►
From YouTube: 2020-11-10 meeting
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
A
B
A
C
By
the
way,
I
don't
know
if
this
was
the
case,
but
10
minutes
ago
I
cannot
tax.
I
could
not
access
to
the
to
the
dock
of
this
call,
but
unexpectedly
I
could
after
a
few
minutes
I
could
actually
same
here.
It
was
weird
yeah.
B
I
was
able
to
edit
it
last
night,
but
after
you
post
that
message
in
getter
carlos
I
I
took
a
look
and
it
looks
like
somebody
changed
it
to
restricted
access,
only
short
list
of
people.
So
I
I
added
you
like
literally
a
few
minutes
ago
and
then.
D
B
Clicked
on
the
link
to
change
it
to
public
editable
like
it
used
to
be
so,
but
I'm
not
the
guy,
that
did
it.
So
I
don't
know
where
that.
C
F
C
C
C
C
B
H
Wanted
to,
unfortunately,
I
can't
stay
on
the
whole
call
I
gotta
run,
but
at
the
maintainers
call
yesterday
we
were
discussing
just
ga
timeline
and
I
thought
it
might
be
a
good
idea
to
just
reiterate
that
here
and
maybe
get
some
clarity
from
the
group
before
we
go
put
this
on
the
status
page.
H
Yeah
tweaks
so
just
to
review
what
we're
actually
gaining
here
and
what
this
plan
will
be.
First
off
is
trying
to
ga
tracing
there's.
Actually
some
terminology
question
I
have
here.
I
was
talking
about
ga
and
tracing
in
gitter
and
that
surprised
some
people,
because
people
were
thinking
ga
meant
open,
telemetry,
1.0
with
tracing
and
metrics.
H
So
my
first
question
is:
what
terminology
should
we
use
here,
and
this
has
already
been
cited?
That's
great.
I
just
want
to
make
sure
that
it's
clear
to
everyone
before
I
put
it
out
there
when
we
say
we're
tracing
is
ga
or
gen
tracing
is
a
has
a
release
candidate
tracing
his
ga.
H
I
H
I
think
ted
is
asking
about
something
else:
yeah,
so
there's
the
spec
freeze
and
then
we're
going
to
put
out
release
candidates
eventually
for
for
tracing
only
so
there's
a
set
of
packages
that
we
want
to
have
finalized
on
the
tracing
side
that
are
going
to
get
there
well
before
we
finalize
metrics,
but
we
do
want
to
communicate
to
people
that
we
believe
this
is
final.
It
could
be
that
we
just
stick
to
release
candidate
and
we
don't
say
ga
anything.
H
A
I
think
the
question
here
is:
we
need
to
make
at
some
point
the
big
announcement
that
we
are
generally
available.
1.0
is
generally
available.
Do
we
want
that
to
that
announcement
to
be
simultaneously
for
tracing
and
metrics,
or
we
are
okay
with
saying
open,
telemetry
tracing
1.0
is
now
generally
available,
but
not
metrics.
J
I
would
suggest
the
the
lighter,
because
I
I
think
for
some
of
the
languages
they
need
to
support
logging
integration
eventually,
and
it
doesn't
make
sense
for
them
to
announce
ga
for
both
tracing
and
metrics
and
later
they
announce
another
ga
for
logging.
If
the
philosophy
is
the
ga
always
refer
to
everything,
yeah.
E
Yeah,
I
think,
what
yeah
what
you
say
I
would.
I
would
argue
that
this
is
not
the
right
group
to
even
be
discussing
this
this.
This
is
something
I
think
that
the
the
governance
committee
and
the
technical
community
to
decide
on
what
the
language
is
going
to
be
right
and
that
you
can
figure
out
what
to
do
with
it.
