►
From YouTube: IETF-CELLAR-20230124-2000
Description
CELLAR meeting session at IETF
2023/01/24 2000
https://datatracker.ietf.org/meeting//proceedings/
B
D
E
On
the
on
Saturday,
for
me,
too,
I
do
presentation
about
ffv1.
A
A
Three
minutes
after
Dave
we
have
Dave.
Does
he
have
a
microphone.
B
B
I
believe
we
had
one
meeting
after
after
no
time
to
wait.
B
A
I
probably
put
the
wrong
minutes
then
here
to
confirm.
B
And
Michael
Michael
we
had
a
meeting
on
on
December,
6th
and
I
just
found
that.
B
They
I
will
say
that
they
were
said
to
the
the
the
mailing
list.
A
Okay,
okay,
all
right!
So,
let's
dive
into
these
and
are
there
any
objectives
to
those
to
those
minutes?
Any
corrections.
A
Okay,
I
turned
this
status
thing
into
a
table.
Thank
you,
Kodak
parmie.
Thank
you.
Codex
control,
ffe1
flock
still
waiting
for
Shepard
right
up,
I
think.
Yes,
okay
and
you're.
The
shepherd
right.
Yes,.
A
B
A
B
E
A
I'll
put
that
then
okay
matroska,
so
I
saw
that
lots
of
merges
happened
in
the
last
week.
We
had
the
80
evaluation
come
back
and
let's
talk
about
that
in
the
matroska
slot.
Any
other
updates
to
any
of
these
documents
that
we
should
mention
from
anybody.
F
B
A
A
Having
done
that,
let's
go
on.
Let's
talk
about
matroska,
so
documents
returned
from
the
80
with
comments.
We
had
lots
of
issues
merged.
Do
you
want
me
to
pull
up
Steve?
Do
you
want
me
to
pull
up
any
issues
that
are
giving
you
problems.
D
700
is
from
the
80
review,
so
basically
I
realized
from
the
review
that
the
ietf
uuids
are
actually
a
the
values
inside.
D
Actually
have
a
meaning,
there
are
Parts
in
it
with
difference,
meaning,
and
this
is
not
what
we
have
in
matrosca
and
it
was
never
like
that
I,
don't
think
we
should
move
it
that
way.
So
basically
the
UniQue
Ideas
we
have
in
matroscare
are
not
actually
you
your
ideas,
so
I
went
back
and
reverting
the
change
that
we
made
to
do
that.
D
So
that's
the
full
request.
I
think
it's
correct.
The
only
issue
is
whether
we
keep
on
naming
them
uuids
or
we
go
back
to
New
IDs
as
before
as
well.
It
might
be
confusing
for
people
if
they're
called
uuids.
They
might
expect
that
they
have
a
special
meaning,
but
then,
if
they
read
the
space,
they'll
see
it's
not
special
meaning.
D
On
the
other
end,
it's
nice
to
keep
the
name
because
there's
a
lot
of
existing
code
with
formatting
and
things
like
that.
That
could
be
useful.
So.
A
A
Yes,
so
what
I
would
say
is
why
not
say
they
are
uuids
version?
4
uuids
are
recommended
to
be
used,
but
that
readers
must
tolerate,
must
treat
them
as
opaque,
128
bit
strings
or,
and
and
that's
it.
A
That
way,
you
get
to
call
them
uuids
and
it
tells
people
how
to
generate
them,
which
is
just
random
unless
you
want
some
other
version,
which
probably
you
don't
time
ordered.
So
some
of
these
are
are
designed,
for
instance,
to
be
time
ordered
so
that
they
will
insert
into
databases
in
the
particular
order.
For
instance,
you
can
sort
them
and
then
there's
version
eight,
which
is
basically
we
don't
know,
do
whatever
you
like.
B
So
we
want
that
to
be
uuid,
V4.
A
A
D
Yeah
so
I
wanted
to
discuss
that
one
as
well,
because
basically
in
the
spec,
so
we
have
a
lot
of
deprecated
ideas
or
some
a
lot
of
elements
that
were
never
used
that
had
an
idea
assigned
to
them.
D
On
the
other
end,
it's
that
means
we
lose
a
lot
of
very
valuable
ideas,
especially
the
ones
in
one
octet,
one
byte.
