►
From YouTube: Node.js Tooling Group Meeting
Description
A
A
So
if
you're
not
familiar
so
huge
chmod
is
the
thing
to
change
permissions
on
unlike
a
POSIX
operating
system,
and-
and
maybe
there
is
some
true
there's
some
sort
of
equivalent
windows
but
I-
don't
know
anything
about
it,
and
so
there
is
like
a
chi
H
mode
in
node
and
it
allows
you
to
change
the
permissions
of
a
single
file,
but
a
lot
of
the
times.
I
find
myself.
You
know
in
the
shell
using
this
thing
and
I'm
like
well
I
want
to
change
ownership
of
this
whole
directory.
A
All
the
way
down
and
or
I
want
to
change,
not
yeah
I
mean
there's
the
the
permissions
here
or
ownership
I.
Guess
that
would
be
CH,
ger,
/
I,
don't
know
all
sorts
of
different
shown,
I
don't
know,
but
wouldn't
it
be
great
if
we
could
do
that.
Recursively,
like
we
do
make
do
recursively
like
we
do
remarkably
and
Isaac
from
and
Kim
mentioned.
This
would
be
cool
I.
Think
in
the
slack.
So
would
it
be
cool?
What
do
you?
What
do
we
think
yeah.
A
C
A
C
A
C
The
prom
with
frame
Ref,
it
wasn't
faster,
but
I,
think
we
could
have
gotten
it
faster
if
we
figured
out
the
appropriate
abstractions
like
the
because
we
you
want
to
be
doing
well,
one
rim
ref,
terrible
because
of
all
the
hacks
for
windows.
So
let's
ignore
it,
but
imagine
that
you
just
had
to
recursively
chmod
a
bunch
of
stuff
which,
like
in
the
simple
case
and
you
just
trust,
CH
mods
working,
there's,
not
a
million
hacks.
C
A
C
D
Would
just
say
that
unless
there's
like
a
tangible
benefit
for
it
to
be
in
C++,
then
I'd
rather
see
as
much
as
possible
B
JavaScript,
because
putting
the
code
and
c++
kind
of
hides
it
away
from
a
lot
of
developers.
I
mean
like,
for
example,
I'm
I
know,
C
I'm,
really
not
good
with
C++.
So
if
it
were
C++,
then.
A
You
saying
the
benefit
to
writing
in
C++.
If
it's,
you
know
if
he
uses
threads
properly
and
it
does
work
in
parallel.
The
benefit
is
performance.
It
will
be
faster
than
if
we
do
it
in
JavaScript.
But
yes,
somebody's
got
to
write
that
and
maintain
it
and
personally
I'm
I,
just
I.
Don't
have
the
skill
in
C++
to
do
that.
A
C
We
could
have
a
we
could
advocate
in
that
these
pull
requests
like
first
we'll
figure
out
a
semantic
structure.
What
what
folder
they
go
in?
Where
do
they
sit
next
to
the
file
system
helpers,
and
then
we
could
have
the
argument
if,
if
there's
a
compelling
reason
like
someone
needs
to
someone
needs
the
performance,
we
can
consider
moving
it
to
C++
at
a
later
date
like,
and
we
can
do
that.
C
B
C
Doesn't
block
us
from
doing
it
eventually
right
like
as
long
as
we
figure
out
the
API
surface,
then
someone
who's
like
I,
don't
know
like
maybe
NPM
needs
to
chmod
20,000
things
on
some
systems,
so
there's
a
benefit
to
moving
into
C++.
Let's
wait
until
we
have
that
user,
most
people,
our
price
catch,
modding,
eight
things.
B
A
So
Ben,
so
you
not
only
need
to
like
use
this
thread
pool
like
when
you
call
C
chmod,
and
so
you
run
all
this
work
in
parallel
and
sepal
plus,
you
need
to
implement
the
chmod
rsync,
which
does
the
thread
work
but
pretends
that
it's
synchronous,
which
is
exact,
sync
hell
and
is
also
yeah.
You'll,
have
fun,
go
ahead
and.
C
A
B
C
A
A
D
A
A
D
C
Would
be
willing
to
put
a
hand
up
potentially
because
I
I'm,
the
one
who
inflicted
the
one
off
approach
on
the
world,
so
it'd
be
kind
of
nice
to
make
it
more
generic.
Oh
wait!
No
I'm
on
the
wrong
one!
