►
From YouTube: IETF110-CBOR-20210308-1430
Description
CBOR meeting session at IETF110
2021/03/08 1430
https://datatracker.ietf.org/meeting/110/proceedings/
A
And
that
also
brings
us
to
the
official
start
of
this
meeting
with
people
having
come
in
welcome
to
the
zebra
meeting
of
the
online
itf
110.
A
This
is
an
itf
meeting.
That
means
that
the
node
well
applies
both
here
and
in
the
accompanying
venues.
If
there's
anything
unclear
about
this,
please
please
ask
about
it
or
just
re
read
through
it,
and
especially
with
ipr,
con
stuff,
just
ask
about
it
before
we
get
into
a
situation
that
might
otherwise
be
difficult.
A
Yeah
this
be,
this
meeting
will
be
recorded.
We
have
the
jabber
session
on
on
your
left
side.
We
will
use
a
queue.
So
if
you
have
something
to
comment
on
just
please
use
the
hand
button.
It
would
be
two
slides
further
along,
I
think,
and
if
we
when,
when
later
we
are
in
the
discussion
part,
please
feel
free
to
just
unmute
your
audio.
If
you're,
jumping
back
and
forth
between
discussion
points,
one
slide
back,
please,
as
as
as
for
today,
we
have
I'm
confusing
myself.
A
Sorry
as
for
today,
this
is
the
last
meeting
where
francesca
is
joining
us
as
a
working
group
chair,
given
that
she
is
now
taking
on
the
role
of
the
area
director,
thanks
for
working
with
me,
and
congratulations
on
the
ad
role
and
thanks
for
staying
with
us
in
that
same
role.
Speaking
of
staying
with
us
she's
taking
the
pres
she's,
taking
the
position
from
barry.
A
A
That
is
the
time,
duration
time,
duration
period
time
period
from
carson.
Then
keo
will
tell
us
about
few
things
about
how
we
might
do
ordered
magnc
maps
in
sieber
and
discuss.
We
can
discuss
especially
the
topic
of
the
relevant
tech
lengths,
and
that
will
also
be
the
introduction
point
to
the
to
other
notable
texts
that
custom
will
talk
about
briefly.
A
That's
the
28
ai
28
thing,
I
suppose
yep
it's
already
on
the
extended
agenda
yep.
Thank
you.
A
So
if
there's
no
more,
no
more
comments,
I'll
jump
right
into
the
topic
of
oid,
this
document
has
recently
passed
the
working
group
last
call
and
I've
completed
the
shepherd
write-up.
That
francesca
has
been
as
kind
as
to
largely
prepare,
as
the
working
group
blast
call
completed,
without
any
any
objections
and
with
good
support
and
reviews.
A
There
is
one
potential
issue
that
we'll
we're
still
looking
out
at.
That
is
that
john
lennon
hasn't
replied
to
the
ipr
question
yet,
but
there's
still
ample
time
until
this
really
needs
to
be
done
completely,
and
until
then
I
mean-
and
in
general
I
don't
expect
that
there
will
be
large
complications.
It's
more
his
more
procedural
matters.
A
If
there
is
no
other
question,
if
there's
no
other
points
on
that
francesca,
please
please
go
to
carson's
slide
and
we
can
start
with
the
topic
of
cgi
control,
carson.
C
So
yeah
this
is
an
old
acquaintance
which
has
been
around
for
a
while.
It
does
three
things:
it
defines
some
some
computed
literals
using
dot
cat
and
dot
plus
it
defines
how
to
use
abn
f
in
a
cdl
model,
and
that
defines
that
feature-
and
we
were
kind
of
done
with
this,
but
we
hadn't
got
any
feedback
from
implementation
experience
so
so
I
finally
broke
down
and
actually
generated
an
implementation.
C
C
Maybe
it
will
take
too
long
to
to
update
this
internet
standard,
so
we
decided
to
instead
put
something
into
the
control
operator
draft,
and
now
we
have
beside
the
cat,
a
denting
cat,
which
was
briefly
called
badge,
but
then
it
was
called
dead
cat
and
then
people
said
they.
They
would
prefer
a
vegetarian
option
here,
so
we
didn't
use
dead
cat
but
abbreviated.
This
2.,
which,
by
the
way
is,
is
a
comic
figure
that
people
of
my
age
are
rather
familiar
in
in
germany
anyway.
