►
From YouTube: 03-28-2018 Node.js Foundation Modules Team Meeting
Description
A
And
we
are
live.
This
is
the
March
28th
meeting
of
the
node.js
modules
team,
everyone
you
can
find
the
in
issue
52
the
agenda
for
today
and
within
that
agenda
there
is
a
dock
that
we'll
be
using
for
keeping
notes
in
that
dock.
If
everyone
who
is
on
the
call,
could
please
put
their
name
in
so
that
you
know
your
mark
does
present
for
the
meeting.
A
Without
volunteers,
I
will
try
to
do
it,
but
if
any
of
you
all
can
help
out
with
this
it's
hard
for
me
to
do
everything
so
kicking
off
the
meeting
on
our
agenda.
We
have
a
handful
of
different
things.
We
have
approving
PRS.
We
have
the
meeting
notes
from
the
last
meeting
since
we
don't
have
quorum.
We
can't
move
this
forward
Rebecca.
A
Sorry
there
are
three
PRS
against
core
that
I
wanted
to
make
sure
or
two
that
when
we
already
talked
about
that
I
wanted
to
make
sure
people
are
aware
of
there's
one
module
fix
e
stack
error
when
throwing
undefined
or
null
or
null.
This
is
semver
patch.
Arguably,
a
bug
fix.
If
you
want
to
put
eyes
on
it,
please
feel
free
Jordan
had
flagged
on
it,
but
I
can't
dig
it
up
Timothy's
who
who's
on
the
call
Timothy.
Can
you
give
us
like
a
very
short
extra
bit
of
context
on
this
one
yeah.
B
A
Thanks
Timothy,
so
if
anyone
has
thoughts
on
that,
please
feel
free
to
chime
in
on
the
issue.
I'll
follow
up
with
Brad
and
make
sure
that
they're
following
up
on
the
issue,
the
next
one
is
issue
1897
to
loader,
allow
importing
wasm
modules.
So
you
know
this
is
one
that's
been
opened
since
the
24th
of
February,
so
just
over
a
month,
this
isn't
super
new,
but
we
haven't
discussed
it.
We
was
asking
for
feedback
from
the
Wiseman
group,
Lyn
I'm
sure
this
is
something
that
you
would
have.
Thoughts
and
opinions
on.
A
It
does
not
seem
like
this
is
something
that
is
moving
forward
at
any
particular
pace,
but
just
want
to
make
sure
again.
You
know
people
know
that
this
stuff
is
happening
on
court.
So
some
quick
announcements
for
those
of
you
are
not
aware.
I
opened
an
issue
number
48
to
let
you
all
know
about
the
node
foundations
travel
fund,
so
the
node
foundation
has
a
travel
fund.
A
If
you
are
working
for
an
employer
who
I
I
forget
the
exact
rules
but
I
think
there's
rules
around
your
employer
not
being
like
a
sponsor
of
the
foundation
itself,
but
there
are
as
long
as
you
fall
within
certain
requirement
guidelines.
You
can
apply
to
the
foundation
to
get
travel
funds,
so
we
have
an
upcoming
collaborator
summit
happening
just
before
jazz
county
you.
A
So
if
you
were
planning
to
go
to
jail,
scoff
for
you
and
you
would
like
to
join
us
at
the
collaborator
summit
and
your
employer
will
not
be
helping
to
fund
your
trip,
the
foundation
can
help
you
with
flights
and
accommodations
and
all
the
information
is
documented
in
there.
So
really
great
service
is
really
great,
especially
you
know,
students
or
people
who
are
working
as
independents.
We
want
to
just
make
sure
that
that
you
know
everyone,
who's
participating
has
a
chance
to
be
there.
A
I,
don't
know
how
many
people
on
the
call
right
now
or
listening
are
planning
to
attend.
Gia's
come
on
for
you,
but
I
would
love
us
to
do
an
in-person
modules
team
meeting
during
that
time.
If
we're
able
to
reach
some
sort
of
level
of
quorum,
so
you
know
just
keep
your
eyes
open.
We
can
maybe
start
planning
something
more
closer.
The
next
one
is
the
online
module
summit
for
April
3rd,
so
that
would
be
next
Tuesday.
A
A
So,
oh
here,
here's
the
schedule,
so
this
has
this
going
for
a
couple
hours.