E
I
mean
we
could
talk
all
day
about
it
and
we
have
a
ton
before,
but
when
morgan
comes
along
in
other
meetings
and
says
no
we're
not
gaying
until
this
date-
and
I
mean
it's
clear
that
the
everyone
needs,
the
the
people
at
the
top
quote-
top
need
to
be
aligned
and
provide
a
consistent
message
to
everyone
about
what
the
strategy
is,
because
I
think
that's
the
real
problem
here.
H
We
have
a
general
idea
of
what
we're
doing
right,
we're
trying
to
finish
tracing
first
and
we're
going
to
try
to
finish
metrics
and
then
we're
going
to
try
to
finish
logs,
and
there
has
always
been
the
desire
to
have
1.0
mean
tracing
and
metrics,
because
that
was
the
original
scope
of
this
project
with
the
merger
of
open
tracing
and
open
senses.
So
I
will
say
we
are
definitely
sticking
to
that.
H
Logging
is
something
that
will
come
post
1.0
since
that's
a
new
signal,
that's
being
added,
there's,
surely
going
to
be
other
new
things
that
we
add,
post
1.0
as
well,
but
getting
getting
tracing
and
metrics
at
the
door
means
we've
completely
replaced
open
census
and
everything
that
that
it
did,
and
people
can
definitely
use
this
in
production.
H
For
those
two
signals,
what
I
think
there's
less
clarity
on
is
how
we
talk
about
things
leading
up
to
that,
and
mostly
I
want
to
make
sure
we
get
our
terminology
right,
but
people
don't
want
to
talk
about
it
in
a
bigger
group,
I'm
happy
to
to
just
go
off
with
the
tc
and
the
gc,
and
we
can.
We
can
sort
this
out
there.
I
do
like
maintainer
input
on
on
this.
So
do
people
have
any
strong
opinions
about
what
what
standard
or
good
naming
conventions
would
be
before
we.
J
Go
off
and
think
about
it.
So
so
one
thing
I
want
to
call
is
from
from
the
like
the
c,
sharp
and
c
plus
plus
a
lot
of
people
ask
the
status
simply
because
they
want
to
decide
whether
they
could
take
a
bat
on
the
api
or
we're
going
to
make
breaking
changes,
and
I
I
think,
if
it's
not
ga,
it's
totally
fine.
We
need
some
terminology
to
let
people
know
this.
J
J
Of
ga,
so
my
understanding
of
ga
is
something
that
that
we
give
the
commitment
to
customers
we're
not
going
to
break
them
and
the
ga
has
to
come
with.
Well
we're
going
to
start
the
support
and
what's
the
end
of
life.
So
that's
what
people
would
expect
and
whether
the
product
like
open
telemetry
as
a
whole
product
has
the
feature
parity.
I
think
it
might
be
something
different
from
ga,
but
no
strong
opinion
here.
H
Right
yeah,
I
think
in
general,
the
term
generally
available
is
kind
of
confusing
in
a
project
like
this,
when
we're
actually
like
there's,
never
any
point
where,
like
everything,
all
gets
released
everywhere
at
once,
but
I
agree
with
you
riley
that
one,
the
point
of
switching
from
to
from
zero
to
one
is
to
indicate
that
the
1.0
branch
has
long-term
support
guarantees
of
some
kind
for
for
the
public
apis
on
tracing
and
metrics.
H
J
Yeah
and
another
thing
I
think
in
general,
we
follow
the
semantic
convention
like
december
2.0,
so
they
have.
They
have
a
good
definition
of
some
of
the
like
major
version,
minor
version
with
the
semantic
there,
so
it'll
be
great.
If
we
can
align
instead
of
telling
a
very
different
story
and
confuse
people
yeah.
H
I
mean
I
do
think
this
matches
semver
like
we
won't
be
changing
the
major
version
number
yeah
until
we
break
these
public
apis,
which
hopefully
will
not
be
for
a
very
long
time.
Okay,
all
right!
Well,
that's
I
think,
that's
enough
for
me.
I
will.
