►
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
B
A
With
the
hangouts
link
that
is
also
linked
from
the
document,
that's
remove
participation
link,
and
this
is
how
we
will
also
present
instead
of
moving
around
she.
My
cable
death
does
involve
before,
but
this
is
all
fairly
there
was,
and
with
that,
we
thought
we
can
clip
it
off
these
session
here
with
a
recap
of
what
we
did
and
last
team
back
so
good.
A
So
the
working
mission
is
you
provide
methods
to
observe
and
improve
aspects
of
application,
performance
of
user
agents,
features
and
api.
That
in
practice
means
that
we've
worked
both
ways
in
which
we
can
measure
measure
performance,
so
that
developers
can
measure
their
second
performance
and
make
sure
it
gets
better
rather
than
worse,
as
well
as
methods
to
actually
make
them
actively
better.
A
A
That's
a
feature
that
we
thought
was
ready
to
ship
and
move
to
recommendation
last
year,
but
then,
during
last
year's
session,
discovered
that
there
are
a
lot
of
rules
that
seem
critical
to
tackle.
So
this
year
we
basically
plugged
all
those
goals
evident,
get
supported,
entries
and
type
specific
parameters,
performance
entry
registry
and
also
buffered
flag
for
elf
reads.
A
D
A
With
navigation
timing,
we
discovered
that
timing
allow
origin
that
was
controlling
some
of
the
and
priests
loaded
to
the
previous
page.
Didn't
really
make
a
whole
lot
of
sense,
so
replace
all
those
checks
wouldn't
see:
Morgan
checks
and
user
timing.
L3
is
now
a
thing
where
we
have
arbitrary
arbitrary
payload
is
part
of
user
timing,
which
is
something
that
people
have
been
hacking
around
for
a
while.
A
So
it's
for
you
to
see
that,
unlike
out
there
preload
with
SR
I
now
work
spoke
with
spec
wise,
as
well
as
shipping
in
chromium
and
finally
not
working
work,
but
work
that
would
like
incubation
work
that
was
presented
here.
Multiple
times
is
element
hanging
and
largest
essential.
Things
are
both
things
that
shipped
and
we're
looking
forward
to
see
data
that
will
bring
I'm,
not
gonna,
outline
all
the
closed
issues
here,
but
we
had
a
lot
of
them.
A
A
It
was
the
last
Deepak
moved
back
to
working
graph
because
we
discovered
that
a
bunch
of
things
are
missing,
but
now
it's
I
believe
ready
to
use
that
to
some
recommendation
status,
resource
timing,
l2
still
has
one
or
two
open
issues
that
require
work,
but
none
of
that
is
contentious,
so
I
believe
it
will
be
very
like
we'll
be
ready
to
move
along.
Very
soon
same
goes
for
an
application.
Timing,
l2
and
HR
time
is
a
slightly
more
than
what
story,
but
it's
basically
ready
to
assuming
some
objections
have
been.
A
A
Worth
when
he'll
be
here,
okay
as
far
as
relevant
incubations,
so
we
already
mentioned
element
timing
and
lunch
and
tensile
paint.
We
also
have
Advent
timing
and
first
info
delay
as
relevant
incubation
support
presented
to
the
group,
and,
yes,
maybe
we'll
have
some
more
like
it,
but
in
the
end
of
the
day,
as
far
as
market
adoption
or
the
features
that
we're
just
working
on
it's
a
lot.
This
is
these
are
all
numbers
from
Chrome's
use
counters.
A
B
A
community
to
that
I
could
call
out
here
a
once
performance
observer.
So
a
lot
of
the
work
that
went
in
that
we
did
over
the
course
of
last
year.
The
you
all
mentioned
is
around
improving
your
economics
of
the
front
cylinder,
but
even
as
we're
iterating
about
it
was
really
great
to
see.
The
adoption
like
I
could
really
shoot
off
the
watch
here,
the
other
one
we
just
it's
a
little
bit
older.
