►
From YouTube: Rust and WebAssembly Working Group Meeting 2019-04-04
Description
A
A
A
There's
like
a
nice
little
graph
is
visualization
of
how
things
work,
but
yeah.
So
take
a
look
at
that.
The
other
question
that,
if
you
care
about
this
I,
don't
want
to
like
maybe
debate
it
right
now.
But
you
know
I'll
point
you
to
thread
is:
there's
a
question
of
how
to
version
like
the
umbrella
glue
crate
compared
to
each
of
the
toolkit
crates
that
it
reacts
ports
and
whether
they
should
be
all
together
in
lockstep,
which
would
kind
of
be
the
easiest
thing
from
a
maintenance
perspective.
A
But
it
means
that
like
if
any
crate
has
breaking
rope
chain,
then
like
we
arbitrarily
have
to
breaking
change
everything
else,
and
so
there's
a
few
questions
there.
And
if
you
have
opinions
about
that,
please
comment
in
the
thread
and
then
after
that,
there's
like
been
a
bunch
of
new
proposals
or
movement
on
various
proposals
and
I've
listed
I.
Think
most
of
them
here.
A
A
Wanted
to
see
like
how
LLVM
like
what
their
API
is
for
people
who
are
writing
optimization
passes
are
to
maintain
debug
info
and
kind
of
like
what
their
philosophy
is.
It's
it's
just
like
kind
of
prior
research
yeah.
They
have
some
nice,
like,
like
properties
like
if
you
just
like,
take
in
some
IR
and
then
spit
the
IR
back
out,
like
you
always
get
the
perfect
info.
A
If
you
like,
reorder
stuff,
you,
the
perfect
abelian
fo
like
without
having
to
do
any
work
on
your
own
and
then
it's
only
like
if
you're
actually
making
changes
do
you
have
to
like
like.
If
you
make
some
changes,
then
everything
you
didn't
change
still
has
bug
info
and
then,
if
you
did
make
some
change
to
something
like
you
can
choose
to
also
add
to
the
info
or
say
use
the
bug.
Infer
from
there.
A
B
So
after
I
feel
like
a
lot
of
inactivity,
I
think
things
have
been
kicking
right
back
up,
which
is
awesome.
We
sent
out
ODOT
800,
which
is
cool,
I,
wasn't
a
huge
release,
but
because
we
had
a
feature
which
was
the
renaming
of
the
generated
names
that
landed
in
wasn't
by
Jen,
we
went
to
the
minor
bump
I
we
had
a
small.
It
got
a
little
delayed
because
of
some
shenanigans
in
md
book
and
relative
links.
B
B
B
We're
gonna
do
an
8.1
release
because,
with
the
non
brushed
up
environment
changes,
it
ended
up
breaking
folks
on
docker,
and
so
Drager
already
has
a
fix
for
that
in
and
another
cool
announcement
is
that
trigger
is
now
Co.
Maintainer
of
wasn't
pack
he's
been
doing
so
much
awesome
work.
It
felt
like
it
had
to
happen
and
super
great,
and
so
one
of
my
biggest
to
do
is
has
been
to
get
that
release
checklist
out
and
so
I'm
gonna
be
working
with
him
on
this
Oda.
B
It
felt
one
release
so
that
he
can
do
a
whole
bunch
of
that
and
we
can
write
down
the
steps
so
we're
moving
that
forward
and
so
Oh
dot.
Nine
is
going
to
be
our
sprint
release
and
we
have
a
fair
amount
of
work.
To
do.
I've
been
doing
a
lot
of
the
background
work
for
the
cargo
generate
stuff.
It
turns
out
that
the
cargo
generate
binaries
I
was
making
was
actually
suffering
from
an
issue.
Our
tool
chain
ran
into
long
ago
about
not
having
like
exe
on
the
windows.
B
Binary
I,
don't
know
if
you
remember
those
issues,
but
that's
getting
fixed
and
I'm.
Releasing
a
new
version
of
cargo
generate
today
and
then
I
should
be
able
to
get
my
branch
to
put
cargo
generate
into
wise
and
pack
up.
Then
we
have
the
NPM
depth
thing,
which,
thanks
to
Alex
filing
a
really
awesome
issue,
I
think
that
should
be
actually
pretty
quick
to
get
out.
