►
From YouTube: 2022-10-27 meeting
Description
Open cncf-opentelemetry-meeting-3@cncf.io's Personal Meeting Room
B
Yeah
yeah,
the
little
kid
is
doing
great,
maybe
a
week
or
two
ago
we
got
him
sleeping
through
the
night,
but
this
week
he's
been
doing
this
new
thing
where
he
like
wakes
up
at
five
in
the
morning.
A
C
B
It's
been
a
little,
it's
been
a
little
yeah
I've
been
awake
for
a
while.
I
guess
is
what
I'm
trying
to
say.
Yeah.
C
D
Hi
good
morning,
Bryson,
nacho,
santosher
and
I
have
some
questions
on
the
events
based
on
logs.
So
maybe
you
know
I
can
talk
whenever
you
want
me
to
sure
yeah.
D
Okay,
I
I
think
I'll
keep
it
short.
So,
as
you
know,
the
the
hotel
spec
now
supports
events,
Standalone
events,
and
these
are
not
span
events
or
metric
events.
These
are
Standalone
events,
which
means
any
events
that
occur
when
a
span
is
not
in
progress.
I
think
you
can
still
represent
them
as
Standalone
events.
They
are
essentially
log
records
with
some
specific
attributes.
Okay,
to
determine
that
there
are
events.
A
D
So
in
in
the
browser
instrumentation,
what
we
are
doing
is
we
have
identified
the
different
events
that
each
of
us
different
vendors,
are
already
in
their
existing
products.
You
know
we
have
and
we
are
coming
together
and
consolidating
preparing
a
Consolidated
schema
for
these
events,
and
we
are
saying
that
for
some
of
these
events
they
are
they
come
as
fans,
some
of
them.
They
come
as
Standalone
events,
maybe
some
Representatives
attributes
so
we're
doing
that
exercise
so
I'm
wondering
if
we
should
do
something
similar
here.
D
So
my
question
was
in
like
before
open
Telemetry.
You
know
if
you
had
products
here,
for
let's
say
you
know:
you're
I,
you
know
instrumenting
iOS
apps,
you
know
you
probably
you
know,
didn't
use
spans
right.
They,
they
were
all
messages,
or
maybe
events
you
know
whatever
you
used
to
call
them
are.
Are
those
all
covered
in
open
Telemetry
already
today?
Are
there
are
still
a
few
additional
ones
that
that
you
know,
you
think,
are
better
represented
as
events.
B
I
think
that
there
are
quite
a
few
things
that
would
be
useful
as
events
that
we
just
don't
track
at
this
point.
Just
due
to
the
you
know,
I
guess
the
maturity
of
the
Swift
SDK
right,
and
so
we
just
I
mean
we.
Don't
we
don't
have
a
logging,
we
don't
have
a
logging
SDK,
yet
even
and
our
metrics
is
quite
out
of
date,
so
there's
just
yeah
there's
a
there
there's
a
lot
of
things
that
you
know
I
think
we
would
like
to
get
in
there.
D
Correct,
okay,
okay,
so,
unfortunately
the
Android
there's
no
Android
forum
here
in
open
Telemetry
there
is
plunk,
has
an
Android
SDK
for
open
telemetry
and
they
have
asked
others
know
whoever
is
interested
to
like
they
are
open
to
contributing
that
to
open
Telemetry,
but
somebody
has
to
still
work
on
it.
So
there
are
no
open
discussions
going
on
in
that
space.
But
I
believe
that
both
Android
and
iOS
there
might
be
some
events
that
could
be
common
I
could
be
wrong.
I
I
haven't
worked
on.
D
You
know
on
these
things
before,
but
you
know:
do
you
think
it
makes
sense
to
list
down
the
events
of
interest
from
different
vendors
and
then
consolidate?
Come
come
up
with
a
common
schema
so
that
we
can
then
we
need
to
add
them
to
the
spec
semantic
conventions
for
the
for
the
attributes.
In
those
events.
