►
From YouTube: Jitsi Community Call - April 23, 2018
Description
The Jitsi Community Call is held every other Monday at 10:30AM Central US time at https://meet.jit.si/TheCall
A
So
it's
been
another
two
weeks
as
usual,
we'll
go
over
our
team
and
share
some
progress
and
then
we'll
move
on
to
the
open
mic.
So
if
you
have
any
questions,
please
out
on
14
minutes
and
you
have
a
chance
to
ask
them
later,
I
don't
have
anything
to
talk
about
from
my
work
personally,
so
I'm
gonna
go
Astrea.
The
team
Zoli
I
see
you're.
Here
you
want
to
talk
about
your
work.
B
Kalau
so
I
don't
have
much
progress
on
my
side.
I
hit
a
little
bit
of
vacation
and
did
some
work
on
the
deep,
linking
UI
flash
API.
Let's
say
this
should
be
merged
now,
so
I,
don't
suppose
somebody
is
gonna
want
to
use
it,
but
maybe
it's
gonna
be
useful
one
day
for
G
to
meet
electron,
for
example,
or
maybe
somebody
wishes
I'm,
just
not
accustomed
up
there
and
I
think
that's
all
from
my
side.
C
C
So
right
now
we
have
a
fix
which
only
catches
the
exceptions
and
we
may
not
always
go
into
the
picture
in
picture
mode
in
in
those
cases
where
the
app
would
crash
I'll
be
working
on
a
proper
fix
this
week,
other
than
that
I
fixed
some
problems
with
the
Android
tests
in
torture
and
probably
something
else
but
I,
don't
remember
other
things.
This
one
wanted.
D
Specifically
for
this
community,
because
we
had
a
number
of
cases
where
work
on
mobile
actually
broke
stuff.
Sorry
work
on
weapons
actually
broke
stuff
with
mobile.
It
may
or
may
not
get
caught,
so
there's
a
good
chance
that
may
actually
reach
users.
Hopefully
this
is
to
help
a
lot
in
that
direction.
So
it's
actually
quite
exciting.
E
E
Everyone
was
involved
in
the
development,
only
led
it
and
it's
gonna
hit
once
you
get
the
permission,
is
gonna,
be
in
your
calendar
and
show
your
meetings,
which
have
a
URL
that
we
recognize
and
he's
going
to
show
them
in
the
app.
So
it's
super
easy
to
join
those
without
click
on
calendar
notifications
and
then
links
inside
them.
But
it's
also
gonna
show
you,
while
you're
in
a
conference,
a
notification
for
your
next
conference
for
people
who
do
those
things
and
like
one
over
two
and
it's
been.
E
It's
gonna
be
easy
to
talk
on
the
notification
and
going
to
that
subsequent,
so
awesome
right
there.
We
do
have
a
few
crashes
on
Android
related
to
the
calendar.
What
with
all
Android
versions,
devices
in
calendar
format,
so
we
did
at
least
subsequent
hotfix
for
the
calendar
and
in
the
weekend,
or
it
is
one
or
hotfix.
You
know
to
catch
the
Sun
picture-in-picture
exception.
That
was
crushing
us,
it's
a
intermediate
step
or
oh,
no,
it's
not
a
final
solution,
but
pictures
and
pictures
on
the
support
on
Android
date
anyway.
E
F
Yeah,
sorry
thanks,
so
what
I
wanted
to
add,
because
people
some
top
nicely
is
that
is
that
there
is
an
additional
change
that
is
being
reviewed.
A
merge
right
now
is
that
that
was
a
that
was
a
refactor.
That
web
is
using
the
Redux
store
to
store
such
things.
Now
that
seen
the
pipeline
being
reviewed
merged
soon,
and
the
next
things
that
will
come
from
me
is
a
couple
of
free
photos
regarding
current
arcing
and
yeah
I.
Think
that's
it
and.
G
E
And
everyone
did
their
work
in
order
to
bring
that
into
release,
but
it's
a
sparrow.
