►
From YouTube: 2023-03-27 meeting
Description
cncf-opentelemetry meeting-2's Personal Meeting Room
A
A
Check-In
and
then
I
see
we
have
some
topics
for
deeper
discussion
today
so
quickly
going
through
the
Sig
check-in.
It
looks
like
we
have
a
number
of
updates
on
the
spec
Carlos
I
see
that
you're
on.
Do
you
want
to
walk
us
through
these.
C
Actually
I
did
so
those.
B
C
C
C
The
other
is
about
experimental
semantic
conventions.
We've
had
requests
in
respect
when
new
instrumentation
was
added,
where
it
wasn't
clear
to
maintain
us
whether
this
should
be
added
to
the
spec
first
and
then
be
implemented
or
the
other
way
around
and
the
old
tip
224
would
propose
a
process
that
would
would
ease
this,
and
also
since
we
are
on
the
topic
of
semantic
conventions,
the
other
one
is
about
defining
what
semantic
conventions
actually
enforce
and
what
the
stability
guarantees
there
are.
C
B
I
put
it
there
yeah.
This
is
a
tab,
configuration
configuration
tip,
sorry
and
it
has
enough
approvals.
It
went
through
a
lot
of
feedback,
I
think
we're
in
a
great
position.
Now
we
have
enough
approvals.
I
just
want
people
to
take
a
last
look
in
case.
There's
something
you
know
important
there
I
would
expect
honestly
only
minor
things,
probably
for
further
discussion.
As
I
said,
there
was
Heavy
feedback
and
it
was
it
has
all
of
it
being
addressed.
Just
for
your
information
and,
of
course,
remember
that
we
will
still
have.
B
We
will
still
have
discussions
on
the
actual
details
once
these
changes
hit
the
specification
So.
The
plan
is
to
merge
it
in
the
next
couple
of
days.
Now
today
in
this
Monday
people,
were
you
know
on
the
weekend.
You
know
chilling
out
so,
but
maybe
tomorrow
maybe
Wednesday,
let's
see
and
finally
yeah
we
are,
we
will
have.
Oh
that's
typical
of
kicking.
You
out
great
great
timing.
A
B
A
B
Right
sorry,
go
ahead;
no!
That's
all!
In
logs
there
by
the
way,
there's
nothing
in
logs
in
proto
we
can
at
least
I
can
probably
mention
that
there's
a
stability
guarantees
mentioned.
Who
put
this
one
here
Armin?
Maybe
you.
C
Yeah
I
did
so
that
one
is
about
defining
which
operations
are
considered
to
be
allowed
additive
changes.
This
is
something
that
will
effect
maintain
us.
So
please
take
a
look
there.
A
Okay
and
these
stability
guarantees,
while
I
was
signing
back
in
Carlos,
you
mentioned
logs,
is
this
specific
to
logs,
or
is
this
across
all
of
otlp.
A
Okay,
let's
keep
going
looks
like
PHP
is
automatic
instrumentation
in
progress
for
tracing
very
cool;
no,
no
major
updates
for
Java
python
had
1.17.0
released
last
week.
Moving
on,
we
have
the
second
metric
API
release
candidate
I
take
advantage
basically
the
second
release
candidate
for
the
metrics
API
for
go.
It
was
released
last
week,
and
so
you
can
click
here
to
see
more
about
that
very
cool
looks
like
we're.
Really
really
close.
A
No
major
updates
for
C,
plus
plus
collector
0.74
was
released
last
week
and
coms
has
a
new
blog
post
on
otel
at
kubeconi
U
in
the
works.
Great
I
think
I
I
read
that
last
week
and
approved
that
and
the
related
SDK
configuration
there's
the
Otep
that
we
mentioned
above
Tristan.
You
have
a
topic
here
want
to
walk
us
through
it.
D
Yeah
I,
just
so
I
put
this
in
slack
earlier
and
got
some
feedback
about
General's
fan
attributes
which
are
things
like
the
line
in
the
code,
the
name
of
the
file
module,
name,
class
name,
things
like
that
and
what
I
was
asking
was
across
other
languages.