I'm
talking
about
variables
existing
down
to
the
lower
level,
not
detecting
exit
I'm,
not
putting
my
hand
up
for
detecting
exit
right.
A
A
And
so
I
mean
I
think
that's
a
good
place
to
start,
and
you
know
I
think
at
this
point.
It
probably
just
comes
down
to
a
pull
request.
So
you
know
if
we
have
this,
this
list
of
events
that
are
or
events
in
the
in
the
broader
sense
that
the
need
to
trigger
this
event
emitter
event.
So
we
have
that
so
I.
Think
next
step
is
probably
a
PR.
A
A
For
him
to
further
probably
explain
how
things
are
going
to
be
difficult,
but
yeah,
so
if
anybody
does
need
anything
from
Sam
and
is
having
a
difficult
time
reaching
him,
then
I
can
try
to
reach
them
as
well.
Okay,
so
next
thing,
then,
is
the
source
map
support
and
then
I
assume
wrote
here
haven't
figured
out
who
to
talk
to
you
about
a
blog
post.
Yet.
E
C
D
C
Yeah
I'm
working
on
that
I,
don't
think,
there's
any
reason
to
upgrade
to
the
chromium.
Libraries
I
think
it's
more
that
they're
two
different
genus,
then
that
week,
that
one's
better
than
the
other
one
there's
there's
one
fix,
I'd
like
to
pull
in
from
chromium,
which
is
sorting
the
order
of
statements
in
a
source
map,
but
it's
kind
of
a
silly
nice
to
have,
because
you
have
to
go
to
your
way
to
not
have
them
sorted
basically,
because
the
waste
source
Maps
work
is
there
additive?
C
So
if
you
have
say
you
have
line
250
and
there
are
four
statements
on
it.
Each
statement
adds
on
the
offset
of
the
prior
statement.
While
it's
calculating
the
source
map,
that's
how
it
impacts
it,
so
you
would
have
to
actively
do
something
silly
like
making
it.
So
the
next
statement
is
negative
with
respect
to
the
prior
statement,
to
get
it
out
of
order.
So
you'd
have
to
be
doing
like
negative
offsets
to
kind
of
force
yourself
into
a
situation
where
source
maps
are
unsorted,
but
chromium
does
sorted.
That's
the
only
difference.
C
A
C
A
C
A
C
C
I,
like
the
idea
of
hacking
on
it,
I
didn't
set
up
a
meeting.
Cuz
I
got
too
busy.
Ike
had
coffee
with
Isaac
and
he
hates
the
idea.
So
that's
a
window
to
my
sales
a
little
bit
but
I
mean
Isaac,
doesn't
I
respect,
Isaac's
opinion,
but
it's
just
one
opinion,
so
you
could
still
hack
a
little
bit.
A
E
Good
though
cuz
I
don't
know,
I've
looked
at
a
bunch
of
go
CLI
programs
that
are
using
their
thing
and
I
don't
really
like
it.
So
maybe
I
look
at
the
go.
You
know
CLI
interfaces
that
people
have
built
in
them,
I
see
them
as
lacking,
but
they
don't
have
like
a
lot
of
community
solutions
with
better
ones.
Cuz
like
everybody's
like.
Why
would
you
do
that?
E
E
B
A
C
A
A
C
Isaac
Isaac
released
a
want
version
1.0
of
McGirt
the
other
week
and
in
doing
so
pointed
out
a
bunch
of
deltas
between
Victor
and
our
implementation
of
mcdr
recursive,
one
of
the
main
ones
being
that
Victor
rhetoric
derp
returns
the
first
directory
created
or
not
or
no
directory.
If
no
directories
were
created
so
that
you
can
kind
of
work
backwards
and
figure
out
what
exactly
it
did
for
you.
C
A
C
A
A
C
C
A
Anybody
should
research
why
you
can't
pass
owner
group
to
all
of
these
FS
functions.
Why
not?
Because
it
would
seem
to
me
it
just
that
it
would
be
nice
if
you
could
do
that
and
I
think
other
things.
Let
you
do
that
like
I,
don't
know,
maybe
I'm
thinking
a
Python
has
something
like
this.
Where
you
you
talk.
C
C
A
C
C
D
If,
if
the
maker
could
change
the
owners,
then
it
would
do
that
as
it
creates
each
directory
when
it's
doing
it
recursively,
whereas
the
way
would
work
right.
