►
From YouTube: 2022-09-20 meeting
Description
Instrumentation: Messaging
A
B
A
B
So
in
the
spec
meeting
this
morning,
santosh's
browser
editions
so
browser
language
and
browser
mobile.
I
think
it
was.
A
B
Yeah
so
the
it's
now
using
contrib
and
js,
so
I
removed
api,
but
there's
someone
in
the
contrib
history
who
no
longer
has
the
cla
sign.
So
it's
blocked
again.
B
A
A
B
Feedback
from
that
so
and
the
person
who
gave
me
admin
privileges
ping
me
this
morning
saying:
okay,
are
you
ready
to
get
up
yet?
And
I've
replied
not
yet
because
of
this
cla
issue.
So
I
see
okay,
because
if
I
lose,
the
I've
been
probably
just
gonna,
be
like
another
week
to
get
admin
privileges
back,
I'm
trying
not
to
give
it
up
it's
just
taking
too
long.
B
B
A
That
doesn't
matter.
I
was
wondering
if
he,
if
he's
around
and
was
going
to
join
too
today,.
B
Yeah
his
name's,
not
not
in
the
meeting
list,
and
I
don't
remember
seeing
him
on
the
on
the
call
yeah,
okay,.
B
D
C
D
Going
to
join
did
we
get
started.
A
I
was
wondering
the
same
thing,
but
that
doesn't
look
like
he's
going
to
join.
Okay.
D
All
right
we
can
get
started.
Let
me
let
me
share
my
screen.
I
put
a
bunch
of
stuff,
you
know
I.
I
only
got
time
to
work
on
this
yesterday,
but
put
a
bunch
of
questions
discussion
topics
for
today.
Let
me
share
the
screen
and
bring
up
the
agenda.
D
I
changed
the
change
of
devices
just
wanted
to
make
sure
okay,
so
yeah.
These
are
the
things
so
before
we
get
started
with
this,
I
want
to
quickly
update
folks
on
you
know
what
I've
been
doing
and
things
did
create
in
a
bunch
of
links
and
fill
out
stuff.
That's
that's
missing,
and
things
like
that.
I
think
there
is
a
lot
more
gaps
to
be
filled
here.
I
need
to
follow
up
with
dyna
tracer
elastic
csc
if
they
want
to
contribute
anything
where
the
schema
has
already
been
documented.
D
Here
I
put
links
and
stuff.
There
is
also
new
things
that
people
added,
I
think
t2.
I
think
you
added
mobile
stuff
right.
D
Oh,
I
see
okay,
okay,
gotcha,
oh
siri,
okay,
perfect!
Thank
you!
Apologies
for
that!
Okay!
So
that's
great,
I
think
at
some
point
we
need
to
get
to
mobile.
Also,
if
there
are,
I
put
a
question
out
there.
Also
there's
probably
common
events
across
browser
and
mobile,
we'll
probably
have
to
chat
about
it
at
some
point,
see
what
they
are.
A
page
view,
I
believe,
is
one
of
those
things.
An
interaction
is
probably
another
one.
D
We
want
to
see.
We
want
to
define
a
common
event
or
two
separate
events,
or
something
like
that.
So
for
now
we'll
go
with
browsers,
I
think
okay,
so
that's
basically
it
and
then
the
spreadsheet
looks
like
this.
You
know
added
a
little
bit
more
details
and
stuff
all
the
things
that
we
discussed
about
last
time.
All
the
events,
the
main
events
are
captured.
D
I
also
threw
in
filtering
here
so
the
way
I
think
we
could
work
with
this
spreadsheet.
Is
that
pretty
much
every
time
we
meet?
We
should
probably
do
this
as
a
team.
What
I
would
like
to
do
is
that,
let's
take
aws,
for
example,
filter,
remove
the
blanks
and
look
at
all
the
fields
that
aws
is
capturing,
make
sure
that
they
all
have
a
corresponding
open,
telemetry
field.
If
not,
you
know.
That
is
something
that
you
know.
D
We
should
probably
discuss
about,
or
essentially
say
you
know
what
that
is
just
vendor
specific
move
it
to
some
custom
properties
or
something
that's
the
exercise.
We
should
probably
do
for
every
one
of
these
vendors,
if
you
take
microsoft,
for
example,
just
remove
blanks.
I
know
that
these
are
these
are
my
fields,
our
fields,
you
know,
which
ones
are
already
mapped.
D
We
need
to,
you
know,
probably
chat
about
other
things
and
then
see
where
they
fit
and
don't
fit,
and
things
like
that,
so
this
is
probably
what
we
have
to
go
through
for
every
event,
any
questions
or
thoughts
on
the
process.
I
I
know
it's
a
liberal
process,
but
I
guess
this
is
the
best
that
I
could
think
of.
We
have
to
do.
D
Okay,
yes,
all
right
cool.
So
now,
let's
switch
to
our
agenda
and
kind
of
go
through
these
questions.
Post
dope
dark
load
resource.
I
know
we
talked
about
it
last
time
I
apologize
I
tried
to
and
I
raked
my
brain
yesterday
couldn't
come
up
with
a
reason
as
to
why
and
stuff
t2.
