►
From YouTube: 2019-03-21 Rust and WebAssembly Working Group Meeting
Description
A
B
A
A
A
A
Jumping
into
status
updates,
so
the
glue
toolkit
has
been
seeing
some
action,
but
it
is
hard
to
summarize
the
action
because
it
has
not
been
very
you
know,
poll
requests
D.
So
far,
it's
been
a
lot
of
discussion
and
kind
of
figuring
out.
What
is
the
scope
and
kind
of?
How
do
you
want
to
design
each
of
these
crates
in
the
tool,
kit
and
I?
Think
we've
come
to
like
some
pretty
good
consensus.
So
far.
If
I
were
to
like
very
quickly
summarize,
it
would
be
like,
wherever
possible.
A
A
C
B
For
us
to
like
correct
so
a
multi-threaded
allocator
issues
and
I
test
it
again,
the
ran
example
and
actually
it
works,
and
as
just
as
fast
as
the
previous
iteration
and
so
I'll
be
cleaning
that
up
and
sending
Apolo
plus.
You
know
that
we
can
start
getting
that
into
Rio
and
then
once
we
should
get
an
example.
We
can
start
talking
about
like
what
do
we?
What's?
An
official
crate
for
reusing,
ran
on
the
web,
because
right
now,
it's
kind
of
a
bunch
of
hacks
and
not
exactly
the
easiest
thing
to
use.
A
C
Wasn't
pack
there's
discussion
on
this
marketing
PR
that
I
had
on
the
Westpac
repo
about
some
new
stuff
for
a
landing
page.
It
kind
of
has
three
calls
to
action.
That
talked
about
the
targets
that
we
have
for
our
tool
chain
and
just
kind
of
helps
expose
them.
A
lot
of
the
motivation
for
this
is
the
fact
that
there's
an
overwhelming
sense
from
people
that,
in
order
to
use
this
tool
chain,
you
need
to
use
webpack,
but
you
don't
have
to
so.
B
C
B
This
is
where
the
changelog
is
pretty
terse
right
now
for
what
for
resin
binding,
but
it
was
pretty
recent,
but
I
think
maybe
I'll
go
back
and
edit
in
some
more
wording
around
that,
because
it's
it
didn't
have,
as
we
haven't
had
really
done
like
a
big
splash
or
like
we
didn't
have
a
lot
of
messaging
around.
That's
those
are
pretty
quiet
really.
That
was
a
pretty
major
feature.
C
Well,
as
we
trade
off
doing
releases
of
each
other's
tools,
maybe
I
can
give
some
changelogs
love
to
show
as
behind
you
yeah.
So
that's
the
update
for
me
and
yeah
I'll
make
that
PR
to
the
website
and
then
I'll
share
it
out
in
all
the
channels.
Cuz,
you
know
marketing
and
design,
there's
always
lots
of
feedback.
So
please
please
give
us
that
feedback
that'll
be
great.
D
Would
you
like
to
ask
yeah
I
just
updated
the
link
and
it
seems
like
a
pace
at
the
wrong
one,
so
nobody
was
able
to
look
at
this
beforehand.
So
I
had
this
entire
experiment
awhile
ago,
where
I
was
trying
to
use
oxy
PNG,
which
is
a
rust
part
of
an
old
si
tool,
called
up
to
PNG
that
optimizes
PNG
image
is
actually
great.
An
oxy
PNG
is
like
better
code
quality
and
uses.
Threading
and
I
wanted
to
use
it
in
wasm
for
the
squashed
project
and
I'm
working
on,
and
it
turned
out.
D
I
stumbled
over
a
lot
of
things
where
I
think
we
I
want
to
at
least
hear
out.
If
you
have
thought
about
these
problems
at
all.
So
the
first
one
is
that,
as
far
as
I
know,
that
madam
Siskel
is
currently
considered
not
standardized
and
like
somewhat
of
an
under
the
hood
feature,
I'd
love
to
see
that
get
standard,
standardized,
I
think
it's
extremely
useful
and
I
realize
it's
a
bit
weird
in
terms
of
the
platform
trouble
like
wasn't
ready
to
unknown
unknown.
D
That
probably
would
have
to
introduce
something
new
for
that,
but
it's
also
proven.
Incredibly
useful
to
just
be
able
to
use
rust
libraries
in
wasum
lands
Sally.
The
first
thing
I
would
be
interesting
to
me
and
the
other
thing
that
I
was
looking
into
is
that
it
causes
likely
uses
threads,
which
we
don't
have
in
webassembly
yet
or
rust
assembly.
