►
From YouTube: 2022-11-02 meeting
Description
cncf-opentelemetry meeting-2's Personal Meeting Room
B
C
So
I
saw
that
the
ram
is
going
to
be
a
little
bit
late.
I,
don't
know
like
if
the
agenda
items
were
added
by
him.
D
But
those
are
probably
from
yesterday
Carry
forward.
A
A
E
C
So
we
discussed
I
guess
we
need
an
unique
ID
for
this.
Having
discussed
using
service
instance,
ID
I
guess
so
like
I,
don't
I,
don't
no
like.
If
impression
ID
is,
is
it
needs
to
be
a
separate
event
from
navigation
or
route
changes,
because.
C
So
page
view
event
is
going
to
be
just
like
the
hard
navigation,
or
is
it
going
to
be
around
change
too?.
D
So
yeah,
that's
what
we
discussed
yesterday
and
so
we
we
we
can,
we
can
discuss
and
decide
it
can
be
either.
D
And
similar
to
the
page
view
like
the
the
non
route
changes
we
will
have
do
we
want
to
have
both
a
name,
an
event
and
a
span
or
just
an
event
or
just
a
span,
so
at
least
in
in
our
product.
Customers
have
to
enable
single
page
application
that
that
this
application
is
a
spa
application,
and
only
then
we
will
generate
the
route
change
events,
because
otherwise
you
know
there
will
be
route
changes
that
are
not
really
relevant.
D
D
A
D
Yeah
go
ahead,
yeah
identifying
the
timing.
Parts
are
I,
think
are
complicated.
So
do
you
want
to
just
leave
it
out
or
do
you
want
to
take
it
up.
C
That's
a
good
question
so
I,
so
we
have
at
neurologically
I
think
we
do
something
similar.
We
have.
We
have
an
event
for
page
view,
but
the
page
view
event
is
only
the
first
page
navigation,
the
hard
navigation,
and
then
we
have
a
separate,
separate
events
for
route
changes.
C
The
you
know,
like
the
page
View
events,
duration,
is,
is
measured
differently
than
for
the
rod
changes.
The
page
view
essentially
is
just
the
time
to
the
window
load
event
through
the
navigation
timing,
information
and
then
the
route
change
is
measured
by
basically.
C
C
A
D
B
B
What's
been
wrong
with
anyone,
but
that's
at
the
point
where
we
change
the
impression
ID
as
well,
because
we're
monitoring
the
location,
change
event.
C
B
A
B
B
D
Yeah
I
think
it
depends
on
whether
we
want
to
create
a
span
to
include
the
timing
aspects,
and
if
we
do,
then
it
makes
sense
for
this
to
be
a
separate
event
from
page
view,
because
that
way
for
the
for
the
for
the
base
Pages,
you
know
you
have
an
event
and
a
span
and
similarly
for
the
route
changes
you
you
have
a
event
in
the
span,
but
if
it,
if
we
decide
to
do
only
an
event
for
for
an
impression
for
the
purpose
of
an
impression
that
hey
the
user
visited
this
page,
then
just
an
event
is
enough
or
a
common
event
is
sufficient.
B
Yeah
and
I
guess
it's
going
to
depend
on
the
application,
whether
they
want
or
need
a
span
like.
If
you're
popping
up
a
dialogue,
it's
pretty
much
going
to
be
like
a
zero
like
bang,
assuming
you're
doing
like
a
route.
You
know
a
hash
on
the
you
think
that
dialogue
always
I,
guess
this
comes
back
to
the
impression
ID.
B
So
the
way
we
have
impression
ID
is
while
we
have
it
as
the
trace
ID,
because
we
don't
have
to
end
the
span
because
we're
just
sending
events
it
we
effectively
have
that
that
impression,
ID
for
the
entire
life
cycle
of
the
page,
which
we
can't
do
when
we,
if
we're
saying
it,
that
is
the
span,
ID.
Sorry
the
trace
ID
of
the
span,
because
we
can't
really
leave
the
oh
I
guess
we
could.
B
You
could
create
a
a
span
at
the
start
and
you
don't
end
it
until
unload
or
a
route
change,
but
then
everything's
going
to
accumulate
in
memory.
So
any
span
events
associated
with
that
don't
get
sent
into
that
last
point:
I'm.