Perhaps
you
could
talk
a
little
bit
about
it
and
see,
which
is
the
equivalent
event.
C
Yeah,
that's
that's
just
because
of
the
current
hotel
instrumentation.
Only
loading
expands
at
the
that
happens
before
document
load
event,
so
we
need
a
separate
one
just
to
have
the
ones
afterwards,
I'm
like
totally
fine
merging
all
of
those
together
into
one
resource.
Maybe
even
just
one
can
generate
http
requests
spend.
D
C
A
D
See
I
see
so
hey.
C
D
Hey
hey,
so
we
just
get
you
know
we
we
just
got
started
santosh
and
I
was
just
catching
up
focus
on.
You
know
the
the
rearrangement
and
updates
that
I've
been
doing
to
the
dots
and
stuff
we
can.
Probably
you
know
it's
it's
a
recording
right.
You
can
probably
listen
to
it,
not
much
changes
here
and
stuff.
D
E
D
C
So
there's
there's
the
hotel,
interrupt
hotel,
instrumentation
and
then
we
add
an
additional
one
that
is
after
after
document
load.
So.
D
D
Okay
sounds
good,
so
this
is
probably
the
hotel
instrumentation.
The
document
load
is
the
order
instrumentation
that
maps
to
page
view
and
then
the
rest
is
basically
browser
timing.
That's
it
right.
We
should
be.
My
question
is:
should
we
compare
this
event
to
timing
and
you
know
figure
out
what
what
things
you
need.
D
D
Okay,
let
me
let
me
take
a
comment
here.
I
believe
I
need
to
get
open,
telemetry
document
load
resource
fetch.
Do
we
have
documented
here
yeah
that
one.
E
Yeah,
I
think,
to
keep
it
simple.
We
can
call
it
resource
fetch
itself,
but
what
we
can
do
is
the
resources
fetched
initially
will
have
a
parent,
which
is
the
document
load
resources
fetched
later
will
be
on
their
own.
E
E
Yeah
so
that
so
they're
talking
about
beyond
the
document,
the
load
right,
the
the
onload
is
complete.
E
Right,
so
they
want
to
keep
the
parent
open
to
collect
more.
You
know
more
child
spans
yep
finish
up
to
put
it
too.
Okay,
so
that
I
I
think
there
are
different
ways
to
model
this
and
we
will
model
the
most
common
or
rather
something
that
most
of
us
agree
on
there.
E
There
will
be
variations,
so
I
I
don't
know
how
other
instrumentation
teams
manage
that,
so
we
could
give
options
or
hooks
for
people
to
extend,
but
I
think
whatever
a
few
of
us
agree
on,
I
think
that
can
be
the
best
instrumentation
so
so
in.
In
other
words,
we
can
we
can
keep
the
existing
functionality,
as
is
if
we
believe
that's
going
to
satisfy
the
use
case.
First,
for
you
know,
most
of
us.
E
I
actually
added
something
to
the
excel
sheet
ram.
Could
you
go
to
the
excel
sheet
and
go
to
the
go
to
the
login
click
on
that
main
yeah.
E
Is
that
yeah
here
I
have
listed,
you
know
what
are
the
events
spans
and
possibly
metrics
that
we
will
emit
and
let's
look
at
the
center
one?
The
page
navigation
is
the
document
load
and
it
has
the
page
fetch
or
the
document
fetch
as
we
call
it
currently.
E
So
the
resource
fetch
can
be
part
of
the
initial
page
navigation.
But
at
some
point
the
page
navigation
is
complete
right,
but
then
post
that
you
know
the
resources
can
continuously.
You
know
be
called
later
as
well,
so
whatever
I
can
even
add
the
fetch
and
exit
charge
to
be
part
of
the
initial
page
navigation
as
children.
E
E
Any
calls
that
happen
could
be
considered
as
child
spans,
but
they
can
be
in
so
so
the
way
they
will
work
is
they
will
have
their
own
independent
instrumentation,
and
if
it
happens
to
have
a
page
navigation
document
a
span
in
progress,
they
will
consider
it
as
a
parent,
but
otherwise
they
are
independent.
E
D
When
you
say
same
name,
can
you
clarify
so
page
instead
of
page
fetch
resource
fetch
and
just
fetch,
you
basically
say
that
it's
just
a
single
event,
type
called.
E
No,
no,
that's
not
what
I
meant.
I
meant
that
so
in
the
other
document
there
was
a
post
doc
load
resource
right,
correct.
So
that
is
the
row
number
10
resource
fetch.
D
E
E
It's
just
that
in
in
one
case
it
has
a
parent.
In
another
case
it
doesn't
have
a
parent.
D
Okay,
so
my
question
is
so
far
we
haven't
talked
about
it.
You
know
what
what
is
that,
even
so
right
now
in
the
excel
spreadsheet,
that
event
doesn't
show
up
here,
so
maybe
I'm
confused
so
which
you
know
we
need
a
separate
event.
I
guess
right
or
is
it
resource
timing.
E
So
I
have
the
timing
I
have
put
under
events
for
each
of
these
pans.
I
have
also
put
a
timing
event.
You
know
which
you
know.