C
So
with
this
little
edition,
it's
it's
actually
possible
to
to
write
the
live
cat-
yeah,
that's
still
not
vegetarian,
so
I
know
so
with
that.
It's
actually
possible
to
have
the
examples
that
that
have
been
in
the
document
before
but
didn't
compile
with
the
implementation,
and
I
think
we
are
essentially
done
next
slide.
Please.
C
So
the
the
implementation
status,
as
I
said,
is
the
entire
draft
is
implemented
in
the
cdl
tool
about
half
of
it
is
implemented
in
the
tool
from
andrew
weiss,
but
I
think
he
hasn't
had
much
time
to
actually
follow
through
on
his
promise
to
to
put
in
a
bnf
there
as
well,
and
I
took
the
the
that
proposal
and
generated
a
dash
of
three
before
we
got
a
lot
of
discussion
in
the
group.
C
So
we
can
back
this
out
if
we
don't
like
that,
but
I
think
it's
good
to
have
one
document
that
looks
like
what
what
we
would
actually
send
to
the
isg.
So
the
the
next
step
is
either
to
to
get
pre-working
blast
call
reviews
or
to
do
a
working
last
call
to
to
finish
this
document.
C
A
Just
to
get
a
general
feeling
how
many
here
have
read,
have
read
and
read
or
implemented
the
the
an
earlier
version
of
this
document,
given
that
dash
o3,
israel
is
quite
new
kind
of
showing
showing
hands
briefly
will
help.
A
D
People,
how
do
we
show
hands?
I
think
you
can
just
join
the
queue.
Oh
yeah
yeah.
We
don't
need
the
whole
tool.
A
Yeah,
okay,
I'd
like
to
have
a
bit
more
feedback
on
the
mailing
list
as
well
on
this,
because
that's
so
far
so
far,
that's
not
terribly
many
people,
but
generally
the
I
don't.
I
don't
hear
any
loud
complaints
either.
So
unless
something
comes
up
on
the
mailing
list,
I
think
that
this
is
a
what
you
propose
sounds
like
a
way
we
can.
We
can
continue
on.
D
C
D
A
C
Yes,
next
slide,
please
so
we
we
had
a
couple
of
discussions
in
interims
that
led
us
to
add
suffix
packing.
So
far
we
only
had
prefix
packing
and
full
shared
item
packing
and
we
added
an
example,
the
example
from
2017
when
this
actually
was
written
up
for
the
first
time
and
what's
missing
right
now
is
the
small
details
on
the
setup
side.
C
So
we
know
how
to
decompress
or
unpack
a
packed
zebra
data
item
if
we
have
set
up
tables,
but
we
don't
have
the
whole
gamut
of
possible
approaches
for
table
setup
yet
so
the
basic
idea
was
to
to
do
a
basic
version
of
that
and
provide
extensibility
points,
so
people
can
add
their
way
of
doing
this
if
needed.
C
C
So
right
now,
I'm
I'm.
I
don't
really
have
a
very
crisp
use
case,
but
we,
we
certainly
could
add
tags
like
that
later.
C
So
we
are
now
in
the
familiar
position
that
we
we
have.
A
draft
that
already
mostly
works,
maybe
needs
a
little
bit
more
details
on
the
table
setup,
but
we
need
to
get
more
experience
with
using
it.
C
I
have
an
implementation
from
the
2017
sprint,
but
that
really
needs
to
be
adapted
to
what
we
have
now
and
probably
need
some
some
further
implementation
work.
So
we
we
are
in
the
same
position.
We
were
with
the
control
draft
months
ago
and
also
today
it
became
in
the
dispatch
group.
It
became
obvious
to
me
that
we
have
to
have
some
additions
for
streaming
packing.
C
So
how
do
you
run
a
streaming
application
where,
while
you
are
streaming
a
sibo
data
item,
you
notice
that
there
is
something
that
comes
up
repeatedly
and
would
need
a
table
entry.
I
think
that
can
be
added
later,
but
that's
certainly
something
to
look
at
in
on
the
implementation
side.
C
C
A
A
brief
procedural
question
other
than
other
than
suit,
who
have
done
something
different
now
anyway,
is
there
anything
that
puts
the
timeline
on
this.
C
My
point
of
view
is
that
there
are
a
lot
of
json
based
protocols
that
could
easily
convert
it
to
sieber
and
and
get
significant
improvements
from
that.
