►
From YouTube: 2019-03-07 Rust and WebAssembly Working Group Meeting
Description
A
Oh
welcome
to
another
arrest
and
webassembly
working
group
meeting,
alright,
so
right
off
the
bat
I
just
want
to
mention
that
the
template
for
our
meetings,
our
schedule,
template-
has
changed
a
bit.
I've
updated
this
to
reflect,
essentially
our
goals
for
the
for
2019
from
the
roadmap
yeah,
so
RFC
triage.
There
is
a
proposed
amendment,
the
RFC
process,
which
is
mostly
relevant.
If
you
are
a
core
member
of
the
working
group,
so
read
that
and
take
a
look
if
you'd
like
and
jumping
along
into
status
updates.
A
So
the
glue
toolkit
is
the
name
for
the
modular
toolkit
that
we
intend
to
build.
It
is
not
really
ready
to
be
built
yet
I've
been
trying
to
create
a
bunch
of
like
contributor,
Docs
and
kind
of
like
checklists
and
and
things
like
this,
and
so
I
will
continue
doing
that
and
hopefully,
by
this
time
next
week,
it
will
actually
be
ready
for
people
to
be
contributing
to
it.
B
Yeah,
given
the
opportunity
to
talk
about
this,
we
do
actually
have
a
little
bit,
which
is
so.
This
is
a
PR
linked
here,
which
is
enabling
rayon
to
be
compatible
with
whether
simply
threads
and
everything,
and
so
I
have
the
work-in-progress
so
starting
to
get
to
get
that
used,
and
so
I
have
like
a
very
ad
hoc
demo
of
actually
using
rayon
for
the
ray-tracing
example
that
we
currently
have
so
it's
making
progress.
It
still
is
very,
very
buggy
and
still
working
through
some
things.
B
There
one
was
fixing
the
memory
allocator
in
the
state
of
libraries
I'm
just
waiting
for
that
to
get
the
nightly.
After
that
we
can.
They
probably
make
some
more
progress
here.
So
it's
good
it's
coming
along,
but
it
still
gonna
take
a
little
bit.
Take
it
kind
of
like
a
rayon
style
demo.
Also
nomograms
tell
demo
nice.
A
B
Not
yet
that
was
I've
gotten
to
the
point
where
I
can
hit
the
button
in
like
50
percent
of
the
time,
it
actually
works
to
get
a
measurement,
and
so
it
was
like
it
wasn't
really
that
much
slower
than
before,
but
I
suspect
that,
like
doing
some
coarser
grained
parallelism
rather
than
every
single
pixel
is
gonna,
make
it
just
as
fast
and
as
the
previous
version,
if
a
little
bit
faster,
so
I'm
not
actually
too
worried
about
performance.
That
was
just
something
I
wanted
to
bring
up.
A
So
the
alligator
thing
for
the
people
who
aren't
familiar
is
that
the
main
thread
can't
block
and
like
weight
on
a
mutex
or
a
con
bar,
or
anything
like
that,
so
allocation
on
the
main
thread.
The
alligator
is
shared
with
all
the
worker
threads
as
well
right,
it's
so
like
if
another
worker
thread
is
using
the
alligator
in
the
main
that
I
wants
to
allocate.
What
do
you
do?
That's
essentially
the
short
version
of
the
problem:
debugging
I,
don't
think
we
have
any
updates
here.
B
One
thing
we
might
want
to
put
like
try
and
figure
out
more
in
the
long
term
is
the
whole
point
of
walrus
for
internally
being
used.
It
wasn't
vengeance
for
a
large
part
of
its
preserving
to
work
to
big
information
and
so
I
wonder
if
Nick,
if
there's
something
you
and
I,
we
can
do
it's
like
to
kick
off
some
contribution
areas,
something
like
that
mm-hmm
or
if
that's
just
like.
B
A
I
think
it
would
be
good
to
sit
down
and
come
up
with
a
plan
for
exactly
like
what
we
want.
The
final
leg
code
to
look
like
in
Morris
and
then
at
least
write
that
up
and
then
see.
If
we
can
chop
it
into
the
pieces
that
can
and
gonna
be
handed
out
to
folks
who
don't
have
the
whole
picture
in
their
head.
That
makes
sense.
A
C
I
definitely
missed
my
deadline
that
I
had
set
for
myself
to
get
was
impact
about
seven
out
by
February
15th,
but
that
has
the
kind
of
nice
benefit
of
the
fact
that
we
can
probably
get
the
one
data
CLI
output,
PR
that
alex
is
made
in
for
that
release
and
definitely
hit
March
15th,
if
not
really
kind
of
enter
this
week.