Could
you
know,
depending
on
the
configuration
it
could
go
as
a
span
event
or
a
standalone
event?
E
E
D
So,
okay,
I
think
I
think
I'm
following
sorry
to
get
you
off.
I
I
think
so.
This
sounds
like
there
is
not
much
to
define
in
terms
of
the
event
schema
when
we
talk
about
these.
Is
that
correct?
There
is,
if
it
is,
then
we
need
to.
You
know,
define
it.
So
I
guess
the
exercise
that
we're
going
through
is
that
these
guys
have
so
many
attributes
or
fields
that
they
have
to
capture,
and
we
need
to
figure
out
what
the
schema
looks
like.
So
this
is
where
the
heavy
work
is.
D
If
we
are
saying
hey,
we
also
need
a
span
to
calculate
the
time
taken
for
any
activity
and
we
want
to
stitch
them
together
from
distributed
in
a
tracing
type.
You
know
model
we
wanted.
We
want
the
tree
structure,
that's
a
little
bit
of
a
separate
thing.
I
think
you
know
tell
me
if,
if
I've
got
it
wrong,
unless
this
page
navigation
also
has
a
whole
bunch
of
attributes,
we
have
to
collect
as
part
of
that,
and
we
need
to
figure
out
what
the
semantics
is,
what
should
go
in
there?
D
What
should
not
go
in
there?
We
should
probably
table
that
discussion
for
a
little
bit
later.
B
D
B
I,
I
think
sorry
mountain,
I
I
think
it's
they're
related,
but
it's
that's
not
what
we're
trying
to
do
right
now
and
in
terms
of
the
page
navigation
on
the
page
load
that
could
be
controlled
by
the
application
so
that
all
the
resources
in
terms
of
when
is
page
load
finished
deemed
and
that
could
be
based
on
the
framework
so
yeah,
I
think,
for
now.
A
No,
no,
so
I
just
want
to
say,
like
I
basically
agree
with
you
what
you
said
as
well,
like
that's
how
I've
been
thinking
about
it.
Essentially,
the
events
events
are
easy
to
capture.
They
can
capture
all
the
data
that
we
need.
The
only
thing
that
they
do
don't
capture
is
the
relationships
but
like
for
for
like
resources,
and
you
know
the
navigation
timing,
page.
All
that
you
can.
A
You
can
capture
all
that
with
events
right,
so
so
so
I
think
you
know
when
we
yeah,
like
ram,
like
you
were
saying
like
right
now,
we're
talking
about
the
schema.
That
really
is
applies
to
those
events,
the
the
spans.
A
I
I've
been
thinking
about
them
as
like
as
like
yeah,
if
you
do
want
to,
if,
if
the
instrumentation
is
capable
of
of
capturing
those
relationships
and
like
you
really
care
about
those
relationships
like
you,
can
there
could
be
an
an
additional
like
plug-in
that
you
can
add
so
like,
for
it
would
be
useful,
definitely
useful
for,
like
the
page
load,
where
you
can
see
like
the
the
tree
of
things,
the
resources
that
were
uploaded
during
page
load,
fetch
xhr
for
sure
I
wasn't.
A
I
raised
my
hand
a
little
bit
earlier,
like
my
comment
was
about
the
resource
fetch
spam
online
10,
like
I'm,
not
really
sure
that
has
any
value,
because
it
comes
from
the
it's
gonna
come
from
the
browser
api
and
it
can't
be
connected
to
anything
else.
C
I
disagree
that
it
can't
be
connected
to
start
concurrent
instrumentation.
Isn't
it
isn't
connected?
It
is
possible
to
observe
using
mutation
observer.
I
mean
making
it
technical
now
and
it's
that
later,
but
it
is
possible
to
observe
using
mutation
observer
for
when
an
image
or
html
or
script
tag
has
to
be
put
onto
a
page,
and
you
can
call
that
observer
during
context
and
euchronously
to
determine
that
yeah.
These
elements
were
added
during
this
balance
span.
A
B
That
really
depends
on
the
parents,
the
parent
instrumentation,
that's
tracking,
the
unload,
in
terms
of
when
does
the
unload
finish.
So
if
that
resource
online
10,
it
was
an
image
that
happened
to
be
loaded
well,
after
the
fact
that
the
unload
fit,
then,
then
there
is
nothing
there
to
associate
it
with,
which
is
why
I
I
think
there
for
now
we
should
table
it
because
we're
talking
about
how
do
we
define
that
tree
structure?
B
On
top
of
the
events
and
like?
I
think
we
need
to
do
the
events
first
and
it
does
come
back
into
yeah,
like
the
user
interaction,
the
one
that
I
see
ram
moving
around
on
at
the
moment.
It's
a
case
of
if
you're
doing
a
I
know,
you're
using
react
and
you're
doing
a
page
navigation,
so
you
hook
the
router
as
part
of
the
user
event.
B
You
would
have
the
the
user
interaction
span
that
gets
created
while
it
goes
and
fetches
everything
and
then
it
just
automatically
gets
linked.
So
while
there
probably
isn't
going
to
be
an
instrumentation
that
does
a
resource
fetch
and
creates
a
child
span,
it's
it
will
just
as
santosh
was
talking
earlier,
it
would
automatically
inherit
whatever
the
current
parent
span
is.
