►
From YouTube: 6 September 2018 Meeting
Description
The Rust WebAssembly Working Group meeting from 6 September 2018.
B
A
So
this
has
been
kind
of
a
contentious
topic
because
there's
a
lot
of
guiding
values
that
present
internal
inconsistencies
and
conflicts,
and
so
the
RFC
kind
of
talks
about
two
parts.
One
is
like
how
assuming
I
have
an
NPM
package?
Can
I
like
talk
about
it
like
import
or
require
it
into
my
rest
and
that
that
syntax,
I
think,
is
pretty
uncontroversial.
A
B
Okay,
let's
look
at
once
again
the
blocking
release
candidate
list
so
number
one
thing
on
that
list
is
determine
how
to
express
jeaious
dependencies
and
rest
crates,
and
now
we
have
an
RFC.
So
that
is
very
awesome
to
see.
In
fact,
maybe
we
can
just
point
people
at
that
RFC
and
close
this
issue.
Yeah.
B
A
B
D
A
E
B
B
D
D
D
B
E
E
We
want
to
make
sure
there
is
some
story
for
the
callback
types
in
my
videos,
so
you
can
set
the
mental
stirs
and
all
that
good
stuff
and
we're
making
good
progress
on
that
we're
just
supposed
to
appear
to
kind
of
enable
us
actually
do
the
most
of
it
and
once
that
lands
it
should
be
pretty
easy
actually
to
edit
the
web
system.
I.
Think,
though,
that
that's
pretty
much
everything
yeah.
B
Yes,
so
the
check
the
checkbox
for
the
feature
gates
so
for
the
people
who
aren't
familiar
with
the
feature
gate
thing.
This
is
basically,
if
you
want
to
use
some
interface
in
web
sis,
there's
a
feature
for
every
single
interface
and
so
by
default
we
won't
really
do
any
code
generation,
and
then
you
enable
the
things
that
you
want,
and
this
brings
compile
times
down
from
what
were
they
like
20
seconds
to?
If
you
only
do
like
one,
you
know
a
handful
of
seconds.
E
B
E
B
B
B
B
A
A
E
In
terms
of
timing,
so
this
releases
on
the
13th
it's
coming
up
next
week
is
the
rc1
I
think
of
the
2010
release.
But
most
of
all,
none
of
us
am
biogenic
composite
unstable,
yet
October
25th
is
when
all
the
features
will
ship
to
stables.
We
will
actually
compile
and
stable
at
that
point,
and
so
we
have
between
October,
25th
and
December
6,
I
think
so.
I
kind
of
tested
out
our
whole
experience
on
stable
wrists
with
wisdom
and
everything,
and
so
like
that
battle-ax.
That,
I
think,
is
the
best
time
for
making
sure
like.
E
We
have
a
really
solid
experience
like
almost
ready
to
go
and
like
that.
That's
just
pure
polish
period,
and
so
in
terms
of
blocking
for
this
release
candidate,
it's
a
little
less
critical
because
we're
still
we're
all
like
effectively
nightly
only
because
not
really
that
many
people
are
using
beta,
so
I
think
it's
okay
to
have.
So
it's
a
lot.
A
few
things
slide,
but
October
25th
I
think.
Is
there
much
bigger
deadline
where
we
we
need
to
make
sure
that
everything
is
nice
and
solid
for
October
20,
cuz,
yeah.
B
Agreed
and
like
since
most
of
our
stuff
isn't
actually
shipping
in
see
it's
all
a
bit
like
hazy.
It's
just
to
keep
us
kind
of
on
track
and
making
sure
we're
keeping
these
things
in
our
head.
I
think
the
other
thing
I
would
mention
is
like
I
did
not
forget
about
lesson
pack,
but
in
our
2018
gold
status
updates
like
there's
a
lesson
pack
thing
so
I
figured,
we
could
just
oh
I.
B
A
A
That
data
pup,
who
is
not
here
today
is
working
on,
which
is
one
of
the
big
changes
is
we're
now,
if
possible,
down
downloading
the
pre-built
binary
of
the
cargo
commands
like
plasma
engine,
but
we
are
not
respecting
locked
files
at
the
moment
and
that's
kind
of
important.
So
before
ODOT
5
goes
out,
we
need
to
have
that
done
and
then,
in
the
meantime,
there's
a
bunch
of
other
stuff
going
on
I
have
a
very
large
Doc's
PR
that
needs
to
get
it
merged.
