►
From YouTube: 2022-09-28 meeting
Description
cncf-opentelemetry@cncf.io's Personal Meeting Room
A
I
can
probably
just
jump
in
and
the
the
CLA
issue
for
the
sandbox
that's
effective
result.
So
I
think
the
person
went
and
actually
signed
the
CLA
again,
which
made
us
pass.
So
I
was
already
pushed
in
that
that
first
PR
and
I
have
another
PR
to
effective,
merge
the
next
set
of
updates.
So.
B
A
I
I
just
went
back
to
it
on
Monday
and
it
was
signed.
I
went,
excellent
and
I
think
Mark
and
yourself
had
signed
it
off
so
I
pushed
it
then
and
then
ran
the
script
again
and
it
generated
the
next
PR
which,
instead
of
being
2400
files,
was
a
little
bit
smaller.
B
Perfect
I
will
I
was
actually
just
about
to
let
you
know
that
I
reached
out
on
slack,
because
the
ticketing
system
is
slow
and
apparently
not
checked
very
often,
but
I
will.
Let
them
know
that
it's
resolved.
A
Yeah,
so
that's
good,
so
in
terms
of
the
sandbox
I
will
probably
get
back
to
it
next
week
to
actually
start
converting
that
sandbox
into
something
that's
compilable
domain
and
then
once
that's
done,
we
can
then
start
creating
branches
and
make
progress.
B
First
topic:
I
have
is
the
metrics
GA
as
per
usual,
essentially
documentations.
All
that's
left.
I
spent
a
lot
of
time
on
Monday
working
on
a
website,
doc
Spike,
which
I
hope
to
finish
this
afternoon
and
send
that
over
to
Philip,
so
that
he
can
Wordsmith
it
a
little
bit
and
get
it
on
the
website.
That's
the
only
GA
issue
we
have
left
so
I
did
schedule
the
review
from
the
technical
committee.
B
They
said
that
they
will
talk
about
it
in
their
meeting,
which
I
believe
is
actually
today.
It
might
actually
be
right
now,
so
that
should
be
coming
up
fairly
soon
and
we
should
be
able
to
release
the
metrics
SDK
as
a
1.0
in
the
next
week
or
so.
Certainly,
the
next
couple
of
weeks
and
I
I
viewed
kubecon
as
an
important
deadline,
so
I'd
like
to
to
have
it
out
and
released
and
and
in
use
by
then
so
that
we
can
make
an
announcement.
B
Okay.
We
have
the
same
topics
for
after
GA
that
we
had
last
week.
The
only
updates
here
are
on
the
Prometheus
exporter.
Support
for
resource
attributes.
Mark
actually
started
working
on
this.
So
that's
in
progress
now
and
I
started
a
prototype
for
the
high
resolution.
Histogram
I,
don't
know
if
it
will
be
ready
for
the
ga
release.
B
A
B
B
All
right
next
topic
is
the
ongoing
effort
to
merge
the
API
into
the
main
repo.
It
is
now
merged
back
into
the
learn,
a
mono
repository.
So
that's
all
good.
The
building
is
working.
B
B
So
we
wouldn't
want
to
cut
a
release
that
hasn't
been
properly
tested
with
the
old
versions
of
node,
but
that's
kind
of
The
Next
Step
here
I
will
also
begin
the
process
of
migrating
all
of
the
issues
from
the
API
repo,
which
should
be
pretty
easy
and
letting
all
of
the
people
who
have
open
PR's
know
that
they
need
to
reopen
them
on
the
main
repository,
which
probably
will
make
them
all
very
annoyed,
but
it
is
what
it
is
I
guess.
B
B
B
The
root
cause
of
this
is
that
spans
are
started
using
the
start
span
API,
but
without
providing
a
time
stamp,
and
then
there
ended
using
a
manual
time
stamp.
This
was
not
a
problem
before,
because
everything
was
using
the
same
monotonic
clock,
but
now
that
we
have
the
anchored
clock
implementation
to
defend
against
clock
drift,
you
can
end
up
in
a
situation
where
you're
using
a
different
clock
for
ending
the
span
than
you
are
for
starting
the
span,
which
can
result
in
inconsistent
timings.
B
It
happens
to
affect
Express,
because
the
spans
in
question
are
very,
very
short,
like
a
few
nanoseconds,
so
the
clock
drift
is
a
lot
more
likely
to
actually
affect
those.
B
B
This
will
cause
a
slight
change
in
Behavior
because
the
middlewares
used
to
have
zero
durations
bands
and
now
they'll
just
be
very
short,
so
they'll
be
a
few
milliseconds
or
whatever,
but
I
wouldn't
expect
that
to
be
a
problem
for
anyone.
So
it's
already
had
a
couple
of
reviews.