The
final
thing
that
at
least
I
have
on
my
plate.
Is
this
thing
that
we've
been
holding
onto
for
a
while,
which
is
this
isomorphic
target?
B
If
you
aren't
familiar
with
this,
it's
people
have
been
asking
to
be
able
to
create
an
NPM
package
that
would
work
both
like
you
know,
as
an
es
6
module
as
that
an
es6
module
and
on
node,
and
just
like,
we
have
all
of
those
options
but
having
wise
impact
create
a
single
package
that
has
all
those
things
in
it.
As
far
as
hitting
the
o
dot
9
deadline,
I'm
worried
that
one
might
slip
with
it
being
on
my
plate.
B
Then
the
next
two
things
they
don't
really
have
owners
right
now
is
was
mopped
integration
and
creating
an
NPM
package
to
NPM
install,
wasn't
PAC
that
last
one
in
particular
like
if
we
have
JavaScript
folks,
like
that's,
probably
like
a
great
like
get
started,
contributing
to
wise
and
PAC
one.
It's
not
really
about
wise
impact,
it's
more
about
creating
an
NPM
package,
but
that's
where
we're
at
with
stuff
I,
don't
know.
If
folks
have
questions
or
comments
see
for.
C
Expand
of
it
we're
we're,
trying
we're
we're,
trying
really
hard
to
get
the
support
and
upstream,
to
use
their
binaries
as
opposed
to
having
a
local
work
ourselves,
I'm
working
off
of
that
branch
and
so
I'm
working
through
that.
There's,
it's
it's
in
progress
and
it's
moving
slowly,
but
it's
it's
getting
cool
I
will
comment
on
the
issue
about
was
in
pack
for
was
in
pack
using
azimoff's
when
it's
ready
to
go
and
that'll
have
information
about
the
releases
and
what
URLs
to
use
and
all
that
good
stuff
cool.
B
I
was
gonna
say
for
anybody
who
wanted
who
would
want
to
start
working
on.
This
was
mopped
integration.
You
could
start
a
PR
with
the
assumption
that
was
a
mopped
is
like
in
your
path
and
then
the
whole
like
install
was
mop.
Spit
from
binary
can
be
done
a
little
bit
later,
but
there's
like
definitely
some
stuff
with,
like
configuration
and
command-line
arguments
that
we
can
do
and
like
we
can
just
kind
of
assume
it
will
be
there
and
then
do
that
at
the
end
would
be
my
my
strategy.
B
A
B
I
mean
the
distance
between
just
because
I've
recently
done
this
with
cargo
generate
and
a
bunch
of
other
things,
I'm
also
I
have
a
tool
that
now
uses
plasm
pack
is
one
of
these
things,
writing
it
as
if
you
can
assume
it
in
path
and
then
switching
it's
used,
binary
install
is
actually
not
really
a
difficult
developer
experience.
It's
pretty
good,
so
kudos
to
all
the
binary
install
to
tell
the
purpose
that
worked
out
well,
a.
D
B
So
one
I
think
that's
a
cool
idea
to
spend
who
is
not
on
this
call
because
I
think
he's
at
his
orientation
for
CloudFlare
right
now.
Sven
has
a
namespace
I
forget
exactly
what
it
is,
but
it's
like
vaguely
webassembly,
rust,
related
I
think
he's
been
kind
of
building
everything
out
that
he's
been
working
on
there.
So
we
may
need
to
reconcile
that
but
I
think
it's
a
reasonable
idea.
I
don't
know
if
we've
squatted,
rust,
wasm
or
anything
yet
are.
D
B
B
I
guess
just
along
those
lines,
I
think
we've
started
doing.
Some
of
this
I
was
poking
around
at
the
governance
repo
a
couple
days
ago,
because
I
wanted
to
you
know,
set
up
the
way.
I
was
impacting
and
make
sure
dragger
was
on,
like
maintainer
team,
and
we
probably
will
want
slightly
more
teams
defined
in.
B
The
one
thing
I
have
is
like
so
I
am
of
the
things
that
are
on
the
plate
for
Oda
9.0.
That
I'm
worried
will
slip.
Is
that
isomorphic
packages
thing
and
I
can
imagine
that
that's
that
could
be
multiple
PRS
or
it
could
be
one
big
one.
