►
From YouTube: 2022-05-04 Monthly Multicast W3C Community Group Meeting
Description
Notes here: https://docs.google.com/document/d/1eoGvNcbh26wUPEGZOHJWzd4Y1gKZIAUx3_0NWaghNwg/edit
Meeting series here: https://github.com/w3c/multicast-cg/tree/main/meetings
A
All
right
recordings
on
thanks
everyone
for
coming.
I
was
a
little
bit
hoping
that
dom
and
eric
would
make
it
today.
Oh
and
kyle
yeah,
here's
kyle,
hey
kyle,.
A
Like
I
was
saying,
I
was
a
little
bit
hoping
dom
and
eric
would
make
it
today.
There's
one
item
I
wanted
to
add
to
the
agenda,
which
was
which
was
I
I
actually
got
a
ping
offline
from
ramp
from
giovanni
at
ramp.
A
I
should
say-
and
I
asked
him-
I
think
I'll-
be
asking
a
few
others
who
have
shown
interest
in
this
work,
because
they
want
to
use
it
to
contribute
to
a
use
cases
document
where
you
know
what
we,
what
we,
what
I
want
to
gather
is
is
like
a
document
where
we
have
statements
from
a
number
of
companies
that
are
interested,
saying,
here's
what
it
would
mean
to
us
if
multicast
was
in
the
browser,
I
wanted
to
start
collecting
those
so
that
we
can
show
people
at
like.
A
Right
but
since
they're
not
here,
I
guess
we
might
skip
that
part.
I
want
I
wanted
it
in
there,
but
maybe
I'll
just
do
it
next
time
that
they
do
show
up
or
send
them
offline
messages.
A
So
I
guess
the
the
bulk
of
what
else
was
on
the
agenda
was
just
kind
of
going
over
where
we
got
to
on
on
sort
of
getting
started
with
the
you
know,
the
specification
and
a
demo
that
will
that
we'd
like
to
run
I'm
really
happy
with
the
progress
thanks
everyone
who
worked
on
it.
That
was
it's
it's
refreshing
to
see.
You
know
a
bunch
of
check-ins
happening
and
it's
been
a
lot
of
work.
This
happened,
I
think
so.
That's
that's
great.
A
And
I
guess
what
do
we
do?
I
put
all
the
items
in
there
that
I
wanted
to
say
happened
so
yeah,
it's
been
productive
thanks
to
mac
leo
and
max
leo
and
kyle.
We've
got
a
bunch
of
check-ins
happening.
We
made
a
fork,
we're
you
know,
adding
stuff
in
there
we've
been
max
especially
has
been
really
helping
to
tighten
up
the
initial
draft
specification
dock
for
the
quick
extension
I'm
a
little
worried.
The
quick
working
group
is
gonna,
think
it's
too
complicated
and
they're
not
gonna
want
it.
A
I
think
I'm
intending
to
propose
it
as
experimental,
which
I
think
is:
okay,
oh
and
anything
else.
We
should
really
get
into
I'd
almost
like
to
just
sort
of
open
it
up
as
a
as
a
rest
of
the
working
session.
Unless
we
have
questions
like
things
that
we
would
that
we
would
next
address
in
the
next
working
session,
we
have
things
like
clarifying
what
the
demo
will
look
like.
A
A
I,
I
think,
getting
the
push
api
accepting
some
data
first
is
like
the
first
step
of
that,
but
but
we
can
just,
but
I
think
it's
something
that
can
proceed
in
parallel
as
soon
as
there's
something
that
can
forward
through,
like
it
doesn't
have
to
have
the
multicast
transport
part
working
in
order
to
have
a
useful
demo
app
that
runs
in
a
browser
that
can
receive
pushed
data
from
you
know
from
the
server
initiated
unidirectional
data.
The
way
we're
trying
to
to
send
so,
I
guess
yeah
kyle
max.
B
A
I
want
to
finish
the
frames
there's
like
still
four
or
five
of
them
missing.
I
think
I
did
all
the
control
frames
right
unless
the
count,
I'm
not
really
sure
what
distinguishes
a
control
frame
from
from
a
not
control
frame,
the
stream
boundary
might
be
a
control
frame.
