►
From YouTube: 2022-10-07 meeting
Description
cncf-opentelemetry meeting-2's Personal Meeting Room
A
Foreign
for
the
recording
I'm
going
to
wait
till
about
five
after
to
start
see
if
anybody
joins
me,
if
not
I,
will
record
kind
of
what
I've
done
where
I
will.
B
Be
going
with
this
and
hopefully
an
opportunity
to
let
other
people
follow
in
my
stead
so
that
I
don't
have
to
be
the
only
one
doing
this.
A
Well,
it
looks
like
I
might
be
doing
this
solo,
so
let
me
share
my
screen
and
show
what
how
this
will
be
run.
Typically,
So.
B
Currently,
I
am
going
to
put
the
notes
in
the
same
area
as
the
go
Sig
meeting.
If
this
becomes
too
noisy,
we
can
exercise
them.
B
Like
other
the
other
meetings,
these
are.
This
is
where
we'll
record
the
attendees
and
agenda.
The
last
item
on
the
agenda
will
be
to
work
the
no
status
issues
in
order.
If
anybody
who
comes
in
wants
to
add
a
particular
one
that
they
want
to
talk
about.
First
then
we'll
put
that
off
the
top
of
the
agenda.
There's
no
reason
to
do
that
after
we
try
and
work
all
the
status
issues,
because
I'm
the
only
one
here
I'm
going
to
introduce
the
the
beta
board.
B
This
is
already
created,
as
you
can
see,
I've
added
a
couple
labels
or
I've
added
a
label
on
called
triage,
and
it
gives
us
the
selection
of
triage
done.
B
B
That's
the
goal
of
this
I've
also
in
this
view,
exposed
the
any
PRS
that
are
attached
to
the
issue,
so
we
don't
have
to
put
the
PRS
in
the
what
you
would
call
it
in
the
the
project
board,
but
they
need
to
be
attached
to
an
issue
for
them
to
be
tracked.
Otherwise,
and
the
goal
with
this
is
to
go
through
each
of
these
issues
ask
the
same
set
of
questions.
Does
it
need
to
be
here?
B
Can
it
be
worked
or
does
it
need
new
subtasks?
Is
it
blocked
by
anything
like?
Is
there
something
that
needs
to
happen
before
we
can
actually
work
on
this
one?
And
finally,
whether
or
not
there
is
an
active
PR
on
it.
So
those
questions,
if
it
doesn't
need
to
be
in
the
in
the
beta,
then
we
re-home
it
either
uncategorized
for
right
now,
like
out
out
of
any
project
or
later.
B
If
we
find
it,
if
we
make
a
metric
SDK
release
project,
we
can
put
it
there
if
it
needs
to
be
broken
down
in
subtext.
That's
going
to
be
an
action
item
for
after
the
fact,
if
there
are
is
an
open
or
if
it's
blocked
it
will
go
into
the
block
column
here.
B
If
it
is,
if
it
has
a
knack
of
PR,
then
we
put
it
in
the
Pro
in
progress
and
then,
finally,
if
it
has
none
of
if
it
falls
into
none
of
the
other
categories,
it's
in
to
do
because
it
should
be
ready
to
be
worked,
but
we
just
haven't
started
yet
so
I
will
go
through
one
that
I've
already
triaged
a
little
bit
because
it
was
related
to
a
PR.
I
was
already
working
on
and
it
was
already
linked
to
this.
B
So
there
is
for
this
one.
There
is
a
name
Collision,
essentially
in
the
default
case,
when
we
import
the
otel
Prometheus,
it
will
collide
with
the
client
golang
Prometheus.
This
causes
some
friction
for
new
users.
B
I've
taken
the
stance
like
I
think
we've
we've
come
to
an
agreement
that
as
long
as
this
doesn't
happen
in
the
default
case,
it's
okay,
if
it
collides.
So
there
is
a
PR
open
already
to
do
that.
So
I
appropriately
put
triage
done
and
put
it
over
into
the
in
progress.
I've
also
found
that
there
was
a
couple
issues
that
have
APR
outstanding,
so.
A
B
A
Of
I'm
going
to
take
it
out
of
the
the
beta
for
right
now
and.
A
Downscope
one
less
issue
return
a
pointer
instead
of.
A
B
Know
what
we'll
call
this
triage
done
and
put
this
into
in
progress
if
that
PR
doesn't
doesn't
land,
because
it
doesn't
actually
help
with
the
performance,
then,
potentially
we
can
revisit
this
and
remove.
It
seems
a
bit
strange
to
need
to
import
Prometheus
directly,
actually,
I
think
this
is
just
a
duplicate
of
this
one.
