►
From YouTube: 2022-07-20 meeting
Description
cncf-opentelemetry@cncf.io's Personal Meeting Room
B
B
All
right,
I'm
gonna,
move
this
down
below
this
stuff.
Just
a
real
quick
note.
I
am
out
for
this
meeting
next
week.
So
is
amir
and
rono,
so
this
meeting
will
be
cancelled
next
week.
So
just
so
everybody
knows.
B
First
item
is
the
metrics
ga
buyer?
Are
you
in
attendance?
Let's
see,
I
don't
see
him
on
the
call,
so
I'm
taking
over
the
in-memory
metric
exporter
pr,
just
because
it's
been
a
little
bit
of
time
since
it's
been
updated
and
you
know
it's
a
small
thing
that
should
just
be
merged.
C
B
B
I
believe
mark
also
has
a
pr
for
30
90
in
coming
fairly
soon.
Hopefully,.
A
B
Okay,
there
are
a
couple
of
other
upper
grabs
issues
here,
so
if
anybody
is
looking
for
something
to
do
and
wants
to
help
with
ga,
this
is
a
good
place
to
look
for
places
where
we
need
help.
B
I'm
not
going
to
go
through
the
individual
prs,
like
I
have
been
doing
just
they're
in
there
they're
easy
to
find
for
those
that
don't
know
the
otlp
json
short
keys
pr
in
the
proto
repository
was
closed.
B
It
wasn't
worth
the
headache
that
the
short
keys
proposal
introduced
in
terms
of
processing
times
and
having
to
ship
a
map
of
keys
and
things
like
that.
So
it's
actually
good
timing.
For
us,
the
proto
http
exporter
was
updated
in
the
last
week,
has
not
been
released,
but
it
actually
uses
compiled.js
instead
of
loading
protofiles
now,
which
means
it
should
work
in
the
browser
it
uses
the
send
with
http
from
the
base
module,
which
does
support
sending
in
browser
and
since
the
proto
files
will
no
longer
be
required.
B
It
should
work
in
the
browser
there
aren't
tests
for
that
yet,
but
just
adding
tests
and
making
sure
should
not
require
any
additional
work
to
make
that
work
in
the
browser.
So
that's
a
a
decent
win.
B
A
It
cool
exactly
that's
something
that
we
can
definitely
help
out
with
and
how
like
testing
it,
and
then
I
guess,
reporting
back
or
creating
issues
or
adding
any
tests
is
useful.
B
Yeah,
that
would
be
great.
It
should
be
fairly
easy
to
just
enable
the
browser
tests
and
then,
in
terms
of
testing
an
unreleased
exporter,
it
can
actually
be
a
little
bit
annoying
yeah.
I
I
think
that
just
doing
like
npm
run
pack
and
then
installing
it
from
the
guitar
ball
should
work
just
fine.
B
If
that's
not
working
for
you
feel
free
to
reach
out
to
me
in
slack,
and
I
can
release
it
as
a
like.
I
can
do
an
alpha
release
to
npm,
if
that
makes
it
easier.
B
All
right
anything
else
before
I
move
on
there.
B
Okay,
I
know.
Last
week
we
talked
about
the
contrib
rate,
limiting
I
was
supposed
to
look
into
what
the
options
are
from
a
policy
standpoint,
and
I
have
not
had
time
to
do
that.
So
I
have
no
update
here
jamie
any
update
on
your
site
or
same.
A
B
Yeah,
of
course,
so
just
waiting
on
me,
okay,
santosh,
it
looks
like
you're
adding
a
spec
pr
here.
This
is
the
follow-up
from
the
otep.
Then.
B
Okay,
is
there
anything
specific
about
this
pr?
That
needs
to
be
reviewed
from
a
js
perspective,
or
are
you
just
looking
for
general
feedback
yeah.
C
There
is
yeah,
there
is
one
question
I
have
I'm
trying
to
learn
how
the
zone
concept
works
in
the
browser.
C
So
one
specific
question
that
I
wanted
to
check
with
folks
here
is
that
when,
let's
say
as
an
example,
let's
say
a
span
is
in
progress,
but
then
an
independent
event
handler
gets
in
a
callback
and
then
you
try
to
create
logs
or
events
from
that
you
know.
Do
we
you
know,
want
to
associate
the
fan
context.
