►
From YouTube: 2022-09-28 meeting
Description
cncf-opentelemetry meeting-2's Personal Meeting Room
A
A
B
B
C
C
C
C
All
right
remembers:
where
did
I
put
the
except
spreadsheet.
C
Okay,
so
yesterday
we
finished
talking
about
the
page
view,
largely
it's
defined
I
think
the
next
steps
are
to
go,
and
you
know
Define
start
thinking
about
implementation.
How
we're
going
to
capture
these
things.
You
know
where
some
of
these
common
things
are
going
to
come
from
resources
and
you
know
stuff.
C
We
also
decided
to
pick
up
fetch
this
session,
our
last
next
one.
So
I
did
pre-populate
some
of
these
things
from
vendor
fields
and
try
to
match
them
up
and
things
so
we'll
go
through
down
the
list
and
start.
You
know
populating
these
two
columns
for
our
open,
Telemetry
room
events
makes
sense
right.
I,
I
see
a
few
new
participants.
I
I,
don't
know
if
you
guys
need
a
quick
intro
on
what
we've
been
trying
to
do,
if
not
I'll
just
jump
right
in,
but
if
you
think
it'll
be
useful.
C
All
right,
I'll,
I,
think
I'm
just
going
to
do
some
Jump
Right
In
you're
just
going
from
here,
and
we
we
all
need
a
timestamp
one.
You
know
part
of
discussion,
probably
is
what
time
are
we
capturing
here
for
reference
for
page
view,
we
said
it'll
be
the
time
when
the
event
is
created
by
the
instrumentation
similar
to
that
I.
Think
we
need
a
description
for
this
time
here.
B
So
there
will
be
two
timestamps
right.
This
will
most
likely
be
a
span
and
therefore
it
will
have
two
timestamps.
C
Okay,
I'm
just
so
there's
again
two
events:
it's
you
know
that
we
need.
We
need
to
capture
fetch
turning
and
then
the
span
for
fetch.
Also,
so
you
know
just
recollecting
the
discussions
that
we've
had
before.
B
So
the
event
for
fetch
timing
need
not
have
a
timestamp.
It
would
just
have
a
pointer
to
the
the
span.
C
C
This
is
all
the
pitch
timing.
I
just
realized
we're
not
talking
about
the
actual
spans.
The
corresponding
event
is
what
we
are
referring
to.
So
that
would
be
the
time
when
the
event
is
stamped
correct,
I
think
it's
D2
as
a
teacher
or
somebody
else
typing
that
or
Tito
I
think
yeah
I
believe
that
you
know
this
is
not
the
span
that
we're
talking
about
right.
This
is
an
event
that
we
would
be
firing.
C
B
A
B
And
and
then
for
that
span
there
are
two
timestamps.
A
C
Start
an
end
yeah
yeah,
okay,
yeah,
so
that
I
think
you
know,
will
you
know
it
comes
as
part
of
a
you
know
when
you
fight
a
span,
you
you
you're
gonna,
have
to
do
a
start
and
end
I'll
be
part
of
that
automatically
I.
Don't
think
we
have
to
discuss
it
too
much.
So
this
really
is
the
event
understanding
timestamp,
just
like
this
I
would
think
so.
I'm
just
gonna
grab
that
put
it
here.
B
So
RAM,
if
we
agree
that
there
are
going
to
be
both
event
and
span,
maybe
you
could
have
another
tab
right
away
as
we
are
going
through
this
list
for
for
the
for
this
panel
as
well.
B
C
Could
do
that
I
think
we
did
chat
about
that?
We
decided
we
would
do
the
events
first
then
come
to
spans.
What
else
would
be
defined?
So
all
of
the
standard
stuff
that
the
span
captures
would
be
as
part
of
the
span
event.
Is
there
anything
else?
I
mean
no,
no,
no
and
I
say
Spanish
at
the
span
itself
right.
What
else
would
be
captured
as
part
of
that
so.
B
My
understanding
was
that
the
span
is
the
one
that
will
have
you
know
the
core
information.
The
event
is
exclusive
to
the
timing,
information,
correct.
C
All
the
attributes
that
that
you
want.
B
Yeah
so,
for
example,
the
fetch
timing
event
need
not
even
have
the
URL,
because
all
that
will
be
part
of
the
fetch
span
and
then
the
fetch
timing
will
only
have
the
span
ID
to
point
to
the
span.
