►
From YouTube: 2023-02-02 meeting
Description
cncf-opentelemetry@cncf.io's Personal Meeting Room
A
C
B
All
right,
let's
start,
let's
start
with
the
big
topic.
B
It
seems
very
reasonable
to
me
my
the
obvious
hesitation.
Is
it
breaks
everything.
E
Yeah
I
was
spending
some
time
looking
into
this
and
yeah
it
looks
like.
Basically,
the
bigger
question
is:
do
we
want
to
can
aligned
with
elastic
common
schemer
here
in
foreign
makes
sense
to
do
all
these
renamings
because
of
that
so
I
I,
just
I,
don't
know
like
a
from
the
perspective
of
the
whole
open,
telemetric
and
a
spirit
right.
It
was
built
from
the
two
projects,
everything
that
it
probably
makes
sense
to
also
bring
elastic
common
schema
here,
because
it's
also
like
it
was
a
community
before
it
looks
like
this.
E
C
B
I
think
it
it
primarily
came
from
people
on
the
log
side,
since
that's
primarily
where
elastic
schema
is
used.
B
It
is
just
for
my
my
understanding.
Is
it's
just
a
field
definitions,
as
opposed
to
the
open,
Telemetry
semantic
Dimensions,
which
also
have
more
structural
conventions.
B
Maybe
we
could
start
with
that
question,
though,
like
elastic
schema
aside,
Riley
made,
you
know
an
argument
here
to
that
by
bringing
them
under
URL,
for
example,
even
within
just
open
Telemetry
conventions,
it
would
unify,
provide
more
unification.
C
F
No
no
strong
opinion
here,
I
I.
Think
if
you
look
at
the
RFC,
the
general
recommendation
is
URI
is
the
right
term
and
also
like
Wikipedia,
although
it's
not
the
authoritative
Source,
but
URI
is
more
like
the
the
professional
way
of
calling
that
URL
is
more
like
a
conventional
way
of
calling
it
I
have
no
strong
opinion.
I
I
feel
like.
If
you
want
to
be
nourished,
we
can
always
pick
the
terms,
but
these
are
something
we
expect.
People
to
use
and
The
Operators
will
just
see
something
in
their
final
data
store.
F
So
URL
is
something
like
that
people
would
know,
and
there
don't
seem
to
be
a
lot
of
confusion.
So
no
big
problem
from
that
by
the
way,
I
have
a
question
for
trust.
My
question
is
this:
group
probably
is
too
small
for
such
kind
of
discussion.
I
I
figure
probably
like
this
is
something
we
should
have.
The
among
the
emissions
which
John
storage
is
driving
and
I
always
have
a
very,
very
clear
personal
opinion
that
I
I
think
the
current
open
parliamentary
semantics
convention.
F
If
we
don't
make
some
fundamental
change,
it's
going
to
fail-
and
this
is
why,
like
last
year,
when
fed
asked
me
like,
can
we
stabilize
things
by
day
from
last
year?
I
thought
him.
My
personal
observation
is
there's
no
way
and
if
you
want
me
to
drive
it,
I'll
have
two
fundamentally
change
how
it
works.
If
I
work
as
a
Microsoft
internal
stream,
I
know
how
things
would
work
and
I
know
exactly
it's
honestly,
my
assumption
actually
worked
open
symmetry
cement
examination
never
really
work,
we're
still
struggling
so
soon.
F
C
C
Metrics
from
it,
so
maybe
elastic
common
schema
worked
in
some
scenarios.
It
didn't
work
in
what
we're
trying
to
achieve
and
it
by
look
of
it,
it's
not
going
to
so.
It
feels
like
we
are
adding
some
chaos
without
much
reason:
I
I'm,
all
in
on
preview
and
attributes
we
have
and
trying
to
make
them
better,
but
saying
that
Rashid
align
with
common
schema
makes
no
sense
to
me
saying
that
Riley
doesn't
believe
in
semantic
conventions.
C
Doesn't
make
me
optimistic
about
this
effort,
so
maybe
Riley
you
can
become
a
bit
more
I,
don't
know
optimistic
about
it
and
try
to
make
it
work
rather
than
trying
to
make
it
not
work.
F
Yeah
so
the
way
I
look
at
Open,
Country
cement
conventions
applying
for
the
elastic
economy.
Schema
is
I,
think
elastic
on
the
schema
provided
a
core
of.
