►
From YouTube: Diagnostics WG meeting
Description
A
A
B
A
A
So
that's
a
permanent
thing,
I
think
the
discussion
it
was
originally
going
to
be
at
one
and
then
as
I
schedule
them
out
over
the
the
shift
from
daylight
savings
time.
It
moved
back
an
hour
and
so
I
wasn't
exactly
sure
whether
people,
you
know
adjusted
things
with
timezone
if
it
was
sort
of
a
West,
Coast
centric
view
of
time
or
not
so
we
just
kept.
It
came
up.
There's
some
conflicts,
so
there
was
no
opposition
to
moving
it
so
going
forward.
It
should
be
a
1:00
p.m.
A
and
we're
gonna.
Do
it
every
two
weeks
with
some
exceptions,
I
think
the
the
next
week
or
the
next
one
is
skipped
because
it's
a
holiday
week
and-
and
so
the
next
meeting
will
be
on
the
10th
of
January.
If
I
recall
correctly,
so
there's
that
so
they
sync
context
formalization
stuff,
so
we've
got
a
draft
I,
just
put
it
up
on
a
pub
github
thing:
I'll
share
the
link,
it
is
rough
and
but
people
are
welcome
to
look
at
it.
Provide
feedback,
I.
Think
the
other
thing.
With
this
we
can
actually
talk
about
I.
A
That
we
opened
after
the
Vancouver
was
it
interactive
thing?
That's
just
sort
of
tracking
and
saying
like
hey,
you
know
there
was
this
this
paper
that
mark
wrote
and
you
know
we
want
to
try
and
take
that
to
kind
of
the
next
level,
get
a
little
more
democratize
and
then
figure
out
how
to
move
it
forward
and
we're
sort
of
tracking
it
there
right.
B
A
A
A
The
summit
I
think
that's
on
the
agenda.
There's
national
media
Michael
to
get
a
detailed
schedule,
posted
and
Tiger.
Also
some
questions
about
that.
I
missed
the
rest
of
the
topics
it
sounded
like
there
was
some
discussion
around
168
there's
an
open
issue
on
it
missing
tracking
or
there's
some
progress
being
made.
There.
A
B
D
C
B
B
A
I
mean
he
can
provide
the
most
context,
I
think
I,
I,
don't
know
that
the
discussion
is
going
to
be
any
different,
which
is
that
he
doesn't
have
enough
of
I.
Think
if
I
remember
correctly,
he
was
saying
he
didn't
have
a
the
background
knowledge
in
in
the
what's
it
called
the
nan
AP
on
equal
number,
Native
abstraction
thing
and
that
those
people,
weren't
necessarily
I,
don't
know
if
responsive
is
the
right
word
motivated
or
you
know,
had
the
time
to
kind
of
figure
out
what
the
right
thing
to
do,
though,.
B
Right,
I
guess
it's
like
from
my
perspective.
We
need
somebody
who's
interested
in
pushing
on
the
async
hook
side
to
do
something
like
submit
a
PR,
even
if
it's
not
complete
or
something
right.
If
we
ask
them,
if
we've
asked
them
in
guys,
they
can
get
involved,
we're
not
getting
a
response.
The
only
way
I
can
see
is
like
we
actually,
you
know
get
somebody
to
do
with
PR
and
I.
Think
most
likely,
unless
we
have
another
volunteer
in
this
group,
is
somebody
from
the
sink
Hawks
team
has
gotta,
say:
okay,
I'm
gonna
go!
B
B
B
A
So,
okay,
so
moving
on
so
number
129.
This
is
the
issue
I
opened,
you
know
some
some
feedback,
some
questions
about
domains,
I
think
there's
another
issue
that
somebody
opened
around
domains
and
what's
the
future
of
that
I,
don't
know
that.
There's
a
clear
answer:
there,
Natalie
wolf
from
New
Relic,
had
some
comments
about
some
performance
stuff,
which
is
actually
quite
interesting
and
maybe
a
little
concerning
just
in
terms
of
the
degredation
when
async
hooks
is
turned
on
and
it's
no
opping
right-
and
this
is
the
key.
A
B
D
Let
them
do
that.
That's
a
good
idea
at
elastic.