A
The
problem
with
not
having
the
URL,
though,
is
if
you,
if
the
event,
gets
sent
separately
from
the
span,
because
the
resource
timings
won't
be
available
until
after
then.
How
do
you
link
them
together
at
the
backing.
B
B
You'll
have
to
somehow
save
that
information
somewhere
and
add
it
to
the
event.
B
Because
you
need
to
link
them
yes,
okay,.
C
B
It's
the
same
there
as
well.
It
are
rather
it's
similar.
C
Yeah
this
one
I
agree,
yeah
I,
think
I.
Think
that
makes
sense.
The
one
thing
I
would
like
to
bring
up
is:
there's
no
guarantee
these
events
would,
you
know
ultimately
make
it
right,
so
if
just
span
makes
it,
is
it
useful
in
and
of
itself?
C
If
just
this
thing
you
know
through
logs
or
spending
if
it's
span
event
anyway,
these
are
two
they're
going
to
be
separate
entities
right
if
one
of
them
is
lost,
the
other
one
Standalone
does
it
make
sense
is
another
thing
that
we
should
probably
think
about,
especially
for
clients.
Things
can
become
lazy.
B
So
in
this
case,
if
if
the
span
is
lost,
but
the
event
remain
is
present,
then
that
is
not
so
useful
because
it
if,
if
that
case,
you're
anticipating,
then
you
might
you
know
we
can
consider
duplicating
the
info
okay,
but
if
the
span
is
present
and
event
is
lost,
then
at
least
you
know,
you
have
the
impression
level
info
that
you
know
that
FH
was
made,
but
you
don't
have
the
timing
info.
A
C
Is
okay,
I
think
that's
fair,
so
we
can
yeah.
We
can
create
a
a
you
know
a
span
right
next
to
it
fetch
span
and
you
know,
go
through
that
also
yeah
yeah.
So
it's
it's
a
unique.
You
know
good
that
we
picked
it
up.
It's
going
to
be
interesting
to
see
how
this
comes
about
yeah.
C
C
C
C
C
We
should
have
the
duration,
so
that's
actually
a
good
question,
so
some
of
the
vendors
have
a
duration
and
then
some
of
them
captured
a
lot
of
these
details
also,
so
that
will
not.
B
Be
derived
with
end
minus
start.
B
B
C
Barbie,
do
you
have
a
separate
duration
field
for
you
guys
today?
Thank.
A
You
I
I
should
have
added
it
in
it's
duration,
underscore
Ms.
However
you're
right,
it's
derived,
so
that's
okay,
I
just
wasn't
sure
what
what
the
norm
I
need
to
look
up
like
what
we
do
in
other
places
in
in
hotel,
but
yeah
I
agree
if
it
can
be
figured
out.
That's
fine.
C
Okay,
I,
don't
even
need
to
I,
don't
think
we
even
have
comparisons
here,
yeah,
okay,
yeah!
This
is
going
to
be
weird.
You
know
stay
with
us
guys,
first
time
we're
doing
something.
That's
got
equal
and
events
equivalent
artifacts
in
both
okay
request,
URL
I
think
that's
part
of
the
span
also
right.
C
Okay,
so
I
guess
I
can
delete
that
cut
yeah.
This
is
going
to
be.
You
know
you
see
what
I'm
doing
right,
we'll
just
put
it
as
part
of
the
span.
What
would
be
the
field
name
again?
Somebody
help
me.
C
Yeah
request
method
is
probably
not
going
to
be
part
of
span
or
would.
B
It
be,
it
will
be
HTTP
dot
method.
You
can.
C
This
is
the
parameters
I
believe
I
couldn't
find
again,
I.
Did
it
late
last
night,
I
didn't
think
too
much
about
it
and
stuff
just
go
on
for
pattern
matching
pretty
much.
This
is
the
HTTP.
You
know
post
and
get
parameters
that
you
would
do
for
Ajax
correct.
Sometimes,
can
you
talk
a
little
bit
about
what
this
field
is
for
you
guys.
C
C
Okay,
I'll
just
leave
it
alone.
You
know:
okay,
yeah,
that's
fine
response
status
would
be
part
of
the
span.
Also
yeah.
C
Okay,
I
think
we
can
go
xhr
metrics.