B
C
B
I
mean
I,
I
think
that
it
would
be
nice
to
have
those
associated.
Certainly
the
logs
could
have
the
expand
context
on
it.
I
have
not
looked
at
this
pr
to
see
if
the
same
thing
would
apply
to
events,
but
I
I
would
think
so.
If
that's,
if
that's
I
don't
see
any
reason,
we
could
not
implement
that
in
javascript.
C
No,
actually,
the
events
could
be
totally
independent
right,
not
related
to
the
you
know,
span
in
progress.
So
let's
say
the
you
know
while
and
jack's
call
is
in
progress.
Somebody
clicks
the
button
right
and
that
results
in
so
it
could
be
confusing
to
the
you
know:
consumers
of
the
telemetry
when
they
look
at
that.
You
know
that
exception.
C
Correct
so
when
we
create
the
the
logs,
in
this
case
the
the
log
emitter,
the
logger
object,
has
an
option
to
associate
the
place
context
or
not.
So
there
is
an
open
item
in
the
in
this
pr.
C
Where
I'm
debating
that
that
option
be,
you
know
by
default
false,
and
you
know,
and
the
reason
is
you
know,
I
I
personally
prefer
that
the
instrumentation
developer
is
explicitly.
You
know
aware
that
they
need
the
trace
context.
You
know
for
that
situation,
so
I'm
I'm
curious.
You
know
what
the
opinion
is
on
on
this
flag.
Yeah,
I
mean.
B
I
think
that
I
would
default
it
to
true
if
there's
a
just,
because
there's
a
span
in
progress
does
not
mean
that
it
would
be
an
active
span.
So
I
think,
if
there's
an
active
span
in
the
in
the
execution
context,
then
yeah.
I
think
that
I
would
like
to
have
the
events
and
logs
correlated,
but
if
the
user
clicks
a
button,
that's
not
like
in
the
context
of
a
spam
just
because
there
are
spans
behind
the
scenes,
doesn't
mean
that
they're
active
in
that
particular
execution
context.
C
B
C
B
C
Okay,
yeah,
if
it's
I'll
make
the
change.
B
A
Probably
just
my
normal
waffle
about
the
sandbox,
so
I
I
expect
to
have
a
pr
out
for
the
auto
emerging.
So
this
is.
This
is
really
the
the
script
that
will
take
the
js
and
the
js
api
repos
and
move
them
in
their
entirety
into
a
subfolder
into
what
I'm
calling
a
staging
folder.
This
is
before.
I
then
move
it
into
the
final
package,
one
which
is
the
initial
pr
that
I
sent
out
in
terms
of
its
structure
for
compiling.
A
So
this
is
like
a
just
to
get
the
history
in
in
there.
So
the
only
thing
which
we'll
have
to
talk
about
daniel
is
how
do
we
get
the
bot
token
into
the
sandbox,
because
I
I
have
constructed
the
scripts
using
a
I
again
end
up
creating
a
microsoft,
open,
telemetry
bot
and
I've
been
testing
with
that
against
my
local
fork.
So
it
should
all
work,
but
we
need
to
get
it
in
there.
A
So
I
I'm
just
finalizing
some
of
the
messages
that
the
gets
included
in
the
pr
text,
so
hopefully
I'll
have
that
out
today
before
you
go
away.
B
Okay,
I'm
not
gone
for
like
the
entire
week
by
the
way,
I'm
just
missing
that
day.
So
it's
not
like
I'm
going
on
a
long
vacation
or
anything
I'll
be
available.
A
Yeah
so
I'll
drop
a
couple
of
links
into
the
section
you
just
created,
so
you
can
sort
of
see
what
my
fork
is
doing
to
get
an
idea.
So
I
get
a
little
bit
of
a
heads
up,
but
hopefully
I'll
be
sending
that
pr.
Today
I
tried
to
do
it
yesterday,
but
when
I
did
some
final
last
minute
testing
after
moving
it
away
from
my
butt
to
the
open,
telemetry
bot,
it
failed.
So
I'm
just
finalizing
that.
B
Okay,
I
guess,
as
a
sort
of
related
note,
none
of
the
other
maintainers
are
here,
so
we
really
can't
make
any
decisions,
but
I
I
I
would
like
to
consider
moving
the
api
back
into
the
mono
repo
when
we
originally
moved
it
out.