It's
it's
a
non-trivial,
it's
a
non-trivial
amount
of
time.
So
for
those
of
you
who
are
planning
to
participate,
you
know
this
is
creeping
up
on
us.
We
should
probably
send
an
email
to
the
group
and
make
sure
that
everyone
is
still
you
know
available
for
that
time.
So,
I'm
gonna
post
the
link
in
the
chat
really
quickly:
Gus
Kaplan,
John,
David,
Dalton,
guy
Bedford,
Brad,
ferry
Aslam,
Clarke,
myself
and
Justin.
A
Okay.
So
next
thing
up,
we
have
discussions
membership
requirements.
This
is
one
that
we
keep
floating
past.
We
haven't.
We
didn't
end
up
spending
time
on
it
last
week,
seeing
as
how
we
have
less
attendance
this
meeting,
it
seems
apropos
to
maybe
have
a
slight
conversation
about
this.
You
know
as
time
progresses,
and
this
is
something
that
you
see
a
lot
in
these
meetings.
Participation
falls
off
with
less
than
a
with
less
than
a
quarter
of
the
group.
Here
today
we
are
a
far
cry
from
quorum
and
I.
A
Don't
think
that
that's
gonna
block
us
from
getting
the
work
that
we
want
to
get
done
today,
but
especially
as
we
move
closer
to
wanting
to
be
chartered,
and
actually
you
know
own
the
work
that
we're
doing
within
the
foundation
having
some
sort
of
rules
around
participation
is
necessary.
So
some
examples
of
things
we
can
measure
are
meeting
attendance
participation
in
the
repo
participation
in
votes
in
code
review.
There
was
a
suggestion
about
potentially
allowing
people
to
be
members,
but
not
count
towards
quorum,
which
I
have
mixed
feelings
about
creating
like
specific
governance.
A
Just
for
this
group,
that's
out
of
line
with
the
rest
of
the
project,
I
put
10
minutes
on
the
agenda
to
discuss
this.
We
probably
don't
need
all
of
it,
but
I'll
just
kind
of
leave
it
to
the
floor.
Do
people
think
that
this
is
something
that
we
need
to
start
thinking
about
right
now?
Is
it
too
early?
If
we're
going
to
do
it?
What
kind
of
stuff
should
we
be
focusing
on?
Please?
D
Just
since
no
one
else
jumped
in
that
quick
I
mean
as
somebody
who
hasn't
been
involved
in
the
node
processes
and
is
trying
to
kind
of
represent
the
web
point
of
view
in
this.
In
these
meetings,
the
specific
governance
about
you
know
a
member
who
doesn't
count
to
quorum.
I,
wonder
if
that
would
kind
of
apply
to
me
like
I,
don't
necessarily
see
like
I
want
to
be
a
member
of
this
group,
but
I,
don't
necessarily
see
my
role
as,
like.
You
know,
voting
on
pr's
that
are
going
into
node.
A
So
to
just
speak
directly
to
that
and
then
CJ
I'll
go
to
I'll
go
to
you.
That
is
essentially
what
the
concept
of
an
observer
is
that
we
discussed
before,
and
maybe
the
language
wasn't
like
well
defined.
But
technically
an
observer
is
someone
who
is
you
know
a
member
of
the
group
would
get
added
when
things
are
added
can
attend.
All
the
meetings
can
participate
in
all
the
discussion,
because
arguably
anyone
can
participate
in
any
of
the
discussion.
E
D
A
A
A
A
D
D
This
is
nothing
even
that
exciting
is
just
kind
of
standard
of
how
they
work,
but
this
ends
up
blowing
up
on
the
web,
because
the
web,
the
HTML
spec,
has
reserved
these
so-called
bare
specifiers
in
the
spec
and
said
that
their
specs
to
throw
right
now.
So
you
can
only
import
or
export
with
specifiers
that
use
full
URLs
or
a
path,
an
absolute
or
relative
path.
D
This
definitely
doesn't
work
very
well
for
situations
where
you
want
to
write
a
module
that
is
consumed
by
other
people,
because
you
don't
know
where
your
package
is
going
to
be
installed
and
it
particularly
breaks
down
in
the
near
future
when
node
has
module
support
and
you
want
to
write
a
module
that
is
loadable
on
node
in
the
web.