I
will
round
up
the
gcntc
to
make
a
I'll
get
a
proposal
together
and
bring
it
back
to
the
group
after
we
collaborated
a
bit
on
it
cool
all
right.
Thanks
see
you
later.
B
Okay,
I'll
move
on
to
the
next
agenda
item
triaging
with
new
issues.
K
B
C
Spec
logs
after
ga
well,
this
is
not
only
related
to
logs,
but
there's
yeah
and
a
very
specific
intersection
point
basically
he's
saying
that
we
should
like
flush
is
expected
to
be
something
like
an
expensive
operation.
So
the
the
the
fact
we
are
using
force
as
part
of
this
operation.
It's
like
it's
not
needed.
A
Why
is
this
saying
vlogging
well?
Is
this.
C
C
L
I
I
know
I
think
we
should
just
mark
it
as
allow
for
ga
and
we
can
discuss
this.
I
I
I
think
it's
it's
an
important
topic,
but
it's
not
critical.
Even
if
we
stay
with
this
name,
I
don't
think
it's
going
to
be
the
end
of
the
world.
C
C
No,
no
probably
just
leave
a
q.
I
will
leave
a
comment
that
this
isn't
on
the
sdk
side,
not
on
the
api.
This
is
small.
This
also
applies
to
traces.
B
I
just
add
that
one,
this
yeah,
okay,
there
you
go,
it's
got
the
sdk
and
then
we
will
move
on
to.
B
I
I
think
this
is
this
is
something
that
we
will
offer.
I
think
people
will
be
able
to
configure
both
exporters
otlp,
which
is
push
and
prometheus,
which
is
pulled.
E
A
E
A
A
A
This
is,
and
I
think
it's
p1
or
people
probably
also
sdk.
Yes,.
N
N
Really
trying
to
pin
down
metrics
like
the
next
month,
I
think
we
should
piece
this
open.
It's
pretty
critical!
It's
it's
going
to
be
a
problem.
If
we
don't.
B
And
as
a
p1
who's,
a
suitable
assignee.
For
this
me.
B
C
C
It
feels
like
allow
for
ga.
I
think
this.
This
is
like
set
of
improvements.
We
can
take
not
on
the
api
or
sdk
itself,
but
on
the
specification
side.
C
F
Yeah,
I
would
make
that
at
least
allowed
for
ga
and,
as
you
see
on
the
right,
there's
already
a
pro
request
targeting
this
and
it
has
been
been
accepted
so
far,
there's
no
one
rejecting
it.
So
that
looks
good.
C
F
Yep
sounds
good,
but
no
doubt
it
would
so
you
wouldn't
reject
it
right
right.
Okay,.
D
B
Here
we
go.
Okay,
actually
do
we
have
time
you
guys
want
to
take
two
more
I'm
just
running
over
the
time
box.
N
P
D
F
B
B
Following
on
from
that
of
the
two
that
are
not
metrics
related,
I
think
we're
trying
to
come
to
resolution
on
from
the
maintainers
meeting
for
using
the
trace
api
contacts
and
baggage
freeze
in
this
br.
We
talked
about
it
at
the
maintainers
meeting
and
I
think
we
got
some
momentum,
but
just
a
second
ago.
A
A
The
the
one
about
the
attributes
now
values
in
the
attribute
in
the
arrays
and
the
other
is,
I
can't
remember
so
far
about
the
attributes.
I
think
I
just
merged
one
of
them.
Okay,
there's
there
are
two
of
those.
P
B
So
this
is
on
track
for
merging
later
today.
Is
that
is
that
the
expectation
yes.
P
B
This
last
issue
that
I've
put
up
this
last
item
I
put
up
on
the
agenda.
I
noticed
that
this
pr
was
marked
as
after
ga,
but
then
it
got
some
movement
and
changes
that,
even
though
it's
marked
as
after
ga.
I
A
And
it's
better
to
close
that
versus
it's
a
reminder,
right
program.