A
B
B
I
think
I
would
like
to
review
those
PRS
because
I
feel
like
it
doesn't
make
sense
to
really
support
azzam,
snip
or
azzam
ops
or
anything
until
we
can
like
configure
them
in
a
cargo
tamil,
because,
like
I
mean
we
have
some
ops
like
kind
of
there's
sane
defaults
for
that,
but
there's
definitely
not
sane
defaults
for
Azzam
snip,
like
if
you're
using
azzam
snip.
It's
like
you
need
to
be
very
involved
in
the
process
and
there.
B
B
B
I
would
like
to
see
like,
without
looking
at
the
actual
code
like
what
we
want.
The
configuration
to
look
like
and
then
like
make
sure
that
the
code
actually
does
that
and
then
get
into
like
you
know
the
nitty-gritty
of,
like
you
know,
is
it
idiomatic
or
whatever
right,
but
like
I,
feel
like
that
first
step
of
like
agreeing
on
what
the
configurations
shape
should
be
is
like
probably
probably
the
first
thing
we
should
do
after
cutting
the
release
sounds.
B
B
F
B
F
A
It
one
thing
that
I
do
think
is
important
for
us
to
tell
as
a
story
like
at
some
point
is
we
have
two
templates
and
the
trick
about
them
is
that
they
both
actually
use
our
workflow
and
use
webpack
at
the
moment
kind
of
so
we
should
people
are
gonna,
ask
like
which
one
should
I
use,
and
why
and
I'm
not
sure
we
have
a
good
story
for
that
right
now.
I,
don't
think
it's
urgent,
but
certainly
it
seems
like
a
story.
We
need
to
tell
yeah.
B
B
Also
help
when
we
get
like
more
templates
as
well
like
parcel
and
roll
up
and
stuff
and
then
it'll
be
I.
Think
more
clear.
Just
like
that.
There's
a
go!
There's
the
like
template
for
when
I'm
using
webpack
and
a
template
for
when
I'm,
using
roll
up
and
like
hopefully
like
more
choices,
will
make
things
more
clear.
Yeah.
A
A
like
init
template
and
that
MPM
init
template
should
probably
just
take
an
argument
as
to
what
bundler
you
want
to
use,
because,
right
now
the
create
was
an
app
template
will
just
default
to
web
pack,
and
so
my
thought
was
that
that
thing
will
to
webpack,
roll-up
or
parcel
you
know,
and
and
so
the
real
question
is
is
like.
Who
do
you
want
running
the
tool
chain?
Do
you
want
WebP
actor
in
the
tool
chain?
Or
do
you
want
like
to
run
the
tool
chain?
A
A
F
A
A
A
I
know
I,
so
I
understand
that
the
bundler
like
webpack
there's
a
Wesen
pact
plug-in
now,
yeah
I,
think
there's
a
question
of
whether
or
not
you
want
the
tool
just
tool
chain
to
actually
be
invoked
by
node
or
if
you
want
to
just
invoke
the
rust
tool
chain
and
then
like
shoot
it
off
into
space,
and
then
the
JavaScript
developers
use
a
node
environment.
So,
like
the
real
distinction
for
me
here
is,
do
I
want
to
be
doing
a
ton
of
like
do.
A
A
B
B
Some
deals
not
here
to
talk
about
the
website
and
nothing's
really
happened
anyway,
not
much
there
with
the
bugging
and
buts.
We
just
talked
about
so
I
think
we
can
move
on
from
2018
status
updates,
and
so
the
next
thing
in
the
agenda
I
added
is
this
is
something
that's
come
up
and
people
have
filed
issues
on
both
wasm
pack
and
Watson.
B
Because
we'd,
like
you
know,
module
imports
are
different.
You
know
yes,
modules
versus
like
commonjs
require,
and
it's
you
know.
Sometimes
you
can
do
that
like
UMD
thing,
where
you
like
stuff
things
on
like
a
global,
but
it's
hard
and
we
don't
really
have
a
good
solution
for
this,
and
it
seemed
like
this
is
something
that
everyone's
running
into
who's
like
having
initial
success
and
then
when
they
try
to
actually
like
get
to
like
deployment
or
whatever
right.
This
is
the
issue
that
people
run
into
so.