We
haven't
merchand
the
PR
yet
for
adding
agent
cooks
to
our
APM
solution,
but
when
we
do
it,
we
we
decided
to
actually
do
a
feature
flag
for
now,
so
people
can
switch
it
on
and
off
if
they
want
to,
and
part
of
part
of
adding
this
to
our
APM
solution
is
also
to
do
benchmarks.
So
I
will
add.
I'll.
Add
this
also
those
benchmarks
to
the
issue
as
well,
when
we
get
them.
D
A
B
A
So
I'm
gonna
be
anything
to
talk
about
the
summit.
It
looks
like
we
locked
in
some
dates
that
I
just
figured
12
13
right
like
we
have
there's
a
proposed
agenda
down
and
in
the
comments
I
was
thinking.
Maybe
we'd
hoist
that
up
into
the
into
the
issue
you
and
you
know,
I,
don't
know
if
we
want
to
go
through
and
actually
it
like
blocks,
sometimes
out
for
some
of
these
things
or
I.
Don't
exactly
know
the
structure
this.
A
B
B
The
challenge
with
that
sometimes
is
of
course,
like
I'm
interested
in
all
these
different
ones.
So
it's
kind
of
hard
to
choose
your,
but
that's
you
know
we
can
think
of
whether
you
know
that's
the
kind
of
approach
that
we
want
to
do
or
we
can
try
and
drive
through
all
of
them
together
as
a
larger
group,
there's
no
real,
fixed
requirement
of
how
we
do.
This
is
really
how
we
think
that
we
can
best
get
through
the
content
and
make
it
the
most
productive.
I
was
thinking.
B
You
know,
was
kind
of
like
a
9
to
5
each
day
too
much
longer
than
that.
You
know
you
kind
of
get
burnt
out.
Of
course,
like
the
first
night,
we
can
go
out
for
dinner
together
and
continue
to
do
that.
The
last
day.
Sometimes
it
ends
up
petering
out
a
little
bit
earlier.
Although
I
think
we
have
lots
of
lots
of
content
here,
so
I
could
easily
see
you
know
going
the
whole
time.
A
B
Yeah
so
I'm
happy
to
kind
of
take
what
I've
got
there
we
through,
like
James,
had
asked
to
have
some
sort
of
demo
as
part
of
it
sort
ake.
What's
there
try
and
put
it
into
that
structure
of
like
okay
morning
we're
gonna,
you
know
this
isn't
the
fixed
list,
but
this
is
our
starting
list
of
you
know
morning
will
do
this
will
break
it
out
into
the
afternoon
and
the
next
morning
into
these,
and
of
course
we
can
adjust
the
purpose
of
that
first
half
morning
is.
B
B
The
other
sort
of
action
I
have
in
there
is
I
gonna
figure
out
whether
I
need
information
from
people
in
terms
of
who's
showing
up
I,
suspect,
I'm
gonna
need
like
a
name
company,
maybe
something
like
that,
but
I'll
post,
that
in
there
as
well
in
terms
of
the
things
I
booked,
the
room.
I
still
don't
have
food,
so
I'm
trying
to
get
approvals
to
bring
out
bring
in
lunches
in
the
worst
case,
there's
a
cafeteria
right
on
site.
So
we
can
just
go
and
eat
on
our
own.
I
I
booked
the
room.
B
B
The
room
is
they
I
have
to
look
at
the
count.
The
the
room
I
went
looked
at
is
big
enough
that
you
know
if
we
have
five
people
more,
it's
gonna
be
fine.
Great,
even
ten
I
think
would
fit
into
that
room.
Great
I
think
it's,
it's
sort
of.
You
know.
Iii
said
it's
sort
of
up
for
20,
but
when
you
look
at
the
room
itself,
it's
like
yeah,
okay,
we
could
find
some
chairs
and
not
have
a
problem,
there's
actually
several
rooms
in
there.
E
B
Money
be
able
to
use
that
for
break
out
okay,
it
might
be
the
challenge
if,
if
we
don't
have
the
other
room-
and
we
all
you
know
breaking
out
into
four
different
groups
in
the
same
room
as
a
little
bit
her
as
said
than
done,
but
I,
we
did
that
in
past
one.
