►
From YouTube: 2022-08-09 meeting
Description
cncf-opentelemetry@cncf.io's Personal Meeting Room
A
A
B
C
You
hear
me
now:
yep,
perfect
yeah.
Thank
you.
So
within
messaging's
sake,
we're
trying
to
find
the
right
way
to
represent
to
trace
batching
scenarios
and
the
problem
there
is
that
each
individual
message
within
a
badge
should
be
traceable,
and
it
means
that
each
message
should
have
a
unique
context
and
the
question
we,
it
seems
it's
wider
than
the
messaging
seek
and
we
want
to
get
by
their
opinion
on
is
different
options
and
how
we
do
this
and
the
couple
of
options
we
have.
C
We
don't
think
we
didn't
include
some
esoteric
options,
but
basically
the
two
options
are
that
we
create
a
span
for
message
and
it's
unconventional,
because
at
least
in
some
cases
it
will
have
zero
duration
and
won't
represent
much
except
the
the
context
and
the
other
approach
we
are
trying
to
evaluate.
It
is
if
we
don't
create
a
spend
and
we
just
create
a
context
per
message
and
in
both
cases
we
would
add,
links
to
this
context.
C
Right,
we
don't
need
to
link.
We
don't
need
spam
to
create
a
link
to
the
context
yeah,
so
we
really
want
to
get
by
their
community
opinions
on
this
and
help
us
make
some
progress
here.
We
are
worried
that
if
we
continue
working
it,
it
blocks
our
progress
on
supporting
like
matching
and
messaging
spec.
C
A
Yeah,
that
makes
sense.
I
I
see
by
the
way
that
there's
no
preference
over
any
of
the
two
options
yet
so
it
means
that
there's
a
lot
of
discussion
to
you
know
waiting
there.
A
Okay,
thank
you
so
much
for
that.
Yeah.
Let's
continue
discussing
the
offline
unless
somebody
wants
to
say
something
here,
but
they
don't
see
any
of
the
messaging
group
or
not
many
of
the
members.
C
A
D
So
what
you
said
about,
if
there
is
any
preference,
we
don't
have
any
preference.
I
also
joined
the
meeting
with
dude
miller,
but
there's
my
feelings
that
some
people
are
really
opposed
to
because
they
already
brought
multiple
times.
D
The
zero
duration
expands
and
also
people
brought
the
cost
factor
of
it
that
if
you
have
a
batch
with
a
lot
of
spans
and
you
just
create
a
lot
of
spans
and
and
not
really
they're,
not
really
useful,
and
also
that
they
make
the
visualization
messy
and
and
and
noisy
and
so
on,
but
there's
no
real,
strong
anything
for
we're
like
the
dominance
that
we're
just
trying
to
get
and
gauge
the
opinion
of
of
the
broader
audience.
A
Yeah,
that
makes
sense.
I
think
we
hear
it
like
that.
We
started
playing
with
that
recently,
so
we
will
probably
provide
some
feedback
on
that
because
on
the
second
option,
it's
fun
less
context.
I
see
that
there's
a
cons
regarding
that.
Probably
this
will
be
more
complicated
indexing
wise
on
the
back
end,
so
yeah.
B
Okay,
probably
a
a
general
thought
like
in
the
the
rum
sake
we
had
something
similar,
which
is
why
we're
using
logs
to
send
events
and
as
part
of
an
event
we
have
a
domain
and
then
a
name.
So
one
possible
option
would
be
that
you
have
a
domain
which
is
messaging.
I
don't
know
what's
what
else
is
in
the
message,
but
logs
do
have
the
advantage
of
supporting
nested
objects,
so
you
don't
have
to
try
and
flatten
them
to
send
them
down.
C
B
Anyhow,
yeah
a
just
a
general
thought
that
yeah
yeah,
because
I
see
one
of
the
comments
in
here
is
have
you
thought
about
the
transport?
So
that's
what
triggered
my
thought
on
that.
One.
C
I
think
the
transparent
wise
we
have
all
options
there.
We
have
stan
links
that
have
all
we
need.
We
have
spans
and
I
think,
transferred
white
is
is
well
thought
out.
We
don't
need
anything
there,
okay,
cool
regarding
events,
the
esoteric
options.
We
thought
about
that.
If
we
don't
need
a
span
and
it's
a
zero
duration
span,
maybe
it's
an
event.
Then
we
would
need
a
convention
on
how
to
express
the
context
through
event
that
usually
events
don't
have
a
unique
context.
E
E
C
Yeah,
so
if
we
don't
yes
pants,
but
we
still
want
a
notification
of
some
sort
that
we
can
use
event,
but
then
it's
a
new
concept
where
we
would
say:
okay,
events
can
have
a
unique
context
and
then
we
would
need
a
convention,
an
api
for
this.
E
C
B
B
C
All
right
and
then
we
need
an
api
that
could
create
an
in
inventories
context
right,
it's
a
special
sort
of
event,
not
every
event
has
it
it's
less
convenient,
but
the
convention
to
say,
okay.
