►
From YouTube: WebRTC Working Group Virtual Interim September 25, 2018
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
A
A
A
This
is
a
reminder
that
we
have
a
very
early
draft
of
WebRTC
next
version
use
cases,
and
we
would
like
people
to
take
a
look
at
it.
We've
had
a
foot
up
reviews,
a
few
issues
filed
and
thank
you
for
those
who
have
read
it
and
commented
on
it,
but
we
would
like
more
people
to
take
a
look
at
it
and
file
issues,
so
we
can
move
forward
on
it
and
improve
it,
and
here
are
a
few
of
the
comments
that
people
have
made,
and
so
you
may
want
to
take
a
look
at
it
see.
C
A
C
A
D
F
The
specs
already
mentions
Commission's,
but
we
need
some
additional
language
in
the
permission,
spec
special
circumstances
for
or
differences
such
as
a
from
camera
microformats
Queen
sure,
is
that
the
spec,
our
spec
specifically
forbids
granted
permissions,
but
denied
permissions
are
okay
and
for
those
of
you
who
have
seen
the
the
drop-down
in
Firefox
and
wondered
why
there's
a
checkbox
for
remember
this
decision.
When
you
check
it,
it
says
Firefox
does
not
a
lot
of
permanent
access.
F
The
reason
for
that
is,
you
can
still
hit
the
don't
allow
button,
maybe
not
the
best
UX,
but
that's
why.
So?
That
means
we
need
a
new
permission
name.
So,
in
addition
to
camera
microphone,
we're
proposing
display
screen
was
so
runner-up,
but
it
display
seems
to
be
reasonable,
as
I
mentioned,
get
use
them
into
a
spec
already.
The
algorithms
already
assumed
this
so
there's
already
language
in
display
there
and
they
get
in
the
screen.
Sharing
spec,
there's
also
already
text
assess
the
user.
F
Agent
must
not
create
a
permission
stores
entry
with
a
value
of
granted,
so
hopefully
that
should
be
sufficient.
So
the
open
questions
is
whether
this
is
sufficient
to
cover
both
video
and
any
related,
audio
or
I.
Think
Martin
suggested
that
it
would
be
not
necessary
to
have
independent
audio
and
video
for
this
and
I
tend
to
agree,
but
but
and
because
the
effect
would
be
basically
that,
if
you
block
one,
you
block
both
and
I
think
there's
a
separate
slide
about
audio
and
I.
F
Guess:
there's
a
question:
if
there's
even
an
audio,
only
use
case
for
this,
but
so
for
now
the
recommendation
is
to
just
have
a
simple,
simple
entry
for
display
that
would
cover
existing
video
and
future
audio,
and
this
is
still
the
US
agents
responsibility
to
explain
and
prompts
what's
being
shared.
So
this
again
is
only
the
only
part.
That's
needed
in
the
permission
expect
for
this
is
to
be
able
to
persist,
block
grants
basically
and
to
fulfill
the
permission,
algorithms,
otherwise
and
yeah.
That's.
A
B
B
So
yeah,
so
you
use
cases
typical
use
case
would
be
your
you're
holding
a
presentation
and
you,
when
I
share
like
it's
an
audio
and
video
clip,
so
that
I
can
imagine
doing
audio
only
to
share
music
like
that
right.
So
I
before
I
made
a
proposal
where
I
thought
you
know,
keep
things
simple:
let's
just
share
all
the
audio,
but
talking
to
some
people
but
different
ways.
People
want
to
implement
this,
and
even
the
limitations
are
very
OS
like
something
like
share
system.
B
Audio
might
not
even
be
possible
on
all
operating
systems
and
just
extensions
where
you
opt
in
or
out
to
a
audio.
You
know,
depending
on
the
window
or
tab
and
my
best,
some
of
the
technical
constraints
and
the
differences
between
OSS
and
and
how
browser
might
want
to
do
this
differently.
I
think
that
we
should
say
as
little
as
possible
asked
what
audio
is
into
that
and
to
unlock
all
the
future
audio
related.
B
F
F
B
B
F
My
concern
is
that
as
a
way
to
signal
but
I'm
guessing
a
web,
app
that
doesn't
expect
audio
should
not
receive
it.
So
it
makes
sense
to
use
audio
:
true
to
say
that
I
would
like
audio.