So
in
the
case
of
line
10
when
there
is,
there
is
no
parent,
that's
exactly
what
it
would
look
like.
C
Yeah-
and
that
was
by
the
way-
actually,
I
think
it
yeah
sometimes
interactions
sometimes
talk.
Mutations
will
happen,
that
we
don't
have
the
current
context
current
context
at
the
moment,
but
like
when
we
do
there's
no
point
in
not
doing
that.
A
Okay,
so
so
it
sounds
like
so
I
wasn't
aware
I
don't
know
exactly
the
details
on
how
you
would
connect
the
resource
span
from
the
observer
to
a
context.
The
current
context,
but
I
take
a
word
that
is
possible,
so
it
would
be.
It
sounds
like
that
span
would
be
useful
only
for
things
like
interactions
like
where
you
would
want
to
see
that
part
of
some
other
context
like
bigger
higher
context
than
that
resource
yeah.
B
Like
I,
I
I
keep
logically
thinking
of
this
a
span
is
a
grouping
context.
So
if
you
know
what
the
group
of
things
you're
doing
structure,
how
we
do
that,
I'm
not
worried
about
that
at
this
point,
that's
when
we
once
we've
defined
all
the
events
and
we
figure
out
okay.
This
is
what
we
want
to
send.
Then
we
can
start
drilling
into
the
detail
of
okay.
How
do
we
make
sure
these
things
are
linked?
Because
what
we've
got
in,
like
columns
being.
A
Okay,
so
I
just
just
to
kind
of
complete
that
whole
circle.
I
would
say
that
the
resource
fetch
span-
maybe
it
can
be
connected
to
like
a
parent
span,
but
it
can't
it
can't
propagate
context
downstream
right
so
like,
unlike
the
fetch
and
xhr.
D
You
know
great
discussions.
I
wanted
to
almost
comment
about
it,
but
you
know
trying
to
try
not
to
comment
on
spans,
so
we
need
two
sets
of
things.
You
know
one
is
to
capture
data
or
details
about
you
know
what's
happening,
you
know
like
the
yellow
stuff
right,
yellow
the
orange
stuff,
and
then
there
is
this
relationships
between
things.
That's
happening.
D
We
we
need
both
of
them
and,
like
we
talked
about,
let's
separate,
separate
them
out,
and
I
I'm
deliberately
attempting
to
do
that
because
if
you
put
them
together
it'll
it
it'll
be
a
long
time.
I
think
I'm
just
trying
to
keep
it
focused
on
the
left
side.
Let's
define
those
things.
D
If
you
want
to
run
that
in
parallel,
you
know
somebody
else
could
run
it
or
we
could
do
it
subsequently
after
we
are
kind
of
defined
most
of
the
events
and
stuff,
I
think,
if,
if
everybody's
in
agreement,
then
we
can
separate
them
out
and
focus
on
the
yellow
items,
defining
semantics
cool.
Thank
you
all
right,
so
going
back
here,
give
me
one
sec.
D
Let
me
just
okay,
perfect.
Okay,
I
need
to
capture
that
here
before
I
forget.
D
Okay,
I'll
I'll
try
to
define
it
later.
Okay,
so
I
think
that
I
think
that
makes
sense
so
the
I
think
I
got
the
answer
for
this:
the
dark
load
resource
we
should
basically
compare.
D
D
So
we
did
talk
about
mobile
events-
I
I
know
seared
right
or
yep
yeah,
sir
added
it.
They
didn't,
follow
the
the
pattern.
I
think
I'm
just
asking.
I
think
we
should
just
do
a
browser
like
that.
A
mobile,
slash,
app
startup
mobile,
slash,
app
crash
and
stuff.
D
D
So
that
one's
done
this
is
a
question
we
could
probably
take
that
toward
the
end.
There's
more
medial
things.
I
think
I
did
touch
on
this
a
little
bit.
They
probably
are
common
events
across
clients.
We
should
figure
out.
D
Are
we
going
to
use
a
single
event
schema
for
both
or
is
it
going
to
be
two
separate
things
and
santos
confirmed
this
main
tab?
You
were
going
to
attempt
to
do
sort
of
a
tree
structure
of
the
yeah.
E
Yeah,
that's
what
I
tried,
but
then
ended
up
with
this,
because
the
tree
is
only
for
a
trace
right.
You
know
the
events
in
it
are
the
spans
in
a
trace
right,
whereas
here
I
I
think
many
of
these
things
are
independent.
The
only
you
know
a
single
trace
relationship
between
spans
is,
is
you
know
the
page
navigation
and
the
page
fetch
and
resource.
I
think
that's
the
only
thing
I
could
determine.
I
think
I.
D
See
I
I
think
we
could
you
know.
Definitely
you
know,
map
these
events
to
the
navigation
timing
or
something
you
know
if
in
you
know,
you
bring
up
network
tab
in
a
chrome
or
something
like
that.
You
could
see
the
timeline
view
of
things
as
they
happened.
D
For
example,
page
navigation
can
never
or
you
know,
starts
or
fires.
