►
From YouTube: 2022-09-07 meeting
Description
cncf-opentelemetry meeting-2's Personal Meeting Room
A
Okay,
I
think
we
can
probably
get
started
if
you
have
anything
that
you
would
like
to
talk
about.
Please
add
it
to
the
agenda.
A
There
are
two
things
that
I
that
I
didn't
want
to
talk
about.
One
was
based
on
hold
on
it's
based
on
this
conversation
or
this
pr
that
the
sanctuary
and
ebenezers
are
working
on.
A
A
topic
that
we
discussed
at
the
very
beginning
of
of
this
group,
which
was
how
to
identify
or
classify
browser
and
mobile
telemetry,
and
at
that
time
like
we
thought
we
would,
we
would
introduce
browser,
specific
or
client-side
specific
attributes
that
would
identify
the
telemetry
but
like
an
app
app
dot,
name
app.version,
and
but
that
was
essentially
it
was
not
not
it
was.
A
It
was
shut
down
by
the
tc,
basically
at
that
time,
in
in
favor
of
having
a
common,
it's
in
favor
of
using
service,
dot,
name,
service
path,
version
for
everything.
Even
though
service
does
not
semantically
redescribe
well
client-side
applications.
A
You
know
I
there
is
actually.
I
just
noticed.
A
A
This
this
this
pr
that
sometimes
she's
working
on
that.
Basically
it's
about
the
question
there
is
should
be.
You
know,
like
four
back
ends,
ingesting
browser
telemetry.
Should
they
classify?
How
should
they
classify
it?
Should
they
classify
it
by
the
presence
of
browser
attributes
or
should
should
they
classify
by
the
value
in
process?
Runtime
name
attribute.
A
So
that's
what
I
wanted
to
discuss.
If
does
anyone
have
any
thoughts
or
opinions
or
questions
about
this
topic.
A
A
And
then
the
other
thing
that
I
wanted
to
just
bring
up
really
quick,
although
I'm
gonna
share
my
screen
real
quick.
A
The
other
topic
is
that
so
we
have,
I
have.
What
I
have
been
working
on
is
the
the
logs
logs
api
analog
sdk
in
the
javascript
repo.
I
think
the
logs
api
is
very
close
to
being
merged.
It's
just
failing
some
tests
right
now
that
I
need
to
figure
out,
but
it's
getting
approvals
now,
which
is
great
with
that.
I
started
working
on
the
sdk.
A
And
I
do
have,
which
is
this
is
work
in
progress.
It's
in
my
fork
right
now,
if
I'm
looking
for
help
on
this.
So
if
anyone
is
interested
in
in
this
work,
please
let
me
know
I
can
give
you
access
to
to
my
fork
and
we
can
collaborate
on
this.
A
A
That's
all
I
had
nev
nev.
I
thought
I
saw
that
you
had
some
discussions
yesterday,
anything
worth
bringing
to
this
group.
C
No
santosh
was
there
yesterday
as
well,
which
is
where
we
talked
about
his
pr
ram's
just
back
for
holidays.
It's
not
here
again
today.
It'll
probably
take
him
the
week
to
catch
up
on
all
his
emails
before
he
picks
up
on
the
rum.
C
C
C
But
yeah
that's
pretty
much
it.
The
sandbox,
which
was
about
to
put
in
here,
is
still
blocked
waiting
for
the
cla
to
be
signed.
C
I
had
to
pivot
back
to
some
internal
work,
so
I
haven't
really
done
anything
with
with
hotel
I'm
trying
to
get
the
foundation
done.
My
the
basic
plan
is
come
now
next
semester,
which
is
the
end
of
this
month.
A
Yeah-
and
I
I
had
to
just
for
for
everyone's
for
transparency
like
I
I'm
having
to
scale
back
a
little
bit
on
hotel,
I
still
can
dedicate
some
time
to
it,
but
not
as
much
as
in
the
past.
So
yeah
I
mean
dedicating.
Resources
to
this
work
is
challenging
for
sure.