If
we
had
something
like
packed
out
there.
So
to
me,
it's
really
as
soon
as
possible
and
not
wait
for
a
really
strong
use
case
to
come
up.
E
Yeah,
that's
true
so
especially
coming
was
working
on
sebor
based
protocols
and
some
emerging
ones.
That
would
be
a
that
is
the
benefit
here.
I
think
the
the
story
we
can
tell
is
when
someone
is
coming
from
a
string
latin
protocol
and
these
strings
tend
to
be
repetitive
to
some
extent
this.
E
This
the
procedures
here
in
fact,
are
basically
the
solution
to
then
really
make
use,
not
only
in
the
scaffolding
of
sibo,
that
is
more
concise,
but
also
the
values
that
you
inherit
from
the
old
world,
and
I
think
that
that's
that's
really
really
addressing
this
issue
here.
So
that's
why
I
think
carson
is
right
with
saying
we
don't
need
a
lot
of
proof
this.
This
makes
a
lot
of
sense
by
itself.
A
Christian
with
my
chair
head
off,
if
we
look
in
to
make
making
this
useful
for
for
js
for
documents
coming
from
jason,
I
think
that
there
will
be
a
case
where,
where
there's
a
temptation
to
recreate
the
original
json
by
making
kind
of
by
default,
deferring
things
to
the
seaboard
packing,
which
is
to
keep
string
identifiers
where
we
would
regularly
use
integer
identifiers
and
then
let
the
zebra
packing
do
its
do
its
magic.
A
I
don't
have
a
clear
opinion
on
on
whether
that's
a
good
thing
or
a
bad
thing,
or
we
should
encourage
this
or
not,
but
we
should
be
aware
of
this.
F
I
know
that
suit
has
always
been
already
been
referenced
as
having
done
something
different,
but
I
I
just
want
to
say
again:
I
really
wish
that
this
had
been
available
when
suit
was
drafted,
because
it
would
have
made
life
much
easier.
F
C
To
add
to
this
yeah,
I'm
aware
of
this,
this
little
danger
of
people
using
their
json
based
protocols,
unchanged
instead
of
adapting
them
to
sibo,
but
on
the
other
end,
maybe
that's
exactly
what
we
want
them
to
be
able
to
do
so.
It
really
depends
on
on
whether
your
working
environment
is
really
so
so
constrained
that
you
you,
you
need
to
take
out
the
last
byte.
Then
probably
you
want
to
have
integer
labels,
but
if
what
zebra
pack
can
do
for
you
is
is
good
enough.
C
A
Fair
point:
thanks
for
thank
you
who
is
going
to
implement
this,
who
could
write
a.
A
D
And
brandon
great
anybody
who
is
planning
to
implement
this.
D
D
A
With
that,
we
are
pretty
much
pretty
much
on
time
to
proceed
on
to
the
documents
that
are
new
to
the
working
group
or
newly
adopted
being
file
magic.
Michael,
please
go
ahead.
G
Hi,
I'm
michael
richardson,
so
I
these
were
adopted
last
week.
I
guess
after
a
presentation
in
one
of
the
virtual
interims
in
february
I
came
across
this
as
a
as
a
need,
as
I
was
current,
converting
a
bespoke
protocol.
Well,
next
slide,
please!
Actually
it's
all
on
the
slide.
G
So
I
was
converting
some
bespoke
protocol
that
involves
passing
c
structures
across
a
thing
across
an
ipc
and
also
storing
them
on
disk
as
part
of
a
unit
test
case
system
and
I'm
converting
it
to
cbore,
and
I
kind
of
really
wanted
to
know
what
they
were,
and
it
also
came
out
of
the
discussion
that
we
were
having
in
cosi
about
the
various
different
kind
of
containers
for
crl,
like
things
or
public
keys
or
private
keys.
G
And
my
observation
is
not
the
peak
space,
it's
a
bit
of
a
no
one's,
quite
sure
what
extensions
to
use
for
what
files
and
a
lot
of
the
time
you
don't
necessarily
get
a
clear
idea
of
what's
going
on,
and
you
don't
necessarily
want
to
leave
your
your
private
keys
world
readable.
So
you
can
find
out
that
they
are,
in
fact
the
private
keys.
So
this
is
where
it
came
out
of
next
slide.