So
now
we
can
zoom
into
someone's
desktops
to
him
and
like
see
all
the
details
and
move
around,
because
obviously
the
substances
are
gonna
fit
onto
a
mobile
device
crimson
time.
So
we
can
pinch.
Zoom
move
around
also.
G
B
E
Also
got
you
like
future
points
is
already
said,
and
things
that
we're
working
on
we're
currently
working
on
UI
redesign,
we're
like
moving
buttons
around
changing
some
of
the
toolbox,
so
so
is
working
on
that
actively
and
we're
gonna
bring
it
so
ish
the
test
even
have
a
development,
we're
gonna
split
local
participant
from
the
remote
participants
like
just
like.
We
have
on
work
where
you
can
scroll
through
the
remotes,
but
the
logo
is
gonna
stay
in
his
position.
That's
what
solid
it
is!
E
Well
he's
got
a
PR
for
that
young
and
we
have
internal
to
work
on
adding
directory
search
so
that
we
can
tap
on
the
button
which
is
like
plus
person
as
daikon,
and
then
we
can
that's
gonna
raise
an
event
to
dice,
DK
consumers
and
they
can
plug
in
the
home
search
UI.
In
our
case,
the
default
is
a
shared
sharing
sheet
which
allows
you
to
copy
the
URL
and
like
send
it
to
all
different
labs
depending
on
the
operating
system,
but
some
times
we
want
to
have
something
else
like
the
rectory
search.
G
H
A
I
H
A
C
G
So
one
of
the
things
ona
to
point
out
for
our
new
to
Bri
was
we
actually
made
a
mistake
with
our
versioning
for
a
little
while
with
the
Debian
packages
which
we
have
now
corrected
in
the
public
repository
the
issue
of.
Basically,
you
were
going
to
end
up
pulling
all
older
versions
of
the
jibbery
because
of
the
numbering
scheme,
even
though
the
newest
version
should
have
actually
been
1.2,
and
we
should
have
been
blown
away
all
of
those
older
demo,
revisions
that
were
incorrectly
numbered,
so
that's
resolved
now.
G
If
anybody
has
any
issues
trying
to
pull
jibbery
from
the
repo
I
also
have
a
PR
open.
That
explains
how
to
do
it
in
into
the
Dupree
readme,
so
that
should
make
it
a
little
bit
more
user
friendly
and
easier
to
consume
other
than
that
I've
been
working
to
productionize
the
GB
a
little
more
and
working
on
the
file
recording
stuff
I've
got
some
feedback
to
Brian
that
will
probably
guide
some
of
that
work
forward.
Around
file,
recording
and
gathering
a
participant
information.
A
Thanks
Aaron,
so
one
one
more
note
from
me
before
we
move
on
to
the
open
mic:
it's
about
ji-sook,
as
we
reported
last
time.
We
had
five
slots
allocated
to
us
and
in
about
15
minutes
guru
should
announce
the
accepted
students.
So
as
of
right
now
we
don't
know
what's
going
to
happen,
but
check
back
on
our
mailing
lists
and
on
our
website
soon,
because
we're
have
information
for
this
year.
We're
very
excited
about
this
year's
participation.
D
So
I
had
at
least
three
people
ping
me
privately
during
the
past
couple
weeks,
who
promised
they'll
come
over
I,
don't
know
if
they
haven't
made
it
or
not.
But
let's,
let's
talk
about
one
of
the
questions
that
was
raised,
which
was
around
the
performance
testing.
Someone
was
asking
more
about
what
what
is
the
purse
that
we
were
getting
from?
Hopefully
we're
seeing
and
Brian.
You
were
the
last
one
to
do.
First,
test
on
our
side.
D
D
I
D
I
Yeah
yeah,
that's
where
we
were
maxing
I
want
to
say,
might
even
have
been
600
and
some
change
or
something
but
I'm,
not
100%.
Sure
on
that
and
at
that
point
going
higher
than
that
is
when
we
started
seeing
basically
like
the
UDP
buffer
errors.
So
we
weren't
keeping
up
with
the
queues.
I
believe
was
the
issue
at
that
point,
and.
