►
From YouTube: Jitsi Community Call - February 11, 2019
Description
The Jitsi Community Call is held every other Monday at 10:30AM Central US time at https://meet.jit.si/TheCall. The Jitsi team provides updates and community members can ask the team questions.
See https://community.jitsi.org in the mean time for questions.
Subscribe for our new videos: http://bit.ly/2QVLZCG
Learn more About Jitsi at https://jitsi.org
Try Jitsi Meet now at https://meet.jit.si
See the code at https://github.com/jitsi
A
A
And
we're
life.
Thank
you.
Everyone
for
joining
this
week's
edition
of
our
community
hope
we're
going
to
go
as
usual
through
a
brief
update
of
the
most
recent
changes
that
we've
seen
happening
recently
and
then
we're
going
to
open
the
microphone
for
mensen
questions
from
the
four.
So,
let's
go
quickly
through
the
updates.
There
have
been
a
few
fixes
in
master
things
around.
There's
there's
this
bug
that
some
of
you
may
have
noticed
around
the
tutsi
displayname
not
being
properly
shown
and
people
saying
hello
tipster.
A
B
That's
just
a
few
more
notes,
ones
that
we've
been
working
on
the
connections
of
this
API
on
Android,
so
that
will
be
also
in
the
next
release,
which
is
my
next
note.
We
currently
don't
have
releases
into
stores.
You
can
find
every
every
update
in
master,
but
we
don't
have
a
store
release.
Yet
we
plan
to
have
it
probably
this
week,
but
timing
may
change.
A
Moving
to
other
components,
Damian
has
been
working
recently
on
a
number
of
very
significant
fixes,
with
de
casa,
so
again
make
sure
that
you
update
your
G
gosse's
he's
currently
joining
through
jothee
as
well,
so
we're
seeing
much
better
performance.
Also,
on
a
slightly
different
note.
We
have
the
company
that
provides
us
with
the
PSN
connection
service
for
this
coal
box.
A
A
C
It's
a
pure
bug
on
our
side,
where,
if
a
conference
goes
to
using
two
bridges
and
then
just
using
one
bridge
and
then
stays
in
this
state
for
a
few
minutes,
some
of
the
channels
timeout
and
then
the
communication
between
the
bridges
is
broken.
We
have
a
fix
for
it
in
master
already,
but
it
hasn't
been
employed
immediately.
So
you
might
still
experience
the
bug
and
if
that
happens,
just
choosing
your
conferencing.
A
And
a
good
way
to
recognize
it
is,
if
you
have
Iowans
people
communicating
to
each
other
and
that
sort
of
seemed
to
match
to
your
graphic
location.
Then,
yes,
that's
the
bunny
experiencing.
This
should
be
Boris.
Correct
me
if
I'm
wrong,
but
this
should
be
one
of
the
things
that
are
fixed
with
the
release.
We
were
planning
yep,
probably
even
the
most
important
one.
A
There
was
by
the
way
another
issue
that
we
were
recently
looking
at
with
regard
to
DCP
I,
see
TCP
streams
connected
to
the
bridge
and
ice
performance
or
lack
thereof,
and
that's
something
that
we
were
being
severely
impacted
by
in
the
past
couple
of
weeks.
Boris.
Would
you
like
to
talk
about
this
one
too,
because
I
don't
think
we've
discussed
it
on
the
call
sure.
C
C
So
when
we
did,
we
discovered
that
some
of
these
thread
some
of
these
changes
that
reduce
the
number
of
threads
cause
problems
when
we
were
using
ice
TCP
problems,
meaning
the
all
of
the
available
threads
in
the
pool,
gets
stuck
right
into
a
TCP
socket
which
results
in
ice
failing
for
everyone
else
on
the
bridge.
Because
of
this
we've
reverted
the
ice
for
Jay
version
end
video
bridge.
A
D
D
Anyone
who
is
security,
minded
I'd
love
a
review,
because
that
would
be
great
from
a
security
perspective,
not
necessarily
perspective
of
from
the
DTLS
perspective.