You
know
the
timing
of
when
these
things
could
fire.
We
should
probably
be
able
to
find
that
that
could
definitely
throw
some
clarity
into
these
events
when
people
look
at
them.
D
Okay,
all
right
cool.
Thank
you.
D
The
user
interaction
event
seems
to
be
missing
from
the
dock.
Here
I
don't
know.
If
I
went
back
and
forth,
I
just
couldn't
find
it.
Is
that
something
that
you
could
help
with,
or
you
know
have
somebody
follow
up
and
do
that
youtube.
C
It
was
just
a
broken
link
due
to
the
last
time
last
week's
deletion.
It
should
be
particular.
D
D
Do
we
have
that
documented
here?
D
C
Well,
the
formatting
is
also
again
broken
due
to
the
deletion,
but
these
are
actually
span.
Data
type
fields,
so
stop
span
has
its
names
when
it
says,
instead
of
the
spans
one
is
appearance,
but
as
then
attributes
and
attributes
there
are.
These
are
the
attributes
gotcha.
D
Okay,
that
makes
sense.
So
really
you
know
if
we
were
to
trans
transfer
this
over
to
the
spreadsheet.
This
is
basically
what
I
would
do.
I
think-
and
I'm
not
going
to
do
all
of
that
right
now,
but
just
so
people
understand
what's
going
on
it's
the
user,
user
action
user
interaction
stuff,
I
think
we
would
simply
just
throw
it
there.
E
D
Right,
so
we
did
talk
about
that
logistics
just
before
you
started.
So
one
of
the
quick
things
is
what
I
did
is
added
filtering
for
all
these
things.
So,
for
example,
if
you
take
app
dynamics
right,
you
would
just
do
this
remove
the
blanks.
D
This
is
all
the
fields
that
you
care
about
that
you
capture
today
right
see.
If
it
has
mapping,
if
it
doesn't
have
a
mapping,
you
have
a
problem,
you
basically
bring
it
up
and
say:
hey
guys,
you
know
we
want
event
type.
This
is
why
it's
super
important.
You
must
all
capture
it.
That's
a
discussion
topic
and
then
we
all
decide.
You
know
we
either
accept
your
proposal
or
we
shoot
it
down.
If
we
shoot
it
down,
it
goes
to
custom.
If
we
accept
it,
then
we
come
up
with
a
name
for
that
even
type.
D
That's
that's
how
we
should
just
go
through
this
as
a
exercise
after
I
quickly
go
through
the
questions
I
added
in
the
doc.
This
is
what
I
want
to
do
so,
as
a
you
know,
a
repeat
thing
right.
Every
time
we
meet,
I
want
us
to
essentially
in
a
couple
of
couple
of
times
or
something
start
going
through.
This
pick
an
event
go
through
this
exercise
for
every
vendor
figure
out
this
as
we
go
through
this.
This
will
start
growing.
That's
probably
a
different
angle.
D
Otherwise
we
define
this.
It
could
be.
You
know,
there's
two
different
ways
of
doing
it.
You
know:
where
are
you
going
to
base
it
from?
If
you
took
that
responsibility,
it's
most
likely
going
to
be.
You
know,
like
your
events,
and
that
would
drive
you
discussions
and
stuff.
So
you
know
this
is
what
I'm
thinking
or,
if
I
did
it,
it'll
most
likely
look
like
a
microsoft
event.
A
D
So
this
is,
you
know
this
could
be
an
exercise
we
could
all
do
separately
also
and
then
be
prepared
for
that
meeting.
So
I
know
that
there's
a
whole
bunch
of
stuff
that
I
need
currently
in
capturing
for
page
view
that
is
still
not
defined.
Then
you
know
we
can
drive
those
discussions
on.
You
know
where
it
shows
up
and
things
like
that
wherever
you
see
these
things
already
populated
you
know
when
I
see
common
stuff
across
multiple
vendors.
I
already
put
them
here,
because
I
know
it's
a
common
theme.
E
So
the
ref
uri,
what
what
was
it
page?
Referral
right
actually.
D
Yeah
ref
uri,
for
you
guys,
I
believe
it's
page
referrer,
whereas
it's
refuri
for
others,
I
haven't
you,
know
splunk
once
I
populate
I'll,
be
able
to
see
which
one
needs
to
go.
There.
E
A
D
Okay
yeah,
so
typically
it's
the
referring
pages
uri,
but
it
can
be
blankets
and
it's
an
optional.
It
can
be
blank
if
somebody
just
goes
to
the
browser
and
starts
typing
www,
you
know
google.com
or
something
like
that.
Reference
is
going
to
be
blank
okay,
but
when
from
within
google.com
you
click
a
link,
then
google.com
is
going
to
be
the
reference
for
the
next
page.
That
comes
up.
B
Yeah
as
long
as
the
user
hasn't
blocked
it,
so
yeah
there's
a
couple
of
levels
that
we
need
to
look
at
here.
One
is
this
first
level
of
identifying
the
field,
the
common
field
names.
Once
we
have
the
set
of
common
field
names.
Some
of
these
will
have
the
existing
semantic
conventions,
so
we'll
then
have
to
have
the
discussion
of
yes.