D
Looking
at
the
notes
in
the
ram
events
document
I
think
we
already
discussed
that
the
impression
ID
will
be
a
uuid
yeah.
B
D
It's
not
same
as
Trace
ID
or
span
ID
of
any
of
the
tracerspect,
its
independent,
separate,
ID,
universally
unique.
B
Okay,
so
we'll
just
say
that,
as
a
in
that
case
probably
is
a
a
change
of
resource,
assuming
we
can
get
the
ephemeral
resources
there.
C
So
I
don't
have
one
question
about
this.
Look,
so
if
you
think
of
route
changes
as
page
views
or
impressions,
then
does
that.
C
C
B
I,
don't
think
we
need
to
distribute,
shows
I,
think
the
instrumentation
when
it
comes
to
the
implementation,
will
lead
to
distinguish
how
it
deals
with
those
but
from
a
an
event,
perspective.
I.
Don't
think
we
need
to
I
think
we
need
to
acknowledge
that
these
different
things
exist
and
that
you
know
that
that
will
be
navigation
events
that
will
occur
that
we
may
not
not
want
to
grab,
but
it
comes
back
to
yeah
what,
when
we
report
a
significant
route
change,
how
we're
going
to
report
that.
D
I
I
think
at
least
I
I'm,
not
100
sure
about
this,
but
I
think
we
asked
the
customers
to
to
enable
this
feature.
D
So
if
it
is
a
single
page
application,
then
every
route
change,
irrespective
of
you
know
whether
it
is
you
know
just
showing
a
pop-up
or
not
saying
that
it's
significant
or
not.
You
know
it
will
be
reported.
Yeah.
B
F
So
sorry,
sorry,
for
generally,
that
would
be
a
configuration
based
item
right,
so
we
won't
be
doing
it
by
default.
Correct,
oh.
B
F
C
So
do
we
do
we
want
to
think
of
I
mean
maybe
maybe
I'm
not
recalling
correctly,
but
this
page
View
events
that
we
have
we
have
in
the
spreadsheet
like
do
we
want
to
think
of
every
URL
change
as
a
page
view
event,
or
do
we
want
to
define
a
different
like
a
route
change
event,
that's
different
from
page
view
or
navigation
timing.
B
B
D
Yeah
I
think
there
is
not
a
whole
lot,
that's
different
from
our
page
view.
So
what
do
you
folks
think
about
timing
aspects?
Should
we
put
an
effort
to
get
that
included
or
just
you
know,
ignore
it.
F
D
So
for
a
route
change,
a
route
change
in
a
single
page
application,
it
involves
making
some
API
calls.
You
know
loading
some
black
star
resources,
so
it
it.
It
has
some
the
timing
aspects.
You
know
that
customers
would
be
interested
in
knowing
like
you,
you
navigate
it
to
a
different
route
that
you
know
took
a
few
milliseconds,
so
they
they
will
be
interested
in
knowing
that,
but
the
logic
to
compute
that
is
very
complex.
D
Correct
so
we
can
defer
it
to
a
later
time
and
and
not
bother
about
it.
Now.
D
E
We
just
take
all
of
the
data
that
are
doing
session.
We
look
at
for
the
click
to
consider
test
the
start
of
interaction
and
the
end
of
the
interaction
is
when
there's
just
no
activity
for
like
a
few
seconds.
Yes,
it
doesn't
catch
everything,
because
sometimes
it
goes
wrong,
but
it
works
close
enough.
I.
D
E
F
C
D
Right
so
I
think
the
instrumentation
could
give
a
configuration
option
that
customers
can
overwrite
specifying
hey.
You
know.
If
for
these
route
changes
you
know
ignore,
but
by
default
you
know
you,
you
would
report
all
route
changes
if
a
top
level
flag
is
enabled.
B
B
So
that's
an
implementation
detail,
so
really
it
Martin,
it
sounds
like
you
would
probably
prefer
to
identify
the
fact
that
this
was
a
rap
change
versus
a
yeah.
It
sounds
like
we're
going
to
add
a
flag,
at
least
if
we're
going
to
have
the
single
event
for
the
page
view
that
we
should
have
a
flag
to
say
whether
this
is.