B
Metric
SDK
caching
I
actually
think
that
there
are
a
number
of
PR's
attached
to
this.
There
is
one
PR
already
attached
to
this.
It
looks
like
there
will
probably
be
another
two
PRS
after.
A
B
So
it
looks
like
the
ask
here:
is
the
exponential
histogram.
We
should
be
able
to
create
an
instrument
that
that
exports
that
eventually
ends
up
in
the
exponential
histogram
path.
There
seems.
A
Need
to
support
Expo
and
the
data
package.
A
Expo.
Kids
and
the
exporters
and
finally
create
a.
A
Internal
aggregation,
I,
don't
know
if
that's
going
to
be
I'm
going
to
move
that
into
the
blocked
and.
B
Put
it
as
need
subtasks,
because
I
don't
know
if
that
is
sufficiently
complete.
If
we.
A
B
If,
if
we
verify
that
that
is
enough
of
the
subtasks
to
do
that,
we
can
move
that
into
to
do
so
bound
the
memory
of
com,
cumulative
sum
aggregators.
A
B
The
ask
here
is
right:
now:
the
cumulative
aggregators,
both
I,
think
sum
and
histogram-
do
not
forget
any
attributes
that
they've
ever
seen.
So
if
you
create
multiple
attributes,
they
will
they
will
grow
in
an
unbounded
number.
B
A
A
A
B
I
I
don't
know
how
we
could
actually
work
this
without
following
afoul
of
how
the
specification
might
flow.
A
So,
let's
not
put
that
in
the
beta
at
this
point,
optimized
memory,
allocation
of
cumulative
histograms.
A
Limiting
the
number
of
items.
A
B
A
B
The
history
on
this
is
currently
you.
We
filter
at
the
aggregator
level,
with
a.
B
Sorry
excuse
me:
we
filter
only
not
when
we're
ingesting
metrics,
but
when
we
are
exporting
metrics,
but
that
would
probably
have.
B
That
will
probably
have
some
changes
to
how
our
internal
apis
are
doing.
I
think
we
could
actually
start
working
on
this.
B
Investigate
if
the
aggregator
API
should
handle
attributes,
looks
like
there's
a
lot
of
logic,
bug
baked
into
the
sum
aggregator
that
can
be
treated
at
a
higher
level
than
the
SDK
inside
this
type,
whereas
a
map
could
be
managed
independently
of
the
specific
aggregation
used
when
it
comes
to
managing
synchronization
of
removing
of
sale
entries
from
this
map,
the
logic
ought
to
be
identical
across
aggregators
I
figure.
For
this
reason,.
B
So
it
looks
like
this
is
predicated
on
the
fact
that
this
would
also
be.
A
Part
of
the
the
limiting.
B
Didn't
just
isn't
working
here
limiting
the
cumulative
aggregations
because
it
talks
about
the
reducing
the
scale.
A
Maybe
we
can
Milestone.
A
A
A
I,
don't
think
I
don't
know
if
I
would
be
able
to
take
this
on
I'm
gonna
put
it
as
blocked
need,
subtasks
and
we'll
come
back
to
that
add.
A
To
do
and
done
simple
ones:
I
only
got
about
10
more
minutes.
I
will
do
probably
two
or
three
more
and
then
hop
off.
This
call
I
hope
anybody
watching
on
review
enjoys
it
and
can
follow
along
and
continue
the
song.
A
Let's
see
improve
the
error
message
of
the
metric
data
test
package.
I
honestly,
don't
think
we.
B
For
people
watching
improve
the
error
messages
of
the
metric
data
packages,
if
you
compare
a
complex
struct
and
it's
one
of
the
and
if
one
of
the
like,
if
you
compare
a
resource
metric
and
one
of
the
scope,
metrics
or
one
of
the
values
in
the
scope
metric
are
different,
it
gets
a
kind
of
inscrutable
error
message.
That's
hard
to
parse.
B
B
A
The
the
meter
provider.
B
That
their
identity,
that.
B
If
they,
if
you,
if
they,
if
you
can't
do
that,
you
can't
apply
viewer
or
whatnot
to
that
to
understand
whether
or
not
you
need
to
be
calling
this.
This
probably
could
be
pulled
from
the
beta,
but.
B
If
somebody
wants
to
try
and
take
it,
I
will
take
a
task
to
put
some
subtasks
on
there.
I
will
also
take
as
a
cleanup
task
to
try
and
make
sure
that
these
have
subtasks
and,
if
not
ping,
the
appropriate
people
to
write
the
subtasks.
A
But
it
looks
like
I
got
through
about
half
of
them
in
30
minutes.
So
that's
great.
A
An
hour
have
a
wonderful
day.