A
I
can
talk
a
little
bit
about
this,
because
this
is
something
that
I
brought
up
with
Alex
like
extremely
early
on,
mostly
because
of
an
issue
someone
filed,
probably
after
the
first
release
of
has
impact
where
we
supported
commonjs,
which
I
it's
just
so
amazing.
I'm
like
this
was
like
four
months
ago
we've
been
working
on
this
for
a
while.
So
when
we
were
able
to
have
waz
impact
support,
multiple
types
of
exports
like
you
can
have
it
for
commonjs
we
could
have
a
306.
A
Don't
believe
that
it
can
refer
to
two
different
things,
and
now
this
is
where
I'm
like
hazy,
because
again
this
conversation
was
like
literally
four
months
ago,
but
I
believe
that
the
fact
that
the
wesam
file
refers
back
to
the
module
wrapper
file,
isn't
load-bearing
or
there's
like
another
way
to
do
it,
and
if
we
could
change
how
that
works,
we'd
be
able
to
enable
this.
So
alex
is
unmuted
I'm,
ready.
E
Just
to
make
sure
I
understand
this,
how
does
this
actually
work
in
node?
Like
do
you
publish
like
to
compiled
versions
of
your
Jas
or,
like
you
say,
if
you
use
this
bundler,
here's
all
the
code
for
you
use
this
little
bit.
Here's
a
look
good
for
you,
so
the
idea
is.
We
want
to
publish
multiple
versions
of
wasn't
my
gen
stuff.
We're
like
this
is
the
style
that
that
style
or
like
in
that
publishable
towards
will
do
this
thing,
create
I'd,
say
for
bundler.
Sorry
go
for
it!
Yes,.
G
I
guess
I
would
look
at
packages
that
do
get
used
a
lot
of
times
in
note
and
the
web
like
moment.js,
for
example,
and
a
lot
of
times.
Then
they
do
have
like
their
original
version
and
that's
all
required.
Well,
there's
really
two
scenarios
either
that's
pure
JavaScript
and
then
it's
usually
written
as
like
a
bunch
of
command,
J
s
and
there's
imports,
and
then
they
use
something
like
browserify
to
package
up
another.
G
You
know
uglify,
you
know
small
J
s
package
that
gets
more,
but
then
you
have
the
other
scenario
where,
if
they're
using
something
like
their
typescript
or
babel,
but
a
lot
of
times,
all
their
source
files
are
using
es
m
and
then
like
webpack,
is
able
to
read
that
and
do
all
the
tree
shaking
and
stuff
off
of
that.
So
then
that
points
at
the
original
source
and
then
webpack
takes
care
of
the
Uggla
flying
and
everything
after
the
pack
and
then
they
usually
then
just
have
a
babel
version
or
a
typescript
compiled
version.
G
That
then,
is
converting
into
common
j
s
and
then
they're,
basically
just
hitting
one
or
the
other.
But
yes,
usually,
then,
when
they're
pulling
down
the
package
for
npm,
there's,
there's
usually
you
know
two
options:
whether
they're
a
single
file
or
just
two
different
directories,
but
there's
always
two
different
versions
to
support
both
command
J
s
and
then.
G
But
it
sounds
like
it
was:
it's
yeah,
there's
usually
to
that
Mike
White
was
there's
usually
to
when
NPM
pulls
it
down.
There's
this
browser
version.
This
is
the
node
version
and
sometimes
there's
even
a
third
that
this
is
the
core.
Like
here's,
my
typescript,
and
if
you
want
plasm
to
go
over
top
of
his
web
pack
to
go
tree
shake
that,
then
you
can
even
point
at
this
third
one,
which
is
the
you
know
your
you
know,
new.
G
E
So
in
that
sense,
it's
definitely
not
the
kate,
like
we
can't
publish
one
wesson
file
that
somehow
works
amongst
all
of
these,
because
you're
right
ahead.
Well,
the
wizened
file
has
to
point
to
something
and
it
well.
There
wasn't
file
has
to
point
to
something
and
it
has
to
like
it
has
to
reference
some
path,
and
someone
has
to
eventually
interpret
that,
and
it's
probably
gonna
be
like
web
pack
or
probably
web
pack.
That's
interpreting
it.
I
don't
know
this
is
where
I'm
not
entirely
certain,
where,
like
I
guess
so,
here's
a
question
we
do.