C
Yeah,
so
so
the
way,
the
way
we
handle
it
right
now
and
and
you're
like
I'm,
saying
that
I'm
not
saying
that
it's
correct,
but
the
way
we
handle
it
is
we
so
we
have
this
Legacy
page
view
event,
which
is
just
like
the
hard
navigation,
and
then
we
have.
We
represent
route
changes
by
as
interactions
so
basically
like
the
we
think
of
like
around
changes,
as
always,
starting
with
with
some
some
user
interaction.
C
So
it's
going
to
be
a
click
or
some
kind
of
user
interaction,
and
then
it's
then
it's
associated
with
the
URL
change.
C
So
if,
if
it
happens,
you
know
when
the
page
is
already
loaded,
then
you
have
an
interaction
and
you
have
a
URL
change.
Then
it
represents
a
route
change.
C
F
Right,
yeah,
the
the
timing
you
know
should
be
in
a
different
event,
but
use
it
user
action
should
capture
all
of
this
timing.
Stuff
I
think
yeah.
Our
idea
with
page
view
was
that
just
impression
counts.
That's
that's!
Basically,
it.
F
Most
customers,
the
default
listening
to
route
changes
and
firing
approach
view.
Most
customers
will
not
like
it
from
our
experience,
at
least
they
don't
like
it,
and
then
they
take
control
of
it
because
you
know
like
Ned
was
saying
they
do
some
sort
of
filtering
or
whatever
they
consider
the
page
physically
changed
enough
to
Fire
and
consider
that
a
new
impression
or
something
so
they
really
take
control
and
they
fire
huge
views
themselves.
B
B
So
you
can
say
the
page,
your
top
level
page,
you
would
always
be
zero
and
then
any
location
change
events
could
be
an
ever
increasing
number
or
just
say
true.
False.
B
F
D
I
think
it
there
aren't
many
today,
Boolean
attributes
so
I
think
sequence,
I,
I'm,
okay,
with
sequence,
sounds
you
know
it
could
be
helpful
or
only
thing
is.
You
know
we
have
to
be
careful
that
the
sequence.
B
It
does
complicate
the
an
instrumentation
level
location,
but
you
can
just
say
that
just
starts
at
one.
If
we
say
the
main
page
is
zero.
F
So
here's
my
thinking
on
that
you
know
two.
A
F
Of
the
argument-
and
one
is,
we
don't
know
a
good
use
case
for
sequence,
but
there's
definitely
some
processing
involved
in
things.
It's
not
a
straightforward.
You
know
it's
not
change.
D
Is
it
not,
is
it
not
straightforward,
is
it
complicated
to
implement
you.
F
You
have
to
do
you
have
some
processing
instrumentation,
you
have
two
item
reset
and
you
know
reset
probably
will
happen.
The
heart
rate
valuation
happens,
I.
Think
that's,
okay,
it's
much!
It's
definitely.
You
know,
you
know.
Definitely
the
stamping
the
Boolean.
True
false
is
just
Blended
simple.
You
know
it
can't
go
wrong
at
all
so
and
the
usage
also,
and
then
you
know
you
have
to
let
people
know
how
do
they
use
the
sequence
and
I
just
don't
think.
D
No
I
think
justification
is
easier
for
the
sequence
because
bully-
and
you
know
I
could
argue
that
hey
if
there
is
a
URL
fragment
in
a
hash
argument
in
the
URL,
a
hash
part
in
the
URL,
then
it's
it's
a
route
change.
A
F
E
And
there
are
some
famous
that
prefer
hashem's
events
and
in
some
deployments,
like
the
top
pages.
D
Okay,
I
I
used
to
think
that
any
URL
change
the
browser
will
send
a
request
to
the
server.
But
if
there's
a
URL
fragment
change,
it
keeps
it
on
the
client.
D
E
C
F
For
sure
the
number
or
a
Boolean
is
the
thing.
B
Well,
what
we
could
say:
let's
have
it
as
a
number,
but
it
will
Define
zero
and
one.
So
zero
is
a
page
one
one
will
be
for
subsequent
ones.
So
if
we
end
up
in
a
situation
where
someone
does
want
to
count,
the
the
data
model
already
has
that
ability.
Okay,.