What
our
maintainers
doing
there
if
they're,
including
them
at
all
or
is
it
expected
that
users
include
them?
D
What's
sort
of
expected
from
the
spec
standpoint,
I
guess
it
seems
doable
down
to
its
up
to
the
user
to
include
them
and
we're
just
giving
the
names.
If
that's
I'm,
just
looking
to
see,
if
that's
what
people
take
it
to
be
or
I
think
what
might
end
up
is
another
use
case
for
the
being
able
to
change
spans
in
the
process
there's
easier
because
then
you
could
have
a
filter.
You
know
attribute
filter
and
then
you
could
provide.
D
You
could
say
all
these
general
ones
for
Source
are
available,
but
the
user
can
specify
if
they
want
to
include
them
or
not,
because
I
assume
with
limits
and
billing
or
whatever
people
might
not
want
all
these,
and
that's
maybe
why
people
don't
include
them
by
default.
So
just
curious
if
any
maintainers
have
thoughts
on
that.
A
Yeah,
oh
yeah,
I
wonder
if
these
were
added
to
the
spec
after
a
lot
of
the
stuff
was
instrumented
in
the
other
languages.
I
very.
D
Yeah
so
yeah
that'd
be
interesting
for
people
to
look
at
it
and
think
about
and
get
back
because
yeah
I'm
not
sure
if
the
spec
means
it
to
be
some,
it
I
think
it's
marked
as
recommended,
but
if
it's
something
that's
kind
of
expected
to
be
on
spans
or
if
it's
expected
just
that
we're
providing
these
attribute
keys,
so
users
can
Define
them
if
they
want
yeah.
D
That
sounds
like
people
might
not
have
looked
at
it.
Much.
A
D
I
think
the
next
one
thread
ID
probably
the
same
case
this
one
after
talking
to
some
people
on
slack
also
I.
This
one
I
kind
of
came
away
from
probably
just
leave
it,
and
so
my
use
case
specifically
is
coming
from
it
at
erlang
and
elixir
where
we
have
process
IDs,
which
fits
really
well
in
here
and
maybe
would
be
good
to
put
instead
of
the
actual
OS
thread.
D
Id
actually
means
in
these
examples
and
yeah
I
wasn't
sure
if
people
were
using
it
and
if
anybody
had
any
thoughts
on
whether
it
should
be
OS
thread
or
if
it
should
be
whatever
the
language
wants,
and
if
they
have
more
specific
things
like
an
early
with
process,
IDs
Ruby
has
fibers
go,
go
routines,
have
more
just
VM
specific
terms,
attribute
keys
and
not
even
worry
about
thread
ID.
C
We
fought
about
the
ones
that
that
do
have
a
meaning
outside
of
the
respective
execution
environment,
so
that
would
be
something
that's
visible
to
the
US
to
the
OS
and
light
threads
and
such
they
might
not
be
of
of
much
help
there,
but
it
should
or
could
be
discussed
if,
if
someone
will
propose
them.
D
C
D
The
threat
ID
currently
is
inconsistent,
so
I'm
not
sure
if
it's
worth
adding
an
attribute,
so
those
can
be
all
consistent
or
if
it
also
just
doesn't
really
matter
that
one
because
I
could
see
just
leaving
it
as
thread
ID
and
erlang
will
add
erlang.pid
in
Ruby,
they'll,
add
ruby.fiber
and
we'll
just
and
Java
we'll
just
stay
as
it
is
I,
don't
know.
You
know.
A
Any
other,
not
some
other
maintainers,
like
I'd
love
to
hear
from
people
on
Java
or
JavaScript
or
python
or
others
like,
is
this
part
of
the
spec
you've
seen
and
chose
not
to
implement,
because
you
think
there
are
good
reasons
to
or
or
is
it
just
that
it
was
a
new
part
of
the
spec
we
never
got
around
to
it.
Well,.