This
is
the
name
of
the
attribute
that
represents
context.
B
C
Okay,
I
think
somebody
wants
to
discuss
something.
Maybe
please,
if
you
have
an
opinion,
if
your
backhand
works
on
links
and
you
want
to
support
batching,
please
comment
comment
on
this
issue.
If
we
don't
solve
it
now,
we
will
have
to
solve
it
later.
I'm
sorry,
thank
you.
A
Yeah,
thank
you
so
much
for
that.
Let's
continue
discussing
that
offline.
Okay,
then
next
item
hotel
sdk
enabled
we
need
more
reviews.
We
discussed
that
last
week
it's
about
being
able
to
disable
the
sdk.
A
A
So
do
we
have
python
golang
maintainers?
Here
I
mean
the
java
crew
already
approved
these.
So
that's
that's
great,
but
we
really
need
maintainers
to
take
a
look
at
this.
E
It's
probably
worth
mentioning
at
this
point
that
support
for
this
is
optional,
and
so
far
only
the
java
sdk
seems
to
be
interested
in
supporting
it,
because
actually
they
already
do,
but
they
want
to
have
it
formalized,
but
it's
not
clear
if
any
of
the
other
sdks
would
ever
support
it.
But
if
they
do,
it
would
be
good
if
it
was
specified
innovative
work
for
them.
A
B
Okay,
it
makes
sense
I
can
tell
you,
I
supported
this.
I've
tried
to
enable
the
other
sdk
into
your
java
application,
and
this
would
have
been
very
useful.
A
So
let
me
provoke
yeah,
go
ahead.
Sorry.
A
Perfect,
okay,
I
will
contact.
I
don't
see
python
nor
golang
maintainers
here
in
the
call,
but
I
will
contact
them.
I'm
here,
oh
perfect,
so
any
opinion
on
that
tyler.
F
We've
been
thinking
about
it.
I
think
I
have
some
follow-up
to
do.
We
were
looking
at
the
comments
that
were
looked
to
be
addressed
and
I
think
they've
been
addressed,
but
I
haven't
followed
up
on
what
they
were.
I
don't
know
if
there's
real
value
in
in
the
go
set
up
just
because
you
have
to
do
so
much
to
instrument
it
but
I'll
take
a
look.
A
A
Okay,
in
that
case,
the
follow-up
at
least
would
be
for
for
python
really
well,
I
will
rally
you
around
no.
B
A
Oh
yeah
you're
around.
I
wonder
whether
you
have
experience
sorry,
some
feedback
from
the.net
or
c
plus
plus
cruz
on
this
one.
A
A
Okay:
let's,
let's
talk
a
little
bit
of
lying
about
that
yeah,
we
can
listen
to
you.
Maybe
we
can
listen
to
you
later,
okay,
in
any
case
yeah,
it's
just
enough
reviews
waiting
for
tyler's
and
we'll
contact
the
python
maintainers
and
from
other
sigs.
A
Let's,
let's
take
a
look
at
that,
as
I
said
before,
it
has
enough
reviews
so
yeah
we
should
take
a
decision
very
soon.
The
next
one
is
antoine,
who
I
think
just
left,
but
basically
it's
about
adding
process
threats.
Metrics,
it's
a
single
matrix.
A
Yeah,
so
basically
gemma
did
you
want
to
mention
your
request.
Changes.
G
Yeah,
I've
been
working
a
lot
in
the
runtime
metrics
space,
particularly
in
the
go
group
right
now,
and
I've
been
studying
them
a
lot,
and
I
think
that
this
qualifies
as
one
of
the
language
implementation
details
that
is
a
runtime
metric
and
go.
We
don't
have
threads
that
you
know
it'd
be
silly
to
try
and
count
that,
but
we
do
have
a
runtime
metric
already,
you
know,
go
routine
count
is
already
there,
and
I
think
that
this
is
one
of
those
language
specific
items.
G
You
might
have
a
runtime
that
just
doesn't
expose
threads
at
all,
and
so
I
think
it
belongs.
There.
A
G
A
Okay,
yeah!
That's
what
antoine
had
mentioned.
Also
that,
like
he,
believes
that
process
threats
is,
you
know,
won't
be
using
a
namespace
okay,
perfect
any
other
commenter.
A
A
We
have
a
few
reviews
there,
a
few
approvals
there
are
some
still
there
were
some
comments
from
bogdan
and
jack.
So
please
review
that
again.
If
you
can.
A
Well,
thanks
rayleigh
for
replying
to
my
gmat.
Please
comment
on
the
thread
count
pr,
okay,
moving
on
then
there's
one
more
that
I
added
here.
Well,
a
few
more
one
of
them
is
about
changing
the
inclusivity
of
exponential
histogram
bounds
from
lower
to
upper
inclusive
bounds,
just
to
work
better
with
permit
use.
I
think
geometry.
You
were
asking
for
feedback
last
week
we
still
want
dives
yeah.
G
This
is
a
really
technical
issue
that
we
study
it.
