►
From YouTube: Node.js Diagnostics WG meeting - 2018-09-19
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
C
A
A
E
I
would
imagine
like
if
you
have
more
collaborators
available
in
person
like
what
other
topics
we
want
to
talk
about
with
the
broader
collaborative
base
and
what
topics
you
want
to
drive
and
sort
of
call
for
help
from
a
broader
collaborator
based
on
or
we
could
run
it
like.
A
just
another
meeting
like
we
have
here
with
the
same
set
of
suspects
that
you
open
this
meeting,
I
prefer
the
former
I.
E
Former
was
the
idea
that,
given
that
I
expect
there's
a
broader
collaborator
base
attending
at
the
collab
summit,
I
think
it
might
be
a
good
idea
to
give
an
overview
of
what
are
the
main
things
that
tastic
working
group
is
working
on
and
what
are
the
areas
where
we
need
help
from
the
Baroque
laboratories
and
so
basically
trying
to
get
more
help
from
the
rest
of
the
collaborators
on
some
of
the
bigger
issues
that
we
have.
That
needs
somebody
to
drive
those
things.
Oh.
A
A
A
Yeah
I
mean
I
think
at
a
minimum.
Maybe
we
could
just
sort
of
you
know
capture
kind
of
the
issues
that
we've
been
talking
about
here
and
then
I
think
her
oles
comment.
It's
like.
Is
there
anything
else
that
mean
what
he's
interested
in
and
and
then
opening
it
up
to
whoever's
there
that
wants
to
sort
of
talk
about
it
and
then
just
kind
of
pulse
through
these
things
see
if
we
can
generate
some
excitement,
perhaps
somebody
to
dive
in
and
help
out
all
the
stuff
like.
F
F
F
So
like
it
might
be
a
matter
of
saying
you
know,
we
could
prioritize
the
the
overview
and
sort
of
call
for
additional
help.
If
that
discussion
goes
long,
you
know
just
keep
doing
that,
but
if,
if
we
finish
that,
then
you
know
having
a
few
working
type
things
we
want
to
work
on
would
probably
also
be
good.
If
so,
you
know,
if
there's
things
like
I
think,
like
the
async
terminology,
it
probably
wouldn't
probably
would
benefit
from
some
FaceTime
right.
You.
A
A
A
So
I
don't
know
if
Mark
is
gonna,
be
attending
its
mark
on.
A
A
F
D
You
know
I
was
gonna,
say
that
I'm
Bennett
I'm
these
names
for
a
while
just
as
I
had
conflicts
but
I,
like
you
know,
I
didn't
talk
to
leave
separately
about
like
getting
up
to
speed
on
the
topic,
sudden
I
thought
here
then
I
couldn't
put
together
a
line.
Think.
D
A
D
F
G
F
D
D
A
A
F
I
I
think
we
want
to
two
weeks
from
today:
we'd
want
to
know
what
we're
doing
I
don't
think
it
needs
to
be
overly
formal
if
we
don't
want
to
make
it
right.
Yeah
like
this,
it's
a
it's
a
slot
for
the
group
to
get
together.
It
can
be
a
working
session.
It
can
be
a
presentation,
I
think
if
we
have
somebody
if
we
know
who's
going
to
be
there
that's
kind
of
important
and
then,
if
we
have
the
list
of
initiatives,
it's
a
matter
of
like
do.
F
We
have
somebody
who's
willing
to
stand
up
and
talk
about
that
initiatives,
and
we,
you
know
I
if
we
want
to
put
together
slides,
that's
great,
but
I
also
think
if
people
are
busy-
and
we
don't
want
to
do
that-
just
having
somebody
actually
get
up
and
try
and
communicate
what
it
is
and
motivate
people
to
get
involved.
This
would
work
too
and.
E
F
G
D
A
You
know
someone
else
start
to
kind
of
repeat
some
of
the
things
that
you
know.
We've
been
trying
to
communicate
regarding
async
contact,
so
I
think
if
you
want
to
do
it,
if
you
want
to
help
out
I
think
that's
great
I
do
have
a
DAC
I
need
to
do
a
few
tweaks
on
it.
It
tries
to
paint
some
pictures
of
things
that
I
think
would
be.
You
know
effective,
but
but
let's
see
how
things
shake
for
me
and
then
we'll
discuss
more
in
two
weeks
on
it.
Okay,.