F
F
F
I
guess
the
the
basic
question
is:
do
we
do
we
want
to
capture
type
or
sequence,
that'll
that
will
dictate
the
data
type
yeah.
B
I
guess
Martin
you're
probably
are
there
anyone
here
who
actually
has
that
today?
What
would
your
preference
be.
C
Oh,
my
my
original
thinking
was
to
have
two
separate
events,
one
for
page
view
and
one
for
about
change.
But
if
you
have
just
one,
then
our
current
implementation
would
be
to
have
a
type,
so
there
was
would
be
an
enum.
We
actually
have
yeah
I
mean
the
way
we
handle.
It
is
a
little
bit
different
like
we
have
like
I
said
like
we,
we
handle
it
as
a
interaction
and
we
have
a
type
which
is
like
it's.
C
C
Yeah
I
think
so
we
have
impression
for
page
view,
I
mean
I'm.
Finding
the
way
to
be
honest,
I
guess
I,
guess
the
the
type
would
be
more
explicit.
D
So
another
argument
is:
if
it's
going
to
be
separate
instrumentations,
would
it
be
a
confusion
that
they
are
generating
the
same
event.
B
Looking
further
down
I
I
missed.
So
if
you
go
down
to
like
lines
59
onwards,
would
we
need
to
capture
user
consent,
end
user
ID
session
ID?
That
I
would
say
we
would
would
we
need
to
capture
all
these
events
for
all
these
fields,
for
the
location
change,
and
some
of
that
probably
depends
on
how
the
back
end
is
like
from
a
Microsoft
perspective,
because
today
we
deal
with
everything
as
a
page
view,
we
would
probably
want
to
catch
them.
C
So
I
just
think
I
want
to
say
one
more
thing
about
this
is
so
the
way
the
way
I've
kind
of
envisioned
kind
of
future
is
and
goes
back
again
like
one
more
time
about
the
type
of
the
URL
change
yeah.
So
so,
like
you
know
like
at
the
very
basic,
the
basic
instrumentation,
you
can
obviously
listen
to
the
location,
change
and
then
capture
everything,
and
you
can
you
can.
C
You
can
maybe
interpret
every
every
URL
change
as
a
page
view,
but
you
could
also
have
like
a
framework
specific
instrumentation
that
listens
to
you,
know
events
or
like
some
instruments,
the
framework
and
like
knows
exactly
what
represents
a
URL
change
or
route
change,
and
then
everything
else
would
be
just
noise
right
like
you
want.
You
want
to
distinguish
those
significant
navigation
events,
so
yeah,
so
I
guess
now.
If
you
have
one
instrumentation
that
does
the
base
instrumentation,
that
captures
everything
and
one
that's
framework
specific
like
where
they?
B
I
I
think
you'd
recommend
that
I,
don't
you
either
have
the
location
change
one
or
you
have
your
framework
specific
one.
You
wouldn't
have
them
the
problem
that
you're
talking
about
as
well
as
double
firing,
and
you
have
like
a
rather
change.
One
would
start
would
kick
in
when
it's
finished
it
would
fire
a
PQ.
But
if
you're
listening
to
the
route
change
and
the
router
happened
to
change
the
the
URL,
then
you'd
get
a
second
one,
so
I
I
think
you
would
be
saying
it's
one
or
the
other.
E
B
F
D
On
it,
no
yeah,
you
could.
F
Do
it
yeah
all
right,
so
let
me
just
share
screen.
Anybody
sees
the
stuff.
B
So
I
guess,
then
we
have
to
Define
what
what
the
type
types
would
be.
Yeah
hard
versus
soft
initial
versus
subsequent
locate,
yeah.
F
B
I
I
tend
to
call
them
virtual,
but
I'm,
probably
showing
my
age
with
like
some
of
the
Frameworks
we
used
to
do
in
the
past.
B
D
Yeah,
in
our
case,
it's
we
call
it
a
virtual
page.
D
It's
a
separate
event.
So
in
this
case
it
will
be
same
event.
But
but
what
is
the
attribute
name
going
to
be
just
to
type.
F
D
Okay,
so
it's
it's
going
to
be
an
enum
right
with
zero
and
one
for
initial
page
and
the
virtual
route
change.
A
B
One
yeah
I
think
it
was
a
combination
of
the
two
is
what
we're
getting
what
we're
discussing
there?