This
seems
like
what
we
capture
correct,
request,
metrics
and
stuff.
That's
some
others
capture
them.
Let's
see,
oh
I
think
we
are
the
only
ones
that
have
so
much
detailed
stuff.
Now
can
you
speak
to
it
a
little
bit
here?
So
this
is
essentially
we
break
that
object
up
into
multiple
entries,
multiple
attributes
and
save
them
right.
A
Not
today,
so
looking
at
this
and
and
the
general
discussion,
I
I,
think
pulling
this
stuff
out
of
the
event.
Is
the
wrong
approach
like
if
you
look
at
the
by
pulling
this
out
we're
effectively
going
back
down
to
just
the
resource
timing
which,
if
we
end
up
there,
that's
good.
But
if,
if
you
look
at
the
resource
timing,
we
do
need
a
start
time.
A
URL
and
all
the
rest
not
quite
sure
why
it's
empty.
C
C
A
C
B
Think
the
confusion
there
is
that
the
the
only
field,
only
difference
only
addition
in
span
is
that
there
are
two
timestamps,
but
but
otherwise
everything
else
is
present
in
the
event
as
well
and
and
I
thought.
B
Another
reason
for
splitting
these
as
event
and
span
was
only
for
the
purpose
of
modeling,
but
eventually
it
is
going
to
be
a
runtime
option
or
a
in
an
option
that
individual
users
would
configure
where
the
event
information
could
be.
You
know
either
part
of
the
span
as
a
span
event,
or
it
could
be
a
standalone
event.
So,
and
in
that
case,
if
it
is
part
of
the
span,
you
don't
need
to
duplicate
the
info.
C
I
I
don't
know:
if
we
can,
you
know
it
could
be
a
user
configurable
thing
you
know.
I
would
like
to
you
know
for
us
to
discuss
a
little
bit
about
a.
C
Each
vendor
I
understood
if
it's
even
if
it's
a
vendor
option,
then
you
know
the
systems
that
are
mapping
at
the
back
end
I,
don't
think
we'll
be
able
to
a
vendor,
chooses
to
I,
don't
understand
how
we
would
put
things
back
together
or
extract
information
and
stuff.
C
It's
sometimes
it
could
be
in
two
separate.
You
know
all
they've
used
all
the
words.
The
events
are
the
two
separates
of
the
databases
or
two
separate
artifacts.
Let
me
just
call
them
like
that.
It's
gonna
be
two
separate
roles
right
and
sometimes
they
could
be
in
the
same.
It
just
doesn't
make
sense
we
have
to
be.
C
We
have
to
concretely
say
if
you
see
and
let's,
let's
assume
we're
going
to
use
logs
as
my
concept
for
sending
these
these
good
good
events
from
the
page
view
and
then
spans
for
sending
the
spans
okay,
they
have
to
be
in
the
same
bucket.
They
say.
Sometimes
you
have
to
look
at
the
logs
to
you
know
for
your
information.
You
know,
oh,
if
the
vendor
configured
it
to
put
them
in
the
two
separate
ones
you
have
to
look
at
both.
It
just
doesn't
make
sense.
B
Maybe
nav
you
can
speak
to
that,
because
I
think
the
current
implementation,
you
know,
puts
them
as
event
in
the
span.
It's
just
that
you
know
there
are
individual
events
and
we
are
consolidating
into
one
event,
and
my
preference
for
our
company
also
is
to
you
know,
have
it
as
one
event
but
inside
the
span.
A
A
In
the
event,
once
we've
identified
the
data
that
we
want
in
the
event,
then
we
can
have
the
discussion
of
okay,
we're
going
to
send
this
like
this,
which
is
why
I'm
saying
we
should
keep
the
timestamp
and
the
URL
and
everything
here,
because
conceptually
this
is
the
data
at
some
level
we're
going
to
have
to
have
a
mapping
recommendation
for
saying,
okay.
A
Well,
when
there's
a
fetch
request,
we
are
going
to
recommend
that
it
gets
sent
as
a
span
and
then
it
may
have
these
other
events
attached
to
it
and
then
from
a
back-end
mapping
perspective.
This
is
where
you
get
your
fields
from,
but
if
we
haven't
got
the
fields
defined
in
the
first
place,
it
starts
getting
messy
like
we're.