F
B
B
If
the
implementation
guarantees
that
it's
only
video
but
maybe
but
but
I
mean
some
suggestions,
you
have
a
picture
where
you
pick
one
tab
and
then
in
the
middle
of
the
call,
you
just
switch
to
another
tab
and
think
they're
places
where
you
start
with
no
audio.
And
then
you
have
audio
later
and
we're
good
to
have
an
audio
track.
B
F
B
What
audio
to
include
I
thought
I
thought
it
would
make
sense
to
produce
silence
so
anything
similar
to
how
you
can
mute
the
microphone,
because
because
if
you
are,
for
example,
if
you
want
to
share
a
window
and
I
think
the
application
can
say
I'm
interested
in
both
audio
and
video,
but
if
you're
an
hour
rating
system
or
an
implementation,
where
it's
not
possible
to
give
you
that
audio
is
okay,
I,
don't
think!
If
you,
if
you
had
a
failing,
then
we
needed
to
have
a
lot
of
mechanisms
for
probing
but
impossible.
B
We
should
be,
then
maybe
you
have
to
have
different
constraints
and
what
I'm
thinking
is,
when
you
just
say,
get
display
media
that
the
user
hasn't
made
their
choice.
It
right
like!
So,
if
you,
if
you
say
that
you
have
to
be
able
to
support
audio,
but
then
you
have
to
exclude
the
window
option,
even
though
the
user
might
ever
let.
F
B
F
Sorry,
a
question
there
right
now
because
existing
implementations
today,
if
you
put
in
all
your
true
that
will
just
be
ignored
right
right,
so
might
be
a
problem,
we're
compatibility
if
we
or
like,
if
you
did
this
today
with
video
and
audio
true,
you
would
get
back
a
stream
that
only
has
a
video
track
right.
Yeah.
B
So
well,
so
in
that
would
be
a
user,
don't
be
an
argument
for
not
returning
a
silent
audio
stream
if
we
can
ensure
that
there
will
never
be
any
audio
in
the
case
of
you've
never
implemented
this
run.
That's
true,
on
the
other
hand,
are
we
in
a
stage
where
we
have
to
worry
about
backwards?
Compatibility
like
this,
the
spec
is
pretty
fresh,
a.
A
F
A
B
B
F
A
B
C
F
E
Well
and
I
guess:
Safari
right
yeah.
This
is
un,
so
having
to
implement
a
silent,
audio
track,
good
simulation,
odd
and
then
the
rotors
need
to
evaluate
to
detect
body
silence,
or
would
it
be
just
muted
that
seems
great
for
the
bloopers
and
not
great
for
in
content,
exciting
good
for
nothing
right.
C
F
B
F
B
I
mean
so
maybe
maybe
first
app,
but
maybe
I
get
the
request
and
the
usage
I
choose
to
share
a
window.
The
operating
system
doesn't
allow
me
sharing
any
on
you
there,
but
during
the
call
the
browser
has
some
UI.
That's
that's
nice
say
like
no.
No
I
want
to
share
a
tab,
I'm
just
changing
the
source,
and
in
that
case
now
I
do
have
audio
and
there
and
let
you
just
start
producing
stuff,
so
the
application
you
have
access
to
a
track.
B
E
B
F
F
Once
you
pick
a
source
there,
javascript
observable
properties
of
that
source
and
if
they
change
from
under
you,
I
think
that
violates
the
spec
in
some
way
like
we
had
this
with
front
and
back
cameras
on
phones
and
the
way
around
it,
I
guess
would
be
to
make
a
virtual
device.
That
is
the
combination
of
the
two
and
then
you
could
maybe
change
the
which
way
the
camera
is
facing,
based
on
the
the
device
sort
of
the
facing
mode
constraint.
B
E
Supports
them,
for
we
could
edit
that
for
camera,
you
have
the
background
on
a
program
cameras
and
we
could
and
then
the
monitor
you
have
a
switch
so
that
we
don't
need
to
end
JavaScript
to
actually
switch
switch
off.
Devices
that
are
getusermedia
is
providing
you,
this
ability,
through
constraint,
user
environment
and
it's
attached
to
different
devices,
so
the
device
ID
will
be
somehow
different.
F
A
F
I,
like
you
one
point
that,
since
the
end
user
is
going
to
pick
a
source,
if
they
pick
a
source,
I
will
never
have
audio
I
think
we
should
return
a
stream
with
just
a
video
track
and
you
know
audio
track
right.
