►
From YouTube: CBOR WG Interim Meeting, 2021-02-24
Description
CBOR WG Interim Meeting, 2021-02-24
A
Which
makes
this
also
an
official
itf
meeting,
so
the
note
well
applies
and
if
there's
anything
unclear
about
it,
please
talk
to
francesco
me.
A
That
being
said,
there
has
been
some
discussion
on
the
topic
of
ordered
maps
on
the
mailing
list.
Recently.
My
understanding
is
that
this
is
something
that
there
is
interest
in
the
working
group
to
have
this,
especially
with
respect
to
seaboard
being
a
more
general
format
that
that
general
purpose
languages
can
use
to
serialize
their
their
data
items
right
now.
The
documents
that
are
around
there
are
more
more
out
of
the
working
group.
Suggestion
texts,
so
is
there
is,
are
there
is
there?
B
So
I
already
have
some
embryonic
text
on
the
the
other
map
tags
in
the
notable
tags
document.
I
don't
remember
how
far
I
got
in
writing
this.
So
for
me
that
would
be
a
natural
thing
to
add
there.
B
So
we
would
just
identify
the
contributors
here
and
just
add
it
to
that
draft,
but
that's
not
the
only
way
to
to
skin
this
cat.
C
So
there
is
some
text
right.
I
I
mean
I've
not
really
followed
super
closely,
but
I've
seen
some
texts
from
keo
and
joe.
B
B
B
Yeah,
so
I
I
just
paste
it
in
and
form
it
in
a
way
that
fits
the
rest
of
that
draft,
and
I
think
we
we
can
have
that
well
before
110.
So
we
can
look
at
that
and
then
at
110
decide
how
to
do
this,
so
the
the
for
me.
The
interesting
observation,
is
that
there
are
two
objectives
here:
one
is
the
objective
to
actually
be
able
to
to
tag
a
map
as
order
preserving,
and
the
other
objective
is
to
work
with
platforms
that
have
two
different
map.
B
Types
of
course,
depending
on
on
how
these
map
types
actually
look
like
it
may
be,
you
may
want
to
have
different
properties
of
the
tanks.
So,
for
instance,
we
already
have
one
tag
that
just
says
all
the
keys
in
this
map
are
text
strings.
B
Platforms
that
have
maps
with
text
strings
like
javascript
objects,
that's
a
hint
that
they
don't
have
to
go
for
the
full
javascript
map,
treatment.
A
I
think,
but
I
think
the
four
default
behaviors
will
be
just
as
as
tricky
with
the
order
preserving
map
just
as
well,
because
different
languages
have
different
defaults
and
different
native
types
for
mappings
and
occasionally
change
the
semantics.
Just
in
between
versions.
B
Yes,
but
in
the
list
you
cannot
talk
about
keys
and
values.
So
saying
this
is
really
a
key
value.
Thing
makes
the
difference
between
just
putting
this
as
an
array,
so.
B
Okay,
I
think
the
only
really
interesting
or
difficult
question
here
is
how
many
of
these
do
we
want
to
have
and
which
of
these
is
actually
worth
going
for
a
1.1
or
one
plus
one
or
even
a
one,
plus
zero
tag,
yeah
and
that's
of
course,
very
much.
A
judgment
call
so
yeah.
It's
not.
We
don't
have
a
corpus
of
documents.
We
can
just
run
against
that
and
say:
oh,
it's
an
obvious
technical
decision
that
we
don't
want
to
do
this
or
that
we
want
to
do
that.
A
Or,
or
requests
from
anywhere
in
the
constrained
constrained
area,
where
people
where
people
want
to
use
something
that
is
not
a
native
seaboard
type,
because
things
like
serializing,
a
native
python
or
pro
or
whatsoever
language
object,
is
not
something
that
would
typically
go
into
the
constrained
domain
that
we
usually
preserve.
The
small
types
for.
B
A
Okay,
I
think
it
would
be
good
here
to
to
put
ask
him
directly
to
well
where
he
sees
the
use
case
for
this,
because,
if
what
I've,
what
I've
seen
on
the
main
instrument
was,
was
a
yes,
we
need
this,
but
no
concrete.
We
need
this
force
or
we
need
short.
We
need
your
text
with
it.
A
C
A
A
Sizes,
any
any
other
points
on
the
on
the
map
types.
A
Michael,
just
before
mentioned
that
he
would
run
a
bit
late,
so
I
postpone
the
the
file
magic
and
network
network
address
topics
to
when
he's
here,
which
makes
the
next
topic
a
cddl
control.
Is
there
anything
I've
seen
there
is
the
new
document?
I've
got
to
be
honest,
I'm
not
completely
up
to
date.
What
the
what
the,
what
the
updates
there,
where
custom?