F
G
What
exactly
you're
referring
to
in
in
Java
is
there
a
conversation?
I
can
go
look
at
or
if
we
talk
about
newer
versions
of
java
and
this
being
a
virtual
thread
versus
an
OS
yeah.
D
In
I
think
it's
in
hotel,
maintainers
I
was
talking
with
Tyler.
Who
was
explaining
that
in
Java
that
current
thread
ID
will
potent
in
newer
versions
will
potentially
be
a
virtual
thread?
That
is
just
an
attribute
essentially
on
thread.
Instead
of
being
like
a
separate
like
in
Ruby,
you
have
fibers
and
go.
D
You
have
go
routines
and
erlangua
processes,
so
they're
completely
separate
from
an
OS
thread,
but
in
Java,
apparently
that's
you're
still
using
a
thread
which
could
be
virtual
or
could
be
an
OS
thread,
but
yet
there's
a
conversation
not
too
far
back
in
Slack,
where
I
asked
about
this
and
Tyler
went
over
it.
C
Also
funny
called
attributes,
there
was
a
conversation
on
that
one
yeah.
C
So
when
I
added
these
like
many
many
years
ago,
the
idea
was
mostly
about
manual
instrumentation,
so
that
you
would
put
it
there
in
in
manual
instrumentation.
If
you
don't
have
any
any
better
idea
or
for
like
close
to
debug
log
purposes.
C
But
as
check
pointed
out,
there
is,
for
example,
in
Java
and
attribute
extractor.
That
would
put
it
on
on
certain
spans
and
it's
not
not
always
enabled
by
default,
but
some
instrumentations
use
it
so
there,
if
you
want
to
follow
up.
There
is
a
it's
like
a
conversation
on
the
topic
as
well.
D
C
And
as
for
the
requirement
level,
that's
on
there,
so
it
says
recommended
for
these,
but
that
unless
the
the
like
introduction
on
the
overall
page
here
was
changed,
I
think
that
one
should
say
that
within
each
category,
if
you
have
one,
it
is
recommended
that
you
also
follow
the
others,
but
it
doesn't
mean
that
you
necessarily
have
to
put
net
top
transport
or
or
a
code.function
on
each
and
every
span.
So
it's
always
recommended
within
within
the
category
or
within
the
the
individual
convention.
H
B
You
feed
that
sounds
good.
I
would
like
to
have
five
minutes
yes
of
today.
Ask
all
the
depths.
Well,
not
all
of
tips.
E
B
Give
you
control
perfect.
Let
me
then
share
my
screen.
We
only
have
three
and
I
think
it
makes
sense
before
they
start
piling
up.
You
know
there
we
are
perfect,
so
we
only
have
three
new
since
the
last
time
we
triage
things.
B
Let
me
start
with
the
with
the
sorry
what's
happening
with
the
most
obvious
one,
which
is
the
one
that
Armin
already
mentioned
regarding
the
separation
what's
happening
anyway,
which
is
this
of
the
majority
students
about
separating
the
semantic
conventions
from
specification.
We
don't
have
to
discuss
it
now,
but
basically,
for
you
know
to
simplify,
let's
say
things
between
the
specification
and
the
semantic
conventions.
We
don't
have
to
to
be
breaking
specification
every
time
we
do
a
change
in
the
somatic
conventions.
We
want
to
do
this
change,
as
you
can
see.
B
There's
a
lot
of
discussion
going
on.
This
is
very
important,
also
as
part
of
the
of
I
could
say
any
of
the
stability
we
we
want
to
provide
on
this
many
conventions
and
that's
why
and,
as
you
can
see,
there's
no
discussion,
so
I
would
like
to
put
a
label
of
a
priority
zero.
You
know
I
think
if
we
want
to
do
this
change,
it
has
to
be
done
sooner
later
than
later.
C
B
C
B
Yeah
correct
and
please
take
a
look
as
I
as
you
saw,
there's
a
lot
of
discussion
and
yeah.
We
still
need
to
find
agreement.
What
the
discussion
is
going
on
this
one.