D
Basically
and
I
was
wondering
if
there
is
interest
or
any
thought
surrounds,
enabling
a
mechanism
in
rust
to
like
shadow,
a
standard
lip
module
with
something,
as
provided
by
the
current
crate
and
yeah,
just
curious
about
these
thoughts
and
if
there's
anything
being
worked
on
where
I
could
weigh
in
or
help
out
with
something
on
the
size.
I.
A
A
There's
going
to
be
like
proper
sis
calls
and
everything
after
that,
but,
like
my
understanding
of
Westminster's
call,
is
that,
like
one
person
just
did
it
as
a
proof
of
concept
and
it
because
it
was
useful
for
them
and
like
it's,
not
something
that
we
want
to
make
like
a
foundation
on
top
of
or
even
like
deal
with
or
maintain,
I.
Think
it's
more
about
the
history
of
that
than
idea.
The.
B
Pr
and
it
was
never
intended
to
become
stable
just
for
a
proof-of-concept
to
see
how
it
could
be,
could
could
work,
and
so
what
we,
the
the
thinking
here
is
we
want
to
enable
everyone
else
to
do
this.
So,
like
you
can
like
it's.
Incredibly,
it's
not
nearly
the
same
user
experience,
but
you
can
always
make
your
own
standard
library
on
Chrissa.
Its
own
has,
like
its
own.
Citizens,
calls
its
own
stuff,
but
obviously
I
mean
this
is
kinda.
What
resume
engine
is,
but
like
it's
a
pretty
big
investment.
B
It's
like
it's
not
nearly
as
nice.
If
it
is
the
center
library
business,
so
general
ideas
have
experiment
in
there
and
so
coming
into
the
standard
library
itself
and
becoming
part
of
a
standard
target
would
involve
standardization
far
beyond
just
us
like
as
the
rest
in,
but
simply
working
group,
but
like
probably
the
community
group
I
in
or
like
buy-in
from
like
other
languages
like
C
and
C++
or
like
go
and
blaze
over
and
all
those
stuff
and
so
like
having
a
agreed
upon
set
of
sis
calls
or
an
agreed-upon
set
of
importance.
B
But
kind
of
people
would
be
needed
for
that,
but
in
the
meantime
the
use
case
for
taking
existing
software
and
just
running
it
as
is,
is
not
really
so
much
with
the
unknown
unknown
target
is
intended
for
its
that's
actually
much
more
in
the
wheelhouse
of
the
unknown
description
target,
because
descripted
is
all
about
taking
your
existing
application,
compiling
it
to
the
web
and
running
it,
as
is
with
a
few
minor
tweaks
as
necessary,
and
so,
if
that's,
the
goal
was
to
take
all
these
existing
functionality.
Scripted
target
is
definite
the
way
to
go.
D
Gotcha
I
took
me,
honest,
I,
think
I
mean
sargo
and
things
were
already
making
that
part
quite
easy,
because
I
just
had
to
recompile
with
feature
enables
if
I
wanted
to
go
the
route
and
use
an
on-set
feature,
it
was
actually
quite
easy
to
do
so,
say:
I
guess
I
am
I,
am
more
interested
in
the
shadowing
of
sine
of
lip
modules
in
a
crate.
If
that's
been
thought
about
in
the
past,
I.
F
A
D
Absolutely
no!
This
is
not
a
generic
solution.
This
was
tailored
towards
the
program.
I
was
working
fine.
So
that's
why
I
think
like
this
would
be
something
that
the
crate
would
provide
itself,
because
only
the
crate
would
know
what
is
a
feasible
shim
in
that
context
right.
So
it's
not
something
really
like
here's,
a
generic
polyfill
for
threads
and
wasm
right
that
that
would
probably
almost
impossible
to
do,
but
sometimes
for
simplicity.
It
could
be
nice
if
you
just
like
you
know
what
the
Senate
Library
is
doing
too
much
or
is
relying
on
pictures.
F
Kind
of
feels
to
me,
like
that's
sort
of
the
domain,
that
futures
is
currently
targeting,
but
it's
like
here's
this
one
unit
of
computation,
and
then
you
have
an
executor
that
sort
of
says,
like
oh
we're,
single
threaded,
we're
multi-threaded.
This
is
how
we
like
run
them
over
the
underlying
system,
so
hopefully,
as
future,
at
least
this
is
how
I
view
it
as
futures
become
more
widespread
or
standardized.
F
D
B
D
Cool,
that's
been
very
insightful,
that'll
be
my
first
office.
