►
From YouTube: Jitsi Live Stream
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
B
So
three
major
things:
we
worked
a
little
bit
on
Firefox
earlier
in
the
past
couple
of
weeks,
we
added
there
was
a
very
serious
problem
where
Richard
allocation
was
completely
broken
you,
if
a
single
just
this
single
fire
expert
is
no
was
wasn't
the
call
everything
was
getting
messed
up
and
that
could
fix
by
adding
multitrack
bit
rate
control
in
the
bridge,
which
sounds
fancy,
but
it's
not.
He
were
we
fixed
the
hack.
We
were.
B
We
removed
the
hack
that
we've
had
in
the
bridge
to
allocate
we
do
it
for
the
first
track
of
a
specific
participant,
and
that
and
Firefox
has
multiple
local
tracks
because
of
unified
plan
so
yeah
it
was
that
was
broken
for
Firefox
by
adding
the
model
track,
get
rid
control.
We
we
mitigated
this,
so
that's
one
thing,
then
yeah.
We
took
the
decision
to
disable
lip-sync
because
of
the
way
the
chrome,
synchronizes
extremes,
and
actually
it's
not
only
chrome.
This
is
going
to
be
the
standard.
B
Apparently
it's
synchronized
streams
based
on
the
RTP
play
time,
which,
in
other
words
based
on
there,
you
have
the
RTP
timestamps.
You
have
a
specific
offset.
You
have
a
specific
time
stamp
it
on
the
frequency
you
can
calculate
how
much
time
is
a
Civic
stream
has
played,
and
this
is
how
the
super
nice
room.
B
So
this
discrete
problems
with
our
implementation,
and
we
don't
have
the
time
right
now.
Well,
we
don't
have
managed
to
fix
this
issue,
so
we're
going
to
disable
and
think
because
it's
for
broken-
and
it
probably
creates
more
problems
than
it
solves,
including
on
lip
singing
itself,
specifically
yeah
who
doing
the
loops
in
itself.
B
Yeah,
so
that's,
let's
another
think
and
Lady
letter
do
the
past
week
we
have
implemented.
We
have
tuned
our
billet
allocation
algorithm
to
allocate
high
high
frame
rate
to
favor
high
frame
higher
frame
rates,
as
opposed
to
high
resolutions
for
the
onstage
participant-
and
this
is
currently
in
here
in
full
request
and
it's
being
reviewed
by
Boris,
tell
us
how
this
works,
George,
so
yeah,
so
so
the
I
you're
trying
to
allocate
for
the
onstage
participants.
We
have
five
our
granularity
levels,
so
either
you
get
720p
30fps.
B
Then
you
go
to
363
at
30fps,
then
you
go
to
180
P
at
30fps,
15
fps
over
25
years
and
well,
nothing
which
is
actually
six
granularity
levels.
So,
instead
of
going
to
the
change
here-
and
that
is
that,
instead
of
going
from
720p
30fps
to
720p
15
fps,
you
favor
the
frame
rate,
you
don't
want
to
drop
the
frame
rate.
You
drop
the
resolution
and
you
go
to
360
period
30fps
this
well.
This
is
actually
something
to
mean
that
we
need
to
determine
with
her.
B
A
I
think
that's
that's
good!
Now,
and
basically,
once
this
PR
gets
merged,
we
think
we're
in
a
pretty
good
shape
on
the
adaptivity
side,
so
we're
probably
going
to
push
out
a
stable
built
there
and
have
a
new
deployment
on
me
GT
likely
tomorrow.
Maybe
even
today
knows
alright.
Thank
you
very
much.
George
Damian
has
been
working
on
solving
a
number
of
G
gas.
He
issues
Damian.
Would
you
like
to
talk
a
little
bit
about
that.
C
So
we
have
been
working
mostly
on
G
gaseum
fixing
some
stuff
there.
There
were
a
number
of
problems
like
leaking
threads
because
in
the
trans,
when
were
when
you
launch
the
G
gas
within
translator
mode,
the
how
do
translator
was
not
close.
There
were
some
issues
with
in
a
situation
where
peer-to-peer
is
enable.
Then
there
is
one
participant
in
the
room
and
if
joining
the
chicas
participant,
you
guys
basically
receives
two
session
initiates
and
one
of
them
is
faulty,
and
this
leads
to
leaking
protocol
providers
and
memory
basically
yeah.
C
A
C
B
E
Yes,
so
actually
we're
not
sending
them
in
because
they
are
still
in
the
peer
review
process,
but
we
started
logging
on
star.
That's
like
betrayed
bandwidth,
packet
loss
and
primate
and
stuff
like
that
separately
for
jvb
and
beautopia
connections,
and
we
log
them
to
analytics.
We
plan
to
make
some
comparison
between
beautopia
and
jzb
modes.
