►
From YouTube: Diagnostics WG meeting - August 28, 2019
Description
A
B
A
B
A
A
Proposal
you
inspect
stuff
like
Alexei
is
not
here,
so
I
would
expect
that
we
can't
really
discuss
it.
Yes,
last
update
was
in
June.
So
let's
move
to
the
next
one
proposal
to
drive
domestic
working
group
initiative
as
to
us
user
journeys.
This
is
still
blocked
on
the
survey,
which
is
mainly
blocked
on
me.
No
progress
here.
C
A
B
A
Mean
we
should
propose
new
meeting
times
and
I
guess
people
against
ought
to
vote,
but
to
be
honest,
it
was
only
like
two
three
people
who
who
kind
of
shared
interest
in
this
marine.
What
would
be
new
time
which,
which
is
great
because
then
we
would
double
the
size
of
the
people
yeah,
but
it's
still
not
perfect,
yep,
so
I
believe
in
a
longer
term.
Maybe
maybe
you
should
do
other
things,
but
yes,
we
should
start
with
the
meeting
time.
Yeah.
B
A
B
A
B
B
D
D
D
D
A
B
D
D
A
D
So
this
is
debugging.
Aid
is
a
is
a
small
project
where,
where
I
experiment,
with
tools
to
to
try
new
things,
I
could
do
to
help
our
diagnosis
potential
problems
with
an
application
running
on
the
server
I.
Guess
it's
mostly
useful
for
the
server,
but
might
also
work
for
the
IOT
use
cases
and
similar
and
I'm
gonna
go
through
the
ideas
here
and,
if
you're
interested
in
any,
we
can
explore
them
a
bit
more
I'm
just
going
to
demo
them
first
and
explain
what
they're
using
so
the
first
one.
D
We
need
to
move
these
new
controls
to
see
something
myself.
The
first
one
is
a
recent
thing:
I
tried
to
to
get
something
similar
to
a
network.
Log-In
node,
it's
a
very
tiny
that
when
you
run
your
code
it
logs
to
the
console
outgoing
requests,
it
tells
you
what
the
URL
was
and
gives
you
a
stack
trace,
hopefully
pointing
to
the
last
bits
of
code
that
were
active
in
order
to
send
the
request.
D
So
this
should
be
helpful
to
identify
if
the
request
is
going
out
at
all
or
if
something
unusual
is
happening
in
the
application,
where
some
module
that
you
didn't
expect
is
sending
out
HTTP
requests.
I
started
exploring
this
topic
looking
into
ways
to
detect
or
even
stop
modules
that
you
didn't
want
to
send
things
out,
I'm
sending
things
out,
and
this
is
effectively
just
giving
you
the
URL
and
nothing
else,
because
it
uses
a
man-in-the-middle
package
to
if,
on
anything
that
uses
sockets.
D
Okay,
so
HTTP
is
using
that
and
if
you're
opening
sockets
directly,
it's
also
using
that
okay,
yeah
okay,
so
I
would
be
interested
if
you
know
of
any
prior
art
anyone
trying
to
do
that
other
than
the
old
good
old,
node
inspect,
which
had
a
network
panel,
which
is
no
longer
possible.
As
far
as
I
know
now.
D
This
is
this
is
a
tool
that
I
think
I
already
mentioned
before
this
is
this
is
leveraging
a
library
that
detects
when
event
loop
is
blocked,
but
it
doesn't
do
it
the
old
funny
way
with
set
timeout,
but
instead
it
uses
asking
hoops
for
that.
So
it
can
also
give
you
a
stack
trace.
So
I
have
an
example
code
which
has
a
function
that
loops
over
some
strength
and
catenation,
and
it
detects
it
in
various
places.
D
First,
there's
the
function
logs,
a
metric
from
console
time
how
long
it
was
looping
and
then
this
is
the
outputs.
This
is
the
time
that
asking
cooks
my
words
and
a
stack
trace.
So
this
is
the
fact
that
we
giving
you
stack
traces
pointing
to
the
last
asynchronous
junk
before
the
loop
blocking
happen.
D
It's
been
tested
by
a
few
people,
so
there's
a
separate
repository
where
it's
called
blocked
at
and
this
one
it's
just
wrapping
it
into
something
you
can
require,
because
the
idea
is
that
you
run
nodes.