Okay,
let's
go
to
this
one
I,
don't
know
whether
jmic
is
around
or
somebody
from
the
sampling
group,
but
basically
this
expands
the
probability
some
probabilistic
sampling
about.
You
may
remember
that.
B
Well
in
case
you
were
involving
that
the
that
that
section
is
experimental
at
the
moment
and
that
probabilistic
sampling
defines
basically
use
values
that
are
powers
of
two
and
now
basically
they
are
trained
to
expand
that
that
include
non-power
of
two
values
as
well.
B
B
B
And
in
case
you
were
curious,
that's
this
is
the
issue
by
the
way,
which
is
that
we
want
to
also
expand
a
little
bit
more
because
there's
there's
some
section
regarding
Tracy
the
Radio
based
on
blur,
which
the
maintainers
may
be
familiar
with,
and
basically
we
want
to
expand
this
to
cover.
We
want
the
probability
sampling
to
be
expanded
and
cover
these
two.
You
know
so
that's
important
work
but
yeah,
given
the
the
have
calendar,
I
would
say:
Priority
One.
I
B
Which
is
I
hope,
maybe
about
this
one.
So
basically,
I
can
probably
just
read
well
you're.
Reading
yourselves
there.
It
proposes
addition
of
a
chain
ID
attributes
to
span
unlocks
that
supports
another
processing
analysis
and
can
be
implemented
as
part
of
an
extension
I
think
nobody
has
reviewed
that,
including
me.
Sadly,.
A
B
Right
right,
okay,
in
that
case,
probably
let's
take
this
offline.
I
will
review
about
myself
after
the
call
I
don't
want,
as
I
said
before,
I
don't
want
this
to
be
piling
up.
At
least
we
want
I
want
the
community.
Like
you
know,
Ted
was
trying
to
to
do
basically
just
be
aware
of
that.
There's
this
hotel,
instead
of
just
having
them
and
realizing
yeah,
no
I,
think
it's
good
good
to
go.
B
So,
let's
review
the
offline,
that's
more
realistic,
yeah,
okay
other
than
that.
I
think
that
we're
done
then,
and
please
review
especially
the
one
by
Josh
surette
on
separating
somatic
conventions
from
the
spec
I.
Think
that's.
A
And
we
have
one
more
topic:
Aaron
had
suggested
in
the
chats
based
off
of
last
week's
conversation
about
answering
stack,
Overflow
questions
in
each
Sig.
Aaron
had
said
what
about
the
stack,
Overflow
questions
that
kind
of
Falls
with
the
cracks,
if
they're,
not
language,
specific
or
not
specific,
to
The,
Collector
or
or
something
with
an
existing
Sig
for
I.
Think
this
form
could
be
a
good
place
for
just
quickly
reviewing
those
once
a
week
and
either
I
mean
theory.
A
If
there's
just
a
few,
we
could
answer
them,
or
at
least
like
farming
out
who
who
could
answer
them
so
I.
Think
it's
a
very
good
idea.
I've
got
stack
Overflow
open
now,
so
we
can
just
try
and
go
through
it
the
first
time.
What
I've
done
here
is
I
just
searched
for
open,
Telemetry
I.
A
Think,
there's
work
being
done
right
now
to
get
us
like
a
like
to
standardize
everyone
on
a
single
tag,
given
that
this
is
all
new
I
just
did
like
a
string
search
for
open
Telemetry,
just
because
I'm
guessing
a
lot
of
these
won't
be
tagged
and
I
sorted
by
newest
sort
of
flying
at
the
seat
of
my
pants
here,
like
there's,
certainly
going
to
be
some
of
these
that
might
be
specific
to
certain
Solutions
or
back
ends.
I
suppose
we
could
cover
those,
but
I
feel
like
like
this.
A
One,
for
example,
is
probably
better
answered
by
someone
in
grafana,
because
that's
probably
where
their
challenges
are
same
with
this
for
x-ray,
but
if
we
go
through
these
and
categorize
them
and
again
this
first
time,
we've
done
this
just
sort
of
a
rush
thing,
we'll
probably
come
up
with
a
better
format
for
this,
so
we've
got
one
here.