A
Okay,
the
only
other
thing,
that's
that
I
wanted
to
ask
is
you
know
bryce
you,
you
started
coming
back
to
this
group.
I
know,
sir,
you
are
interested
in
mobile,
I
don't
know
if
anyone
else
on
this
call
is
interested
in
mobile,
but
like
we
so
this
work
that
we're
working
on.
I
think
the
next.
A
You
know
the
thing
you
know
our
focus
is
gonna,
be
the
cement
or
is
the
semantic
conventions
for
browser
which
ram
is,
is
driving
and
I
wonder
like
if
we
should
think
about
having
similar
discussions
from
mobile
and
if
there's
anything
that
I
can
help
to
facilitate
that.
D
And
is
the
desire
to
have
all
of
them
in
the
same
dock?
I
I
feel
like
that.
It
dock
is
going
to
explode,
it's
just
it's
already
pretty
long,
there's
already
and
like
web
and
mobile,
don't
always
match
up,
except
for
maybe
sort
of
network
events
for
them.
Yes,
yeah.
C
And
that
was
the
intent.
Is
we
needed
to
start
somewhere
to
see?
What
is
the
crossover
so
yeah?
I
wasn't
getting
to
it
so
ram
decided
to
jump
on
in
and
help
out
there.
C
Yeah
we've
gone
right
around
circles,
always
going
to
define
the
events,
but
one
of
the
big
things
we
need
to
define
the
events
to
get
moving,
especially
now,
we've
got
the
api
spec
going
and
martin's
pushing
through.
With
the
you
know,
an
implementation
we
can
start
building
on
top
of.
D
Yeah,
I
think
once
we
have
some
momentum
around
the
events,
we
should
absolutely
pick
up
the
conversation
of
mobile
and
see
how
it
sort
of
fits.
Because
I
mean
they
don't
look
all
that
different.
There
are
some
things
that
are
unique
to
mobile
in
general,
but
you
know
a
screen
transition
and
a
page
change
for
all
practical
purposes
would
look
and
feel
similar,
but
the
sim
con
might
be
different,
like
a
page
is
not
a
construct
in
mobile
for
the
most
part.
C
E
D
E
Class
or
no,
I
just
wanted
to
mention
that
today
it
is
my
first
time
joining
this,
this
group
and
I'm
also
interested
in
mobile,
so
I'm
also
working
at
elastic
same
as
price,
so
we'll
I
guess
we'll
be
taking
a
look
at
those
semantics
together.
C
Yeah,
the
dock
as
it
is
at
the
moment,
is
really
just
trying
to
gather
what
the
common
events
might
be
and
what
everyone
calls
the
fields.
So
we
can
then
start
saying.
Well
you
call
it
fred.
I
call
it
joe,
let's
call
it
bob
in
hotel
so
that
we've
got
mappings
for
everyone
who
wants
it.
If
it's
common,
let's
make
a
comment
for
everyone.
B
Caesar's
working
on
android.
C
Okay,
yeah,
I
don't
think
the
java
sdk
supports
logs
properly,
yet
it
definitely
doesn't
support
nested
attributes,
which
is
one
of
the
reasons
for
building
the
events.
Api,
analogs.
E
A
Yeah,
I
think
I
think
the
jack
berg
was
the
maintainer
on
java.
I
think
he
started
working
on
the
logs
sdk.
I
can
check
with
him
how
far
he
has
gotten
with
that
and
once
once
that's
available
I
mean.
Maybe
there
would
be
good
good
opportunity
like
to
do
some
prototypes
for
for
android
some
android
events.
B
Yeah,
there's
still
no
logs
development
on
this
or
the
swift
side
of
things.
B
I
I
don't
have
to
talk
with
nacho
the
metric
sdk.
The
metric
implementation
is
really
out
of
date,
so
that
might
be
a
higher
priority
than
logs.
F
On
the
topic
of
logs
and
events
for
android
and
ios,
is
there
any
list
of
events
that
you
plan
to
capture?