H
G
A
J
Okay,
just
a
follow-up
question
on
the
test,
because
I've
been
testing
also
with
like
a
c4
x-large
machine
in
US
and
my
problem.
There
is
that
I
can
also
reach
500,
megabit
or
even
higher,
but
the
CPU
is
is
what's
going
to
run
out
and
it
happens.
You
can,
if
I,
set
up
a
test
where
I
actually
stream,
only
one
stream
into
the
conference
and
I'd
say
like
40
50
users,
just
listening,
then
I'm
running
out
of
CPU,
while
I'm
at
75
80
%,
but
by
just
listening.
You.
J
I
D
E
A
A
I
J
J
D
J
D
D
F
F
D
K
Community
Colin
thought
I
would
join
to
ask
a
few
questions,
although
I'm
realizing
that
it's
quite
technical
and
I've
sort
of
had,
in
my
mind
too,
describe
a
business
case
that
we're
interested
in
because
I'm
working
for
a
big
IT
company
here
in
Greece
I,
don't
know
if
this
is
the
place
to
do
so.
I
don't
want
to
waste
anybody's
time.
I
think.
D
D
G
K
D
Fairly
quickly,
so
there
are
multiple
systems
of
thousands
of
users
any
more,
including
Atlassian's,
but
many
others
that
use
GT
and
and
are
dealing
fine
with
it.
Obviously
they
don't
run
on
a
single
server.
So
it's
not
only
about
the
scalability
of
the
GT
server
itself.
You
have
to
build
your
infrastructure
appropriately.
What
it
comes
down
to
with
GT
is
that
it
has
been
scalability
that
partially
gets
you
there.
You
can
run
multiple
video
bridges
behind
our
conference,
Orchestrator,
which
we
call
cheek
awful.
D
This
is
basically
the
thing
that
manages
sessions
and
puts
them
on
different
video
bridges
and-
and
we
typically
do
installations
where
one
machine
hosts
one
gqo,
full
server,
one
prosody
server
and
then
those
machines
for
multiple
other
machines
that
that
are
bridges,
something
like
link
to
that
add
bridges
and
remove
them
on
the
fly.
And
then
you
can
put
another
layer
in
front
of
the
gqo
for
instances
they
use
h8
proxies
and
use
those
agent,
the
Praxis
to
front
multiple,
what
we
call
shards,
a
shard
being
one
geo,
for
instance,
post
multiple
bridges.
D
So
you
can
have
multiple
of
those
shards
behind
the
layer
of
har
axis
which
can
get
to
you
very
very
very
far.
This
is
there's
no
practical
limit.
Well,
actually,
that's
not
true.
There
is
a
practical
limit,
but
it
you
really
have
to
work
hard
to
hit
it
now
in
terms
of
scalability
of
individual
calls.
I
don't
know
if
that's
something
that
you
were
asking
about
or
not,
but
we
currently
consider
that
realistically,
it's
hard
to
use
the
system
for
calls
with
more
than
50
people
and
even
50.
D
H
D
That
work
is
increased
currently
in
progress.
We
expect
to
be
supporting
hundreds
of
participants
for
coal
somewhere
Tory
end
of
the
summer,
but
yes,
that
only
last
part
pertains
to
number
of
participants
for
conference
everything
else
that
I
said
about
the
shards
and
how
you
run
to
me.
They
shave
boxes
that
directly
answers
your
question.
K
Yeah
great
thanks:
what
would
be
its
perhaps
it's
a
difficult
question
to
answer,
but
what
would
be
the
best
year
is,
let's
say
in
in
developing:
let's
say
the
the
infrastructure
and
managing
the
whole
development
of
the
software
on
the
software
side
to
finish
the
project.
It's
a
difficult
question:
what.
D
C
K
K
Let's
say
providing
this
availability
in
the
market,
first
of
all,
in
our
higher-end
models
and
then
moving
slowly
towards
the
mid-range
and
perhaps
the
the
lower
range
motors.
So
we
do
have
the
the
opportunity
to
provide
this
service
only
in
a
subset
of
devices.