So
the
essential
problem
that
we
need
to
solve
to
support
this
is
we
need
to
turn
this
name
into
a
path,
and
we
have
two
pieces
of
information
that
we
can
use
that
we
so
imagine
this.
D
And
base
URL
referral
URL
of
the
module
and
we
know
the
specifier
we
want
to
solve.
So
we
want
to
implement
this
function
and
in
the
case
of
three
J
s
there
we
might
have
our
base.
Url
is
like
a
index
J
s,
and
this
falsifier
is
three,
and
so
this
will
look
familiar
probably
to
anybody
who
followed
along
with
the
loader
spec
in
the
past
and
the
resolve
hooks
and
possibly
maybe
some
hooks
they're
being
proposed
to
follow
familiar
now.
This
is
not
the
type
of
hook
that
returns
to
you.
D
You
know
the
instantiated
module
itself
or
anything,
but
just
turns
a
name
into
a
path,
and
so
essentially
all
package
name
maps
are.
Is
the
data
required
to
have
a
standard
fast
implementation
of
resolve,
specifier,
so
I'm
gonna
get
into
just
a
couple
of
the
details
to
outline
what
it
does,
but
if
you
want
to
look
at
the
proposal
and
haven't
already
it's
up
on
github
in
Dominic's
area
there
under
package
name
maps,
so
please
load
that
up
and
I'll
get
into
a
little
bit
about
what
it
is
concretely.
So
we
can
talk
about
it.
D
So
a
package
name
map
is
a
JSON
file
and
the
core
part
of
it
is
that
it
just
lets
you
describe,
basically
where
packages
are
so
in
the
previous
example
that
we
want
to
import
three.
We
have
to
define
what
three
is,
and
so
we
just
have
a
package
is
stanza
here
and
then
we
have
a
three
section
and
say
the
main
maps
to
this
kind
of
theoretical
module
version
of
three
that
that
has
a
top
level.
Three
touch
is
file.
This
should
be
also
familiar
looking
to
anybody.
D
D
Yes,
so
you
can
have
situations
where
you
want
to
load.
You
know
sub
modules
more
than
just
the
main
file,
and
in
this
case
you
can
specify
the
path
that
a
name
resolves
to
separate
from
its
main
and
now,
if
you
or
three
/
and
some
other
path,
it
knows
to
resolve
three
to
this
path
and
then
append
the
path
from
the
specifier
and
then,
of
course,
you
can
import
their
define
mappings
from
multiple
packages,
and
so
this
gets
us
started
in
our
road
to
you
know
how
do
we
support
names
on
the
web?
D
But
you
know
we
know
that
it's
very
common
for
people
to
have
workflows
where
this
wouldn't
be
quite
sufficient.
This
is
kind
of
standard
capability
for
what
we
have
for
a
lot
of
system.
Letters
now-
and
we
know
that
you
know
when
using
NPM
and
possibly
other
package
managers
that
they
allow
for
this
nesting
of
packages.
If
we
want
to
get
projects
to
load
natively
on
the
web
without
having
to
run
through
a
bundler.
We
have
to
support
this
use
case
in
the
file,
so
package
name,
Maps,.
D
D
D
But
we
think
the
best
scenario
right
now
is
to
use
a
script
tag
with
a
type
of
package
map
so
that
you
can
inline
on
your
package
map
as
JSON
there
or
you
can
refer
to
an
external
file
with
script
source
so
to
just
very
briefly
kind
of
cover
the
concepts
that
were
in
the
sample
code.
There
there's
two
main
ideas
that
we
have
here
and
one
is
it
package
and
packages
are
keyed
by
package
mean,
and
then
they
have
the
metadata
in
them
that
you
need
to
resolve
a
package
name
to
a
path.
D
So
they
have
the
path
of
the
package.
They
will
have
the
sub
path,
the
relative
path
to
the
main
file
and
potentially
some
other
metadata
in
there.
Depending
on
how
rich
we
want
to
make
resolution
work,
and
then
we
have
a
scope.
So
a
scope
is
keyed
by
URL
fragment,
so
scopes
are
looked
up
based
on
the
base.
You
are
of
the
module
containing
the
specifier
and
they're
used
to
look
up
a
specific
name
to
path.
D
Mapping
and
scopes
can
nest
and
that's
how
you
get
this
overriding
information
or
behavior
and,
of
course,
they
inherit
from
their
parent
scopes.
