►
From YouTube: 2022-11-23 meeting
Description
cncf-opentelemetry@cncf.io's Personal Meeting Room
B
We
can
wait
a
couple
of
minutes
to
see
if
people
join
but
I
think
it'll
be
a
smaller
meeting
than
normal.
B
B
B
I
guess
we
can
get
started.
You
share
my
screen.
B
Okay,
yeah
so
just
to
start
off
I
think
it
will
be
a
fairly
short
meeting
today.
So
again,
if
you
want
to
add
topics
today
would
be
a
good
day
for
it.
B
First
item
that
I
have
on
the
agenda.
Is
the
clock
drift
PR
I
opened
a
new
PR.
That
is
a
slightly
different
approach.
B
You
can
see
a
short
description
of
what's
going
on
in
the
linked
comment
in
the
pr
one
issue
with
it,
as
pointed
out
by
t2t2,
is
that
if
you
do
not
provide
a
start
time,
but
you
do
provide
an
end
time,
you
can,
you
can
have
a
situation
where
the
start
time
is
corrected,
but
the
end
time
is
not
particularly
the
fetch
instrumentation.
B
Does
this
I
think
that's,
probably
okay,
the
fetch
instrumentation
can
be
modified
fairly
easily
to
either
provide
a
number
instead
of
a
HR
time
which
would
be
corrected
or
to
provide
both
the
start
and
the
end
using
the
performance
API,
but.
B
Just
asking
people
to
please
review
this
PR
and
take
a
look
at
the
test,
sandbox
that
t2t2
created
and
let
me
know
which,
which
method
you
prefer
I-
think
T2
T2
you've
been
doing
the
most
thorough
review
of
both
of
these
PRS.
Do
you
have
a
feeling
of
which
you
would
prefer
at
this
point.
B
D
It's
me
Abnet.
Can
you
hear
me
yes,
I
hear
you
yeah
is
this.
This
PR
has
been
approved,
it's
about
allowing
adding
customer
service
to
document
load,
but
we
were
not
able
to
merge
it
because
of
this
issues
and
I
we
I
needed
some
more
help.
I
think
Amir
he's
here
has
been
reviewing
it
and
he
may
have
more
context
on
that.
One.
So
I
have
tried
to
like
like
compile
this
on
on
my
local
branch.
C
So,
like
Auto
instrumentation
a
web,
it
merges
the
config
like
the
typing
for
the
config
for
all
the
instrumentations
yeah
and
the
one
configure
option
called
the
apply
attributes
to
spend
I
think
yeah.
It
has
different
typing
for
xhow,
I,
think
and
so.
D
C
D
What
what
I
did
was
just
it.
It
was
using
the
previous
pocket,
the
previous
versions
of
the
face
and
the
x6r
instrumentations
in
the
in
the
package
that
just
for
the
auto
instrumentations,
so
I
put
I
updated
that
one
to
use
the
latest
one
and
after
that
I
when
I
run
it
locally.
It
was
working
fine
like
it.
It
was
not
compiling
before,
but
now
it's
compiled,
but
is
still
as
an
issue
when,
when
it
comes
to
the
CI,
that's
land.
C
D
B
I'll
have
some
time
after
this
meeting
I
can
try
to
reproduce
it
locally.
On
my
machine
too,
and
see
if
I
can
help,
you
find
it.
Okay,.
C
B
C
Yeah
so
someone
added
the
I'm
not
sure
we're
still
familiar
with
the
SQL
commento
I
was
not
I,
read
about
it
in
this
BL,
so
it's
basically
a
way
to
to
attach
a
context
to
SQL
queries
as
a
comments.
So
someone
added
this
to
the
pg
instrumentation
and
so
I
reviewed
it.
It
looks
good
but
since
it
modifies
the
query,
so
I
think
we
need
to
be
extra
careful
about
this
one
so
and
if
people
want
to
take
a
look
and
make
sure
it
does
nothing
problematic
about
it,
and
please
please
take
a
look.
C
B
Oh
post
crash
I
got
it
okay,
yeah
I
I
was
vaguely
aware
of
SQL
commenter
I
think
it
was
donated
by
Google
to
the
open
teletube
project
a
while
ago,
which
I
think
is
what
this
is
probably
linking
to
yeah.
