►
From YouTube: WebPerfWG F2F TPAC 2019 - day 2 late afternoon - issues
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).
B
Just
saw
that
on
it
so
then
I
figured
I
should
come
back
then
in
the
area,
but
I,
don't
really
know
how
you
know
useful
I
will
be
so
as
I
understand
it.
The
question
is
whether
is
your
is
reload
navigation
wise.
So
the
question
was
the
backward.
Was
that
cord
specifically,
because
whether
it
is
a
history,
navigation
should
be
true
or
false
back
and
it
redirects
to
a
different
URL.
B
Instead
of
getting
fresh
content
to
try
to
optimize
that
and
then,
if
it's
a
reload,
it
should
kind
of
more
aggressively
go
to
network.
So
getting
things
from
the
cache,
because
a
signal
from
the
user
that
the
page
is
like
broke
grand.
Alright,
the
user
wants
fresh
content,
so
that's
kind
of
how
we
ended
up
with
these
two
things.
I
think
some
sites
are
using
it
today.
It
is
history
navigation
to
decide
to
go
to
cache.
A
What
we
had
in
mind
is
for
us
and
for
navigation
timing.
The
use
cases
basically
I
want
to
be
able
to
split
my
reporting,
in
my
met,
the
metrics
that
I've
collected
based
on
the
mutation
types,
because
the
performance
characteristics
of
navigation,
reload
or
a
back
forward.
Navigation
are
different,
so
I
want
to
reduce
noise
by
splitting
out
those
tags
and
from
that
perspective,
if
a
backwards
navigation,
then
redirects
to
a
brand
new
page.
That
feels
like
a
navigation
from
that
perspective.
A
A
B
Of
the
brand
name,
pH
is
really
relevant
for
the
service
record.
It
just
is
kind
of
like
it's
taking
a
user
intent
in
deciding
what
content
is
show.
So
it's
like.
Oh
it
because
you're
just
trying
to
go
back
I'm
just
gonna
show
whatever
we
had
in
the
cache.
So
we
don't.
We
don't
slow
down
that
back
operation.
It's
possible
would
be
if
cache
that
it's
not
that
important
I,
don't
know
it
would
be.
If
cache,
would
we
even
do
this
redirect
if
the
pages
may
have
captioning?
B
Probably
this
wouldn't
happen
right,
you
just
yeah,
but
this
was
created
and
then
before
that
well,
okay,
that
background
funds,
you
it
yet
below
them
through
the
bf
yeah,
there's
this
possible
that
that
it's
not
in
the
VF
cache
right
yeah.
When
you
go
back
so
yeah.
In
that
case
it
would
still
be
relevant.
So.
B
A
C
B
Casual
car
and
you
feel
like
you're,
saying
that
we're
actually
focused
on
user
and
preferred
me
yeah
for
this
the
exposed
these
flags
exposed.
We
were
thinkin
user
intent.
More
then
I
guess
so.
Ten
verses
like
the
infrastructure,
yeah.
Whatever
not.
The
question
is
from
the
bfk
everything
reads:
it.
D
A
G
A
Where
do
those
definitions
sit
from
your
perspective,
because
navigation
timing
has
an
enum
would
open?
Yes,
like
there's
an
enum
with
a
navigate
reload
back
forward
and
meta-narrative
render
that
we
have
to
find
that
we
probably
shouldn't
have
defined,
and
so,
if
there
is
already
an
HTML
or
fetch
definition
of
what
it
or
each
of
one
of
them
means
you
can
probably
just
point
that
yeah.
B
B
A
B
H
From
implementation,
how
do
we
get
this
like
today?
Is
it
like
on
the
roof
us
project
play?
Well,
we
ain't
remit
in
one
case
in
point,
we
I'm
from
the
spec
aspect.
Why
do
like
we
get?
The
is
reload
is
navigate.
You
have
a
flag
movie
set
of
Benjamin.
Is
that
come
from
like
a
internal
report?
It's
not
an
offensive
in
anymore.
It's
a
light.
Son
requests
now,
okay,
so
from
a
spec
point
of
view,
this
is
what
a
redirect
happens.
You
just
copy
that
flag
to
the
new
requests
to
maintain.
D
A
A
B
H
B
H
H
H
A
Should
pick
one
just
pick
something
if
one
of
them
is
easier
than
the
other,
although
you
could
also
say
like
you
could
similarly
say
that
in
case
of
a
redirect,
if
it
was
a
back
forward,
now
becomes
a
navigate.