It
sends
you
a
warning:
why
are
you
not
paying
attention
so
the
intent
is
not
too
close.
The
intent
is
to
remind
you,
but
but
if
you
keep
because,
if
you
keep
ignoring,
I
guess
maybe
it's
not
important
right.
Nobody
cares,
but,
but
here.
F
I
think
that
the
reminder
works
really
well,
because
it
notches
people
and
then
quite
often,
people
pay
attention
to
it.
We
could
disable
closing.
That
would
be
an
easy
thing
to
do
or
make
it
dependent
on
the
label,
but
I
would
keep
the
reminder
that
one
works.
A
I
A
B
Go
and
priority
p2
p3.
N
Q
F
B
Okay,
yeah:
okay,
thanks
that
totally
takes
care
of
my
items,
I'm
happy
to
keep
sharing,
but
if
someone
else
wants
takeover
driving,
just
let
me
know
and
I'll
relinquish,
we
can
go
on
to
the
next
item.
C
Actually,
I
would
love
if
you
keep
on
driving
only
I
mean
at
least
sharing
your
screen.
That's
very
helpful!
Thank
you!
So
much
all
right.
So
basically,
yes,
I
I
added
the
next
item,
which
is
what
are
the
required
issues
for
tracing
for
for
for
freezing
it?
It's
basically,
two
of
them
just
wanted
to
discuss
it
to
discuss
them
briefly,
yeah.
The
first
one
is
the
otlp
board.
I
Okay,
I
did
some
experiments.
I
was
able
to
share
the
port
and
I
was
also
able
to
in-test
to
do
some
of
the
certificates
mtls
stinky,
but
I'm
not
very
experienced.
Does
anyone
have
good
experience
with
mtls
that
can
confirm
that
my
test
is
doing
the
right.
A
Yeah,
let's,
let's
call
for
help
broadly
on
guitar
as
well.
If
no
one
is
able
to
help,
I
will
try
to
spend
some
time
myself.
Okay,.
C
L
Can
you
quickly
scroll
to
the
comment
on
the
pr?
I
think
it
was
suggested
to
always
produce
a
new
span
id
and
then
we
could
store
the
last
sample
span
id.
L
L
So
I
think
always
generating
a
new
spell
id
is
something
we
can
do,
but
I
don't
think
we
can
still
link
the
last
sampled
parent
in.
N
I
The
original,
or
something
like
that
I
need
to
to.
We
need
to
take
a
look
at
the
nodes
and
probably
document
here
and
see
you
ready
to
comment.
I
L
How
we
want
to
go
for
it,
I
can
adapt
the
pr
to
say
to
or
generate
a
new
id
or
make
can
make
an
alternative
pr.
Whatever
you
like.
C
Yeah,
well,
I
guess
the
first
thing
I
saw
that
you
commented
something
last
week,
so
I
need
to
poke
judy
to
review
what
you
said
once
that
is
discussed,
then
we
can
decide
how
to
go
forward.
C
Thank
you
great
yeah,
it's
very
good,
because
these
are
the
two
last
items
so
fantastic.
Okay.
We
can
go
to
the
next
one,
a
man
requesting
clarification
on
pr
comment
from
by
the
way
by.
I
C
A
A
I
R
I
S
I
can
I
can
talk
to
morgan,
I
mean
I
work
with
him
so,
like
let
me
know,
let
me
know
what
message
you
need
to
get
to
him
today
and
I
can
ping
him
but
yeah.
What
like?
What
do
you
have
agreement
on?
What
you
want
the
name
of
the
port
to
be
because
then
I
can
just
be
like
hey
here's.
What
the
name
of
the
port
should
be.
Let's
make
sure
we
click
the
button.
A
A
R
A
Q
So
I
I
can
ask
somebody
in
easter
team
because
they
have
a
lot
of
experience
with
service
meshes
and
like
port
forwarding,
so
maybe
they
can
advise
very
quickly
like,
but
I
need
to
know
exactly
what
to
ask.