But
that's
the
only
update,
I
don't
have
any
update
is
to
win
this
week,
won't
make
it
into
master,
but
certainly
some
point
both
hopefully
well
before
from
74,
it's
stable
as
that's
when
there
deprecating
support
for
well,
not
know.
A
D
Sure
we're
still
hoping
for
a
master
at
the
end
of
March
I've,
a
lot
of
the
majority.
If
the
the
future
work
is
done,
Forrest
is
going
to
be
working
on
some
small
things
and
bringing
octo
back
on
the
new
bridge.
That's
one
of
the
notable
things
that's
missing
and
some
of
Georgia's
most
fixes
have
not
been
ported
yet
I'm
in
a
mode
of
going
back
in
doing
some
kind
of
memory,
analysis
and
general
performance
analysis
and
and
taking
care
of
some
of
the
kind
of
scaffolding
code.
D
A
E
A
A
Switching
to
do
and
there's
this
bug
specifically
around
you
appear
to
be
using
one
device
but
you're
actually
using
another,
and
then
you
have
to
go
and
dis
elected
and
recently
did
in
order
for
this
to
actually
happen
so
or
plugging
in
your
devices
and
like
we,
we
don't
actually
show
them
to
me
immediately
or
things
like
that.
So
there's
a
good
chunk.
The
work
that's
going
to
happen
soon
as
part
of
this
chunk,
we're
also
going
to
make
it
possible
for
API
users
to
set
device
that
is
from
the
outside.
A
This
is
specifically
useful
when
you're
planning
and
using
when
you're
planning
an
integration
of
the
GT
me
API
in
an
app
that
has
its
own
device
management,
central
device
management,
so
that
would
be
good
place
or
you
have
an
integration
in
an
app
that
doesn't
necessarily
have
other
audio
calls,
but
really
just
thing
that
audio
video
device
selection
should
happen
in
your
global
preferences.
Then
this
would
also
like
to
do
that.
A
There
has
been
also
significant
progress
on
spot.
Nothing,
we're
not
yet
at
the
point
where
we
would
be
encouraging
the
community
to
tested
ready,
it's
progressing
very
nicely,
so
everyone
get
ready
for
it.
Get
ready,
get
your
communities
ready.
You
tune,
have
native
room
support
with
with
duty,
whether
it's
what
are
using
midgets
or
whether
you're
having
your
own
deployments
you'd,
be
able
to
support
rooms
with
their
own
calendars
and
such
that's
all
I
had
for
for
the
team.
Does
anyone
from
the
team
want
to
add
stuff
that
I
doubted,
lay
forgotten.
A
A
F
A
G
A
Need
in
order
to
get
something
to
understand
that
this
evening
and
talk
to
PSTN,
but
yes,
the
PSTN
provider.
For
me,
jitsi
is
it's
vaccine
plant
and
specifically
one
of
the
interesting
things
there
is
that
you
have
support
for
SSRC
mixing.
So
you
don't
need
the
gas
e
to
actually
go
in
and
do
the
audio
mixing.
You
can
just
send
all
the
streams
to
back
to
your
supervisor.
That
way,
right.
F
When
people
dial
in
like
through
a
landline
into
a
jitsi,
meet
you're,
actually
paying
per
minute
for
them
to
be
able
to
do
that
correct
with
Box
pricing,
that's
correct.
Okay
and
my
last
question
is
I-
have
some
questions
that
I
don't
need
to
necessarily
ask
in
this
call,
but
I
wanted
to
see
if
I
could
reach
out
to
someone
that's
kind
of
the
Guru
on
the
mobile
side.
I
have
some
things
we're
trying
to
work
on
and
before
we
go
down
a
certain
path.
F
A
F
F
It
real
quick
I
mean
we
already
have
a
web
RTC
solution
for
one-to-many
type
things,
and
we
have
we've
integrated
the
Google
Translate
into
it.