Is
there
anyone
here
who
would
be
interested
in
taking
a
stab
at
that?
It's
gonna
involve
a
lot
of
like
understanding
modules
in
Jas
kind
of
situation
and
like
build
steps.
B
B
Mean
being
able
to
merge,
package.json
is
going
to
be
a
thing
that
comes
with
the
MPM
depths
thing
and
as
far
as
steps
go,
I
definitely
think
that
NPM
dependency
bit
is
going
to.
You
know,
fill
Nestle
complicated,
but
it
seemed
at
least
in
my
head
to
be
a
thing
to
do
before
doing.
This
is
isomorphic
packages
thing,
because
now
we're
modifying
package.json
a
whole
ton
and
like
how
we're
gonna
deal
with
those
dependencies
for
each
of
these
different
ones
might
be
different.
B
I
think
the
original
iterative
step
that
we
had
thought
of
for
that
isomorphic
one
was
that
we
would
generate
three
different
packages
like
have
three
different
directories,
or
that
package
directory
would
have
three
directories
inside
of
it.
That
has
the
three
different
ones,
I'm,
not
sure,
I
love
that
but
I,
don't
know.
If
we're
okay
with
it
slipping,
it
would
be
the
top
priority
for
Oh
10,
but
maybe
maybe
the
the
first
two
shouldn't
be
too
difficult
and
that
we
still
have
three
weeks
I.
A
Alright,
so
I
had
one
thing:
I
wanted
to
just
sound
off
against
everyone
with
regards
to
the
Azzam
weight,
tracker
and
iteration
with
glue
I
think
it
would
be
nice
for
the
glue
project
to
be
able
to
measure
like
every
individual
utility,
crate
kind
of
like
what
it's
weight.
Impact
is
just
to
keep
an
eye
on
it
for
ourselves,
but
I
think
this
would
quickly
add
up
to
a
lot
of
test
cases
for
the
weight,
tracker
and
I.
Don't
know
how
people
feel
about
that.
B
C
A
B
B
C
Related
so
much
to
the
jas
ecosystem
that
we're
working
in
right
now,
but
if
you're
curious
about
rezian
rust,
we
I
don't
know
well,
I,
guess
one
thing
I
wanted
to
ask
as
well
as
so.
This
y'see
target
came
out
of
this
concept
of
Y
Z,
which
is
a
system
interface
or
web
assemblies.
A
portable
set
of
API
has
to
do
with
the
file
system
and
all
that
good
business
did.
C
We
want
to
consider
this
being
underneath
under
the
working
group
like
do
we
want
to
have
discussions
related
to
SEO
and
like
the
team
repo
and
maybe
have
repos
around
that,
or
do
we
still
want
to
do?
We
prefer
to
relegate
this
more
so
to
some
other
body.
I,
don't
know
what
but
someone
else.
In
terms
of
like
most
of
what
we've
been
doing
right
now
and
what
Ruston
webassembly,
then
the
working
group
has
been
largely
JavaScript,
related
and
kind
of
the
web
related
aspect
and
not
so
much
the
embedded
everywhere
else
kind
of
aspect.
A
Yeah
so
I
think
different
parts
of
it
make
sense
for
us
to
deal
with
like,
like
I,
see
a
future.
It's
not
exactly
clear
like
what
the
details
are.
The
path
to
get
there
is
but
I
see
a
future
where,
like
our
web,
targeting
wasm
is
using
the
Y
Z
target
because
it
gets
you
a
lot
of
things.
Kind
of
by
default,
like
printing
the
standard
out
or
whatever,
and
so
like
in
that
sense
like
it
is
very
relevant
to
even
the
use
cases
that
we've
been
focusing
on
I.
A
C
That's
a
good
point
when,
in
the
sense
that
I
don't
think
we
will
need
Y
Z
as
a
target
may
not
require
as
much
of
an
ecosystem
around
it,
as
well
as
the
unknown
unknown
target,
because
we
need
tooling
requires
in
mind.
You
know
mass
spec,
but
all
that
is
not
really
necessary
on
the
Y
Z
target,
depending
on
what
you're
doing
and
then
but
yes,
I
have
as
of
right
now,
I
don't
really
actually
have
any
open
questions
about
the
wet
target
of
I.