G
Please,
and
actually
the
other
other
part
of
this
was
that
I
realized
that,
as
you
have
pieces,
that
you
are
constructing
and
a
source
code
tree
and
you
may
be
incorporating
a
zip
or
into
your
source
code
as
an
object
that
you're
going
to
send
that
you
may
also
like
to
know
in
your
source
code.
What
is
what
without
actually
necessarily
wanting
to
transmit
it,
and
so
this
was
was,
was
evolved.
There's
a
discussion
on
the
list
about
this
and
a
seabor
sequence
was
was
suggested
next
slide.
G
Please,
and
so
I
wrote
the
first
version
with
a
seabor
sequence.
The
discussion
subsequent
during
the
adoption
call
was
that
we
would
like
to
have
a
zebra
sequence
and
a
straight
c
board
tag,
and
so
the
document
now
does
that
next
slide,
please.
G
So
in
the
case
where
you
have
a
keyboard
tag
and
you
wrap
your
object,
then
you
start
with
the
keyboard
tag
55799,
which
magic
our
file
already
knows
as
cbore,
and
then
you
allocate
an
a
four
byte
tag.
Number
from
the
first
come
first
served
area
and
you
put
that
afterwards
and
then
you
wrap
your
seaboard,
and
so
this
one.
This
here
is
the
one
that
I
allocated
to
my
project,
which
was
open
swan.
G
So
if
you
decode
this
is
in
the
file,
it
says
opsn
in
in
the
file
next
slide,
please.
G
So
the
other
version
of
this
is
that
you
have
a
seabor
sequence,
and
so
we
need
a
new
tag.
The
proposal
is
five:
five,
eight
zero
zero,
so
that's
one
above
the
previous
one
and
it
would
seem
to
have
all
the
properties
that
we
needed
to
have
or
suspect
to
unicode
utf-8,
utf,
16
type
things,
and
in
this
case,
with
your
cbor
sequence,
you
need
to
wrap
the
tags
around
something,
and
so
you
wrap
it
around.
G
The
byte
string
bor,
which
happens
to
have
three
characters,
long,
which
happens
to
encode
as
a
c,
and
so
the
bytes
9
through
12,
or
I
guess,
8
through
11
of
your
file
counting
from
0,
are
actually
the
letter
c
b,
o
r.
So
if
you
dump
the
file,
you
will
see
the
word
seabor
in
it,
which
may
suggest
to
you
that
if
you
have
some
unknown
thing
that
maybe
applying
a
seaboard
decoder
would
be
a
good
idea.
G
This
is
a
seabor
sequence
and
it's
12
bytes
long,
and
so,
if
you
don't
wish
to
transmit
this
over
your
network
protocol
after
having
read
it
from
the
file,
then
you
advance
12
bytes,
it's
always
12
bytes.
You
can
mem,
compare
it.
If
you
want
and
then
you
start
there
and
then
you
have
your
object
as
it
is.
So
this
is
the
two
different
versions
next
slide,
please,
so
you
get
what
you
want.
G
I
created
for
my
thing:
a
file
file
output
that
did
the
right
thing
and
it
worked
great.
I
actually
delved
into
the
file
programs
seaboard
decoding
logic
and
it's
a
little
more
complicated
than
I
think
it
ought
to
be.
But
I
don't
really
not
really
our
concern
next
slide.
I
think
that's
the
last
slide
yeah.
G
So
I
guess
the
one
of
the
questions
is
whether
or
not
we
should
allocate
this
55800
as
a
working
group
early
allocation
or
whether
I
should
simply.
I
ask
the
iana
to
pointing
at
this
document,
to
please
go
through
the
right
review
and
allocate
it
since
it's
in
a
space
that
could
go
either
way.
I
think
it
would
be
cleaner.
It
was
an
early
working
group
earlier
allocation,
but
it
could
go
any
either
way.
It
doesn't
matter.
A
Working
group
earlier
location
sounds
like
something
that,
in
my
opinion,
we
should
do
it
more
often
anyway.
So
that
would
be
a
good
starting
point
comments
from
carson.
C
Please
yeah:
we
cannot
do
early
allocation
here
because
you
only
can
do
earlier
location
on
things
that
would
ietf
would
need
ietf
review
to
to
be
allocated,
but
in
this
case
it's
an
fcfs
base.
So
we
just
just
go
ahead
and
request
that
allocation
for
this
draft,
like
today,
yeah
okay,
I'll,
do
that
at
the
end
of
the
working
group.