A
Okay,
so
next
thing
here,
there's
this
best
practices
guide,
there's
yeah
again
at
the
risk
of
being
over
committed
here.
I,
don't
know
if
anybody
has
any
updates.
I
started
trying
to
do
some
stuff
around
memory,
leak,
analysis
and
I
hope
to
make
some
progress
there
over
the
next
week
or
so,
and
have
basically
a
forum
being
like
hey,
here's
kind
of
an
example
and
then
driving
through
the
example
and
how
to
use
chrome,
dev
tools
to
actually
arrive
at.
Oh.
This
is
my
leak.
A
F
A
A
A
F
B
B
B
F
Think
it's
the
only
one,
the
only
one
yeah
we
don't
have
anything
in
the
higher
tiers.
Yet
there's
at
least
one
that
we
probably
can
already
move
up
like
the
CPU
profiler,
some
some
flavor
of
that
which
I've
been
worked
on
right
and
fixed
I.
Think
there's
testing
in
core
for
that.
So
we
can
probably
move
that
up
but
yeah.
Currently
we
don't
have
anything,
that's
actually
higher
up
in
the
listing
and
then.
F
A
F
E
F
F
You
know
it's
as
if
possible
tests
would
be
run
at
least
nightly
a
note,
see
I
so
I
think
normally
I
wouldn't
I,
wouldn't
say
that
like
it's,
it's
got
to
the
point
where
we'd
normally
added,
but
if
it's
adding
coverage
for
stuff
that
we
do
want
to
cover-
and
we
don't
have
any
other
way
to
cover
it
right
now,
I
could
see
it
in
that
context.
So.
A
F
B
Like
how
do
we
differentiate
between
when
we
want
to
add
this
to
level
3,
just
because
oh
it's
nice
to
have
the
test
coverage,
yeah
and
and
and
how
do
and
in
other
cases
right,
because
we
have
other
means
of
insurance
as
coverage
we
have
sidqin,
which
is
like
yeah
exactly
is
for
this
purpose.
So
what?
How
could
this
be
able
to
system
instead?
Would
that
make
sense?
Why
I
think
this
list
is
mainly
for
things
that
I
guess,
of
course,
no
clinic
is
diagnostic,
related,
obviously
yeah.
B
F
Well,
I
think
in
terms
of
that
overlap,
I
am
NOT
concerned,
because
this
this
could
justify
why
it's
part
of
Scituate
right
like
if
we
decide
we
want
testing
to
cover
these
things.
I,
don't
think
it
matters
like
it.
This
is
not
specific
at
all
in
terms
of
how
the
test
coverage
is
added
in
Siddim
could
be
part
of
that
coverage.
B
F
If
sitchin
is
the
best
way
to
test
it,
then
adding
it
to
decision
make
sense
or
if
adding
some
core
tests
or
adding
a
separate
job
makes
sense
that
the
the
the
bigger
concern
is
like
in
some
cases,
and
we,
you
know
we're
still
I-
guess
struggling
a
little
bit
as
like
in
some
case
we're
trying
to
make
sure
an
API
still
works.
So
I.
Don't
think
anybody
would
argue
that
having
it
some
testing
that
makes
sure
that
the
the
API
that
generates
the
data
for
the
flame
graphs.
B
F
B
F
A
Hey
I
got
there's
some
concern
here.
The
clinic
is
too
broad.
It
touches
too
many
things.
We're
a
smaller,
more
targeted
test
would
be
better
scope
here,
there's
some
existing
flame
graph
tool
that
can
be
added
to
the
matrix
that
would
be
more
targeted
and
prevent
prevent
clinics
flame
graphs
from
breaking.
Is
it
appropriate
to
test
this
V
acid
Jim,
and
can
we
add
tests
in
an
API
level
to
ensure
that
flame
graph
api's
are
working
correctly?
That
is
a
test.
Aren't
validating
clinic
itself,
but
rather
a
testing
API
is
clinic
depends
on
is.
F
That
a
fair,
summary,
yeah
and
I
guess
maybe
just
an
additional
comment,
would
be
that,
like
those
smaller
like
the
API
tests
would
have
a
better
chance
of
moving
up
to
even
higher
levels
right
like
if
we
had
a
if
we
had
an
API
test
to
test
the
flame
graph,
the
the
API
for
that
that
the
flame
graphs
need
or
the
data
generation
that
I
could
see
pushing
up,
trying
to
push
up
to
say.
Well.