We
did
it
so
because
the
api
was
meant
to
be
more
stable
than
the
core
repo
and
we
wanted
to
be
very
deliberate
about
making
changes
and
that
sort
of
thing.
Since
then,
the
core
repo
has
gotten
significantly
more
stable.
B
We
have
a
lot
better
processes
around
it
and
it's
making
it
difficult.
I'm
currently
trying
to
work
on
the
attach
and
detach
api
context,
api
methods
and
it's
difficult
to
prototype
them
in
the
actual
implementations.
A
Why
emerging
exactly
yeah
so
that
we
can
spin
faster
as
well
yep.
B
So,
with
none
of
the
maintainers
here
we
can't
make
a
decision
on
that
today,
but
just
so
you're
aware
that's
likely
coming
unless
somebody
has
a
very
good
reason
to
say
no.
A
Yeah
my
script
should
handle
that
we
would
just
change
the
script
to
say,
don't
bother
dragging
in
the
api
anymore
and
it'll
just
get
merged
into
the
auto
merge,
so
that
that
should
be
working.
D
A
D
Go
ahead
yeah,
so
I
posted
it
in
the
chat
I
put
together.
Our
I've
been
building
a
lot
of
applications
at
my
company
u-ship
using
cloudflare
workers
essentially
is
a
very,
very
small,
javascript
runtime
that
runs
at
the
quote-unquote
edge
and
also
has
some
unique
constraints,
which
meant
we
couldn't
use
the
opentelemetry.js
sdk
by
default,
some
notable
things
being
that
there
is
no
concept
of
global
states,
which
is
a
big
assumption
that
the
open,
telemetry
sdk,
makes
in
a
lot
of
places
as
well
as
they
don't
have
an
equivalent
of
a
performance
api.
D
So
we
had
to
patch
that,
but
after
getting
over,
those
pumps
ended
up
implementing
this
sk,
which
by
and
large,
is
fully
functional
and
implements
ot
lp
http
support
for
both
json
and
protobuf
and
actually
implements
the
log
spec
a
very
hand,
rolled
implementation.
Since
I
couldn't
lean
on
the
js
packages
for
that,
but
it
does
exist,
and
so
I'm
also
happy
to
help
upstream
that.
B
Yeah
that
sounds
fantastic,
particularly
the
logs
work.
I
think
that
if
we
could
get
a
jump
start
on
the
logs
implementation,
that
would
be
awesome,
but
also
you
know,
removing
any
sort
of
platform
dependencies
that
we
have
is
always
a
good
thing,
making
it
easier
to
run
in
different
environments.
B
This
reminds
me
of
well
man,
I
don't
remember
his
name
pa.
I
think
it
was
presented,
something
that
ikea
was
working
on
in
a
similar
vein
and
they're
working
on
upstreaming
it
as
well,
but
I
think
he's
working
through
internal
challenges
in
terms
of
upstreaming
upstreaming,
that
to
open
source,
I'm
not
sure
how
to
get
a
hold
of
him.
He
may
be
he's
definitely
on
the
cncf
slack,
so
it
may
be
worth
reaching
out
to
him.
D
Yeah,
absolutely
I
think
there
there
are
assumptions
that
both
the
browser
and
the
node
packages
make,
which
simply
don't
work
in
workers
and
node
are
excuse
me
demo,
but
there's
having
worked
in
it.
I
think
there
are
some
accommodations
that
probably
could
be
made
since
most
of
the
api
is
pretty
good
about
allowing
you
to
pass
in
context.
So
it
doesn't
depend
on
that
global
state,
and
I
think
the
biggest
thing
is
since
the
since
workers,
both
in
deno
and
the
cloudflare
terminology,
are
very
like
the
browser.
B
A
D
A
D
B
Yeah
I
mean
whatever
you
can
upstream
is
obviously
very
welcome.
D
Yeah,
I
think
there
was
also
two
relevant
other
things
that
came
up
which,
on
the
note
of
the
protobuf
encoding,
we
we
are
actually
doing
exactly
what
you
mentioned
above
and
that
we
are
using
the
compiled
protobufjs
for
protobuf
support
and,
aside
from
the
bundle
size,
it's
been
working
pretty
well
so
far
we
ended
up
or
ended
up
pushing
it
into
its
own
separate
import.
So
it's
an
opt-in
behavior
because
of
the
bundle
size
increase,
but
so
far
it's
been
performing.