Probably
in
order
to
merge
that
I'm
gonna
need
to
pull
it
down
and
regenerate
some
gifts
and
check
that
out.
So
that's
the
thing
that
I
need
to
do.
C
I
think
there's
also
another
one
for
fixing
Wes
and
pick
on
on
rust
up
setups
that
we
should
probably
be
able
to
get
in.
It
looks
like
there's
something
that's
causing
all
of
our
tests
to
fail
currently,
which
I
think
is
a
car
go
test
lock
tissue
if
I'm,
seeing
things
correctly,
but
I'll
need
to
check
that
out.
But
that
is
where
that
is
at.
C
B
So
it
was
I've
talked
a
couple
of
weeks
ago
about
kind
of
tracking
webassembly
sizes
over
time.
That
is
not
a
profiling
aspect,
just
the
how
bar
how
big
our
binaries
and
GS
things
are,
and
so
I'm.
The
second
link
here
is
a
github
pages
generated
website
which
has
no
contextual
information.
It's
just
a
bunch
of
numbers,
but
the
idea
here
is
that
this
has
been
periodically
updated
over
the
past
week
automatically,
and
so
this
is
just
kind
of
we
fire-and-forget
and
the
rough
idea
is
that
there's
a
number
of
benchmarks
and
metrics.
B
We
want
to
track
on
it.
So,
for
example,
the
first
one
here
is
Twiggy,
which
is
Twiggy.
The
repository
has
a
resume
thing,
and
so
it
just
CDs
into
that
and
runs
Westpac
build
and
then
measures
everything.
And
so
this
measures,
like
the
top
line,
is
just
the
actual
Azzam
coming
out.
It
wasn't
by
gin
and
then
that
orange
line
is
then,
if
you
actually
gzip
it.
B
So
in
theory,
we
can
get
and
it
may
be
a
sense
of
like
a
100
byte
increase
in
wisdom
doesn't
matter
that
much
because
it's
two
bytes
of
an
increase
in
something
else,
but
so
I'm
not
sure
how
useful
this
is.
But
I
wanted
to
say.
This
is
I
just
wanted
to
put
this
up
as
something
that
I'm
working
on
and
see.
If
all
those
have
thoughts
of
how
to
make
this
more
useful
than
it
currently
is,
which
is
a
bunch
of
flatlines.
B
One
thing
I
would
actually
point
out
as
well
as
if
you
at
the
top
there's
this
check
box
saying,
show
absolute
data.
So
if
you
uncheck
that
you'll
actually
see
that
on
the
game
of
life,
for
example,
there's
a
really
big
spike
of
an
increase
in
size
and
then,
if
you
mouse
over,
that
it'll
show
you
that
changes
were
actually
related
to
as
a
bind.
B
A
B
B
So
the
GS
got
a
little
bit
bigger
and
then
also
the
Y's
and
got
a
little
bit
bigger
because
now
we're
putting
in
realloc
and
things,
but
so
I
really
know
how
to
interpret
this
we're
like.
If
you
show
relative
data
to
from
the
left
hand
mouse
point,
you
can
actually
see
some
nice
bikes
and
what's
going
on
there.
But
if
you
go
back
to
the
absolute
data,
you'll
notice,
how
you
can
barely
tell
any
difference
at
all,
so
I
don't
really
know.
B
C
Haven't
done
dough
too
deep
into
this
y-axis
or
some
seeing
it,
but
maybe
we
should
default
to
showing
the
relative.
So
that
seems
like
the
more
useful
view,
at
least
on
my
immediate
look
and
then
just
because
it
kind
of
it
gives
us
that,
like
change
over
time
view,
which
is
what
I
think
we
were
mostly
looking
for
at
least
for
us
on
the
team.
That
would
be
like
my
first
suggestion
said.
B
C
If
there
was
a
way
to
look
at
like
the
versions
and
a
way
to
like
click
into
like
a
change
log
or
a
release
that
might
be
kind
of
nice,
because
then
we
can
because
like
when
you
see
changes,
the
first
question
you
usually
ask
is
like:
oh
there's
a
change.
Why
did
that
change
happen
and
I
see
that
there's
a
link
to
the
rest?
See
changes
oh,
never
mind.
Actually
you
have
this.
If
you
hover
over
I
did
see
for.
B
B
I
mean
one
thing:
I
wanted
to
point
out
as
well
as
iosef
I,
personally
like
the
idea
of
a
relative
as
opposed
to
absolute,
but
one
of
the
downsides
of
that
is
that,
if
this
is
it's
unlike
unlikely
to
show
good
information,
where
it's
probably
just
going
to
be
a
bunch
of
like
hey,
look
we're
always
up
into
the
right
or
like
we're
getting
a
little
bit
more
Evernote
over
and
over.
