►
From YouTube: IETF-CORE-20230510-1400
Description
CORE meeting session at IETF
2023/05/10 1400
https://datatracker.ietf.org/meeting//proceedings/
A
B
B
C
B
And
small
meeting
right,
the
time
and
I
don't
know
if
you
discussed
this,
but
Christian
told
me
that
he
has
Fabian
around
so
he
will
be
somewhat
Limited
in
his
interaction
capabilities.
A
Yeah
worst
case,
listening
only
probably
typing
can
work
here
and
there,
but
it
will
do
our
best.
C
A
Adjustments,
no
so
let's
get
started
properly
and
welcome
everyone
to
this
interim
meeting
of
the
co-working
group,
I'm,
Martin
law,
chemical
chairs
and
custom
Borman
and,
being
this
an
official
meeting
and
not
well
applies,
get.
A
C
D
A
B
A
B
So
my
plan
was
to
have
a
dash
13
submitted
for
this
meeting.
Of
course,
that
didn't
quite
work
out
but
I
think,
except
for
two
things
that
still
need
to
be
done.
We
have
that.
So.
Can
you
increase
the
point
size,
some
more
right,
better
or
even
more
use?
The
whole
screen
always
use
the
old
screen
more
more.
C
B
So
interesting,
interestingly,
all
the
changes
got
focused
into
two
areas,
so
we
have
one
area
that
you
are
showing
right
now,
which
adds
rules
that
go
beyond
the
cdd8
specifications
or
the
CDA
specification
gives
you
the
overall
structure,
but
it
cannot
express
some
things
and
we
already
had
text
about
trailing
now
values.
Of
course
we
could
write
cdgl.
That
removes
those,
but
this
is
very,
very
tedious.
B
So,
instead
this
is
in
in
text,
but
the
text
is
now
probably
turned
into
a
bullet
list,
so
the
first
two
items
are
not
new:
they
are
just
changed,
so
it
fits
into
the
Buddhist
and
there
are
two
new
items.
B
One
is
that
we
had
this
discussion
about
an
empty
pass
versus
a
knife
path
and
since
both
seem
to
have
the
same
meaning
in
both
actually
only
consume
one
bite
in
in
the
encoded
sibo
data
item
we
kind
of
flipped
the
coin,
and
to
me
it
seems
that
that
we
are
closer
to
the
actual
semantics
if
we
do
require
an
empty
array
here,
so
this
fits
with
the
semantics
that
the
empty
array
has
in
the
CI
reference.
The
path
is
actually
given.
B
It's
just
empty,
so
that's
what
I
wrote
up
here.
So
if
you
have
an
empty
path
in
a
CI,
this
is
not
about
CI
references.
This
is
about
absolute
Cris.
Then
you
put
in
an
empty
array.
So
if
people
want
the
the
other
side
of
the
coin,
I'm
totally
open
to
that,
this
just
seemed
more
consistent
to
me.
C
B
Good
now,
if
meet
eggy
worth
listening
to
me,
no
I
cannot
select
the
chat
foreign.
B
B
This
is
the
discussion
we
had
about
an
entirely
empty
CIA
reference
which
doesn't
seem
to
be
particularly
useful,
and
this
is
now
explicitly
ruled
out
here.
So
that's
not
a
CIA
reference.
We
want
to
Interchange.
B
So
that
results
one
other
item-
I-
think
that
that's
item
two
out
of
issue,
number
50
and
I-
think
the
other
sub
issues
of
that
issue
were
already
covered,
so
I
I
made
it
close.
The
entire
issue
of
that
is
wrong.
Christian,
please
reopen
it.
B
B
So
this
lists
a
number
of
of
ways
in
in
which
a
CRI
can
be
invalid
and
and
one
of
those
is,
it
not
even
be
well
formed
as
a
CI,
but
there
are
other
requirements,
and
so
of
course,
it
might
be
using
extensions
that
are
so
not
supported
by
the
implementation
and
the
the
new
thing
here
in
this
subsection
is
that
we
now
have
the
concept
of
an
unprocessable
CI.