If
you
have
a
key
value
of
error,
and
you
see,
the
key
is
called
like
your
thought,
something
then
what
exactly
that
I
mean-
and
this
is
something
if,
if
anyone
already
defined
that
in
the
public
like
like
the
the.
F
Feel
defining
a
similar
thing
in
a
different
way,
probably
wouldn't
give
a
lot
of
new
value.
However,
I
don't
think
open
climate
system
and
Convention
has
a
lot
of
things
added.
On
top
of
this,
for
example,
ECS
doesn't
even
care
about.
If
you
have
some
HTV
stack,
then
there's
a
rich
High.
How
do
you
model
them
as
parent
child,
because
it
doesn't
even
have
this
choosing
concept
to
take
the
year
right,
so
this
type
of
thing
I
thought
it
Max
different
things
and
which
one
should
be
treated
as
a
required
thing.
F
This
type
of
thing
is
something
I
think
open.
Elementary
commission
is
adding
value.
So
I'll
tell
you
my
ideal
situation
like
in
open
climate.
If
we're
trying
to
introduce
some
new
key
value
pair,
which
will
work
with
elastic
comments,
can
make
sure
we
align
there
and
and
maybe
elastic
common
screen.
My
time
time
eventually
merge
with
open
function.
I
understand
it
might
be
some
concern
on
the
logistics
or
even
like
the
ultimate
Insurance
who
owns
this.
F
So
there
might
be
some
regulation
and
I
feel
like
if
the
elastic
is
calling
something
http.url
and
works
on
something
called
hpe.uri
or
something
it
doesn't
seem
to
be
a
reasonable
thing
for
us
like.
Why
do
we
want
to
be
different
and
what's
additional
value?
So
this
is
something
probably
we
can.
We
can
solve
and
leverage
elastic
common
skin,
but
everything
else.
How
do
we
model
things
and
which
one
is
required,
which
one
is
optional,
I
I
think
open.
Telemetry
is
in
a
very
good
position
for
that,
so.
B
Let's
talk
just
briefly
about
what
it
means:
alignment,
what
alignment
even
means,
because
I
could
see
one
I
one
piece
would
be
that
they
have
the
same
components
but
just
different
names
and
there's
a
mapping.
You
know
of
just
a
name
mapping,
but
they
have
the
same
definitions
behind
them
and
the
other
is
you
know,
a
higher
level
of
alignment
which
is
actually
trying
to
make
the
names
equal
the
same,
where
possible.
F
Yeah,
the
naming
thing
is
not
something
that
concerned
me.
A
lot
I
I
think
there's
a
philosophy
behind
that.
If
you
look
at
ECS,
it's
composable,
so
if
you
have
some
even
thing
like
this
is
Riley's
personal
event
and
I
won't
have
something
called
URL,
and
this
is
clearly
an
HTT
URL
with
ECS
I
know
how
to
call
it
with
open
slab,
which
I
have
no
idea.
I
have
to
define
the
entire
Wireless
events
right
so
so
ECS
is
basically
the
same.
F
C
Okay,
so
what
you're
saying
is
that
when
we
Define
semantic
conventions,
we
say
HTTP
should
be
described
like
that,
but
ECS
is
coming
from
a
different
perspective.
It's
coming
from
from
those
attributes
means
that,
but
how
they
use
together
with
other
attributes,
is
undefined.
It's
not
a
part
of
the
specification
so
like
if
it
had
attribute
registry.
If,
for
example,
we
had
General
Network
attributes,
they
are
like
this:
they
there
are
some
groups
of
them
that
they're
the
years
together,
but
basically
it's
just
an
individual
thinks
that
any
spec
can
refer
to.
A
C
Like
basically,
if
we
have
a
general
set
of
registry
of
attributes
and
then
semantic
conventions
can
Define
their
well,
they
should
contribute
probably
to
the
central
place,
but
that
they
can
refer
to
this
thing.
But
essentially
then
the
fact
that
somebody
creates
a
span
with
messaging.system
attribute
doesn't
mean
much.
C
You
can
combine,
messaging.system
and
I,
don't
know,
http.flavor
it
won't
spend
and
backends
would
not
be
able
to
understand
what
is
it
right?
Yeah.
B
So
I
think
what
you're
saying
vanilla
if
I
can
Claire
confirm
is
that
the
open,
Telemetry
semantic
conventions
would
provide.
You
know
this
is
what
a
HTTP
client
span
looks
like,
and
these
are
the
attributes
which
make
sense
are
recommended
on
this
band.
It
wouldn't
be
like.