But
so
it
may
not
be
in
terms
of
that.
C
Don't
know
also
maybe
showing
speed
somehow
on
this,
which
is
like
a
huge
nightmare
and
will
require
a
lot
more
design
might
help
to
show
that
there's
like
a
trade-off
between
size
and
speed.
It's
actually
it's
true,
yeah
I,
think
I
think
trying
to
give
some
context,
but
in
general
I
think,
just
being
honest
would
be
useful.
The
absolute
numbers
are
nice,
but
I'll
admit
when
I
first
landed
on
the
page
it
kind
of
looked
like.
Maybe
it
wasn't
working.
A
A
C
Thing
that
we
we
maybe
could
do
that's
like
a
much
better
idea
than
like
the
equivalent
is,
is
like
I
assumed
that
we
probably
have
some
sort
of,
maybe
not
a
specific
one,
but
like
kind
of
a
magic
number
and
our
heads
where,
if
we
saw
it,
we
go.
Oh
that's,
definitely
too
big,
and
so
perhaps
if
we
can
just
like
gave
some
sort
of
like
like
when
we
all
see
the
graph
like
kind
of
hit.
That
point
like
we
all
kind
of
have
like
a
gut
check
that
goes.
Oh,
we
like
that's,
probably
wrong.
C
We
should
go
check
something
maybe
just
giving
that
type
of
context
would
help
people
see
the
graph
also
to
be
like.
Basically
it
says
we
see
this
graph
and
even
though
it's
going
up
into
the
right,
we're
not
worried
to
like
give
them
some
sort
of
like
understanding
of
like
oh,
this
is
good
versus
bad
versus.
This
is
expected
versus
not
yeah.
D
Fun
metric
that
I
sometimes
applies
like
how
many
day
and
how
many
data
packets
can
we
fit
it.
That's
like.
Oh.
We
now
need
five
PCP
packets,
that's
like
if
you
don't
considered
a
growing
thing,
they're
like
cool
we're
now
down
to
three
and
it's
like:
oh
cool,
benchmark
or
achieving
might
be
fun.
I
know.
B
Like
one
thing,
that
I
think
would
be
really
cool
is
to
have
annotations
on
here.
I
would
love
to
see
like
like
I
want
to
after
we
see
a
spike
or
an
inflection
point
say
like
oh,
this
is
why,
like
I
want
to
annotate
on
here
saying
Oh
228
like
oh,
whatever
the
new
Western
budget
version
is
like
this
is
the
exact
PR
that
did
it.
This
is
exactly
what
we
know
and
we're
like
and
either
here's
the
bug
fix
over
here's
the
person
that
oh
praise,
because
they
made
it
twice
smaller
or.
B
A
So
next
agenda
item
is
basically
in
the
leading
up
to
the
rest,
2018
edition
we
kind
of
synched
our
sprints
with
the
rust
release
cycle.
It's
like
every
six
weeks.
We
kind
of
like
did
a
little
planning
session
and
chose
what
do
we
think
we
could
do
in
the
next
six
weeks
and
tried
to
have
you
know
soft
time
limits
for
ourselves
and
ship
things
and
I'm
wondering
if
people
feel
like
that's
a
useful
kind
of
organizational
tool
to
do.
A
C
Like
that
idea,
I
think
it's
nice
to
have
a
sense,
particularly
when
you
want
to
get
contributors
to
like
have
some
sort
of
plan
to
be
like
these
are
the
things
to
work
on
and
being
able
to
flag
those
I
think
works
and
I'm
trying
to
do
a
monthly,
really
sesh
with
wasm
PAC.
But
six
weeks
should
be
definitely
people
like
that's
exactly
where
we're
at
right
now,
but
I
like
the
idea
of
having
some
sort
of
cadence.
So
we
can
feel
like
we're,
making
progress
or
something
so
I'm.
A
big
fan.
A
A
The
last
time
we
did
this
was
having
a
single
issue
label
that
we
can
use
across
all
of
our
organizations
repositories,
because
that
way
you
can
see
the
whole
list
of
everything
in
one
place
and
like
I,
was
sorting
that
by
at
least
recently
updated,
which
was
really
useful
for
like
finding
things
that
no
one
else
was
working
on.
Essentially.
C
So
we
got
the
bookshelf
merged.
Hey
I
opened
another
PR
to
see
if
we
could
get
the
Lawson
bookshelf
merged
into
the
roast
bookshelf,
and
there
was
a
little
bit
of
discussion
there
that
they
would
prefer
to
not
link
to
external
documentation
that
if
we
would
like
to
have
a
rest
and
webassembly
book
brought
in
the,
we
should
create
a
sub
module
that
we
can
import
and
the.