B
F
B
E
B
So
if
you
include
this
constraint,
then
again
starting
point
to
still
implementation,
that's
what
everyone's!
So
you
can
get
any
audio,
but
you
cannot
get
audio
that
includes
the
requesting
tabs
audio.
So,
for
example,
if
you're
doing
time
capture,
then
app
implementation
can
return
only
that
audio
for
the
implementation
can
return,
like
any
other,
like
that
fold,
full
browser
windows
audio
as
long
as
doesn't
contain
this
audio.
That
would
cause
it
go
and
it's
again
it's
up
to
the
implementation.
B
B
F
That
sounds
like
a
special
use
case.
So
I
know
some
sites
uses
the
peridot
in
uses
screen
sharing
to
record
their
session,
but
I
believe
in
those
cases
I
would
argue
they
don't
really.
You
don't
really
need
tap
audio
capture
because
you
already
have
the
audio
and
you
can
probably
do
with
Web,
Audio
and
other
means
to
get
that
audio.
B
F
F
Well,
my
suggestion
would
be
to
not
do
this
because
the
following
reasons,
basically
because
expect
expressly
forbids
it-
there's
already
language
in
there,
and
this
is
also
why
I
get
displayed
in
media
did
not
support
constraints
for
the
longest
time
until
very
recently,
and
we
recently
reintroduced
constraints
only
thanks
to
the
strong
language
against
letting
the
the
application
influence.
The
user
selection,
yeah.
E
Yeah
I
cannot
agree
that,
for
instance,
hangout
might
have
some
UI
telling
the
user
hey
Joe
in
Chrome,
particularly
this
way,
and
you
might
actually
click
to
phone.
You
get
the
window
this
way,
and
so
it
can
provide
contact
to
the
user
to
actually
tell
him
what
if
we
decide
as
part
of
the
UI
of
the
web
page,
and
hopefully
that's
good
enough,
maybe
it
would
not
be
the
case
and
in
which
case
you
might
want
to
continue
thinking
about
better
ideas.
That
I
would
agree
with
the
proposal
there
to
not
support
that.
B
I've
a
curious
about
what
the
use
cases,
if,
because,
if
it's
the
matter
that
they
want
to
have
one
button
for
sharing
the
one
button
for
share
tab,
one
button
like
they
make
the
picker
part
of
the
applications
UI
just
so
that
the
you
know
they
get
the
same
experience
regardless
of
browser,
then
I
think
it's
fine,
but
if
they,
if
they
want
to
add
this
constraint
because
they
want
to
avoid
you
know
certain
choices
or
problematic
because
of
implementation
details
like
echo,
then
it's
not
okay,
then
there
should
be
a
different
constrain.
Well,.
F
I
could
speak
to
some
of
the
use
cases.
I've
heard
is
that
the
applications
want
to
have
a
document
sharing
feature,
so
they
very
specifically
want
to
only
pick
documents,
but
unfortunately
they
would
also
include
Google
Documents,
so
in
that
would
require
about
it,
browser
sharing,
which
is
the
the
scary
one.
If
you
go
to
next
slide,
I
basically
I
should
mention
the
background
here.
Is
that
not
all
sources
are
equally
safe,
so,
specifically
any
any
surface
that
has
a
browser
surface
visible
on
it,
as
a
potential
is
extra
unscary.
E
I
can
see
overuse
cases
like
you,
you're
it
to
help
desk
and
you
actually
want
to
only
capture
a
window.
But
it's
from
a
given
app,
so
filtering
on
are
getting
the
picker
with
four
windows
will
have
somehow
reserved
to
the
committed
choice
of
a
user
which
might
be
good.
But
again
the
web
page
might
have
UI
to
provide
guidance
to
the
user.
But
here
the
begins
done
right.
All.
F
E
B
So
I
think
this
will
be
quick.
What
happens
if
the
window
is
closed
or
the
monitor
is
disconnected
and,
and
my
cycle
strikes
already
assessed,
but
now
we
have
this
mute,
which
is
for
temporary
and
ended,
which
is
permanent,
and
you
know
if
you
close
the
window
I
think
you
should
just
change
the
ready
state
and
higher
on
ended
and.
E
A
B
B
E
B
F
F
We
already
overload
muted
in
peer
connection
to
know
whether
you're
receiving
bits-
and
you
know
the
direction.
You
can
change
the
transceiver
direction
and
you
will
get
muted
that
way.
I
guess,
there's
no
reason.