B
Now,
at
the
end
of
this
section,
this
is
actually
diluted
a
little
bit
by
pointing
out
that
you
don't
have
to
have
code
that
detects
that
a
CRI
is
unprocessible
for
you,
if
you
are
not
actually
planning
to
do
the
the
processing
that
would
make
it
unprocessable.
So
the
the
duty
of
checking
validity.
B
The
last
paragraph
now
on
the
screen,
the
duty
of
checking
validity
is
with
the
recipients
that
rely
on
this
validity
and
an
intermediate
that
does
not
use
the
detailed
information
might
actually
be
able
to
process
the
CI,
even
if
it
doesn't
understand
a
specific
feature
that
is
going
on
in
the
the
some
detail
of
this
UI.
B
So
you
might
be
able
to
perform
reference
resolution
without
actually
understanding
percent
encoded
text,
for
instance,
and
it's
not
not
your
duty
to
check
that
the
CRI
is
valid
according
to
your
subset
of
of
the
extensions
that
you
implement,
and
that
means
that
we,
that
the
consumer
has
the
ultimate
responsibility,
but
also
that
the
producer
has
the
responsibility
to
create
something
that
actually
is
interoperable
and
that's
true
for
both
basic
Cris,
where
at
least
we
have
to
check
for
a
valid
UDF
age,
which
is
already
required
for
for
checking
the
validity
of
the
sibo
and
for
extensions
that
might
have
additional
requirements
like
the
minimal
use
of
percent
encoded
text
elements.
B
So
I
think
that
that
is
a
useful
to
point
out
that
there
is
no
need
to
actually
do
excessive
checking,
in
particular,
if
you
don't
even
have
the
parts
that
you
would
need
for
for
checking.
So
we
we
can
have
intermediary
intermediaries
that
check
less
and
kind
of
avoid
ossification
by
by
having
each
intermediary
check
everything.
B
Okay,
so
this
this
addresses
part
of
44,
but
the
rest
of
issue
44,
where
we
write
up
what
minimal
use
actually
is
that
that
still
requires
some
some
text
to
be
written.
So
that's
the
the
one
big
thing:
that's
Missing
for
doing
Dash,
13.
B
So
the
the
interesting
part
here
about
unprocessable
Cris
is
that
you
can
still
handle
them.
I
mean
you,
don't
have
to
blow
up
all
your
processing
just
because
you
come
across
an
unprocessable
CRI.
B
It
is
just
not
processable
as
a
CI
by
itself,
so
you
should
have
some
some
code
in
your
implementation
that
allows
you
to
handle
unpossible
Cris
to
get
maximum
robustness.
B
And
the
the
text
that
is
now
the
second
on
the
screen
is
indicating
how
unprocessable
Cris
compare
they
compare
a
little
bit
like
not
the
numbers
do
in
in
floating
Point
implementations,
except
that
we
allow
Cris
with
identical
theorizations,
to
be
treated
as
identical,
and
we
may
need
to
tweak
the
word
serialization
here.
B
Okay,
so
I
think
that
that
is
the
biggest
new
piece
here,
but
it's
also
something
that
that
a
sensible
implementer
would
probably
have
done
already
now,
a
few
scroll.
B
In
the
direction,
so
we
can
see
more
of
of
the
early
text
of
that
section
right
which
is
scrolling
down
for
me
and
scrolling
up
for
other
people.
You,
you
see
the
the
little
editors
comment
here
with
the
starting
with
a
so
I
think
Christian
was
was
pointing
out
that
life
is
easier.
If
you
don't
have
indefinite
length
items
in
extensions
and
I
think
actually
we
could
go
ahead
and
say
sia
eyes
shouldn't
be
using
indefinite
links
representations
at
all.
B
So
this
just
together
with
some
some
new
text
that
is
further
down.
Essentially
it
says,
don't
use
indefinite
links
in
extension,
so
it
becomes
easier
to
skip
them.
If
you
don't
understand
them.
B
B
On
the
other
hand,
an
implementation
that
that
writes
an
indefinite
length
array
open
and
then
starts
processing
a
path
bite
by
bite
that
that's
not
entirely
weird,
so
yeah,
let's,
let's
maybe
have
a
little
bit
more
discussion
about
this,
and
the
final
change
is
at
the
end.