Now
you
get
the
first
directory
that
was
created
and
then
you
recursively
change
the
owner,
so
you're
iterating
the
directory
tree
again,
not
that
that's
gonna
be
a
huge
performance
thing.
E
But
I
think
the
argument
of
making
the
user
not
have
to
do.
It
is
really
good
because
it
avoids
the
very
common
user
error
of
doing
the
two
operations,
one,
the
second
one
failing
and
not
having
to
you
know
not
correctly
cleaning
up
so
hiding
that
little
bit
of
complexity
inside
of
the
you
know,
core
implementation
would
be
I,
think
a
pretty
big
win
for
the
user
experience
I.
A
Think
there
mate
there
might
be
a
question
of
okay.
If
you
do
this
and
something
fails
like
how
should
it
fail,
like
I
feel
like
what
were
we
talking
about
before,
where
it
were
it's
where
you
run
into
this
problem
with
well,
how
exactly
should
what
should
the
behavior
be
upon
failure,
like
do
you,
okay,
now
I've
just
made
all
these
directories,
but
but
changing
the
owner
failed,
and
so
that's
just
what
you
have
now
remove.
E
A
A
A
C
D
A
A
D
C
Yeah
and
the
Devils
in
the
details
too,
like
we
don't
have
when
I
was
fixing
some
bugs
around
user
permissions
for
Victor.
We
have
like
two
tests
and
all
of
nodejs
for
permissions
because
so
hard
to
write
tests
for
Windows
permissions
as
part
of
it.
You
need
to
use
some
ACL
program
and
it's
just
completely
different.
So
it
seems
like
it's
not
a
popular
thing,
to
build
features
out
around
or
write
tests
for
those
features.
If
you
are
building
them,
it's
like
two
tests
were
shown
in
the
whole
codebase.
A
Like
just
not
even
in
JavaScript,
let's
do
it
from
the
shell,
because
I've
got
like
a
meet
like
a
media
server
with
a
bunch
of
crap
in
it
and
I
want
to
I,
want
to
make
sure
like
the
right
user,
for
the
right
process
has
access
to
this,
that
it's
a
thing
that
you
do.
But
what
are
the
real
use
cases
provisioning
and
then
maybe
package
manager
or
something
right,
I,
don't
even
know.
C
C
A
E
E
C
That's
a
good
question:
I
think
we've,
the
ones
that
tend
to
stay
on
the
agenda
have
been
ones
that
were
more
personally
involved
in
I.
Think
and
I
think
that
that's
not
to
say
that
all
the
issues
aren't
equally
important
is
just
like
you
say
it
would
be
good
to.
If
anyone
wants
to
take
ownership
of
any
of
the
issues
we
can
make
sure
they're
predominantly
talked
about
in
any
of
these
meetings.
It's
I
don't
know.
I'm
personally,
like
super
excited
about
the
packaging,
for
instance,
but
don't
know
much
about
that
space.
A
Yeah
I
think
that's
it
really.
I
mean
and
and
once
things
have
this
this
label
it
they
tend
to
like
there's
an
inertia
and
they
tend
to
retain
the
label.
But
yeah
I
mean
if
there's,
if
there's
something
that
you
you
want
to
talk
about,
it
like
distributing,
see
like
tools
and
self-contained
binaries,
throw
a
label
on
it.
We'll
talk
about
it,
but
you
know
I,
think
those
things
really
need
more
of
a
champion.
A
E
I'm
not
ready
to
step
up
for
that,
but
I
will
be
I
intend
to
ship,
something
which
will
be
a
self-contained
binary,
but
I'm
probably
going
to
do
with
PKG
or
one
of
the
you
know
other
open.
You
know
non
core
that
for
that,
but
it'd
be
interesting.
If
that
was
something
moving
forward
and
and
like
I
said
I'm
not
ready
to
champion
it,
but
I
just
wanted
to
know
where,
where
those
other
ones
were
super
City
Club
support
as
well.
A
Be
good,
yeah
and
I
mean
that
wait.
What
oh
so
no.
He
did
comment
and
said
something
but
hood.
Oh
he
wanted
to
get
that
done,
which
is
funny
if
he
thinks
arguments
are
a
bad
idea
but
sure,
and
if
he
wants
join
the
meeting
and
and
talk
about
it,
that's
fine
too,
but
like
until
then
it's
probably
not
going
to
get
the
label
either,
but
I
don't
know.
A
E
Fine
I
didn't
need
it
to
have
the
label
I.
