►
From YouTube: 2022-08-03 meeting
Description
cncf-opentelemetry@cncf.io's Personal Meeting Room
B
Okay,
so
yeah,
I
said
on
the
first
item
here:
please
add
any
agenda
items
that
you'd
like
to
talk
about.
I
left
some
space
at
the
bottom,
for
that
first
thing
is
a
quick
metrics
ga
update.
There
has
not
been
much
movement
on
this
in
the
last
week.
B
B
A
B
B
That
is
no
longer
true,
so
we're
going
to
promote
at
least
the
trace
exporters
for
protobuf
and
grpc
to
stable
in
the
near
future.
B
So
if
anybody
has
any
concerns
about
that,
now
is
the
time
to
bring
it
up.
I
they
have
been
effectively
stable
for
a
long
time.
We
haven't
changed
the
interfaces
on
them
or
anything
like
that
in
quite
a
while,
but
just
so
you
guys
know
those
will
be
officially
promoted
to
stable
in
the
near
future.
Here.
B
A
related
point
on
that
is
that
the
protobuf
exporters
are
going
to
get
browser
support.
Currently
they
only
work
in
node,
but
because
of
some
changes
that
were
made
by
legendicus
recently,
namely
generating
javascript
code
for
the
proto.
Instead
of
consuming
the
the
protofiles
directly,
we
should
actually
be
able
to
introduce
browser
support
fairly
easily.
Here.
B
I
think
it's
going
to
be
a
matter
of
packaging
and
testing,
but
there
was
a
github
user
that
volunteered
to
do
this
wednesday
of
last
week.
I
know
we
didn't
have
a
meeting,
so
we
didn't
talk
about
it,
then,
but
I
have
not
seen
any
update
on
that.
Yet
I
don't
know
if
this
github
user
is
on
the
call
today.
I
don't
think
so,
though,
but
expect
that
to
come
in
in
the
near
future.
B
If
I
haven't
heard
anything
from
him,
probably
by
the
end
of
the
day,
I'll
probably
ask
for
an
update
actually
I'll
just
do
it
right
now.
A
B
B
Okay,
the
first
real
topic
of
discussion
here
I
am-
I
opened
this
issue
asking
if
we
should
rename
the
sdk
for
metrics.
B
B
The
the
base
in
the
name
comes
from
the
tracing
packages
so
in
tracing
we
have
the
the
trace
base
package
and
then
we
have
trace
web
and
trace
node,
but
in
metrics
we
don't
have
that.
We
just
have
a
a
single
package.
That's
meant
to
be
used
directly.
B
So
initially
I
was
thinking
we
should
just
remove
base
from
the
name
here,
but
then,
as
a
part
of
the
discussion,
we've
also
talked
about
removing
sdk
from
the
name
which
is
calling
it
open.
Telemetry
metrics,
so
we'd
have
metrics
api
and
we
would
have
metrics.
B
In
my
opinion,
I
think
that's
a
little
bit
more
straightforward
for
users
and
there's
a
linked
issue
down
here
where
somebody
was
confused
over
the
sdk
name,
it's
a
little
bit
too
late
for
tracing,
obviously,
but
not
too
late
for
metrics,
and
I
was
curious.
If
anybody
here
has
opinions
on
this
or
some
of
the
maintainers
have
have
been
discussing
this
a
little
bit,
but
I
don't
know
if
there's
anyone
here
that
feels
strongly
that
we
should
or
should
not
do
this.
B
If
not
that's,
okay,
all
you
know
talk
about
it.
Offline
and.
A
B
B
Now,
big
n
is
not
supported
on
all
platforms,
particularly
some
older
versions
of
node
and
some
some
older
browsers,
but
if
we
introduce
it
as
a
separate
api,
then
our
users,
who
aren't
targeting
those
platforms-
you
just
simply
not
use
it.
We
just
have
to
document
that
these
apis
are
only
usable
on
platforms
that
support
bigent,
which
is
most
modern
platforms.
B
There's
been
some
discussion
here.
I
don't
know
if
anybody
on
this
call
has
looked
into
this
or
has
has
been
following
this
discussion
or
has
opinions,
but
if,
if
you
do
now
is
the
time
to
to
voice
those
opinions
or
questions,
if
it's
not
clear
what
is
going
on
here,.