You
require
a
certain
tool
from
the
package,
run
your
app
and
there's
nothing
else
to
know
about
it.
You
just
get
additional
output
from
the
application,
so
I
don't
need
to
do
anything
specific
to
do
to
get
this
to
run
another
experiment.
D
Another
experiments
very
recent
is
is
using
probably
somewhat
of
a
hack
I
noticed
that
for
requiring
modules,
even
if
you
require
an
internal
module,
node
is
going
to
check
the
require
cash
for
it.
So
I'm
I'm,
adding
a
getter
and
setter
to
a
given
ID
in
the
require
cache,
and
that
lets
me
in
this
case,
give
you
a
stack
trace
to
every
location
where
the
net
module
was
required.
D
Okay,
so
yeah,
so
this
this
seems
also
to
be
useful
to
detect
when
a
package
wants
to
send
a
bail
request,
they
probably
need
to
require
net
or
get
it
as
a
transitional
require
somewhere.
So
you
get
the
full
stack
trace
and
you
find
out
about
all
the
locations
where
this
internal
modulates
use.
Okay,
this
one's
pretty
boring
this
just
wraps
around
the
hooks
for
promises,
but
this
is
a
tool
that
someone
can
install.
They
don't
have
to
know
what
to
do
in
code.
D
So
you
get
information
about
particular
functions
that,
where
callbacks
to
a
sent
chunks
or
promises
in
any
way
and
this
measures
the
time
and
displays
the
time
here,
I'm
working
on
getting
these,
be
something
undefined
to
be
the
function
name.
The
numbers
here
are
coming
from
the
async
context,
so
this
is
an
experiment
to
visualize
the
they
are
synchronous
flow
of
your
code
in
in
performance,
tracing
and
notice.
How
long
a
synchronous
execution
takes
on
a
graph?
D
So
this
is
somewhat
similar
to
the
cross
potentially
or
to
some
of
the
tools
from
near
form,
but
this
is
actually
three
lines
of
code
at
this
point:
I
guess
so:
I'm
I'm
checking
if
this
is
even
feasible
to
do
say
as
a
tiny
tool
so
yeah
these
are
my
experiments.
I
wanted
to
do.
I'd
get
feedback
on.
If
you,
if
you
think,
there's
room
for
that.
If
any
of
these
seems
interesting
to
introduce
to
either
audience
I'm
gonna
continue
to
look
for
what's
possible
with
new
stuff
yeah,
that's
wrong.
Sure.
B
A
D
The
information
that
would
normally
get
from
a
bigger
developer
tool-
and
all
you
need
to
do-
is
require
and
choose
the
right
tool.
So
this
would
be
lightweight
and
quick
way
to
get
some
information
out
of
your
app
and
potentially
these
these
experiments
could
be
useful
for
building
something
else
like
if
we,
if
we
don't
have
a
network
panel,
it
still
in
a
few
months
or
a
year.
D
A
D
Additional
logs
that
you
wouldn't
normally
get
without
putting
them
in
place
or
a
lot
of
instrumentation,
it
gives
you
context
information.
So
when
you're
running
an
app
that
supposed
to
be
supposed
to
be
making
some
requests
and
based
on
that
performing
in
the
connection,
you
either
have
to
patch
your
code
to
see
if
the
request
is
even
coming
out.
D
If
you
don't
add
enough
logging
at
the
right
time
or
he
can
use
one
of
the
tools
to
display
all
the
up
outgoing
traffic,
also
in
reviewing
your
dependencies
that
you've
added
to
your
app,
which
seems
reasonable
to
check
that
they're,
not
making
external
requests.
So
this
is
this
is
aimed
at
the
developers
in
the
in
the
development
stage,
not
a
tool
for
production
debugging.
Unless
you
know,
unless
you
want
to
disturb
your
application,
because
I
don't
think
async
cooks
are
recommended
to
run
in
production.
At
this
point
right,
yep.
D
D
A
D
A
Would
definitely
reach
out
to
Mattel
Calina
both
because
he
works
for
me
from
so
he's
working
on
the
bubble,
probe
and
all
the
others.
Since
what
you
mentioned
and
he's
also
member
of
the
TSC
and
the
note
contributor,
so
I
think
if
he
can
get
his
a
time,
then
their
faith
is
something
which
which
can
be
can
be
carried
over.