So
resolution
we've
kind
of
taken
a
couple.
Attempts
at
writing
down
a
real
resolution,
algorithm
gonna,
take
we're
still
kind
of
just
gonna
start
fresh
at
some
points
or
the
proposal
as
it
stands
doesn't
really
have
a
concrete
resolution
algorithm,
but
the
basic
way
that
it
works.
Is
that
there's
two
main
steps
here?
D
Go
pack
prefixes
from
your
referral
URL
until
you
can't
anymore,
and
then
that
gets
you
a
scope
and
then
the
next
step
is
once
you
have
a
scope.
You
need
to
look
up
the
package
metadata
for
the
main
that
you
have
to
resolve
and
you
do
that
by
recursing
in
the
other
direction
by
your
cursor
box.
So
you
look
in
the
packages.
Stand
up,
stands
up
for
a
package
name,
that's
a
prefix
of
your
specifier
and
if
you
find
it,
then
you
have
the
metadata
you
need.
D
If
you
don't
need
to
recurse
up
and
then
there's
the
details
about
how
do
we
construct
the
URL?
So
you
have
a
current
scope
path
that
you've
built
up
as
you
recurse
down
the
scope
tree
and
then,
when
you
found
a
package,
you
found
a
package
path
and
then
you
either
have
the
main
field
or
you
have
the
sub
path
after
the
package.
Name
is
so
you
piece
those
together
to
get
the
complete
you.
A
D
Awesome
I'm
basically
done
here
some
important
details.
The
most
important
is
that
the
details
aren't
set
in
stone.
Also,
the
you
know
the
web
has
behavior
defined
for
lawn,
bear
specifiers
for
path
specifiers,
and
it's
very
unlikely
that
we're
gonna
apply
resolution
to
pass
that
begin
with,
say,
slash
or
dot
slash
and
that
it's
a
goal
to
support
composition.
So
you
can
build
up
a
mapping
out
of
multiple
mapping
files,
but
we
don't
quite
know
how
that
might
land
yet
and
then
last
I
didn't
finish
this
slide.
D
But
you
know
it's
been
said
that
we
can't
implement
node
module
resolution
on
the
web,
because
you
can't
take
the
overhead
of
pinging
the
network
and
doing
the
node
search,
but
one
interesting
way
to
look
at
the
package
name.
That
proposal
is
just
is
that
it's
a
partial
serialization
of
your
filesystem
metadata,
so
it
lets
you
perform
something
very
close
to
node
module
resolution
on
the
metadata
synchronously
so
that
you
don't
have
to
hit
the
network
all
right.
So
that's
what
I
have
for
describing
it,
and
hopefully
we
have
some
time
for
questions
or
conversation.
A
A
G
D
H
Another
thing
I
was
looking
at
it,
and
this
is
something
I
didn't
click
until
you
were
just
describing.
It
is
that
this
map
is
very
much
like
a
source
map
where
it
doesn't
have
to
be
human
writable.
It's
just
something
I
think
maybe
it
might
sell,
sell
better
if
you
say
like
hey,
like
tools
can
generate
this.
That
way
like
it
could
implement
the
well
if
its
index,
without
an
extension
versus
an
extension,
it
could
have
all
those
variations
put
in
there
in
the
map
generated,
and
then
it
would
know
how
to
resolve
them.
D
Yeah
I
mean
I,
the
web
is
a
big
big
place
and
it
might
even
be
you
know
it's
definitely
like
more
kind
of
diverse
in
terms
of
developer
opinions
and
styles
than
mode
and
I
I.
Think
that
trying
to
keep
it
at
least
a
little
bit
hand
rideable
for
some
viewers
is
important.
The
other
thing
is
we.
This
is
gonna,
be
on
the
critical
path
for
loading
resources
to
like
lender
your
initial.
D
You
know
parts
of
your
page
right
and
the
more
kind
of
flexibility
that
we
have
in
terms
of
resolving
specifiers,
the
more
metadata
we're
going
to
need
to
put
in
that,
and
this
might
make
it
a
lot
bigger.
So
there
are
certain
capabilities
that
I
think
there's
gonna
be
pushback
on
to
be
able
to
keep
the
mapping
simple
or
even,
if
generated
by
tools.
My.
H
A
One
thing
to
quickly
add
and
then
two
to
Rebecca
you
would
be
able
to
do
deep,
searching
into
modules.