E
We
want
to
publish
like
when
you
publish
for
us
code
NPM,
do
you
want
the
default
version
to
be
like
the
nodejs
compatible
stuff?
And
then
this,
like
second
version,
is
saying
like.
Oh,
if
your
web
pick
but
read
all
this
stuff
instead,
so
we
have
like
basically
the
common
chance
version,
which
is
the
default
and
then
the
the
web
pack
version,
which
is
like?
Oh,
the
CSM
trees.
You
can't
do
all
that
so.
A
There's
a
new
key,
the
one
thing
I
would
say
is
like
that
key
only
matters
to
blunders
and
only
by
convention.
So
it's
not
actually
load-bearing
like
from
the
NPM
side
like
browser,
is
like
a
brand
new
key.
The
fact
that
it's
and
the
docs
at
all
actually
surprised
me,
but
it
doesn't
actually
do
anything
unless
using
I.
G
Just
like
its
when
it
gets
loaded
it
needs
to
know.
Is
it
that
I'm
you
know?
Is
it
so
that
somebody's
calling
me
because
I'm
there
is?
Is
the
JavaScript?
That's
calling
me
main
or
is
the
JavaScript?
That's
calling
me
browser
essentially
is,
although
as
an
pilot,
it
needs
to
know,
and
then
it
should
bill
to
point
back
at
the
appropriate
one.
This.
F
E
Yes,
the
the
instantiation
of
the
actual
module
happens
explicitly
like
will
replace
the
module
with
like
a
tiny,
J's
shim,
which
will
use
all
over
so
much
ASAP
I
and,
like
it
zooms
it's
like
synchronous
and
Intuit,
so
doesn't
matter,
but
then
for
the
the
es6
version.
It's
relying
on
the
bundler
to
know
how
to
actually
instantiate
the
west
of
modules.
So
it
actually
has
no
instantiation
code
at
all,
and
it
just
assumes
it
always
said.
E
G
E
G
Yeah
cuz
I
guess
I
was
thinking
so
yeah.
There
really
there's
three
options,
then
there's
I'm
being
instantiated
as
a
comment
by
a
common
J's
gem
I'm
being
instantiated
by
a
front
from
the
Wiseman
spots
perspective,
I'm
being
accentuated
by
this.
This
yes
Jim,
I'm,
being
instantiated
without
any
shim
or
I'm
being
instantiated
by
a
common
test.
Show.
Is
that
right,
I
believe
so
no.
E
G
Yes,
I
guess
I'm
wondering
if
we
can
say
okay,
if
there's
no
signal
passed,
then
we
assume
that
it's
a
bundler
and
then
just
passed
the
signal.
If
you're
doing
you
know
comma
J
s
or
you
know,
node
or
or
ESM
yeah
I
think
I
think
it
just,
and
this
is
the
problem
is
yes,
is
to
support
backwards,
to
make
sure
that
everything's
backwards
compatible
at
least
for
quite
a
while
all
NPM
packages
are
pretty
much
gonna,
be
assumed
to
have
here's
my
common
J's
version.
G
B
A
A
lot
of
complexity
here
and
there's
a
lot
of
really
good
stuff
being
said
in
this
meeting
right
now.
That
I
would
want
written
down,
because
this
is
all
actually
documentation
that
we're
gonna
need
to
explain
to
folks.
So
I
would
love
to
like
have
a
design
meeting
that
maybe
produces
an
RFC
for
this.
So.
A
B
D
B
A
A
D
A
G
For
a
short-term
solution,
yeah
I
mean
it's
like
we're,
not
really
talking.
At
least
you
know,
I'm
just
starting
so
I'm,
sorry,
if
I'm
mister
sitting,
but
it
sounded
like
you
know
before
we
were
talking
about
the
goals
for
2018.
That
optimization
was
not
one
of
them.
So
I
could
almost
say
that
this
getting
rid
of
the
one
is
something
that
could
be.
You
know,
make
it
work
for
people,
and
then
optimize
is
a
way
that
that
can
later
be
classified
is
that
we
want
to
reduce
the
size
of
the
package
that
we
generate.
B
G
F
G
G
There's
not
I
mean
I
looked
there's
not
a
XML
great
pure
raw
sex,
muck
right
right
now
that
does
everything
that
the
Stax
parser
todd's,
so
a
I
think
it's
a
great
great
to
just
have
if
it
gets
poured
it
over
and
then
B,
then
it's
something
that
I
think
these
really
low
level
system
crates
if
they
have
a
drop
in
wasm
replacement,
especially
because
there
can
still
be
a
JS
fallback.