C
B
I
mean
I
do
definitely
think
that
there's
gonna
be
like
I'm,
not
sure
if
I
would
necessarily
want
there
to
be.
Another
group
spoke
up
because
I
feel
like
that,
would
end
up.
We
have
this
thing
and
Russ
where
the
groups
kind
of
become
silos
and
cross
team
communications.
Rough
though
I
know
what
working
on
it,
but
I
could
imagine
that
there
be
at
least
a
fair
amount
of
overlap,
at
least
as
far
as
people
interested
and
then
like
skills
and
resi
has
the
polyfill
for
the
browser
and
I
don't
know.
B
D
B
A
A
A
B
A
C
B
C
Okay,
well
then,
along
those
lines,
one
thing
I
would
status
update
for
Wednesday
is
that
we
do
have
a
132,
unknown
Y
Z
target
on
nightly.
Now,
if
you
restaurant
a,
we
can
add
that
target.
You
should
be
good
to
go
and
it's
not
currently
set
up.
It
may
or
may
not
work
with
us.
Imagine
it
wasn't
packed
I
kind
of
doubted
that
us,
but
it's
not
really
in
any
way.
So
if
you
want
to
play
around
with
it,
it's
there
PR
just
landed
today
to
add
some
more
actually
add
the
file
system
support.
C
E
I
was
thinking
about
it
previously,
but
before
I
do
that
I
wanted
to
ask
one
question
about
the
lousy
stuff,
and
maybe
this
is
actually
central
to
the
conversation
that
it's
been
taking
place
right
now,
but
how
much
overlap
are
we
expecting
there
to
be
between
the
bind,
JNT
ecosystem
and
huazi,
because
it
sounds
like
huazi
is
kind
of
a
more
standardized
approach
of
saying
here
are
the
api
it's
that
are
available
by
the
system.
That's
providing
the
interface
versus
bind.
E
C
Would
say
that,
for
the
near
term,
we
probably
won't
see
a
ton
of
overlap,
and
this
is
where
the
main
use
cases
for
y'see
right
now,
I
think
is
basically
fastly
and
then
local
wagon
time.
Development
are
kind
of
like
the
portable
binary
aspect,
and
so
that
use
case
hasn't
really
developed
to
the
point
where
you
want
to
have
a
lot
of
cross-pollination
between
the
web
and
YC
and
so
but
I
suspect
that
will
develop
over
time
and
then
also
a
fundamental
limitation
of
ossie.
C
Is
that
the
way
that
it's
architected
it
does
require
a
polyfill
to
run
on
the
web?
And
so,
if
your,
if
your
goal
is
the
absolute
smallest
piece
of
web
Assemblyman,
J
has
to
run
on
your
application,
whereas
he's
probably
not
what
you
want,
because
that
polyfill
will
inherently
be
adding
some
cost.
Now,
if
you
can
sum
up
the
cost
that
I
think
work
great,
but
so
that's
where,
at
the
very
immediate
from
the
outset,
I
think
it.
C
It's
gonna,
be
tough
to
have
to
see
a
lot
of
overlap,
but
as
Nyx's
and
I
suspect,
as
many
browsers
will
go.
The
intention
of
y'see
in
the
long
term
is
that
everyone
kind
of
natively
implements
it.
So
node
would
natively
put
this,
so
you
wouldn't
what
is
amazing.
We
way
list
necessarily
unknown
cuz.
You
could
just
do
raw
sits,
calls
a
little
more
powerful,
even
faster
than
the
gia
surgeons,
and
maybe
the
web.
Maybe
browsers
will
eventually
have
a
native
implementation
of
Y
Z.
So
very
long
term.
We
might
see
a
lot
of
overlap.
B
Mean
I,
don't
think
I
guess
I
was
just
thinking
like
the
way
I
could
see
like
quickly
there
being
some
overlap
is
with
these
binaries.
We
were
building
out
all
this
HTML
kind
of
CSS
JavaScript
shenanigans
building
you
eyes
for
these
native
binaries
I
could
see
people
going
kind
of
like
an
electron,
II
kind
of
route,
and
so
the
UI
is
in
these
binaries
and
then
us
doing
this
web
interface.
B
Stuff
I
could
see
a
really
huge,
quick
win
if
we
could
get
that
working
sooner
than
later,
because
people
would
probably
want
to
use
that
and
building
UI
frameworks.