A
Yeah,
I
have
a
quick
question.
Sorry,
so
if
it
I'm
assuming
that
this
line
here,
you
were
talking
about
what
happens
if
the
big
n
is
bigger
than
64
bits.
B
B
It
actually
talks
about
different
bitness
overflows,
like
16-bit,
integers
and
32-bit
integers,
which
of
course
can
be
interpreted
or
which,
which
can
be
represented
as
a
64-bit
integer
easily,
but
we
don't
do
any
overflow
at
all
right
now.
So
if
you
have
a
counter
that
goes
above
the
max
safe
integer,
where
do
we
go,
then,
if
you're
only
adding
like
one
to
it,
eventually,
the
no
the
the
sum
will
stop
changing.
B
So
the
the
overflow
is
what
we
originally
were
talking
about,
and
then
the
bigent
support
was
kind
of
the
the
solution
to
it
like.
If
you're
going
to
have
numbers
that
go
that
high,
then
you
should
use
big
int
and
then
they
would
be
represented
in
proto
as
a
64-bit
integer
with
you
know,
correct
overflow,
which
is
easy
to
do.
There's
a
there's.
A
function
call
on
bigend,
specifically
to
just
represent
it
as
a
64-bit
integer.
A
B
So
there
is
like
a
big
end:
you
could
do
as
into
n.
You
can
actually
represent
it
as
any
bitterness
and
it
returns
whichever
so.
It
returns
a
javascript
or
returns
a
big
into
still,
but
if
you
are,
if
your
number
is
2
to
the
64
plus
1
right,
so
if
it's
outside
of
that
64-bit
integer
range,
it
would
represent
it
as
a
very
large
negative
number.
B
Instead
of
you
know
a
very
large
positive
number,
which
does
fit
inside
the
64-bit
signed
range,
and
then
you
would
just
convert
that
to
assign
64-bit
integer
in
the
proto,
as
you
normally
would.
A
Okay,
maybe
I
need
to
read
a
little
bit
about
it
offline,
I'm
wondering
if
you
also
considered
like
when
it
when
it
hits
the
max
safe
in
letting
it
overflow
back
to
the
smallest
value.
So
we
sort
of
just
would
follow
the
guidelines
for
the
general
overflow
rules.
B
Yeah,
so
we
did
consider
that
that
would
be
when
you
get
to
2
to
the
53.
You
just
wrap
it
back
to
negative
2
to
the
53.
I
guess
right
so
that
would
be
here
stay
within
the
javascript
safe
range.
B
A
Yeah
so
you're
saying
you
might
think
that
you
subtract
it
load
further
than
you
than
you
actually
did.
B
Yeah
or
that
we
added
like
an
extremely
large
number,
all
of
a
sudden
okay,
another
option
was
instead
of
wrapping
to
negative
you
just
reset
it
to
zero,
then
in
the
most
back
ends,
I
think
would
would
detect
that
as
a
reset
and
the
worst
that
would
happen.
Is
you
lose
a
single.
B
Addition
potentially,
which,
depending
on
your
opinion,
is
a
big
deal
or
isn't.
I
think
it's
better
than
than
a
53-bit,
integer
being
wrongly
interpreted
as
a
64-bit
integer,
but
still
not
completely
ideal.
A
Sense,
I
think
I'll
read
through
as
well,
but
it
sounds
like
you,
like.
You
all
covered
everything.
B
Yeah
we've
been
discussing
this
for
for
quite
a
while.
I
know
we
didn't
have
a
meeting
last
week,
so
there's
been
quite
a
bit
of
discussion
and
some
of
the
other
maintainers
have
been
on
vacation
but
have
been
responding
on
github
anyways.
So
we've
had
more
more
async
discussion
than
usual
this
week.
A
B
B
Yeah,
does
anyone
have
additional
questions
about
that
or
anything
that
you
think
we
haven't
thought
of.
B
Okay,
the
next
item
here
is
the
logs
and
events
prototype.
It
looks
like
somebody
else
added
a
couple
little
items
here.
I
suspect
that
that
is
martin
yeah.
B
Would
you
like
to
talk
about
this?