E
A
Right,
thank
you
very
much
and
potentially
we're
looking
into
rewriting,
which
is
something
that
Baba
when
I
was
discussing
rewriting
debris
in
Java,
so
they
would
use
the
same
components
as
the
rest
of
the
GT
infrastructure,
which
should
make
it
easier
to
handle
for
people
that
are
familiar
with
the
rest
of
the
the
GT
project.
So
it's
not
going
to
be
I
mean
the
current
implementation
is
obviously
totally
valid,
but.
A
A
Lenny
yawning
fish
to
have
been
working
on
on
various
aspects
around
the
user
interface
we're
getting
close
to
final
versions
of
our
call
interfaces.
A
Could
one
of
you
guys
maybe
talk
a
little
bit
about
this,
we're
now
using
external
services
that
you
can
configure
to
consult
numbers,
dialing
numbers
and
show
them
as
part
of
the
interface
and
to
admit
I
actually
lost
a
little
bit
track,
of
which
one
of
those
we
have
already
merged
and
mature
still
in
peer
review.
But
so.
F
We
have
Mars
the
IANA,
we
have
merged
the
incoming
call
you
I
I,
it's
not
enabled
yet,
though
the
one
that's
in
a
PR
and
it's
it's
going
to
be
merge,
probably
today
or
tomorrow,
is
the
outgoing
call,
your
Y,
which
is
updated,
which
is
enabled
here
in
only
GT,
and
we
have
added,
as
you
said,
Emile
a
drop
box
with
the
countries
that
are
available
and
some
checks
on
the
phone
numbers
that
are
supported.
We.
A
Should
probably
add
MB
files
for
for
how
we
expect
those
services
to
behave
so
that
other
people
can
do
their
own
interfaces.
There
sure.
A
You
and
we
were
going
to
have
some
some
surprises
from
laying
layout
perspective
in
the
near
future,
but
we're
going
to
keep
that
a
surprise
for
for
the
near
for
the
near
future.
All
right,
let's
see
Lu
Bo,
and
so
would
you
like
to
give
us
an
update
from
mobile
and.
G
Yes,
the
weeks
I
think
merge
the
JSON
web
token.
Support
in
mobile
am
almost
close
to
having
the
rink
overlay,
which
is
GWT
power
as
well
on
mobile
and
I'm
gonna
start
looking
instance.
Competing
and
book
exists
currently
with
the
missing
beta
channel
size
to
see
videos,
even
though
I
mean
now
the
only
mode,
and
so
can
talk
about
his
progress
in
58,
which.
H
H
Know
so,
anyway,
in
the
past
couple
of
weeks,
I
work
mostly
on
the
case,
so
that
is,
we
are
making
our
mobile
applications
a
very,
very
thin
and
by
very
very
thin,
any
like
less
than
100
lines
of
code
application
that
loads
a
view
from
an
SDK
so
that
other
applications
can
embed
it.
I
have
an
open
PR
for
iOS
and
I'm
90%
done
on
the
Android,
so
there's
only
90
percent
left
and
also
work
a
little
bit.
H
You
know
this
uphill
battle
towards
getting
some
decent
chrome
build
working
to
update
from
we
where
we
seem
to
be
happy
with
58,
with
a
couple
of
back
ports
that
are
needed
for
stability
purposes,
so
that
it
doesn't
crash
like
the
desktop
thing
and
probably
when
we
have
kind
of
released
it
on
crave,
pre-release
the
plug-in,
and
we
are
internally
testing
this
build
ourselves
and
when
those
fixes
hit
the
58
branch,
if
we're
still
happy
with
them,
we'll
probably
make
you
deal
with
it.
I
think
rightly.
A
All
alright,
so
you're,
not
I,
think
that
we
can
complete
status,
update
part
of
the
comb
unless
anyone
else
from
the
team
wants
to
add
something.
D
Well,
I've
got
a
request
for
ditching
me:
you
guys
are
distributing
Debian
and
Ubuntu
binaries
that
files
per
enable
in
nine
Lee.
Could
you
guys
also
have
the
swords
builds
for
that,
because
I'm
constantly
trying
to
recruit
those
which
is
somewhat
of
a
pain
as
I
am
doing
similar
stuff
for
a
lot
of
different
projects
and
I?
Guess
you
guys
have
to
do
them
anyways
as
part
of
the
nightly
cycle
that
you're
already
doing
so?
Could
you
add
those
files
to
that
download
section?
D
D
D
Basically,
what
I
do
for
the
yoga
fire
plug-in
that
I'm
writing
is
I,
take
that
entire
source
package
and
and
now
I'm
talking
about
what
does
it
make
make
make
make
make
source
package
right
so
I
take
that
I
extract
it
and
I
replace
the
index
file
with
some
of
my
custom
code
and
that
gets
shipped
as
as
part
of
the
plug-in,
which
also
includes
the
video
bridge
key
cofo
and
some
other
components.