So
it
allows
people
say
in
Germany
that
wanted
to
watch
a
webinar
to
be
able
to
select
German
type
in
German
the
people
it
would
be
in
another
language
say
the
United
States
would
get
it
in
English,
so
everybody
thinks
everybody
else
is
typing
in
their
languages.
In
this
translation
we
wanted
to
be
able
to
do
that
with
the
jitsi
app
and
we
didn't
easily
do
it
with
the
browser-based
version.
F
A
B
We
currently
have
it
in
the
in
the
in
the
reader
state
that
comes
as
signaling
from
GGC,
and
then
we
have
a
component
that
is
called
there.
It
actually
future
called
parens
scription,
which
you
can
check
out.
The
entire
code
is,
is,
is
box
there.
Nowhere
else,
it's
not
that
long.
So
it
is
it
done.
We
basically
put.
F
B
F
And
yet
answer
your
question.
We
would
want
to
translate,
of
course,
the
closed
captioning,
but
we
would
also
want
to
be
able
to
have
someone
to
type
in
their
native
language
and
then
it
get
translated
into
whatever
language.
Other
people
select
that
that
technology
exists,
but
I
didn't
know
how
to
integrate
that
into
the
mobile
app
and
I
didn't
know,
control.
A
F
A
A
Just
say
that
word
in
the
XMPP
server
that
we
use
is
called
prosody.
You
know
as
in
the
poetic
thing
and
and
it's
written
in
lua,
and
it
has
a
very
lightweight,
very
easy
to
use
module
mechanism.
So
you
can
just
not
easily
add
new
modules.
They
have
access
to
the
messages
that
are
being
chained,
exchanging
chat,
you're
only
interested
in
chat,
I
I
would
start
there,
because
you
don't
need
to
worry
about
any
specifics
around
mobile
and
browser
yeah.
Basically,.
I
C
Don't
have
a
full
implementation,
yet
we
started
work
on
it.
You,
google,
Summer
of
Code.
Last
year
we
have
the
transcription
part
which
takes
the
text
that
the
audio
passes
it
through
Google
Google's,
voice-to-text,
API
and
there's
a
closed
captions.
We
were
also
planning
to
have
a
module
which
this
translation,
but
we
haven't
finished
with
that.
So
that's
another
way
to
go
and
potentially
could
also
do
the
chat
this
way,
but
obviously
there's
multiple
ways
to
do
the
chat.
J
This
is,
this
is
Louis
I
told.
A
J
Had
two
questions,
one
I
think
may
actually
be
resolved
with
the
fix
you
mentioned
earlier,
where
somebody
calls
in
via
juga,
see
dial
em
bridge
and
the
it's
not.
The
display
name
isn't
always
set
to
the
user's
phone
number.
Sometimes
it's
empty,
and
sometimes
it
has
the
phone
number
and
it's
non-deterministic
yeah.
J
A
So
the
problem
is
that
I'm
trying
to
remember
it
was
something
to
do
with.
We
wanted
to
make
sure
that
we
owe
to
expire,
gee,
gassy
science.
Obsessions
people
are
hung
up
on
the
on
the
PSC
inside
and
we
want
to
make
sure
that
they
don't
linger
behind
yeah
I'm
drawing
a
blank.
Does
anyone
remember
why
we
we
did
it
that
way?
A
J
So
that
there's
no
there's
no
way
around,
that's
fine
I
just
wanted
to
make
sure
I
wasn't
doing
so.
A
A
K
K
L
A
G
A
From
browser
it,
just
you
would
sit
there
and
in
order
for
Jack
offer
to
do
to
connect
anything
to
anyone,
there
needs
to
be
at
least
two
people
that
are
talking
to
each
other.
So
my
recommendation
for
addressing
this
would
be
to
see
if
maybe
an
easy
way
to
address
it.
I'm
I'm,
I'm
kind
of
shooting
from
the
hip
here
so
not
I,
not
necessarily
have
the
details
of
how
this
would
work,
but
maybe
you
could
Park
the
call
on
your
server
get
some
music
on
hold
there
and
then
track
the
200.
