►
From YouTube: IETF-CBOR-20230125-1500
Description
CBOR meeting session at IETF
2023/01/25 1500
https://datatracker.ietf.org/meeting//proceedings/
A
A
A
A
A
B
A
B
Now.
This
is,
of
course,
always
an
interesting
question.
When
you
have
a
document
that
itself
is
extensible
to
a
certain
extent,
then
part
of
that
may
be
Rock,
Solid,
stable
and
part
of
that,
maybe
things
that
people
put
in
because
they
think
they
will
need
them
and
the
transition
from
an
information
document
to
to
a
standard
strike
document,
of
course
raises
the
question
how
we
handle
the
second
category.
B
Well,
there
have
been
some
GitHub
discussions
about
the
clock
quality
and,
in
particular
about
the
fact
that
the
clock
quality
is
defined
in
terms
of
an
RFC
that
actually
doesn't
Define
anything
at
all.
But
then
just
references,
IEEE
1588,
which
is
behind
a
paywall.
So
many
algebrae
standards
are
available
through
the
gets
IEEE
program.
But
this
one
isn't
so
that
that
ruffled,
some
feathers.
A
A
Sure,
okay,
so
then
why
don't
we
say
that
on
the
next
version
you
make
it
standards
track
and
I
will
change
the
data
tracker
to
go
to
standards
track,
and
we
will,
when
you
post,
that
to
the
mailing
list,
we'll
make
a
another
quick
call
of
whether
there
are
any
objections
to
that
change.
Does
that
sound,
reasonable.
B
B
So
the
next
line
event
start
equals
time
tag.
Dot
e
time
is,
is
my
proposal
for
this
when,
when
we
give
something
a
name
like
time
tag
here
or
if
we
know
don't
do
that
when
we
just
use
the
the
file
name
for
that
purpose,
then
we
can
reference
this
as
prefix
dot,
whatever
it's
called
in
the
reference
file.
B
B
Yes,
Michael,
the
the
semicolon
makes
this
comment
in
acidity
L1
implementation.
So
you
can
still
run
your
your
city,
di-1
tools
on
this
without
hiccups
and
well
I
I
chose
the
the
hash
mark
because
that's
something
that
people
know
from
from
C
for
for
this
pragmaster
a
comment
and
then
of
course
the
stylistic
question
is
whether
there
should
be
a
space
be
behind
the
hash
or
not.
B
We
can
discuss
that
so
probably
I'm
going
to
do
a
little
mailing
list
poll
for
that,
but
otherwise
the
syntax
is
probably
going
to
steal
python
syntax
a
little
bit
more
than
than
C
language,
syntax
and
I.
Think
it's
also
important
to
make
sure
that
we
don't
allow
arbitrary
file
system
references
here
so
right
now,
this
is
very
limited
in
what
you
can
have
after
the
world
include
so
what's
after
the
word
include
essentially,
is
is
limited
to
to
cddl
identifiers.
B
So
the
the
problem
with
the
wholesale
include
is
that
you
get
all
these
additional
rules
that
you
actually
don't
need
so
right
now,
I'm
thinking
about
reducing
this,
to
just
the
rules
that
are
actually
imported,
plus
the
the
transitive
closure
of
of
the
reference.
So
everything
that
that
is
needed
to
to
make
this
usage
time.
Tag.Etime
is
also
imported
from
the
include,
but
not
the
whole
document,
because
there
may
be
other
things
in
there.
B
So
for
the
existing
rfcs,
it's
pretty
obvious
how
to
handle
this,
except
that
there
is
one
one
question:
how
do
we
handle
the
Prelude
I'm
actually
in
the
current
implementation?
I
just
have
the
word
prelude
for
that,
because
it's
really
very
special,
but
we
could
also
identify
this
as
appendix
D
of
RFC
8610
I'm,
not
sure
we
need
this.
So
at
the
time
when
I
actually
import,
all
the
the
rfcs
I
will
try
to
check
what
we
actually
need
there.
B
So
in
in
your
pipeline,
you
would
include
a
command
like
this,
maybe
with
some
additional
Flags
such
as
warning
about
undefined
and
things
like
that,
and
the
the
question
really
is:
where
do
we
find
the
the
reference
document?
So
where
do
we
actually
find,
for
instance,
RFC
in
1952
or
the
the
internet
draft?
B
That
is
in
a
reference
in
the
example
here
and
what
what
we
actually
need
is
the
CDL
for
them
and
I
I
plan
to
put
in
some
processing
that
allows
you
to
get
CDL
out
of
most
documents,
XML
text
or
whatever,
but
of
course
that
won't
work
reliably
in
in
all
cases.