It
would
be
just
that
you
wouldn't
be
able
to
drop
the
file
extension
or
load
specific
folders,
but
you
could
do
like
lo
slash
module,
dot,
MJS
or
J
S
or
modules.
Slash,
folder,
slash,
folder,
slash,
folder,
slash
module.
D
F
Yeah,
it's
it's
worth
noting
that
in
the
very
near
term,
like
the
next
three
to
three
months
or
so,
NPM
intends
to
have
an
install
path
for
ESM
for
browsers.
That
does
not
use
node
modules
that
and
like.
If,
if
something
like,
this
were
to
happen,
that
would
be
the
place
where
it
would
generate
that
at
install
time.
F
Well,
I,
don't
think
we
could
produce
it
from
a
package
lock
or
at
least
not
sufficient
for
full
cross
compatibility,
because,
as
Myles
says,
you
have
to
write
it
in
a
specific
way
for
it
to
work
in
both
places,
whereas
the
Installer
can
actually
like
look
at
how
it's
written
and
produce
the
mapping
at
the
needs
to
just
by
looking
at
the
JavaScript
so
like.
If
this
worked.
F
D
F
Just
that
we
were
going
to
be
generating.
If
this
were
to
become
a
standard,
we
would
be
generating
this
as
part
of
your
install,
so
it
should
be
transparent
to
to
end-users
I
agree.
F
Trying
to
fit
like
to
me,
this
feels
like
the
sort
of
thing
that
should
be,
should
not
have
a
lot
of
logic
in
it.
It
should
be
pretty
dumb
and
that,
if
you
want
more
logic
that
that
it's
probably
time
like
there
needs
to
be
an
escape
valve-
and
this
doesn't
provide
one
like
my
biggest
concern
about
this
is
this
is
being
used
as
a
well
instead
of
having
loader
hooks.
Let's
have
this
because
you
know
Gmail
can't
use
loader
hooks
and
it's
like.
Well,
yes
having
a
fast
path.
F
D
You
know
intimately
know
the
entire
history
of
the
loader
spec,
but
it
did
seem
to
fail
at
least
after
many
many
many
years
and
I.
Don't
know
how
likely
we
are
to
get
something
like
loader
hooks,
I.
Think
there's
been
at
least
from
my
point
of
view.
There's
been
more
like
a
warmer
reception
to
this
idea.
Our
kind
of
working
theory
was
that
people
aren't
going
to
be
able
to
write
like
interesting,
loader
hooks
on
their
own,
like
that
often,
and
that
mostly
there's
gonna
be
a
loader
hook.
D
D
A
J
Just
wanted
to
clarify
something.
So
when
we
talk
about
loader
hooks,
usually
we're
talking
about
the
request
data
hook,
not
the
what
WG
loader,
which
is
an
open
issue
on
the
HTML
standard,
that's
stalled
because
of
a
perpetual
chicken-and-egg
problem.
They
refused
to
move
forward
with
the
hoe
because
they
require
userland
experimentation.
Just
as
impune
people
are
stating.
We
can't
experiment,
because
we
don't
have
the
hooks
so
there's
no
way
for
us
to
progress
on
it,
because
we
are
blocked
by
something
that
we
cannot
do
in
either
direction.
Okay,.
A
H
I
share
my
screen:
I
can't
switch
between
desktops.
Alright.
Would
you
like
to
open
the
document
and
then
I
can
just
talk
through
it
since
I.
A
A
A
H
Cool
alright.
So
last
week
we
were
in
over
a
handful
of
these
use
cases,
the
ones
up
here
without
names
associated
with
them.
Afterwards,
some
more
use
cases
were
then
filled
in
so
I
figured.
We
could
just
go
through
some
of
them.
I,
don't
know
if
you
wanted
me
to
just
like
read
loud
or
step
through
there,
how
many
people
have
actually
reviewed
the
document.
I
can't
see
hands
so
I'll,
just
assume,
yea
or
nay,
Lots,
no,
okay,
so
I'll
start
with
Jordans.
H
It
says:
Francis
has
a
large
application
and
wants
to
be
able
to
granule
a
migrate
from
CJ
STS
M.
This
is
talking
more
about
doing
it.
Let's
see
piece
by
piece.