Okay,
the
impression
before
you
came
on
around
I,
think
we
were
talking
about
having
that
as
a
one
of
the
ephemeral
resources
as
a
good
okay,.
B
F
Maybe
before
moving
on
to
this
just
real
quick
question,
we
I
think
the
last
thing
that
we
were
working
on
was
using
either
action
right.
We
haven't
completed
this
so
accurate
is
this
where
we
have
to
go
to,
if
we
are
starting
to
look
at
more
schemas
sounds
like
Edward.
Does
anybody
remember.
B
D
On
the
user
action,
I
I
had
some
questions,
I
think
they
are
there
in
the
dock.
Does
it
also.
F
D
Do
we
also
capture
Mobile
in
on
the
mobile
browsers?
You
know,
you
have
touch
events
and
you
also
have
drawing
and
dragging
things
you
know.
Would
you
capture
all
those
as
well
I.
F
Would
think
so?
Yes,
there
was
a
pipe
here,
so
this
is
what
we
put
there
fashion
time,
yeah
just
to
define
the
different
types
of
user
interactions.
D
Okay,
because
if
it's
there's
a
coordinates
and
if
it's
a
drag,
then
you
you
would
need
both
the
begin
and
end.
D
When,
when
you
draw
something,
it
typically
has
a
start
to
coordinate
and
an
end
coordinate
right,
so,
whereas
we
are
capturing
only
one
set
of
coordinates,
the
line
12.
B
I
see
yeah
the
way
I've
done
that
in
the
past.
Is
you
actually
deal
with
the
drag
as
a
two
separate
events,
so
I
think
the.
F
A
D
C
I
was
just
gonna,
say,
I
think
as
a
side
as
a
side
comment.
I
think
this
also
should
be
configurable
too,
because
it
could
be
very
noisy
and.
A
C
Very
useful,
if
we
capture
everything
so
by
default,
like
you
would
capture
just
like
the
basic
ones
like
click
and
but
if
you,
if
you
care
about
things
like
dragging
and
you
could
could
add
them,
I
agree.
Yeah.
F
Yeah
there's
so
there's
this
there's
this
thing
called
tagging
that
we
do
that
we
have
for
internal.
You
know
SDK
I,
don't
believe
our
external
SDK
has
that
which
is
you?
Can
you
know
you
can
Mark
certain
elements,
as
you
know,
do
not
track
so
they'll.
Never
fire
interaction,
events
at
all,
so
that's
yeah!
So
there's
a
DMT
attribute
that
you
can
throw
on
elements
and
then
I
believe
mayor
can
correct
me
if
I'm
wrong.
F
We
have
also
have
a
set
of
elements
upon
which
will
capture
interactions,
not
just
random
interactions
and
stuff.
We
you
know
scroll,
everything
is
configurable
like
Martin
is
asking
things
are
all
configurable,
but
the
configuration
is
at
that
detail
level.
F
You
know
buttons
are
always
included,
anchors
are
included,
and
you
know
a
few
other
things
right.
You
know
I
think
yeah
input,
input,
input,
boxes,
I
think
are
included
also,
so
one
of
the
few
things
that
we
captured
in
an
action
from
and
then
if
the
users
do
not
want
a
particular
thing
that
they
need
to
be
noisy
or
whatever
they
can
throw
a
dnp
attribute
on
it.
F
If
we
honor
those
things,
then
we
can
reduce
the
noise,
but
that
would
require
HTML
to
have
a
bunch
of
stuff
that
that
the
SDK
would
look
for
or
the
instrumentations
would
look
for.
Yeah.
B
Yeah,
these
are
additional
things,
so
this
is
the
click
analytics
plugin
that
I
dropped
in
a
few
weeks
ago.
So
effectively
you
have
a
basic
set
of
config
that
says
what
you
should
and
shouldn't
listen
to,
and
then
when
so,
it
ends
up
listening
to
like,
for
example,
all
of
the
math
clicks.
But
it
looks,
it
looks
at
the
element
that
you're
that
you're
clicking
on
and
then
walks
up
the
the
Dom
and
find
attributes
on
any
of
the
elements
that
tell
it
not
to
do
anything.
It
doesn't
fire
an
event
simplistically.
D
Yeah
I
think
I
think
to
Martin's
Point
I.