B
E
B
Coming
here,
another
one
that
I
was
pleasantly
surprised
about
the
server
timing
because
in
one
hand,
in
severe
and
each
sort
of
thing,
but
you
have
a
kind
of
definitely
finding
some
really
cool
pieces
in
the
industry,
so
I
think
that's
another
one
that
should
100%
divvy.
This
is
one
percent
of
page
lows
in
Chrome,
where
we
observed
that
a
future
issues,
so,
if
like
five
percent
of
page
loads,
have
servers
having
some
sort
of
semi
hesitation,
which
is
pretty
much
more
than
evident
stuff
for
survivors,
just
a
fitness
in
context.
A
Okay,
we
also
had
some
process
improvements,
so
we're
now
recording
all
of
our
meetings
and
placing
them
online.
We
have
improved
minutes
as
well
as
at
least
the
intent
is
to
publish
them
fairly
soon
after
meetings.
So
that's
never
and
we're
also
hoping
to
have
some
sort
of
are
describing
facilities
for
the
videos
to
avoid
both
anchoring
down
people
with
scribing,
as
well
as
having
minutes
that
are
not
necessarily
percent
accurate
or
biased
towards
what
people
heard
them.
A
What
people
didn't
hear,
so
we
were
hoping
to
have
basically
used
the
recordings
in
order
to
take
up
some
auto
spending
service.
That
will
give
us
transcripts
of
every
word
said
and
thanks
for
us,
Tegan
here
or
you
can
get
into
them
on
that,
and
we
also
had
a
very
successful,
at
least
from
my
perspective
at
least
afaik.
In
June,
we
had
great
participation
and
cross-pollination
between
browser
books
and
Industry
folks
a
lot
of
great
presentations
and
remains
limited
progress
on
frame
timing
on
device
classification
and.
A
Beyond
that
single
page,
apps
seems
like
the
most
important
subject
that
we
make
progress
on.
I
need
to
make
much
more
progress
on
in
the
near
future,
and,
yes,
we
definitely
like
to
have
another
one
in
2020,
oh
well,
trying
to
plan
this
one
out
ahead
of
time
so
that
we
have,
you
know,
dates
sooner
rather
than
later
and.
B
B
We
got
a
lot
more
joining
the
input
and
the
feedback
has
been
actually
really
useful
to
help
us
carry
those
features.
I
think
one
of
the
accident
relaunched
the
bulk
of
the
face
to
face
recording
on
the
flight
here,
really
good
input
there
and
ready
for
some
reasonable
initiatives.
I
made
some
really
diverse
and
I
see
a
nest.
I
just
mentioned
other
need,
I
mean
that.
So
that's
something.
B
A
So
the
agenda
for
today
is
mostly
comprised
of
we
basically
split
up
in
the
two
days
into
design
section
and
issues
section
where,
unlike
last
year,
where
we
have
many
many
issues
to
work
through
this
year,
we're
in
a
bit
of
a
better
shape.
So
we
just
reserved
half
a
day
for
issues
and
have
more
time
for
presentation
than
this
special
news
regarding
new
stuff,
which
is
always
more
exciting,
then
you
know
talking
about
why
things
are
broken,
so
we
have
the
full
day
today
we're
planning
to
talk
about
event.
A
J
K
A
We
can
talk
to
Stephane
and
see
if
he
can
make
it
it's
here
and
can
you
yell
again
so
yeah
sure
we
can
around
this?
We
also
theoretically,
the
day's
agenda
ends
at
like
we
planned
up
until
5:00.
We
have
the
room
till
6:00,
so
we
can
like
at
least
last
year
we
overflowed
so
we
have
that
possibility,
but
we
can
also
move
things
around
and
maybe.
A
A
D
D
D
So
like
the
Harvard
X
n,
then
we
expose
the
timestamp
right
before
we
start
running
the
event
handlers.
Those
are
the
timestamp
when
we
have
finished
running
the
event
handlers
and
we
expose
the
next
pains
of
after
to
the
event.
Handlers
have
one.
So
that's
curation
curation
will
be
the
difference
between
the
next
page
and
the
event
hardware
times
time.