So
I
think
we
just
have
to
work
through
that.
Oh
yeah,
my
next
action.
You
know
when
I
get
time
like
I
said
probably
early
in
January
I'll
go
through
count
the
count
the
people
put.
B
D
A
B
A
Okay
case
number,
107,
so
async
context,
so
we're
done
with
the
summit
stuff.
We
can
move
on
so
async
context,
formalization
and
and
diagnostic
support.
So
Marc's
been
working
on
this.
This
markdown
document
it's
in
a
public
repo,
now
I
posted
the
link,
it's
still
rough
but
feel
free
to
you
know,
take
a
look
at
it
and
chime
in
with
stuff.
If
people
want
to
start
offering
some
things,
I
think
that'd
be
great.
A
A
A
The
graph
and
the
transitions
in
the
graph
write
this.
If
you
have
a
dag
and
these
events
are
happening
and
are
creating
edges
and
nodes
or
changing
state,
and
things
like
that,
we
should
be
able
to
visualize
that
in
some
kind
of
a
slideshow
and
then
we
can
find
some
way
to
kind
of
link
to
that.
So
you
can
say,
hey
here's
promised
at
all,
and
this
is
what
the
graph
that
you,
you
would
see
if
you
execute
a
bla,
simple
block
of
code,
that's
using
promised
at
all
or
here's
a
set
interval.
A
F
B
A
E
A
E
A
The
yeah
I
hope
that
I
hope
that
it
it
can
I
hope
that
it
can
solve
a
lot
of
problems.
I
think
that
we
all
have
uh-huh.
So
the
the
other
comment
is
like
we
put
it
in
a
in
a
separate
repository,
just
cuz.
We
didn't
want
to
deal
with
permissions
around
you
know
having
to
make
updates
to
it.
Well,
it
was
undergoing
what
should
end
up
being
a
significant
number
of
revisions
so
and.
A
A
A
There's
an
interest
in
this
I
think
from
Microsoft
side,
because
we've
got
some
stuff,
that's
monkey-patching
the
hell
out
of
things,
and
so
we
have
this.
You
know
diagnostic
channel
library
that
hopefully
is
someplace
that
you
can
have
subscribers
to
some
of
these
different
events
so
that
everybody
doesn't
have
to
go
in.
You
can
add
stuff,
but
we
don't
have
somebody
on
our
end
right
now.
That's
sort
of
that
has
the
capacity
to
actually
drive
that
forward.
It
is
on
github
I.
Think
it's
linked
on
this
and.
B
B
B
An
issue
on
you
know
alternatives
to
monkey
patching.
Maybe
we
don't
put
it
on
the
agenda
until
we
find
somebody
you
can
actually
push
it
forward,
but
it's
kind
of
there
a
reminder
that
says
yeah.
This
is
something
as
a
group
we're
interested
in.
We
just
don't
have
bandwidth
and
maybe
you
know
I.
We
should
make
sure
that
that's
on
the
list
for
the
agenda,
the
diagnostic
summit,
we
can
kind
of
discuss
prior.
You
know
where
all
these
things
fit
into
the
priorities
right.
D
A
There's
two
things
right:
one
that
you
need
to
know
where
these
asynchronous
transitions
are
happening
right
and
that's
one
case
where
people
are
going
in
and
monkey-patching
right
and
that
I
think
is
where
they
sing
close
should
address
or
the
intention
of
it
to
address,
and
the
other
issue
is
say.
For
example,.
A
A
That
you
know
people
are
catching
a
press,
API
is,
and
monkey-patching
and
Postgres
and
Redis
and
whatever,
and
and
not
because
of
the
context
transitions,
but
because
they
would
need
additional
information,
and
so
the
intent
with
the
diagnostic
channel
I
think
is
to
get
that
additional
information
so
that
they
don't
have
to
go
through
and
monkey
patch
everything.
So
maybe
it's
not
a
significant
problem.
I,
don't
know
exactly
all
the
details
of
what
other
APM
vendors
are
doing
in
this
space,
but
I
know
that
it
came
up
in
in
our
situation.
So
certainly.
B
A
A
Think
it's
a
question
is
well
I,
don't
know
it's,
it
would
be
ideal,
I
think
if
everybody
who's
writing.