F
G
B
E
B
B
H
F
E
So
the
complexity
here
is
that
there's
three
distinct
ways
you
can
build
flame
graphs
on
node
like
one
is
to
write
a
native
module
that
uses
the
C
purpose.
C
pod
API,
the
other
one,
is
to
use
the
inspector
protocol
to
get
a
CPU
profile
from
there.
The
third
one
is
trace
events.
These
events,
you
can
also
get
the
CPU
profiler
tumid
group,
sampled,
on
the
treatments
general
right,
but.
E
Yeah
perfect
purpose
and
entirely
different
from
the
CP
voice
profiler.
But
what
I
was
getting
at
is
that
I
think
when
we
move
something
to
a
particular
tier
I
think
it's
we
have
to
be
clear
which
yeah
I
will
be
trying
to
harden.
Thank
you
three
distinct
ways
of
doing
it
and
and
resonances
experimental
inspector
has
a
different
use
case
than
an
eerie
module
right
so
and
and
I
don't
think.
Just
adding
no
tunic
is
is
targeted
enough
because
they
could
change
implementation
underneath
the
covers,
and
now
we're
maybe
testing
a
different.
E
If
yeah,
then
we
originally
were
testing
so
I.
Think
that
that's
where
I
was
coming
from
from
the
point
of
view
of
when
you
do
testing
I
think
it's
better
to
start
with
us
with
a
simpler
test,
rather
than
a
big
application
as
the
first
one
that
you
have,
because
it
just
has
more
complexity.
It
has
more
moving
parts
and
if
something
breaks,
it's
harder
to
figure
out
why
it
broke
yeah.
A
E
D
A
Sure,
but
I
thought
part
of
this
was
to
say
well
what
are
the
set
of
tools
that
we
need
to
say
you
know
we're
continuing
to
work
and
not
getting
broken
on
each
release
and
then
the
question
is:
how
does
clinics
fit
in
there
as
it
right?
So?
In
that
sense,
it
seems
valid.
I
mean
I
totally
buy
your
point
that
there
should
be
API
tests
are
validating
the
eyes
and
clinical
dependency
on
on
and
I
total
agreement.
A
That's
easier
than
sort
of
this
broader
test
of
like
does
clinic
work,
but
I,
don't
know
that
those
are
necessarily
like
those
are
mutually
exclusive
things
right,
but
that's
fine
I've
got
a
comment
on
here
on
this
specific
issue
that
I'm
gonna,
post
and
I.
Think
if
you
guys
disagree
or
I
didn't
capture
the
discussion
correctly,
then
you
know
chime
in
and
and
and
tell
me
what
was
what
was
wrong
about
it
and
it'll
be
interesting,
I
think
what
we'll
have
to
say,
yeah.
F
A
H
I
think
the
main
part
of
the
PR
is
to
use
resources
side
by
side
with
AIESEC
IDs
and
some
assing
operations
recycled
the
resources
so
that
they
can
be
shared,
which
would
mean
that
they
would
have
different
I,
think
IDs,
but
the
same
resource
which
doesn't
make
it
work.
Well.
I'm
wondering
whether
this
should
like
the
scope
of
this
should
be
widened
so
that
we
should
explore
where
they're,
not
recycling
the
resource
is
something
feasible.
I.
H
C
It's
it's
pretty
much
entirely,
just
due
to
HTTP
parser
reuses
the
handle
and
no
one
actually
likes
the
HTTP
parser
and
there's
been
lots
of
discussions
of
replacing
it
was
like
a
JavaScript
implementation
or
something
so
I
think.
Eventually,
we
just
need
to
do
something
about
it
like
at
the
very
least,
we
could
just
wrap
the
HT
HTTP
parser
in
something
so
like
each
use
of
it
gets
its
own
handle
I'm
thinking
somehow,
even
though
like
under
the
covers,
it's
really
nothing.
H
So
so,
from
our
perspective,
this
is
actually
one
of
the
things
that
could
still
improve
performance.
By
a
lot
we
have
been
an
especially
Benedict
has
been
trying
to
squeeze
blood
from
stone.
Essentially,
we
have
also
been
approaching
to
z29
to
change
this
back
in
a
way
they're
just
one
cycle,
we,
we
use
I,
think
a
wage
and
we've
been
looking
at
emitting.
Temporary
promises
from
triggering
promise
works,
but
I
think
the
biggest
gain
that
you
can
get
is
from
just
PRN
that
haven't.
