►
From YouTube: Diagnostics WG - March 2017
Description
A
A
A
A
A
B
D
E
A
A
A
Okay,
welcome
everyone.
We're
already
live
on
YouTube,
sorry
for
a
bit
of
confusion,
getting
the
started.
I'll
I'll
figure
that
out
for
next
time,
so
we're
ready
to
start.
This
is
the
Diagnostics
working
group
meeting
for
March
2017
go
ahead
and
put
your
name
in
the
attendance
list
if
you
can
get
into
the
document.
A
Let's
today's
agenda
is
on
our
notes.
So
let's
just
look
quickly
through
the
previous
meeting.
We
who
programmatically
exposed
evade
inspector
URL
natural
we
concluded
on
that,
but
that's
relevant
to
what
that
the
final
text
will
be
I'll,
make
debug
an
alias
for
inspect
I.
Think
that's
on
our
agenda
again
today
and
basically
it's
going
to
be
half
hour
discussion
I
have
a
feeling,
I'm
switching
the
CLI
debugger,
which
I
believe
yeah
with
no
to
inspect.
A
A
Ace,
or
at
least
we
don't
need
to
worry
about
upstream,
although
I
think
there
was
an
issue
a
couple
days
ago
that
about
that
async
hoax
initial
implementation.
If
we
haven't
think
discuss
about
that
again
today,
we
can
yarm
joined
us,
and
so
did
a
note
or
sorry.
I
forgot
your
full
name,
and
then
we
have
the
two
tracking
issues
for
acing
hooks
and
trace,
and
then
finally,
we
had
a
guide
on
debugging
and
a
blog
on
what's
been
happening
with
the
diag
Ouija
that
went
out
a
few
days
after
our
last
meeting.
A
C
A
Actually,
what
I
was
thinking
is
that
this
should
depend
on
whether
it's
5.8
or
5.7,
meaning,
if
the
word,
if
we
land
5.7,
then
we
should
not
a
lius.
Yet
we
might
as
well.
Let
people
use
the
old
one
if
it's
still
there.
If
we
go
5.8,
then
the
seams,
it
seems
reasonable.
At
least
I
have
no
objections.
I,
don't
know
if
any
up
and
when
else's
opinions.
A
E
A
E
C
So
it's
my
so
feeling
here
is
that
I
think
it'd
be
good
to
use
the
supported
yet
I
debugger,
rather
than
an
unsupported
one.
In
a
eight
point,
X
ltf
blank
so
say:
I
thought
that
was
the
motivation.
I
I,
don't
have
dominguez
key
strong
feelings,
however,
so
thud
be
willing
to
give
a
of
other
people
have
strong
opinions.
Aight.
A
Dep,
I
have
the
same
concerns
yawn,
that
if
we
that's
gonna
be
hard
for
us
to
maintain
I
hear
that
is
anybody
concerned
about
pushed
back
from
the
ecosystem.
I
mean
I'm
I
from
Microsoft's
point
of
view
were
we're
fine,
we're
fine
with
whatever
the
community
wants
to
do.
Are
we
concerned
about
people
in
the
ecosystem
that
are
depending
on
the
old
one?
I
mean
own
inspector?
Is
the
only
one
I
can
think
of.
C
Yeah
I,
don't
inspectors
really
yeah
I,
think
I,
don't
think
that
has
tremendous
amount
of
use.
Any
more
people
have
moved
on
to
alternatives
such
as
dev
tools
or
weapons,
webstorm
or
wheel,
Studio
code.
All
of
these
tools
are
working
well,
so
I
don't
know
that
the
community
really
is
dependent
or
not
in
the
better
anymore
I.
E
Mean
we
did
give
give
plenty
of
warnings
and
I
mean
it's
not
like.
The
no
inspector
people
aren't
aware
of
this
change,
like
I
think
that
they
are
painfully
aware
that
this
is
coming
and
I
think
points
me
yeah
and
six
months.