So
if
you
can
scroll
up
some
more
or
scroll
down
depending
on
which
part
of
the
world
you.
B
Good
enough,
yep,
okay,
so
here
we
have
the
fix
that
dots
in
hostname
items
in
host
name
elements
are
not
reserved,
so
normalization
will
turn
back
any
percent
encoding
that
you
might
have
had
there.
So
that's
just
not
possible
to
have
a
DOT
in
in
the
host
name
element.
So
this
is
now
you
know
clarified,
and
the
last
one
we
see
on
the
screen
is
that
mandate.
That
extensions
should
not
use
well,
must
not
use
indefinite
length
encoding.
B
Yeah,
if
we
go
go
the
whole
way
and
say
sia
eyes,
I
intended
not
to
use
indefinite
language
representation,
I
think
that's
fine
and
we
can
remove
that
here.
B
Yeah,
we
need
to
extract
a
little
bit
more
of
item
44
of
issue
44,
so
we
we
have
a
better
description
of
what
a
minimal
representation
is
and
maybe
to
a
few
editorial
small
things.
And
then
we
have
a
dash
13.
B
I
definitely
wanted
want
to
submit
it
this
week,
so
if
we
can
get
a
Shepherd
right
up
done,
we
we
might
have
some
overlap
of
the
working
class
call
with
my
vacation.
B
Takes
excuse
me
not
the
shepherd
right
up.
The
chairs
preview
takes
longer
than
that.
That's
not
a
problem
either.
B
A
Yeah
I
was
planning
to
do
another
review
of
this
anyway.
It
may
be
problematic
exactly
in
the
next
few
weeks
because
of
other
deadlines.
Yeah.
C
D
B
A
And
in
case
you
miss
it
from
the
chat
on
the
previous
topic
on
the
extensibility
and
indefinite
land.
Christian
is
also
fine,
with
none
at
all
approach.
Yeah,
okay,.
B
Yeah,
that's
a
little
bit
of
a
simple
weirdness.
What
Cris
really
Define
is
a
data
model
specification
and
what
representation
you
actually
use
in
the
encoding
is
kind
of
a
layer
below
but
saying.
If,
if
you
are
handing
around
to
your
eyes
on
their
own,
please
send
them
in
in
a
definite
length
encoding.
That's
a
good
thing
to
do,
but
if
it's
part
of
a
gigantic
data
structure
like
a
cosmic
or
something
actually
requiring
the
the
CI
item
in
the
crossword
to
be
definitely
length
encoded.
B
Is
that
puts
a
requirement
on
the
implementation
that
that
may
be
really
hard
to
do,
because
there
may
be
other
things
that
that
I
mean
there
may
be
an
image
of
the
code
for
a
device
in
the
Hedge
crossword.
So
you
may
have
some
need
for
for
indefinite
links
encoding
for
other
parts
of
the
cosmet.
So
it's
hard
to
instruct
the
sibo
encoder
to
to
actually
avoid
indefinite
length
encoding
for
the
actual
Cris
that
that
are
happening
in
that.
So
that's
why
I'm
always
a
little
bit
careful
with
mandates
like
this.
A
A
A
B
Right
yeah,
so
hi
Christian,
I
added
you
to
the
conscious
contributors.
B
A
A
A
A
There's
No
Agenda
items
yet
and
we'll
call
for
agenda
items
on
Thursday
and
next
week,
following
up
with
this.
If
anything
has
to
be
brought
up,
is
an
option.
Otherwise
we
may
have
a
seed
depending
on
any
progress
on
that,
but
I
suspect
no
but
hi
may
we
have
something.
D
D
But
I
don't
have
the
contact
people
so
I
will
contact
people
for
the
reviewers
in
saiana,
so
but
I
think
it
was
custom
who
will
look
into
the
contact
people
from
those
groups
right
if
I
tell
them
wrong.
That's
my
memory.
Maybe.
D
D
B
Yeah
that
might
be
done
in
an
off
off
meeting
call
just
just
informally
or
in
the
interim,
let's
find
out
whether
people
again
this
has
been
around
for
a
while.
So
we
we
first
have
to
check
whether
they
are
even
the
the
right
people
to
discuss
this
in
the
first
place,.