Getting
messy
here
we're
jumping
between
two
tabs
trying
to
copy
things
around
rather
than
saying.
C
C
I
think
that
makes
sense
yeah
yeah
thanks
thanks
for
clarifying
and
kind
of
resetting
your
position,
yeah
I
agree.
Let's,
let's
not
talk
about
implementation,
and
just
say
these
are
the
things
that
we
want
to
capture
mechanism.
As
a
next
step,
we
can
Define
yeah.
A
C
I
agree
any
any
thoughts,
folks
I'm
about
to
hit
Ctrl
Z
and
get
everything
back
and
and
then
use
a
description
say.
Oh,
this
could
be
the
recommendation
or
you
know
something.
This
could
be
the
expense
span
stuff
and
so
on
and
so
forth.
C
Let's,
let's
go
back
to
this
and
say
you
know:
timestamp
is
not
not
a
thing,
it's
basically
span
start
and
span
and
that's
it.
C
C
Then
that
one
was
the
thing
that
I
just
added
okay,
so
HTTP
method.
C
No
parameters,
so
we
just
leave
it
like
that
and
we're
not
going
to
define
it
and
that's
what
we've
been
doing.
If
it
is
vendor
specific,
we
just
said
I
think
it's
response
status.
Did
we.
C
Yeah
I
think
we're
we
can.
We
continue
talking
about
this
Etc
metrics
resistors
nav.
This
is
where
we
kind
of
went,
went
to
two
different
tabs
or
whatever
did
you
have
a
preference
of
break
them
apart.
A
So
these
are
effectively
like
the
the
stuff
that
Martin
originated
with
the
the
resource.
Timings.
This
is
effectively
is
what
this
is.
So
apart
from
the
Ajax
not
missing,
that
was
that
was
just
a
flag
that
I
added
to
say:
I
couldn't
get
I,
couldn't
find
the
resource
timing
for
this
event,
so
that
pick
up
okay,
they
were
potentially
available
but
affected
the
the
resource,
timings
weren't
available
in
either
in
a
timely
manner
or
at
all,
because
there
is
a
limit
to
the
the
resource
timing,
size
so
I
think
for
the
page
navigation.
A
We
split
these
out,
but
I
listed
them
specifically
as
ajax.perf,
because
we
do
send
them
as
a
Json
coded
blog,
which
I'm,
assuming
is
what
is
happening
with
that
metrics
one.
A
B
It
is
a
map
currently.
C
Okay,
so
the
question
yeah,
the
question
is:
do
we
have
Quinn
here
for
AWS,
don't
think
so?
So
can
some
other
vendors?
You
know,
perhaps
you
know
T2
and
will
we
talk
about?
How
do
you
guys
do
you
know?
Do
you
not
need
these
things
I'm,
just
wondering
we
don't
capture.
A
I
think
what
we
capture
is
especially
around
like
the
HTTP
specific
stuff
is,
is
really
more
about,
like
all
of
the
metadata
that's
available
like
about
the
request,
so
yeah
I
would
just
really
like
to
capture
basically
anything
that's
available
about
the
the
request.
Gotcha.
C
Just
sounds
like
that.
This
is
basically
the
research
timing
or
the
page
timing.
Stuff
Martin
worked
on,
should
we
just
transfer
it
over
here
essentially
say
we're
going
to
have
that
has
broken
down,
looks
like
Cisco
and
Microsoft.
C
C
A
Don't
know
where
I
think,
if
you
send
us
excite
our
metrics
down
to
the
Ajax
perfect,
that's
probably
where
it
lines
up
yeah
and
then
yeah
his
teacher
has
got
there.
I
think
we
just
pushed
them
over
and
we
we
met
them
as
individual
things.
A
C
A
C
Okay,
do
you
have
a
link
to
Martin
Stark
I?
Don't
have
it
handy,
yes
or
or
if
you
can,
you
know
you
know
paste
them
here.
That'll
be
great,
yeah,
okay,
so
I'll
just
Mark
these
paste
them
and
then
you
know
change
the
color.
C
Okay,
we
could
you
know,
while
that's
you
know,
we
can
do
it
later,
I
guess
now,
yeah
I
think
we've
got
to
do
the
discussion
here,
but
this
is
again
the
event
type
and
stuff
yeah
I.