Maybe
we
could
start
speaking
out,
you
know
what
attributes
you
know
it
must
have
and
what
names
it
should
use.
B
B
F
Yeah
yeah,
I
think
I'm
I
was
talking
about
consistency
between
ios
and
android.
B
B
A
And
it
probably
probably
crashes
like
how
do
you
do
either
android
or
swift
capture
anything
but
crashes
right
now,
probably
not
right.
B
Yeah
not
at
the
moment,
it
is
something
that
we
want
to.
We
want
to
add
in
the
future,
but
I'm
not
sure
how
it's
going
to
fit
into
hotel.
E
Yeah,
we
definitely
are
gonna
need
it.
I
think
we've
mentioned
something
about
maybe
having
some
sort
of
special
locks
for
crashes,
but
having
we
haven't
discussed
it
further.
F
I
I
I
think
for
for
crashes,
I
I
like
at
least
internally
at
our
company.
We
were
thinking
that
there
has
to
be
a
generic
mechanism
to
upload
the
logs
to
some
destination,
and
then
you
generate
an
event
with
a
metadata
of
that.
F
So
so
we
could
have
a
standard
between
ios
and
android
that
whenever
there
is
a
crash,
so
it
is
typically
picked
up
next
time
the
app
opens.
F
So
at
that
point
it
will
upload
that
to
some
destination
and
the
url
and
the
other
metadata
you
know
can
be
then
passed
on
in
an
event.
B
Yeah
should
we
just
add
it
to
the
existing
one,
the
hotel
or
the
semantic
conventions
that
we
already
have.
F
I
think
that
is
browser
the
one
which
nav.
C
I
think
ram
was
looking
to
create
a
central
source
because,
like
we've
already
got
loads
of
things,
so
I
think
right,
you
know,
add
at
least
the
the
name
there
you
know
and
we'll
probably
eventually,
let's
put
them
out,
but
let's
just
leave
at
least
for
the
main
table.
There
that's
got
the
events
listed.
F
F
So
there
is
a
document
that
nav
and
ram
from
microsoft.
I
think
they
have
started
where
we
have
listed
from
from
different
vendors,
the
list
of
events
as
we
capture
in
our
respective
organizations
in
respective
products.
F
So
we
have
listed
the
details
of
the
payload
and
then
we
are
going
to
identify.
You
know
how
to
present
them
in
a
common
way,
but
this
table
has
the
names
of
the
events,
so
you
could
add
an
entry
here
but
link
it
to
your
new
document.
F
B
Oh,
I
see
I
get
you
so
sir
says
that
she
has
a
lot
of
events
or
they
have
already
defined
a
lot
of
events
over
there.
Do
you
think
you
could
fill
out
this
table
and
add
all
those
in
here?
So
we
know
what
you
know
start
from
a
common
place.
D
Yeah
we
can,
we
feel,
like
I
filled
up
this
document
before
but
yeah.
We
can
add
about
a
bunch
of
lists.
I
mean
yeah
sure
we
can
add
a
bunch
of
them
here.
We've
actually
already
defined
them
in
our
product
documentation.
Does
this
need
to
go
to
yet
another
dock,
or
is
it
okay?
If
it
links
to
our
documentation.
F
I
think
it
is
fine
yeah.
A
Yeah,
I
think
we'll
be,
I
think,
we'll
be
good
to
do
like
once
once
we
once
we
actually
get
into
these
details
and
to
start
prototyping
these
these
events-
and
it
sounds
like
in
for
for
mobile,
at
least
like
we
will
be
able
to
do
it
only
in
android.
A
So,
okay,
so
any
other
thoughts
or
comments
on
this
topic.
D
F
So
so
there
are
multiple
topics
on
on
that.
One
right,
I
think,
firstly,
is
anyone
using
that
class
today,
our
like?
Is
it
going
to
be
a
breaking
change.
F
F
I
I
feel
that
process
runtime,
I
think
they
were
introduced,
looks
like
not
long
ago,
maybe
six
months
to
one
year
ago,
but
they
are
not
so
applicable
to
the
client
side.