Just
don't
wanted
to
know
if
there's
a
reason
why
those
ones
which
also
don't
seem
like
they're
getting
active
updates
in
the
couple
last
few
weeks
didn't
get
it
yet.
A
few
that
also
aren't
getting
updates
the
most
couple
weeks
did
get
it
just
to
make
sure
I
understood
the
process
the
team
had
been
using
for
that
yeah.
A
A
All
right
yeah
does
anybody
got
anything
they
want
to
bring
up
or
Show
and
Tell
or
whatever
my
show
and
tell
I
suppose
I
think
I
can
do.
It
is
I
spent
like
a
long
time
writing
documentation
for
port
toolkit,
which
is
this
tool
that
works
with
diagnostic
reports
and
so
I
like
I
wrote
a
new
piece
of
documentation
and
it
took
way
longer
than
I
should
have,
and
so
that
is
what
I
have
to
show
off
and
unless
you
use
it,
you're
probably
not
going
to
be
interested
in
it.
A
But
that's
what
I've
been
up
to
so
like
I
went
and
I
meticulously
documented
each
and
every
command
line
flag
and
the
ways
they
interact
and
all
sorts
of
things
and
so
and
I
used
and,
of
course,
the
the
document
site
document
site
is
built
with
Gatsby
and
so
I
use
lots
of
NM,
DX
and
essentially
I.
Think
I
should
have
use
11
D,
but
Here
I
am,
and
so
that's
that's
what
I've
been
up
to
and
we
also
anything
they
want
to
show
I.
C
A
A
C
A
You
can
do
it
that
way
there
you,
you
do
okay,
so
it's
experimental
still,
so
you
still
have
to
use
the
flag.
You
can
tell
it
to
automatically
create
one
of
these
things
on
like
an
uncaught
exception
or
for
whatever
reason,
but
there's
several
different
ways
you
can
trigger
it.
You
can
just
do
it
programmatically,
if
you
want
to
you,
can
have
it
respond
to
a
signal.
A
E
E
A
But
yeah
the
big
one
is
post-mortem
debugging,
you
know
there's
if
you
want
to
have
it
trigger
some
other
way.
You
can
get
one
of
these
reports
from
a
running
process.
It
doesn't
have
to
die
to
create
a
report
if
you
can't
like,
for
whatever
reason,
attach
a
debugger
to
your
script.
You
know
running
somewhere
yeah,
so
you
can.
You
can
use
a
diagnostic
report
instead.
E
E
A
And
I'd
like
to
get
this
out
of
experimental
but
yeah
like
right
now,
for
example,
it
there's
no
option
to
generate
a
report
upon
unhandled
exception
or
no
untitled,
rejection,
sorry,
but
yeah
so
I
don't
know.
If
that's
something
that
needs
to
be
done,
because
you
can
give
it
another
flag
to
cause
an
unhandled
except
or
you
know,
you
know
what
I
mean
you
can
give
the
this
other
flag
to
action.
We
quit
instead
of
doing
what
it
normally
does.
A
I
don't
know
if
it's
needed,
but
and
I
don't
really
know
I'm
not
too
too
involved
with
the
Diagnostics
group.
The
working
group
that
works
on
this
stuff,
maybe
I,
should
be
but
yeah
I'd
like
to
see
it
get
out
of
experimental.
It
works
I'm,
not
sure
why
I
think
we're
just
we
want
to.
We
need
more
people
to
try
it.
It's
just
one
of
those
things
like
we
need
more
people
to
try
it
and
see.
A
If
there's
anything
it's
missing,
there
is
anything
that
it
could
be
doing
better
because
otherwise,
like
the
underlying
thing
that
it
does
it
does
this
very
well
and
it
does
it,
you
know
pretty
much
in
a
bug-free
way,
but
it
you
know
what
exactly
how
that
API
is
shaped.
You
know,
maybe
there's
other
other
use
cases
for
this,
that
we're
not
thinking
of
and
those
aren't
gonna
come
up
and
until
people
actually
try
it
so
I'm.
E
A
I
mean
I:
I
did
actually
show
this
to
the
diagnostic
group.
I've
been
in
like
a
couple
meetings,
and
so,
but
I
just
haven't
been
able
to
kind
of
make
them
and
yeah,
but
I'm
sure
he
knows
of
this
I.
Just
I,
don't
know
if
he's
had
a
chance
to
use
it,
but
they'll
be.
That
would
be
good
to
get
feedback
from
him.
Yeah.