►
From YouTube: WebRTC Working Group Virtual Interim June 28, 2017
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
It
doesn't
have
to
be
incredibly
detailed,
just
figure
out
what
we
what
we
said.
We
wanted
to
do
about
each
one,
okay,
all
right,
so
here's
the
list
of
things.
For
today
we
obviously
have
a
long
list
of
issues
for
discussion.
Hopefully
we
can
get
at
least
well
into
it.
We're
going
to
start
with
Yanni
bar
and
then
move
to.
Taylor
has
a
ton
of
issues
with
the
object
model
and
then,
if
time
permits,
I'll
have
the
various
miscellaneous
issues
that
I'll
I'll
try
to
get
through.
A
D
D
Basically,
the
spec
was
kind
of
vague
about
what
it
said
to
act
as
if
I
don't
see
very
similar
kind
had
already
been
created
that
wasn't
clear,
whether
that
only
affected
quite
often
behavior
or
if
it
actually
called
a
transceiver,
and
the
other
issue
was
that,
if
you
called
great
offer
more
than
once
would
offer
to
receive
the
legacy,
behavior
was
too
late.
We
called
create
offer
again
without
offer
to
risk
you
something
it
would
remove
previous
instances
where
so
that's
hard
to
implement
for
legacy
and
I
think
so.
D
Acet
text,
so
basically
whenever
so,
we
could
simplify
the
text
a
little
bit
from
four.
So
again
when
this
is
false,
nothing
happens,
but
when
offered
to
receive
video,
true
or
audio
true
is
specified.
So
the
language
is
whenever
this
is
given
a
non
false
value
and
the
career
connection
has
no
non.
Stop
send
receive
or
receive
on.
D
The
audio
transceivers
for
audio
create
offer
must,
as
its
first
step
in
both
the
equivalent
of
a
transceiver
audio
on
the
peer
connection,
object,
except
that
this
must
not
update
the
negative
goshi
ation
needed
flag
and
provide
this
does
not
sale
proceed
with
trade
offers
regular
steps
and
if
anyone
wants
to
wordsmith,
so
this
doesn't
sound
like
a
robot
talking,
that'd
be
great,
but
that's
pretty
much
it
any
questions.
Yeah.
E
So
I
mean
this
Decker
I
mean
the
entire
Pro
totus
Acree
offer
supposed
to
be
that
it
does
not
have
this
kind
of
side
effect,
so
it.
D
Right,
I
think
that
the
CR
also
renamed
sis
to
legacy
config
settings
so
I
think
and
when
it
was
discussed
an
issue
it
seems
like
we
thought
that
this
would
be
an
okay
side
effect
for
the
few
cases
where
people
still
use
this
flag.
Instead
of
since
there
is
now
the
new
sanction
way
to
do.
This
would
be
to
call
for
the
jail
client,
the
polite
receiver,
with
a
string
directly.
E
D
G
E
D
If
this
other
method
is
older,
being
called,
but
the
problem
with
only
acting
like
it
and
great
offer
is
the
new
architecture
requires
a
lot
of
things
to
exist
afterwards,
like
a
transceiver,
a
sender
and
receiver
object.
So
if
you
go
ahead
and
don't
create
any
of
those,
but
you
call
a
crate
offer
with
offer
to
receive
video
and
offer
consider
audio
true.
D
E
J
F
That
set
local
needs
to
create
transceivers
and
that
affects
data.
So
basically
I
mean
alternatives
to
fix
this
issue
or
either,
like
we
add
new
logic
in
JSF.
We
add
these
like
special
transceivers
in
what
were
to
see
one
of
those
that,
like
only
impacted,
J
sub
algorithm
but
aren't
returned
by
by
any
other
API,
or
we
do
this,
and
this
seems
like
the
easiest
alternative
for
sewing.
I,
don't
think
I'll
satisfy
the
backwards,
compatibility
and
use
case,
and
we
should
just
move
on.
E
Okay,
I
mean
I,
guess
as
I
say,
this
seems
like
a
pretty
major
of
rotational
change
adjacent
in
this
house
for
you
because
they
said
the
entire
point.