You
know
they
are
more
applicable
to
the
server
side
processes.
On
the
client
side.
We
don't
use
the
term
process
so
much
even
on
the
mobile
side.
Correct
me.
If
I'm
wrong,
I
think
they
are
typically
only
called
apps,
but
but
internally.
A
So
I
mean
I
think
I
had
I
had
a
pr
a
long
time
ago
on
adding
the
the
app
app
attributes-
and
this
was
just
got
discussed
there
and
I
think
it
was
armin
who
said
who
was
basically
arguing
that
you
know
you
still
have
around
time.
You
know
in
the
browser
right
so
you're,
just
saying
that
the
runtime
is
the
browser.
F
F
But
in
the
presence
of
browser
namespace
attributes,
do
you
really
need
you
know
this
runtime?
I
would.
F
I
I
saw
it,
and
you
mentioned
that,
what
if
the
browser
you
know
at
some
point
in
future
stops
giving
even
the
client
things
data
right.
A
Yeah
so
my
my
question
was
all
of
the
browser
attributes
right
now
are
are
optional
and
and
so
like
there
is,
and
some
of
them
are
based
on
the
new
api,
the
user
agent,
and
we
know
that
you
know
some
browsers
do
not
support
those,
and
then
there
is
the
fallback
browser.useragent,
which
is
the
legacy
user
agent
string.
A
So
I
don't
know
like
if
there
are
user
agents
out
there.
That
would
not
be
able
to
provide
either
of
those.
F
A
F
I
think
if,
if
the
browsers,
you
know
change
their
specs,
I
think
we
can
change
it
in
the
future.
We
will
have
to
change
it
in
the
future,
but
until
then
I
I
wouldn't
expect
browsers
to
stop
supporting
that
api.
F
Yeah,
but
the
other
client
hints
api
navigation
that
the
navigator
object
provides.
That
will
continue
to
be
there
right
now.
C
For
now
yeah
the
fact
that
it's
in
the
spec,
it
will
continue
for
a
extremely
long
time.
I
think,
will
be
a
lot
older
and
grayer
by
the
time
it
gets
deprecated.
C
Yeah,
italy,
if
they
deprecate
it
and
remove
it,
it'll,
be
because
of
some
security
issue
that
someone's,
exploiting
or
they'll
replace
it
with
something
else.
Yeah
a
bit
like
the
brands
today,
like
the
user
agent,
is
getting
frozen,
but
you
can
look
up
the
the
brands
and
the
high
entropy
basic
functions
to
get
the
equivalent
data
to
figure
out
the
version
yeah.
F
And
martin,
if
that
api
goes
away,
it
will
have
larger
implications
and
we
are
a
smaller
component
that
gets
affected
right.
The
I'm
sure
there's
a
lot
of
business
logic
that
the
applications
have
that's
built
around.
You
know
this
api.
C
But
yeah
in
in
general
for
the
process,
runtime
name
there
may
be
some
back-end
systems
that
pivot.
Based
on
that,
we
don't
at
microsoft
at
this
point
because
we
don't
use
otel
for
browser,
so
it
may
be
a
case
of
still
populating
at
least
the
process,
runtime
name
with
browser,
but
then
still
have
the
browser
attributes
as
well.
Although
user
agent
is
debatable
because
that's
going
to
come
down
with
the
request
anyway,
anyway,
there's
going
to
be
lots
of
discussions
versions
already
problematic,
so.
A
So
so
here's
my
question
like
do
we
need
to,
like
I
think,
santosh
was
proposing
the
deprecating
the
process
run
time
for
browser.
Do
we
need
to
do
that?
I
mean
do
we?
A
C
C
It
doesn't
really
do
very
much
at
all
yeah
because
it
gets
you
know
petabytes
a
day,
so
it's
in
really
the
downstream
consumers
and
how
they
process
that
so,
ideally
having
a
process,
runtime,
name
saying,
browser
or
javascript
or
whatever
yeah
having
something
that's
common
across
all
signals
and
all
platforms
would
be
useful,
but
either
way
when
we
go
to
deprecate
this,
whether
we're
removing
it
or
adding
it.