D
B
Let's
see
we
do
have
gzip
as
an
option
and
I
believe
that's
using
like
the
gzip
package
to
do
that,
which
is
unfortunately
a
fairly
large
dependency.
D
I
haven't
done
the
precise
measurements
on
the
performance
overhead
of
that,
but
it
would
probably
be
it
would
be
an
interesting
either
automatic
or
opt-in,
behavior
to
potentially
use
the
compression
stream
and
the
browser
if
it's
available.
So
that
would
be
an
option
to
reduce
those
sizes
without
the
massive
bundle
or
performance
overhead.
That
external
provides.
B
Yeah,
I'm
not
sure
where
exactly
to
look
at
it,
look
for
it
on
short
notice,
but
I'm
pretty
sure
it's
using
the
external
grpc
module
just
to
be
available
in
more
environments
but
yeah.
Obviously,
if
we
can
use
built-in
browser
products,
then
that
is
better.
B
I
don't
know
if
there's
a
way
to
like,
I
don't
want
to
ship
two
different
exporters,
one
for
chrome
and
one
for
other
browsers.
I
wonder
if
there's
a
way
to
only
yeah,
I
don't
know.
I
think
there
could.
A
But
that's
still
the
issue
yeah!
That's
also
one
of
the
issues
that
the
sandbox
is
going
to
play
with
so
yeah.
Looking
at
your
sdk
workers
that
looks
like
good
stuff.
A
B
Okay,
thank
you
for
that.
Anyone
else
have
questions
about
that
before
we
move
on.
A
I
I
guess
I
I
see
two
additional
members
from
my
team
that
have
joined
for
the
for
the
first
time
they
plan
on
coming
along,
quite
often
so
hector
and
jackson
jackson's
new
to
to
microsoft,
but
he's
joined
our
node.js
team
and
hector
is
our
lead
on
our
node.js
stuff.
So
welcome
guys.
B
B
A
More,
we
own
the
node.js
side
of
things
in
azure,
so
we
actually
right
now
next
month
we're
going
to
publish
this
beta
package.
That
is
basically
application.
Insights,
but
internal
is
using
everything
open,
telemetry,
so
we're
using
a
bunch
of
experimental
packages,
a
bunch
of
the
packages
that
you
guys
generate.
A
B
All
right
well
welcome
to
the
group.
Thank
you.
I
guess
let's
take
15
minutes
and
do
some
bug
triage.
B
B
B
B
B
This
is
what
the
grpc
plugin
is
there.
Anybody
here
using
the
grpc
plugin.
B
B
A
There
it
sounded
like
they
were
using
a
an
embedded
mode.
If
they
didn't
have
this
ua
compatible
tag,
it's
actually
running
an
i8
which
definitely
won't
work
for
open
telemetry
runs
on
ie7
by
default,
but
the
fact
that
I
didn't
respond-
I
don't
know.
B
A
Okay,
so
when
you
create
a
net
app
and
you
host
put
a
drop
in
a
browser
and
you're,
so
therefore
you're
expecting
to
be
running
a
hosted
browser
by
default.
Ie
will
be
in
ie7
mode
unless
you
include
the
meta
tag,
ua
compatible
to
force
it
up
to
ie11
mode.
B
I
got
it
okay,
yeah,
so
hotel
definitely
does
not
support
ie8
or
ie7.
So.
B
A
B
Yeah,
so
I
think
this
has
actually
been
solved
if
I.
A
B
A
B
B
This
is
very
similar
to
yeah
this
issue,
which
was
recently
opened,
and
I
think
that
they
came
up
with
the
same
work
around
yeah.
So
if
you
put
an
empty
headers
map,
it
forces
it
to
use
xhr
instead
of
send
beacon.
B
You
would
prefer
to
not
use
the
zipkin
receiver
in
jaeger.
B
B
I
don't
think
so
all
right,
and
that
should
mean
that
they
are
all
assigned,
or
at
least
have
pr's,
I'm
going
to
assign
who
made
this
pr.
B
Oh
closed
pr,
that's
no
good
instrumentation
basically
falls
partially.
We
believe
this
has
been
fixed
labels,
information
requested,
and
since
I
asked
the
question
also
myself
there,
it's
kind
of
annoying
that
it
shows
this
as
a
linked
pr,
even
though
it's
closed
all
right.