F
E
I'm,
not
I'm,
not
sure
I,
don't
I,
don't
find
that
particular
person.
I
mean
I
mean
yes,
you
can
just
you
just
say
that
such
of
sugar,
but
it's
actually
a
critical
change
in
the
meaning
of
trade
off
well.
K
F
E
It
is
like
the
only
reason
to
do
all
your
reason
to
do
this
as
far
as
I
can
tell
it
on
the
theory
that
somehow
there
are
some
legacy
device
and
some
legacy
software
out
there.
That
actually
depends
on
this
behavior
we're
dis,
calling
this
convention
is
there
any
sub
software
either
in
client
or
server
client,
or
a
provocation?
Lots.
F
A
B
F
E
F
If
you
leave
the
SSRC
field
undefined,
then
implementation
will
choose
an
SRC
on
its
own
and
it
will,
you
know,
handle
conflicts
beyond
the
required
changes
automatically,
whereas
if
you
specify
that
sexy
explicitly
then
need
to
handle
those
kinds
of
issues
yourself
and
more
to
see,
you
know
the
it's
always
the
implementation
user
agent,
that's
responsible
for
choosing
s
sorties,
so
it's
always
using
the
former
behavior.
But
yet
there
is
an
SRC
field
in
these
various
RGB
encoding
parameters
structures.
F
So
you
know
that
creates
an
issue
if
you
wanted
to
implement
web
RTC
on
in
terms
of
RTC,
and
it
also
could
cause
practical
issues.
If,
for
example,
you
call
get
parameters
and
it
returns,
the
current
SRC
is
being
used,
but
the
meantime
something
happens.
That
requires
it
to
change
and
you
call
set
parameters
with
the
old
SSR
site
parts
RT.
Does
that
drawing
valid
modification
error
or
is
the
field
just
ignored
and
I
would
say?
A
A
F
F
K
H
To
clarify,
if
I
understand
it
correctly,
basically,
what
you're
saying
is
that
there's
nothing
in
those
objects
that
would
influence
what
you
wanted
to
do
in
terms
of
creating
local
description.
So,
basically
we're
saying
if
you
get
a
you,
get
an
offer,
there's
nothing
that
would
appear
in
those
objects
that
would
affect
how
you
created
the
local
description
and
applied
it.
H
H
Yeah,
but
so,
for
example,
that
can't
be
anything
in
there
about
I
mean
I,
can't
remember
what's
in
those
up
in
the
ice,
transport
and
Nicholas
transport
objects,
but
do
they
have
like
you
know,
version
numbers
or
something
we
say
well,
hey
he's
sending
me
GCLs
wishing
one
and
I
don't
want
to
do
that.
Well,.
F
And
I
mean,
for
example,
with
ice
transport.
You
have
the
I
assume
the
remote
you
password.
Now
you
check
correctly
so
yeah.
You
could
argue
that
you
know
when
you
set
a
remote
description.
You
may
want
to
inspect
the
dice
parameters
and
use
the
I
server
interface
to
do
that.
But
in
that
case,
why
are
the
transport
attributes
notable
in
the
first
place?
Oh
I
guess
because
you
can
call
a
transceiver
and
then
this
right
so
of
Barnard
with
with
our
GC.
F
Is
it
possible
to
have
nice
transport
that
you
only
have
remote
info,
no
vocal
info,
no
yeah?
So
again
it's
an
RTP
mapping
issue,
but
I
don't
see
I,
don't
see
there
being
a
use
case
for
having
you
know
knowing
the
remote
you
guys
password
before
you
have
a
and
have
done
an
auto
answer.
No.
H
F
F
F
F
No,
you
don't
get
a
new
ice
transport
because
basically
I.
F
Isis,
you
know
written
in
such
a
way
that
when
you
do
a
nice
restart,
you
can
continue
sending
media
on
the
existing
candidate
hair
that
was
working
and
then,
as
soon
as
eyes
completes
for
you
know
the
next
generation
candidates
you
can
seamlessly
switch
to
the
same
media
on
that
pair.