We
just
sent
them,
as
you
know,
the
existing
attributes
and
then
the
third
one
in
the
case
of
the
event
type
is
like.
B
D
Sorry
now
apologies,
the
last
one.
I
didn't
follow
quite
what
even
type
you
know
the
point
that
the
media
body
runtime.
B
Okay,
so
in
the
example
that
you
gave
with
ap
dynamics
where
they
with
the
event
type
yeah,
it
was
a
case
of
if
we
decide
that
we
want
to
have
event
type
as
a
standard
hotel
field.
B
The
the
second
level
of
discussion
saying
hey,
we
have,
we
have
something
called
event
type,
but
then
we
also
have
what
are
the
values
of
that
event
type?
Is
it
just
going
to
be
open
or
is
it?
Is
it
just
a
string
or
is
it
going
to
be
an
enum
or
or
what
but
like
our
behaviors,
it's
like
is
event
type
behaviors
or
or
you
know,
if
we
had
behaviors,
what
would
the
values
be?
That's
really
all
I'm
saying
that's
the
third
level.
D
E
And
the
browser
line
30
3132
those
will
go
into
the
resource
on
the
hotel
resource
object,
so
they
they
can
be
removed
from
here.
D
D
Exactly
so
so
that's
the
mechanics,
you
know
in
my
mind,
I
I
you
know,
I
think
that's
what
nevi's
also
saying:
okay,
we,
you
know,
we
don't
care
the
you
know.
How
are
we
going
to
set
them?
How
are
we
going
to
capture
them
or
anything
like
that?
We
first
just
say:
these
are
the
things
I
need
and
I
don't
care
how
we
capture
them,
how
we
send
them
over
the
wire
or
to
the
back
end
or
anything
like
that.
First,
let's
do
that.
D
E
D
Yeah,
that's
the
this.
This
is
exactly
what
you
know
the
way
I
see
it
is,
you
know
we're
building
up
to
you,
know,
figuring
out
and
building
the
muscle
on
and
how
we
can
work
on
this
together.
I
believe
we
understand,
you
know
what
tools
we
want
to
use
and
now
I
believe
now
on.
It's
gonna
be
rinse
and
repeat
once
we
all
go
through
one
or
two
fields,
we'll
all
know
exactly
what
we
do
as
a
team.
D
So
it's
just
a
matter
of
going
through
so
right
now,
I'm
just
still
populating
stuff.
You
know,
gathering
these
big
chunk
of
things
to
you
know
show
an
example
or
whatever,
which
I
believe
I
did
you
know
we
can.
We
can
start
going
through
one
by
one
and
then
discuss
okay,
so
then
yeah
we
can
do
that
exercise
see
how
you
know
where
it
takes
us.
I
tried
to
populate
page
view
with
most
info,
so
we
could
go
through
this
as
a
as
a
team.
D
What
I
would
do
is
aws
doesn't
have
a
page
view.
Schema
that's
another
problem.
I
need
to
go
fill
that
up.
Give
me
one
quick
second
I'll
yeah.
We
need
to
go
through,
adb,
schema
and
find
equivalence
for
things.
I
think
they
dumped
a
bunch
of
stuff.
It's
not
there
who
in
aws
is
active
again
quinn,
clint,
right,
yeah,
yeah
yeah.
I
did.
I
did
chat
with
him.
Does
he
show
up
to
tomorrow
meetings.
E
D
Okay,
so
I
guess
I'll
have
to
think
about.
E
Slack,
you
know
not
lately.
D
Okay,
I'll
I'll
tag
him
you
know,
maybe
you
know
somebody
else
could
also
bring
him
to
see.
If
you
know
he
could
join
tomorrow
or
next
week,
or
something
like
that,
we
need
that
that
inform.
D
D
A
D
And
they
don't
seem
to
have
a
they
have
a
network
request,
performance
error
and
then
something
called
just
an
event.
This
probably
is
a
interaction.
Target
element
seems
like
an
interaction
thing.
A
D
Yeah
button
and
stuff
okay,
so
I'm
making
some
guesses.
I
didn't
want
to
start
populating
with
a
lot
of
cases
and
stuff
so
I'll
I'll
reach
out.
D
We
need
somebody
from
aws
and
anikum
real,
quick
going
through
other
stuff.
I
think
that's,
basically
all
the
exact
you
know,
questions
and
stuff.
Let
me
go
back
to
here.
We
can
go
through
this,
like
I
said
how
we
could
we
can
start
with
cisco.
D
And
then
look
at
you
know
this
time
stamp.
I
think
it's
pretty
obvious.
I
did
capture
comments
here
as
like
descriptions.
I
think
this
is
an
okay
place
to
do
that.
That
is
simply
the
client
time
stamp
right
for
you
guys
the
events
time
correct
it's
the
same
for
us
also,
martin.
Can
you
confirm
that's
the
thing
yeah,
it's
just
even
local
time,
right.
A
Yeah
so
yeah,
and
just
to
clarify
like
in
when
you,
when
you
is
it
going
to
be
like
a
relative's
timestamp
to
the
navigation
start,
or
is
it
going
to
be
unix
timestamp
like
absolutely
next
time.
