►
From YouTube: IETF112-WPACK-20211110-1600
Description
WPACK meeting session at IETF112
2021/11/10 1600
https://datatracker.ietf.org/meeting/112/proceedings/
A
A
A
Just
a
reminder:
we're
gonna
record
this,
please,
if
you're,
not
speaking,
make
sure
your
your
microphone
is
muted.
If
you
wanna
come
off
turn
your
video
on
when
you're
speaking,
that's
fine.
This
room
is
small
enough
that
it
shouldn't
really
crash
it.
There's
some
links
there
for
the
useful
things
for
the
links
for
the
meat
echo
audio
only
in
cody
md.
So
next
there
we
go
click
on
the
next
slide.
This
is
the
note
well.
A
This
relates
to
you
know
all
the
the
ipr
notices
that
we
have.
You
know
how
to
working
group
progress
all
kinds
of
stuff
I'll
get
to
the
annual
harassment
things
later
in
the
code
of
conduct
the
copyright
patents.
Basically,
if
you
know
anything
about
ipr
in
the
space,
you
declare
it
before
you
participate,
and
that
includes
basically
communicating
in
any
way
to
the
working
group.
Next
we're
gonna
go.
A
A
A
Yeah
use
our
best
engineering
judgment,
find
the
best
solution
for
the
whole
internet
and
contribute
to
the
ongoing
the
itf
and
remember
that
we'd
like
this
to
apply
everywhere,
not
just
in
not
just
on
the
mean
echo
channel,
but
also
in
the
mic,
jabber,
media,
etc
and
the
mailing
list.
For
that
matter
all
right.
I
kind
of
know
that
that's
us,
the
agenda
is
pretty
straightforward.
Like
I
said
I
did
request
an
hour
and
they
gave
us
two,
so
I
basically
just
split
it
down
the
middle.
A
We
have
two
documents
that
we've
adopted.
We
have
more
that
we
need
to
adopt,
but
these
are
the
two
that
we're
starting
with
and
kind
of
just
split
them
between
use
cases
in
bundle.
I
do
note
that
I
put
felipe's
name
on
the
bundle
draft.
I
think
it's
probably
just
going
to
be
jeffrey
and
that's
really
it
we
didn't
have
any
particular
slides.
A
B
Switch
first,
yes,
there
is
one
thing
that
we
skipped
right
over
when
that
is,
we
need
a
scribe.
Please,
oh,
yes,
take
notes
into
the
cody
md.
Hopefully
other
people
keep
an
eye
on
it
and
the
chairs
will
definitely
go
over
it
afterward.
But
somebody
to
get
the
initial
set
of
notes
into
the
cody
would
certainly
be
helpful.
B
A
B
B
D
Right
so
the
the
use
cases
document
is
basically
the
the
content
that
you
all
have
seen
before.
It's
still
on
my
plate
to
to
split
out
the
the
controversial
use
cases
into
an
appendix
and
leave
the
rest
in
the
in
the
main
section.
I
believe
that
the
primary
non-controversial
use
cases
are
basically
the
the
prefetch
of
of
content
unit,
of
like
sub
resources,
and
so
so
just
about
everything
else
will
will
go
into
the
appendix.
D
C
D
Yeah
at
least
roughly
bright
priority-
it
was.
It
was
not
kind
of
in
the
in
the
top
set
of
things
that
that
I
was
interested
in
when
starting
the
the
work
kind
of
we
were.
We
were
looking
at
at
ways
to
replace
or
ways
to
to
compete
with
native
apps,
where
you
can
share
the
the
installation
file,
but
it
seems
like
there's
kind
of
we
can.
D
B
B
Do
you
have
thoughts,
perhaps
that
you
could
even
then
share
with
larry
on
the
list
for
others
to
digest,
because
larry
keep
coming
back
to
this,
my
multi-part
thing,
and
so
far
we
haven't
seen
energy
in
the
group
to
go
down
that
particular
path.