If
the
duration
is
long,
then
we
expose
a
performance
event.
Any
entry
with
it.
Also
that.
D
D
D
Their
word
happens
in
a
separate
context
and
aside
that
who
and
when
Tim
said
example,
unions
cases,
even
if
you
are
not
using
a
framework
or
when
something
happens,
that
requires
resource
watching
and
that
happens
asynchronously,
and
you
need
to
wait
for
it
to
meaningfully
know
that
oh,
this
is
now
going
to
be
painted
what
I
want.
So
you
click
right
in
image
edit
to
the
dumb
and
then,
where
you
say.
Okay
now,
I
went
to
the
next
thing
after
that,
because
that's
when
the
event
was
processed,
another
one
would
be
image
the
biggest
letter.
D
If
the
video
for
the
new
part
of
the
video
that
wants
to
be
displayed
and
after
that's
fetched
and
done,
then
you
can
query
the
next
thing
to
know
when
you're
gonna
start
playing
the
video
again
another
one
enter
a
query:
send
a
query
to
the
server
and
once
the
server
has
completed
its
responses,
and
you
know
that
the
next
paint
should
be
when
the
you
are
doing.
The
results.
D
D
D
So
here
is
that
when
an
event
is
dispatched,
we
wait
for
it
promises
to
be
resolved
so
for
each
which
call
to
measure
until
you
have
a
promise.
You
wait
for
the
next
and
after
the
promise
has
been
resolved,
and
that
gives
you
some
times
and-
and
you
add
the
timestamp
to
the
map,
with
the
key
being
the
name
of.
D
D
Then
we'll
create
the
map
facing
curations,
which
will
be
exposed
in
the
performance
event,
timing
entry
and
for
curation,
which
is
the
value
that
we
use
to
determine
whether
the
event
is
longer
or
not,
and
it
just
because
we've
used
longest
of
durations,
we're
competing
in
a
sink
Eurasians.
So
the
largest
timestamp
in
a
sec,
durations
question,
easy
I,
understand.
J
D
F
L
D
There's
certainly
a
security
problem
I'm,
not
sure,
though,
if
just
having
the
last
one,
not
the
next
page,
wholly
useful,
but
it'd
be
right
that
if
the
security
problem
is
in
your
event
handler,
you
just
register
a
bunch
of
promises
like
callosum
and
then
anytime,
you
want
the
next
paint.
You
just
resolve
the
promise,
however,
notice
that
you
only
get
the
entry
at
the
end,
so
it
is
that
the
wait
for
all
the
promises
to
be
resolved.
Yes,
certainly,
that's
a
security
concern
that
needs
to
be
considered.
We
clearly
need
to
the
security.
F
D
D
The
reason
we
have
the
map
yeah
instead
of
simply
just
one
measure
until
is
because
it
could
be
multiple
vendors,
for
example,
having
their
own
handlers,
and
we
don't
want
them
to
affect
too
much
the
measurement
of
each
other.
If
that
makes
sense,
because
always
then
you
get
your
easing
direction,
you
have
no
idea
what
it
means
right.
B
F
D
B
B
L
B
A
F
B
D
D
F
J
D
Be
caused
by
other
things
here
is
the
security
concerns
of
just
exposing
next
things,
which
is
really
what
we
are
exposing
here.
We
want
to
do
it
pain
only
certain
scenarios
where
it's
meaningful
for
the
developer.
You
know
the
next
Camus
just
expose
next
pandemic
Oh.
Will
this
work
from
JavaScript.
J
F
Then
we
resolve
that
promise
at
an
arbitrary
point
in
time
right,
which
means
you
can
measure
an
arbitrary
frames
next
page
time
right,
I
mean
what
you
need
to
do
is
just
keep
clothing,
that
you
want
to
write
and
that's
where
Nichols
never
discussing
earlier
I,
don't
even
expose
the
next
paint
for
every
promise.
I
think
you
would
maybe
Jake
when
the
last
promise
is
resolved.