J
A
Know,
you're
you're
you're,
basically
waiting
for
jiggity
to
send
back
a
200.
Ok,
so
you
could.
You
could
connect
your
user
and
on
the
PSTN
or
on
your
sip
line.
You
could
connect
your
user
and
and
send
them
early
media
or
something
like
that
and
then
once
you
get,
the
200
back.
Ok
from
juga
see
you're
good
and
you
can
connect
them
directly.
J
A
M
D
I
D
L
L
A
Not
not
really
we're
think
it's
problem,
it's
it's
it's
one
of
the
things
that
we
should
start
tackling
within
the
next
couple
of
months.
Okay,
but
we're
not
really
rushing
it.
Given
that
I
was
checking
the
other
day,
even
Google
Mead
seems
to
be
using
Plan,
B
and
yeah.
We're
not
we're
not
really
in
the
rush
right
now.
But
yes,
it's
lovely!
Oh,
do.
A
N
Good
good
happy
Monday,
so
we've
made
a
lot
more
progress
in
our
yeah
yeah.
Well,
we
make
the
best
of
it.
We've
made.
We
made
a
lot
more
progress
on
getting
our
recorder
RTP
and
pull
working
in
jvb
with
some
of
the
changes
we've
talked
about.
So
a
couple,
quick
questions
that
things
we've
run
into
so
so
one
is
there.
Is
there
a
way
like
a
consistent,
repeatable
way
of
adjusting
the
quality
of
opus?
So
you
know,
I
looked
at
the
opus,
RTP,
spec
and
I
know:
there's
been
with
settings
that
we
can
set.
N
N
Right
now,
just
Chrome,
yet
so
yeah
so
we're
it's
not
even
clear
whether
chrome
will
actually
obey
the
the
bandwidth
settings
or
any
of
these
other
settings,
but,
and
then
also
I
was
curious,
whether
you
know
within
the
I
forget
like
the
source,
one
of
the
packet
jingle
packets
or
spected
info
extension,
whether
the
bandwidth
settings
there
is
is
something
that
we
should
be
leveraging
to
shut
the
bandwidth.
You
know
we're
gonna
minder
saying.
Is
that
what
you
can
be
dynamic?
A
I
I
think
that
Simon's,
that
you're
hearing
and
please,
if
every
want
anyone,
wish
to
say
something.
Please
just
interrupt
me,
but
so
far
what
we
found
is
that
this
is
very
obscure.
We've
needed
to
set
parameters
in
opus
and
some
of
them
used
to
be
with
FAQ.
We
wanted
effect,
always
be
enabled,
and
we
have
this
parameter
in
there,
but
it
seems
that
at
one
point,
problem
just
started
ignoring
that
and
they
started
playing
some
more
complex
logic
about
one
FAQ
is
honor
enough,
don't
specifically
know
about
bandwidth.
A
N
A
N
Okay,
just
curious,
so
so
the
other
question
is
we
actually
perchance
came
across
I.
Guess
Boris
is
all
that
old
thesis
from
five
years
back
that
that
talks
about
the
record
or
achieve
people
which
is
really
really
helpful.
Really
useful
document
doesn't
better
understanding
the
strategy
there,
because
there
were
a
few
things
that
I
just
wasn't.
I
wasn't
fully
clear,
I'm
like
exactly
what
the
rationale
for
the
silence
effect
was
or
packet
buffer.
So
one
thing
that
that
I
think
we
were
a
little.
N
According
to
that
document,
it
says
that
the
packet
buffer
basically
makes
the
jitter
buffer
kind
of
useless
like
the
packet
buffer
is
basically
the
makes
it
is
effectively
the
jitter
buffer
and
is
significantly
large
enough
to
be
able
to
handle
and
address
some
of
the
issue
that
we
were
running
into
yeah.
We
could
really
I
guess.
The
question
is
like
we
from
from
our
experiments.