Oh,
you
can
throw
URL
dot
anything
on
here.
These
are
the
specific
url.d
ones
that
we
support,
but
then
that
that
that's,
your
main
concern
am
I
understanding.
That
right
is
that
people
can't
just
throw
random
attributes
onto
these
spans.
C
B
Riley,
is
it
fair
to
say
that
that
small
detail
is
what
you're
concerned
about
of
the
alignment
of
just
the
attribute
the
attributes
than
at
the
attribute
level,
or
is
there
a
bigger?
Yes,.
F
I
I
I,
don't
feel
I
fully
understand
what
the
marriages
explains.
So
I'll
I'll,
probably
just
reuse
the
example.
So
here
we
have
HTTP
URL
and
in
other
places
you
might
have
some
like
prpc
dot,
endpoint
or
something,
and
if
there's
a
desire
for
people
to
provide
some
logs
and
they
have,
they
have
touched
all
these
places.
Are
they
going
to
use
the
same
attribute
or
you're
going
to
keep
using
different
ideas,
because
it
looks
as
easy
as
they
took
a
different
approach?
They're
saying
this
is
a
domain.
C
Yeah
so
I
think.
The
key
question
is
that
we
here
we
discuss
either
traces
or
metrics
and
in
both
cases
we
say:
okay
for
HTTP,
we
should
collect
following
information.
This
is
for
tracing.
This
is
for
metrics.
This
information
should
have
specific,
unnamed
span
kind
and
a
set
of
attributes.
Let's
say
if
you
don't
provide
the
http.net
attribute,
it's
not
the
validation,
TP
instrumentation,
because
it's
a
required
attribute.
D
C
Then,
for
what
you're
talking
about
is
more
like
logs,
that
people
are
going
to
throw
in
some
attributes,
but
then
the
only
thing
that
has
defined
as
names
like
the
the
attributes
itself.
It's
not
that
what
a
set
of
attributes
you
need
to
to
yeah.
F
F
I
understand
I'm
I'm,
saying
that
like
it'll,
be
great
if
the
the
spans
and
the
logs
and
metrics
we
use
the
same
attribute
name
for
the
same
concept
if
we
buy
these
have
different
things
because
in
open
Library
it
might
like,
when
people
use
examples,
they're
going
to
report
something
as
logs
and
if
they
have
a
different
name.
C
Totally
this
I
agree
with
so,
and
it
doesn't
say
anything
about
this
ECS,
yet
it's
so.
What
we
need
is
that
when
people
instrument
things
they,
if
they
see
something
similar
for
logs
or
for
not
specified
spans,
they
should
probably
reuse
one
of
the
existing
attributes
rather
than
introducing
a
new
one.
A
B
These
seem
pretty
reasonable
to
me
to
you
know:
change
to
URL,
even
the
HTTP,
you
HTTP
user
agent
kind
of
make.
This
make
some
sense
to
me
like
that.
It
could
be
reused
elsewhere
than.
B
C
What
one
knife
changed,
a
bunch
of
things
to
net
I
tried
to
fight
technologists
that
use
something
like
X4
over
did
for,
but
I
failed
to
find
anything
except
HTTP,
and
this
is
where
I
don't
know
how
to
have
the
balance,
because
if
it's
something
very
specific
to
a
certain
technology
over
abstracting,
it
would
not
help
anyone
it
might
hurt.
C
And
if
we
are
going
to
align,
we
will
have
this
conversation
about.
Basically,
everything
in
elastic
common
schema.
E
E
But
if
you
like,
I
want
to
make
it
more
General,
then
it
will
be
a
mixture
of
again
HTTP
plus
some
common
attributes.
Some
Network
layer,
attributes
or
you
know,
URL
or
URI,
is
something
which
is
most
likely
like
a
technology
agnostic
right.
So
it's
not
like
I,
just
just
like
a
jail
specification.
So
what
in
this
case,
what
HTTP
submitting
conventions
mean
right,
because
we
can
just
generalize
all
these
attributes
and
it
will
be
really
really
specific
for
HTTP
then
like
for
headers
response
codes
and
something
like
this
right,
but
well.
B
I
think
the
https,
the
hotel,
HTTP
spec,
would
still
have
all
of
this
stuff
right.
It
would
maybe
missing
something,
but
I
feel
like
the
proposal
on
the
table
is
just
to
rename
these
some
of
these
Concepts,
which
could
maybe
be
shared.
A
B
B
E
Question
yeah
I
just
wanted
to
say
that's
the
stability
question.