Think
in
addition
to
that,
the
instrumentation
should
also
be
configurable,
so
that
customers,
they
know
their
app
and
then
they
can
choose
to
enable
you
know
more
events,
more
situations,
but
by
default
I
agree
that
we
should
only
capture,
maybe
just
the
click.
A
B
Okay,
then,
that's
totally
up
to
the
the
interpretation
itself.
I
think.
C
A
D
D
Any
xhr
activity
that
that
this
event
results
in
how?
How
is
that
captured.
D
E
Because
if
we,
when
use
users,
use
your
instrumentation,
we
user
interaction,
instrumentation,
then
inside
the
event
listener
it
sends
it
sets
the
span
context
to
that
span
of
the
introduction,
so
any
expense
United
inside
that
unit
action
I
have
the
parent
of
user
deduction
automatically.
D
C
B
I
I
see
all
these
as
events
on
the
basis
that
you
know,
people
and
instrumentations
can
create
spans
and
because
we're
acquiring
events,
those
events
will
be
tagged
with
those
you
know
the
parents
relationships
so.
D
Yeah
I
think
here
it's
it's
a
different
from
a
page
view
right,
I
think
here
we
don't
know
if
the
click
event
is
is
going
to
result
in
in
child
spans
correct.
So
you
don't
know
whether
to
start
this
fan.
So
so
so,
in
the
case
of
a
page
view,
we
know
that
we
are
going
to
represent
the
timing
information
in
the
span,
and
so
you
start
the
span
put
that
plan
context
in
your
event
and
then
ship
the
event
out
and
then,
when
the
span
is
complete,
you
send
it
out.
D
F
F
Wouldn't
we
use
the
same
page
in
a
spam
for
all
the
things
that
happen
inside
of
instead
of
it,
isn't
it
the
natural
relationship.
B
Yeah,
the
the
problem
is
that
there
generally
there's
no
span
that
lasts
the
entire
lifetime
of
the
page,
because
nothing
happens
with
the
Spanish
or
you
end.
It.
F
B
B
E
B
Which
is
what
the
impression
ID
is
about.
Yeah.
C
F
B
Which
could
be
done?
Yeah,
I
I
would
think
that
would
be
the
instrumentation's
job
to
start
to
spend
so
logically,
I
I
just
think
of
a
span
as
a
grouping
mechanism.
So
not
everything
will
want
a
grouping
mechanism,
but
some
things
logically
make
sense
to
have
a
grouping
mechanism
for
this.
F
E
F
I
would
assume
you
stamp
the
span
ID
to
that
particular
interaction
event
or
whatever,
and
then
you
decide
to
throw
away
the
span,
because
nothing
really
happened
and
that
will
call
didn't
happen.
It's
probably
something
you
know
something
changed
within
the
HTML
or
something
like
that.
Your
hide
it
either
section.
Let's
say
after
you
click
a
button.
You
decide
to
draw
with
a
Spam,
but
the
interaction
event
has
a
span
ID
in
it
as
a
parent
right,
a
reference
added.
What
happens
to
it?
Is
it?
F
Is
it
not
confusing,
or
is
it
okay
or
do
you
not
stamp
that
span
information
on
the
interaction
event
also.
C
So
we
basically
think
of
so
we,
so
this
is
old,
instrumentation,
that's
really
not
not
based
on
spans
as
Hotel
defines
them,
but
so
like
we
have.
We
we
when
I
say
spam.
I
basically
mean
like
the
top
level
piece
of
work,
which
is
the
interaction
in
this
case
like
the
I
we
I
would
say,
like
the
top
level
span,
is
the
interaction
and
then
like
any
child.
C
Child
spans
would
be
like
the
Ajax
calls
that
happen
during
that
interaction,
and
the
way
is
like
if
and
we
only
actually
capture
so
like
we,
we
started
started
like
the
root
span
for
the
interaction
and
then,
if,
during
while
the
the
span
is
is,
is
active
while
we're
waiting
for
some
work
to
be
finished,
we
also
see
that
the
route
changed,
then
only
then
we
will,
we
will
save
it,
but
so
yeah
so
like
I
mean,
if
you
wanted
to
capture
all
interactions,
then
that
wouldn't
apply
and
I.