It's
like
there
may
be
someone
who's
already
playing
with
it.
C
F
It
is
not
a
mandatory
field.
Anyways
right,
I
think
like
do.
We
have
a
confirmation
that
all
the
spans
generated
by
the
by
the
backend
instrumentation
they
generate
this
attribute.
C
Been
going
to
take
every
every
bit
of
code
to
figure
that
out
and
if
you
don't
have
the
resource
detector
loaded,
then
it's
not
going
to
be
there,
so
that
depends
on
what
distros
are
including
and
what
people
are,
including,
for
example,
parsley
have
now
open
sourced
their
definition.
This
came
up
in
the
javascript
sig
last
week,
where
previously
ikea
built
their
own
who
were
built
on
top
of
fastly.
I
haven't
gone
to
look
out
there
to
see
whether
they
populate
these.
These
values
now
they're
not
using
hotel.
C
C
Yeah
so
you're
gonna
find
vendors
who
would
like
to
say.
Okay,
I
want
to
distinguish
between
my
my
server
my
function
app,
my
python,
my
java
server
versus
my
client
or
clients.
C
So
again
it's
it's
going
to
be
up
to
the
destroyers.
C
Like
what
what
does
your
company
do?
Do
I
use
this
at
all
today.
F
No,
no,
we
we
haven't
yet
no
not
yet.
I
think
we
will
be
based
basing
it
on
what
we
decide
here.
C
F
So
sorry,
what
did
you
say?
What
do
you
use
to
distinguish.
D
I
think
there's
something
with
like
an
sdk
version
and
it
shows
up
as
javascript
web
sdk.
F
Yeah
those
are
again
required
attributes
at
resource
level,
so
they
will
always
be
there.
A
So
one
thing
to
note,
though,
is
that
the
source
of
open
telemetry
data
may
not
be
always
the
official
sdks.
D
F
So
this
is
a
sample
resource
from
the
browser
today
and
I
think,
is
talking
about
these
three
attributes.
So
these
four
are
mandatory.
D
A
So
so
I
so
maybe
the
language
is
what
you're
looking
for.
A
C
Okay,
so
how
about
how's
that
going
to
be
hit
by
or
is
it
just
feel
that
so,
okay,
multiple
trains
of
thought
going
on
in
my
head
a
couple
weeks
ago
there
was
a
spec
meeting
about
supporting
enum
values
in
otlp
and
they
were
going
to
ban
the
use
of
names
in
preference
of
the
of
the
number.
C
This
particular
case
looks
like
the
the
number
is
the
string.
So
it's
probably
okay.
F
F
So
yeah,
I
think
when,
when
somebody
exports
in
the
json
format,
you
know
these
will
be
numbers,
but
then
there
is
no
clarity
on
the
custom
values
that
are
not
part
of
the
enum.
Today.
I
think
I
think
they
are
they're
they're,
saying
that
it
needs
to
be
added
to
the
enum
before
you
can
use
it,
because
otherwise
there
is
no
way
for
the
producer
and
the
consumer
to
to
coordinate
on
what
that
custom
value
could
mean.
E
E
Sorry,
no,
I
just
want
to
add
if
it
helps
like
in
the
case
of
andre.
Actually
this
this
week
we've
been
discussing
about
process
name
and
inversion,
because
we
also
are
sending
you
know
the
language
and
sdk
language
and
name
and
that's
that's
useful,
but
I
I'm
really
not
sure
about
how
how
it
would
work
with
web,
but
at
least
in
the
case
of
mobile,
it
could
help
to
provide
a
bit
more
of
information
about
the
the
environment
itself,
where
the
code
is
running.
E
So
for
the
case
of
android,
we
could
mention
stuff,
such
as
the
android
api
level,
which
it's
only
available.
You
know
at
runtime.
It
comes
with
the
os
yeah,
so
it
could
be
helpful
for
for
that
kind
of
stuff.