We
can't
also
mute
when
the
source
is,
but
actually
now
we're
just
talking
about
meeting
the
track
which
actually
does
not
cause
mute.
It
never
mind
yeah.
E
E
F
B
F
E
B
F
F
C
B
C
Put
a
little
note
same
as
should,
and
we
should
think
about
the
reconnect
case.
Some
more
I
can't
read
Rick
Lee
there.
Oh,
is
this:
handle
disconnection
a
motoring,
big
monitors
differently,
yeah,
here's
a
feel
just
that
it
turned
off
and
they
were
still
just
stupid.
The
drivers,
but
nowadays
motors
can
actually
go
away
and
you
are
no
guarantee
that
they're
coming
back
at
the
same
place.
Yeah.
E
Every
annum
to
be
different
from
you,
it
cameras
as
well
because
get
you
the
media,
usually
once
you
know
what
once
you're
granted
access
to
one
camera
when,
if
you're
disconnected
and
reconnected,
you
just
go,
get
you
the
media
again
and
there's
no
prompt,
but
there.
If
you
end
the
track,
you
actually
need
to
call
get
the
media
again,
which
means
a
new
prompt
which
might
not
be
very
good
for
users.
So
that's
why
I'm.
B
Yes,
so
what
I
thought
in
the
issue
was
suggesting
that
we
make
it
black
if
it's
minimized
and
I
think
that's
that's
a
valid
option,
but
I
also
think
you
could
see
minimized
windows
as
a
special
case
of
it
being
hidden,
because
you
know.
Okay,
if
you
don't
see
the
window,
but
that's
that
it's
also
the
case.
If
you
put
another
window
over
it-
and
in
that
case
we
don't
make
it
black.
B
So
if
we
want
to
be
consistent
with
oh,
it's
just
it's
just
hidden,
it's
still
still
being
sent
and
I
think
it
would
make
more
sense
to
not
do
anything
and
just
keep
sending
frames
and
I
wanted
to
group
to
discuss
which
option
I
I
prefer
not
to
make
it
black
because
I
think
that's
now,
that's
like
muting,
your
your
stream
and
I,
don't
think
might
be
on
purpose
to
do
that.
It
I'm
sharing
a
window
I'm!
Okay
with
you,
seeing
it
I'm
minimizing
I'm
problem,
just
context.
A
B
B
F
A
F
A
A
An
edge
when
you
minimize
it
it
sends
black,
but
we
have
because
we
we
are
not
screen
scraping.
You
can
actually
show
a
covered
window
if
you
select
a
window,
it'll
show
whether
it's
covered
or
not,
and
if
that
required
special
operating
system
functionality
to
be
able
to
do
that
previously,
we
weren't
able
to
do
that
so.
E
E
And
when
it's
minimized
by
looking
about
commute
and
you'd,
even
see
that
the
JavaScript
for
it's
about
TCP
data,
it's
very
visible
from
the
web
applications
that
something
happens
and
probably
you
got
the
user
visible,
get
display
media
or
it's
minimized,
and
can
we
act
upon
it?
It
is
black
plane
when
it's
political
for
web
applications
to
understand
what
yes,
what's
happening.
B
It
could
also
be
up
to
the
browser
like
one
browser
might
explicitly
have
a
mute
and
news
button.
Another
browser
might
just
rely
on
minimizing
or
not
minimizing.
I,
don't
think
we
need
to
make
this
decision,
but
we
should
say
that
if,
if
we're
not
producing
frames
for
the
captured
area,
we
we
should
mute
and
unmute.
If.
F
C
E
Yeah
there's
a
special
case
of
covert,
which
is
when
the
screen
saver
is
kicking
on
or
when
you
are
actually
are
logging
out
for
something
like
that
and
I
did
an
experiment
with
chrome
and
Chrome
is
still
capturing,
for
instance,
even
though
you
finish
lock,
you
can
do
and
I
don't
know,
but
for
a
fox,
maybe
Firefox
is
the
same,
not
culture
so
the
mighty.
So
it
might
be
good
for
the
stack
to
mention
some
cases,
even
though
we're
like
a
implementer.
B
B
B
E
Yeah,
so
hopefully
it
will
be
quick
and
fast,
so
get
this
play.
Media
I,
haven't
looked
at
all
implementations
in
browsers,
but
gadget
play
media
has
the
same
concern
as
getusermedia.
So
probably
when
it's
not
a
second
contact
by
default,
it
should
not
be
functional.