But
if
you
had
some
some
relevant
thoughts,
it
would
certainly
be
welcome.
D
A
A
A
E
E
Okay,
so
I'm
alex
from
realia
we've
been
working
in
partnership
with
io
feel
free
to
reach
out
after
the
meeting.
If
you
have
any
questions-
and
this
is
a
short
update
on
the
work
that
we
have
been
doing
on
the
use
case
of
resource
preloading-
using
bundle
responses,
you
can
have
a
look
at
the
explainer
for
more
details.
E
In
the
last
meeting
of
the
group,
we
discuss
some
changes
to
keep
the
spec
as
simple
as
possible
and
to
focus
on
consensus
functionality
like
reloading,
while
also
allowing
for
additional
sector
sections
that
may
extend
the
format
for
specific
use
cases
related
to
the
implementation
of
this
spec.
There
is
a
node.js
module
to
create
and
read
bundles,
and
one
of
the
things
that
we
have
been
doing
over
the
past
weeks
has
been
to
update
this
tool
to
support
the
new,
updated
version
of
the
spec,
while
also
keeping
compatibility
with
the
previous
version.
E
E
So
that's
what
we
have
done.
Our
client
parses,
the
declarations
in
the
html
file
that
lists
the
resources
that
the
page
wants
to
preload
and
then
sends
an
http
request
with
a
specific
header
that
lists
the
resources
that
are
not
in
the
case.
In
the
case
of
the
prototype,
we
are
using
a
service
worker,
so
the
cashier
is
the
service
workers
case,
and
then
the
bundles
or
those
requests
are
served
by
a
small
web
server
that
supports,
where
abandoned
preloading
with
subsetting
and
is
able
to
create
bundles
on
the
fly.
E
E
One
benefit
of
this
prototyping
strategy
is
that
it
wouldn't
take
much
to
try
out
different
syntaxes
for
this.
We
have
a
couple
examples
at
the
moment.
In
the
experience
repository
the
first
one
is
a
very
basic
test
where
some
resources
are
pre-loaded
and
you
can
use
the
dev
tools
in
the
browser
to
see
how
those
resources
are
retrieved
and
then
one
by
one.
They
are
made
available
in
the
cashier,
so
they
are
retrieved
in
one
request,
but
then
they
become
several
entries
in
the
service
workers
casino.
E
The
the
second
example
is
a
small
react,
application
that
uses
webpack
as
a
bundler
tool
and
in
this
case
bundle
preloading
is
already
able
to
improve
the
performance
because
the
resources
can
be
loaded
in
parallel
in
one
single
request,
whereas
by
default,
what
webpack
does
is
that
it
loads
the
javascript
and
then
the
javascript
loads.
The
rest
of
the
resources,
so
it
happens
in
sequence,
which
is
slower.
E
So
what
have
we
learned
so
far?
Well,
first,
that
this
approach
works
for
prototyping
and
for
trying
out
a
solution
for
this
use
case.
It
already
allows
us
to
test
different
scenarios
where
the
solution
might
be
used
and
it
gives
us
an
easy
way
to
experiment
with
different
syntaxes
for
the
html
tags
or
for
the
http
headers.
E
Of
course,
using
a
service
worker
has
its
own
overhead,
but
other
than
that,
we
are
seeing
that
the
solution
has
good
potential
for
improving
performance
in
the
future,
because
we
are
seeing
that
we
can
reduce
the
number
of
requests
response
round
trips.
We
can
parallelize
downloading
resources,
we
can
put
individual
resources
in
the
cache
and
so
on.
E
E
We
want
to
create
more
examples,
so
we
are
really
looking
for
ideas
of
scenarios
where
people
feel
that
this
could
be
a
good
solution
and
something
that
they
would
like
to
try
out,
and
a
final
thought
is
that
I'm
really
wondering
if
this
approach
of
prototyping
and
then
iterating
on
that
can
also
be
applied
for
other
use
cases
that
we
have
on
the
table
and
that's
all
for
my
short
update.