For
that
we
need
to
think
it
was
security,
a
little
bit
more
carefully
that
make
sense.
I,
don't
know
like.
J
N
N
F
During
the
point,
I
think
worst-case,
we
cannot
expose
a
particularly
accurate
meantime
we'll
have
to
fall
back
to
basically,
what
would
give
you
I
wanna
hit,
fell
back
and
for
a
moment
to
the
the
comment
event
to
this
be
part
of
easier
China.
If
I'm
a
web
developer
and
I
wanted
reason
holistically.
But
how
long
do
you
click
event
handler?
Is
my
life's
a
cake?
I
want
to
be
able
to
instrument
both
click
events
that
are
handled
by
the
browser.
F
J
F
F
J
F
F
F
D
F
A
A
F
B
And
web
developers,
so
we
need
a
standard
so
that
we
go.
We
looking,
we
see
a
lot
to
react,
sites
have
bad
invalidity,
and
so
we
go
and
we
have
these
devote
so
we're
gonna
react
to
me
at
this.
You
can
you
can't
just
be
user
timing,
because
the
person
that,
once
the
application
is
different
codes
from
the
person
that
can
do
the
instrumentation,
which
is
Tendulkar
base
again
from
the
analytics,
require
and
such
we
need
some
court.
It's
a
coordination
problem
between
these
two
beautiful
code
bases.
A
F
J
A
F
D
A
F
So
the
proposal
here,
if
I
understand
correctly,
is
we
added
something
either
driving
all
three
lets.
You
say
here
is
my
associated
event.
With
this
measurement,
I
guess
inside
event
is
present.
We
will
also
service
this
in
the
event
timing,
API.
But
this
means
that
if
you
want
to
do
similar
things-
and
you
don't
have
an
event,
you
can
use
this
strategy.
F
A
F
F
B
F
F
F
F
F
F
J
Might
be
in
some
cases,
but
I
think
it
must
measure
duties,
they
don't
matter
and
I.
Think
if
we
use
a
raft,
I
mean
you
know,
I
mean
LaRhonda,
we're
gonna
be
sick,
you're
right
yeah.
They
will
ourselves
eighty
ninety
percent
of
cases
yeah
dot,
that's
something
I
can
go.
Browser
can
implement
bigness
and
stuff
all
the
browser's.
It
has
the
institution
as
well
growing.
F
Applications
to
do
the
asynchronous
measurement
is
like
80
percent
of
the
value.
I
think
one
particular
case
scrolling
is
a
strong
one
where
the
compositing
time
is
often
why
you
don't
end
up
getting
a
smooth
scroll,
at
least
in
chrome
limitation,
and
one
of
these
cases,
I
believe
opening
salvos
event.
Timing
is
scrolled
rank,
however,
I
think
these
worlds
can
work
together
where,
if
you
just
do
have,
unless.
E
H
A
B
H
I
just
have
a
few
slides,
I
figure,
it's
kind
of
an
update
about
what
it
is
where
we
are
testing
spline,
quick
refresher
about
is
including
factors
basically
lets
you
track.
Is
there
been
around
the
idea
being
that
there's
a
lot
of
overhead
when
you're
like
doing
a
injures,
every
comedians
the
browser
get
a
neutral
back.
You
think
it's
one
time.
So
it's.
H
Not
break
it
up,
it's
fun
because
they're
just
trade
off.
If
you
break
up
work
like
inside
this,
a
little
box,
you
end
up
with
like
time
in
between
each
block
and
if
you'd
like
ten
blocks,
maybe
that's
an
editable
blessing
I'd
like
a
few
hundred,
a
few
thousand
like
little
blocks
of
work,
the
time
in
between
each
block
of
gives
the
browser
you're
getting
control
back
that
really
ends
up
and
that
can
slow
down
the
site
quite
a
bit
overall.
So
you
end
up
trading.
D
H
D
H
To
get
rid
of
that
mention
basically
allow
sites
to
have
longer
shows
with
work,
so
they
don't
have
the
extra
latency
issues
I
like
breaking
up
work
even
more,
but
still
if
there
is
an
input
event
check
out.