C
C
C
C
A
I'm
yeah,
so
it's
it's
more
than
just
six
weeks
of
latency
because
it
has
to
ride
the
trains
as
well
right.
So
it's
like
six
weeks
until
the
first
beta
and
then
six
weeks
from
beta
till
stable,
which
I
don't
know,
I
guess
we
can.
We
can
discuss
this
in
an
issue
and
like
read
over
all
the
discussion
and
everything,
but
it
just
seems
like
such
a
long
time
for
our
tool
chain.
That's
moving
relatively
fast
I.
B
Would
imagine
that
maybe
the
book
would
be
good
for
this,
so
like
the
game
of
life
tutorial,
because
it
isn't
changing
at
the
same
velocity
as
wasn't
by
mine,
Jen
was
in
pack,
but
I.
Don't
think
that
those
books
would
be
suitable
for
conclusion
for
stub
module
inclusions
of
a
slang
gross
no
I
I
would
like
to
make
a
stronger
case
for
why
we
should
be
externally
linked
and
at
this
time,
but
I
don't
want
to
reopen
old
wounds
and
that's
not
something
to
discuss
here.
That's
something
to
discuss.
C
So
we
have
this
conversation
briefly
when
we
were
talking
about
the
2018
Edition
website
and
the
docs,
and
just
really
seeing
the
edition
and
I
think
it
could
be
wrong
here,
but
I
think
it's
largely
preferable
for
it
to
be
bundled
in
because
the
experience
of
Rostock,
open
and
being
able
to
do
that
thing
locally
is
what
is:
is
desired
to
be
mapped
to
what
is
published
on
the
bookshelf
and
so
I.
Think,
fundamentally,
that
that's
really
the
reason.
C
But
again
we
do
also
have
the
website
which
links
out
to
all
of
the
other,
docs
and
I.
Think
that's
supposed
to
be
kind
of
understood.
The
trade-off
visibility
was
okay,
but
yeah
I.
Think
it's
totally
worth
asking
and
getting
like
documented
in
an
issue,
but
I
had
the
same
question
because
I
was
like:
why
do
we
have
to
bundle
all
the
books
into
the
like?
Why
do
we
have
to
do
this,
and
that
was
what
I
was
okay.
C
I
do
really
like
the
idea
like,
if
you
think
about
what
is
generally
contained
in
the
rest,
books.
I
think
the
game
of
life
tutorial
is
like
are
very
valid,
like
good
thing
to
include,
and
while
yes,
there's
like
the
reference
API
in
the
bookshelf,
rust,
obviously
moves
slower
than
our
libraries
do
and
so
I
think
just
having
the
game
of
life
tutorial
and
then
like.
We
could
just
create
stub
pages
that
are
like
to
read
the
most
up-to-date
version
of
this.
Like
click
this
link
and
we
still
get
the
visibility
aspect
of
it
or.
C
A
B
That
will
be
it'll.
That'll
require
a
new
release
of,
as
both
wasn't
mentioned,
impact
because
we're
gonna
want.
You
definitely
want
to
use,
wasn't
packing
the
book
and
it's
gonna
need
a
new
flag
for
new
flags,
and
so,
if
you
switch
to
new
modules,
I
think
that's
totally
fine
as
an
interim
step.
I
think,
because
it's
pretty
easy
to
switch
from
no
modules
to
whatever,
whatever
we
call
it
and
so
I
wouldn't
want
to
I.
Don't
want
to
block
the
improvements
on
the
book
just
because
we
haven't
I.
Think.
C
A
new
flag
that
just
like
maps,
something
that
wasn't
by
gen,
does
is
not
too
hard
so
and
since
I
mean
if
it's
something
that
we
wanted
to
do
bike
before
the
next
wasn't
pike
release
I'd
be
willing
to
commit
to
do
that
work.
So
we
can
get
it
out.
So
it's
not
blocking
I
do
think.
There
was
an
issue
just
recently
filed
I
believe
on
wasn't
buying
gen.
C
We
might
I
guess
some
of
this
is
in
some
of
the
RCS,
but
we
might
want
to
do
a
short
review
of
what
we
call
these
targets
and
where
we
expect
them
to
work,
because
it
does
seem
like
there's
largely
a
fair
amount
of
confusion
and
I.
Also
just
based
on
comments.
I've
read
from
folks
I
think
we
all
kind
of
agree
that
we,
like
don't
love.