B
D
B
Yeah,
so
I,
don't
I,
don't
really
have
a
lot
of
I
I.
Don't
have
like
a
very
strong
opinion
about
what
events
would
be
included
so
far
in
the
spec
I'm
sure
that
there
are
a
lot
of
events
that
that
are
defined
for
browser
that
would
apply
well,
you
know,
like
I,
don't
necessarily
think
that
we
need
to
do
too
much.
You
know
hard
hard
thought
about
it
at
this
point.
B
A
Yeah
yeah
I
also
don't
work
daily
with
with
a
Ram
Solutions,
so
yeah
one
one
thing
that
comes
to
my
mind
is
application.
Launch
time.
That's
usually
a
a
really
used
value
because
loading,
the
app
is
always
a
critical
thing.
Yeah
yeah
yeah.
A
D
A
Also
things
like
memory
warnings
is
something
that
could
also
work
very
well
for
for
events
correct
on
mobile,
so.
A
You
that
you
are
getting
to
you
are
getting
too
much
memory
and
it
can
and
healing
you
if
you
take
too
much.
So
it
sends
a
warning
that,
but
you
might
not
send
that
on
time
before
the
the
system.
A
Could
be
very
similar
to
the
web.
Okay,
next
step.
D
A
D
At
an
abstract
level,
I
think
it
is
a
general
class
of
errors
and
and
this
memory
warning
could
be
one
type
of
an
error.
I
don't
know
if
that's
the
right
way
to
model
it,
but
there
could
be
errors
and
you
know
even
crashes
so
for
crashes
when
the
app
launches
next
time.
D
A
Yeah
crisis
in
in
iOS
are
are
string
based,
so
okay,
it
could
use
just
a
string,
biggest
string,
yeah.
D
But
you
don't
want
to
send
megabytes
of
data
in
an
event
so
I
was
I
was
thinking
that
you
upload
the
crash
dumps
to
an
S3
or
some
other
place.
You
know,
that's
URL
is
configurable,
and
then
you
put
that
as
a
metadata
in
an
event
object,
and
that
goes
to
the
otlp.
A
Yeah,
but
the
problem
with
with
crosses
in
at
least
in
iOS
is
that
crosses
are
not
symbolicated
and
things
like
that
so
and
they
are
really
string
based
what
the
system
gives
you.
Okay,
the
system
gives
you
a
load
with.
The
credit
is
just
strings.
Yeah.
A
Small,
it's
kilobytes
I
see
just
kilobytes.
You
can
get,
maybe
one
megabyte
if
you
have
hundreds
of
libraries
and
hundreds
of
threads
running
and
something
crazy,
but
not
above
that.
Yeah,
okay,
okay,
okay,
yeah.
A
Yeah
and
and
for
the
crowd
reporters
that
are
actually
there
I
know
two
of
them
and
they
also
yeah
they.
They
usually
export
the
format
as
Apple
does
with
with
text
okay,
and
you
could
also
recover
those
classes
without
a
crack
reporter,
just
from
the
from
the
metrics
I
think
that
Apple
gives
you
also
crosses
about
your
apps
in
your
device
from
time
to
time,
and
you
can
upload
that.
But
the
problem
with
that
is
that
you
cannot.
A
You
cannot
attach
that
to
to
an
application
to
an
ended
to
a
instance
of
an
application
run
so
I,
don't
know
yeah
I'm,
not
sure.
If
how
it
will
work
with
grasses
on
on
iOS.
It
could
be
a
bit
more
difficult
at
just
sending
events-
okay,
because
it's
very
it's
very
tied
up
to
the
version
the
cross
and
the
yeah,
and
you
need
to
symbolicate
that
you
I
mean
it's
not
not
so
easy,
as
in
another
languages
like
so.
B
A
B
A
Anywhere,
except
in
your
own
devices,
so
yeah,
it's
crazy,
so
I
don't
know
for
crisis
for
iOS,
because
that
that's
a
very
difficult
topic
but
yeah
yeah
for
the
rest,
I
think
it
could.