C
Yeah
but
I
think
it's
like
adding
a
comment,
and
then
you
have
to
escape
some
some
collectors
and
and
do
some
processing
of
on
the
data
and
then
I
think
I
need,
if
we're
not
very
careful,
it
might
open
the
door
for
things.
That
add
something
to
the
query.
So
we
need
to
verify
that
it's
safe,
like
we
need
a
lot
of
eyes
on
it,
yeah
so
yeah.
E
Yeah,
if
you
need
somebody
from
the
Google
side
as
well,
who
helped
work
on
this
I
can
probably
try
to
get
somebody
to
look
at
it.
B
E
The
other
thing
is
I,
think
part
of
the
donation
had
connects
and
I
want
to
say,
sqlize
orm
instrumentations
also
so
there's
like
a
there
should
be
a
reference
implementation
that
does
the
escaping
I.
Don't
know
if
anybody
from
the
JS
Sig
has
been
across
that,
but
I
think
there
was
a
repo
made
in
the
open,
Telemetry
org
that
should
have
that
code,
and
then
there
was
some
like
discussion
about
whether
or
not
it
was
going
to
be
incorporated
into
the
individual
like
instrumentation,
repos
or
stay
separate,
but.
B
Yeah
so
I
I
remember
when
this
donation
was
made
looking
at
it
briefly,
but
at
the
time
just
didn't
have
any
time
to
to
think
about
this.
To
be
honest
but
yeah,
the.
B
B
I
will
try
to
review
that.
It
would
be
great
if
we
could
get
someone
from
the
Google
side
to
take
a
look
at
it.
Also,
though,.
B
B
B
B
Okay,
my
original
thought
was
that
this
wasn't
debug,
but
actually
it
seems
like
it
might
be
if
some
values
are
coming
in
as
an
integer
and
some
are
a
float,
that
definitely
complicates
it.
For
the
back
end,
I
will
add
a
P2
to
this
and
assign
myself
I,
don't
think
our
metrics
SDK
has
a
way
to
I
I
know
we
have
the
integer
option.
I'll
look
into
this
later.
I
think
P2
is
correct.
There.
E
Yeah
I
do
so.
Basically
what
I'm
doing
is
I
create
a
counter
instrument
and
then
I
change,
the
aggregation
to
histogram
and
I
change,
the
name
and
then
I
wanted
to
see
what
happened
with
the
descriptor,
the
instrument
descriptor
and
the
output,
and
it's
sort
of
like
halfway
matches
the
view
halfway
matches
the
instruments
and
you
see
like
if
you
go
down
to
actual
result,
the
name
is
changed
to
match
the
view,
but
the
type
still
like
describes
the
original
instrument.
E
So
I'm
not
I,
wasn't
sure
what
the
actual
intended
Behavior
was
because,
like
my
thought
was
that
this
is
for
somebody
writing
an
exporter.
They
can
say
you
know,
like
hey.
This
view
was
caused
by
this
original
instrument.
It
has
an
error
or
something
like
that
or
the
idea
was
to
have
like
a
descriptor
so
that
you
don't
have
to
look
at
the
actual
data
points
to
understand.
What's
going
to
be
in
there.
So.
B
Yeah,
not
sure
data
data,
point
type
and
value
type
is
probably
it's
probably
value
type,
I'm,
not
sure
what
the
off
the
top
of
my
head
with
the.
B
Enum
values
are
for
those
particular
values,
but
probably
what's
happening
is
type
counter.
It
is
still
it
is
still
a
counter
instrument
because
it
still
has,
like
the
add
method
on
it,
for,
for
example,
it's
the
underlying
data
type
that
changes
in
the
aggregation,
so
maybe
take
a
look
at
what
the
enum
values
are
for
these
two
value.
E
Type
one
is
in
64,
and
then
what
was
the
other
one
you're
asking
about
if.
E
I'm
trying
to
pull
it
up.
Sorry,
so
we
have
value
type,
sorry
value
type
in
is
zero
and
double
is
one
and
again.
E
Which
that's
on
the
top
level.
E
A
Yeah
I
think
I
I
can't
remember
anymore,
but
I
think
I
applied
some
refactoring
that
made
this.