If
it's
stiff
it's
so,
you
can
just
say
hey
if
you're
using
an
older
version
of
node.
If
your
browser
doesn't
support
wisdom.
G
Well,
then
it
just
falls
back
to
the
old
stacks
JavaScript,
which
is
just
slower
and
so
I
think,
there's
a
lot
of
these
low-level
modules
that
could
be
identified
to
be
great
contributions
to
great
things,
to
have
I've
in
the
Rost
ecosystem
and
then
also
could
be
then
published.
Npm
is
lesson
that
would
just
speed
up
everybody's
node
projects.
G
You
know
just
and
then
that
should
be
just
transparent
to
anybody
like
the
guy
who
writes
the
the
Excel
there's
a
really
big
NPM
module
that
that's
Excel,
parsing
and
writing
and
that
uses
stacks
so
I'm
not
sure
how
much
faster
this
would
make
it,
but
it'd
be
really
nice.
If
I
can
you
know
my
thought
was
really
really
nice.
I
can
do
a
PR
to
that
guy,
really
hey!
If
you
used
this
new
node,
it
just
gets
faster
and
everything's
the
same,
and
you
know
and
and
again
I.
G
Don't
think
that
project
should
be
rewritten
in
Ross,
but
it
would
be
I
think,
there's
a
lot
of
these
low-level
crates.
That
would
be
nice
to
have
sorry
low
level
packages.
It
would
be
nice
to
have
in
the
Rost
ecosystem
and
then
also
would
make
everybody's
node
better.
So
that
was
kind
of
the
idea
was
create
a
site
where
people
just
them
yeah.
A
This
kind
of
came
up
and
I
think
it's
funny,
because
I
had
pitched
something
vaguely
like
this
to
Nick
and
my's
manager
till
where
I
said
this
like.
Are
we
wasm
yet
and
it's
funny?
Cuz
Corbin
I
think
you
came
up
with
the
same
URL
and
we
were
kind
of
talking
about
this
like
we
don't
want
to
like
make.
A
Weird,
like
talking
with
this
library
experience
giving
like
so
people,
love,
to-do,
lists,
right
and
so
like
having
like
a
list
of
these
would
be
great
candidates,
or
maybe
not
even
a
to
do
so
like
these
are
things
that
people
have
already
done,
can
kind
of
help
us
with
answering
that
question
of
like
what
should
I
do
with
breasts
webassembly,
and
I
thought
it
was.
It
was
a
great
compromise
between
the
like
rewrite
everything
in
wisdom
versus
not
thing.
I.
B
If
we're
gonna
like
start
putting
random
packages
on
a
like
actual
to-do
list,
kind
of
thing
like
I
would
want
to
have
buy-in
from
the
package
maintainer
'he's,
probably
if
the
goal
is
to
upstream
it
again,
if
the
goal
is
just
to
create
a
rest
version
and
like
you
know,
obviously
anyone
can
do
that,
whatever
they
want
right.
But
if
our
goal
is
to
like
actually
upstream
again,
you
know
we
should
be
collaborative
yeah.
A
B
G
B
G
G
G
I
guess
I
was
more
I
more
had
to
kind
of
approach.
Some
of
the
people
big
depends
on
it
and
and
was
gonna
just
let
them
know
hey.
This
is
a
the
in
a
drop-in
replacement
and
then,
if
it
detects
that
Azzam
isn't
available,
then
it
just
it
already
depends
upon
the
existing
one
and
it
just
drops
that
it
just
uses
that
instead
and
then
the
other
one
I'm
working
on
right
now,
which
is
super,
obscure
and
I'm
sure
nobody
will
care
about,
but
I'm
working
on,
porting
couch
bases
database
driver
protocol
into
wasm.
G
B
Right,
if
you
you
know,
when
you
get
into
a
good
place
where
these
things
are
actually
usable,
and
you
know
you
feel
like
it's,
it's
a
success.
If
you
want
any
help
like
writing
up
the
story
about
that
or
like
just
like
feedback
on
drafts,
or
anything
like
that,
like
I'm
super
happy
to
help
I
think
probably
other
folks
here
are
happy.
Sounds.