B
I
was
planning
on
merging
it,
but
I
figured
I
would
wait
until
after
the
meeting
to
see
if
anybody
had
a
problem
with
the
solution
that
we
came
up
with,
we
will
also
have
to
apply
a
similar
fix
to
the
xhr
plugin,
which
is
doing
the
same
thing
here.
Just
getting
a
manual
end
time,
setting
a
timeout
and
then
ending
the
span
with
manual
times.
B
This
one
is
a
little
bit
more
complex.
I
think
that
the
fix
is
not
as
easy
as
just
not
using
those
times,
but
I
am
looking
into
that
and
I
created
a
discussion
issue
for
the
general
problem
and
how
we
can
solve
this
in
a
more
General
way.
Moving
into
the
future.
There's
quite
a
bit
of
discussion
on
here,
but
to
summarize
the
takeaway
I
guess
we're
considering
the
idea
of
not
using
the
monotonic
clock
at
all
and
only
using
the
the
the
system
date.
B
It
obviously
has
a
couple
of
disadvantages
if
you
have
like
an
ntp
synchronization
during
a
span,
you
may
end
up
with
some
weird
timings
and
some
things
like
that,
but
it
may
be
the
easiest
way
to
solve
this
problem.
The
system
timer
is
not
any
slower
than
than
the
monotonic
clock.
It
is
not
as
high
resolution
but
in
recent
versions
of
node
and
recent
browsers.
B
That
is
not
as
much
of
an
advantage
anyways
because
they
specifically
do
not
allow
super
accurate
times
in
order
to
prevent
again
things
like
fingerprinting
and
user
tracking
and
stuff
like
that,
so
we
may
be
better
off
just
moving
to
the
date
interface
anyways,
but
that's
kind
of
an
ongoing
discussion.
Nothing
has
really
decided
yet
so
I
would
encourage
anybody
with
a
stake
in
this,
which
should
be
most
of
you
to
at
least
read
this
issue
comment.
B
If
you
have
any
thoughts
or
concerns
about
the
proposed
Solutions
and
let
us
know
what
you
think
I
know
that
was
kind
of
a
monologue.
But
does
anyone
have
any
immediate
comments
there
or-
and
it's
not
a
super
straightforward
topic,
but
it's
not
too
I.
Think
too
complex
either.
B
Okay,
I
guess
we
can
continue
that
discussion
asynchronously
when
people
have
time
to
read
the
issue.
B
All
right,
the
next
topic
that
I
have
and
the
last
one
that
I
have
on
the
list
is
the
require
on
the
middle
Singleton
I
linked
it
here.
It
there's
no
real
change
from
last
time,
except
that
it
is
just
in
need
of
more
reviews
before
we
merge
it.
It's
obviously
a
very
important
part
of
all
of
our
instrumentations,
so
I
want
to
make
sure
that
it
gets
enough
eyeballs
on
it
before
we
before.
B
We
merge
it
just
to
make
sure
that
that
nothing
is
missed,
because
it's
not
something
that
I
am
excited
to
have
to
have
a
bunch
of
bug
fixes
all
the
time
I'd
rather
get
it
right
on
the
first
try.
If
we
can.
A
B
Then
let's
spend
a
few
minutes
on
bug
triage
here.
It
looks
like
we
have
two
new
triage
issues.
B
B
B
It
must
be
not
a
primitive
object,
it
must
be
trying
to
set
a.
It
must
be
trying
to
set
the
key
as
an
object
or
something
like
that.
I'm
going
to
switch
this
to
P2.
A
B
B
B
It's
not
supported
and
not
supported,
or
for
the
browser
open
to.
Let
me
instrumentation
of
open
telemetry.js
as
a
browser.
Implementation
of
instrumentation,
based
with
no
enable
method,
tried
the
node
implementation
for
the
browser,
but
it
doesn't
work
yeah.
So
I'm
not
sure
if
this
is
actually
a
bug,
I
guess
I
will
assign
myself
this
real
quick
so
that
I
can
look
into
it
a
little
bit
closer
but
I'm
going
to
assign.
B
Okay,
we
already
talked
about
this
last
week.
I
have
not
had
time
to
look
into
this,
yet
we
already
marked
this
one
as
up
for
grabs,
although
it
looks
like
Rano
assigned
himself.
B
A
B
That
should
be
it
very
short
meeting
today.
Anyone
have
anything
that
they
want
to
bring
up.
A
B
Three
hours
in
two
days,
wow
very
close,
I
haven't
no.
It
was
so.
A
B
Are
the
are
the
implementations
very
different.
A
A
I
just
saw
that
they
both
solved
the
same
thing
and
then
I'll
wait
for
my
instructions.
A
A
B
A
A
B
Right
love,
a
good
short
meeting,
I
guess
thanks
everyone
for
your
time
and
I
will
talk
to
you
next
week.