►
From YouTube: Node.js Tooling Group Meeting
Description
A
A
So
when,
when
I
start,
the
livestream
it
like
holds
up
the
livestream
and
I,
see
myself
talking
delayed
by
a
few
seconds,
and
it's
very
off-putting
anyhow
I
got
rid
of
that,
and
here
is,
let's
see,
might
be
worth
pulling
up.
These.
A
A
A
Last
meeting
it
was
suggested
that
we
look
into
like
pythons
Shu
till
module
because
there's
a
lot
of
stuff
in
there
that
we
feel
it
would
make
sense
to
have
an
in
node
core,
and
so
that's
where,
if
you're
familiar,
that's
where
you
know,
pythons
version
of
of
like
Rim
raft
lives
right,
there's
also
a
copy
tree
function
and
a
couple
of
other
things
and
I
am
gonna.
Guess
there
hasn't
really
been
any
movement
here,
but
I
don't
know.
Has
there
been.
B
A
B
A
B
A
Okay-
and
you
know
what
I
think
we
can
take
the
cream
wrap
off
of
this
list,
then,
since
it
kind
of
folds
into
this
task.
Okay,
so
anything
else
on
the
Shu
till
front
that
that
we
want
to
want
to
discuss
there.
A
They
may
enhance
some
of
these,
but
essentially
I
had
to
draw
the
line
somewhere.
For
example,
like
yards
is
going
to
have
more
features
than
New
York's
parser,
but
I
didn't
want
to
just
just
duplicate
everything.
So
these
there's
nine
on
this
list,
they're
roughly
in
order
of
popularity
I,
think
commander
might
actually
be
the
most
widely
used
one,
but
I
went
through
and
just
picked
some.
A
Common
features
of
these
two
to
kind
of
just
just
get
a
feeling
for
what
what
they're
doing
some
of
the
less
widely
supported
features
I
left
out,
just
because
I,
don't
know,
I
mean
again,
I
had
to
die.
I
could
probably
just
spend
weeks
and
weeks
doing
this
and
doing
every
single
little
thing
and
I
didn't
want
to
do
that.
So
I
picked
some.
If
you
scroll
down
this
comparison,
matrix
you're
gonna,
see.
A
That
all
of
them,
then
this
is
what
all
of
them
do
and
I
think
what
we
want
to
do
with
this.
This
research
was
kind
of
get
get
an
idea
of
what
people
are
using
now
and
what
people
are
doing
now
and
so
there's
a
precedent
if
we
want
to
want
to
push
forward
with
something
so
because
there
are
so
many
different
ways
to
do
argument.
Parsing
and
people
have
different
ideas
about
how
it
should
work.
A
I
think
it's
important
to
just
kind
of
go
with
what
is
popular.
Essentially,
so
all
of
these
nine
modules,
the
he'll,
do
a
few
things.
One
of
them
is
they
all
return
the
arguments
and
some
sort
of
object,
and
again
people
may
have
different
ideas
about
how
that
should
be
done.
What
that
object
should
look
like,
but
they
all
do
it.
A
A
Some
of
them
will
support
number
types,
some
of
them
won't.
Some
of
them
will
even
differentiate
between
floats
and
integers,
but
at
bare
minimum
they
they
have
an
idea
of
they
have
some
notion
of
types
and
they
minimum
support
a
string
type,
and
so
that
is
an
option
with
some
sort
of
value
attached
to
it
and
in
a
boolean
type,
which
is
essentially
a
flag.
So
if
it's
present,
then
it's
yes,
if
it's
not
present,
then
it's
it's
no
or
true
or
false.
A
So
I
mean
yeah,
there's
no
yeah
on
Windows,
for
example,
you
might
encounter
tools
that
expect
a
a
forward,
slash
or
something
like
that.
I
didn't
see
anything
like
that.
Maybe
some
of
them
do
but
I
couldn't
find
it
referenced
anywhere.
So
evidently
it's
not
it's
not
very
popular,
and
so
that's
kind
of
what
we
can
we
can
for
where
there
have
been
exceptions
or
weirdness
I've
tried
to
add
a
little
note
about
it.
A
It
looks
like
the
the
equals
so
like
foo
equals
bar
is
widely
supported.
It
is
just
kind
of
unclear
from
the
documentation
I'm
on
command
line,
arts
and
basi,
whether
if
if
that
is
actually
supported
because
I
didn't
actually
go
and
download
it
and
try
it
out
myself,
that's
what
I
would
need
to
do
because
it
wasn't
mentioned
in
the
documentation
where
it's
probably
safe
to
us.