I
can
look
at
agree
on
that,
and
so
in
that
case
they
are
like
two
main
options
to
make.
Get
this
pan
medium,
not
functional
nonstick
for
context.
One
is
to
do
ad,
get
use
Vidya.
Basically,
if
the
contact
is
not
secure,
you
reject
the
promise.
E
E
B
F
Depress
and
some
of
this
method
anyway,
so
I
like
that,
whether
we
can
do
the
same
thing
for
media
capture,
I
get
using
media
I,
hope
so,
but
getusermedia
doesn't
actually
isn't
actually
secure
contacts
only
today,
Firefox
still
supports
it
in
HTTP,
but
we
would
like
to
change
that.
To
only
have
it
work
in
secure
it
only
work
in
HTTPS,
but
have
been
removing
it
entirely.
Yeah
I
would
probably
break
some
websites,
but
probably
not
that
important,
so
yeah
I
would
support
that
as
well.
If.
E
F
F
F
E
F
B
A
All
right
so
WebRTC,
let's
try
to
get
into
these
so
1858
for
WebRTC.
It
was
about
what
happens
when
an
answer
stops
a
transceiver
that
others
a
bundle
on
just
to
refresh
everybody's
memory.
When
you
have
a
bundle
group,
say
audio
and
video.
The
first
identification
tag
is
known
as
a
bundle
tag
and
that
identifies
the
tagged
M
section
and
there's
a
bunch
of
things
relating
to
this
in
bundle.
A
So
it
says
it's
recommended
that
the
offeror
tag
section
be
one
that
the
offer
believes
is
unlikely
to
the
answer
or
reject
how
it's
supposed
to
know
that
I
don't
know
it's
not
implementable
and
then
in
Section
seven
33.
It
says
when
the
answer
wants
to
reject
a
bundle,
M
section.
If
it's
the
offer
a
tag
one,
it
can't
reject
just
at
M
section
of
a
needs
to
reject
the
whole
offer,
reject
each
bundled
M,
section
or
or
later
create
an
offer
to
disable
it.
So
that's
what's
causing
the
issue
here.
A
So
yeah,
so
some
of
the
API
implementations
create
offer
you
actually
the
text
and
that
spec
Kearney
doesn't
even
say
which
you're
tagged.
Em
section
is
my
understanding
is
that
implementations
make
that
the
initial
transceiver
so
the
first
time
you
call
a
transceiver
that
becomes
the
offer
tagged.
Em
section
is
that
true,
universally.
A
So
so
that's
my
first
question
is
what,
if,
whatever
my
visions
actually
do
so
there's
some
confusion
around
that
and
then
the
second
limitation
is
that
the
answerer
without
parsing,
the
SDP,
can't
determine
whether
the
transceiver
they're
attempting
to
stop
is
the
one
that's
offer
tat.
So
basically,
I
can.
B
A
Well,
so,
if
you're
in
a
stable
state
and
and
say
the
answer,
our
calls
transceiver
dot,
stop
definitely
the
say
on
the
audio.
The
audio
will
definitely
stop
right.
There.
Transceiver
is
now
dead,
but
he
hasn't
generated
an
offer
for
an
answer
right
because
it's
already
stable.
So
it's
it's
not
gonna
I,
don't
think
it'll
automatically
kill
the
video
to
call
stop
on
the
answer.
It's.
A
Right
it'll
cause
negotiation
needed
and
then
well.
It
depends
on
which
side.
So
if
the
answer
creates
does
create
offer,
then
it'll
actually
work
right
out:
okay,
so
yeah,
but
but
if
it's,
but
it's
the
other
side,
if
the
offering
pier
sends
another
offer,
then
it
won't
then
it'll
it.
Basically,
the
answer
will
come
back
with
both
of
the
M
lines,
zeroed
that
right
I
think
so
so
anyway,
it's
I
think
your
point
is
correct.
Henrik
but
it'll
still,
it
won't
immediately
cause
confusion,
but
it
will
cause
confusion.
Perhaps
later
yeah.
A
So
it's
the
intenser
in
a
law
and
if
the
answer
calls
transceiver
that
stop
and
then
offers
right,
that
offer
will
only
have
the
video
in
it
because
you
stop
the
audio
and
that
can
work.
If
it's
the
other
way.
It
won't
work
because
then
he'll
the
offer
will
still
offer
the
audio
in
the
video.