Can
you
give
a
one
sentence
summary,
and
is
there
anything
more
that
you
want
to
mention
there.
A
B
B
So
I
think
these
three,
we
could
just
finish,
and
the
question
really
is:
do
we
wait
for
the
abn
f
takes,
which
are
obviously
more
work
to
to
get
right?
Do
we
wait
for
these
to
to
also
be
implemented,
or
do
we
just
say?
Okay,
let's
do
these
three
first
and
the
other
ones
a
little
bit
later.
A
A
At
least
that's
my
that's,
that's
that's
my
personal
direction.
Here
is
so
I
I'd
say
I'd
say
yes,
but
with
no
urgency.
So
if,
if
this
only
happens
say
some
sometime
after
the
itf,
possibly
around
a
processor
that
would
give
a
bit
of
time
to
for
for
abnf
use
case
to
come
up
and
if
they
don't.
B
Well,
it's
not
a
matter
of
use
cases.
We
have
tons
of
use
cases,
it's
a
matter
of
somebody
sitting
down
and
implementing
it
and
android
wise
has
started.
I
have
started,
but
we
are
not
yet
in
a
stage
where
would
where
we
would
say.
Oh
it's
implemented
and
we
understand
all
the
pitfalls
and
that
I
really
prefer
to
have
things
implemented
before
we
standardize
them.
Okay,.
A
Let
let
me
phrase
the
use
case.
A
bit
differently
is
any:
are
any
of
these
use
cases
urgent
enough
to
make
the
people
who
would
use
it
actually
actually
use
it,
and
not
only
just
use
it
as
a
notational
hint
or
use
it
in
the
term,
in
the
sense
that
they
need
implementations
for
it.
B
And
actually
have
a
tool
that
does
that.
But
it's
not
it's
not
good.
I
cannot
recommend
that
at
the
moment,
so
I
I
mean
we
have
lots
of
abnf
in
in
the
itf
and
I'd
rather
pave
a
way
where
people
can
say.
Yes,
we
can
use
the
existing
ata
abnf
for
the
text,
parts
of
my
protocol
and
cdl
for
the
structural
parts,
and
that
comes
up
all
the.
A
But
on
the
large
scheme
of
things,
the
the
point
probably
stays
that
if
we
need
at
least
those
parts
and
the
and
don't
get
traction
on
the
others,
then
splitting
it
by
something
like
april
would,
in
my
opinion,
be
a
good
idea.
B
C
I
think
we
can
put
on
waiting
for
implementation
and
maybe
putting
on
some
interim
after
content
and
make
point
and
see.
What's
this
other
point.
A
So
when
you
say
that
that
that
the
dependent
project
wants
to
be
done
by
this
year,
does
this
mean
past
its
own
working
group?
Last
call
by
this
year
or
past
isg
last
call
or
like
published
rfc.
B
A
A
A
A
A
So
what
just
to
repeat
what
I've
said
earlier,
the
working
group
adoption
calls
are
looking
good
for
for
file
magic
and
network
address.
So
without.
A
Kind
of
anticipating,
too
much,
I
think
we
can
start
working
on
this
on
this,
at
least
at
least
in
in
discussion
parts.
So
there
there
have
been
a
few
questions
on
the
mailing
list
on
the
topic
of
what
do
we
need
in
terms
of
why,
where
the?
Why
are
the
particular
numbers
chosen
in
terms
of
for
file
magic?
I
think
that
this
has
been
answered.
D
So
I
think
that
the
only
important
part-
I
hope
you
can
hear
me,
I
think-
for
for
final
magic.
We
just
need
agreement
that
we
want
to
have
a
seahorse
sequence
and
a
wrapped,
and
that
we
said
we
wanted
a
new
tag
for
the
super
secret
place:
five,
five,
seven,
nine
nine.
As
far
as
I
can
tell,
there's
nothing:
five,
five,
seven,
five,
five,
seven
nine
anything
would
work
because
the
last
bite
is
not
if,
as
long
as
it's
in
that
area,
it's
not
a
valid
unicode.
D
Second
fight.
But
I
guess
the
other
test
is
whether
it's
something
else,
but
I
don't
think
it
is
and
I've
I've
kind
of
checked
that
already
no
their.
B
D
Yeah
800
or
98,
or
whatever
you
want
yeah
any
any
direction,
that's
totally
arbitrary.
What
I
did
look
up
was
that
numbers
in
that
air
range
are
are,
are
three
bite
would
be
three
bite
utf-8
except
the
second
bite
is
not
valid.
It
must
start
with
two,
the
two
ones
anyway,
so
this
range
does
not
have
it
in
that
area,
so
it
would
not
be
a
valid
utf-8.