B
Yeah,
I'm
pretty
sure
it's
a
string,
and
I
was
thinking
about
this
the
other
day
like
any
time
we
define
time-based
stuff
for
open
telemetry,
which
probably
should
have
it
in
a
time
format.
So
if
we
use
like
unix
time
in
milliseconds
we're
going
to
hit
a
big
wall
once
we
hit
2032
2034
depending
on
bit
wise,
because
that's
when
we
ran
out
of
32-bit
bits
to
represent
the
number
of
millions
milliseconds
since
1970.
A
C
E
Yeah
yeah,
so
what
does
the
timestamp
represent
the
time
when
the
the
page
fetch
was
initiated
by
the
browser
or
when
the
page
view
event
is
created
by
the
instrumentation.
A
B
A
B
B
Sorry
yeah
right
right,
but
yeah.
There
are
two
time
stamps
the
second,
so
one
is
when
the
job
when
the
javascript
says
create
the
event,
and
this
the
other
one
is
when
it
really
did
occur.
So.
D
I
think
I
typed
it
right
occurred
and
I
always
get
that
typo.
Okay,
I
I
I'm
pretty
sure
this
is
a
requirement
across
the
board.
You
know
once
you
start
filling
out
from
other
vendors,
it
will
naturally
flow
through.
I
think,
and
then
url.
This
is
typically
the
pages
url
you
know
not
having
splunk
here
populated
I
apologize
t2,
you
know
having
it.
There
would
have
been
really
useful,
but
I
don't
want
to
spend
everybody's
time.
You
know
trying
to
copy-paste
stuff
here
and
things,
but
I'll
do
that.
D
Url
is
the
current
pages
url
that
we
are
talking
about
and
things
it's
it's.
I
think
it's
simple,
but
you
know
also
do
we
do
anything
in
terms
of
you
know
truncating
or
anything
like
that.
Is
it
the
fully
realized?
It
shows
up
with
query
string
and
all
these
things.
B
I
think
for
us
we
do
sanitize
it.
Okay,
there
is
another
field
that
we
have
buried
in
data,
where
I
think
we
can
have
the
full
url
and
that's
because
of
pai.
Like
one
thing
we
haven't
really
started
to
think
about
yet
is
you
know
the
pii
and
eudb
type
data
so.
D
Yeah,
actually,
actually,
the
question
is
and
more
like
you
know,
the
the
usefulness
of
you
know
capturing
this
field
and
stuff.
Typically,
they
use
it
for
bi.
You
know
they
want
to
look
at.
D
You
know
the
aggregate
values,
for
you
know
the
views
on
a
particular
page
or
something
like
that.
So
if
you
capture
the
query
string,
each
one
is
going
to
look
different.
They
you
know
the
back
end.
Whoever
is
doing
the
processing
will
have
to
come
and
truncate
all
these
things
and
just
look
at
the
scheme
and
the
domain
and
then
the
relative
page
name
right,
foobar.com
slash,
you
know,
folder,
slash,
page.aspx
or
something
like
that,
not
the
query
string.
Typically,
we
shave
up
the
query
string
to
aid
in
easy.
D
You
know
aggregation
of
the
back
end
and
stuff.
Do
you
guys
know
what
you
do
in
your
site?
Do
you
capture
the
url
as
is
or
do
you
use
do
you
do
something
you
know?
Do
you
massage
the
url
before
saving
it.
D
Okay,
there
is,
you
know,
one
another
topic
I
would
like
to
talk
about
as
part
of
that
there
is
also
something
called
an
id
which
I
believe
shows
up
in
our
side.
I
can
talk
about
it
if
not,
that
concept
had
come
up
before
I
don't
know
what
it
ended
up
being
so
anyway,.
A
Yeah,
so
we
do
remove
the
query
stream.
We
don't
sanitize
the
path,
though
so.
Okay
we've
had.
I
think,
we've
had
maybe
one
couple
cases
in
the
past
where,
like
customers
were
putting
some
sensitive
data
in
their
past
and
and
then.
D
A
D
A
D
D
Okay,
the
thing
that
I
want
to
talk
about
is
there
is
also
this
concept
of
id.
I
believe
it
exists
in
our
side.
Do
you
remember
that
now
there
was
a
there's,
an
id
field
here
somebody
had
an
id.
I
remember
seeing
it
page
id
right
here.
Aws
has
a
page
id,
so
this,
I
believe,
is
a
unique
id
given
to
any
of
these
pages.
So
this
actually
takes
away.
D
D
Page
number,
you
know
246
just
give
me
an
aggregate
of
all
the
impressions
for
that
or
something
like
that.
Is
that
something
that
we
want
to
capture
as
a
thing.
So
this
comes
up
only
if
you
know
the
question
is:
where
do
you
get
this
value
from?
Is?
Is
the
problem
there
so.
E
I
feel
like
this
is
something
that
the
back
ends
should
do
it's
a
consideration
for
them.
I
see
the
agent
I
I
don't
know
whether
they
should
be
concerned
about
it.
D
I
think
sorry,
I
think
martin
might
be
asking
about
impression
id,
which
I
believe
is
something
it's
a
complete
client
in
a
true
client
concept,
every
impression
unique
id
you
have
to
create.