So
I
think.
N
O
Users,
don't
want
two
parts
either
josh.
Is
your
code
base
go
yeah.
I
N
Me
josh:
could
you
in
parallel
just
ping,
grpc
or
sto
or
morgan,
and
just
like
get
a
definite
like
yeah,
we're
experts
on
this,
and
we
know
how
to
make
it
work,
not
just
that
we
know
how
to
make
it
work,
but
it's
technically
supposed
to
work
and
that
the
go
code
base
actually
works.
We
know
that
all
that
works.
I
So
sergey
sergey
one
thing
to
if
you
want
to
ask
easter
folks
is
if
they
can
confirm,
they
were
able.
I
mean
lifestep
said
they
have
they
have
these
very
ugly,
but
maybe
istio
has
a
better
solution.
My
solution
is
using
grpc,
but
again
I
I'm
very
newbie
in
mtls
and
I
I
wrote
a
test
and
somebody
would.
I
would
like
to
confirm
that
it's
working.
N
S
Morgan
just
confirmed
he's
filing
this
today,
so
hopefully
he
does
the
right
thing.
Just
fyi.
I
D
C
D
I
Okay,
sure,
but
usually
this
should
be
in
the
next
meeting.
I
know
I
know
aws
folks
are
following
me
in
all
the
meetings
and
are
blaming
me
in
all
the
meetings,
but.
D
C
So,
okay,
so
I
guess
we
can
go
to
the
next
one
yeah.
This
is
a
pr
created
used
to
revert
some
of
the
changes.
So
basically
we
don't
have
third
party
propagators
and
lolita
is
she's
working
on
a
follow-up
or
she
will
be
working.
C
So
this
is
just
to
call
out,
I'm
probably
suggesting
that
we
go
on
mercedes
and
we
follow
her
rita's
follow-up
and
I
think
she's
not
in
the
call.
R
R
C
I
would
probably
add
a
small
node
as
part
of
my
pr,
so
we
can
merge
it
and
then
you
know
like
note:
there's
a
follow
up
coming
or
something.
C
I
C
C
Okay,
there,
you
are,
you
know
anyway,
so
the
next
one
is
yes,
this
one.
I
am
a
little
bit
curious.
I
think
it
changed
makes
sense.
But
if
I
remember
correctly,
there
was
a
specific
use
case
in
which
you
are
sending
array
attributes
and
then
you're
sending
five
and
five
and
like
because
they
are
like
you
know,
siblings,
so
to
speak,
and
then
you
are
expected
that
some
of
them
may
be
null.
C
A
L
I
But
so,
for
example,
this
the
array
says
is
primitives
and
I
think
I'm
following
back,
I'm
going
back
to
to
andrago's
argument
that
primitives
in
general
cannot
be
null
like
a
double
cannot
be
null
unless
so,
if
you
are
using
depends
on
the
language,
but
even
in
java,
if
you
are
using
an
array
of
double
of
the
primitive
double
with
the
lowercase
d,
not
not
the
the
capital
d,
you
cannot
be
not.
U
U
L
And
and
I
think
for
javascript,
the
old
wording
still
applies,
because
you
cannot
prevent
a
race,
so
you
should
pass
them
on,
as
is
to
the
exporter
like
before
yeah.
V
I
Yeah
that
that
makes
a
lot
of
sense
tristan.
But
what
do
you
do
in
the
exporter
if
you
find
there
are
not.
L
A
L
N
I
want
that
interesting
question.
I
I
mean,
I
think
everyone
here
knows
I
believe
knowles
are,
but
I
don't
want
to
re-litigate
that
one
tristan.
You
said
something
about
dynamically
checking
the
array.
When
you
add
something-
and
I
can't
imagine
what
api
that
would
be,
I
don't
believe
there's
ever
a
case
where
we
are
adding
through
an
open,
telemetry
api.