You
know
if
you
do
study
it,
please
approve
or
disapprove.
This
is
a
it's
been
written
before.
I
don't
want
to
speak
much
about
it
in
this
meeting.
This
is
a
political
issue
and
a
hardly
a
technical
one.
At
this
point,
so
please
consider
the
the
issue.
A
Yeah,
okay,
make
sense
great,
I
mean
technically
we
can
merge
that
we
have
a
pair
of
approvals
from
very
involved
people
yeah.
G
Yeah,
so
my
concern
is
that
josh
surf
and
no
one
from
google
has
approved
this.
It
I'm
afraid
to
start
tying
this
request
of
theirs
to
all
the
requests
that
we
have
from
prometheus
things.
That
could
be
done
better
in
our
opinion
or
my
opinion.
G
So
I
don't
want
to
make
this
a
bargaining
chip.
I
just
want
to
do
the
right
thing
for
prometheus
and
I'd
like
to
also
ask
them
to
do
the
right
thing
for
us
certain
things,
I'm
not
talking
about
them
right
now,
but
there's
a
there's
a
list
of
things
that
are
difficult
for
us
because
of
the
way
prometheus
interoperates.
A
Okay,
yeah,
I
see
a
pair
of
comments
from
riley
that
he
prototyped
this
in.net,
it
works.
Well,
ceo
will
be
doing
another
review
pass
and
he
also
really
asks
joshua's
review
and
he
will
do
that
after
he
comes
back
from
holidays.
Does
anybody
remember
like
right
now,
when
is
that
happening.
A
F
G
Do
you
see
the
associated
issue?
I
mean
there's
a
lengthy
discussion
where
josh
did
provide
his
feedback,
so
it's
worth
it's
worth
reading
through
the
math
has
definitely
been
validated
by
a
few
people.
At
this
point.
Thank
you.
G
A
Perfect,
thank
you
both
now
for
slacking,
david
and
josh.
Okay.
The
next
item
then
add
user
agent
to
the
otlpix
forwarders.
Basically,
it's
just
adding
a
header
when
you
export
data
using
the
autumn
speaks
borders,
and
this
already
happens
in
the
collector.
According
to
the
alex,
that's
the
alex
pr
sorry,
it
has
a
few
reviews.
It's
relatively
a
minor
item.
We
need
more
eyes.
A
So
please
review
that
and
the
next
one
was
about
adding
hardware
metrics,
which
was
already
merged.
This
was
very
specific
about
hardware
metrics
and
it
was
interviewed
forever,
but
it
got
enough
reviews
and
advice.
So
please
follow
up
in
case
you're.
You
are
an
expert
or
are
very
interested
about
this
one.
We
can
always
do
follow-ups.
A
Okay,
the
next
one
dashboard
opens,
let's
just
open
some
images.
F
Hey
yeah,
so
I
just
wanted
to
introduce
it
briefly
and
if
you
worked
on
open
census
or
are
invested
in
helping
people
migrate
off,
please
take
a
look.
The
motivation
for
this
is
that
I
think
we
should
specify
metrics
bridges
soon.
Right
now
we
have
a
spec
for
trace
bridges
and
before
I
did
that,
I
wanted
to
provide
some
context
on
how
we
expect
the
bridges
to
be
used
both
by
users
and
by
third
party
libraries
that
are
currently
written
with
open
census
instrumentation.
A
Yeah
great,
I
think
you
will,
by
the
way
try
to
you,
will
have
to
get
some
additional
reviews
there,
because
tips
require
like
five
or
four
and
unless
you
can
find
open
competitions,
people
that
can
review
that,
but
anyway
yeah.
I
am
happy
to
review
that
later
this
week.
E
Hi
sorry
for
getting
so
late
today,
I
still
have
the
request
about
the
enable
property-
I
don't
know
if
you
guys
have
discussed
it
in
this
meeting.
A
Yeah
we
mentioned
that
it
has
enough
approvals,
but
I
would
like
to
get
some
you
know
hint
from
maintainers.
The
job
java
crew
is
happy
with
that,
but
yeah.
So
I
asked
tyler
who's
in
the
call
to
consider
reviewing
this.
I
will
contact
the
python
crew
as
well.
A
Yeah
I
mean
armin
mentioned
reminded
us
that
this
feature
would
be
optional.
So
not
all
sdks
would
be
like
super
required
to
implement
this,
but
if
anybody
wants
to
do
that,
they
are
good
to
to
do
that
yeah
anything
else.
You
want
to
comment
on
the
technical
side.
Maybe.
E
No,
I
think,
after
the
the
discussion
on
the
pr
I
I
think
I'm
happy
with
the
options
that
were
took.
I
I'm
now
more
concerned
about
merging
it
and
get
it.
A
Yeah
yeah,
that
makes
sense
yeah.
I
will
contact
here
a
developer.
We
have
from
the
python
maintainer
group,
so
hopefully
we
can
get
some
eyes
on
that.
One
thanks
very
much.