C
A
B
For
what
it's
worth
early
on
in
rust
history,
we
used
to
require
full
sign-off
from
everyone,
and
it
was
I
think
it
was
a
little
bit
before
1.0
or
sometime
after
1.0.
Then
we
ended
up
starting
some
really
realizing
that
it
was
just.
It
was
way
too
hard
to
get
a
full
sign-off.
It
ended
up,
for
it
was
always
a
reason.
B
One
person
thing
one
one
person
there
was
taken
up
or
would
take
a
long
time
to
get
around
to
it,
and
it
ended
up
slowing
down
the
process
a
lot
more
than
it
was
so
I
think
it's
worked
out
really
well
and
wrestling
rested
to
say
that
we
need.
We
don't
need
100%,
but
you
need
a
very
reasonable
amount
of
time,
for
everyone
was
to
say
everything
than
otherwise
most
people
to
be
on
board,
and
so
that
that's
a
big
example
I
think
I've
seen
of
not
having
100%
sign-off.
C
Interesting
to
look
at
the
precedent
set
by
rust,
because
that
is
for
merging
in
RFC,
but
it's
not
for
stabilization
and
because
we
have
a
more
lightweight
process.
We
kind
of
allied
those
and
I
do
think
that
you
know
I
mean
again,
it
depends
on
which
team
you
are
looking
at.
But
it
is
my
understanding
and
perhaps
I'm
wrong
here-
that
the
language
team
requires
everybody
to
agree
before
stabilization
is.
B
C
It
yeah
so
I'd
love
to
comment
on
this
RC
and
just
to
like
state
the
elephant
in
the
room.
It's
like
this
is
about
me
being
unavailable
for
the
last
couple
months
like
for
sure
and
I.
Think.
The
one
thing
that
I
just
want
to
make
sure
that
we
were
really
careful
about
is
I,
agree
that
there's
a
failure
mode
here
and
I
think
that
the
things
proposed
in
the
RFC
are
not
necessarily
unreasonable.
D
C
C
I
realized
that
and
I
got
pinged
and
I
was
like
alright
I'm,
just
gonna
like
just
say
you
to
ship
it,
but
I
also
think
that
one
of
the
other
interesting
differences
between
this
team
and
some
of
the
other
teams
is
how
incredibly
small
the
core
team
is,
and
if
we
look
at
the
makeup,
I
mean
even
if
we
think
about
last
year,
there's
also
incredibly
homogeneous.
All
three
people
on
the
core
team
were
full-time
employees
of
Mozilla
and
so
I
think
I
would
be
significantly
more
comfortable
emerging
this
RFC.
C
If
we
were
able
to
expand
the
core
team
beyond
people,
because
I
don't
want
there
to
be
a
situation
and
again,
this
is
not
suggesting
that
anyone
is
acting
in
a
malicious
way,
but
we
don't
want
to
be
in
a
situation
where
we
have
a
process
that
will
preference
folks
who
are
working
full-time
on
this
specifically
and
again,
I.
Don't
think
anyone
would
do
that,
but
I
think
it's
important
to
just
like
kind
of
keep
an
eye
on
that,
and
obviously
you
know
solve
it
before
it
ever
would
become
a
problem.
I.
A
C
And
I
guess
I
would
also
just
say,
like
to
the
extent
that
things
are
being
held
up
in
a
way.
That's
unacceptable,
like
I,
definitely
want
to
try
to
incur
this
team.
Like
we're
small,
like
we're
new
we're
close,
we
have
like
a
great
opportunity
to
just
be
able
to
like
talk
to
each
other,
so
we
should.
We
should
do
that.
So
it's
like
things
are
going
really
slow
and
we
don't
like
that,
like
we
should
go.
C
Hey
person
like
what's
up
like
you
having
a
bad
time
like
what's
going
on
and
I
think,
particularly
for
a
new
project
like
we
obviously
can
legislate
this
now,
but
working
on
building
that
interpersonal
communication
is
gonna,
make
the
team
more
robust
in
the
long
term
than
any
type
of
policy.
We
could
possibly
write
I.
D
Can
share
a
quick
thing,
which
is
the
node.js
collaborator
summit
is
going
to
happen
in
May
in
Berlin,
so
I'll,
probably
just
a
tenth
last
year,
also
attended
last
year,
and
it
was
a
at
least
one
session
about
azzam.
So
I
was
thinking
if
anyone
has
anything
or
comes
up
with
anything
over
the
next
two
months.
That
might
be
useful
to
discuss
with
no
jazz
people
I'm
happy
to
act
as
an
ambassador
to
the
no
jazz
project
and
like
represent
the
rest.
Plasm
working
groups
views
what.