Some
of
these
libraries
that
you
know
have
some
such
information
to
produce
could
just
kind
of
instrument
their
own
things.
Push
it
out
on
a
you
know,
some
published
data
on
some
channel
and
then
whoever's
interested
can
just
subscribe
to
that
channel
and
read
that
data
and
then
have
you
know
some
interesting
outcome
from
that.
A
B
We
haven't
had
time
to
discuss
or
think
about
it
enough
yet,
but
you
know
how
that
discussion
are
thinking
they
get
to
the
point
where
it's
like
yeah,
it
just
doesn't
make
sense,
and
people
are
happy
with
monkey
patching
or
know,
there's
something
to
be
done
there.
Even
if
we
can't
do
it
right
away.
D
There
was
also
a
thing
that
I
think
erased
on
Twitter
recently,
which
is
so
we
have
now
that
I
think
is
called
the
providers,
API
needs
and
crooks,
which
is
so.
If
you
have
a
user
land
view,
you
can
call
back
you're
supposed
to
use
that
API
to
fire
asynchronous
calls,
though,
obviously
people
are
doing
it
yet.
So
it
would
be
part
of
our
job
to
also
go
out
and
reach
out
to
these
people
and
end
these
module
openers
and
make
them
utilize
that
API
maybe
make
PRS
and
stuff
like
that.
D
But
until
all
that
gets
gets
done,
it
would
be
nice
to
have
a
list
of
known
modules
that
break
a
break
the
continuity.
So,
for
example,
like
the
one
we
all
know,
is
a
generic
cool,
for
example.
Right,
so
maybe
that's
also
an
issue
we
should
open
to
to
kind
of
eight.
If
people
are
like,
we
that's
a
good
idea
to
kind
of
keep
tabs
on.
These
are
the
modules
that
is
that
are
currently,
let
me
know,
or
like
sinners
sort
of
speaking
of
breaking
breaking
all
the
EPM
vendors
and
and
continuously
storage,
etc.
D
A
D
Identifying
is
always
hard
part,
but
I
mean
we
all
have
probably
at
some
point
stumbled
upon
a
module
that
that
it
breaks
whatever
we're
doing
so.
I'm
I
just
was
just
basically
two
step
one.
It's
basically
just
lets
them
have
a
common
list
that
we
all
can
share
and
add,
add
to,
and
then
from
that
we
can.
D
You
know,
start
to
approach
the
owners
of
these
modules
or
make
PRS
to
these
modules
to
utilize
the
brightest
API
native
cooks,
but
but
but
having
the
list
is
like
step
one
and
and
that
list
will
never
be
complete,
but
hopefully
will
cover
the
most
huge
modules.
Right
and
I
mean
I'll,
be
happy
to
to
start
it
off
and
take
whatever
we
know
and
input
in
there,
and
then
we
probably
missed
some
and
then
others
can
add
to
it.
Yeah.
D
D
A
A
B
B
A
A
A
E
So
Kelvin
has
been
working
on
updating
or
bringing
the
set
of
macros
within
no
more
update.
So
that's
the
first
step
so
they're
all
the
things.
That
array
will
be
it
that
I'm
not
waiting
in
order
to
point
them
in
synchronized
and
then
there's
some
macros
people
have
requested.
We
added
that
I
don't
have
an
implementation
v8,
so
I
would
be
a
thing
to
follow
on
with,
and
the
third
step
would
be
try
to
figure
out
a
longer-term
path
or
maintain.
B
A
B
B
But
there's
some
back
and
forth
discussion
around
how?
How
can
you
keep
this
metadata,
up-to-date
and
there's
sort
of
two
pieces?
One
is
that
you
know
there's
data
which
is
tied
to
the
structures
used
within
v8.
So
as
those
change
you
have
to
update
that
metadata.
Now
that
potentially
can
be
automated,
because
I
know
we
did
that
in
one
of
the
tools
we've
both
within
IBM.
The
other
part,
though,
is
relies
on
knowledge
of
how
you
sort
of
stitch
those
structures
together
to
get
interesting
things.