So
I
would
sit.
You
know
conceptually.
They
make
sense
that
a
single
eyes
transport
object,
whereas
with
TTLs
you
know
the
details,
associations
are
independent
and
the
way
our
GC
has
written.
F
You
would
need
to
create
on
your
detail
s
transport,
for
every
detail,
s
Association,
so
there
it
makes
more
sense
to
add
Museum,
UT,
less
transport
objects,
so
then
I
would
say.
Basically
if
you
set
a
remote
description
with
a
new
fingerprint,
then
you
get
a
new
detail.
S
transport
object
that
you
know
that
already
was
one
before.
I
F
Items
are
in
simplest
cases,
yeah
well,
I.
Think
you
want
to
make
it
the
same.
That's
the
point!
So
then,
maybe
it's
that
you
only
have
DTLS
transport
when
you
set
an
answer.
F
E
Vector
I
actually
worry
about
this
issues
like
I'll,
be
inconsistent
between
ice
and
GT
OS
and
like
they
actually
have
Rolo
Livonia
in
terms
of
like
being
able
to
keep
doing
things
while
things
are
happening
and
or
they
should
and
like
the
ice
properties
are
cheating,
I
mean
I'm,
not
I
mean
device,
transfer,
brandeis
properties,
change
every
wonder
you,
including
the
parameters
trace
music
stored
nearly
off
like
like
oh,
it's,
the
same
way
strands
pour
some.
E
A
J
A
J
E
Let's
take
the
simple
case
where
you
just
changed:
a
transport
which
is
toe
reverts
to
illegal
right
I
just
do
like
restart.
Why
I
don't
understand
the
present
argument,
for
why
doesn't
create
a
new
waste
transport
it
seems
to
be.
It
really
is
a
new,
a
strange
call
from
Fresno
ice
and.
F
J
A
J
A
A
F
E
E
E
J
I
F
E
E
That's
a
really
serious
problem,
because
they're
not
yet
intentions.
Do
that
and
and
that's
why
that's
when
translation,
why
I'm
holding
to
discuss
on
this
value
is
discuss
on
the
overview
occupation,
IES
sheets,
so
we
regional
that
question
as
well,
but
I
think
this
little
bit.
This
is
relevant
to
RFC.
F
I
F
E
But
I
think
50,
buddy,
okay,
okay,
let's
just
get
a
step
back.
Db
honking
model
like
old
scribes,
has
room
for
what
I
transport
ones
and
parameters
on
price
and
and
so
like.
When
I
query,
right,
query
for
the
parameters
or
say
you
frag,
it's
got
to
say
one
thing
or
the
other
and
I've
been
doing
nice.
We
started
item
postings
corrective
those
crackers
and
either
correct,
and
so
whether
you
say
I
saw
the
objects
out,
or
you
say,
yeah
I'll,
just
just
slightly
change
their
parameters.
E
The
bottom
line
is
you're,
no
longer
access
to
the
existing
ice
context
or
your
honor.
An
axe
is
a
new
one,
so
I
mean
if
you
want
to
have
access
to
both,
but
you
have
to
eat
a
certain
data
structure
which
has
that
allows
you
to
access
both
not
when
you're
size
only
becomes
it
nice
and
UTC.
The
other
is
amundo
done
suppose.
My
part.
J
E
F
E
E
F
H
Stepping
back
and
looking
at
it
from
the
users
from
the
JavaScript
code
perspective,
it
will
be
kind
of
odd
to
have
if
you
had
already
got
hold
of
it
and
one
of
these
objects
before
you
did
a
nice
restart.
It
will
be
kind
of
weird
for
that
object
to
have
its
values.
Change
on
the
fly
because
you've
done
a
nice,
restart,
I,
think
I,
told.
J
H
H
J
E
Now,
I
really
understand
what
you
think
is
you
change
the
DCA
law
subject?
Well,.