C
So
let
me
add
that
I
didn't
think
much
of
of
this
the
idea
of
generating
this
draft,
but
I
I
really
like
what
we
have
now
and
I
think
that
that's
just
a
very,
very
small
piece
of
really
best
current
practice
that
we
could
use
in
all
of
our
little
sibo
protocols.
G
A
If
there's
no
one,
I
I
do
have
one
one
question
that
I
that
I
maybe
didn't
just
get
precisely
from
your
presentation,
is
there
anything
to
be
really
decided
between
the
proposed
hybrid
solutions
or
is
this?
Is
this
just
both.
G
Are
possible
yeah
both
are
proposed.
That
was
the
discussion
during
the
adoption
was
that
people
felt
they
wanted
both
they
were
like
well.
Why
are
you
using
a
sequence
when
I,
when
I
just
want
to
tag
and
my
my
content,
is
what
fought?
What
originally
for
why?
I
thought
that
was
a
good
solution
that
was
proposed
in
earlier
discussion
was
that
it
allows
you
to
tag
it
sanely
in
the
file
without
actually
making
it
mandatory
that
you
are
transmitting
it
across
the
wire.
G
So,
while
removing
you
know,
eight
bytes
of
tag
is
easy
to
do.
If
you
speak
seabore,
if
actually,
what
you're
doing
is
you
have
a
somewhat
cbor
ignorant
program,
who's,
simply
loading,
something
and
transmitting
it,
because
it's
let's
call
it
a
cbor
certificate
or
something
else
that
goes
in
another
protocol
or
is
being
converted
into
a
byte
string
for
another
sieber
protocol.
Then
maybe
you
don't
actually
want
to
open
that
up.
G
Maybe
there
are
also
signature
issues
about
what
you're
what
you've
signed
there,
and
so
the
reasonable
thing
would
be
that
the
seabor
sequence
would
be
something
that's
easy
to
cleave
off.
You
just
take
off
12
bytes
and
you
know
exactly
what
they
are
and
if
they
don't
match,
then
it's
the
wrong
file.
G
Okay,
so
the
other
document
again,
while
converting
things
I
needed
to
have
ip
addresses
in
my
process,
and
I
support
v4
and
v6
and
I
thought
well,
I
should
have
a
tag
for
that
in
the
grass
protocol,
which
is
not
yet
an
rfc,
but
any
hour
now
had
the
same
problem
and
it
was
too
bad.
We
didn't
do
that.
Think
of
doing
that
at
the
time
there
so
next
slide.
G
Thank
you
so
tag
260
and
261
is
in
the
iana
registry,
and
I
thought
oh,
there
must
be
something
there
and
I
looked
at
this
and
I
was
kind
of
dissatisfied
with
this
for
a
couple
of
reasons.
I
didn't
like
that.
It
also
supported
ethernet
that
it
implied
the
version
number
by
the
size
of
the
the
thing
which
I
thought
was
potentially
quite
wasteful.
G
If
you
are,
you
know
transmitting
slash,
32,
prefixes
or
something
like
this
and
and
then
I
read
the
261,
and
I
didn't
understand
the
description
at
all
as
to
how
to
make
this
into
a
prefix
at
all.
So
both
of
these
were
allocated
by.
I
guess
document
required
and
the
document
is
a
is
a
wikipedia.
A
a
wiki
entry
in
github,
so
I
just
thought
it
wasn't
well
enough
to
document
a
261.
I
I
didn't
like
it,
and
so
I
needed
prefixes
as
well
as
stuff.
G
So
these
are
essentially
the
proposal.
That's
there
there
will
be
two
tags,
one
for
v6
and
one
for
v4
and
the
one
for
v6
has
a
byte
string
array
and
the
prefix
has
an
array
containing
an
integer
which
is
the
prefix
length
and
then
a
byte
string
with
the
the
with
the
as
much
as
many
bytes
as
the
prefix
as
you
care
about
in
both
cases,
trailing
zeros
may
be
omitted,
so
you
may
omit
as
many
zeros
on
the
right
as
you
like,
and
for
prefixes.
G
This
is
typically
many
so
often
for
a
64-bit
prefix.