Let's
say
just
so:
we
can
make
the
infrastructure
work
to
build
it
faster,
and
then
we
can
worry
about
supporting
more
customers
in
the
in
the
future.
K
So
we
do
with
the
the
Android
application
to
to
integrate
with
our
voice
recognition,
technologies,
I
say,
for
example,
that
call
someone
some
customer
could
speak
to
the
device.
Ask
the
device
to
call
someone
and
the
device
fugitive
would
call
one
of
his
contacts.
So
in
that
respect,
and
only
having
one
to
one
video
calls
and
using
the
the
smallest
possible
effort
on
the
infrastructure.
D
D
Secondly,
I
would
not
start
with
one-to-one
support
only
because
you
don't
even
need
JIT
c412
once
now,
jitsi
would
natively
optimize
one
two
once
by
offloading
them
of
the
bridges
and
and
and
and
connecting
people
directly,
if
that's
possible,
using
turn
servers
if
that's
necessary
and
only
switching
automatically
to
a
bridge
when
a
third
party
joins
I
would
very
very
strongly
advise
you
to
not
split
out
one
to
one
and
multi-party
calling,
because
it
would
probably
be
more
work
for
you
to
do
the
split
than
anything
else.
D
A
side
question
which
you
didn't
ask
but
I'd
like
to
give.
You
is
also
make
sure
that
you
use
I,
don't
know
if
you're
currently,
if
you
currently
have
plans
on
whether
you'll
be
using
the
bridge
alone
or
you'll,
be
using
the
entire
GT
mid
suite
would
strongly
encourage
you
to
use
the
entire
suite
through
the
iframe
API
that
we
have
so
whatever
embedding
you're
doing.
You
can
embed
the
Android
application
entirely
so
that
you
use
the
full
suite
of
of
infrastructure
and
I
puts
you
in
the
same
use
case
that
we
are
at
Atlassian.
D
D
Basically,
how
you
a
few
thousand
users
split
it
that
way.
Is
it's
not
a
lot
of
users?
That
will
probably
mean
at
best
a
couple
of
tens
of
simultaneous
calls
and
if,
if
you're
expecting
one-to-one,
which
is
usually
what
services
mainly
do,
probably
more
than
60%
of
your
calls
would
be
one-to-one
anyway.
Regardless
of
how
you
limit
that
service,
then
you
know
resource
wise.
Even
a
single
server
would
support
that.
G
Also,
try
me,
if
you're
thinking
about
running
a
communication
system,
it's
always
about
monitoring,
so
you
know
what's
wrong
before
your
customers
do
so
that
usually
takes
people
resources
as
well
as
computer
resources
and
I,
would
highly
recommend.
You
study
that,
as
as
you're
considering
running
a
service
so
that
you
gain
trust
to
your
customers.
L
Do
we
have
anyone
else?
Hey,
hey
Emil,
it's
Chad!
Actually,
I
have
a
quick
question:
hey
Chen
I
have
a
client
that
wants
to
run.
Jitsi
me
just
has
a
mobile
web
app,
not
not
using
the
native
clients,
it
looks
like
the
meet
application
by
default
that
detects.
If
it's
a
mobile
browser
and
then
forces
you
go
and
download
or
it
basically
doesn't.
Let
you
use
me
like,
and
you
have
an
option
to
like
include
download
links
or
not
for
the
native
app.
Is
there
a
way
to
actually
allow
it
to
run?
L
E
E
D
Right
yeah,
if
you,
if
you
guys,
if
you
tell
if
you,
if
you
send
a
note
to
the
list
or
or
to
us,
we
should
be,
we
should
be
able
to
point
you
to
the
place
in
the
cool
where
that's
happening,
but
there's
a
bigger
question
to
consider
here,
which
is
the
fact
that
the
reason
we
turned
that
off
is
because
off
the
box.
It
really
doesn't
look
that
great
in
a
browser
on
a
mobile.
It's
not
a
responsive
design
and
it's
kind
of
challenging
to
use
yeah.