A
I,
don't
think
it's
safe
to
assume
they
do,
but
it's
it's
likely
that
they
do
I
just
couldn't
confirm
it
just
because
it's
so
so
such
a
popular
way
to
wait
to
do
this.
So
that's
kind
of
the
that's
kind
of
the
idea.
I
just
wanted
to
see
where
the
where
user
Lane
was
at
right
now
with
this
and
what
we
could
use
as
kind
of
a
bare
minimum
bare
minimum
functionality
for
something
like
this.
So
what
what
do
you
think
I.
C
Think
it's
a
positive
data
point
it.
It
seems
like
there
is
sort
of
consensus,
and
at
least
subsets
like
there's,
no
obvious
contradictions,
at
least
in
this
group
like
something
that's
the
girl
like
that
separates
that
turns
into
turns
it
into
not
the
subset
of
of
the
whole
right.
Yeah
I
think
like
a
forward,
slash
which
I'm
not
in
favor.
C
D
Yeah
I
mean
we
can
essentially
just
treat
is
like
well
this
this
flag
showed
up
by
you
know:
here's
the
list
of
places
for
that
flag
showed
up,
and
then
I
guess
I
owe
ya.
That
would
be
the
other
one
is
passing
in
values
to
doubled,
out
double
flags.
Yeah
supporting
equals
is
pretty
universally
supported.
Supporting
not
having
equals
seems
to
be
the
the
corner
case.
There.
D
A
So
right
so
positional
I'll
just
explain
this
quickly,
so
positional
is
something
that
you
can
specify
that
doesn't
have
an
option
before
it.
Typically,
this
might
be
a
file
name.
So
if
you're
in
like
mocha
food
Jas
food
is,
is
your
positional
argument
commands
think
like
get
commands,
so
it's
get
some
sub
command
and
then
and
then
you
pass
a
bunch
of
arguments
and
that's
that's
what
a
command
is.
A
To
be
clear
down
below
in
the
types
comparison
count
means,
if
you
say
like
V,
V
V,
that
would
be
the
value
of
V
would
be
three
very
attic,
meaning
you
can
supply
the
same
thing
multiple
times.
So
if
there
was
a
flag
you
know
file,
you
could
say,
should
I
file
fou
file
bar,
that's
just
and
you
would
get
back
at
an
O,
an
array
or
something
like
that.
C
A
A
D
D
C
C
A
Right
if
there's
right,
it's
if
anybody's
gonna
scream
bloody
murder,
it's
because
it's
it's
missing
a
certain
feature.
It's
not
because
it's
doing
something
that
they
believed
to
be
as
completely
wrong,
because
as
bad
there's,
no
there
are
no
conflicts
here.
There's
no
like
incompatibilities!
It's
all
some
subset
of
this
wider
set
of
features,
so
so
so
in
that
and.
C
A
D
C
C
D
I,
don't
think
like.
If
I
was
writing
this
I,
don't
think
my
design
would
be
the
same
for
those
two
different
goals
like
if
I'm
building
this
to
be
built
off
of
then
I
want
it
to
expose
like
the
parsing
to
sit
like
you
know,
literally
like
here's,
the
list
of
stuff
I
got
in
the
order.
I
got
it,
you
decide
what
you
want
to
do
with
it.
D
C
That
I
think
that's
a
secondary,
I,
think
yeah
I.
Think
or
you
know,
my
understanding
is
that
the
first
goal
primary
goal
is
to
have
something
reasoning
out
of
the
box
for
people
who
don't
who
fancy
or
varsity
like
the
use
case.
Is
you
build
a
tool
that
takes
one
argument
and
then
you
know
something
comes
in
and
you
need
to
take
a
second
argument
and
suddenly
you
have
you
have
your
own
argument
cause
instead
of
from
the
get-go
having
something
just
using
it
and
then
replacing
it.
A
Right,
you
know,
I
mean
I
would
say
absolutely
both
of
those
are
goals,
but
but
the
highest
priority
is
just
let's
get
something
that
works
out
of
the
box.
For
these
simple
cases,
and
it's
not
it's
not
so
simple
that
it's
useless
and
I
think
if
we
take
a
subset
of
these
these
features,
it
won't
be
useless
because
I.
G
Think
that's
I
think
it's
a
great
idea
that
I'm
thinking
at
Google
right
now
for
our
code,
samples,
we're
actually
removing
yards
because
it
requires
people
install
yards
to
run
our
code
samples.