But
in
the
case
I
don't
know
if,
in
the
case
of
web,
every
signal
has
a
user
agent
parameters.
So
maybe
is
that
the
issue
that
it
could
be
redundant
with
user
agent.
F
I
think
here
we
are
talking
about
the
browser,
so
there
there
there
is
a
component
called
browser
detector,
which
you
know
I
checked
with
a
few
folks
nobody's
using
it.
Maybe
you
know
the
the
author
who
introduced
it.
You
know
they
are
using
it
and
and
that
that
adds
a
bunch
of
process.runtime.name
process.runtime.version.
F
You
know
with
with
some
values
from
the
browser,
so
so
that
I'm
I'm
suggesting
that
we
stop
using
in
in
favor
of
you
know
these,
because
it's
it's
essentially
the
same
data.
That's
added
there.
E
Okay
got
it
yeah,
but
that
seems
more
for
definitely
more
for
web
environments.
F
Yeah
so
so
the
question
is
like
I:
I
suggest
that
we
drop
those
attributes
pending
I'll
I'll,
find
I'll
reach
out
to
the
author,
somehow
and
and
then
find
out
what
they
think.
I'll
actually
have
them
comment,
but
from
the
folks
here
is
there.
Do
we
have
a
consensus
or
do
we
still
need
to
discuss.
A
So
I'm
still
not
clear
like
what
the
recommendation
should
be
for
classifying
on
the
back
end
like.
I
definitely
agree
that,
like
the
the
process,
runtime
version
should
be
should
be
probably
removed
from
the
spec
to
be
the
user
agent,
because
that's
the
duplication
and
they
tend
to
be
long.
F
Yeah
that
we
will
never
know
right
because
the
back
end
there's
there's
no
spec
about
it.
A
F
F
Going
we
could
bring
it
up,
we
could
bring
it
up
in
the
in
the
news
working
group,
that's
being
started
on
the
instrumentations
and
70
conventions.
Bajaj,
I
think
there
his
agenda
is
includes
the
backend
considerations.
F
I
I
I
think
I
read
in
his
proposal
that
today
you
know
there
is
literally
very
little
consideration
given
to
the
back
ends
during
the
design
of
the
you
know
these
agents.
A
F
We
we
could,
we
could
skip
this
deprecation
part
to
a
later
time.
We
could
defer
it.
We
could
proceed
with
this
current
pr
to
introduce
these
fields
and
we
could
get
back
to
the
deprecation
part
later.
C
So
where's
the
resource
detector,
which
briefing
are
you
talking
about.
D
F
F
F
So
this
comment
here
indicates
that
you
know
somebody
has
to
explicitly
invoke
this
detector.
C
Looks
like
was
added
six
months
ago.
Well,
at
least
the
history
only
goes
back
six
months.
I
thought.
F
F
Yeah
so
I
looked
at
who
added
it.
D
F
Okay,
so
so
maybe
martin,
we
could
proceed
with
the
pr
for
now
and
come
back
to
the
deprecation
part
later
then,
once
we
have
more
information
sounds
good.
F
Have
any
yeah,
I
have
one
other
quick
topic.
F
I'm
adding
it
to
the
agenda,
so
I
I
created
an
issue
asking
for
comments,
so
I
like,
we
want
an
attribute.
F
For
specifying
the
current
page
url
in
all
the
spans
and
events
generated
you
know
from
that
page,
so
so
today
you
know
we
do
have
a
span
for
the
xhr
and
fetch
user
interaction
and
several
other
you
know
spans,
but
it
doesn't
tell
you
you
know
on
which
page
they
were
generated.
F
F
Ideally,
you
know
that,
like
for
all
the
telemetry
generated
from
that
page,
you
know
you
you
have
to
send
it
only
once
so
it
makes
sense
to
send
it
at
the
topmost
level.
But
given
that
that's
not
progressing,
so
I
recommended
that
we
put
it
in
the
scope
attributes.