I'm
not
sure.
I
think
it's
about
whether
it
goes
whether
it
contributes
to
the
to
the
flow
control,
because
it's
not
part
of
a
stream
or
something.
A
And
also
like
the
the
stream
boundary
frame
that
maybe
should
be
control,
I
thought
maybe
it
was
associated
with
the
stream
sufficiently
that
that
it
shouldn't,
but
anyway,
so
there's
still
some
more
frames
to
add.
So
I
was
going
to
finish
doing
that.
I
took
a
little
a
quick
little
break
to
to
take
a
look
at
the
at
the
push
command
because
because
leo
had
some
questions
and
was
that
a
reasonable
kind
of
here's,
something
that
kind
of
works
or
kind
of
does
something
I
wanted
to
get
that
integrated.
A
But
I
gotta
get
back
to
the
frames
and
finish
those
up.
I
think,
but
then,
after
that,
we
gotta
start
like
doing
logic
with
the
frames
or
something,
and
we
also
have
to
get
inbound
push
data
right.
So
those
are
kind
of
the
two
keys
for
getting
this
part
working.
But
then
the
other
piece
is
going
to
be
dependent
on
the
inbound
push
data
and
that
would
be
doing
something
in
a
web
transport
web
app
which
which
would
be
able
to
take
the
multicast
as
well
yeah.
B
A
That
yeah,
it's
the
it's
the
quick
e-pull
server
stuff
so
that
that's
basically
a
sort
of
event
driver.
A
A
B
Yeah
yeah
no,
but
I
felt.
E
B
A
A
You
know
give
it
the
the
sock
opt
call
that
that
does
the
join
and
and
stick
the
read
on
the
into
the
into
the
e-pull
server
for
the
client.
I
think.
A
Actually,
I
didn't
look
at
that
on
the
client
side
yeah.
That
is
a
good
question,
especially
because,
like
as
it
stands
now
we're
not
we
don't
know
if
it's
the
epo
or
not
right,
probably
isn't.
A
Okay,
so
we
should
decide
for
the
client
side
whether
we're
gonna
do
the
yeah.
We
probably
have
to
do
the
simple
server
if
we
or
the
simple
okay,
so
I
don't
know
how
it
works
on
the
client,
you're
right,
okay,.
B
Sorry,
no
last
thing:
okay,
so
decline,
but
shouldn't
it
be
inside
the
kitchen.
Implementation
like
shouldn't
the
kitchen
implementation
open,
the
sockets.
A
So
it
sort
of
takes
it
outside
of
the
it's
like,
not
part
of
the
core.
I
think
where
the
sockets
are
maintained.
So
there's
like
it's,
it's
like
one
of
the
test
tools
is
sort
of
how
it's
structured
right.
So
the
client
itself
is
a
test
tool.
You
know
once
it's
inside
chromium,
then
it'll
be
something
something
totally
different
again
right.
So
the
idea
is
that
it's
going
to
be
like
most
of
the
quick
core
library
stuff
is
about.
A
You
know
once
you've
got
a
packet,
then
how
do
you
interpret
it
and
all
that
kind
of
thing
right?
But
then
they
do
include
some
drivers
and
there's
a
few
different
drivers
right.
There's
the
the
simple
server
and
the
e-pull
server
that
are
that
are
both
sort
of
treated
similarly
but
have
a
different
way
that
they
get
kicked
off
and
started
depending
on
which,
which
executable
you
build
or
you're
running
right.
A
A
A
And
kind
of
post
links
to
where
you
think
is
the
right
thing
or
or
make
a
a
you
know
a
prototype.
I
think
you
can
go
ahead
and
put
something
in
that
response
to
the
join,
but
we
do
have
a
join
frame
there
and
you
can,
you
know,
make
it
make
it
send
that
as
the
sort
of
initial
initial
thing.
A
As
a
sort
of
10-pack
kind
of
a
deal
but
yeah,
I
think
you
could
start
experimenting
with
that
and
just
see
what
works
that'd
be
great
cool.
Okay,
that's
a
good
next
step,
all
right
now,
kyle.