We
is
already
out
for
a
while
and
if
they
really
wanted
to
like
to
make
the
additional
features.
Work
like
HTTP
request,
inspection,
I,
think
that's
one
of
the
few
things
that
has
left
it
is
not
in
not
supported
by
definite
inspect,
but
it's
the
heart
of
an
inspector.
D
A
A
There
anything
too,
so
let
me
put
it.
Let
me
just
slightly
change
that
we
know
we
don't
want
to
surface
the
old
interface,
even
if
it's
5.7,
because
we
don't
have
to
support
it.
Is
there
any
concern
about
the
alias
breaking
something
meaning
if
somebody
it's
a
different,
slightly
different
behavior
with
inspect,
would
it
be
better
to
just
remove
it
entirely.
E
I'm
kinda
depends
on
where
you
want
to
end
up
in
the
end,
like
I
would
be
fine
with
just
changing
the
name
inconsistently
to
be
dash.
It
inspect,
we
added
a
sure
to
inspect
break
already.
So
there's
this
precedent
for
just
renaming
the
thing
right.
So
we
we've
put
a
lot
of
work
in
making
the
replacement
from
off
the
park
with
inspect
work.
E
So
I
think
that
just
removing
it
would
definitely
be
a
tempting
alternative,
because
it
would
just
remove
any
kind
of
confusion
that
could
arise.
On
the
other
hand,
if
someone
is
using
the
command
line,
debug
up
to
just
get
back
random
script,
it
would
force
them
to
be
aware
of
something
that
they
don't
need
to
be
well,
so
I'm
torn
yeah.
C
It's
basically
I'm.
Definitely
when
you
can
together
we
should
remove
debug.
But
if
somebody
has
like
a
lot
of
shell
scripts
like
a
term
that
basically
do
this
at
least
we're
giving
them
one
more
major
psych
eval
mornings
before
we
remove
it.
But
having
said
that,
I
think
it's
I,
don't
think
it's
going
to
be
a
substantial
amount
of
change
for
anybody
to
do
in
order
to
react
to
this
so
so
I'm
also
con
I
I
definitely
am
a
gen
tempted
to
say
that
we
should
move
debug.
D
C
A
E
E
So
I'm
not
sure
what
we
can
discuss
specifically
about
this
issue.
It's
I
think
one
of
the
things
that
we
shoot
discussing
to
be
the
12k
comic
came
up
already
earlier
is
which
parts
do
we
want
to
deprecate
/
remove
in
8.0,
and
when
do
we
lent
them,
because
we
have
I
think
one
or
two
weeks
left
to
land
any
breaking
changes,
which
means
that
if
we
want
to
remove
something,
we
need
to
do
it
fast.
E
E
C
E
Yeah,
so
I
would
I
wouldn't
like
to
alias
dash
dash
debug,
because
desert
debug
is
really
about
a
specific
protocol
and
aliasing
aliasing.
That
protocol
seems
to
break
more
than
fixes.
At
least
that
would
be
my
expectation
see.
Lacrimal
is
a
little
different
because
it
wraps
its
own
protocol
client
inside
of
the
thing
it's
ponch
right,
so
it
owns
both
sides,
so
aliasing
dubaku
man
is
not
quite
as
bad
but
aliasing
directly,
but
seems
fragile.
C
Yeah,
I
agree,
I
think,
that's
a
pretty
sound
argument,
but
it's
an
independent
question.
So,
if
it's
not
so
you
turn
a
bad
question
whether
it
should
be
removed.
In
eight
point,
oh-
and
I
think
the
argument
from
josh
is
that
actually
there
is
two
such
as
visual
studio
that
are
actually
dependent
on
it
that
still
haven't
migrated
over
to
the
new
protocol.
They.