F
So
the
advantage
of
the
scope
attributes
is
that
you
know
every
instrumentation
library.
F
If
it's
generating
multiple
signals,
it
will
at
least
be
common
to
them.
So
on
a
given
page,
if
there
are,
let's
say
three
xhrs
and
two
fetch-
and
maybe
you
know
a
few
other
events
so
for
each
type
it
will
be
repeated
only
once
it
will
be
specified
only
once
so.
That's
my
proposal
and
I
think
t2
mentioned
in
the
ticket
that
you
know
he
still
prefers
that.
F
Rather,
you
know
that
we
should
once
again
make
an
effort
to
push
for
the
ephemeral
resource
attributes
because
it
may
not
be
easy
to.
You
know
change
this
later,
but
I
I'll
reply
to
his
comment,
but
I
think
the
way
the
spec
for
the
scope
attributes
is
written
today.
F
It
is
it
it
is
a
replacement
to
the
span
attributes
as
in
if,
if,
in
the
absence
of
the
same
attribute
in
the
span
level,
you
know
the
attribute
from
the
scope
actually
flows
into
the
span.
F
F
So
we
could
define
a
similar
language
when,
when,
whenever
we
come
up
with
the
you
know,
the
ephemeral
resource
attributes
that
they
are
a
generalized
version.
F
A
If
not,
then
this
might
be
probably
our
best
option,
but
I
think
you
know
I
think,
from
implementation
perspective
it's
going
to
have
to
since
it's
something
that's
common
to
all.
Instrumentations
then,
like
I
went
like
we
probably
would
need
something
like
a
like
a
processor
that
just
adds
it
for
automatically
to
all
instrumentations
on
that
level.
F
Yeah
yeah,
I
think
in
that
issue
we
talked
about
some
implementation
too,
like
it
has
to
be
part
of
some
base
class
where
it
applies
to
everything.
A
And
then
the
other
thing
that
I'm
not
hundred
percent
clear
about
this
is
the
by
the
compatibility
like
in
the
future
like
if,
let's
say
that
we
get
ephemeral,
resources,
resource
attributes
in
the
spec
then,
but
we
we're
already
sending
them
a
scope
attributes.
Is
that
an
issue
or
not?
I
don't
know.
F
Yeah,
so
so
that
that's
what
I'm
saying
that
so,
given
that
the
scope
attributes
are
a
general
version
of
the
span
attributes,
we,
we
should
extend
that
at
the
resource
level
as
well,
where
the
at
least
the
ephemeral
attributes
are
are
more
generic
version
of
the
scope
attributes.
F
F
And
and
and
once
the
back
ends
implement
the
spec,
you
know,
then
they
should.
They
should
accept
in
this
attribute
at
the
resource
level.
A
Yeah,
okay,
I
feel
like
this
is
like
has
broader
implication,
that
I
think
we
probably
need
like
someone
from
tc
just
to
give
us
their
advice,
but
I
mean
in
theory
it
sounds
makes
sense
to
me:
yeah,
okay,
okay,.
F
Actually,
speaking
of
tc,
I
wanted
to
ask
you,
you
know
there
was
one
more.
F
Like
yeah,
you
remember
this
issue.
F
Because
this
will
ensure
that
we
get
a
tc
member
assigned.
A
Yeah,
so
I
did
brought
it
up
to
the
to
the
tc
a
few
weeks
ago
and
they
said
they
were
going
to
discuss
it
in
their
in
their
meeting,
but
I'm
not
sure
if
anyone
has
been
actually
I.
I
did
not
attend
the
specification
meeting
last
couple
weeks
so.
C
Hasn't
come
up
in
the
spec
meeting
riley's
most
likely
the
tc
member
will
be
allocated.
I
think
daniel's
on
the
tc
as
well.
So
we
can
ask
him
in
there:
okay,
yeah.
F
So
could
you
take
it
up,
martin,
with
daniel.
D
A
A
Does
anyone
have
any
other
last
last
minute
comments
or
questions?
If
not,
then
we
can,
we
can
call
it.