It
has
historically
not
been
terribly
successful
interests
so,
like
we
have
a
lot
of
stuff
that
probably
people
would
want
to
use
so
I.
A
Had
a
couple
of
things
to
say
about,
like
the
cost
of
the
polyfill
I,
think,
can
be
reduced
a
lot
by
doing
some
sort
of
like
tree
shaking
a
dead
code,
elimination,
I,
think
there's
multiple
ways.
We
could
do
this
like
we
one
way
it
could
be
like
wasm
bind
gen
could
look
at
which
has
the
imports
you're
relying
on
and
then
provide
the
GS
like
the
polyfills
on
a
as-needed
basis
for
individual
imports,
and
I
think,
like
webpack,
does
that
kind
of
stuff,
as
well
as
other
J's
bumblers,
do
too
so
I?
D
What
I
would
love
to
see
is
some
sort
of
a
future
that
I'm
envisioning
is
actually
more
in
a
wagon
pack
side
where
currently
provides
the
nice
facilities
for
testing
in
the
browser
with
the
drivers
and
all
that
stuff,
like
I
reaiiy
really
like
that
experience
there
for
like
end
users
would
be
awesome
if
we
could
provide
something
similar
for
huazi.
Now,
you
don't
really
have
to
care
like
what's
going
on
on
their
hood,
just
like
oh
cool
a
target,
mozzie
I
can
test
it.
D
B
I
definitely
think
that
with
wise
impact,
like
people
are
already
upset
that
they
can't
make
packages
to
target
all
of
the
JavaScript
things,
people
are
probably
going
to
it.
Ii
like
the
wedgie
runtimes
to
take
off
people
are
gonna,
want
to
be
able
to
leverage
that,
and
so
I
feel
like
I
mean
I,
guess,
there's
a
universe
where
we
make
another
tool,
and
it
does
the
West
the
wedgie
stuff,
but
like
I,
feel
like
I
could
see
a
future
where
it's
bit
squirrely
until
it
wasn't
that
I
was
trying
to
do
I.
C
Think
another
way
to
summarize
one
of
this
is
that,
if
wise
imagine
is
stepping
on
Y
Z's
toes
or
the
other
way
around,
that's
a
bug
we
want
to
fix,
so
we
want
these
should
always
be
interoperable
and
there
should
always
be
if
you
want
overlap,
it
works.
If
you
don't
want
overlap
pitch,
it
also
work
yeah.
C
B
I'm
not
sure
I
have
to
do
more
research,
but
would
people
be
against
like
having
a
section
of
this
meeting
where
we
like
check
in
on
some
of
the
community
group
stuff?
That
would
like
affect
us
when,
when
I
pitched
and
did
a
little
one
on
the
host
bindings
proposal,
which
has
a
new
name
now
right,
I
keep
not
remembering
it
like.
A
B
B
E
B
B
I
would
just
strongly
encourage
you
to
if
you
could
just
open
an
issue
and
like
the
easiest
thing
to
do
is
just
like
dump
what
you
think
those
interfaces
should
be,
and
then
we
could
might
kind
of
pick
it
apart
from
there
long
ago,
like
way
before
react,
even
existed,
I
worked
with
PhoneGap,
which
was
like
the
og
like
native,
but
like
actually
just
browser
stuff
on
the
phone
and
I
always
thought
that
that
kind
of
technology
was
super
empowering
and
react
native
is
at
least
popular
enough.
That
people
have
strong
opinions
about
it.
E
I'm
happy
to
put
some
stuff
together,
I
guess
I'm,
just
kind
of
wondering
where
a
good
place
would
be
to
start
some
discussion
on
that
was
PhoneGap
yeah.
That's
that's
like
the
canonical
example
and
then
react
native
uses
a
very
similar
mechanism
for
being
able
to
run
the
bundled
code
on
you
know,
Android,
iOS
or
I.
Think
Windows
is
supported
as
well,
not
sure
discussion,
I,
think.
B
Starting
it
in
wasm
pack
makes
the
most
sense
and
then,
depending
from
there
like,
we
can
poke
at
it
and
like
there
may
be
some
stuff
that
we
think
that
actually
should
be
wise
and
bindings
job.
And
then
we
can
like
drop
those
down
to
where's
my
dread.