You
don't
like
it,
but
I
think
that
the
use
case
from
our
perspective
tends
toward
navigate,
but
it's
not
very
strong,
so
I'm
fine
with
aligning
with
whatever
you
end
up
with
and
then
reading
that
navigation
pipe
off
of
the
document.
A
H
H
Safe
little
bit
like
what
we're
taking
I
have
state
the
Kings
trying
to
track
like
when
was
the
first
read
or
accent
like.
Doesn't
that
cross
origin
and
stuff
like
that
that
source
worker
doesn't
have
that
to
worry
about,
like
we
only
care
about
the
current.
The
current
state
of
the
requests
not
like
previous
three
directors,
yeah.
C
So
think
about
this,
a
little
bit
more
and
actually
think
I've
I've
gone
back
on
my
previous
comment,
but
I
think
the
intention
user
is
maybe
more
important
from
the
issue
typing
a
bad
person.
Okay,
look
you
want
to
measure.
How
fast
are
my
navigations
and
I
know
that
these
words
navigations
typically
characteristics
I,
want
to
understand
that
and
if,
if
I
have
an
expectation
that
clicking
the
back
button
is
generally
faster
than
doing
a
reload,
but
my
statistic
actually
in
many
cases,
click
the
back
button
to
slow
as.
C
C
A
C
C
H
H
A
H
Syria
race,
what
was
previously
the
previous
state
of
the
document
in
the
navigation
timing?
It's
like
all
the
the
state
of
the
document
is
frozen
in
bf
cash,
all
the
heap
everything,
but
when
you
come
back
out
you're
overriding
the
bits
that
are
in
navigation,
climbing
entry,
I'm,
actually
not
sure,
that's
gonna
be
a
cash
case.
You
don't
get
new
them
each
time.
You
just.
H
Forward
number
like
you
would
need
to
create
some
new
thing
to
say,
like
yeah,
I'm
performance
observer
fires,
like
you
know,
yeah
yeah
like
something
indicate
you'd
get
the
page
show
with
that
persistent
that's
how
you
could
fire
something
like
I
can't
reach
I
got
a
hat.
Something
went
back
track
but
page,
but
you
wouldn't
yeah.
H
C
H
Yes,
I
didn't
know
if
you
were
adding
a
spec
language
to
change
that
I've
mentioned
typing
Hydra.
To
me
to
be
honest,
I
you
know,
but
do
you
I
mean
I
the
way
this
the
F
caches
tradition
doesn't
use
as
a
browser.
Automation,
optimization
that
sites
cannot
detect,
accepts
really
page
showed
up.
Persisting
was
true
because
of
that
state
state
and
generally
the
state
of
when
all
the
things
I
should
see
should
not
magically
change
when
it
comes
back
out.
H
Processing
devote
when
we
have
cash.
No,
no
there's
no,
there's
no
event,
there's
nothing!
There's
no
fetch
event,
there's
no!
It's
it's!
Just
as
if
the
page
you
got
frozen
and
it
comes
back
to
life
and
it
can
tell
from
like
timestamps
being
different.
Perhaps
it
they
lost
some
time.
The
time
margin
is
still
the
original
yeah.
So
yeah.
H
A
B
H
F
J
A
E
A
B
L
A
A
C
C
Yeah
I
mean
but
I
didn't
understand
a
way
to
learn
that
information
other
than
serviceworker
I.
Guess
it's
not
the
first
one,
but
you
can
make
the
same
argument
about
reload
like
if
reload
is
always
going
to
be
a
navigation
it
should
have
similar,
especially
if
people
are
like
answers
worker
not
going
to
cash.
C
Every
now
similar
performs
characteristics
to
the
first
load,
and
so
then,
potentially
the
real
use
of
it
is
just
understand
your
behavior
understanding,
the
performance
of
those
types
of
those
but
I,
don't
know
it
could
probably
be
probably
worth
asking
somebody
who's
using
it
to
see
so
or
defiantly
abiding.
What
are
these
cases
for?
These
are
spots
in
these
fields
and.
D
B
H
C
H
Saying,
like
the
group
decided
we're
not
going
to
change
the
existing
navigation
timing,
entry
coming
out
of
the
bf
cache.