I
was
just
putting
it
on
the
the
agenda
here
so
that
people
would
be
aware
that
it
exists,
but
I
don't
know
if
yeah
any
further
than
that.
A
Yeah,
thank
you
yeah
the.
So
there
is
a
there
is
an
corresponding
pr
in
the
spec
repo
to
for
adding
the
events
api
and
that
pr
is
still
in
progress,
so
it's
not
merged
yet.
So
I
open
this
open
this
implementation
as
a
draft
for
now
until
the
specs
are
done,
but
I
don't
anticipate
you
know
it's
pretty
straightforward,
so
I
don't
anticipate
there
to
be
many.
A
You
know
any
changes
really
so
yeah
I
if,
if
folks,
are
interested
in
in
logs
logs
or
events
api,
please
take
a
look
at
this
in
comments.
Quick
question,
martin.
I
know
that
there's
some
locks
and
events
as
a
code
for
python
and
other
languages.
B
B
Yeah,
in
addition
to
that,
I
know
that
there
were
some
prototypes,
including
this
javascript
prototype,
which
existed
in
a
different
repository.
There
were
some
prototypes
during
the
the
otep
process.
A
So
I
I
I
did
try
to
I'm
working
on
cleaning
this
up
and
I've
I've
added
tests
to
make
it
as
complete
as
as
possible
once
the
spec
is
ready.
So
this
can
be
ready
as
well.
One
one
thing
that
I
I
just
a
quick
question
it.
B
So
all
of
the
tests
are
timing
out,
usually
when
that
happens,
it's
because
in
some
package,
like
the
the
done
callback,
is
not
called
now
the
the
test
itself
should
time
out
after
I
think
two
seconds
usually,
but
I
think
in
some
packages
we
have
disabled
the
timeouts.
B
A
Well,
I
think
definitely
the
api
design-
that's
probably
number
one.
I
don't
know
which
I
don't
know.
Maybe
I'm
looking
for
your
guidance
here
like
should
we
wait
until
the
spec
is,
is
actually
merged
for,
like
a
full
review.
B
I
mean
most
of
the
time
we
wait
for
spec
to
be
merged,
but
if,
if
this
is
a
prototype
that
the
that
you're
working
on
the
spec
and
this
at
the
same
time,
I
I
don't
see
any
reason
to
not
review
this.
Because
of
that
I
mean,
I
know
as
part
of
the
specification
process
they
want
to
have
prototypes
and
if
the
prototypes
don't
get
reviewed,
then
so
yeah,
I
I
think
we
can
do
both
at
the
same
time,
especially
since
I
don't
expect
that
this
is
a
particularly
complex
api.
B
A
B
B
Okay-
the
next
item-
here
I
don't
know
if
it
actually
is
affecting
anyone
here,
but
the
current
js
build-
is
failing
on
windows
platforms
because
of
the
code
generation
in
the
protobuf
exporters.
I
guess
it's
just
a
a
matter
of
using
the
tool
in
a
slightly
different
way.
It's
not
that
it
doesn't
support
windows
or
anything
like
that.
It's
just
that
all
of
the
maintainers
are
on,
I
think,
macintosh
machines.
So
it's
tough
for
us
to
know
when
things
are
breaking
on
windows.
B
So
as
a
result
of
that
we're
going
to
add
a
windows
build
target
in
the
ci
just
so
that
we
know
when
the
project
is
failing
to
build
on
windows,
I
don't
have
any
particular
anything
else,
particularly
to
say
about
this
other
than
that
this
is
coming,
and
I
don't
know
if
it's
affected
anyone
here.
I
know
that
we
do
have
some
microsoft
employees,
so
it's
possible
that
I
know
at
least
some
of
you
are
using
windows.
A
Yeah,
there's
a
pr
also
to
fix
the
issue.
It's
actually
working
for
me,
so
I
believe
it
needs
to
be
merged,
needs
some
approval.
It
will
be
good
to
push
that
one
forward,
I'm
actually
adding
the
changes
locally
to
unblock
me,
but.
B
B
Yeah
compile
as
we're
going
to
find
in
windows
after
this
update.
Okay,
I
have
not
approved
this
mostly
because
I
don't
have
a
windows
machine
to
test
it
on.