You
will
omit
at
least
eight
bytes
to
the
right,
because
they're
considered
zero,
and
if
you
look
at
this
one,
you
realize
that
there's
only
48
bits
present
in
this
64
bit
long
prefix,
because
the
next
16
bits
are
zero,
so
they're
omitted.
So
that's
good!
So
next
slide
please
and
so
the
same
thing
for
ipv4.
G
G
I
I
don't
personally
have
a
great
need
for
that.
I
don't
need
to
pack
thousands
of
prefixes
into
a
network
kind
of
thing
and
do
something
with
them.
I
would
be
happy
with
just
putting
the
tag
on
each
one,
particularly
if
it's
not
too
long,
but
if
someone
else
has
that
need.
What
I
would
see
is
that
this
could
be
accommodated
within
the
same
tag
by
just
saying:
oh,
look,
it's
an
array
and
then
you're
you
could
do
that,
but
I'm
not
personally
interested.
G
G
And
I
would
like
to
do
an
early
allocation
is
a
oneplus
one
tag
justified
for
this,
so
that
would
be
a
two
byte
tag.
260
and
261
are
in
that
space
and,
if
so,
I'd
like
to
allocate
as
soon
as
possible
next
slide,
I
think
that's
it
yeah
and
then
there's
a
notable
tag
document
which
it
was
mentioned
that
this
we
could
merge
together.
But
well
I
one
way
the
other.
I
don't
care
as
long
as
the
early
allocation
happens
already,
because
I
don't
want
to
wait.
A
D
G
G
G
Okay,
so
in
my
code
the
prefix
consists
of
an
array,
a
number
and
then
reading
address
with
with
trailing
pres
zeros
and
noting
what
the
prefix
length
is
expected.
So
to
me,
it's
common
code,
so
it's
free.
G
Secondly,
I
think
that
decoders
need
to
have
some
instructions
as
to
what
to
do
if
there
are
bytes
missing
and
if
that's
drop,
that's
okay,
but
one
way
or
the
other.
They
need
to
be
told
what
to
do.
If
there's,
if
there's,
if
that's
happening,
so
I
think
implementing
the
trailing
zeros
on
the
receiver
is
not
so
terrible
and
if
encoders
don't
drop
them,
that's,
okay!
It's
okay
to
not
drop
the
trailing
zeros!
G
G
G
A
A
A
brief
question
from
me
again
with
chair
head
off
on
the
topic
of
the
existing
existing
tech
260.
A
C
Thanks
custom
yeah,
so
that's
really
the
the
job
of
the
notable
text
document
we'll
get
to
in
a
couple
of
minutes
to
to
generate
and
to
hold
explanations
on
on
how
these
tags
actually
work
together
or
in
this
case,
don't
work
together,
because
we
don't
expect
them
to
to
turn
up
in
the
same
environment.
I
would
prefer
to
keep
the
the
actual
tag.
Definitions
very
crisp,
so
implementers
only
get
that
information.
They
need
to
implement
this
and
if
they
actually
need
a
tour
guide
through
the
world
of
tags.
A
Sounds
good
to
me
in
terms
of
document
reviewing,
which
is
something
that
will
come
up,
who
has
plans
to
implement
or
review
this.
A
Okay,
that
will
we'll
have
we'll
have
to
get
a
bit
more
active
participation
also
on
the
to
to
to
finalize
this
eventually,
but
with,
I
think
we
can.
At
least
we
can
have
a
good
way
to
continue
from
here.
C
So
this
is
another
old
acquaintance.
Next
slide
please
so
this
has
been
around
for
a
while
and
actually
it's
an
active
use.
So
since
this
is
a
tag
from
the
from
what
used
to
be
the
fcfs
space,
it
no
longer
is
fcf
space,
but
at
the
time
it
was
fcfs
it's
allocated
and
and
people
are
using
that.
But
now
we
have
some
some
additional
requirements.
C
So
one
thing
that
we
added
was
a
time
scale
indication
the
draft
already
hinted
that
this
is
about
to
happen,
but
I
think
we
now
understand
what
we
want
to
have
there.
One
can
boil
the
ocean
there,
but
we
came
up
with
a
pretty
simple
thing:
that's
probably
proficient
for
most
applications
and
we
added
clock
quality,
uncertainty
and
guarantees
to
to
address
specific
use
cases.
C
So
this
is
about
the
time
tag
itself
and
the
the
other
thing
that
that
was
open
in
the
previous
versions
was
actually
defining
a
representation
for
duration
and
period.