C
E
E
A
C
C
A
E
A
A
A
C
That's
an
entirely
different
thing,
so
so
the
protocol
is
a
decent
based
protocol
that
has
been
duplicated
for
a
long
time
and
that
the
thing
that
got
got
removed
in
5.8,
the
debug
context
is,
is
a
is
an
entirely
different.
Certified
java
script
and
C++
API
in
v8,
and
that's
going
to
be
around
till
the
end
of
the
year.
A
A
A
C
Yes,
let
me
think
about
the
timing
of
the
deprecation,
so
if
it
goes
away
at
the
end
of
the
year,
no
10
would
be
the
first
one
that
removes
it.
So
I
guess
no
date
should
issue
a
deprecation
morning.
Although
I
mean
it's
a
v8
API,
so
the
rule
of
two
summer
majors
doesn't
strictly
apply,
but
it
so
would
be
good
to
do
it.
In
noted,
it's.
A
C
E
C
A
D
A
E
E
Don't
you,
if
there's
anything
that
we
need
to
talk
about?
I
think
both
of
these
are
kind
of
waiting
for
PR.
I
think
the
ya
activate
inspector
with
you
back
process.
That's
the
one
that
kind
of
depended
on
where
we
go
with
our.
How
do
we
land
this
discussions
and
I?
Think
that,
based
on
what
we
just
said
landing
it
in
note
8.0
be
the
way
to
go
right.
So
in
the
opener,
if
you
kill
user
one
anode
process,
it
should
start
starts
the
inspect,
rohtak
all
greatly
the
breaking
change
in
no
date
right.
A
F
E
A
A
A
Yeah
I
think
everyone
knows
my
opinion:
I'd
love
it
to
say:
WS
colon,
slash,
slash,
IP,
address
colon
port
/
uuid
and
leave
it
at
that
and
then
align
redirecting
to
a
page
but
I'm.
You
know
whatever
we
want
to
do,
but
we
should.
We
have
to
take
out
the
experimental.
So
we
should
finalize
this
I
think
in
the
next
couple
weeks,
yeah.
E
A
If
we
only
say
the
port
and
not
the
IP
address,
so
there's
two
there's.
If
so,
there's
actually
two
things
either
the
IP
address
might
be
relevant
because,
for
example,
like
I
think
in
the
issue
has
come
up
with
the
docker.
For
example,
if
you
listen
on
127
001,
you
can't
connect
because
it
it
dr.
does
whatever
it
does
to
connect
you
to
a
proper
IP
address.
A
E
So,
yes,
that
is
kind
of
important
at
the
same
time,
at
these
leaders
do
your
coat
and
chrome
devtools,
don't
really
care
about
the
hewa
t,
because
they
both
hit
the
/
jason
endpoint
of
just
find
it
themselves
and
actually
printing
out
the
UID
makes
it
harder
to
restart
and
reload
right.
So
it's
the
thing
that
you
actually
maybe
did
wouldn't
want
people
to
explicitly
copy
somewhere,
because
it
means
that
they
need
to
go
back
and
forth
between
the
console
and
they're
developing
tool
for
each
time.
They
restart
the
process,
because
the
unity
keeps
changing.
C
C
F
E
A
That's
a
good
point.
Actually
I
was
kind
of
wondering.
Cuz
I
know
that
we
were
concerned
that
people
would
want
to
copy
paste.
The
chrome,
devtools,
URL
and
people
are
doing
that
now.
Is
there
a
way
that
chrome
devtools
could
take
just
a
janier
like
WS
colon?
I
don't
know
if
that's
an
official
scheme
but
like
WS,
IP,
port
uuid
and
just
copy
that
into
into
chrome
and
it
you
know
it's
really
just
you
know
interpolating
that
into
the
the
chrome
devtools
URL.
Would
that
be
possible.
A
F
D
C
C
F
E
It
takes
to
add
one
well
we.
So
if
you
go
into
chrome
inspect
today
and
you
go
behind
discover
network
tiger
targets
to
configure
you
can
always,
you
can
already
enter
a
hostname
colon
port
combination
and
it
will
automatically
discover
the
node
process.