Don't
believe
we
need
to
capture
that
separately,
especially
if
it's
going
as
a
span
right
or
do
we
need
need
the
type
or
something
like
that
capture.
C
It'll,
be
the
outgoing
outgoing
request,
or
something
like
that
question
is
we
could
most
likely
be
sending
open,
inventory,
Event
open,
Telemetry,
you
know
in
a
span
when
we
do
that.
Do
we
need
to
say
this
is
for
fetch
timing
somewhere.
B
We
can
get
to
that
later,
RAM
because
I
think
that's
the
part.
We
differ
right
whether
these
events
will
be
sent
as
this
information.
These
messages
will
be
sent
as
spanner
event.
Okay,.
C
C
C
This
I
think
we
should
agree
on
what
it
is
for
pitch
standings.
All
right.
Are
we
going
to
use
pitch
timings,
slash,
reject,
standings
or
something
like
that.
A
How
about
we
just
call
it
fetch
now,
because
I
I
highly
suspect
when
we
look
at
we
do
the
same
thing
for
xhr,
it's
going
to
look
the
same.
Okay,
it's
from
Microsoft's!
First
software
perspective.
They
are
the
same,
but
it
would
be
interesting
to
see
from
other
vendors.
Okay,
the
only
thing
we
have
extra
is
I
think
we
are
identified
that,
whether
it
was
xhr,
Rock,
h.
C
Yeah,
do
we
need
a
description
here?
We
basically
we
put
a
description
only
if
you
know
that
is
deemed
an
event,
I
mean
the
thing
that
we
want
to
capture
this
product.
Hotel
I,
want
to
say,
yeah,
that's
a
vendor
specific.
We
don't
even
need
to
care
about
it.
C
Okay,
this
is
a
status,
this
is
interesting
status
code
and
then
we
have
a
success.
True
false
I,
don't
know
you
know
I
remember
from
several
years
ago.
This
was
you
know
it's
a
sort
of
a
duplicate
thing.
People
didn't
want
to
do.
You
know
the
parsing
of
the
response
status
quo
and
figure
out
if
it
was
success
or
not
at
the
back
end.
C
That's
why
it
was
there,
but
do
we
do
we
believe
we
need
that
anymore?
No.
A
C
Actually,
this
is
a
sorry
and
I
I.
Think
I'm,
you
know
convicting
myself
and
I
step
in
the
discussions
are
coming
back,
so
this
possibly
there
are
systems
would
return
at
200.
But
still
the
operation
was
a
failure.
I
don't
think
it's
Norm,
but
when
they
do
that
the
success
is
the
one
that
you
know
you
that
defines
whether
it
is
successful
or
not.
C
A
Correct
we
will
okay,
but
as
part
of
the
first
mapping,
then
we
can
probably
say
no
but
I.
Actually
yeah
I
suspect.
C
Okay,
so
let's
leave
it
out
for
now.
C
This
is
the
schema
version.
If
I
remember
correctly,
that's
what
we
discussed
for
this
guy
here,
there's
a
version
field.
What
happened
to
it?
I,
oh
I,
guess
we
didn't
capture
it.
That's
interesting!.
C
C
It
was
it
was,
we
are
yes,
oh
I
see
what
you're
saying,
okay,
but.
A
Yeah
we
did
a
discussion
yesterday.
This
is
the
the
schema
version.
So
what
version
of
the
event
is
it
and
yeah
I?
Think
we
end
up
dropping
it
on
the
basis
that
I
mentioned
pigrens
approach,
the
lack
of
aversion
means
is
version
one
and
then
later
we
could
add
the
version.
A
C
This
one
here,
yeah
I,
think
we
still
have
to
go
through
in
a
page
view,
and
you
know
there's
a
few
things.
I
just
noticed,
I,
don't
know
if
we
did
go
through
these
guys,
the
SDK
that
were
epoc
sequence.
We
did
not.
C
Yeah,
let's,
let's
not
jump
between
these
two,
but
you
know
this
version.
Let's,
let's
chat
about
that
version.
So
that's
the
schema
version
right,
yeah,
yeah,.
A
Sorry,
which,
at
this
point
we
we're
still
defining
version
one
so
but
having
a
placeholder
for
where
this
would
be
Morgan.
Do
you
recall
for
other,
like
spans
logs
events,
Etc
the
schema
discussion?