If
you
have
any
questions.
Thank
you.
E
Yeah,
one
of
the
things
that
we
want
to
try
out
is
some
benchmarking,
because
we
want
to
see,
for
example,
what
happens
if
the
client
requests
a
lot
of
resources
in
in
one
go?
Does
that
have
a
very
big
impact?
Does
that
create
a
very
large
http
request
and
has
to
travel
to
the
server
and
so
on?
So
those
are
the
kinds
of
mortgage
cases
that
we
that
we
want
to
try
out.
B
B
D
The
the
biggest
question
that
I
that
I
discovered
recently
was
how
to
interact
with
the
new
http
working
groups,
binary
hdp
format,
the
kind
of-
and
I
forwarded
it
to
the
list.
But
the
the
direction
that
that
seems
appropriate
to
me
is
try
to
try
to
converge
them,
even
though
that'll
that
probably
involves
several
changes
to
the
to
the
bundle
format.
B
B
But
yes,
javon
has
said
in
chat.
Convergence
here
would
be
good
minimizing
the
differences.
So
until
we
get
a
little
bit
more
information,
though
about
what
challenges
might
be
involved
there,
you
know,
I
think
you
and
martin
have
to
start
seeing
how
you
can
work
it
all
out
and
then
what
issues
that
that
might
bring
up.
D
A
A
A
A
B
D
B
A
Yeah,
that's
my
understanding
as
well
that
if
you,
if
you
go
ahead
and
update
the
document
and
then
people
will
comment,
then
I
think
you
can
engage.
People
will
be
engaged
to
be
like
yes,
that
one
is
in
or
out
that
the
ones
that
are
still
in
the
front
at
the
front
part
of
the
document,
and
then
we
can
actually
get
to
whether
we're
done
it
might
be
as
easy
as
just
moving
things,
and
that
would
be
great
and
then
we
could
focus
our
energy
elsewhere.
A
I
highly
doubt
that
I'm
sure
this
would
be
something
that's
controversial,
but
I
yeah
it's
basically
just
reorganize
the
document
and
then
we'll
go
from
there,
because
now
it's
a
working
group
document
that
one
of
the
reasons
why
you
want
to
submit
the
old
version.
The
way
it
is
is
so
that
when
you
submit
the
next
version,
that's
the
01
version
of
the
working
group.
You
can
see
what's
changed
as
part
of
the
working
group
document
process
and
all
that
stuff.
A
The
only
thing,
the
only
other
thing
I
would
add,
is
just
to
make
sure
that
you
know
do
it,
do
it
in
the
github,
so
it's
open
and
people
can
see
what's
going
on,
and
if
we
have
any
questions
we
can
just
refer
them
to
the
list.
We
can
cc
the
the
list.
We
probably
should
go
ahead
and
get
wpac
mailing
list
set
up.
So
there
are
weekly
summaries
sent.
I
think,
mark
nottingham
has
a
cool
tool.
B
Okay,
the
audio
didn't
actually
come
through,
but
it
was
brief
enough
that
I
thought
that
that
was
and
felipe
must
have
done
the
same.
You
have
to
give
it
a
half
a
second
for
the
audio
actually
channel
to
connect
but
sean.
I
think
we
are
probably
in
agreement
that
much
like
we
expected
with
only
needing
an
hour
session.
I
think
that
we're
gonna
be
ending
early
here
at
only
half
an
hour
right.
Do
you
have
any
other
business
you
need?
John?
A
Do
have
other
drafts
in
the
working
group
in
our
statement
of
work
that
we
agreed
with
the
isg
and
so
at
some
point
we'll
need
to
begin
engaging
on
those,
and
I
guess,
there's
probably
some
other
things
that
need
to
get
done
first.
But
if
you
want
to
start
working
on
those
drafts
feel
free,
you
know
we
basically
will
start
the
process
the
same
way.