A
value
understanding
was
that
you
can
set
a
list
value
and
you
would
check
that
list
for
for
consistency,
but
there's
never
a
add
to
array
that
was
already
annotated.
I
believe.
V
Do
you
mean
an
array,
as
in
already
an
attribute
value?
I
mean
we
don't
check
the
because
you
can
add
attributes
to
a
span
that
already
has
some
attributes
right.
N
V
Merging
we
were,
we
were
yeah
merging,
so
you
could
later
on,
like
in
an
http
request
at
the
beginning,
you
can
set
a
bunch
of
attributes
about
stuff
and
at
the
end
you
got
to
set
some
attributes
when
the
request
is
over,
so
you
just
append
them
as
additional
attributes.
I
No,
no,
I
think
adding
yeah
go
ahead.
Bogan,
please
josh!
I
think
I
think
what
josh
wants
is
the
following:
let's
assume
you
have
attribute
with
a
key
foo,
and
that
is
that
is
an
array
and
has
the
one
and
two
as
values,
so
I'm
adding
initially
four
as
with
an
array
value
with
one
and
two
and
later
I'm
adding
again
or
I'm
setting
again
four
with
values.
Three
four
is
the
result.
I
One
two
three,
four,
three,
four,
oh
just
three,
four,
that's!
I
think
that
confirms
what
the
value
is
immutable,
so
there
is
no
wording
of
the
values
it's
merging
on
the
map
of
attributes
on
the
key
like
with
different
keys,
we
keep
both.
But
yes,
it's
overriding.
If
this
is
the
same
key,
so
I
think
that
answers
your
concern.
Is
it.
N
I
V
Right,
oh,
but
one
thing
on
this
topic,
though
I
noticed
the
the
proto
buffs
for
attributes,
don't
force
homogenous
like
arrays
or
anything.
It
would
kind
of
be
nice
if
it
did
just
to.
If
we
want
homogeneous
because
then
the
exporter
would
force
itself
into
only
sending
the
ones
that
fit
whatever
type.
A
We
this
has
been
discussed.
We
don't
want
to
force
that
on
the
protocol
level,
because
it
prevents
us
in
the
future.
If
we
decide
to
support
non-homogenous
arrays,
we
will,
it
will
require
changes
in
the
protocol.
Breaking
changes
in
the
product.
Okay,
so
the
protocol
is,
is
a
superset
of
the
data
that
the
api
requires.
At
the
moment
you
are
able
to
represent
more
than
is
necessary,
but
not
less
or
exactly
what
is
necessary
right
now.
V
V
S
Okay,
so
basically
there's
this
tool,
which
takes
the
ammo
files
and
like
generates
semantic
conventions
from
it.
Okay
and
right
now
there
are
rpc
conventions
defined
for
trace,
which
define
all
the
attributes
that
you
like.
We
recommend
people
provide
or
like
encourage
people
to
provide
to
denote
rpcs,
and
I
threw
together
a
pull
request
to
just
kind
of
mirror
that
for
metrics
what
was
raised
in
the
pull
request,
which
is
interesting,
was,
I
you
know,
put
the
same
like
semantic
table
for
rpc
to
share
with
tracy
and
then
the
actual
code.
S
That's
generated
references
tracing
like
it.
References
span
ids
in
the
metric
specification
right.
Also,
the
tool
only
has
one
namespace
for
all
ids.
So,
even
though
the
symantec
conventions
are
under
trace
rpc
because
it
has
the
id
or
pc,
I
tried
to
make
an
id
rpc
for
metric
and
then
all
hell
broke
loose.
So
I
guess
what
I'm
asking
is.
I
have
two
questions.
One
is:
should
we
be
sharing
semantic
conventions
for
attributes
across
metrics
and
tracing?