It
looks
like
syntax
at
a
time
without
forcing
to
change
their
own
code
base
and
without
creating
a
break
and
change.
Simper
major
change.
I
I
he's
not
here
to
talk
about
it.
So
I
can't
really
go
into
this
but
I.
H
A
H
D
A
K
H
H
Longer
paths
and
I
think
webpack
has
a
plug-in.
That
also
does
something
like
that
too.
So,
if
we're
looking
for
things
among
ecosystem,
they
currently
do
something
like
that.
There's
a
web
pack
plug-in
and
then
there's
that
module
remaps
require
paths.
Okay,
so
fine
we'll
go
to
Justin's
a
little
bit
lower
okay.
So
these
are
web
web
scenario
and
web
use
cases.
Let's
see,
John
a
web
developer
wants
to
create
a
module
that
imports
dependencies.
Yes,.
D
So
they're
mostly
concerned
about
specifics,
how's,
my
mic
great
yeah,
yeah
they're,
concerned
around
specific
cases
that
you
know
we
would
really
like
to
enable
where
we
don't
require
tooling
necessarily
to
go
from
to
map
from
kind
of
a
node
world
to
a
web
world
where
things
just
kind
of
naturally
work
with
native
support.
So
the
first
one
is
simply
the
John
wants
to
write
a
module
that
imports
any
dependencies
and
works
natively
on
the
web
without
using
any
build
tools.
This
works
today
with
paths,
but
it
doesn't
get
work
with
names.
H
You
fill
that
in
with
with
that
that
Colonel
there,
where
that's
the
that's
the
piece
that
you
want,
which
is
not
with
paths
but
with
the
the
node
resolution,
because
that
would
help
a
little
bit,
because
when
I
first
saw
that
too
I
was
like
well,
people
can
already
do
this
today.
I
have
a
yes
package
and
know
that
that
has
pads
that
also
works
in
the
browser.
So
yeah.
D
H
D
H
D
H
D
Let's
see
the
next
one
with
Sean,
this
is
kind
of
a
expansion
of
the
first
one,
so
she
wants
to
use
modules
from
NPM
on
the
web
natively
without
doing
any
source
transformations.
Maybe
she
likes
her
dev
tools,
editing
capabilities
or
something,
but
she
doesn't
want
the
other
tools
that
she
used
to
understand
where
the
dependencies
are
so
the
idea
IDE
should
be
able
to
open
the
dependency
if
they
do
command
click
or
whatever
their
linter
should
be
able
to
pull
in
or
locate
files
to
make
sure
that
the
imports
are
actually
valid.
D
The
build
system
would
have
to
find
files
in
order
to
bundle
them.
So
this
speaks
to
the
ability
of
tools
to
agree
upon
what
a
specifier
means
yeah
and
then
the
next
one
is
about
doing
things
in
situations
where
you
might
not
have
build
tools
available
at
all.
So
Amy
just
wants
to
write
a
jsbin
that
uses
some
models
from
NPM.
D
H
D
Down
in
cases
where
there's
no
specific
things
in
modules,
the
next
one
is
a
Jeff
here-
wants
to
deploy
his
app
to
production,
but
he
would
like
to
take
advantage
of
a
CDN
for
common
modules.
So
you
either
get.
You
know
them
to
serve
to
take
the
bandwidth
hit
or
to
have
highly
cashed
common
modules
out
there
from
URLs.
So
he
might
want
to
transform
this
graph
to
point
to
modules.
D
The
next
one
alex
is
just
a
library
author
who
wants
to
write
a
package,
that's
intended
to
be
used
on
the
web
and
node,
so
it
needs
to
be
compatible
with
those
and
then.
Finally,
this
is
kind
of
something
that
just
came
up
today
when
we
were
trying
to
use
modules
from
NPM.
Sara
would
like
to
use
dev
Bills
of
modules
at
dev
time
and
frog
builds
at
prod
time,
probably
without
rewriting
her
source.
D
D
H
They
just
need
more
D
or.
A
K
Okay,
good
all
right,
so
I
think
we've
covered
the
the
first
one.
This,
the
second
one
was
Thomas,
which
was
about
being
able
to
use
aliases
for
dependencies
and
I.
Think
that
a
Cugini
map,
if
summarize,
that
one
very
well
the
third
one
for
Peter,
is
that
not
only
can
you
imagine
aliases
to
get
into
a
package,
you
can
also
imagine
then
some
kind
of
indirection
so
that
you
have
only
named
entry
points
into
that
package.