There
is
an
input
event
and
go
back
and
like
the
browser,
this
match
back,
maybe
reorder
how
you're
doing
work.
So
you
can
respond
to
the
bit
of
that
corner.
So.
C
Shirt
so
yeah
the
current
cryoballoon
vision,
our
site,
isolation,
obviously
isn't
really
viable.
If
you
want
shits
on
all
platforms,
so
we've
been
going
back
to
the
drawing
board
on
Planet
Waves
of
leverage
like
cluster
hit
testing,
which
is
present
on
most
UAE's.
To
get
this
to
state
where
we
can
do
per
frame,
organized
motion,
and
so
far
things
are
gonna.
H
So
far
so,
like
I
said,
it's
gonna
try
Chrome,
so
we've
tested
it
out
on
Facebook
all
right.
So
the
thing
we
tested
was
a
very
nice
implementation,
mouthful
scheduling.
All
we
did
is
basically
the
three
Macbeth
loans
insight
inside
of
that
framework.
If
we
saw
that
there
are,
who
would
check
his
impending
it
anytime
not
for
America
sighting
was
going
to
learn
more
and
if
there
was
it,
wouldn't
stop.
Let
the
even
get
the
special
kind
of
you.
H
H
Deadlines
but
be
able
to
take
advantage
of
like
doing
actual
reordering
with
this
one
thing,
but
it's
not
kind
of
the
scale
to
get
like
valuable
yeah,
so
yesterday
only
I'd
say
but
I
mean
so
there
is
also
great
basically
like
we're
able
to
see
there's
an
event
pending
we're
able
to
reduce
event
queue
time.
So
that's
the
time
from
like
when
surrounds
record,
basically
like
the
horizontal
stamp
as
when
the
u.s.
gets
the
the
event.
H
H
A
A
A
Appending
for
two
ways,
one
is
to
get
a
sense
of
how
often
we
are
potentially
affecting
these
are
experience
I
when
we're
doing
work
is
there
in
both
ending
that
we
should
be
getting
another
way
up
and
then
based
on
kind
of
an
assessment
potentially
taking
some
active
changes,
just
like
Facebook
was
talking
about
to
whatever
and
cue
it
up
to
later.
At
some
other
point,
so
the
boomerang
builders.
A
A
That
very
little
knows
the
median
amount
of
time
that
we
were
doing
a
the
browser
being
able
oppresses
imposing
development
the
whole
second
overall,
the
different
events
that
we
were
looking
at,
but
breaking
it
down
into
specific
terms
like
the
onload
event
or
visibility.
Changes
that
there
were
other
there
were
some
cases
were
the
beginning
was
up
to
twenty
five
to
forty
milliseconds.
A
A
How
are
we
doing-
and
we
definitely
would
like
to
keep
this
if
it
becomes
given
that
in
order
to
trial
and
keep
it
in
our
analytics,
but
just
as
a
general
feedback
mechanism
for
ourselves,
and
then
it
will
take
a
little
a
some
results.
It
will
probably
take
a
little
bit
of
work
on
actually
efficient.
We
rigged
up
these
tasks
in
the
meaningful
way
or
considering
planning
I'm
using
Elizabeth
pending
to
do
that.
Work
as
well
so
does
it
again
might
be,
would
like
to
get
all
over
stuff
as
quickly
as
possible.
A
A
B
Support
that
types
integrate
any
counseling
or
whom
you
wear
a
piece
of
home
and
also
the
movement
based
in
this
matters.
They
always
send
it
quick
or
not
depends
on
how
the
biggest
handlers
your
members
and
they
love
steak,
knives,
bounty
hunter
down
and
what
thought
of
any
state
state
and
all
of
us
then
people
of
the
management
at
that
time.
So
basic
question
is
how
he
chose
instead
of
or
event
types
ain't.
What
I
click
with
something's
would
be
religion,
particularly
and
point
out
just
more
raw
events
down.