The
only
issue
is
that
you
need
to
configure
it
for
each
individual.
Ip
port
combination
then
wait
for
it
to
pop
up
and
then
click
on
it.
Instead
of
having
you
know,
one
thing:
it's
immediately
connecting
to.
C
E
A
C
C
A
Okay,
I
am
so
what
I'll
propose
is
that
I
will
fix
up
my
thing
to
do
what
I
suggested
before
I
know
that
yon
you
mentioned,
maybe
not
the
uuid
and
eugene,
and
you
might
prefer
the
chrome
devtools
URL.
Let's
have.
Let's
have
that
conversation
there,
and
and
if
we
can't
reach
a
consensus,
then
we
can
ask
the
CTC
to
make
a
decision
for
us
what
it
does.
Everyone
put
their
own
thing.
F
A
A
B
I
wanted
to
mention
one
thing,
so
it
an
issue
came
up.
It
was
opening
as
the
Diagnostics
a
group
a
few
days
ago.
It
turns
out
that
the
the
macros
that
we
added
fell
so
they
were,
they
were
pulled
in
from
v8
and
they
fell
out
of
date
with
v8,
causing
them
to
no
longer
work
and
I'd
like
to
get
to
a
point
where
we
can
have
some
kind
of
instrumentation,
of
course,
so
that
we
can
test
that
they
continue
functioning
so
that
we
have
a
test
that
will
catch
catch.
B
This
on
v8
upgrades,
so
I
have
a
pull
request
together
as
soon
as
5.7
lambs.
That
will
update
the
macros
to
be
in
line
with
with
the
version
of
tracing
that
that
v8
5.7
has,
and
that
will
fix
them,
but
hoping
to
follow
on
to
that
with
at
least
some
small
instrumentation
of
core
to
get
kind
of
get
get
things
started
on
discussing
how
course
should
be
instrumented.
I
I
can
open
the
pull
request
that
I
original
for
the
patch
that
I
originally
used
when
I
was
playing
around
with
trace
event.
B
Yes,
the
second
one
would
add
some
trace
points
in
core
the
first
ones
that
make
the
most
sense
I
think
are
in
piecing
crap.
But
that's
when
we
can
talk
about
where
the
best
place
to
put
them
are,
but
it'd
be
great
to
have
have
them
appear
somewhere
in
node
core,
so
that
we
can
have
a
regression
test
for
the
two
makes.
A
A
A
Okay,
next
one
last
thing
on
our
list.
Actually
today
is
a
sink
hooks
album.
If
you
didn't
know,
Trevor
opened
a
new
PR
a
few
days
ago.
Oh,
he
said
yesterday
in
the
CTC
call
that
he
really
overhauled
it
I
haven't
looked
at
it
yet
is
there?
Are
there
things
we
want
to
discuss
here?
I,
don't
know
if
Trevor
zone
is
Trevor
on
the
call.
No.
A
A
Well,
I
want
to
bring
something
up
so
I
made
an
issue,
I
hope
with
an
issue
that
can
we
just?
Can
we
change
this
meeting
to
like
a
fixed
day?
If
that's
ok
with
everyone
will
make
it
easier
to
schedule,
we
can
have
it
that
after
the
tsc
meeting
like
like
this
or
before
on
the
last
Thursday
of
the
month
or
the
last
DSC
meeting
of
the
month,.
A
A
A
Yeah,
okay,
I'll!
Do
that
and
you
know
run
the
basically
anything.
That's
tagged.
Diag
agenda,
I,
just
run
broads
nude
meeting
agenda
thing
copy
that
in
there.
So,
if
you're
in
a
repo
or
you
are
in
a
group
that
doesn't
have
that
label
go
ahead
and
just
put
it
on
because
it
automatically
goes
across
all
the
repos
okay.