A
I
think
the
the
the
property
that
that
we,
that
the
original
tag
info
was
to
be
invalid
in
all
the
utf,
all
the
unicode
encodings,
including
utf-16
and
utf-32.
D
Okay,
well
I'll
check
utf-16.
I
didn't
do
that
and
see
if
that,
if
that
that
isn't
whether
it's
invalid
in
that
too,.
D
B
B
D
D
And
then
on
the
topic
of
ip
address
tags,
if
you
want
to
go
to
that,
there
was
some
quest
for
do.
We
want
ability
to
be
able
to
tag
entire
array
of
them
and
I'm
gonna
say
that
I'm
indifferent,
not
I
don't
have
that
use,
but
I
understand
that
other
people.
A
D
Yeah
and
and
so
for
the
for
the
person
that
wants
to
have
an
array
of
prefixes
so
like
cider,
plus
a
prefix.
D
If
they
have
many
thousands
of
them,
then
it's
quite
likely
that
they
have
a
fair
bit
of
overlap
in
the
creep
in
the
initial
parts
and
that
they
may
also
wish
to
sort
them
by
longest
to
shortest.
In
which
case
there
may
be
better
in
code
than.
D
Yeah,
so
there
may
be
a
better
encoding
like,
for
instance,
you
you
could.
You
could
certainly
not
use
jersey
by
doing
deltas
on
the
prefix
length,
but
you,
but
you
may
also
you-
can
also
avoid
repeating
the
you
know,
64
bits
of
the
initial
part
of
the
prefix,
which
you're
changing
all
the
time,
the
rest
of
it.
That
may
just.
B
Yeah
that
that
may
actually
be
a
bit
of
over
optimizing
here.
It
is
supposed
to
be
simple,
so
I'm
not
sure
that
coming.
D
D
B
Yeah,
so
I
I
went
ahead
and
did
another
version
of
the
time
tab
tag
document
right
now.
We
are
getting
more
and
more
requests
out
of
the
attestation
environment,
red
working
group
and
so
on
for
for
how
to
to
put
time
into
these
data
structures,
and
they
they
need
a
little
bit
more
than
than
tag
one.
So
the
timetable
tag
document
seemed
appropriate,
but
they
also
need
things
out
of
existing
rfcs
like
three
one:
six,
one
and
the
precision
time
protocol
stuff
and
so
on.
B
So
we
added
some
of
that
and
that's
one
aspect.
We
probably
have
to
run
this
document
by
the
tick
tock
working
group
before
or
during
last
call
at
least
about
probably
before-
and
we
have
to
make
sure
it
actually
solves
the
problems
that
that
the
rats
people
are
having,
but
I
think
otherwise
we
are
in
pretty
good
shape
and
I
would
expect
that
there
will
be
documents
that
are
going
forward
that
use
this
relatively
soon.
B
They
just
need
a
little
bit
more
information
about
the
quality
of
the
time
tag.
Oh,
if
somebody
tells
you
that
this
happened
at
1705,
you
really
want
to
know
how.
How
sure
are
they
about
that
and
what's
the
the
accuracy
or
the
uncertainty
about
that
time
and
that's
what
we
added.
B
A
So
that's
probably
a
good
point
in
time
to
ask
people
in
the
working
group
to
have
a
look
at
this,
because
at
some
point
we
will
want
to
discuss
whether
this
should
be
become
a
working
group
document.
And
for
that
should
people
should
have
read
it.
B
So
from
my
point
of
view,
this
is
now
getting
ready
for
an
adoption
call,
because
it's
no
longer
a
big
swiss
cheese.
A
But
that's
something
that
can
easily
be
that
doesn't
need
to
be
fully
solved
before
at
the
point
this
this
is
asked
to
be
adopted
into
a
working
group.
B
Right
unless
somebody
says
that
that
they
are
absolutely
opposed
to
working
on
this
in
this
work
group,
unless
the
registry
issue
is
resolved,
but
I
hope
that
is
not
the
case
wouldn't
expect.
A
A
A
But
but
asking
but
getting
getting
around
asking
asking
about
adoption
of
this
document
is
something
that
I
could
easily
see
being
on.
That
list
sounds
good.
B
C
Not
yet,
I
guess
during
the
atf
week
where
I'll
be,
you
know
both
roles
at
the
same
time,
but.
C
I
think
this
time
both
me
and
barry
will
be
like
overlapping
the
whole
week,
because
there
is
a
lot
of
overlaps,
also
in
in
art
working
groups.
So
we
will
both
act
things
that
we
can
follow.
All
the
groups.
A
Well,
okay,
then,
in
this
case,
see
you
all
in
in
online
and
francesca.
If
you
have
a
few
minutes,
then
we
could
we
could
can
we
could
keep
working
on
basically
exactly
that
thanks
everyone
and
see
you
there.