So
my
current
plan
is
to
put
all
rfcs
into
the
implementation,
because
rfcs
don't
change,
it
should
be
easy
to
extract
cddf
from
rocs
that
have
cddl
include
them
in
the
gym
and
then,
of
course,
the
only
problem
is.
B
B
So
there
could
be
other
places,
so
it
could
try
to
get
things
from
the
web
or
anything
but
I'm
I'm,
not
so
sure
about
how
that
that
works
with
actual
workflows,
because
you
don't
want
all
the
surprises
that
come
from
referencing
something
on
the
web.
So
right
now.
This
is
the
the
set
of
things
that
that
is
supported.
B
B
C
Sorry
I
was
we
didn't
want
to
interrupt
so
I
think
that
yeah,
you
should
have
another
place,
cddl
path,
so
I
gather
you're,
basically
proposing
that
you're
going
to
do
the
file
extract
from
RFC
to
make
a
database
of
cddl
files
and
I
guess
I'm
gonna
suggest
mechanically.
Maybe
you
want
to
make
that
a
separate
gym.
B
Yeah,
that's
probably
not
needed
so
much
for
rfcs,
because
these
are
actually
pretty
stable,
so
yeah
you're
talking
about
the
yeah.
The
problem,
of
course,
always
is.
While
you
are
working
on
documents,
they
are
nowhere
in
the
web
or
or
anywhere.
So
that's
why
the
current
directory
is
in
this
list
and
you
want
to
have
a
way
to
to
point
to
a
different
place
than
the
country.
B
Yeah
Russ
comments
that
the
extraction
sounds
a
lot
like
young
catalog.org
for
young
modules,
so
maybe
at
some
point
we
we
need
that
right
now
we
are
talking
about
maybe
30
or
50
documents,
so
I
think
it
it's
still
somewhat
manageable
and
also
a
little
bit
of
of
hand.
Processing
may
be
needed
for
the
the
oldest
RCS
that
do
something
with
CD
edit.
So
maybe
that's
an
appropriate
thing
to
include
in
the
gym.
B
D
What
I'd
like
to
understand
is
if,
if
we
already
need
this
Con
this
down
conversion
from
cddl2
to
cddl1,
why
does
the?
Why
does
the
pragma
need
to
go
through
something
that
used
to
be
a
comment
to
me
this
this
style
of
hiding
something
in
what
used
to
be
a
common
smells
a
lot
like
HTML
comments
that
say
if
Internet
Explorer
version
equals
and
I'd
like
to
forget
about
those,
so
a
document
that
imports
from
somewhere
will
not
be
usable
Standalone
anyway.
D
B
Yeah
I,
don't
don't
have
a
really
strong
opinion
on
that
I
think
it
is
useful
to
be
able
to
use
City
A1
tools
on
on
those
files
just
because
there
is
such
a
diversity
of
towards,
but
yeah.
You
may
be
right
that
the
usefulness
of
of
that
will
decline
over
time,
but
on
the
other
hand,
why
don't
we
just
say
semicolon?
Hash
is
a
really
nice,
a
combination
that
we
all
like
and
and
then
we
don't
really
need
to
have
this
discussion.
I,
don't
know.
B
B
This
is
all
implementation
details,
but
I
think
these
are
important
details,
because
if
we
cannot
offload
these
things
to
the
implementation,
we
may
have
to
address
them
in
the
language.
So,
if,
if
we
need
something
in
the
language
to
actually
access
the
web
and
and
get
something
from
there
and
do
extractions
from
there,
that's.
B
Yeah,
the
the
problem
is
that
that
I
finally
broke
down
and
said
this
will
require
a
modern
Ruby
version,
because
working
with
with
Json
structured
data
is
just
so
incredibly
ugly
if
you
have
to
do
it
manually.
So
the
the
support
in
the
recent
Ruby
versions
of
this
decade
is
really
useful
for
that.
So,
if
you
don't
have
at
least
Ruby
3.0,
then
you
may
want
to
look
for
for
installing
that.
B
D
B
Okay,
so
maybe
there
will
be
something
like
a
head
to
path
directive
or
something
like
that,
but
I
think
we.
We
can
see
that
when
we
actually
start
using
it
and
and
see
that
Things
become
unwieldy,
and
that's,
of
course,
very
important
for
for
me
to
actually
get
this
feedback
when
people
start
using
it.