So
you
know
having
something
like
this.
We
would
probably
immediately
use
for
just
simple
code
samples
and
stuff
stuff,
like
that,
from
the
point
of
view
of
yards
parser.
How
many
yards
is
more
of
a
visual
visual
library
yards
parsers.
G
D
E
G
C
Again,
from
a
little
bit
of
perspective
of
been
working
on
a
lot
of
the
node
Court
ruling,
which
is
Python
because
of
bootstrapping
almost
every
one
of
those
tools,
if
you
look
at
the
git,
history
has
been
there's
like
one
hour
added,
then
the
second
argument
added
now
I'm,
just
looking
for
a
syringe
and
I
think
I'm
holding
any
fine.
So
it's
it
makes
a
ton
of
sense
from
again
from
the
Thule
perspective.
That's.
G
C
What
I'm,
seeing
or
at
least
the
answer,
I'm
giving
for
people
who
come
with
this
is
it's
derived
from
the
small
core
perspective
is:
does
the
feature
in
core
give
a
significant
amount
of
benefit
over
having
it
so
userland
module
if
it
can
be
easily
implemented
as
usual,
and
with
what
I
point
to
like
a
very
good
library
that
does
it
and
there's
not
a
lot
of
like
this
incremental
feature?
Requests
persisting,
it
happens.
D
G
Yeah
I
was
almost
wondering
maybe
a
conversation
worth
having
any
more
detail
elsewhere,
but
but
would
we
consider
adding
this
into
a
standard
modules
namespace
like
I,
wonder
if
this
has
benefit
as
a
standard
module
I'm
just
having
this
on
my
mind,
because
I've
been
talking
to
Daniel
Arenberg
about
tc39,
wanting
to
standardize
on
some
modules,
I
wonder
if
I
wonder
if
we
could
move
it
into
a
written
standards
document
and
then
it
kind
of
moves.
The
argument
away
from
the
node
issue
tracker,
which
could
be
nice
right,
I.
G
Yeah
I
think
we're
I,
don't
want
to
change
the
topic
conversation,
but
we're
kind
of
doing
an
experimental
one
right
now
with
UID
v4,
just
to
see
what
it's
like
to
work
with
them
and
write
a
document
for
it.
There's
a
learning
curve
like
I.
Don't
want
to
advocate
that.
That's
what
we
do
for
everything
in
the
tooling
working
group
at
this
time.
A
To
when
we
implement
this,
it
I
don't
want
to
do
it
in
such
a
way
where
it
is
you
can't
hack
at
it.
So
yeah
I
want
to
prioritize
hey.
This
should
work
out
of
the
box
should
be
easy
to
use.
But
you
know
if
there's
maybe
a
class
or
some
thing
that
that
a
user
could
go
and
and
in
subclass
and
and
overwrite
the
behavior
I
mean.
Maybe
that
would
suffice
for
people
get
started,
I
don't
know.
A
Maybe
it's
maybe
that's
just
too
tough
to
do
properly,
but
of
course,
then
maybe
it
just
could
be
well
we'll
have
this
class
internally.
We
won't
expose
it
for
now
until
we're
sure
about
what
this
like
hackable
API
could
look
like,
because
I
don't
what
I
would
like.
You
know,
library
authors
to
be
able
to
build
on
top
of
it,
but
I
also
don't
want
to
design
a
stupid
API
on
accident,
because
that's
really
easy
to
do.
I.
G
Was
gonna
say
even
if
we
don't
like
to
do
the
full-on
try
to
write
a
tc39
standards
approach,
maybe
start
with
a
readme
document
that
just
describes
the
perfect
world
and
a
few
of
us
noodle
on
are
not
the
perfect
world
would
like
to
describe
the
API
in
the
abstract
or
certain
degree,
and
what
extension
would
look
like,
etc,
etc.
Rated
this
design
document,
because
it's
probably
pretty
easy
to
actually
program
the
command-line
parser.
A
A
G
G
D
Yeah,
this
is
simple
enough,
then,
that
I
kind
of
feel
like
running
code
would
help
us,
but
I,
think
you're,
right
I
think
it
should
be
a
smaller
discussion
but
like
for
the
you
know
like
it
consumed
like
if
I
wanted
to
have
something
that
was
maximally
consumable
by
things
than
it
would
be
like
a
token
stream
is
what
it
would
produce
and
then
and
then
you
process
that
token
stream,
and
it's
done
some
pre-processing
on
it.