But
duration
is
a
time
delta
like
like
five
seconds
and
the
period
is
an
interval.
So
it's
like
the
start
of
the
meeting
up
to
five
minutes
into
the
meeting.
That
would
be
a
time
period,
so
those
are
now
defined.
C
Actually,
the
the
time
period
looks
a
little
bit
different
than
it
used
to.
So
we
will
need
to
ask
ayanna
to
actually
update
the
the
registration
for
that
with
respect
to
the
the
data
type
assumed,
but
I
think
that's
a
worthwhile
change
to
make.
C
So
there
is
no
implementation
experience
with
duration
and
period
yes,
but
yet,
but
it's
written
up,
so
people
can
start
using
it.
So
at
this
point
I
I'm
not
in
a
particular
rush
moving
this
forward.
On
the
other
hand,
there
are
protocols
over
there
in
in
the
reds
context
that
that
could
use
a
slightly
more
powerful
time
tag.
C
So
if,
if
there
is,
if
we
see
that
documents
that
could
use
this
become
ready
for
approval,
we
might
want
to
to
finish,
or
at
least
snapshot
this.
This
document
as
well,
so
that
that's
my
status
and
I'm
waiting
for.
E
E
Working
of
course,
on
items
that
direly
need
this,
but
I
think,
for
example,
we
have
protocols
here
in
the
itf
already
that
go
beyond
the
typical
annotation,
that
this
is
a
posix
timestamp,
or
that
this
is
the
standard
spring
time,
and
so
so
they
think
that's,
that's
these.
The
the
attributes,
carson
is
optional,
but
they
really
really
help
in
practice.
E
When
you
want
more
information
about
your
time
and
every
everybody
is
coming
back
to
to
not
only
me
but
sometimes
me
and
is
asking
how
do
I
phrase
my
timestamp
if
I
need
this
additional
information
and
discuss
over
and
over
again,
so
this
is
just
a
just
a
speaking
from
experience-
and
this
is
a
side
comment
that
this
is
really
useful.
When
you
can
cherry
pick
your
your
kind
of
fraction,
your
resolution
and
other
attributes,
you
you
want
to
attach
to
your
time
value,
and
so
that's
successful
comment.
A
So
there's
been
quite
a
bit
of
discussion
in
in
early
in
on
earlier
versions
of
this
on
the
mailing
list.
So,
given
that
this
was
announced-
and
I
still
think
this
is
a
good
idea-
I'd
like
to
ask
you
for
a
hum
in
for
for
the
adoption
of
this
document
into
the
sibo
working
group,
you
sure
that
you
should
have
see
the
raise
hand
now.
A
A
A
Okay,
thank
you.
So
I'm
seeing
11
11
hands
from
the
raised
here,
which
is
given
that
we're
only
30
participants
quite
a
bit
just
to
check
for
the
for
the
opposite
question,
any
any
any
opposition
to
you
having
this
in
the
working.
A
A
A
Thank
you
before
we
go
to
notable
tags
which
are
in
the
sequence
of
slides,
pretty
much
the
next
one
I'd
like
keo
to
give
to
tell
us
about
the
topic
of
ordered
maps,
which
is
something
that
has
not
really
been
an
itf
activity,
but
he
has
been
kind
enough
to
come
here
and
present
the
work
that
is
so
far
a
document
that
would
just
register
in
the
specification
requirement
section
but
would
could
still
profit
from
the
discussion
here.
Thank
you
for
coming
here.
Q
floor
is
yours,.
H
Okay,
thanks
very
much
yeah,
so
I'm
kia
smallwood,
if,
if
you
skip
to
the
second
slide,
so
I'm
maintainer
of
the
seaboard
2
python
library
originally
written
by
alex
gronholm
in
in
finland,
and
I
submitted
a
giant
patch
that
implemented
the
deterministic
encoding
back
when
it
was
when
it
was
called
canonical,
and
he
asked
if
I
wanted
to
maintain
the
library.
So
I
accept
it.
H
H
H
The
benefit
of
this
is
that
it's
gives
you
efficient
iteration
over
all
items
and
in
python
at
least
it's
fairly
fairly
compact
next
slide.
Please.
H
H
Originally,
that
was
purely
coincidence,
but
since
I
think
python
3.8
that
has
been
documented
behavior
in
that
the
built-in
dict
preserves
insertion
order,
but
the
difference
that
order
dict
has
is
that
it's
got
some
extra
methods
that
allow
you
to
move
things
around
and
also
when
you're,
comparing
for
equality.