J
J
There's
a
while
life
survival,
there's
a
real
issue
with
having
two
different
nice
transports
that
you
want
to
receive
packets
on
at
the
same
time,
which
is
what
you
want
to
do
during
a
nice
restart
and
only
and
having
a
detail,
less
transport
only
associated
with
one
of
them,
because
now
you
can't
send
the
packets
from
both
ice
transport
into
one
detail
transport.
Unless
you
somehow
represent
this
relationship
to
the
application,
which
I
think
is
even
more
confusing.
You're
saying
details
transport
as
many
as
transports.
I
J
I
F
Happen
so
if
I,
okay
as
JavaScript
I
read,
you
know
DTLS
transport
.,
you
know
whatever
the
attribute
is,
then
you
just
call
transport
and
at
some
point
in
time
it's
the
old
object
at
some
point
in
time.
Is
the
new
object?
I
guess
you
know,
and
then
do
we
say?
Basically,
when
the
new
lies,
transfer
object
becomes
connected,
cute
ask
to
change
the
attribute
or,
but
then
what
happens?
If
you
receive
with
details,
packets
from.
J
E
An
old
Kennedy
pair
after
that
right
I
mean
I,
guess
I,
guess
I'm
terrible.
What
is
it
perfect?
It
is
the
purpose
of
these,
the
purpose
of
data
structures
to
accurately
reflect
what
is
happening
on
the
wire
and
if
it
is
then
then,
like
I,
don't
understand
how
I
mean
Orion
details
in
higher
levers
and
how
accurate
I
think
is
happening.
The
wire
to
be
like
you
know
here
is
a
you
know
here.
Here
is
nice
transport.
It
contains
candidates
from
two
generation
device.
E
J
A
E
A
J
A
J
Can
I
think
we
can
agree
to
part
of
it,
which
is
that
the
details
transport
needs
to
change
at
the
point
in
time,
and
the
answer
is
important.
I
think
there
was
agreement
on
that.
It's
just
the
question
of
whether
that
triggers
I
think
the
question
is
whether
a
nice
restart
creates
a
new
I
was
trying
to
put
the
right,
but
the
part.
A
J
J
A
A
F
B
F
K
A
A
I
D
A
F
Alright,
this
is
weird,
is
song
about
okay,
so.
F
So
the
question
is
what
happened?
Basically,
there's
some
weirdness,
if
you
said,
negotiated
to
false
but
still
choose
an
ID,
because
RTC
web
data
protocol
requires
that
the
identifier
in
the
day,
open
message
is
based
on
your
DTLS
roll.
Please
clean
your
application
could
say:
I
want
in
data
channel
with
ID
zero.
F
But
then
later
your
DTLS
scroll
indicates
that
you
should
have
used
odd
IDs
and
what
happens
that
currently,
that
would
be
a
violation
of
RTC
web
data
protocol
and
I
propose
that
we
just
throw
an
error
if
the
application
happens
to
do
that,
which
is
not
really
something
they
should
be
doing.
The
first
place.
K
F
It
would
make
life
easier
and
I'm,
not
sure
why
there
are
two
fields
in
the
first
place,
as
opposed
to
just
saying
you
know.
If
the
ID
is
undefined,
then
it
does
invent
negotiation
and
the
user
agent
chooses
IDs
and
if
ID
is
defined,
then
you
do
out-of-band
negotiation
I'm,
not
sure.
Why
even
possible
to
pick
the
ID
and
doing
banned
ago,
she
ation,
but
that's
how
the
SEC
is
right.
Now.
K
F
Okay,
well,
I'm,
not
sure
about
Firefox,
but
chrome
currently
does
allow
you
to
do
this,
so
if
you
know
it
could
break
existing
applications,
I
guess
that's
true.
If
we
throw
air
as
well
but
I
mean,
would
you
propose
to
just
throw
an
exception?
If
you
have
an
IDE
that
set
button,
you
go
shade
false.
F
J
J
K
A
C
A
F
The
issue
is
that
it's
only
a
recommendation
that
you
choose
specific
SCTP
priority
values
based
on
the
enumeration,
so
implementation
could
choose
some
other
value,
and
this
integer
value
is
what
goes
in
the
day
a
channel
open
message.