Our
topic
I
want
to
block
anyone
else
when
I
tell
something
but
I
have
another
one
I
think
go
for
it.
Alright,
it's
in
a
similar
realm.
It's
the
from
wasm
ABI
trait
that
exists
in
resin,
fine,
Jen,
I
think
and
I've
recently
looked
into
that
a
little
bit
more
I.
D
Don't
actually
know
why,
but
it
seems
as
far
as
I
know
that
I'm
not
able
to
implement
myself
as
a
great
author,
currently
I,
don't
remember
what
the
error
was,
but
for
some
reason
I
ended
up
yeah
at
a
dead
end.
I
was
worrying
if
this
will
change,
because
it's
very
useful
to
be
able
to
define
myself
how
values
go
from
JavaScript
to
my
rust
structures
and
vice
versa,
generically.
B
So
right
now
you
you
technically
can
have
an
external
implementation,
it's
kind
of
really
wonky
to
do
so,
and
it's
it's
got
a
terrible
interface.
It's
not
intended
to
be
used
at
all
that
anyone,
it's
only
only
for
like
kind
of
internals
of
the
macro,
buy
some
budget
macro
right
now,
but
I
do
suspect
we're
probably
going
to
want
to
stabilize
this
over
time.
One
of
the
gotchas
is
that
you
you,
you
really
do
want
like
fine-grained
control
in
some
circumstances,
so
for
how
exactly
your
data
is
marshal
between
javascript
and
web
assembly.
B
But
one
of
the
problems
is
that
you're,
fundamentally
limited
by
what
the
CLI
itself
does.
So,
like
the
Wesen
mind
you,
the
westin
bhaijan
program
mogul,
would
actually
emit
all
the
JavaScript,
and
that
has
some
certain
set
of
function,
which
is
entirely
not
extensible
and
so
trying
to
figure
out
how
to
bridge
that
gap
or
like
you
do
want
to
customize
it,
but
you
still
have
to
always
customize
it
within
the
context
of
what.
C
B
In
minority
news
like
this,
is
you
can't
have
like
an
arbitrator
like
that
macro
which,
like
plasm,
binds
Milotic
when
it's
its
own
timing,
like
dynamically
run
code,
that's
like
if
we
wanted
to
kind
of
officially
stabilize
the
from
as
a
baby
may
tree,
although
it
it
is
currently
technically
stable,
it's
got
a
lot
of
documentation.
Saying
please
don't
use
this
because
it
probably
will
change.
We
would
probably
try
and
fix
that.
B
D
Right
I
would
go
back
to
it
and
see
because
if
that
is
acting
possible,
then
I
must
have
done
something
wrong.
I'll,
take
a
look
and
see
if
you
want
y'all
just
write
on
my
feedback
and
then
we
can.
If
we
want
to
fill
that
in
to
stabilizing
that
you
said
that
some
information
or
not
all
these,
that's
you
yeah.
C
Alex
I
was
gonna,
say
I,
get
this
question
kind
of
a
lot
and
so
I
wonder
if
it
might
not
be
useful
to
have
an
issue
where
we
kind
of
do
the
brain
dump
of
the
tricky
bits
about
it.
Kind
of
like
what
you
said
and
like
to
the
extent
that
if
there
are
people
out
there
who
would
be
interested
in
championing
it
like
just
kind
of
like
making
it
an
issue
as
something
we
can
like
point
to
folks
yeah.
They
can
say
because
this
this
is
one
of
the
questions.
B
B
A
C
H
C
C
So
the
question
which
I
am
pretty
sure
will
require
an
RFC
and
we
probably
won't
solve,
but
it's
worth
introducing
today
is,
would
we
be
interested
in
supporting
like
proprietary
targets
in
the
tool,
train
and
I?
Don't
even
know
entirely
how
I
feel
about
that,
but
it
certainly
would
be
easier
to
use
the
tool
China
on
cloud
floor
workers.
C
If
we
didn't
have
to
manually
edit
the
JavaScript
or
tell
the
computer
to
manually
edit
it,
so
that's
like
kind
of
an
open
question,
I
don't
know
if
people
have
like
in
it
feels
or
thoughts
about
that
I'm
also
trying
to
move
closer
workers
to
use
es6
modules,
which
would
be
much
nicer
than
what
they're
doing
right
now.
But
yeah
I,
don't
know
what
notes
think.
G
C
F
C
Sure
I
mean
speaking
of
getting
tied
on
this
I
also
think
it
opens
up
a
new
target
for
Glu.
If
you
wanted
to
do
all
of
that,
but
just
do
it
in
where
the
assembly
is
and
rust
and
webassembly
instead.