C
Guess
you
would
you
would
you
would
end
up
with
a
lot
of
a
lot
of
very,
very
tiny
spans
like
if
they,
if
they
don't
result
in
any
like
Ajax
calls
or
any
additional
work
or
any
child's
fans,
then
it
would
end
up
a
lot
of
yeah
right,
because
then,
then
the
span
would
be
essentially
just
just
the
the
duration
of
the
Callback.
The
synchronous
function,
yeah.
F
B
So
it
sounds
like
we
could
say
the
instrumentation
can
choose
to
create
the
span
or
not,
and
it
could
probably
use
some
of
the
HTML
markup
like
the
DNT.
That
Ryan
was
talking
about
to
say,
well,
I'm,
not
doing
anything.
So,
therefore,
I
don't
bother
creating
a
span
because
yeah,
creating
and
starting
and
stopping
a
span
for
every
single
user
interaction
that
results
in
that
things
will
be
Memory,
wasteful
in
CPU
as
well.
F
C
No
I
was
just
gonna
say
that
the
this
very
much
like
seems
the
same
to
me
as
the
discussion
we
had
about
the
initial
page
load
and
having
a
span
for
initial
page
load,
and
you
know
like
you,
which
is
optional
and
like
if
you
want
to
have
it
great
and
but
it
used
to
you
will
still
get
the
events
and
if
you
have
a
span
in
progress,
then
you
just
attach
these
events
to
the
span.
C
So
the
same
thing
here
like
you,
could
have
video
kind
of
instrumentation
that
just
by
default
generates
these
events.
It
could
be
like
the
interaction
event
and
the
you
know
the
page
view
event
or
the
route
change,
and
you
could
have
instrumentation
that
also
starts
to
span
and
then,
if
they,
if
they,
both.
If
the
span
exists,
then
these
these
events
are
associated
with
that
span.
F
And
and
pardon
me
I
think
I've
forgotten
how
we
would
associate
the
spans
with
the
actual
events
Can.
Somebody
remind
me
of
that.
C
B
As
part
of
the
API
and
fantastic
he
wants
to
please
correct
me:
one
of
the
things
that
the
logging
stuff
will
do
under
the
covers
is
when
the
event
gets
fired.
It
will
get
linked
with
what
it
will
optionally
get
linked
with
the
current
spanner
okay.
F
B
B
F
So
yeah
going
back
to
this
one:
are
we
settle
down?
This
click
coordinates,
then
you
know
if
we
capture
so
scroll
or
something
like
that
now
this
is
most
likely
going
to
be
useless.
Is
that
okay.
B
F
B
Which,
at
some
point,
we
I
think
we
are
going
to
have
to
Define
to
that
level.
I
would
like
to
avoid
doing
that
in
cement
conventions.
We
just
say
we,
the
semantic
mention
says
we
have
this:
the
user
action
event
and
then
we're
going
to
have
other
documentation
say
the
user
action
event
can
represent
these
things.
F
Okay,
I
was
thinking
you
know.
Should
we
create
this
as
a
more
generic,
you
know
thing
for
coordinates.
Instead
of
Click
coordinates
it's
what
I
was
going
for
if
you're
going
to
Define
another,
you
know
format
for
semantics
for
capturing
those
things
for
different
things.
B
G
I,
don't
think
so,
yeah
I,
don't
think
we
do
this
I
just
I'm
curious,
like
what
people
use
this
for.
B
Mostly
it's
for,
if
you're
doing
new
exchanges
and
you're
doing
a
b
testing
you
can
actually
see
if
the
user
is
interacting
with
your
elements,
because
this
is
all
relative
to
the
to
the
page.
So.
G
But
would
we
like
so
far
for
a
click?
Would
we
also
capture,
like
so
I,
think
we
have
like
the
element
path
as.
G
So,
like
that's
a
specific
element,
got
it
yeah
and
yeah
I'm,
not
sure
I
feel
like
it.
I
would
say
like.
B
It
is
problematic
because
of
the
yeah,
depending
on
everything
else,
but
yeah
I
have
had
teams
in
the
past
use
this.
F
Yeah
there's
one
one
example
that
I
can
think
of
as
I.
Don't
believe
that
product
it's
an
intern
but
I've
never
made
it
outside
was
a
plug-in
or
an
extension.