This
is
definitely
JS
specific,
so
I
don't
think
we
would
tackle
this
as
a
group.
G
Could
I
make
a
suggestion
that
maybe
yes,
please
put
a
tag
on
it
for
like
open,
Telemetry
JS
so
that
they
can
find
it
or
maybe
just
add
the.
A
Js
tab.
Yes,
that's
a
good
idea:
I
I
mean
I'm
signing
a
stack
Overflow,
but
I
am
not
usually
a
person
who
comments
on
it.
Can
you
walk
me
through
how
to
add
a
text
I.
F
A
G
I
I
I
Ahead,
so
how
do
you
like
it
so
open
Telemetry,
all
together,
Dash
JS
or
what
was
the
suggestion
here.
A
I
I
Yeah
I
agree
with
you,
I
mean
I,
think
I
mean
it's
been
a
while,
but
I
think
we
can
a
future
based
on
multiple
tags
at
once,
so
you
can
have
a
open
time
to
close
Express
filter
exactly
subscription
or
RSS,
even
and
and
instead
of
having
like
one.
F
Yeah
I
think
that
there's
a
two
questions
there
that
I
heard
one
is
if
we
want
that
hyphen
in
open
symmetry
I,
don't
I
think
that
ship
might
have
sailed
unless
somebody
wants
to
go
and
edit
that.
F
And
then
the
other
thing
is
that
having
an
open,
Telemetry
Dash,
whatever
language
specification,
would
help
go
specifically
because
there's
also
collector
questions
that
get
bundled
with
the
language
and
I.
Don't
know
if
there's
any
like
Java
might
be
another
one
where
there's
like
Auto
instrumentation
versus
the
actual
core
SDK,
but
I'm,
not
100
sure
on
that
one.
F
Yeah
that
sounds
good
to
me,
too.
Yeah.
A
F
I
So
about
the
dash
there's
something
called
synonyms
for
like
it's
a
tag:
synonyms,
yeah,
okay,
I
can
try
to
create
a
synonym
for
produce
one
here
as
well.
Okay,.
A
Perfect,
it
looks
like
this
one's
tagged
pretty
much
correctly
over
here
for
C
sharp,
interesting
okay,
this
one's
about
graphql
reporting
traces,
probably
here
with
Express
okay,
so
this
one
should
be
tagged
as
JavaScript.
It
already
has
Express
on
it.
A
A
This
is
specific
to
those
back
ends
that
looks
correctly
tagged
with
both
those
looks
like
this
one's
closed.
Hey
I
can
recognize
who
answered
this.
That's
Dan
down
looks
like
that
one's
answered
and
normally
we'll
just
go
back
one
week
because
of
the
first
time
it
was
going
to
scroll
to
the
end
of
the
page.
What's
the
current
date,
27
okay
use
case
for
multiple
tracers,
this
one's
specific
to
go
and
it's
tag
is
go
and
looks
like
we've
already
got
someone
from
on
this
call.
A
Answering
it
spring
boot
application
connection
looks
like
Michael's
already
answered
that.
A
H
A
A
This
is
still
language,
specific
okay,
so
we
should
tag
the
right
language
on
this.
Oh
typescript,
never
mind
I
apologize
for
openings,
okay,
so
it
looks
like
right
now.
We
just
have
a
few
that
need
to
be
tagged,
but
in
future
weeks
we
might
have
some
where
the
questions
are
about
basically
about
spec
or
about
non-language
or
non-component
parts
of
open,
Telemetry
and
we'll
go
through
this
every
week.
Aaron.
Thank
you
for
suggesting
and
reminding
us
to
do
this.
A
Okay,
well
with
that
I
think
we're
actually
done.
I'll
see
we'll
see
each
other
all
throughout
the
week
and
then
see
everyone
next
week.
Next
Monday
on
this
call
have
a
great
week.