You
has.
E
C
E
E
And
I
could
I
a
key
was
going
to
talk
about
the
same
topic.
Do
so
I
was
talking
to
Matteo,
because
the
returns
of
Matteo
have
not
been
making
this
meeting,
because
this
timeslot
just
doesn't
work.
So
if
you
were,
if
we
are
going
to
have
a
discussion
on
this,
this
meeting
time,
fraud
doesn't
work.
So
what
I
am
going
to
propose
is
that
I
already
talked
to
young
and
Matteo
about
this?
E
Is
that
I
send
out
a
doodle
for
a
one-off
meeting,
specifically
on
this
topic
and
I
propose
we
do
it
next
week,
because
we
don't
have
a
diagnostic
work.
You
know
maybe
next
week
so
I'll
send
out
a
doodle
and
I
can
open
an
issue
and
be
that
you
can
get
the
right
participants
involved
in
the
discussion
so
that
we
can
actually
extract
it
again.
My
complaint
is
they.
H
E
F
C
F
A
Last
comment
was
that
you
know
with
the
approach
of
whatever
daylight
savings
time
or
whatever
the
it's
call
right
is
gonna
fix.
That,
for
him
was
what
I
saw
it,
but
if
there's
other
people
that
want
to
attend
that
would
like
it
and
it's
like
I
kind
of
feel
like
we
do
this
every
three
or
four
months,
and
we.
A
F
It's
still
worth
revalidating
every
so
often
if
it's
the
right
time,
the
only
other
suggestion
is
like
an
altar
day
time
slot,
so
that
you
know
one
time
it's
one
hour,
the
next
time
it's
a
different
hour
and
that
way
it
maximizes.
You
know
you
don't
get
everybody
every
time,
but
more
people
have
a
chance.
There's.
G
B
G
B
C
C
A
A
E
I
can
give
status
so
I
have
been
investigating
some
of
the
thread.
Races
that
else
and
I
think
I
have
some
of
the
fixes
out
now
and
the
idea
is
to
get
I
mean
to
fix
some
of
these.
The
hugheses
I've
been
playing
increase
event
and
making
it
like
crash
intermittently
on
shutdown
so
that
we
can
have
something
that's
reliable
enough
to
be
used
by
like
node
clinic
and
other
things.
E
There
is
another
project
that
chrome
is
looking
at,
that
the
v18
will
also
be
exploring
I
I
understand
called
perfetto.
That
I
talked
about
I
think
last
time
around
and
the
idea
would
be
prepared.
Oh,
is
a
is
a
library
that
basically
extracts
tracing
code
into
a
standalone
thing
that
both
Android
and
chrome
are
switching
to,
so
that
will
get
rid
of
some
of
the.
So
the
complex
complexity
of
how
we
hooked
up,
v8
and
node
at
the
moment
with
tradesmen-
and
my
hope
is
that
this
change
can
be
made
without
any
so.
E
Yeah,
so
that
the
status
at
the
moment
and
and
I
actually
I
would
really
get
there's
some
things
that
can
be
done
with
tracer
meant
and
new
training
is
one
example
and
there's
another
one.
That
I
was
working
on
the
other
day
with
a
colleague
of
mine,
which
was
just
trying
to
understand
file
operations,
IO
operations
I
think
there's
a
good
set
of
tools
that
could
be
built
on
top
of
it.
So
it'd
be
nice
that
if
we
at
some
point
may
get
some
demos
off
what
all
is
possible
with
race
events
at
some
venues.
A
E
It's
a
standalone
library,
as
opposed
to
so
right
now.
The
way
we
have
hooked
up,
v8
and
node
is
that
we,
it
exposes
some
super
classes
with
some
way
interface.
An
API
contract
implements
and
override
some
of
the
sub
classes,
but
the
kree
Aughra
fee
of
when
things
are
supposed
to
happen
and
what
are
the
constraints
on
on
certain
operations
happening
was
not
clear,
and
this
is
what
what
our
code
suffered
and
that
was
causing
all
of
these
flaky
test
failures.
E
E
A
This
issue,
84,
is
just
it's
kind
of
got
a
little
bit
raw
I,
don't
know
how
much
of
these
check
boxes
are
still
valid.
I
think
some
of
these
things
are
done
and
they
can
be
checked
off
and
some
of
them
me
yes,
there's
probably
some
new
ones
and
there's
probably
some
things
that
can
be
removed
right,
I,.