You
can
add
to
your
your
Chrome
or
ads
and
visit
your
page.
That's
already
been
tagged
and
instrumented
with
other
details
that
Extinction
would
essentially
show
hot
areas
of
the
page,
or
you
know
things
that
are
mostly
interacted
and
things
in
real
time.
F
So
let's
say
your
website
or
you,
you
know
it's
already
the
website's
already
out
there
collecting
data
and
things
you
bring
up
the
website
and
then
open
up
the
extension.
F
Of
course
you
have
to
have
the
right
data
access
and
stuff
using
these
click
coordinates
they'll,
be
able
to
overlay
on
your
page
to
say
you
know
the
person
does
your
kicks
that
happen
on
Section,
six
page
so
hard
page
versus
Hearts
areas
versus
cold
areas
and
stuff
that
seem
to
be
pretty
useful
for
it's
essentially
a
b
testing,
but
I
was
talking
about,
but
in
real
time
there
was
one
use
case
that
I
can
think
of
with
click
coordinates.
F
Okay
right
I
guess
we
could
do
this
at
that
if
we
start
looking
at
as.
A
F
As
option
I
think
it's
okay,
the
question
was:
if
we
start
thinking
about
other
things,
then
let's
click
coordinate
always
makes
sense.
So
I
guess
but
drag
start
and
stop
to
separate
events
still
coordinates.
Make
sense.
Should
we
just
name,
it
coordinates.
F
Cool
we
got
three
minutes
left.
D
So
in
the
log
Sig
at
10,
never
we
we're
going
to
talk
about
the
event.data.
D
Okay
and
how
do
we
bring
up
the
ephemeral
resources,
discussion
back
into
life.
B
It
would
be
nice
if
Ted
made
it
today,
because
I
I
think
it's
considering
it's
his
PR.
We
probably
need
the
Pokemon
see
how
he
wants
to
progress.
F
One
thing
and
then
mentioned
the
heart
folks
and
stuff
that
we
didn't
have
a
meeting
yesterday
with
data
analyst
data
scientists
team
within
Microsoft.
We
kind
of
give
them
a
overview
of
the
work
that
we're
doing
here
and
invited
them
to
you
know,
perhaps
join
and
get
representatives
to
join
here.
That
could
help
answer
some
of
the
questions
like
you
know.
What
is
the
use
case?
You
know
and
are
we
doing
the
right
definitions
and
things
essentially
a
different
perspective?
F
Hopefully
you
know
we'll
be
able
to
get
a
couple
of
folks
from
that
team.
You
know
if
we
do
they'll
probably
be
observing
for
a
little
bit
before,
jumping
in
and
helping
the
thing.
You
know
my
thought
is
you
know?
Perhaps
you
know
if
you
are
able
to
get
people
from
the
Olympics
side
participate
in
some
of
these
schema
discussions
from
your
companies
that
might
be
useful,
I.
Think,
okay,
the
the
thought
is,
you
know
oftentimes.
You
know
we
look
at
it
from
instrumentation
inside
we
don't
want.
F
F
D
And
and
Ram
before
he
leave
I
I
think
yeah.
We
will
paint
it
on
the
female
resources,
but
but
I
think
there
was
significant
resistance
about
his
Otep
last
time.
Okay,
I
at
one
point:
un
Nev,
you're
you're,
going
to
talk
to
Riley,
okay.
F
But
yeah
we
can.
We
can
follow
up
on
that
yeah
because.
D
I
think
that's
slightly
in
a
hard
topic,
so
we'll
need
some
support
from
TC
members.
Okay,
yeah.
B
I
think
Martin
raised
the
task
to
have
a
TC
member
allocated
to
to
this
I
I
that
seemed
to
have
got
dropped,
yeah.
B
No
I
think
we're
actually
allocated
someone
else.
I
think
it
was
Josh
comes
to
mind.
D
C
D
A
B
Because
yeah
I
think
that
ticket
was
to
get
a
TC
member
allocated
and
we
were
all
saying
it
should
be
Riley.
But
okay.
F
Yeah
all
right
thanks,
sometimes
yeah,
we'll
we'll
chat
about
it
and
description.
Okay,
all
right
thanks,
everybody,
okay,.