N
We
couldn't
really
replicate
that
type
of
behavior,
so
if
we
we
can
increase
our
jitter
buffer
and
that
was
helping,
but
we
couldn't
getting
them
both
to
work
together,
so
the
packet
buffer.
If
we
increase
that
to
be
significantly
large,
the
jitter
buffer
being
smaller
coming
before
it
would
cause
problems.
So
I
was
wondering:
is
there
something
that
we're
missing,
because
it
doesn't
seem
that
there's
a
way
to
basically
effectively
disabled
the
the
jitter
buffer
and
increase
the
packet
buffer
and
have
the
pack
basically
be
the
main
buffer
for
for
RTP
packets?
C
This
should
cause
fmjs
buffer
to
stay
small
because
it's
adapt.
If
and
it
grows
when
it
sees
delays
and
loss.
There
are
some
properties
that
you
can
use
to
set
the
behavior
of
the
jitter
buffer.
You
can
look
them
up
if
you
want
I,
think
there's
one
to
actually
disable
the
adaptiveness
and
certain
set
of
things.
Sighs.
L
O
Then
I'm
just
gonna
add
some
more
so
yeah.
What
we've
seen
is
that,
if
the
jitter
buffer
is
not
cannot
grow
enough,
then
it
still
drops
frames.
So,
even
though
the
packet
buffer
is
can
be
very
large
if
the
jitter
buffer
is
a
little
bit
smaller
and
let's
say
transcoding
or
recording
takes
some
time,
then
the
jitter
buffer
still
will
drop
frames.
Obviously
because
it
it
is
sort
of
the
last
step
in
that
flow
and
yeah.
O
So
so
we
were
wondering
I,
guess
whether
originally,
let's
say
in
the
thesis
or
in
your
when
you
worked
on
this,
did
you
still
have
a
jitter
buffer,
or
do
you
have
a
way
around
that
and
just
not
have
a
jitter
buffer
at
all?
At
least
I
mean
I'm
talking
about
the
RTP
C
source,
the
RTP
source
stream,
jitter
buffer.
That
comes
right
before
the
processor.
A
O
That's
true,
so
if
you
raise
the
max
on
the
jitter
buffer
or
make
them
just
really
buy,
even
without
adaptive,
noise
make
them
really
large.
Yes,
then
that
would
solve
that.
The
only
thing
I
remember
that
the
thesis
mentions
this
problem,
then
that
you
still
have
to
for
your
last
whatever
ten
seconds
or
so
you
would
have
to
sort
of
empty.
C
C
C
N
One
problem
that
that,
when
side
effects
that
we've
mentioned,
of
using
bigger
buffer
instead
of
the
packet
buffer
is,
if
you
have
a
really
high
jitter
buffer,
especially
for
the
purposes
of
recording,
we
run
into
this
issue
where
we
lose
the
last.
You
know,
however,
many
seconds
of
the
buffer
of
the
audio
right,
because
there's
not
there's
no
with
packet
buffer,
their
methods
to
flush
it.
O
N
N
C
C
N
A
F
M
F
P
M
P
M
There
has
been
some
subsequent
updates
to
that
code
because
we
have
been
finding
issues
and
fixing
fixing
them
as
we
go.
So
maybe
I
don't
know
if,
before
we
start
looking
into
this
a
bit
more
closely,
maybe
maybe
it
would
worse.
Therefore,
to
update
to
the
latest
version.
The
very
latest
commit
and
check
again
and
if
the
problem
persists,
then
we
are
going
to
have
a
closer
look
and
I'm.
Gonna
continue
the
email
thread
and
we
can
discuss
about
it
and
see.
What's
going
on
exactly.
G
Heyo,
this
is
Josh
here
from
Miami
got
a
question
with
regards.
Hopefully
you
guys
can
hear
me
now
this
awesome,
so
I've
got
a
question
with
regards
to
benchmarking.
Has
there
been
any
work
to
benchmark
the
performance
of
jitsi
reliability
versus
some
of
the
other
platforms?
I
saw
the
one
comparison
on
quality,
but
just
in
terms
of
packet
loss
and
consistency.