K
K
K
D
K
Okay
and
then,
finally,
our
last
one
is
Vlad,
which
is
about
ensuring
that
we
we
declare
all
of
our
dependencies
explicitly
and
so
and
it's
it's
possible
in
in
regular
vanilla
packages
today
to
accidentally
in
your
in
your
modules,
ask
end
on
something
inside
your
node
modules
that
you
didn't
ask
for
directly,
and
this
is
because
dependencies
can
have
dependencies,
and
so
this
is
saying
that
it
is
useful
to
at
least
have
some
mode,
maybe
for
production,
maybe
only
for
development
where
it
effectively
will
not
load
things
that
you've
not
declared
up
front.
So.
H
K
K
Yes,
I
think
the
the
package
name
maps
and
wasa
serie
cuts
are
a
combination
of
the
resolver
hook
and
and
package
name
map
does
seem
to
provide
solutions
for
each
of
these
use
cases.
I
just
wanted
to
be
very
clear
on
the
problems
that
we're
solving,
so
we
can
enjoy
with
them.
They
map
forwards,
yep,
absolutely.
A
A
F
It
affects
how
NPM
installs
your
modules,
so
your
node
modules
folder
will
only
have
your
directive
it,
so
it
reduces
deduplication,
basically
there's
a
price
associated
with
it.
Currently,
if
you
want
to
flat
install,
you
can't
have
this
without
something
to
the
resolution
algorithm,
and
so
this
is
where
it's
like
becomes
more
complicated
and
it's
why
it's
not
the
default
today.
F
If
we
had
some
other
way
of
achieving
this,
then
I
would
be
inclined
to
have
global
style,
be
the
default
like
if
we
like,
when
we
have
asset,
then
it
starts
to
when
we
have
an
ability
to
install.
You
know
plat
dependencies
in
addition
to
your
node
dependencies,
those
today
things
for
your
web
web
app,
then
I
start
to
go
well.
F
Maybe
node
should
only
should
do
this,
so
what
global
style
does
is
installs
your
direct
dependencies
as
separate
folders
and
your
node
modules
all
of
their
dependencies
are
installed
flat
inside
of
each
of
them,
so
it's
one
layer
deep
rather
than
the
old
in
pm-2
style,
where
it
was
a
arbitrarily
deep
tree,
basically
as
deep
as
possible.
This
is
this
mirrors
the
way
your
your.
A
Do
people
think
at
this
point
it
might
make
sense
for
us
to
just
break
early-
and
you
know,
put
the
word
out
to
people
hey.
Let's
all
keep
working
on
this
document
and
maybe
earmark
more
time
for
next
week.
I
only
did
25
minutes
this
week.
Maybe
would
make
the
majority,
if
not
the
entire
meeting
two
weeks
from
now
focused
just
on
this-
is
that
seem
reasonable
to
people.
H
Yeah
I
I'm
I
have
one
use
case
that
I
like
to
bring
up
it's,
not
mine,
but
it's
a
it's
one
that
I
I
think
speaks
to
me
more,
and
this
is
one
by
Jordan
and
it
says
Karl
is
unable
to
use
the
s
Emmett
in
their
code,
because
the
company
is
still
unwilling
to
change
the
coding
style
and
workflow.
However,
they
want
to
use
Eugene's
gsm-only
package.
H
They
are
able
to
access
it
with
require
and
make
use
of
it
both,
while
on
an
old
version
of
node
that
lacks
C
is
some
support
and,
after
upgrading
to
a
newer
version
of
node
that
has
ESM
support
this.
That
use
case
seems
very
spot
on
with
the
ESM
loader
that
I've
written
and
I
wonder
like
I.
Think
that
that's
going
to
be
something
for
a
transition
period
where
you're
gonna
see
people
wanting
to
do
something
like
that.
H
A
So,
to
speak,
to
like
one
of
the
things
that
I
had
mentioned
earlier
in,
like
the
manifesto
that
I
think
speaks
to,
this
is
like
allowing
people
to
publish
multimode
modules.
Whatever
the
mode
is.