I
believe
that's
that's
a
different
than
the
id
we're
talking
about
example,
says.
D
Yeah
yeah,
let's
say
I
have
a
website
with
three
pages
page,
one
that
html
page
two
dot,
html
page
three.html,
my
page
url
will
look
like
this
www
say:
rom.com,
page1.html
page2.html,
so
on
support.
These
are
the
three
big
long
string
names
that
I
have
to
aggregate
with.
If
I'm
doing
some
bi
in
the
back
end.
Instead,
I
could
also
say
I'm
going
to
call
them
123,
124
125.
These
are
my
page
ids.
That's
it
that's
the
equivalent,
but
the
problem
is
page.
D
Url
is
something
that
every
page
must
have
it's
it's
it's
a
given
right.
You
know,
you
know
where
to
query
it
from
you
know
how
to
get
it.
Just
query.
The
browser
for
the
page
url
you'll,
get
it
id
is
something
that
the
page
creator
the
service
right
has
to
stamp
on
the
page,
probably
in
metadata
or
something
that
we
have
to
go
read.
If
they
don't
do
it
right,
we
won't
get
it.
So
this
is
a
little
bit
of
a
tricky
thing
to
capture.
D
B
This
is
probably
more
an
application
level
thing.
So
when
I
was
on
identity,
we
did
exactly
this.
So
as
part
of
the
javascript
and
the
spa,
we
effectively
had
our
own
set
of
ids
with
your
own
enums,
so
it's
very
application
or
domain
specific.
D
B
So
this
could
just
be
a
case
of
we
just
say
there
may
be
an
id
it'll
be
optional
and
it's
probably
going
to
be
tagged
as
an
attribute
or
we
can
keep
it
as
the
event,
but
it
would
be
yeah
application
defined
like
I,
I
don't
think
from
an
hotel
perspective.
We
can
say
there
is
an
id
and
it
will
always
be
like
one
means.
The
login
page
two
means
the
log
out.
D
We
cannot
do
that
yeah
most
likely
most
likely
and
did
he
say
and
it
will
be
an
attribute
right.
Custom
attribute
or
something.
B
Yeah
custom
attribute-
or
we
just
say
it's,
the
user
page
id-
which
we
describe
it
as
okay.
D
D
So
and
then,
while
we're
at
it,
we
can
probably
talk
about
impression
id.
So
that's
actually
I
did.
You
know,
explain
what
that
was.
Typically
it's.
If,
if
the
four
of
us
or
five
of
us
look
at
the
same
page,
wanted
html,
we'll
all
have
unique
impression,
ids
and
if
I
go
home
and
look
at
page
one.html,
that
will
be
a
different
impression
id.
D
A
D
D
We
had
an
impression
id.
Maybe
I
couldn't
find
it.
You
know,
didn't
yeah.
I
know
where
it
is.
I
know
what
it
is.
We
have
a.
I
didn't
capture
all
of
our
ext
schema
parts
yeah,
the
general
general
parts.
It's
there,
an
int
web
I'll,
get.
E
D
E
D
E
Id
okay,
you
go
from
page
a
to
page
b.
They
both
will
have
a
different,
obviously
yeah,.
D
Yeah,
so
it's
it's,
it's
exactly
the
what
the
impression
that
the
browser
got.
That's
it!
So
if,
if
you
sat
there
and
refreshed
the
example
that
I
gave
you
the
four
of
us
looking
at
it
will
all
be,
you
know
unique,
impression,
ids
and
then,
if
you
switch
to
a
different
computer
impression
id,
even
if
you
just
hit
refresh
on
your
page,
that
will
be
a
new
impression
id.
Okay,
then.
B
E
Yeah,
I
agree
with
that
that
you
know,
even
though
we
don't
like
it,
we
will
have
it's
better
to
use
service
namespace
for
consistency's
sake,.
C
D
So
I'll
I'll
have
to
fill
this
up
for
neural,
like
the
browser
interaction
id.
Is
that
the
equivalent
modern
did
I
get
that.
A
D
Thanks
guys,
yeah,
I
think
this
is
how
we
want
to
do
it.
You
know
if
you
know
if
we
already
populated
the
awesome.
Otherwise,
as
we
discussed,
if
you
guys
can
go
through
look
at
your
schema
and
just
drop
those
things
here.
This
is
great.
I,
like
you,
know
the
way
we
are.
You
know
the
the
direction
this
is
going
in
awesome.
So,
cisco,
you
don't
sorry.
A
I
was
gonna
say
I
would
propose
that
if,
if
all
of
you
are
available
tomorrow
at
we
could
maybe
after
we
go
through
any
any
other
topics
during
the
this
week,
we
could
continue
this
yeah.
D
I
would
like
to
yeah
I'll
I'll.
Join
I've
not
been.
I've
been
skipping
that
thinking
it
was,
you
know,
mainly
to
do
with
actual
implementation
and
stuff.
You
know
they've
had
it
covered,
but
never
clarified
that
schema
discussions
happened
there
also
so
I'll
join
tomorrow.
We
can
definitely
continue
this
discussion
I'll
be
happy
to
do
that.