I
know,
J
MCD
was
talking
about
having
an
attribute
which
is
like
a
schema.version
or
something
but
I.
Don't
know
where
that
landed.
It's
something
like
that.
I
think
that
did
get
merged
in
tigrant
knows
more
than
I.
Do
I
wasn't
really
involved
in
that.
A
But
yes,
there
is
some
strategy
in
hotel
for
schema
versioning.
That
involves
applying
schema
versions.
I
think,
like
a
schema
version,
number
two
spans
and
metrics
and
things
that
come
in
and
then
there's
a
schema
document,
that's
defined
somewhere.
So.
C
A
C
I
also
think
the
same
applies
for
this
other
version.
There
policies
guys
for
jumping
back
and
forth,
but
this
one
should
be
pretty
straightforward.
Stop
it
there,
okay.
C
All
right,
this
is
a
custom
field
for
us
agreement.
A
Yeah,
this
is
like
embedded
user
properties,
hello,
yeah,
yeah,
nested,
attribute
level.
C
C
C
C
And
we
need
to
you
know
we
need
to
capture
this.
You
know
people
agree
right.
B
C
So
this
is
an
interesting
thing
right.
What
about
Pia,
consents
and
stuff?
You
know:
do
we
we
have
to
do
some
sort
of
a
scrubbing
before
we
capture
those
so.
A
C
C
C
I
think
you
know
this
is
a
header
of
things.
I
think
I
split
them
up
like
that
I'm
just
going
to
remove
that
word
feel
and
then
we
can
discuss
about
the
other
two
request:
headers
and
response
headers
I.
Don't
think
anybody
else
is
capturing
them
at
that
level.
C
Okay,
all
right,
so
if
it's
going
to
be
the
span
for
HTTP,
then
it
should
automatically
be
here.
I
think
not
talking
about
Expedition
for
a
second.
This
is
useful
stuff
right.
So
can
you
know
wondering
why
other
vendors
don't
need
that
at
the
moment-
or
you
know,
probably
you
know
some
use
case
or
something
like
I
don't
know.
B
So
this
is
typically
if
the
the
headers
are
like,
like
there
are
some
applic,
you
know
some
products.
You
know
that
capture
the
entire
request
and
response.
B
I
think,
in
that
case,
capturing
the
entire
set
of
headers
is
useful,
but
specifically
in
the
context
of
application
monitoring.
How
do
we
use
these
headers
like,
for
example,
if
let's
say
customers
want
to
derive
a
metric
out
of
one
of
the
values
of
one
of
the
headers,
then
maybe
it's
good
to
pass
it
on
so
I'm
just
thinking
loud
how
it
can
be
helpful
from
a
product
perspective.
A
There
could
be
information
in
the
address
like
which
CD
and
Edge
nodes
are
the
request,
or
something
like
that,
which
would
probably
be
it
will
probably
be
elevated
better
by
user,
which
state
you
know
which
headers
they
want,
because
you
don't
really
want
all
letters,
okay
and
I
think
Ryan
mentioned
yesterday.
In
the
case
of
like,
if
you're
doing
experiments
you
might
be
capturing
the
the
header
based
on
that.
B
Okay,
yeah,
we
don't
send
it
today,
but
I
I
see
the
value
yeah.
We
so
yeah.
B
I
see
so
then
do
you,
how
do
you
like
you,
drop
certain
headers
yeah.
A
Yeah,
so
we
have
a
that
we
accept
and
then
users
can
provide
extra
headers
if
they
want
to
collect.
C
A
And
this
is
where
we
start
getting
into
it
like
it's
not
talking
about
ages
ago,
having
repeated
http.request.header
and
then
the
name
is
just
wasteful.
This
is
where
the
nested
attribute
aspect
of
would
come
in
handy,
but
for
now
we
can
have
it
like
that
in
terms
of
like
this
is
how
it
is
how
it's
currently
defined,
but
when
we
come
to
say
okay,
these
are
the
things
we
want
to
include
and
then
then
we
have
the
discussion
of
how
we
want
to
include
them.
C
So
would
it
be
value,
basically,
that's
it
right.
A
No,
the
value
would
be
attribute
value.
The
key
itself
is
in
the
attribute.
C
Key
gotcha
I
want
to
say
you
know:
how
do
we
capture
this
here?
Is
it
the
corresponding?