And
then
the
answer
will
have
to
come
back
with
nothing
because
you
stop
the
audio.
B
A
So
what
do
we
do
so
recommendation
I'd
like
to
make
is
first
try
to
if
there's
commonality
among
the
browser's
in
terms
of
what
they
do,
such
as
making
the
initial
turn
C
with
a
bundle
tied.
One
say
that
so
at
least
it's
it's
consistent
as
I
note
explaining
the
some
of
these
side
effects
and
then
recommend
that
transceiver
that
stop
only
be
called
when
you're
doing
it
on
all
the
bundled
transceivers,
because
otherwise
I
think
it's
it's
going
to
lead
to
some
surprise.
I.
F
A
A
F
A
A
A
B
B
F
F
F
B
D
C
A
A
C
And
Rick
you
have
it
okay,
so
the
eye
I
took
the
liberty
of
chain
of
changing
the
subject
of
the
of
the
Asya,
because
it
was
wrong
right.
It
was
didn't
reflect
the
content.
The
problem
is
that
our
toffee
parotid
by
the
smoke,
okay,
man
or
what
is
not
documented
if
simulcast
layers
are
different
priorities,
so
it
is
defined
well
defined
between
RTP
streams
that
come
from
different
sources.
C
It's
also
it's
not
defined
with
what
it
means
with
between
encodings
there's
Emma
gasps
and
it's
kind
of
conflicting
with
the
way
we
the
simulcast
is
thought
about
in
that
when
we
use
chemicals
and
protector
when
they
get
congestion,
we
drop
layers.
While
priority
was
described
in
such
a
way
that
all
layers
would
get
some
packets
through,
no
matter
how
rough
the
congestion
was,
but
but
don't
only
the
high
priority.
Linnaeus
will
get
all
the
baggage.
True,
so
I.
C
Put
in
too
quick
possible
prefixes,
which
is
kind
of
yeah
yeah.
Let's
not
have
this
problem
and
want
low
fix
what
the
quick
fix
makes
are
either
we
claim
that
all
encodings
in
the
simulcast
group
mathematics
a
variety
well,
that's
compatible
with
the
mostess
fiber
idea,
which
is
what
we
do
today.
It
looks
kind
of
silly.
C
Another
quick
fix
would
be
to
move
the
prior
to
the
sender,
object
and
say
that
it's
actually
a
setter
property
that
makes
compatibility,
which
is
bad
thing,
but
it
doesn't
look
a
silly
and,
of
course,
a
bit
slower
fixes
to
actually
define
what
behavior
we
want.
Auto
parity
hey
when
it's
different,
because
encodings
I
found
after
reading
through
the
slides
after
writing-
writing
this
one
that
this
actually
and
actually
has
some
relationship
to
the
to
the
issue
of
of
simulcast
that
we
encountered
a
little
further
down
decks.
C
A
A
B
C
So
that
that
was
one
that
to
the
cost
method.
Okay,
but
very
divided-
actually
make
sense,
but
not
like
not
this
Friday
so
and
Bernard
found
some
cases
where
ABI
sand
or
specs
imply
different
different
treatment
based
on
order
in
particularly
if
we,
if
there's
an
application,
asked
for
more
simple
players
than
by
support.
Then
we
drop
the
tail
ones
and
there's
a
prohibition
against
to
changing
the
order.
But
there
wasn't
a
clear
statement
that
that
the
order
in
the
in
the
encoding
by
in
their
group
parameter
was
the
order
from
the
from
the
array.
E
C
And
it
was
also
clear
that
two
there
doesn't
seem
to
be
a
best
way
of
doing
it.
So
the
proposal
is
that
the
proposal
I
wanted
to
make
is
that
order
is
consistent
and
we
stated
state
the
rule
about
simulcast
layer
subtract
it
and
with
a
rule
that
says
which
order
single
cast
the
layers
are
dropped
in
and
might
want
to
reinstate
priority
in
order
to
in
order
to
make
that
decision
and
to
influence
the
decision
that
we
shouldn't
require
any
particular
ordering.
A
C
F
C
A
A
Kind
of
makes
sense
to
drop
the
the
biggest
one
first
or
just
from
you
know
the
multi,
multiple
ikut
of
decrease
principle,
I'm,
just
worried
about
you
know
with
this
start
affecting
congestion
control
stability.
If
someone
put
it
in
the
other
rotor
and
you
started,
dropping
the
smallest
one
or
something.