►
From YouTube: W3C WebPerformance WG triage call March 8th 2019
A
A
C
A
A
B
D
D
A
D
D
A
A
A
E
E
D
E
D
A
E
I'm
actually
asking
the
opposite:
question
editor:
Nicolas
Pina.
You
have
reviewed
these
ten
issues
when
and
as
the
owner
of
the
specification
and
the
active
editor.
Are
you
confident
that,
when
this
proposal
shifts
it's
not
going
to
cause
the
inability
to
solve
these
issues
in
the
future,
is
that
fair
as
a
question.
D
None
of
them
are
blocking
or
like
what
you
said.
Let
me
go
over
them
to
ensure
that
some
of
them
I
think
are
kind
of
obsolete.
For
example,
there's
one
that's
level
three
proposal:
I
can
close
that
and
things
like
that,
so
let
me
go
over
them
and
I'll.
E
A
Not
necessarily
the
good
like
the
right
metric
here,
because
we
have
specs
that
have
been
shipping
for
years
and
unfortunately,
you
have
many
many
issues:
I
open
on
them,
so
this
is
just
to
yeah
going
over
the
issues
to
make
sure
that
nothing
there
will
require
API
changes
that
will
cause
compatibility.
Issues
is
a
perfectly
sensible
thing
to
do,
and
yes
is
something
that
typically
happens
as
part
of
the
shipping
process
of
the
intent
processing
link.
It
seems.
F
A
Which
which
one
the.
D
A
Cool
so
moving
on.
A
We
finally
managed
to
trim
down
the
number
of
open
issues
which
one
is
just
ship
the
thing
and
the
is
that
we
need
to
switch
the
IANA
registration
of
the
timing,
allow
origin
from
like
from
experimental
to
permanent
once
this
spec
has
shipped.
So
these
are
two
administrative
issues.
The
main
problem
that
we
have
is
with
WP
T's.
A
A
So
thank
you
for
like
various
buffer,
full
related
issues
as
well
as
I'm,
not
sure
regarding
the
supported
resource
type,
any
I'm,
not
sure
that's,
but
it's
also
work,
that's
been
done
recently
and
but
we're
still
left
with
a
list
of
seven
different
tests
that
are
not
passing
in
one
implementation,
so
that
are
only
passing
in
one
implementation.
Sorry
and.
A
A
E
G
G
A
And
I
have
to
say
that
there's
also
Lake
eNOS
related
bug
in
Firefox
that
we
are
hitting
constantly
when
fought
like.
When
touching
those
tests,
we
often
hit
flakiness
on
the
bots
on
the
WPT
BOTS,
with
Firefox
nightly.
That
seemed
to
be
a
result
of
an
implementation
issue
where
some
time
stamps
that
are
supposed
to
be
identical
are
racially
identical,
but
sometimes
are
not.
B
A
G
G
F
Can
totally
file
a
issue
but,
as
that's
father
I
know
like
nobody
is
looking
into
it,
I
saw
I,
don't
I
mean
I,
realize
speaking,
even
if
I
want
to
file
a
bug
nobody's
going
to
look
at
this.
It's
not
if
you
look
like
talking
about
a
couple
month
timeframe.
Unfortunately,
I,
don't
think
we'll
have
anybody.
Looking
this
various.
E
F
E
But
just
from
my
perspective,
I
completely
understand
that
so,
where
I
think
we
should
count
ourselves
fortunate
that
Marcus
and
Firefox
has
an
engineer
committed
here
and
that
the
minimal
ask
is
really
just
about
getting
that
investigation
to
a
point
where
Firefox
agrees
with
the
specification
and
has
a
public
bug
which
can
unblock
shipping.
The
specification.
A
A
A
A
Okay,
and
otherwise
they
are
all
yeah
I,
think
all
of
them
are
also
requests
from
an
effing
gastrin.
Basically,
that
ended
up
is
tests
so
yeah.
If
you
could
prioritize
the
review
or
I
could
just
you
know,
lend
them
and
then
we'll
see
if
you
know
assuming
the
like,
let's
hope
that
they
were
passing
in
all
implementations,
or
at
least
in
two
of
them,
and
in
that
case
it
won't
be
a
problem,
but
if
they
aren't
yeah
it's,
it
would
be
something
to
add
to
the
priority
list.
Yeah.
G
A
And
I
probably
should
yeah
yeah
it's
just
that
I.