If
we
need
to
observe
backboard
cache
can
be
on
the
backboard
cache
be
on
a
page
show
like
at
least
my
recommendation.
Could
you
create
some
new
place
them
some
new
event,
we're
like
putting
on
a
show
somewhere?
Don't
overwrite
the
data?
That's
already
there,
because
you
know
I,
think
that
would
be
a
little
bit
magical,
confusing
yeah.
A
A
A
A
A
A
A
G
G
H
A
G
H
To
me,
like
they're,
all
going
to
an
endpoint
and
reporting
observer
is
another
way
to
tap
into
that.
It
was
not
clear
to
me
at
least
reading
them
reporting
spec.
That
intended
was
that
reports
associated
with
documents,
look
like
a
broadcast
mechanism
so,
based
on
the
wording,
I
mean
it's
not
maybe
I,
misunderstood,
I
think
that's
it's
clear.
I.
H
H
G
H
H
Csp
is
clear:
there
is
a
document
or
something
that's
triggering
the
CSP
violations
or
other
report
types.
Don't
have
any
association
of
the
document
necessarily
so
what
happens
with
that
is
seems
like
just
to
say
you
don't
get
reporting
over
at
all
seems
well
I
guess.
I
would
like
to
ideally
have
a
report.
I,
don't
know
if
my
proposal
is
gonna
go
forward
or
not
for
like
they
use
guesses
I
had
I
was
counting
on
having
reporting
observer,
be
able
to
sure
I
have
an
expander,
but.
H
Today,
when
a
we
have
storage
in
the
reservoir
things
in
the
next
TV
cache
storage,
various
places-
and
inevitably
we
have
to
deal
with
the
problem
of
things
break
like
devices
stopped
working,
but
it's
good
flipped
things.
You
know
fire
scanners
come
in
and
change
things.
What
have
you
were
browser
engineers
right
bugs
and
create
corruption?
So
what
about?
What
happens
when
browsers
storage
gets
corrupted
is
undefined
at
any
spec,
so
I'll
do
different
things
and
usually
the
amount
of
throwing
some
like
undefined
exception.
H
A
it's
hard
for
sites
that,
like
they
don't
know
that
corruption
happens
in
so
they
trying
to
do
something
that
looks
like
it.
You
know,
then
what
because
it
wiped,
is
it
still
there?
We
don't
know
so
step
one
so
trying
to
sites
I've
gotten
a
sense
that
there
is
low
confidence
and
the
stability
of
storage,
because
because
of
these
issues,
like
so
step,
one
I
was
trying
to
do
proposed
was:
let's
create
some
visibility
by
creating
inconsistent
reporting
mechanism
from
when
and
corruption.
H
You
know
it
might
not,
but
today
we
we
do
some
of
this
through
throwing
exceptions,
but
we
can
hit
storage
corruption
when
there's
no
method
call
either.
So
it
gives
us
an
app
like
a
good
way
to
provide.
You
know
when
we're
computing
quota.
There
might
not
be
any
like
JavaScript
function.
That's
you
know
that
we
can
reject
from
so
it
gives
us
a
way
to
consider
for
these
corruptions
regardless
of
context,
but
it
also
wouldn't
give
them
a
place
where
they
could
write.
Like
you
know,
what
do
I
do?
H
My
recovery
code
can
go
on
my
reporting
observer
and
be
like.
Oh
I
ran
into
a
corruption.
Let
me
trigger
a
validation
of
all
my
data
and
redialing
and
loading
from
the
server
or
wiping
or
whatever
they
want
to
do
based
on
that,
and
it
would
be
sort
of
a
consistent
way
of
at
least
servicing
the
corruption.
Then
the
next
step
would
be
browser
disagree
on,
should
we
might
origin?
H
Should
we
like
leave
some,
and
some
of
this
comes
into
like
many
buckets
and
other
storage
things
to
happen
before
we
can
come
up
with
a
policy
mechanism,
but
the
first
step
I
was
trying
to
grate
was
like
a
transparency
to
help
give
Sykes
confidence
that
they
know,
like
my
clients,
are,
are
there
and
they're
not
in
corrupt
or
I,
can
see
like
the
corruption
rate?
You
know
in
different
problems
of
the
certify.
Are
there
any
other.
G
E
H
G
Reporting
system,
so
what
let's
say
corruption
happens
and
you
want.