B
B
If
it
seems
to
be
working,
then
we'll
just
go
for
it.
It
certainly
won't
make
it
worse
than
it
is
now
right.
B
Okay,
so
merged
this
next
item.
Some
of
you
may
already
have
been
familiar
with
this.
This
bug
has
been
open
for
a
really
long
time.
The
abstract
async
hooks
context
manager
breaks
removeless.
So
if
you,
you
have
an
event
emitter
that
is
bound
with
the
async
hooks
context
manager
and
you
use
once
to
add
a
callback,
and
then
you
use
remove
listener
to
remove
that
callback.
It
does
not
get
removed,
which
is
obviously
a
problem.
B
So
I
have
a
fix
for
this,
which
prevents
double
wrapping
the
event.
Emitter
listeners,
if
you're
interested
in
what
the
actual
bug
was,
there's
quite
a
bit
of
discussion
about
it
on
the
original
issue,
but
the
fix
is
just
to
the
short
version.
Is
that
once
calls
on
and
because
we
patch
once
and
we
patch
on,
we
end
up
double
wrapping
the
listener
which
breaks
the
the
removal
so
detecting
that
we
are.
You
know
in
a
situation
where
once
is
calling
on
prevents
this.
B
B
Are
tests
you
can
you
can
verify
that
they
pass
now
and
they've
failed
before?
But
please
review
that
as
this
is
a
p1
bug
that
is,
was
actually
reported
by
an
end
user.
It
was
causing
their
system
to
crash.
So
it
is
an
important
one
and
I
believe,
a
relatively
straightforward
fix.
So
please.
A
B
Okay,
missing
semantic
conventions
for
metrics.
I
don't
know
who
added
this
here.
A
Yeah
I
added
this
one.
Basically,
I
think
it
would
be
good
to
have
the
semantic
convention
for
metrics
as
part
of
the
ga,
and
I
realized
that
there's
a
specification
prevented
us
to
do
this.
There's
some
script
in
the
js
repo
that
builds
the
the
package
using
the
jaml
files,
but
there's
some
and
I
look
at
the
specs,
and
I
found
that
there's
a
like
a
pr
fixing
this
in
the
build
tools,
but
it
has
been
there
for
eight
months.
So
do
anyone
have
some
extra
knowledge
about?
A
B
B
A
Yeah
I
shared
the
actual
issue,
someone
created
the
issue
saying
that
it's
not
available,
and
they
mentioned
this
pr-
is
preventing
that
this
one
as
well.
So.
A
B
Yeah
I
mean
our
semantic
conventions
package
is
stable.
I
don't,
I
agree
with
you
that
we
should
have
metrics
in
the
semantic
conventions,
but
I
think
the
metric
semantic
conventions
is
trailing
so
far
behind
the
metrics
sdk
specification.
B
I
don't
know
if
we
can,
I
would
much
rather
have
it
done
with
code
generation.
Let
me
just
say
that
I
guess,
because
maintaining
it
manually
by
hand
is
verging
not
impossible.
It's
just
too
much
work.
B
I
mean
I,
I
don't
know
how
we
can
move
this
forward,
the
I
I'm
honestly
surprised
that
the
ammo
files
aren't
already
there,
because
there's
so
many
symmetric
semantic
inventions
that
already
exist,
I'm
shocked
that
they
added
them
all
without
having
yaml
files.
I
didn't
realize
that
that
had
even
happened.
B
Yeah
to
me
this
seems
more
to
be
a
failing
of
the
specification.
I
I'm
surprised,
there's
no
yaml
files-
I
I
don't
have
anything
else
to
really
even
say
about
it.
A
Lie
idea,
it
looks
like
it
is,
but
my
understanding
is
the
ones
in
the
spec
aren't
stable,
they're
marked
experimental,
for
instance,
like
http.
A
A
A
B
B
Not
a
good
way
to
go
about
it
in
the
future,
though,
so
I
have
to
think
about
that
a
little
bit.
I
guess
the
the
resolution
to
this
particular
agenda
item.
Is
we
really
don't
have
a
resolution.
A
B
Suspect,
that's
probably
not
the
most
satisfying
answer,
but
that's
what
I
have
for
you
right
now.