B
You
have
to
know
the
relationships
right
and
if
those
change,
you
can't
automate
the
update
of
the
the
sort
of
the
instruction
that
says,
I
know
that
I
use
these
structures
to
find
the
data,
because
there's
sort
of
logic
built
into
those
right.
So
that's
the
other
piece
where
the
individual
commands
in
ll
node
are
also
dependent
on.
You
know
how
we
stitches
those
things
together.
So
there's
sort
of
discussion
on
those
two
bits
on.
B
You
know
how
we
might
keep
these
more
up-to-date
more
easily,
but
no
real
conclusions
on
that.
There
was
some
information
about
there's,
there's
a
bit
more
info
on
fixed
offsets
that
could
be
useful
to
the
people.
Who've
been.
You
know,
the
the
people
that
joined
who've
been
the
biggest
contributors
to
that
metadata.
B
You
know
they're
saying
that
might
help
maintain
things
a
little
bit
more
easily,
but
still
there's
just
the
general
I.
You
know
having
read
through
it
I'm
still
not
clear
whether
we're
broken
or
not
broken,
and
we've
pinged.
You
know
who
kind
of
was
the
one
who
reported
it
to
me
a
few
times
just
still
need
to
catch
up
with
him
to
find
out
the
pieces.
He
thought
wasn't
working
just
so
that
we
could
better
understand
that.
A
Okay,
yeah
I
guess,
in
the
back
of
my
mind,
I'm
wondering
if
tools
that
are
at
risk
of
getting
broken
need
to
be
written
to
some
stable
API,
any
way
that
you
write
a
native
module
to
some
stable
API
and
then
v8
can
grab
underneath
it
now.
I,
don't
know
how
reasonable
that
is
of
an
approach
right,
especially
if
you're
talking
about
you
know
things
that
need
to
traverse
internal
structures
right.
You.
B
G
B
A
Okay,
so
there's
nothing
currently
in
v8.
It
says,
like
hey,
here's
a
set
of
primitives
that
expose
in
general
through
some
debugger
thing
that
and
you
and
then
build
abstractions
on
top
of
that,
and
those
things
will
remain
stable.
No
doll-like,
I
looked
at
the
code
and
I
know
the
layout
of
some
struct.
So
if
you
want
to
get
the
count
of
X
here's
how
to
do
it,
yeah.
B
We
know
that
you
know
the
object
is
shape.
The
shape
looks
like
this.
The
offset
at
you
know
address
eight
bytes
from
that.
The
top
is
X
right
yeah
and
that's
we're
sorry,
you
can
you.
You
potentially
can
automate
that
update
by
having
you
know
something
that
scans
the
code
and
says:
okay,
I'm
gonna
do
those
structures,
but
then
the
things
that
use
those
structures
are
harder
to
update
and
in
fact,
when
you
support
multiple
versions
you
may
even
have
to
have
them.
B
Okay,
okay,
so
yeah?
No,
that's
a
deeper
discussion
continue!
You
know.
Hopefully
we
can
spend
some
time
on
that
in
the
summit,
and
this
specific
one
on
like
is
it
now
broken
just
needs
to
continue
to
sort
of
see
if
we
can
get
the
original
report
clarified
and
then
figure
out.
You
know
reproduction
or
not
on
that.
B
B
The
action
really
is.
The
original
report
was
from
Netflix
slash,
you
know,
and
so,
if
we
can
get,
you
know,
callin
columns
chimed
in
saying
he
thinks
things
are
okay,
so
we
just
kind
of
need
to
close
the
loop
on
well,
you
know
is
there's
something
we're
actually
it
does.
It
is
working
okay
or
it
isn't
right
on
this
particular
issue.
This
particular
issue,
because
we
still
don't
have
the
end
user
saying
now:
okay,
you're
right,
it
does
work
or
no
wait
a
sec.
B
D
Have
a
quick
question:
I
just
took
a
look
at
the
async
wrapped
folder
on
our
repo
and
the
stuff
in
there
seems
to
be
horribly
outdated
and
since
the
the
documentation
now
actually
have
been
published
to
the
official
website,
I'm
thinking,
maybe
just
to
remove
this
folder
and
then
make
a
new
one
called
Asian
hooks.
Instead,
how
do
you
guys
feel
about
that.