You
know
you
like,
let's
say,
there's
actually
no
documents
for
some
reason
we
just
erupted,
and
then
we
didn't
make
sense
to
send
a
report
cut
of
dance
to
the
origin
staying
and
why
this
mega,
not
like
is
item
useful
signal.
I
think
it
is.
H
H
You
know
if
we
get
to
the
world,
where
you
thought
alike,
though,
or
you
completely
being
able
to
tell
the
server
that
you
had
a
user
here
at
Bethel
data
and
it's
gone,
you
know,
like
you,
lost
easier
to
storage
ruptured
until
they
login
again
I
think.
Is
it
useful
health
metric
for
that
site
based
on
the
sites
that
I
talk
to.
H
So
the
remember
you've
got
a
corruption,
you
know
so
the
report
that
I
was
proposing.
He
had
it's
a
storage,
for
we
had
corruption.
When
did
we
like
which
subsystem?
Was
it
in
indexing,
the
cache
storage
or
multiple?
What
did
we
do?
Leave
the
data
there
or
wipe
it?
You
know
what
are
the?
What
did
we
like
and
then
maybe
a
debug
field
that
could
be
pasted
into
a
like
a
bug
report
to
the
browser
like
we're.
Getting
like
a
ton
of
these
things
that
all
have
you
know
this
like
browser
specific
string
in
it?
H
Can
you
guys
help
us
debug
this?
You
know
and
fix
it
for
us,
and
so
so
they
get
a
like.
They
get
a
spike
on
their
health
metric
saying
Oh
summer
have
one
percent
of
our
users
getting
storage
corruption
and
they
all
have
like
this
string
that
says
corruption
and
leveldb
open
that
a
like
checksum
failed.
You
know
they
give
that
to
us.
That
tells
us
where
to
look
to
try
to
fix
the
problem.
Okay,
but
then
wouldn't
have
let's
go
say
all
that
happens,
and
then
it's
ready.
H
Than
I
was
having
you
may
dated
that,
because
we
can
then
talk
to
them
and
be
like
tell
us
what
you're
doing
that
might
be
triggering
to
this
corruption,
like
we
get
the
context
with
the
raw
error
error
data
from
chrome.
These,
like
we're
doing
a
very
complicated
dance
to
get
the
data
back
to
ourselves.
Well,.
A
G
Transparency
to
help
people
build
up.
You
know
some
level
of
confidence.
What
I'm
still
struggling
with
is
like
I
get
this
report.
Okay,
ten
percent
of
my
users
start
failing
or
like
have
corruption
issues
the
next
time
the
user
actually
opens
the
site,
presumably
you'd
want
to
like
notified
the
absolute
candor
so
clean
out
the
quarterly
restart
right.
G
H
H
E
H
H
H
I'm,
not
fine,
I
mean
I'm
fine.
If
we
don't
fire
the
reporting
observer.
If
this
happens
well,
there's
no
windows
open,
but
you
wanted
to
fire
at
some
point.
Well,
then,
I
think
there's
an
issue
of
like
we
don't
want
to
fire
multiple
times
for
the
same
function,
and
that
might
be
a
weakness
in
my
proposal.
A
H
H
Like
this
is
the
work
will
be
already
go
through
with
sub-sites
or
like
they
have
a
dashboard.
They
measure
how
often
they
get
like
exceptions
thrown
religion,
guys
what
it
spikes
they
filed
bug.
We
go
talk
to
them
and
we
can
do
this
with
sites
that
we
have
very
close
relationships
with,
but
it
would
really
be
nice
to
be
able
to
give
this
visibility
to
I.
Don't
it
maybe
maybe
it
doesn't
make
sense,
but
it
seems
like
providing
more
transparency,
might
help
improve
confidence.
Mr.
Burch,
just
the
one
on
the
one
I
was
sure
Mike.
C
H
H
Actual
is
that
they
get
a
data
that
they
can
use
to
file
bugs
against
the
browser.
Very
small
wing
majority
come
to
web
does
not
open
bugs
against
browsers
right,
that's
just,
but
the
ones
who
do
certainly
help
help
us
and
I
would
like
to
get
help
like
we
can.
We
can
help
people
whom
we
have
very
close
relationships
with,
but
this
would
allow
us
to
get.
You
know
more
people
who
could
give
us,
but
right.