You
know
it's
a
kvp.
That's.
A
C
Yeah
so
I
was
going
to
say
this
hope
it
makes
sense.
C
A
C
C
Let
me
let
me
clarify
so
the
last
night
on
it
transferred
this
over.
So
this
really
should
be
the
it's
this.
This
denotes
a
kvp.
That's
it
it's
a
collection.
So
if
you
remember
the
top
level
they've
had
I
can
go
back
here.
C
Smoking
I
don't
see.
Oh
there,
you
go
sorry,
that's
a
New
Relic
when
that
go
remote.
There
you
go.
So
this
is
how
you
know:
Nev
captured
it
when
I
transferred
it
over
to
the
thing.
I
just
said
here
to
Value
it's
a
kvp,
it's
basically
what
you're
typing
out
there
teacher.
C
C
A
C
Yeah,
so
the
the
question
is:
do
we
need
looks
like
honeycomb?
Has
some
equivalence
on
the
these
these
things
sort
of
like
metadata
about
the
device
user
and
things
do
we
need
that
as
part
of
I.
A
Yeah,
okay:
is
there
a
mechanism
to
do
that.
A
In
theory,
resources,
but
that's
the
problem
of
ethereal
ones:
okay,
yeah,
so
the
circle
that
doesn't
change
like
SDK
version,
that's
just
a
resource
but
yeah
for
the
ones
that
could
change
like
session
ID,
user,
ID
or
I'd,
say
no
authenticated
user
ID.
Not
necessarily
that's
the
ephemeral
one
from
Ted
that
we
need
to
get
to
the
answer
of,
because
that
will
determine
whether
we
put
it
in
the
event
or
right.
A
C
Got
it
got
it
yeah,
perhaps
what
we?
What
we
could
do
is
you
know
Define
this
here
redundant
and
then
you
know
we'll
realize
this
is
what
the
global
stuff
is.
You
can
put
them
on
in
global
and
then
for
every
event.
We
can
say
this
needs
to
be
part
of
it.
C
C
A
I
I
think,
rather
than
jumping
around
that,
just
like
the
other
approach.
Let's
just
say
this
is
probably
Global
and
then
we
come
back
and
do
a
Passover.
This
and
say
this
is
common
between
them
all.
A
If
it
doesn't
change,
then
it
can
go
into
a
resource.
If
it
does
change,
then
we
need
the
ephemeral
resource
and
if
not
that,
then
it's
going
to
go
in
the
event.
C
A
B
Given
that
we
have
10
minutes
left,
I
want
to
maybe
use
this.
You
know
to
switch
to
the
other
document.
I
I
have
added
some
information
in
in
your
other
document.
C
B
The
bottom
yeah:
can
you
see
if
it's
appropriate?
You
know
you
had
something
in
mind,
not
this
document,
the.
C
No
no
I
just
want
to
capture
you
know.
A
current
state
yeah
give
me
one
quick
sec.
B
B
Yeah,
so
if
you
go
to
the
bottom
yeah
so
so
here,
I
filled
I
took
the
information
from
the
page
view.
The
completed
things
and
I
put
it
here
and
I.
C
B
Table
called
common
resource
browser,
common
resource
and
I
have
categorized.
You
know,
fixed
attributes
versus
varying
attributes.
B
A
B
In
that
case,
then,
we
can
keep
it
in
the
flat
format
in
the
other
Excel,
where
the
session
ID
can
just
stay
in
in
the
same
document.
Don't
worry
about
so.
C
Yeah
I
think
that
I
think
that
makes
sense
yes.
Well,
you
know,
like
never
saying,
will
simply
put
it
in
the
comments.
That'll
be
helpful
for
you
or
anybody
else.
That's
I'm!
Sorry
yeah.
Anybody
else!
That's
transferring
over
here
and
actually
defining
these
these
common
stuff
things
I.
Think
yeah
I
agree.
B
What
I
mean
is
fixed,
so
it's
a
varying
attributes,
let's
say
screen
width
and
Screen
height.
Now,
while
the
browser
is
open,
you
know
you
can
resize
the
browser
in
in
a
same
session.
So
so
a
few
messages
can
go
with
a
certain
dimensions,
and
you
know
a
few
minutes
later.
You
know
you
could
in
the
same
session
you
could
send.
You
know
the
different
dimensions
sure
so
it
will
change.