It
became
becoming
more
trickier
right
so
because,
if
you
want
to
reuse
something
from
General
perspective,
that
General
should
become
stable
first
right
so
make
sure
that
it
will.
It
will
not
will
not
change
on
HTTP
level,
so
only
after
that,
HTTP
can
be
called
the.
B
Do
you,
as
far
as
like
the
URL,
what
if
we
had
an
attribute
you
know
on
the
yaml
per
open,
Telemetry
attribute
that
says
whether
it's
stable
or
not,
so
that
we
could
say
have
a
URL
attributes
Mark,
you
know
just
the
one
stable
that
are
required
for
HTTP
or
lit
even
let's
talk
net
since
this
is
we
have
to
deal
with
this
anyways
yeah.
F
Or
or
maybe
making
it
either
in
the
description,
maybe
we
can
just
either
not
with
being
this
stable
or
not
available,
yet
there's
a
description
so
and
later
the
tool
kind
of
evolves,
but
I
figure
like
instead
of
trying
to
make
the
tool
perfect.
Currently,
what
is
it
like?
The
tool
is
good
enough.
Let's
make.
C
I
mean
adding
stable
property
into
the
yaml
and
generate
like
this
tables
are
generated.
Then
they
have
to
be
generated
like
updating
the
tool.
It's
super
easy
I
can
easily
do
it,
commit
to
anything
that
there
but
I
agree.
That's
a
great
point
to
add
this
property
is
stable
right
and
have
it
obvious
in
the
table.
A
F
I
can
definitely
see
the
value
of
given
how
you
have
knives
and
and
even
if
the
entire
thing
is
stable,
at
least
when
people
want
to
add
something.
We
still
need
a
way
for
people
to
either
the
specifics
in
Market
as
experimental
and
then
push
it
was
stable.
So
so
there's
definitely
a
need.
A
question
is
like
now,
I
know
it's
not
going
to
be
a
huge
amount
of
work,
but
it
seems
like
we
have
a
lot
of
items
so
trying
to
make
sure
I
can
solve
the
items
that
will
work
out.
C
Yeah
and
I
think
it's
it's
an
interesting
also
because
of
the
Danielle's
Otep
with
regard
to
private
and
public
attributes.
It's
somewhat
the
same,
so
maybe
Daniel
can
help
us.
There
I
mean
unless
we
are
ready
to
get
stable.
Unless
we
already
to
stabilize
anything,
we
will
have
to
have
all
the
discussions
first
right.
So
it's
not
urgent.
We
don't
need
to
do
anything
now,
it's
just
something
that
will
need
to
happen.
C
B
F
C
You
know
they're,
assuming
we
do
the
Euro.
There
are
more
things,
for
example
in
ECS
they
have
url.domain,
which
in
our
case
is
not
pure
name,
and
we
haven't
done
the
ECS
analysis
to
see
if
it
supports
everything
that
we
supported
in
our
proxies
and
whatnot.
F
F
Yeah,
my
suggestion
is:
actually
we
ask
Alex
to
help
because
he
has
the
intention
to
to
like
have
EPs
and
open
time
align
as
much
as
possible,
and
he
has
the
best
knowledge
about
ECS
and
as
part
of
the
platform,
he
should
also
spend
time
to
understand
open
language.
So
I
would
say
like
why
not
ask
if
Alex
is
willing
to
do
it
and
if
it's
not
willing
to
do
it,
then
it
like
it's
pretty
clear,
but
in
open
language
we
don't
have
much
choice.
F
A
B
To
that
net
net
attributes,
I
was
trying
to
see
I
hadn't
looked
up
what
the
corresponding
ECS
attributes
were.
There's
like
I
guess.
C
B
Source
Port
yeah
store
subdomain;
okay,
yes,
yes,
and
then
they
have
destination
or
kind
of
the
other
side.
C
F
Something
else
can
help,
because
the
the
fact
that
they
have
this
type
of
domains
is
already
saying,
like
the
things
roll
some
of
the
challenges
here
and
especially
consider
their
open
source.
So
a
lot
of
people
are
determined
something
with
their
own
stuff,
not
necessarily
the
elastic
companies
controlling
everything
so
so,
maybe
like
they
will
learn.
If
open
publisher
is
going
on
with
this
current
approach,
where
we
might
learn,
I
I
feel
there's
a
lot
of
learning.
F
C
I
have
an
overall
feeling
that
we
have
two
choices
here.