C
C
H
A
hundred
different
API
points,
that's
where
this
error
and
code
that's
not
expecting
to
see
these
exceptions
today,
I
mean
it
doesn't
know
the
bug
had
to
go
change
all
those
like
that's.
It
seems
very
impractical
versus
in
a
single
spigot
out
from
an
adoption
point
of
view,
and
also
there
are
cases
where
we
don't
have
a
function
to
be
called
where
we
could
like
we
compute
quota.
H
When
we
start
the
browser
to
see
how
much
space
we
should
allow
you
to
do
it,
corruption,
you
know
we
we
erase
something
or
we
say
you
can't
write
anymore
five-plus.
If
you
don't
like
cause,
we
can't
figure
out
how
much
space
you're,
using
like
we're
kind
of
stuck
and
I
run
like
it
just
seemed
like
a
good
fit
into
being
for
reporting,
API
an
observer
and
I
think.
Maybe
my
mental
model
had
a
mismatch
with
Sun.
A
G
Which
is
so
late,
things
corrupted
I
ain't
as
a
developer
agency,
action
to
fix
stuff
and
today
there's
no
single
event.
It's
just
I
go
to
bed
open
it
and
it's
broken
and
I
try
to
do
it
again,
we're
gonna
yeah.
So
it
sounds
like
there's
a
heat
forced
on
signal
that
been
built
or
should
listen.
To
that
says
sorry,
you
have
to
take
action
well
and
that
bit
is
like
we
couldn't
be
delivered
that
in
here,
referring
to
probably
I
think
that's
a
entirely
new
class
of
you
know
it
because.
H
Issue
of
using
reporting
observer
that
I
think
for
that
use
case
is
that
there's
no
timeliness
guarantee
look.
There's
idea
like
you're
in
the
link.
It's
gonna
be
fired,
synchronously
from
the
corruption
in
the
function
call
like
it
could
be
delayed
on
you
know
your
to
queuing
and
stuff.
I
can
see
the
argument
for
creating
a
separate
surface
area
in
the
storage.
Respect,
I,
don't
think
that's
gonna
happen
because
I
think
most
of
the
motivation
is.
We
want
to
move
to
a
place
where
we
are
on
a
wiping
anyway.
H
So
like
it,
just
we're
not
going
to
look
require
the
user,
the
developer,
to
take
action
pending
action
for
them
unless
they
opt
into
policy.
Okay,
you
know
like
so
please
don't
like
I
am
a
big
company
who
has
paid
a
team
to
you
know
working
on
these
corruptions.
Please
don't
delete
my
data
because
it
has
user
data
and
you
know
I've
spent
a
lot
of
money,
but,
like
most
sites
aren't
going
to
do
that
like
so,
we.
H
G
H
Send
the
notification
is
saying
something:
bad
yeah.
I.
Take
your
point
like
using
reporting
observer
to
enact
cleanup
logic
or
validation
logic.
It
would
be
not
not.
The
original
use
I
do
think,
though
at
least
one
of
the
sites
I
talked
to
said
they
would
want
to
build
their
own
reporting
infrastructure
off
of
it,
and
so
we
need
some
way
to
to
him.
Yeah,
ideally
like
they
might
even
wanna.
H
Put
in
service
rehearsal,
yeah,
but
I,
don't
know
that
we
want
to
start
a
service
for
her
guitar,
the
recurring
observer,
her
Sylvan
related
to
them.
Sorry,
no!
No!
We
have
this
because
they
don't
have
you
for
the
case.
There's
no
page
open
that
the
q2
according
observer
did
or
they
eat
the
use
case.
Weren't
they're
gonna!
Do
it
themselves
instead
of
using
the
recording
endpoint
for
the
carpet.
C
H
Like
during
somebody
in
high
school
yeah
like
we
have
to
queue
how
much
space
each
origin
uses
in
order
to
determine
if
we're
under
pressure
until
he
stopped
to
make
that
so
we
took
every
word
didn't
to
do
that
computation.
So
a
lot
of
times,
there's
no
tab
the
odds
of
us
hitting
rupture
there.
It
doesn't
have
it
as
often
because
if
we
can
get
away
with
just
looking
at
the
size
of
the
file,
there's
certain
cases
we
coded
directory
structure
like
this
doesn't
look
anything
like
what
we
expect
and
like
that.