That
made
it
possible
to
just
see
the
data
point
type
and
and
Export
stuff
before
that
it
would
determine
what
determined
the
data
type
for
the
transformation
to
otrb
in
the
otrp
Transformer
based
on
the
instrument
type,
but
that
wasn't
reliable
with
views
anymore.
So
that
might
actually
be
where
that
comes
from.
A
B
The
type
here
still
says
counter
I
guess:
I'm
gonna
have
to
look
at
the
code
to
see
if
that
is
intentional,
or
if
that
was
just
missed,
when
the
value
type
gets
updated
should
type
also
be
updated.
I
think
I.
A
A
I
think
back
when
I
worked
on
that.
That
is
also
what
the
intention
was.
The
instrument
type
stays
the
same,
but
the
aggregation
changes
so.
D
E
E
C
E
B
E
E
Not
that
I
think
it's
not
useful,
like
I,
think
it's
a
potentially
a
good
feature.
If
you
find
like
an
issue
that
you
can't
support
some
instrument
in
your
specific
exporter,
but
yeah
I
think,
like
other
like,
if
you
were
looking
at
just
the
otlp
protobuf
message,
then
you
would
have
to
look
at
like
that.
First,
entry
in
the
data
points
to
even
figure
out
the
instrument
name,
for
example,
for
the.
E
I
think
that
one
has
the
the
histogram
I
can
I
can
check.
A
Yeah
that
that
should
be
the
the
actual
histogram
data
point
value
in
there.
So.
A
There
but
I
think
with
how
we
have
it
right
now.
It
would
even
be
safe
to
change
the
type
to
match
the
one
with
the
view,
because
that
is
the
only
way
I
think
we
can
move
forward
with
this.
A
E
B
B
E
Yeah
I
think
it's
pretty
clear
from
the
because
the
type
the
interface's
name
and
typescript
is
instrument
descriptor,
even
if
the
key
is
not
so
it's
that's
also
what
I
figured
yeah.
B
E
No,
it
was
actually
more
like
I
I
just
wanted
the
that
this
is
for
the
Google
Cloud
monitoring,
exporter
and
I.
Just
wanted
the
actual
name
of
the
metric
and
I
was
like
okay,
it's
it's
in
the
descriptor,
but
then
the
other
fields
in
the
descriptor
don't
match
the
actual
metric.
So
that
was
just
not
sure.
100,
okay,
yeah.
B
I'd
be
interested
to
see
what
other
sigs
did
here.
I
think
our
our
implementation
is
very
influenced
by
Java
in
particular,
so
they
may
have
this
problem
also
or
if
they
don't
have
this
problem,
then
maybe
we
should
mimic
what
they're
doing,
but
yeah
I
would
say
that
I
would
have
this,
be
the
original
instrument
name
and
then
use
the
name
in
this
value
object
if
it
exists.
E
Yep
yeah,
that
makes
sense
and
just
to
save
you
some
typing
I
put
those
were
the
two
options
I
put
in
the
expected
results.
E
A
You
can
assign
this
to
me.
I
will
look
into
that.
B
And
I
think
that
should
be
it.
It
doesn't
look
like
anyone
added
any
more
agenda
items.
So
if
that's
it
then
oh,
let's
go
through
control.
I'm
sorry
looks
like
two
new
ones
here:
graphql
release
failed,
oh
yeah,
actually,
I'm
Gonna,
Leave,
This.
B
I'm
going
to
leave
this
open
while
I
work
on
I'm
working
on
updating
my
contribci
right
now,
it
hasn't
hasn't
successfully
exported
anything
or
released
anything
in
a
while.
It's
obviously
just
not
working
so
I'm
gonna
try
to
remove
our
dependents
on
release,
please
if
I
can,
or
at
least
the
for
the
released
step.
I
use
it
used
to
generate
the
change
logs
and
versions
yeah.
A
C
C
So
what
he
did
he
created
the
map
so
for
which
redis
command.
There
was
a
number
that
says
how
many
values
to
to
capture
into
DB
statement
so
for
most
of
them
like
for
some
commands,
it's
a
zero
Arguments
for
some,
it's
one
or
two.
You
can
look
at
the.
C
Ability
is
so
this
is
the
list
by
regex,
so
we
just
check
which
one
walks
and
then
he
captures
those
arguments.
C
B
Okay
thanks
everybody
for
your
time,
I'll
talk
to
you
next
week.