If
we
gave.
C
A
And
let
Charlie
this
is
a
first
step
in
money,
we'll
see
how
how
far
along
that
gets.
It
Thanks
all
right.
Thank
you
so
documentation
for
mobile
clients,
that's
a
question
from
Henry
Henry.
Could
you
maybe
talk
a
little
bit
more
about
what
you?
What
sort
of
documentation
you
mean
exactly?
Is
a
developer
documentation,
user,
documentation.
A
A
The
rewrite
you
mean
yes,
we
just
the
first
thing
would
be
to
just
get
it
to
the
Future
parity
with
the
current
version
of
debris
and
eventually,
as
you
know
now,
jibbery
also
can
work
as
a
bi-directional
thing,
so
that
you
can,
you
actually
have
Aaron
invoice
joined
from
a
sip
device
so
to
breeze
doing
the
gateway
into
to
them
and
for
some
reason,
for
me,
they
stay
in
one
ATP,
which
is
interesting
data
channels.
Don't.
K
One
note
about
how
the
jibbery
plan
is
proceeding.
The
component
we
plan
to
replace
in
Java
is
just
a
jibbery
kind
of
controlled
daemon.
The
rest
of
it's
still
going
to
be
exactly
as
is
so
you're
going
to
be
the
new
standalone
machine,
you're
still
going
to
need
an
ffmpeg
you're,
still
going
to
need
all
of
the
things
that
current
jibbery
requires.
It's
just
going
to
be
that
the
control
layer
is
going
to
be
in
Java
instead
of
Python.
So
not
a
lot
is
really
changing.
A
We
are
aware
that
jibra
is
still
sort
of
hard
to
get
running
and
we've
always
we've
had
for
awhile
a
plan
to
just
go
and
make
it
to
take
the
house
virtual
machines
out
there
that
people
can
download
and
configure,
but
time
hasn't
gotten
disclosure.
There
was
this
discussion
about
having
darker
image:
oh
actually,
that
wasn't
about
jeepers.
Even
but
the
part,
the
Challenger
docker
images
is
that
they
they
probably
aren't
going
to
solve
all
of
the
installation
hassles.
So
virtual
machine
images
are
probably
going
to
be
more
useful,
I
suppose.
K
You
think,
with
with
the
current
setup
and
the
new
Java
the
way
we
already
require
certain
pieces,
we
can
probably
get
away
with
having
Debian
packages
that
make
the
installation
at
least
a
little
easier.
It
may
not
be
perfect,
but
we
could
at
least
lead
them
through
some
some
basic
dialogues
and
do
some
requirements
of
Debian
packages
and
I
bet
we'd
get
pretty
far
towards
helping
people
along
there.
We.
A
Could
definitely
try
that
and
yeah
now
that
alsa
loopback
is
part
of
a
boom
I,
guess
that
shows
one
of
the
main
low-level
issues,
yeah
so
yeah.
So
at
some
point
this
is
going
to
come
in
the
next
couple
months,
I
suppose,
but
no
commitment,
really
it
all
depends
on
half
on
everything
else.
We
do
thanks
Sharon,
so
any
other
questions.
A
J
K
Let's
talk
about
what
you're
talking
about,
because
the
the
changes
that
we
need
for
multi
subdomain
have
already
dropped
for
jibbery,
and
it's
really
just
the
parsing,
the
conference
muck
that
it
gets
back
from
jacoco
on
a
request
to
join
a
room
and
concocting
a
URL.
That
includes
a
subdomain.
So
if
you
want
to
talk
about
how
to
do
a
deployment
of
jitsi
with
multi
multi
sub
domain,
we
can
talk
offline
about
it.
But
the
the
support
needed
in
ghibli
should
already
be
there
and.
A
There's
probably
it's
probably
worth
partly
not
Conni
around
that
we
are
actually
departing
from
the
notion
of
subdomains
and
jitsi,
for
one
main
reason
that
with
WebRTC
you,
when
you
do
a
subdomain
based
architecture,
you
basically
put
yourself
in
the
very
bad
position
to
have
to
do
a
nude,
a
new
Chrome
extension
for
every
subdomain
that
you
do
for
desktop
sharing
and
the
whole
point
of
subdomain
was
that
you
could
have
unlimited
subdomains.
That
was
the
point
of
our
of
our
subdomain
support
and
that's
unachievable,
basically
because
of
the
comm
extension
problem.
A
K
Get
the
support
and
Ripper
is
already
there
in
terms
of
the
parse
parsing
out
of
the
sub
domain,
but
the
the
setup
includes,
like
a
pro
CD
plug-in
and
includes
some
nginx
configure
it's
a
bigger
change
than
just
a
some
support
and
ribéry.
So
if
you're
interested
in
that
kind
of
set
up,
let's
talk
about
it.
Offline.