But
again,
I
from
you
is
mostly
like,
like
I
know
that,
like
rust,
will
compile
to
Windows
right
proprietary
target,
but
I
didn't
know
if
it
would
be
something
that
people
would
go.
You
know
immediately
to.
B
C
C
B
The
sense
that,
like
so
having
having
a
new
target,
is
kind
of
just
it's
it's
unfortunate
in
the
sense
of
it's
just
something
else
to
think
about,
like
users
have
to
think
about
it
or
like
configuration
wise
and
like
like,
if
we
could
have
one
rust
target,
that'd
be
way
better
than
having
like
300
that
we
have
right
now.
Obviously,
obviously,
it's
completely
impractical,
but
there
is
there's
benefit
to
get
everyone
on
the
same
path.
B
C
I
mean
I'm
completely
of
the
same
mind
like
I
want.
This
is
why
I'm
like
trying
to
encourage
closer
workers
to
instead
of
adding
the
wasm
as
a
global
resource,
to
use
the
modules
instead,
I
think
that
will
just
like
work
a
lot
better
and
I
think
that's
definitely
something
that's
desired,
but
I
also,
don't
think
will
happen.
C
You
know
immediately
so
like
in
the
meantime,
I
think
that
that
could
be
very
cool
and
people
would
dig
it
and
like
there
is
one
of
the
reasons
using
webpack
is
nice
is
like
for
workers
of
any
type.
There
is
some
kind
of
just
boilerplate
just
for
the
worker,
and
so
maybe
generating
that
too
would
be
kind
of
nice,
because
that
will
always
be
the
same.
C
A
C
I
think
the
reason
I
was
asking
and
I'm
like
confused
about
the
answer
was
because
I
thought
the
argument
would
be
about
supporting
proprietary
company,
but
it's
more
about
just
like
how
we
do
it
and
design
work
so
doing
the
design
work.
Rfc
sounds
great,
but
it
does
not
sound.
Like
the
question
of
do
we
want
to
support
that
target
is
terribly
controversial.
C
Yeah,
it
just
turns
out
that
I
mean
well.
If
you
want
to
do
it
without
webpack
are
the
v8
isolates
are
not
quite
a
browser
right
and
they're
CloudFlare
modified,
v8
isolates,
and
so,
if
we
want
to
try
and
support
this
saans
web
pack,
which
I
agree,
most
people
use
by
pack
and
that's
gonna,
be
awesome,
and
that
will
work
great.
But,
as
some
of
the
orange
website
showed
yesterday,
there
are
people
who
go
to
great
lengths
to
demonstrate
that
you
couldn't
try
and
use
a
thing
with
our
web
pack.
C
C
C
I
thought
about
that,
and
that
was
gonna
be
my
follow-up.
If
people
were
particularly
grumpy
about
trying
to
bake
in
a
proprietary
target
into
our
tool
chain-
and
that
would
be
to
like
make
wise
and
pack
or
wasn't
binder-
and
you
know
pluggable
in
a
way
that
you
could
like
run
it
with
the
you
know,
CloudFlare
plugin,
and
it
would
do
that.
But
if
people
like
I
don't
maybe
we
should
I'm
gonna
write
this
up,
I
think
I
think
this
sounds
like
clearly,
the
questions
been
answered
in
RFC
should
happen.
F
A
We've
just
been
saying
we're
waiting
for
stable,
okay,
yeah,
it's
no
philosophical
thing
or
anything
like
that.
Just
you
know
we
kind
of
fought
tooth
and
nail
to
get
our
tool
chain
and
everything
working
on
stable,
rust
and
don't
really
want
to
undo
that
unless
you
know,
there's
a
really
really
good
reason
like
I
think
that
yeah,
that's
it
that.
A
B
There's
two
things
here:
I
just
wanted
to
clarify:
do
you
mean
so
we
have
the
Y's
imagine
features
Crete,
which
is
kind
of
how
to
run
futures
on
the
web
booth,
and
we
also
have
crates
that
they
themselves
give
futures.
So
are
you
curious
about
each
of
those?
Were
you
wondering
about
the
features
about
three
or
both
I.
B
F
F
There's
sort
of
one
open
issue
right
now,
which
is
whether
or
not
the
Waker
callback
is
sufficient
or
whether
it
should
be
just
Waker
as
an
argument
like
the
call
back
to
wake
it
back
up
or
context
which
includes
the
Waker
there's.
Some
heated
debate
going
on,
but
I
believe,
ladies
and
Liang
are
committed
to
resolving
that
in
the
near
term.