So
it's
a
little
better
than
just
an
array
of
you
know
things
that
came
in
from
the
shell.
A
G
I
think
I
think
my
slate
I'm,
my
slight
leaning
right
now
is
towards
the
I
think
this
is
your
point
to
trace
is
like
the
super,
simple
one
that
as
I
said,
if
I
cleans
up
just
code
examples
throughout
the
world
and
maybe
just
try
to
keep
it
as
constrained
as
we
conceivably
can,
and
maybe
as
nice
as
it
would
be,
to
delete
yards
person.
Maybe
maybe
Rebecca's
right
that
that
shouldn't
be
the
goal
of
the
project.
I.
G
A
G
B
A
A
I'll
invite
him
too,
but
I
won't
expect
him
to
do
anything
so
everybody
else
you're
on
the
hook.
So
all
right.
Well,
that's
that's
good
I'm
glad
to
see
that
kind
of
coming
together.
There's
the
other
thing
on
this
list
is
shims
I
I,
don't
think
we
need
anything
right
now.
You
know
it's
always
I'm
always
interested.
What's
gonna
happen
with
this
whole
namespacing
thing,
but
that
seems
to
have
been
tabled
or
just
nobody
is
really
working
on
it
and
I've
paid
attention
to
it.
C
C
H
A
Bike
shed
it's
a
poop
emoji,
alright,
okay,
so
yeah,
but
anyhow
I
guess
so
somebody
should
do
something
with
that.
Maybe
and
just
see
what
happens
and
that
PR
with
with
hey
I'm
using
this
name
space
like
a
meme,
neener
neener
and
see
what
people
do
so,
maybe
I'll
open
the
floodgates,
maybe
they'll
be
terrible.
Who
knows
but
I
don't
think
we
need
any
shims
for
now.
A
C
C
They
did
what
they
were
factored
out,
a
lot
of
their
internal
code
and
they're
reusing
new
nodejs
core
functionality
and
they
wrapped
it
around
an
API
which
gives
a
better
you
know,
upgrade
guarantee
you
don't
have
to
recompile
it
every
time.
So
essentially
they
turn
themselves
into
a
shin.
But
like
that's
that's
not
for
me.
That's
that's.
A
success
story
know
when,
when
that's.
G
C
G
G
But
maybe
a
similar
topic,
conversation
I
had
a
conversation
this
week
that
came
about
at
work
about
why
we
were
using
exec
or
its
cross
spawn
at
this
point
like,
what's
the
actual
Delta
like?
Why
don't
we
just
make
spawn
work
on
or
accept
work
on
Windows
properly,
so
I
had
this
big
long
conversation
about
what
the
actual
deltas
are
and
there's
like,
I,
don't
know,
ten
things
that
don't
work
well
on
Windows
and
exec
exec
or
response
ink
are
shimming.
G
C
There
is,
there
is
sort
of
an
issue
left
behind
from
a
small
core,
which
is
some
of
the
api's
look
like
system
calls.
Exact,
can
spawn
a
fork
they're,
not
really
high-end,
consumable
API
in
their
design
and
I.
Think
that's
the
problem.
So,
for
instance,
when
you
do
exact
is
like
for
for
spawning
a
new
shell
default.
Shell,
behavior
and
classic
sign
in
windows
have
slight
differences.
C
Windows
does
not
assume
that
the
new
shell
replaces
the
old
shell
windows
starts
with
you
know:
minus
X,
like
with
verbosity
and
then
six
dozen
and
so
having
the
system
call
be.
Opinionated
is
a
little
bit
program
problematic.
The
same
went
with
which
a
banks,
nobody
assumes
there's
a
shebang
in
a
Windows,
File
and,
and
that
has
any
meaning
so
I
would
say
one
of
those
you
know
exec
our
or
the
others
should
be
a
new
thing
with
the
opinions
built
in
because
fixing
those
drug
is
gonna,
be
very
surprising
for
a
lot
of.
G
G
It's
pretty
confusing
that
if
I
pick
up
something
from
the
core
API
that
it's
not
gonna
work
on
Windows,
so
I,
don't
I'm
not
disagreeing
with
you
I,
just
wonder
how
we
would
how
one
might
name
that
API
so
that
people
like
oh
I,
should
just
use
this,
because
the
other
ones
don't
actually
work
like
I,
just
call
it
like
good
spawn
or
something
like
yeah
compromise
spawn.