H
A
A
F
H
E
H
E
G
H
H
This
just
surprised,
because
I
skimming
in
the
reporting
of
the
red-nose
notion
that
reporting
that
reports
were
associated
with
the
document
and
report
they've
been
limited
to
the
reporting
observer
in
that
doc.
I
didn't
I
didn't
catch
that.
So
that's
how
we
rappelled
on.
Like
this
other
blame
proposal,
I
had
kind
of
basing
the
idea
that
they'd
be
report
to
be
broadcast
on
a
different
sourcing,
primitive
roads
or
agent
carriages.
A
G
A
H
So
one
place
we
could
fire
worried,
observe
reports
for
shared
resources
would
be
serviceworker.
We
were
willing
to
wake
up
the
serviceworker
to
do
so.
It
would
be
an
interesting
case
for
something
like
storage
corruption,
because
often
wiping
storage
means
getting
rid
of
the
serviceworker
registration.
H
So
I
don't
have
been
one
to
help
by
use
case
but
like
that
is
a
place
where
we
do
shared
resource
things
yeah,
and
that
could
happen
in
the
background.
Although
then
we'd
have
to
question
whether
it's
safe
trigger
those
cubing
is
probably
better
better
solution.
Specifically
to
this
issue,
I
would
say.
G
The
default
is
that
reporting
that
the
reports
are
fight,
the
government
we
catch
the
errors
associated
with
me
in
this
particular
question.
No,
like
the
iframe
break
up
a
fight
was
not
yeah
mother,
my
suggestion.
We
need
that
yeah
I
would
say
no
there's
easement
simpler
right
could
just
reporting
the
documents
of
Internet.
Now
there
may
be
another
possum
report
us,
which
is
this
thing
that
goes
tributes.
Everyone
yeah.
What
does
it
needs
not
strategies,
and
you.
H
Think
I
think,
depending
on
the
report
type,
if
it's
something
that
can
be
forcibly
triggered,
we
want
to
be
careful
going
to
allow
someone
run
JavaScript
in
the
background
like
if
you
could
arrange,
evil.com
could
arrange
to
get
like
yeah
something
and
to
happen
so
I,
just
just
from
that
risk
I
would
I
would
lean
towards
more
queueing
to
the
next
document
or
the
next
context.
Sorry
comer.
H
H
Like
yeah
I,
don't
know
enough
about
the
different
report.
Types
like
there
was
like
one
time
about
the
storage
corruption
when
I
was
like,
oh
I'll,
be
evict
origins
for
like
quota
usage
railing.
Should
we
also
record
for
that,
but
I'm
like
oh
that
now
we're
talking
about
like
revealing
size
information
like
I,
don't
like
that
seems
like
in
the
realm
of
like
leaking.
H
It's
interesting,
I
didn't
realize
that
we're
logging.
It
sounds
like
that
code
like
if
you
have
assets,
if
you're
on
a
calm
and
your
sub
resource
of
Christa,
be
calm
and
beat
on
that
some
resource
request
fails.
The
report
goes
to
be
calm,
not
dating
I
assume
it
would
fail
the
documents,
like
the
the
documents
origin
earlier
to
some
resources
that
fail.
You
know
if
you
previously
meant
to.
A
G
G
E
G
H
G
Server
right,
oh
right!
Yes,
so
he
comment
like
ap
request
successful.
My
definition.
Server
knows
that
the
reason
resource
was
requested
and
if
it
failed,
that's
a
problem
to
insult.
Basically,
we
wanted
to
just
let
you
know
that
fetches
against
your
word
and
are
failing
for
some
where
they're
coming
from
is
maybe
not
so
admitted.
G
H
Context
or
the
tank
was
kind
of
like
this
doesn't
seem
actual.
You
know
like
similar
to
our
conversation
here
like
how
were
you
actually
to
do
with
these
things?
So
it's
unclear
to
me
if
we're
really
gonna
move
forward
with
it
for
the
1/2
like
at
that
web
apps
group,
I'm
gonna
have
a
session,
hopefully
just
proposing
this
is
the
fruit
transparency
be
like.
What's
our
plan
for
interruptions
in
general,
so
I,
don't
think
it's
energy
use
case.
It's
gonna
happen.
A
E
G
H
H
Need
someone
right,
I
know
I'm
a
big
fan
of
like
if
I'm
gonna
build
something
I
would
like
to
have
someone
who
say
that
they're
gonna
use
it
so
I
know
I
build
it
right.