I,
don't
know
Alex
what'd,
you
think
of
that
again.
I,
don't
entirely
know.
What's
up
I.
A
Have
questions
about
like
I
guess
what
we're
envisioning
here
we
are.
We
thinking
like
how
to
use
wasm
with
react
native
so
that
you
can
kind
of
use
this
everywhere
react
native,
is
use
or
are
you
thinking
like
I,
have
a
wasm
module
with
a
set
of
imports
and
I
implement
those
imports
on
iOS
and
I?
Those
imports
on
Android
and
I
implement
those
imports
on
the
web
to
provide
like
the
same
UI
framework,
yeah.
E
We
obviously
a
part
of
that,
but
then
being
able
to
basically
take
the
same
route
of
any
JavaScript
and
web
assembly
that
we
have
would
run
in
the
browser
on
one
of
these
devices,
but
packaged
as
a
native
that,
which
is
the
approach
that
react
native
in
PhoneGap
and
such
takes.
So
it's
kind
of
our
way
of
saying
hey.
This
is
essentially
the
same
concept,
but
it
would
provide
us
a
way
to
tie
in
to
the
lower
level.
E
Api
is
that
the
devices
provide
and
that's
where
something
like
wise
impact
would
be
beautiful,
because
we
could
just
generate
those
bindings
for
Java
through
the
J&I
or
swift
through.
You
know
a
standard,
C
interface,
so
so
yet
to
answer
your
question
directly.
Nick,
basically
outside
of
the
react
native
context
specifically
are.
E
B
My
understanding
is
that
these
are
kind
of
exactly
like
PhoneGap,
which
is
like
it
would.
It's
like
the
act
native
is
like
react,
pretending
to
be
native,
and
so
that's
why
I'm
kind
of
thinking
that,
like
this
like
what
we
could
end
up
with,
is
like
glue
native
and
it's
largely
a
packaging
issue
and
there's
there
are
some
other
things
and
those
are
the
things
that
I'm
like.
Please
write
those
in
the
issue.
Cuz
I
need
to
figure
out
what
those
things
are,
but
I
think
it's
largely
packaging
yeah,
a
gray.
B
D
So
what
I
was
always
like
curious
about,
like
we
have
waz
impact
and
wasm
bind
gen
to
target
wasum
specifically,
but
I
can
sort
of
see
like
a
parallel
tool
chain
targeting
like
iOS
and
targeting
like
Java,
but
providing
very
similar
semantics.
So
I'm
kind
of
curious,
like
instead
of
like
focusing
on
doing
this
through
wasm
and
emulating
what
react
native
does,
which
is
constrained
by
JavaScript,
who
can't
do
that
if,
instead
building
a
parallel
tool
chain,
might
be
like
a
cleaner
solution?
Maybe
yes,.
B
So
hey,
some
clouds
were
just
released,
a
rust
based,
iOS
app,
which
or
I
guess
it's
in
like
beta
right
now,
but
they're
new,
like
warp
product,
is
all
written
in
the
rest
of
are
getting
high
OS,
and
it
is
extremely
true
that
a
nice
developer
experience
for
that
does
not
currently
exist
in
rust.
At
least
they
didn't
find
it
and
I'm
not
aware
of
it.
B
You
have
sometimes
that,
like
actual
native
situations,
that's
not
possible,
and
my
understanding
is
that
the
UI
frameworks
in
rust
are
not
very
good
and
I.
This
is
not
to
denigrate
anyone.
I
only
know
this,
because
people
complain
about
them
to
me
all
the
time.
I
don't
have
any
actual
first-hand
experience,
but
people
are
like
the
UI
frameworks
I'm
like
I'm
sorry.
B
So
if
we're
like
building
out
this
is
kind
of
coming
back
to
that
whole
y'see
thing
it's
just
like
if
we
can
like
do
HTML
and
that
kind
of
stuff,
and
do
it
well
in
the
rest
webassembly
world,
then
that
gives
this
like
interesting
cheat
to
like
be
able
to
do
UI
with
developers
who
understand
this.
Like
one
specific
thing,
you
are
right
that
it
is
like
a
little
bit
bonkers
that
people
would
choose
to
do
this,
but
this
dream
of,
like
I,
write
web
apps,
but
I
can
also
write.