C
C
I
don't
know
whether
we're
meeting
on
friday
because
of
the
the
wellness
day
or
not,
but
but
if
we
do
then
then
I'll
make
good
progress
before
then.
Okay,.
A
But
you
did
see
you
saw
the
pr
where
I
integrated
leo's
stuff,
so
that
the
protobuf
messages
for
the
for
the
push.
A
I
took
his
branch
and
put
it
in
so
that
it
links
against
the
the
cue
push
driver
and
also
against
the
you
know
it
doesn't.
It
still
doesn't
do
anything,
but
it
it
has
the
capability
to
like
make
those
protobuf
objects
and,
and
then
yeah,
the
the
so
you're
going
to
look
into
making
it.
C
A
grp
is
it
or
you
know.
If
that
doesn't
look
viable,
then
you
know,
try
and
get
it
to
use
a
unix
domain
socket
or
something
like
that.
I
mean
just
come
up
with
a
come
up
with
a
solution
and
get
the
messages
being
transported
into
the
into
the
the
quick
server.
C
A
C
Well,
no,
I
mean
you
know
a
pipe
is
a
natural
thing
to
to
use
if
for
for
for
blocking
stuff,
but
for
non-blocking
things
I
it
just
like
they.
There
are
so
many
weird
corner
cases.
I've
found
with
with
pipes
that
it's
it's
you'll
end
up,
you'll
end
up
with
with
perverse
like
you
know,
corner
case
behavior
and
that's
not
something.
You
want
a
server.
A
Yeah,
I'm
not
sure
if
the
stream
data
should
be
treated
separately
from
the
from
all
the
command
stuff
like
it
might
be.
Fine
to
just
sort
of
have
a
way
to
to.
You
know,
open
a
streaming
context
and
then
send
it
things.
A
A
C
I'm
still
yeah,
I
mean
I'm
still
trying
to
figure
out
grpc
right.
It
may
be
that
there's
a
natural
way
to
do
this
with
that
that
doesn't
involve
proto
buffs.
It's
unclear
to
me
right
because
grpc
and
protobuffs
are
you
know.
Basically,
you
know
two
halves
of
the
same
thing.
So
so
if
it
turns
out
that
there
isn't
or
if
there
isn't
an
accepted
pattern
for
this
kind
of
thing,
or
if
the
pattern
is
hey,
you
should
open
up
a
socket
and
stream
the
data
over
then
we'll.
C
You
know
then
we'll
do
that
instead
sure
I
can.
You
know
I
can
refer
to
some
old
ftp
rfcs,
for
you
know,
inspiration.
A
Let's
not
ftp
is
really
special,
okay
yeah,
I
guess
so.
It
looks
like
it
looks
like
my
appeal
for
use
cases.
I
don't
have
my
my
audience
here
yet
if
I
I'll
I'll
probably
send
an
email
then
to
to
people
like
eric
and
dom
and
see
if
I
can
get
them
to
contribute
just
a
you
know
a
few
paragraphs
about
like
why
it
would
be
helpful
for
them
and
if
you
think
of
anybody
else,
you
know
I
I
sent
something
to
ramp
already.
A
And
then
you
know
anybody
else,
you
know
of
that.
Could
meaningfully
contribute
to
that.
You
know
maybe
ping
them
send
them
to
me.
B
Did
you
contact
richard
or
sam.
A
Oh,
I
did
yeah,
yeah
and
and
lucas,
so
sam
actually
made
some
some
good
comments.
Also
on
the
on
the
the
latest
spec.
The
answer
I
got
was
they
probably
won't
be
authors
on
this
one.
You
know
they're
they're
not
expecting
to
be
authors.
We,
you
know
that's
a
bit
of
an
independent
question
of
whether
we
end
up
collaborating
but
yeah
they're,
aware
of
it,
and
sam
has
been
commenting
on
the
on
the
multicast
on
this
multicast
extension.
A
I
could
continue
to
do
that.
He
had
some
good
comments
too
lucas.
I
still
want
to
hear
from
he
did
say
he
wanted
to
to
be
involved,
but
I
haven't
seen
any
evidence
that
he's
managed
to
make
the
time
to
read
through
the
document.