I'm
I'm
wondering
if
there's
been
any
ongoing
work
to
to
do
benchmarking
against
the
other
commercial
services.
D
We've
done
some
load
testing
in
the
past
to
check,
like
you
know,
bandwidth
and
all
that
and
in
the
past
three
ed
used
to
use
jitsi
hammer
more
recently,
we've
been
more
using
I,
we
haven't
done
anything
official
or
anything
big,
but
often
for
George
or
me
or
Boris
I.
Think
I've
been
using
like
the
selenium
grid,
a
selenium
grid
for
testing,
but
nothing
automated
in
terms
of
gathering
measurements
or
anything
like
that.
G
Thing
thank
you.
I
had
I
had
one
other
question,
which
I
was
just
wondering
if
was
repaired.
For
some
reason
we
have.
One
of
our
members
is
using
a
Mac
and
they
have,
for
some
unexplained
reason,
microphone
unavailable
that
that
comes
up
on
their
on
their
Mac
and
we've
been
trying
to
pin
down
exactly
what
might
that
have
been.
That
makes
her
Mac
different
than
other
Mac's
and
our
an
organization
and
does.
D
It
happen
even
the
first
time
they
used
it
after
rebooting
the
machine
we
haven't,
I'm,
not
sure.
If
we've
tried
to
reboot
I
will
I
will
try
that
so
not
so
I
don't
know.
If
anybody
who
knows
anything
specifically
about
this,
but
in
general
can't
get
microphone
is
usually
like.
There's
some
app
that's
being
greedy
about
the
mic
and
that's
already
acquired
it
and
chrome
can't
get
it.
D
So
if
you,
if
you
reboot
and
it's
fine,
then
it's
probably
definitely
that
if
you
reboot-
and
it's
still
a
problem,
there's
even
still
a
chance
that
maybe
something
is
auto
starting
and
grabbing
the
mic.
Early
I,
don't
know
if
she
is
or
they
have
anything
else,
though
it's
installed.
But
those
are
the
first
things.
I
try
I'm,
not
we're
of
any
known
issue
for
mac
microphone
acquisition,
but
if
anyone
else
is
well.
N
N
I've
Union
big
bits
of
scripts
in
some
cases,
because
it
was
so
consistent
having
the
Google
Hangouts
lobbies
in
a
number
range
of
different
things:
it
just
just
PS
EF
anything
like
that
grep
core
audio
and
then
you
just
kill
core
audio,
and
it
will
automatically
just
restart
that
when
that's
been
the
problem,
restarting
core
audio
without
having
to
restart
your
whole
machine
has
I,
don't
know
if.
N
J
N
D
M
N
Yeah,
no
I,
don't
I,
don't
know
specifically
the
recorder,
but
but
we
yeah,
we
just
noticed
some
weird
weird
behavior
and
rtcp
by
events
we
receive
so
I
was
I
was
just
curious,
whether
that
is
well.
We
can
just
assume
that
that's
always
being
sent
from
the
client
or
whether
the
jb
b
or
FMJ
might
sort
of
win
it.
I
know
connection
goes
bad.
It
might
explicit
just
try
to
send
it
on
behalf
of
a
user
if
it
things
like
that
connections
like
a
stream
is
dropped
or
something
like
that,
but
is.
M
M
And
we
do
relieve
them,
I
mean
at
least
for
the
streams
that
are
being
translated
or
adapted
or
projected
they
are.
The
device
are
being
remade,
but
if
we're
talking
about
the
recorder,
then
FMJ
is
involved
and
I
would
assume
that
FMJ
does
send,
buys
for
the
same
streams
as
boris
has
already
mentioned.
I
haven't
be
like.
N
Actually
thoughtfully
recorder,
we
were,
we
were
intercepting
by
events
and
not
actually
preventing
them
from
getting
passed
through
to
FMJ.
I
have
to
look
at
it
to
be
sure,
but
but
I
can
check
just
curious,
but
yeah
thanks
for
that.