every
new
identifier
but
having
a
root
somewhere,
and
that
is
essentially
what
url
url
based
identifier
gives
you.
You
have
a
route
that
is
a
typically
a
host
name,
and
then
you,
then
you
can
create
your
own
namespace
there.
There
are
some
drawbacks
with
this,
but
I
think
that,
with
let's
say,
GitHub,
which
I
use.
Currently
you
get
a
stable
point
for
having
your
own
namespace.
D
Of
course,
you
can
lose
the
contact
with
your
account
it's
possible,
but
that
is
established
and
and
ISO
they
have
their
own
orn
standard
or
you
are
an
identifier
which
is
their
Roots,
so
IT
addresses
this
kind
of
application,
which
is
or
slightly
different
than
the
origins
of
cboard
I.
Don't
see
it
as
a
sort
of
a
deprecation
of
anything,
it
is
more
or
less
to
make
things
easier
for
those
who
work
with
high-level
applications.
I
do
that
and
I
have
lots
of
objects
and
I.
D
C
D
The
last
page
is
actually
rather
interesting.
There
is
already
a
predecessor,
an
ITF
thing
where
you
actually
do
not
use
a
seaboard
tag
but
and
a
URL,
and
it
could
easily
be
replaced
by
an
empty
Seaboard
tag
with
a
null
argument,
but
they
selected
a
URL
anyway.
So
if,
if
my
proposal
is
bad,
you
could
say,
then
the
the
other
proposal
is
bad
as
well,
because
they
essentially
build
on
the
same
ID
having
URLs
and
do
dereferencing
for
documentation.
C
B
D
Have
a
profile
URL,
which
I
think
is
similar
to
an
object
identifier,
and
they
say
that
it
typically
is
a
URL
and
you
do
dereferencing
and
it
could
definitely
be
replaced
by
a
seaboard
tag.
Iona
registers
with
a
null
argument.
It
would
be
that
would
be
sort
of
the
C
more
like
solution
for
that
problem,
but
they
have
selected
a
URL
and
yeah
I
think
they
have
done
it
for
the
same
reason
as
as
me.
D
D
B
C
C
D
C
B
D
B
D
C
B
C
We
are
yes,
so
the
the
tag
allocation
mechanism
is
structured
into
different
spaces.
Different
ranges-
and
here
is
today's
usage
report
so
of
the
the
first
range
that
has
24
potential
values.
13
are
used
and
11
are
free
of
the
second
range
that
their
30
are
used
of.
The
third
range,
a
very
small
number
I
used,
which
is
essentially
telling
us,
if
only
so
few,
are
used
after
10
years,
then
we
probably
have
the
ability
to
put
in
a
new
tanks.
C
C
But
the
point
is
that,
as
I
said,
OnePlus
2
is
is
a
pretty
empty
right
now.
So
this
is
first
campus
served
as
the
the
higher
ones
and
the
the
bigger
ones
and
I
think
that's
really
important
to
make
this
take
a
location
mechanism
available
for
for
people
who
want
to
get
some
work
done,
so
they
can
get
a
response
from
Ayana
quickly.
C
This
is
about
as
easy
as
getting
a
PIN
number
or
something
like
that,
while
the
the
lower
half
of
one
plus
two
and
one
plus
zero
and
one
plus
one
are
curated
so
that
there
is,
there
are
different
people
looking
at
this
and
trying
to
make
sure
that
that
we
don't
expand
them
too
early
at
a
limited
rate.
So
let's
just
look
at
a
few
other
places
and
I
certainly
will
add
e
to
that
when
I
make
a
new
version
of
the
slide
deck.
C
Let's
look
at
a
few
cases,
so
Json
LD
is
an
interesting
case
because
they
are
essentially
trying
to
do
something
like
semantic
taking
in
the
Json
space
and
of
course,
Jason
does
not
have
extensibility,
so
they
have
to
essentially
steal
a
specific
case,
Json
object
or
map
that
has
a
key
called
ad
type
yeah.
So
that
is
used
for
for
carrying
the
type
information.
C
And
then
there
are
additional
things
like
context.
Information
I
think
we
just
lost
under.
So
this
is
one
one
way
of
doing
things
doing
this
in
a
non-transparent
way.
So
you
cannot
use
ad
type
for
for
anything
else.
That
might
have
wanted
to
use
this
map
key.
When,
when
a
type
is
used
anywhere
in
your
Json
data,
then
then
you
are
making
a
statement
about
the
the
type
of
something
this
has
Uris
as
namespace
and
has
all
the
typical
problems
of
the
referenceable
identifiers
in
the
nodes.
C
C
C
So
I'm
not
going
to
try
to
be
too
precise
about
this,
because
it's
changing
as
we
speak,
but
this
doesn't
try
to
be
General
but
provide
one
specific
data
structure
that
problem
details,
data
structure
and
that's
a
map
or
a
Json
object
and
has
a
certain
predefined
keys,
but
also
allows
application
specific
keys
and
it
uses
a
URI
reference
in
the
type
key.
So
this
is
again
a
place
where
uis
are
being
used
and
7807
Beast
has
changed
this
a
little
bit
by
actually
defining
a
registry.
C
C
In
call,
we
try
to
come
up
with
a
concise
version
of
this
problem
detailed,
so
we
can
have
problem
details
in
a
few
bytes,
and
here
we
replaced
your
eyes
again
with
an
Ina
registry
with
integers,
but
also
providing
a
UI
as
a
fallback
for
for
a
worst
case.