We.
C
E
C
C
This
and
this
and
this
and
we
don't
assume
any
other,
inherited
behavior
from
the
operating
system
like
we
parse
the
shebang
and
we
do
it
Dartmouth.
We
do
our
thing.
We
start
with
we're
both
the
off
by
by
by
design
not
by
default,
and
we
put
everything
as
an
argument.
You
can
control
it
like
you
can
hook.
Maybe
you
know
you
hook
the
shebang,
parsing
you
can
put
a
flag
from
above
city.
You
can
pass
if
you
need.
C
If
you
really
want
to
spawn
a
shell,
you
can
pass
the
shell
and
its
arguments
because
right
now,
like
yesterday,
what
I
was
talking
about
like
if
you
want
to
spawn,
see
and
be
on
Windows,
it
start
with
/s,
/t
and
/c.
You
want
to
play.
You
want
to
start
with
/q,
which
is
shocked,
minus
X,
plus
X.
We
can't.
C
G
Kind
of
like
that,
this
would
might
dovetail
the
really
simple
argument
parser,
because
one
thing
I
think
that's
nicer
about
the
exact
API.
You
correct
me
if
I'm
wrong,
but
the
spawn
API
takes
a
second
argument
of
an
array
of
options
passed
to
it,
whereas
the
exact
can
just
take
a
string
which
will
be
executed.
Yeah
having
an
actual
executed
string
is
in
a
lot
of
cases
nicer
if
we
can
standardize.
A
A
G
A
little
immature
but
I
have
been
talking
to
Daniel
Arenberg
about
his
idea
of
putting
a
module
in
the
standard
space
and
seeing
how
people
we're
talking
about
putting
UID
in
a
standard
module
space
in
the
browser,
because
it's
used
by
like
350,000
different
repos
and
is
a
really
simple
API
and
seems
like
a
good
candidate
for
a
standard
module.
So
I
don't
know,
I.
Think.
E
G
G
C
G
A
G
G
This
is
it
this
is
the
idea
I'm
curious
to
see
what
pushback
we
get
because
like
to
me
like
the
benefit
is
like
some
of
these
things
are
so
wildly
popular.
One
should
almost
think
of
them
as
a
building
block
like
the
UID
is
used
by
three
and
50,000
repos
like,
but
that's
almost
universally
used
for
making
these
little
identifiers
there's
some
other
modules
like
that.
G
That
are
really
simple,
little
utilities
used
by
hundreds
and
hundreds
and
hundreds
of
thousands
of
libraries
I'm
just
thinking
if
we
are,
if
a
standard
module
namespace
is
something
that
is
gonna
happen.
Why
not
use
the
giant
corpus
of
NPM
to
do
some
statistical
analysis
of
what
people
are
actually
doing
with
JavaScript,
rather
than
try
to
event
invent
a
new
standard
module
set
of
api's
like
we've
been
running
this
experiment
for
10
years?
That
shows
what
people
need.
Yes,
I.
A
C
G
C
G
The
benefit
is
like
taking
your
ID.
If
you
were
to
use
math
dot
random,
it's
not
actually
secure.
You
like
you,
need
to
use
a
secure
source
of
entropy.
So
there's
like
there's
moving
things
to
a
platform
level.
You
can
help
protect
people
from
implementation
mistakes
so
for
things
like
cryptography
and
unique
identification
and
and
things
where
they're,
pretty
simple
algorithms,
but
have
foot
guns
that
could
make
applications
that
are
not
as
good
I
think
they
can
be
potentially
good
candidates
for
Standardization.
G
It's
also
kind
of
frightening
to
just
pull
one
module
into
a
standard
module.
Namespace
like
I'd
like
to
understand
what
the
goal
is
and
because,
because
you
know,
everyone's
talking
about
they're
gonna
have
a
not
standard
namespace
and
we
use
it
for
putting
things
into
like
there's.
A
lot
of
open
questions
to
me.
So
I
should
have
brought
this
up.
I'll
bring
this
up
as
a
more
major
topic
in
the
next
movie
and
I'll
show
up
20
minutes
earlier
mm.
A
All
right,
oh
okay,
so
hey
Rebecca,
which
thing
exactly
was
broken
on
that
calendar
like.
A
Interesting
Rebecca,
where
you
logged
in
when
you
clicked
it,
it.
D
A
Yeah
possible
to
like
send
a
screengrab
sure,
okay,
okay,
cool
all
right,
so
I
think
that's
all
we
got.