The
order
is
important,
whereas
for
the
original
dict,
it's
not
important
next
slide,
please.
H
Which
allows
extended
key
types
similar
to
how
you
can
use
like
a
python
tuple
as
a
as
a
key,
you
can
do.
A
similar
thing
with
a
javascript's
map
allows
better
iteration
over
the
keys
than
the
built-in
object.
I
didn't
know
this
before.
I
wrote
this
presentation,
but
the
it's
possible
to
get
key
collisions
with
object.
The
object's
prototype
next
slide.
Please.
H
So
in
ruby,
the
hash
is
actually
ordered
by
default
since
1.9,
but
it's
not
used
for
equality
comparisons.
So
if
you
were
using
this
in
in
ruby,
it
would
be
up
to
the
author
whether
you
felt
that
the
the
order
of
of
your
hash
map
was
important
next
slide.
Please.
H
So
this
originally
came
to
my
attention
when
a
user
of
our
library
opened
an
issue
saying
hey
when
I
serialize
ordered
they
just
get
serialized
as
as
type
5,
which
is
which
is
just
a
plain
map,
and
if
you're
using
a
deterministic
encoding,
then
it
reorders
those
keys,
so
the
original
information
is
lost.
H
It
also
means
that
that
when
decoding
it
becomes
a
plain
dict
when,
when
instantiating
the
object-
and
so
he
requested,
you
know
a
way
to
always
decode
maps
as
as
order
dicts,
but
I
suggested
that
we
use
a
tag
for
it
instead,
so
that
we
could
communicate
to
others
that
the
order
of
this
is
important
next
slide.
Please
so
for
for
us
to
consider,
is
you
know?
H
How
often
will
this
be
used
currently
we're
proposing
tag
279,
which
is
a
as
a
oneplus
2,
and
I
wrote
a
spec
for
it
under
my
github
handle
sikhenrae,
which
you
can
view
there.
H
H
H
A
It
thank
you,
questions
comments,
especially
as
to
the
as
to
the
to
the
tag
length
that
would
make
sense
here
and
to
use
cases
in
the
in
the
constrained
area.
H
C
Yeah
so
we
we
have
had
in
the
last
seven
years.
We
have
had
various
requests
both
for
ordered
maps
and
for
other
multi-maps
and
and
we
never
followed
through
on
these
and
and
maybe
it's
time
to
finally
do
this,
and
I
actually
think
we
need
both
all
that
multi-maps
are
really
needed
when
you
have
a
traditional
tlv
structure,
and
you
are
trying
to
to
represent-
represent
this
in
in
a
zebra
world
in
in
unchanged,
in
an
unchanged
way,
because
tlvs
are
usually
allowed
to
be
repeated.
C
Results,
but
not
necessarily
in
the
way
that
the
reordering
done
by
by
sibo's,
deterministic
encoding
is,
is
providing
so,
for
instance,
in
in
sdf.
C
We
have
definitions
of
of
properties,
and
these
properties
are
very
often
in
a
specification
that
you
look
at
that
is
manually
generated,
ordered
from
from
most
important
to
ancillary
or
in
some
way
and
right
now
there
is
no
way
to
to
actually
preserve
that.
So
that's
one
example
where,
where
other
maps
come
in,
so
I
I
vacate
my
slot
here
so
jeffrey
can
say
something.
B
Can
you
hear
me
this
seems
like
it
would
fit
in
certain
places
in
the
web
packaging
spec,
I'm
not
sure
that
we
would
actually
switch
to
it
because
we're
currently
using
ordered
maps
and
and
making
that
work.
But
but
it
seems
like
a
useful
thing
to
have
had
to
have
around.
A
Thank
you
comments
on
anything
on
the
on
the
tag
length
that
would
seem
meaningful
to.
A
A
We
are
we're
running
out
of
time
right
now,
but
I
see
one
more
plus
one
for
for
one
plus
one
tags.
We
had
on
the
agenda
originally
the
notable
tags
as
well,
and
that
would
need
to
move
on
to
an
interim,
but
we
will
reinstate
interims
after
the
after
this
itf.
So
I'm
I'm
looking
forward
to
the
discussion
there
francesca
handing
over
to
you.