B
It
could
potentially
change
the
same
with
session
ID
and
the
yeah
and
and
the
user
IDs.
Also
I
put
it
in
the
varying
attributes,
because
it's
possible
that
users
can
log
in
and
log
out,
got.
C
It
question
is:
what's
the
what's
the
value
in
splitting
them
up.
B
There
is
currently
a
limitation
in
open
Telemetry,
where
we
cannot
represent
varying
attributes
in
a
resource
resources,
and
we
have
an
Otep
suggesting.
You
know
that
we
had
support
for
this
varying
attributes.
It's
stuck
and
and
that's
the
context
where
we
you
know
earlier,
we
were
asking.
You
know
somebody
from
the
TC
to
help
us.
C
C
The
one
thing
I
remember
is
you
know.
Sometimes,
if
we
add
anything
here,
we
need
to
take
notes
somewhere
and
you
know
come
back
and
make
sure
that
that
makes
it
that
new
field,
or
whatever
makes
it
here
somewhere.
C
Perfect
yeah:
this
is
cool
okay,
Switching
back
here
then.
C
That
makes
sense,
so
this
is
basically
as
part
of
that
translation
is
where
we
need
to
identify,
but
I
guess
that's
Global,
attributes
right.
Okay,
yeah
make
sense.
C
So
now,
let's
talk
about
these
things
here.
Do
we
I
think
we
need
these
things
as
part
of
the
event
right.
A
We
do
yeah
so
session.
Id
is
listed
divisively,
we
talked
about
yesterday.
That's
just
unique
device
ID
and
it
depends
on
the
type
of
runtime
you're
playing
with.
So,
if
you're
running
in
react,
native
or
an
electron
or
something
where
you've
got
access
to
the
device,
ID
that
could
be
the
physical
device
had
the
device
type
is
really
like.
Your
browser,
mobile,
that's
just
kind
of
extreme.
C
So
the
interest
of
time
and
be
efficient,
what
I
can
do
is
Mark
the
ones
that
I
need
to
copy
over
from
phdo
and
I
can
do
that
so
device
ID.
Please
put
some
coloring
there
and
then
device
type
I,
don't
know
if
we
captured
it.
Maybe
I
can
just
go
through
all
these
things.
C
A
Yeah,
it's
actually
customer
for
the
user
as
well.
This
is
just
a
nested
object
that
the
user
gives
us
these
values
right.
We
just
passed
them
on.
Okay,.
B
Where
do
we
list
the
individual
timing
attribute?
Do
we
need
to
list
them
like,
for
example,
if
you
scroll
down
yeah.
A
The
big
Bluetooth
I've
placed
it
in,
but
I
haven't
actually
called
out
the
names
yeah.
C
B
And
that
I
remember
in
one
of
the
calls
you
mentioned
that
today,
like
we
don't
need
to
include
every
attribute
that
the
browser
provides.
You
only
want
to
correct
selectively
pick,
which
ones
you
want
yeah.
C
A
Yeah
but
I
think
we
want
to
Define
it
as
flexible,
so
that
if
another
attribute
comes
up
in
the
future
that
the
browser
happens
to
support,
it
will
also
appear,
which
I
think
is
what
Martin
tried
to
Define.
In
terms
of
saying
it
was
as
per
the
w3c
spec,
but
I'll
go
through
and
have
a
look
at
those
again.
B
C
Okay:
okay,
the
the
next
round
I've
gone
through
these
things,
vendor
specific
stuff
right,
Cisco,
all
of
your
fields
are
discussed
or
mapped
Yes,
except
Etc,
metrics.
B
C
So
Cisco
is
good.
We
can
go
through
a
little
time,
we'll
we'll
probably
do
this
the
next
time
we
meet
go
through
this
or
if
in
a
Peruvian,
did
you
want
to
do
that?
If
you
guys
have
some
Cycles
go
through
that
and
see
if
you
can
either
prepare
or
sort
of
figure
out,
equivalence,
that'll
be
great.
So
if
we
went
to
with
the
current
state,
these
are
the
things
that
are
capturing
today.
That
might
not
be
part
of
it.
C
So
that's
what
we
will
discuss
next
week.
C
Okay,
cool
all
right
thanks
everybody
for
joining
appreciate.
It
yep
see
you
all
next
week,
bye.