We
either
Target
unification
with
ECS
and
when
we
should
abandon
our
plans
to
make
anything
stable
in
the
next
year
or
maybe
two
or
we
should
say
that.
Okay,
we,
the
semantic
conventions,
were
baked
for
enough
time
for
us
to
do.
V1
and
maybe
unification
with
ecs's
future
V2,
which
can
happen
in
the
next
five
years
after
the
graceful
period
for
stable
version.
F
Yeah
I
thought
there
might
be
a
third
option,
given
there
are
two
different
things
going
in
parallel.
One
is
you
want
to
stabilize
HTTP
semantic
convention?
Another
is
easy.
F
We're
going
to
stabilize
this
by
this
by
this
time,
anyways
and
and
let's
have
the
ECS
folks
come
and
help
the
drive.
How
do
we
align
with
ECS
battery
here
and
if
we
failed?
No,
no,
it's
not
going
to
work
even
in
a
smaller
space
than
you
can
imagine
in
a
bigger
space,
it
won't
work
and
we'll
still
carry
with
our
existing
plan.
F
The
only
thing
we're
not
sure
is
how
do
we
align
the
names
with
ECS,
so
you
go
and
and
make
a
change
on
the
current
one.
So,
along
with
ECS
as
much
as
possible,
this
group
will
do
reveal
and
if
that
looks
better
and
solve
that
some
problems
we
have
then
go
together,
pushes
for
April
time
frame.
If
it's
not
working,
then
you
know
the
ECS
is
not
going
to
work,
so
you
have
to
go
back
and
maybe
negotiate
more
with
other
phones.
That
would
help
both
more
streams.
B
B
Or
do
you
want
to
Ping
him
see
if
he
can
make
I
know
he's
in
Germany,
but
maybe
he
can
make
our
Friday
meeting
and
I'll
try
to
get.
You
know
discussion
going
about
just
on
the
URL
side
as
kind
of
floating
this
to
see
you
know
hey.
How
committed
are
we
to
this?
Is
the
community
willing
to
deal
with
this
kind
of
breakage.
F
Yeah
so
here's
my
thing
here:
I'm
going
to
tell
Alice
like
I'm,
very
willing
to
see
UCS
and
open
commissary
cemented
convention
can't
a
lot
as
much
as
possible.
F
B
We
cannot,
we
don't
get
if
we
don't
get
community
Hotel
community
support
for
it
yeah,
but
yeah
I
I
agree
with
I
I'm
I
mean
on
trying,
but
with
the
very
clear
you
know
that
we're
have
a.
We
have
a
very
limited
time
to
to
try
that
and
see.
If
we
get
people
really
truly
on
board.
B
It
will
probably
end
up
talking
about
this,
then
on
Friday,
hopefully
Alex
will
join
matesh,
we'll
probably
join
on
Friday.
If
we
have
time
we
could
chat
about
this
with
him,
but
I
saw
Alid
Miller.
You
replied
on
that
issue.
Thank
you.
B
Was
it
did
they
say
if
they
could
make
it
Friday?
No.
D
B
Yeah
I'll
I'll
ping
him
he
said:
Friday's
work,
the
Friday
time
works
for
him,
so
I
will
ping
him
tell
him.
You
know,
especially
if
we
get
Alex
but
in
even
if
not
the
Friday
discussion
I
think
is
going
to
be
critical
on
the
ECS
topic
and
yeah
I
agree.
We
have
to
get
him
on
board.
F
B
I
I
think
this
is
to
time
sensitive
I,
don't
think
we
have
much
time
to
get
the
open,
Telemetry
community
on
board.
That's
kind
of
what
I'm
more
worried
about
I
know
that
the
ECS
folks
will
be
on
board
with
this
proposal.
It's
whether
the
open,
Telemetry
folks
can
stomach
the
breaking
this
amount
of
breaking
changes.
Yeah.
F
I'm
less
concerned,
because
if
addicts
is
waiting
proposal
there,
then
we
we
have
a
ETI
space
portal
and
we
also
have
a
non-egis
basic
proposal.
Then
from
the
open,
photography,
Community
we're
saying
these
are
only
two
ways:
we're
saying:
there's
no
survey,
surveys,
don't
do
it.
We
don't
have
cement
examination
right,
so
so
folks
come
and
watch
visual
in
your
life
and
and
then
like.
First,
you
lead
this
effort,
so
you
put
that
date
and
at
that
time
like
collect
the
results
and
see
which
one
people
would
like
and
don't
make
decision
here
right.