B
Contribute
limiting
for
releases.
No,
I
don't
have
an
update
on
this.
I
just
haven't
had
time
to
get
to
this.
I'm
sorry.
C
Is
it
this
is
jamie
yeah.
B
C
It
is
it
helpful
if,
like
is
there
something
that
I
could
do
or
or
talk
with
anyone
like
I
don't
know
like,
is
it
like
governance
committee
people
or
anything
like
that,
mostly
just
because
we're
getting
a
little
bit
of
pressure
for
having
the
contrib
release
out
and
I'm
not
sure
how
best
to
help,
and
if
it's
something
like
that
like
should
we
reach
out
to
anyone
else,
because
I
know
you
have
a
lot
on
your
plate
too.
Obviously
we
do,
but,
like
anything,
we
can
do
to
help
with
that.
A
B
I
mean
I
don't
I
guess
this
would
be
a
governance
committee
topic,
I'm
not
sure
whether
it'd
be
gc
or
tc,
but
I
think
it's
more
to
do
with
looking
through
the
github
documentation
to
see
if
there
even
is
a
possibility
of
paying
to
increase
the
rate
limit.
I
don't
know
if
that
would
be.
B
You
know
if,
if
there's,
if
it's
an
undocumented
option,
then
I
certainly
don't
have
any
contacts
at
github
to
help
me
figure
figure
that
out.
So
it's
just
a
matter
of
of
looking
at
the
the
the
payment
tiers
and
figuring
out
which
one
we're
on
I
I
could
pretty
easily
figure
out
which
tier
we're
on.
I
think
the
the
tc
should
know
that
you
know
someone
like
carlos
or
tigran,
or
someone
like
that
should
know.
B
But
I
don't
know
if
there
is
a
additional
tier
that
would
let
us
increase
our
rate
limit.
I
can
say
for
sure
there
is
no
update
on
the
release.
Please
side
so
we
did
create
a
couple
of
issues,
letting
them
know
that
we're
hitting
rate
limiting
and
asking
for
ways
to
reduce
api
consumption
during
releases
and
they
have
not
they've
responded.
They've
been
responsive,
but
there's
no
there's
been
no
movement
on
that.
B
Yet
so
I
can
say
for
sure,
there's
nothing
from
that
side
in
terms
of
the
the
tier
that
we're
on
with
github
and
paying
for
more
headroom,
I'm
not
sure
what
the
options
even
are
yet.
C
Okay,
I
mean,
I
guess
what
I
can
do
if
it's
helpful,
I
can
maybe
try
reaching
out
to
tigran
or
someone
to
see.
I
don't
remember
if
there's
anyone
here
at
honeycomb
2
is
also
remember.
That
would
be
able
to
help
like
if
liz
or
someone
would
know
that.
C
Other
people
are
able
to
peak
too
and
like
again,
there
might
not
be
any
other
options
just
trying
to
see
again,
because
if
it's
like
a
little
opaque
right
like
I
can't
see,
if
there's
anything
else,
we
can
do
or
not
so
just
getting
a
little
more
info
about
that.
I'm
happy
to
kind
of
ask
around
as
well.
If
that
makes
sense.
B
Yeah,
I
I
just
don't
know
like
if
there's
something
that's
not
public,
I
don't
know
who
would
be
the
correct
person
to
even
ask
about
it,
and
liz
is
definitely
on
the
gc
right.
She
definitely
has
contacts
at
the
cncf
who
might
know
so
she
may
be
a
good
person
to
ask,
but
I
don't
know
if
she
has
any
contacts
at
github
to
ask
or
anything
like
that.
I
know
that.
A
B
The
ruby
maintainers
are
github
employees
I
think
I
saw
matt
was
on
the
call
matt.
Are
you
still
involved
in
the
ruby
sig
and
do
you
know
any
of
the
the
github
employees
that
might
be
able
to
help
it
answer?
This
question
for
us
yeah.
A
I'm
I'm
involved.
I
know
there
is
a
member
of
the
observability
team
from
github
that
is
a
contrib
maintainer
reo
valentine,
okay.
I
think
he
might
be
able
to
answer
something:
okay,.
C
Yeah,
I
think
I've
I've
seen
him
on
there.