So
then,
you
know
if
an
implementation
chooses
some
value
that
doesn't
exactly
match
to
then
what
priority
do
get
for
the
data
channel
created
on
the
other
side
like
more
specifically,
what
should
the
priority
attribute
return
and
what
dscp
value
should
I
use.
F
K
K
F
H
F
Well,
I:
after
reading
the
I
forget
which
it,
what
document
is
I,
think
it
was
AQS
document
I,
don't
think
we
can
actually
do
that,
because
okay
is
one
to
one
value
that
you
pass
in
to
create
data
channel
that
affects
both
and
yeah
I'll
go
unfortunate
that
they
choose,
they
chose
different
name
like
one
is
I,
think
very
low
and
other
is
below
normal
subjects.
I.
What
is
meant
to
be
the
same
thing.
F
F
C
F
All
right,
let's
see
so,
as
you
discover
here
with
the
TPMS
first
of
all,
did
we
have.
F
Then
the
the
question
I
had
was
what
happens
in
the
corner
case,
where
subsequent
negotiation
removes
that
format
does
a
communal
and
then
also
what
happens
if
you
call
insert
DTMF
on
it,
I
think
the
application
we
want,
someone
knowing
that
it
actually
did
not
get
inserted
successfully.
So
I
propose
this
for
simplicity
that
once
subjects
created
an
attribute
is
set,
it
doesn't
go
away
and
that
we
add
an
exception,
for
you
know
it
is
different.
I
will
send
it.
F
G
G
A
D
D
I
D
I
J
J
I
C
D
K
A
F
Expected
to
do
then
I'm,
fine
I
think
an
application
trying
to
send
DCMS
combs
and
not
having
the
telephone
rang.
Codec
is
pre
corn
KC
anyway,
so
okay,
I,
think
saying,
is
always
there
for
audio
senders
and
we
throw
an
exception.
If
is
not,
the
format
negotiated
probably
should
be
good
enough
for
anyone.
So.
A
C
A
F
If
not
in
the
same
description,
you
know
at
different
points
in
time.
For
example,
you
know
you
set
one
description
and
then
section
gets
rejected,
but
you're
at
the
track
again
and
there's
a
new
M
section,
but
so
uses
that
track
ID,
because
it's
on
a
different
transceiver,
you
know
there's
a
new
receiver
as
a
new
track.
So
should
you
end
up
with
multiple
remote
media
stream
tracks
that
have
the
same
ID.
F
That
seems
pretty
weird
to
me,
so
I
would
suggest,
and
one
simple
resolution
could
be
just
yes
appear.
Connection
is
trying
to
create
a
remote
media
stream
track,
and
it
already
has
one
of
the
same
ID
then
just
falls
back
to
all
the
generating
ID,
which
we
already
have
in
other
circumstances.
Anyway,.
G
F
F
J
D
F
F
D
F
J
D
Wasn't
this
why
we
switch
to
a
transceiver
mid
because
I
mean
within
a
peer
connection
that
can
be
unique,
but
if
you're
talking
about
the
actual
ID
on
the
track
or
in
the
stream,
those
objects
have
to
coexist,
potentially
with
other
strains
and
tracks
from
totally
different
peer
connections.
So
nonsense,
good
point
entry
to
be
unique.
F
A
J
A
A
So
Cullen
asked:
how
do
you
discover
what
algorithm
identify
our
values,
the
browser
supports
and
I
think
the
answer
to
that
is
you
can't
right
now
other
than
knowing
that
the
minitor
to
implement
algorithm
should
be
there
or
trial
and
error?
I
guess.
My
question
to
the
group
is:
is
there
interest
in
providing
a
solution
to
this.
I
So
it
was
an
interesting
point
here
is
that
it's
not
an
enumerable
in
the
sense
that
you
could
provide
a
complete
enumeration
of
all
the
different
numbers
and
combinations
of
values
that
are
supported
here.
So
that's
difficult,
and
it's
also
true
that
that's
important
and
relevant.