You
know
to
make
it
useful
and
lease
the
people
I'm
talking
to
right
now,
who
are
telling
me
we
would
not
use
it
unless
we
could
like
use
our
own
recording
infrastructure
and
so
I
would
need
a
friend
someone
else
who's
interested
enough.
You
know
everyone
will
use.
The
report
end
point
which
might
be
out
there.
H
D
H
H
D
A
No,
it's
basically
for
that
use
case
to
meet
those
requirements
of
not
creating
a
new
reporting
endpoint.
They
would
need
to
register
reporting,
yeah
reporting
observer
somewhere
and
get
those
reports
magically,
even
if
they
happened
when
no
nothing
it
was
around
which
would
effectively
be
on
top
of.
D
H
D
Being
said,
hey,
we
want
your
browser
to
get
us
some
data
right,
just
like
now
we're
here
along
and
get
before
it
was
a
spots
back
right.
You
would
build
it.
You
just
republished
from
the
browser's.
That's
what
I
thought
you
were
hinting
at
is
that
there
is
an
internal
site
that
once
something
like
network
error,
logging,
yeah,
I'm,
trying
not.
D
D
H
Easier
to
build
something
in
progress,
I
I
was
trying
to
do
a
more
holistic
solution
for
these
problems.
I
think
that
would
give
him
the
tag.
Feedback,
I've
gotten
given
I
miss
every
statement.
Reporting
API
like
I'm
gonna
like
restructure
my
conversation
at
webapps
group
to
be
more
of
a
like.
What
do
we
want
to
do
about
storage
corruption
in
general?
Would
transparency
be
useful,
like
is
it
worth
doing,
work
to
try
to
work
with
reporting
ain't
got
to
like
add
this
new
feature
that
we
would
you
know
to
help
support
this
use
case.
H
D
D
In
the
drifts
above
the
thing
that
happened
here,
the
discussion
yesterday
was
that
there
was
a
discussion
where
both
Mozilla
and
Apple
said
that
they
are
not
in
favor
of
the
concept
right
now,
a
persistent
reporting
as
a
concept.
They
are
very
much
behind
the
idea
of
a
per
document.
Recording
the
dreamer
episode.
L
H
C
C
A
K
B
H
A
A
And
I'm
I,
don't
know
why
can't
I.
Just
probably
the
bigger
thing
is
for
that
pulse
we.
So
our
customers
can
define
conversion
metrics.
Some
of
those
conversion
metrics,
might
be
a
URL
but
check
out
page.
We
don't
increment
those
metrics
on
that
courthouse.
So
if
you
go
to
like
a
checkout
page
and
then
back
then
forward,
we
wouldn't
double
count
that
check
out.
G
A
B
C
H
H
A
G
A
J
J
G
Yeah
that
one
we're
playing
that
reports
of
rather
seriously
no
we're
not
just
implying
that
the
Ranger
or
constructor
interfaces
and
I
predict
that
the
length
issued
that
will
be
open
to
that
should
they,
but
there
are
bigger
days,
long
haul,
so
they
should
be
stretching
pluggable.
If
you
wanted
to
send
it
to
your
worker,
you
could
what
post
my
said,
yeah
but
I
think
they
probably
shouldn't
be
intercepted
by
refrained
River
in
that
case,
just
if
you
need
to
get
them
to
do
something,
the
worker
with
them
engine
or
yeah,.
G
D
D
A
G
A
A
H
G
B
G
G
Actually
that
you
guys
are
getting
averages
I've
seen
this
a
beautiful
cut,
because
people
put
relative
paths
and
then
the
policies
on
some
arbitrary
sums
of
me
and
then
they
can
figure
out
who
I
don't
know
all
the
words
the
source
of
that
so
does
any
time
to
have
absolute
paths.
Always
anyway,
that's.
G
E
E
M
A
Right
yeah
I
wouldn't
want
I,
wouldn't
want
this
to
block
people
from
migrating
hard
work
all
right
to
to
reporting
at
the
same
time,
I
like
do
they
have
an
actual
use
case
as
to
what
like,
if
they
were
migrating
anyway,
they
need
to
change
stuff.
Can
they
ultimately
change
with
I?
Have
like
the
absolute
leaders
any
much.