S
The
reason
I'm
asking
now
is
if
trace
is
stable,
I
don't
want
to
make
any
changes
to
trace
whatsoever,
including
semantic
conventions,
but
this
this
pr
might
make
like
stupid
little
changes
of
just
like
words
to
like
make
things
be
copyable
in
both
places.
So
I
just
wanted
to
check
on
both
those
questions.
I
I
will.
I
Question
first
answer:
yes,
we
should
share
some
of
the
semantic
conventions
that
make
sense
to
be
sure
like
it's
in
the
spec,
it's
specified
clearly
in
the
spec.
I
Second,
the
semantic
conventions
are
not
gonna,
be
freeze
today
or
tomorrow
when
we
are
looking
for
freezing
the
apis.
Yet
we
will
freeze
those
as
well,
but
not
not
yet
so
some
of
the
things
some
changes
are
allowed
still,
especially
if
there
are
editorial
changes
or
clarification,
not
a
problem.
S
I
think
so
yeah
I'm
kind
of
curious
how
much
work
I
put
in.
So
I
guess
last
thing
like
the
second
question
is
really
about.
Should
we
try
as
much
as
possible
to
use
the
same
set
of
semantic
convention
labels,
so
we
reuse
the
exact
same
yaml
file
between
the
two
for
rpc
and
trace
and
rpc
metric,
or
should
I
make
a
general
semantic
conventions
for
rpc
and
then
a
trace
specific
one
and
a
metric
specific
one?
I'm
not
an.
N
Or
a
recommendation,
I
I've
spoken
in
the
metrics
about
how
I
think
what
we,
what
we
might
be
able
to
do
or
could
do
here.
One
set
one
document
or
one
yemo
file.
That
has
the
master
specification
for
all
these
attributes
and
I
think
that
metrics
are
always
a
subset
of
tracing
and
so
that
that,
basically,
what
I
see
is
an
additional
column
on
this
master
document
which
says
I'll
use
this
for
metrics,
yes
or
no,
because
it's
basically
saying
we
expect
high
cardinality
or
we
don't
expect
high
cardinality.
N
L
You
should
come.
That's
a
that's
interesting
idea.
You
should
comment
that
on
on
the
pr,
because
currently,
what
is
suggested
is
to
add
to
each
semantic
convention
group,
a
type
field
that
says
either
matrix
or
span
and
then
all
of
these
metrics
or
spin,
but
not
both,
and
I
wonder
if
across
different
types,
inheritance
or
including,
would
work
yeah.
Maybe
probably
we
need
to
think
about
this
more.
N
That
sounds
good.
We
should
discuss
it
that
was
sort
of
a
like
a
wish,
not
a
demand,
so
maybe
it
doesn't
make
sense.
L
I
think
it's
it
makes
sense.
Yes,
I
see
less
problems
than
with
the
current
approach,
at
least
so.
S
For
the
purpose
of
this
pr,
I
think
that
kind
of
work
is
out
of
scope,
but,
like
I'm
happy
to
try
to
make
progress
or
make
a
proposal
somewhere
else,
if
that
makes
sense
for
what
to
do
in
the
long
run
yeah
anyway.
So
it
sounds
like
there
needs
to
be
more
discussion
and
we're
out
of
time.
So
in
the
interest
of
saving
everyone's
time,
should
I
join
the
metric
sig
to
talk
about
this
or
like
where's,
the
next
place.
We
should
discuss
what
to
do.
N
R
Okay,
yeah
all
right
I'll
work
on
the
proposal.
Then
I
think
that's
a
great
idea
and
then
also
just
commenting
in
your
pr
josh
that,
like
that's
out
of
scope,
we're
gonna
tackle
it
in
this.
You
know
separate
issue
and
then
yeah.
I
all
the
above
I'd
love
to
talk
a
little
bit
more
about
it
and
making
sure
everyone
in
the
metrics
is
aware
of
the
work
you're
going
to
do,
because
I
think
other
people
will
work
semantic
conventions
will
be
benefited
from
that.
So
you.