D
Probably
never
because
when
we
deprecate
something,
that's
because
no
one
used
it
or
used
it
wrong,
but
mostly
never
used
it.
So.
A
What
I
suggest
that
we
do
is
that
we
move
them.
We
mark
them
as
returned,
rather
than
reserved.
A
A
A
So
what
this
means
is
that
we
would
have
to
before
we
started
using
them.
We
would
have
to
have
exhausted
all
the
other
ones
and
yeah.
Then
when
we
went
to
allocate
them,
we
would
know
that
they
were
returned,
and
so
we
might
actually.
You
know
that
point
ask
a
question
like
is
it
has
anyone
seen
such
a
thing,
and
you
know
we
might
get
evidence
that
it's
actually
out
there
and
it
was
act
then
we
might
might
provide
some
like
Telemetry
some
reviews,
the
fact
that
it
might
be
out
there
or.
D
A
So,
let's
mark
them
as
returned
that
way,
I
think
that's
the
right
term
Spencer.
Do
you
think
so
I
think
so.
B
A
We
went
through
this
with
DHCP
option
codes
where,
for
instance,
one
of
them
which
are
also
one
byte
and
and
worse
than
that
half
of
those
were
at
one
point,
turned
over
as
vendor
private
or
something
yeah,
and
then
we
we
said:
okay,
no,
no,
there's
a
different
way
to
do
this,
but
like
like
the
cap,
Port
working
group
got
a
number
assigned
at
first
and
then
discovered
when
doing
an
experimental
deployment
that
polycom
phones
all
crashed
and
it's
because
they
had
used
it
as
a
a
code
for
their
configuration
file
and
well,
they
got
confused
when
the
wrong
thing
data
showed
up
and
they
crashed,
which
is
really
their
own
fault,
but
anyway,
calport
managed
to
get
another
code
assigned
to
it
instead
to
switch
codes
when
revising
the
document,
so
it
happens,
but
you
just
yeah
it
would
have
been
good
if
we
had
known
that
this
was
used
or
if
someone
had
done
it.
D
B
I
did
send
it.
I
did
send
the
No
Doubt
to
the
group
about
remembering
that
this
is
a
the
output
of
a
working
of
the
working
group.
So
if
there's
anything
that
you're
curious
about
in
the
document,
this
is
an
excellent
time
to
point
that
out.
D
A
Martin,
do
you
have
any
Flack
issues
you
want
to
bring
up,
I,
guess
you're
waiting
for
the
write-up.
A
B
We
did
have
the
one
question
that
was
and
asked
today
on
the
mailing
list.
C
C
That's
another
codec
Opus
RFC
a
few
days
ago
and
I
noticed
that
they
had
a
sectional
encoding
so
how
to
encode
certain
stuff
and
I
looked
in
the
MP3
specification.
It
was
also
a
section
about
encoding,
wasn't
really
ever
in
the
scope
but
well.
A
Well,
people
tend
to
focus
on
getting
the
decoder
right
because,
once
you
know
the
rules
for
the
decoder,
then
you
can
innovate
on
the
side
of
encoding
within
those
within
that
space
and
every
knowing
that
everyone
can
do
it.
And
that's
what
Jerome
says
here:
I
guess
and
people
sometimes
don't
want
to
nail
down
the
encoder
too
much
because
they
feel
that
it
unnecessarily.
A
It
unnecessarily
tells
the
decoder
things
that
may
not
always
be
true
right.
Yes,.
C
B
A
Yeah
so
I
would
say
that's
up
to
this
working
group
to
what
extent
we
we
would
like
to
have
an
example
of
doing
this
or
a
reference
and
I
think
it's
okay,
that
it
doesn't
have
to
be
normative.
I
think
that's!
The
whole
point
is
the
encoding
is,
is
done
which,
with
whatever
the
best
technology
is
right
now
and
as
long
as
the
decoder
can
can
parse
it.
It's
it's
good
right.
C
C
I
was
just
curious
whether
for
example,
everyone
was
in
that
line
of
thinking,
so.
B
I
think
that
was
the
only
one
that
was
mentioned
on
the
mailing
list,
though.