It's
tested
I've
written
over
the
last
few
months
that
I
kind
of
forgot
about,
and
they
were
as
part
of
like
the
pr's
open
on
the
WPT
repo,
but
I
haven't
looked
at
them
recently,
so
I
need
to
revive
them,
potentially
rebase
them
and
try
to
lend
them
and
yeah
and
then
we'll
see
if
they
are
passing
or
not
or
I
can
yeah
I
can
test
them
ahead
of
time
to
see
if
they
were
passing
or
not.
A
So,
in
that
case,
I
think
we
can
talk
a
bit
about
navigation
timing
and
the
various
issues
there.
I
would
love
to
also
reach
some
closure
regarding
navigation
timing
in
the
near
future,
so
that
we
can
close
both
navigation
timing,
l2
resource
timing,
l2
and
then
work
on
converging.
Both
one
issue,
which
is
fairly
simple.
We
already
discussed
but
so
issue
95.
A
The
current
wording
is
that
if
any
of
the
redirect
is
passing
unload
event
should
be
exposed,
the
correct
wording
should
be
that
if
all
redirects
are
passing
timing
law,
origin
only
then
should
download
event
be
exposed.
We
agreed
on
that
and
now
I
also
like
we
agreed
on
it
in
a
previous
call.
I
wrote
a
test
as
well
as
a
PR
PR
number
103.
A
A
A
A
D
A
D
So
so,
for
example,
if
worker
does
not
support
tasks,
you
know
you
know
the
worker
performance
and
will
not
have
long
tasks
or
painting
that's
supported.
Ngo
types.
D
A
A
E
A
E
A
A
A
B
A
D
E
E
A
A
Next
one
on
the
list
is
issue.
100
workers
start
should
be
clearly
the
fun
as
applicable
to
the
less
service
workers.
So
this
is
a
leftover
from
the
resource
timing
issue.
It
was
open
initially
on
resource
timing,
but
it
wasn't
really
applicable
there,
but
it
is
applicable
here
there
could
be
in
a
navigation.
There
could
be
scenarios
where
multiple
service
workers
are
involved
and
we
need
to
clarify
that.
It's
the
last
one
which,
like
the
startup
time
of
the
last
one,
is
the
one
that's
reflected
in
worker
start.
A
Those
like
basically
discuss
that
in
a
hen,
hand
wavy
way
and
then
probably
define
it
in
a
tighter
way,
once
we
rebase
everything
off
of
fetch,
but
it
would
still
like
defining
it
in
a
hand.
Wavy
way
would
be
still
a
spec
change,
so
this
would
still
like
I
tend
to
mark
this
as
a
level
two
issue
does
that
make
sense.
A
A
A
And
I'm
not
sure
what
currently
happens
in
HTML
like
what's
the
navigation
type
of
an
iframe
that
like
does
it
inherent
from
the
top-level
frame
or
not
or
from
its
parent
frame,
I'm,
not
100%,
sure
like
what's
the
processing
doing
in
HTML
I
think
we
should
align
with
that,
but
I'm
not
sure
at
the
moment
with
is
and
if
we
can
easily
align
with
it
without
rebasing.
Everything
based
on
fetch
does
that
make
sense.
A
E
A
E
A
A
G
A
A
D
Yeah
so
I
believe,
if
I
understand
correctly,
it's
a
republish.
So
looking
at
issues
outstanding,
there
were
two
and
two
issues:
one
is
for
tests
to
be
clean.
J
C
D
D
B
D
E
D
C
D
So
there's,
for
example,
there
are
several
editions,
even
after
website
entries,
but
for
sex
right.
So.
D
D
I
would
do
it
so
I
guess.
The
question
is
precisely:
what
do
we
want
to
remove?
Is
it
the
buffered
flag,
plus
the
entry
type
specific
parameters?
Is
there
anything
else
we
want
to
remove
from
that?
Alright,
so.
E
Yeah
yeah
cuz.
They
have
to
include
the
entry
type
because
that's
how
the
only
way
to
correctly
write
a
site
that
will
not
break,
and
so
we
chose
to
keep
them
included.
The
reason
we
forced
them
in
l2
was,
if
someone's
going
to
snap
from
one
to
two,
they
have
to
fully
implement,
or
they
start
creating
a
scenario
in
which
websites
can
actually
gently
break
in
certain
browsers.
Oh.
A
D
Yeah
that
that's
that's
fine
by
me,
but
then
in
that
case
some
tests
that
won't
be
passing
and
oh
because
that
those
I
think
that
change
has
not
been
done
in
Safari
and
Firefox.
By
that
change,
I
mean
the
entry
type
specific
parameters
that
it
was
basically
just
add
in
the
dictionary
it's
a
very
easy
change,
so
we
can
wait
for
another
implementation
to
it,
but
then
we
don't
have
a
necessary
test.