I
think
rob
on
our
team
works
with
him
on
some
stuff,
with
ruby.
A
C
B
Okay,
thank
you,
and
I
guess
for
now,
if,
when
you
say
we're
we're
getting
pressure
to
release
katrib,
I
assume
we
is
honeycomb
in
that
instance,.
C
Yeah,
mostly
just
because
there
was,
like
a
you,
know,
a
small
update
that
we
had
made
to
one
of
the
contrib
packages,
and
I
I
don't
know
how
huge
of
a
deal
it
is,
but
it's
more
of
like
there's
kind
of
an
open
issue
on
our
side
that
we
want
to
get
squared
away.
C
I
think
the
customer
might
be
using
like
a
forked
version
or
something
that
has
the
the
patch
that
was
in
there,
but
just
mostly
concerned
about
things
getting
out
of
date
and
hard
to
maintain,
and
you
know,
is
it
like
a
longer
term
thing
too
of?
Obviously,
it's
not
good,
for
you
know
anyone
if,
if
this
is
a
problem,
if
there's
something
that
could
be
done
to
make
it
better,
not
just
for
this
particular
one,
this
is
just
what's
making
it
more
top
of
mind.
Probably.
B
Just
doesn't
have
any
approvals,
so
I
actually
can't
merge
it
nah.
I
don't
know
which
of
these
instrumentations
you're
you're.
Looking.
B
So
generally,
I
do
my
best
to
try
to
merge
these
as
quickly
as
I
can,
but
if
they
don't
have
approvals,
I
I
actually
just
can't
make.
C
B
One
of
the
options
for
release:
please
is
instead
of
creating
a
pr
it
can
just
release
packages
every
time.
There's
a
merge
to
main.
I
just
haven't
done
that
because
it
seems
a
little
bit
risky
and
I
can't
even
really
articulate
why
it
feels
risky
to
me
it
just
does.
B
B
Lose
that
the
build's,
also
failing
here
the
build,
is
failing,
I
think,
in
all
of
contrib
right
now.
So
that's
what
I
was
working
on
before
the
the
meeting
today
is
fixing
the
contrib
release.
So
as
soon
as
I
get
that
done,
I
will
make
sure
to
get
this
release
out.
B
Yeah,
the
the
maintainers
you
typically
work
on
contrib
are
on
vacation
right
now
on
extended
summer
vacations.
So
that's
part
of
the
reason
that
contrib
has
been
going
a
little
bit
slower.
The
last
few
weeks.
C
B
All
right,
I
guess,
is
there
anything
else
here.
Should
we
move
on.
A
B
B
Then,
let's
try
to
go
through
some
of
these
issues.
I
believe
this
is
one
we
talked
about
before,
isn't
it
I
said
I
would
look
into
the
sample
code.
That's
right!
I
had
not
done
that
yet.
B
B
It
doesn't
look
like
she's
on
a
call
today
yeah.
This
is
just
a
matter
of.
I
have
to
look
into
this,
so
we
don't
need
to
talk
about
that
one
today,
exporter
zipkin
does
not
send
content
type
in
default
config.
We
had
asked
for
more
information
here
and
it's
not
yet
provided
getting
a
different
number
of
spans
and
traces
for
the
same
http
endpoint
on
high
load.
B
B
A
B
A
B
You
know
he
used
to
be
a
maintainer
for
those
that
don't
know
who
he
is
and
hasn't
been
around
on
the
project
for
a
while.
So
I
won't
be
surprised
if
we
get
no
response
there.
B
B
I
think
that
this
one
can
just
be
closed,
though.
Let's
see
I
already
assigned
myself.
B
B
B
All
right,
anyone
have
anything
that
they'd
like
to
bring
up
before
I
close
the
callback.
B
Yeah
no
problem,
can
you
send
me
a
link
to
that
on
slack
yep
yeah?
Okay,
am
I
looking
at
anything
specific
on
there
or
just
generally
review
the
pr.
A
B
All
right,
yeah
send
it
to
me
on
slack
and
I
will
I'll
get
that
reviewed,
because
I
don't
want
to
block
your
work,
cheers.
B
A
B
Right,
thank
you
everybody
for
your
time,
and
I
will
talk
to
you
next
week.