C
Else,
I
I
had
nothing
else
for
Flack
yeah
waiting
on
the
shepherd's
write-up
I
did
find
a
few,
because
I
keep
experimenting
with
the
flag
software.
The
the
reference
implementation
I
sometimes
find
some
small
details
usually
do
a
pull
request
for
that.
So.
A
Yeah
I
think
you're.
Fine,
too
I
think
they
should
keep
doing
some
minor
minor
edits,
as
as
we
get
you
know
thing
and
into
review.
So
so
what
will
happen
next
is
we'll
get
reviews
from
other
areas
and
some
of
them
maybe
extensive,
and
some
of
them
may
not,
but
and
then
we'll
get
reviews
from
the
area,
directors
and
our
area
director.
B
Right:
we've:
we've
we've
had
a
couple
of
documents
that
have
gone
all
the
way
through
the
or
you
know
further
than
this
through
the
process
so
far,
so
where
everybody
kind
of
understands
the
the
process
here,
which
is
basically
it
gets
out
of
the
working
group
and
then
goes
to
the
area
director
then
goes
to
the
ietf,
including
review
teams
and
then
goes
to
the
isg
for
final
approval.
A
C
Reading
up
on
how
the
process
went
with
ffe
one
and
look
what
the
timeline
was
so
not
to
get
any
well,
it
might
probably
be
more
a
a
year
or
more
I.
Guess
before
that.
A
Maybe
not
it
doesn't,
it
doesn't
have
to
take
that
long.
I
would
say.
I
would
expect
to
be
finished
by
June
is
what
I
would
say.
B
B
Yeah
I
mean
you're
you're
super
responsive,
and
this
is
something
like
an
80,
some
80,
something
page
spec,
which
you
know
at
least
it's
not
400,
Pages
or
something.
C
A
We've
been
paused
on
it
for
some
time
and
I'm
not
upset
about
that.
I
just
want
to
open
the
floor.
If
there's
something
else
that
nobody.
E
Iowa
I
feel
the
the
link
about
the
first
damn
presentation,
so
the
ffv1
presentation
across
them
with
date
and
time.
E
I
could
say
that
there
is
a
patch
for
ffmpeg
for
having
fav1
in
mxf
for
for
mixing
ffv1
in
the
mixer,
so
it
is
waiting
for
merge,
so
some
some
review
and
match
for
mxf.
So
if
everyone
in
mxf,
because
ffv1
ffmpeg
is
the
reference,
the
encoder
decoder.
So
this.
A
F
E
E
Many
a
lot
of
broadcasters
are
using
that
on
Netflix,
also
at
some
point.
So
it
is
a
very
ugly
container
format,
but
a
lot.
A
A
Setting
a
sandwich
everything
thrown
into
it
right,
four
different
formats.
E
So
it
is
also
used
Now
by
archive,
because
archive
gets
some
files
from
broadcasters
and
so
on.
So
so
there
is
a
request
from
archive
to
have
everyone
in
mxf
support.
So
there
is
already
the
decoding
in
ffmpeg
and
now
there
is
a
patch,
a
batch
proposal
for
a
web
mixing.
E
F
E
Would
I
would
sound
the
a
a
PF
for
that
too?.
F
E
F
I
know,
there's
no,
you
know
not
much
to
discuss
about
version
four
of
ffp1
right
now,
but
I
wanted
to
point
out
there's
a
small
kind
of
wish
list
of
what
could
be
included
in
ffp1
version
4
and
the
issue
tracker
of
the
ffv1
spec
I
think
there's
like
four
or
five
yeah.
It
looks
like
there's
five
items
in
there
right
now.
B
Do
we
do
we
want
to
do?
We
want
to
add
that,
as
a
topic
for
next
meeting.
F
Yeah
I
wonder
if
we
get
Derek
boitness
put
in
two
of
these
issues.
Two
of
them
came
from
redo.
One
was
from
me
the
right
people.
If
we
want
to
show
up.
B
A
A
All
right,
good.
A
B
A
A
A
A
Now
my
youngest
brother
will
be
doing
the
same
thing
as
he
is
expecting
a
child
in
the
spring.
So
this
is
the
last
time
he'll
be
allowed
to
potentially
die.