So
it's
a
pretty
significant
amount
of
effort
to
describe
to
build
a
language
to
describe
what's
actually
supported,
so
I
would
suggest
no.
There
is.
We
shouldn't
be
interested
in
doing
this
simply
because
it's
hard-
and
it's
not
really
that
bad.
I
G
A
I
G
Not
really
I
mean
it
gets
harder.
We
start
talking
about
the
curves,
for
that
I
do
not
think
that
it
would
be
very
hard
to
make
a
language
that
returns.
These
things
I
think
you
wouldn't
want
to
return
the
list
of
every
pot
like
it,
not
every,
not
the
exponential
combination
on
what
you
want
to
return.
Is
you
know?
What
are
you
know?
What
are
the
hashes?
You
support?
What
are
the
you
know
named
curves,
you
support
yeah.
H
A
E
So
one
thing
you
could
do,
though:
I'm
not
sure
it's
a
great
idea.
It
would
be
data
API
to
say
if
I
asked
you
to
get
to
dude,
we
can
certificate.
What
would
you
make
by
default,
which
I
suppose
are
like
I
mean
if
you
worried
people
are
install
time
making
or
if
we
pretty
soon,
because
that's
that
kind
of
value
point
bi,
if
you're
we're
gonna
make
you
know
easy
to
500
I
surveys
are
basically
free,
then
the
nussle,
interesting
API
I
just
told,
was
easier
to
accomplish
here.
G
H
G
I
mean
like,
if
we
look,
this
is
for
money
before
the
application
is
for
given
sets
of
customers
using
the
application,
can
set
of
preferences
set
preferences
right.
So
it's
like
rough
time,
but
if
we
could
say
hey
we're
looking
for
something
that's
supported,
and
you
know
has
at
least
sha-256.
If
that's
the
policy
that
they
want
to
enforce,
we're
only
going
to
choose
things
that
have
at
least
that
and
then
we
can
get
a
list
back
filter
those
lists.
You
suggested
back
filter
through
it
pick
something
that
looks
like
it'll
work
and
go
right.
G
E
But
I
guess
I
guess
I'm
still
like
not
understanding.
Why
Martin
suggestion
doesn't
work?
Okay,
I
mean
I,
guess
I,
don't
I'm,
not
I,
don't
Jack
Martin
I!
Don't
reject
this
proposal.
Less
important
does
but
and
I
guess
I
guess
so.
The
difficulty
I
think
is
that
the
especially
for
RSA,
basically
the
if
it
is
not
big
and
I
suppose
I
give
you
a
memorable
that
very,
very,
very
large
set.
There's
gonna
be
like
every
number
between
2048
and
15
resident
inducing
dozen
right.
Oh.
E
Do
you
easy
media
medium
immunity,
yeah,
I,
ain't
sayin,
so
like
definitely
you
guys
should
nobody's
around
opinioni
is
harder
than
any.
Probably
your
tank
you're
gonna,
like
a
bad
opinion,
so
I
think
you
know
why
I
think
either
we're
going
to
have
to
come
down
on
some
groups
and
well
rules.
Are
you
can
see
language
describe
this?
It's
going
to
be
not
be
not
it's
going
to
be
non.
Generics
may
be
specific.
E
Mainly
you
seem
to
be
on
solid,
like
RSA
is
going
to
be
anything
detritus,
late
or
you're,
an
how
to
query
I,
don't
think
I
don't
really
see
or
around
those
two
I
mean
like
this.
Like
the
language
you
use
describe,
RSA
I
mean
so
like
players
use
describe
our
essays.
All
things
language,
you
describe
DSM
a
CDSA
sure.
F
G
E
I
mean
I
know
they
was
at
ours,
I,
guess
I'm
saying
he
would
require
when
we
then
added
hash
integers.
You
would
easy
to
find
in
another
language
for
this
right.
Basically,
every
single
one
of
these
every
single
chambers
in
jar
with
are
more
acquiring
new
types
under
the
new
structure
to
striped
sure
happen.