So
I
think
that
that's
also
interesting
yeah
and
the
the
under
draft
is
essentially
another
entry
in
in
this
picture,
and
this
is
interesting
because
it
actually
says
it
uses
text
strings.
C
It
only
hints
that
these
text
strings
might
might
usefully
be
uis,
but
it
also
allows
other
text
strings
there,
and
that,
of
course
has
the
advantage
that
it's
easy
to
come
up
with
text
strings.
It's
even
easy
to
come
up
with
your
eyes,
it's
less
clear
how
how
you
can
do
a
reasonable
collision
avoidance
regime
on
these,
so
for
Uris
that
that
requires
some
confidence
that
the
the
owner
of
the
domain
name
is
not
going
to
change
the
meaning
of
that
UI
and
so
on.
C
So
it's
easy
to
to
break
this.
If
you
don't
think
about
these
things,
there's
a
tag
UI
scheme
that
that
is
actually
good
for
for
getting
better
collision
avoidance
here
and
yeah.
The
the
difference
between
this
and
the
c18
system
is
that
the
native
Tech
system
comes
with
sibo
is
well
implemented.
It
has
an
easy
registration
mechanism
and
a
working
collision
avoidance
mechanism
and
yeah
the
disadvantages.
C
You
can't
just
make
up
text
strings
so
yeah,
looking
at
the
coach
as
tag
registration
request,
because
it
actually
uses
the
sibo
tank
system
to
to
put
itself
up
in
the
Civil
type
system.
So
that's
the
right
thing
to
do.
Of
course,
this
is
a
valid
registration
request,
but
but
I
haven't
heard
anybody
question
that
so
the
registry
can
handle
this
registration
request.
C
There
are
two
designated
experts
who
both
are
in
this
meeting,
but
I'm
speculating
here
because
I'm
only
one
of
them
in
the
past,
we
have
accepted
internet
drafts
as
documents
for
the
specification
required
requirement
which
we
have
in
in
the
tags
under
32,
000
and
yeah.
I
think
that
that
is
obvious
here
as
well,
and
of
course
it's
nice
if
that
internet
draft
becomes
an
RFC
at
some
point.
But
it's
not
a
prerequisite.
C
D
All
right,
well,
I,
of
course,
I
can,
if
it
looks
like
I,
am
trying
to
replace
native
taxi.
It
is
an
addition
to
the
Sea
World
System,
it's
not
a
replacement,
and
it
is
just
to
align
it
with
other
things
like
verified
credentials,
ISO
to
add
20,
022
and
similar
systems
that
that
currently
use
URLs,
Uris
or
URLs.
That
is
what
the
purpose
is
to
to
not
force
these
people
into
something
else,
even
if
this
something
else
is
better.
D
D
Well,
I,
don't
know
really
I
have
seen
this
eat
standard
where,
where
you
use
URLs
instead
of
tags,
and
so
this
is
the
only
thing
I
know
of
that,
you
actually
could
fool
the
system.
I
I,
don't
know
really
I
aren't
system
based
on
tables
where
you
have
sort
of
known
identifiers,
regardless
of
what
the
typing
system
you
use,
you
don't
accept
any
kind,
all
kinds
of
objects,
probably
a
set
of
objects,
and
if
somebody
generates
it
comes
with
an
object
with
a
tag,
you
don't
recognize
it
to.
D
A
B
We
have
alternate
types
that
can
be
chosen
so,
for
example,
match
thing
where
you're
you're,
you
have
multiple
ways
to
match
a
string
and
the
tag
system
like
feels
totally
appropriate
for
that,
but
the
scope
of
those
tags
would
probably
be
just
ins
just
just
dedicated
to
this
particular
context.
Right.
Just
this,
this
match
operation,
and
so
rather
than
rather
than
registering
a
tag
because
tags
are
Universal,
we
do
it.
B
A
Thanks
for
that,
are
you
aware
of
the
unfortunately
I?
Don't
know
the
precise
name
of
my
of
my
head
of
the
draft
that
has
been
around
of
The
Proposal
that
has
been
around
for
tags
that
this
thing
that
are
that
could
be
used
for
for
enums,
because
they
these
would
register
precisely
those
tags
where
you
have
a
small
number
of
known
by
the
context,
options
that
an
item
can
have
I
can
look
up
the
the
precise
name
and
add
it
to
the
minutes
later.
D
A
Add
on
add
to
the
mailing
list
thread
comments
on
on
particulars,
where
it
may
be
good
to
adjust
the
text
and
in
parallel
there
is
nothing
that
would
stop
any.
That
would
stop
at
any
point
in
time.
A
request
for
registration
here,
yeah
I,
think
I
think
this
can
go
this.
Can
this
this
should
be
able
to
progress
in
a
in
a
reasonable
time
frame.
A
A
The
registration
will
be
handled
basically
by
you
sending
an
email
to
to
if,
if
I'm
not
mistaken,
that
registration
would
happen
by
you
sending
an
email
to
Ayana
I'm
asking
for
the
assignment
of
a
tech
of
a
tag,
and
if
you,
if
you
have
a
suggestion
for
a
particular
tag,
preferably
in
the
in
the
OnePlus
2
range,
because
otherwise
I
think
this
would
be
excessive
on
on
which
applications
that
might
make
better
use
of
the
OnePlus
One
right,
then
Ayanna
will
forward
this
request
to
Carson
and
me.
Okay,.