Yet
it's
in
its
more
current
form.
A
So
I
did
say
that
if
I
hadn't
heard
back
today,
I'd
probably
go
ahead
and
post
it.
So
I
think
I
will
you
know
towards
the
end
of
the
day,
submit
this
to
the
ietf
data
tracker.
A
So
I'll
have
a
0
version
of
it
that
we
can
that
we
can
start
kicking
around,
but
you
know,
I
think,
there's
still
a
ways
to
go
before
it's
ready
for
you
know
to.
A
So
I
guess
maybe
what
I
should
say
is
towards
the
end
of
this
month.
I'd
like
to
be
in
a
place
to
get
a
few
other
people
from
the
quick
working
group.
To
take
a
look,
I
think,
we're
you
know
we're
the
way,
we're
fleshing
it
out.
It's
going
pretty
well,
I
think
we've
made
a
lot
of
improvements
and
there's
there's
decent
stuff
there
they're
going
to
hate
how
complicated
it
is.
I'm
pretty
sure
it's
you
know
it's
like
10.
A
New
frames
is
like
half
of
what's
already
in
quick,
just
by
frame
count
and
I
think
we're
we're
adding
another
one
too
right,
11
new
frames.
They
had
19,
I
think,
for
the
initial
draft
and
then
there's
like
two
more
going
in
for
multipath
and
there's
one
more
that
went
in
for
the
datagrams.
A
You
know
so
this
is
a.
This
is
a
pretty
heavy
extension.
You
know
in
terms
of
just
how
much
it
adds
to
the
functionality
and
things
that
you
have
to
parse.
That's
part
of
why
I
want
to
make
it
experimental.
I
think
it'll,
take
a
few
rounds
of
it'll.
Probably
take
some
operational
experience
before
we
can
say
with
confidence
that
it's
that
it's
good
the
way
it
is-
and
I
guess
that's
the
other
thing-
I'm
still
working
on
trying
to
get
a
few
events
done
with
multicast.
A
There's
a
lot
of
details
to
work
out
and
people
figuring
out
which
kind
of
event
and
which
kind
of
network
are
gonna,
be
you
know
what
what
timeline
exactly
but
we're
trying
to
pin
down
few
actual
events
or
two
to
start
with
for
multicast
delivery.
A
A
A
Probably
just
to
add
encryption,
the
way
ambi
does
it
or
not,
not
encryption,
just
to
add
the
integrity.
The
way
ambi
does
it
and
maybe
to
skimp
on
that.
For
for
the
first
couple
of
events,
I'm
not
really
sure
I
would
strongly
discourage
that
because
I
don't
I
don't
want
traffic
injected
really,
but
you
know
in
in
general.
I
think
this
will
be
better
received.
A
And
that's
part
of
why
it's
worthwhile
and
part
of
you
know
part
of
the
sort
of
browser
insistence
that
that
that
drives
this
kind
of
work,
I
think,
is
like
we.
We
do
want
to
fit
that
model
and
be
like
properly
encrypted.
A
You
know
I
mean
multicast
as
as
we've
got
as
our
security
considerations.
Draft
you
know
pointed
out
does
come
with
some
special
issues,
there's
a
sort
of
risk
of
key
exposure.
That
scales
with
the
you
know
the
breadth
of
your.
A
You
know
your
client
receive
audience
that
gets
the
same
key,
so
well,
it's
important
to
acknowledge
that.
It's
also
the
case
that
you
know
the
key
does
not
necessarily
get
exposed
so
having
it
encrypted.
A
You
know
I
I
I
yeah
that
that
feedback
from
like
a
year
ago,
now
on
the
on
the
quick
or
on
the
on
the
chrome
net
dev
list
was
well
founded.
I
think
so
all
right
anything
else.
We
should
go
over
then
today.
A
And
I'll
be
asking
for
use
case,
stock
submissions
and
yeah.
If
you
know
anybody
else
ask
for
this.
A
Okay,
if
there's
nothing
else,
then
I
think
we
will
wrap
up
early
thanks.
Everyone
for
coming
and
I'll
see
you
next
time.