I
was
a
little
bit
too
specific
and
in
my
original
language
of
it
and
specifically
where
they
say
you
genius,
ESM,
only
packaged
mm-hmm
and
I
do
think
that
for
core,
we
need
to
find
a
balance
as
to
how
much
we're
trying
to
optimize
and
help
people
for
I
think
we
need
to
allow
and
enable
people
to
create
bridges.
A
But
we
don't
need
to
like
make
all
of
them.
I
know,
for
example,
being
able
to
require
ESM,
since
we
have
an
async
loader,
isn't
really
possible
from
like
a
fundamental
standpoint,
but
we're
getting
really
into
the
weeds,
but
I
think
the
phrasing
here,
especially
that
it's
explicitly
ESM
only
like
I,
don't
want
to
say,
won't
fix,
but
I
think
that
there
are
solutions
that
can
allow
module
authors,
more
flexibility,
but
a
module.
Authors
want
to
only
publish
ESM
I'm,
not
sure
if
that
should
be
no
the
responsibility
to
repave
the
bridge.
Well,.
H
I
think
the
the
bridges
are
there,
it
like
from
user
land,
but
I
just
thought
that
was
it.
That
was
us
like
a
on
the
nose
description
of
the
scenario
that
that
I've
been
working
on
so
I
I
dig
that
I
also
I'd
still
like
to
based
on
that
I'd
like
to
monitor
that,
though,
just
because
it
does
touch
on
things
like
a
synchronous
versus
synchronous,
loading
of
modules
for
things
like
CJ
s
in
a
row.
So
that's
that's!
All
I
just
thought
that
that
was
super
cool
cool.
E
Just
to
sharpen
our
focus
here,
a
little
bit
I
think
the
job
of
this.
This
group,
working
through
these
use
cases,
is
to
figure
out
what
the
common
paths
are,
that
our
users
are
going
to
want
to
do
that.
We
are
going
to
want
to
make
easy
and
if
we
make
the
and
our
secondary
goal
should
be
to
make
the
more
on
you
all
paths
possible
may
be
possible
through
user
land
solutions
like
using
JD
DS
ESM
to
do
this,
but
if
so,
our
job
is
to
discriminate
between.
E
H
The
the
use
the
use
case
by
Jordan
doesn't
really
go
into
that
from
my
my
my
opinion
is
that
it's
doable
today
it
gets
kind
of
into
implementation
wise.
But
what
you
do
is
you
lock
down
the
module?
That's
free
returned,
so
its
exports
and
module
properties
are
frozen
and
that
that
adds
a
little
bit
of
the
insurance
against
people.
A
monkey
patching
the
return
object.
So
then
you
just
you
parse
and
you
validate
synchronously
in
the
the
order
that
you
would
normally
asynchronously
and
since
node
file
operations
can
be
sync,
it
just
shakes
out.
A
Okay,
great,
we
are
over
time
at
this
point
and
unfortunately
Jeremiah
I'm
not
trying
to
skip
on
you,
but
I
actually
have
another
meeting
that
I
need
to
get
to
and
I
hold
the
keys
to
this
so
Jeremiah.
If
your
things
like
one
minute,
maybe
get
that
in
really
quickly,
but
just
as
a
heads
up
I'm
letting
you
know
so.
I
H
So
I
support
that
as
well
and
it's
still
doable
I,
don't
know
what
to
tell
you.
I
mean
I
support
it
and
I
support
top
level
of
Wade
I.
Think
top
level
await,
though,
could
be
used
as
a
band-aid
to
try
to
work
around
the
asynchronous
synchronous
gap
between
common
areas
and
ESM,
so
I,
don't
necessarily
think
it's
it
should
be.
H
That
should
be
it's
it's
a
driving
force,
though
the
main
thing
for
top-level
await
is
convenience
with
like
CL
eyes
and
so
I
think
the
proposal
by
miles,
which
is
variant
3c,
which
is
as
long
as
your
module,
doesn't
have
any
exports.
You
get
top-level
await.
That's
how
I
support
it.
Works
works
well,.
A
Okay
and
with
that
we're
gonna
we're
gonna
close
up
for
the
day.
Thank
you
everyone
for
your
time,
just
one
more
time,
if
you
have
not
put
your
name
into
the
agenda,
the
notes
document,
please
do
that's
how
we
keep
track
of
who's
shown
up
I'll
quickly,
check,
YouTube
I,
don't
see
any
questions
right
now,
so
we'll
move
forward.