G
E
E
Right
so
I
mean
okay,
but
some
has
to
I
mean
I,
guess
I
guess
it's
all
what's
described.
It
was
almost
like
participe
are
that
has
language
for
dislike
my
I
suppose
you
know
we
could
have
that,
but
it'll
just
adapter
that
off
I
guess.
A
E
That's
nicer,
something
else,
a
even
close
that
we
could
do,
which
is
to
say
like
non-native
language
on
and
say.
Basically,
you
know
even
shove
that
RSA
or
you
know,
or
you
saw
I
would
like
occur.
You
know
ECDSA
and
a
curve
and
like
we're
not
going
to
promise
anything
about.
Likewise,
any
other
RSA
permutation
right,
you're,
very
well,
we'll
take
same
parameters
but
you're
on
your
own,
then
we'll
be
just
saw
setting
these
structures
my
side,
you
noticed
by
the
ways
I
just
want
to
say
you
probably
read
this
transfer
PSS.
E
The
Algerian
poor
chip,
thank
you.
A
Okay,
so
here's
the
question
is
because
is
there
interest
in
pursuing
Kim's
proposal
in
a
PR
I.
A
A
I
A
E
G
F
Anything
and
so
yeah
I
can
probably
look.
E
I'm
sorry,
as
I
said,
this
is
I
mean
this
is
a
feature
in
the
JSF
spec.
Hence
it
is
a
dependency
of
university
staff
to
do
so
so
I
don't
know
if
we
give
you
more
than
MEAP
I
guess
if
you
don't
think
we
should
remove
this,
then
we
have
discussed
this.
Also
in
you
know,
in
Prague,
because,
like
this
is
a
feature
that's
been
to
working,
your
last
call
for
JSON.
A
B
A
A
So
issue,
763
and
also
1323,
is
about
the
handling
of
encoding
decoding
errors,
and
this
was
from
subway.
So
there's
a
fundamental
question
here
right
about
the
resource
model
that
we're
using
and
so
in
section
4,
3
2.
It
says:
if
it's
system
has
limited
resources,
you
know
create
offer
returns
an
offer
that
reflects
current
state
so
that
set
local
description
will
succeed
and
it
remains
useable
until
the
fulfillment
callback
of
the
return
promise.
So
the
question
is,
it
says,
remain
usable
by
at
least
until
this
can
errors
occur
later.
A
So
you,
you
know
everything's
great
inside
local
description.
Whatever
you,
you
asked
for
certain
number
simulcasts
and
you
can
you
get
it?
For
example,
you
had
resources
when
you
call
that
local
description,
but
somehow
they're
no
longer
available
when
you
call
set
remote
description,
so
you
said
I
want
to
send
for
simulcast
streams.
You
know
the
other
side
says
sure
I'll
take
them,
you
call
set
remote.
Can
can
you
get
an
error
because
somehow
you
can't
send
tor
anymore
or
apply
constraints
was
called
and
somehow
and
you,
the
height
width
and
frame
rate,
got
changed.
A
Promises
fulfilled,
but
subsequently
somehow
the
included
resources
aren't
available
or
you
call
set
parameters.
You
modify
scale
resolution
down
by
you
know
the
parameters
of
ballots
or
the
promises
fulfilled,
but
somehow
the
encoded
resources
aren't
available.
So
it's
basically
a
question
about
the
the
fundamental
model
we're
using
here
in
in
WebRTC.
You
know:
are
there
errors
these
kind
of
encoding
errors?
Can
they
occur?
Currently
we
don't
have
a
non
error
event
handler
and
the
sender,
receiver
or
an
or
to
see
for
that
matter.
So
I
think
we're,
assuming
that
these
kind
of
errors
cannot
happen.
A
A
K
Have
a
vandal
I
have
a
comment
the
as
far
as
the
set
local
description
to
set
remote
description.
I
think
you
know
the
way
it's
currently
set
up.
I,
don't
think,
there's
any
expectation
that
you
could
get
an
error
when
set
remote
description
is
called
that
the
resources
would
no
longer
be
available
there
and
that
it
would
error.
You.