H
H
The
other
thing
is
like
the
wording
is
like,
if
it
might
who's
bigger
than
like
and
doesn't
want
to
lie
to,
they
probably
might
want
to
just
say,
like
hey,
like
just
randomly
return,
true,
sometimes
very
athletic
security
reasons,
or
other
things
like
that.
So
I'm,
even
thinking
is
like
it's
like
someone
best
effort
like
if
the
browser
to
think
this
Richard,
you
can
return
true,
basically
for
different
reasons.
There's
a
rapper
thinks
there
might
be
look
like.
F
F
J
Yeah,
even
in
capsule
I
think
it's
so
super
vague
that
we
can't
have
both
compatible
implementation
because
of
in
so
and
it's
even
referring
to
basically
any
new
island,
but
especially
interracial
violence
there
in
this
moment,
but
it's
so
bad
if
we
would
implement
this
to
meet
you
basically
event
from
you
source
for
anything
in
a
closed
to
comment.
So
that's
what
I
think
something
more
precise
attending
Stan
yeah.
J
C
So
we
run
into
a
problem
here
when
you
like
to
specify
it
I
think
you
mentioned
in
past,
where
it's
like,
we
can't
read
or
like
simulate.
What's
currently
an
event
cure
with
dusters
will
run
before
the
event
gets
us
patch.
So
doing
things
like
precise
frame
attribution
makes
it
very
tricky,
so
we
can
never
really
get
them
T
that
it
isn't
the
pending
returns.
True,
oh
my
god,
then
we'll
actually
be
dispatch
part.
J
H
To
the
grocer,
that's
healthy,
fun
anywhere,
so
yeah
that's
about
yeah,
but
this
is
like
really
like
place
really
individually.
Why
I
meant
to
really
spec?
Well,
so
there
wasn't
like
there
wasn't
a
great
thing
for
melons
yeah
I
mean
like
we
could
potentially
look
at
like
expanding
the
changing
click
to
like
Mouse
like
animals
down
and
then
expanding.
The
list
of,
like
must
support
types
like
to
get
to
donate.
J
C
I
F
C
B
That
may
include
the
UI
has
be
just
a
team
in
any
other
context,
if
you
have
any
cookies
keys
in
might
move
with
them
in
the
first
place
like
it
was
primarily
to
avoid
must
avoid
you
mentioned
that
it
might
include
that,
so
you
would
assume
that
there
are
some
other
events
that
some
people
might
interested
in.
My
World
War
II
those
events
Oh.
D
B
H
Ideas
like
we
like,
there's
no
like
there's
no
event
that
gets
introduced
like
ideally
browsers,
should
be
able
to
make
you
compatible.
It
wasn't
so
depending
like
right
away,
I
think,
especially
if
against
I
kind
of
like
integrated
into
a
lot
of
scheduling,
any
people
like
come
to
rely
on
it
to
some
extent
like
you,
don't
want
to
want
it
to
be
able
to
be.
D
H
D
Because
I
think
it
might
make
it
easier
to
me,
as
you
see,
yeah
right,
because
you
don't
even
want
people
to
just
queries,
independent
or
events
not
really
important,
and
then
they
like.
Okay,
that's
fine,
let's
keep
going
and
then
we
don't
have
awards
in
science.
We
do
justify
it
by
saying.
Well,
this
is
the
pending
return
false.
So
we
just
find
that
something
hasn't
we
have
to
be
careful
with
which
then
swing
uses
markets.
F
F
J
H
C
M
H
C
E
C
B
F
F
C
B
Components,
mutation,
sec
or
as
a
browser
hood.
There
is
an
area
vencie
fewer
points
in
bowser
and
bits.
We
also
hear
completely
accused
events
and
it
just
it
doesn't
make
currently
using
that's
where
you
show
again
it
in
that
scenario,
there's
a
scenario
called
a
series
of
diseases:
we
don't
have
information
there's
and
that
hits
the
main
trip
of
a
render
to
get
that
only
one
who
is
what
is
a
Olson,
so
you
have
a
lackeys.