It
could
work.
Okay
about
the
you
know
that
we
are.
We
are
still
done
super
logs,
but
we
have
been
talking
recently
on
starting
works
on
locks,
so
it
we
could
have
good
news
in
the
okay,
that's
a
medium
term
about
that.
So,
okay,
okay,.
D
That's
wonderful,
so
yeah
I
think
then
I'll
maybe
come
more
prepared.
Some
other
time.
E
So
there
was
I,
think
one
I
think
I
don't
know
if
Bryce
you
had
mentioned
this
before,
if
the
network
switches
right.
That
is
an
event
as
well
right.
It's
oh.
B
Yeah,
that
would
be
a
good
one.
I
guess
the
problem
is.
A
B
The
mechanism
for
detecting
that
is
expensive,
so
I
don't
know
necessarily
if
we'll
know
exactly
when
the
network
changes,
but
that's
a
potential
one
yeah.
If
we
can
find
a
cheaper
way
to
detect
that.
D
So
so
are
you
referring
to
like
when
a
user
switches
from
Wi-Fi
to
Cellular
yeah,
exactly.
D
So
that
is
yeah,
so
that
is
a
slightly
contortional
topic,
so
I
think
I
I
was
thinking
about
it
in
in
the
context
of
browser.
So
for
browsers
you
know
can't
detect
what
network
you're
on.
D
They
cannot
yeah,
but
I
think
this
property.
This
attribute
is
better
represented
at
the
resource
level
and
resources
currently
do
not
allow
attributes
that
can
change.
You
know.
Resources
are
mutable
today.
D
So
the
reason
I
say
this
is
better
suited
for
a
resources
that
once
you're
on
a
Wi-Fi.
Now
all
the
messages
you
know
will,
instead
of
repeating
their
attribute
against
all
the
attributes,
you
you
can
generate
one
event
saying
hey:
you
switch
to
Wi-Fi,
but
then
you
also
when
you
are
looking
at,
let's
say,
troubleshooting
a
particular
Trace.
D
You
know
you
want
to
see:
hey
was
the
user
on
a
Wi-Fi
or
a
cellular,
so
it
is
better
if
it
is
at
the
resource
so
that
you
can,
then
you
know
it
serves
for
for
all
the
you
know,
yeah.
B
But
the
problem,
the
problem
with
it
being
on
a
resource,
is
that
the
resources
only
applied
at
export
time.
So
that's
why
it's
immutable.
We
do
already.
We
already
monitor
the
network
status
and
that's
applied
at
the
time
of
network
requests
to
network
request
spans
so
but
yeah.
What
you're
saying
is
correct,
yeah.
It
makes
more
sense
to
be
on
the
span.
That's
actually
doing
the
network
request
and
we
already
do
that
this
time,
but
also
detecting
it
as
an
event
is
useful.
A
And
that's
when
you
usually,
when
it
usually
fails,
I
mean
when,
when
it's
really
important,
that
change
is
usually
when,
when
it
happens
during
a
span
or
your
network
request
that
never
reaches
yeah.
C
B
C
A
Yeah
and
also
yeah
I
mean
those
are
the
things
that
that
surely
can
I
mean
different
things
that
don't
usually
happen.
But
but
when.
D
Yeah
no
I,
I
I
think
spans
today
support
events
right,
so
you
could
attach.
B
A
D
I
think
it
has
to
be,
it
has
to
be
emitted
in
multiple
ways,
so
I
think
yeah,
yeah
I
think
the
base
case
is
like
you
know
you
are
in
in
on
one
of
the
you
know:
Network
types
right
as
a
result
or
Wi-Fi
or
something
else
and
in
a
steady
state
you
know
you're,
not
you
know
flipping
the
network
so
often
so
in
a
steady
state.
You
know
that
you
know
you
could
attach
you
could
emit
once
you
know
an
event
when
that
changes.
D
But
after
that,
do
you
want
it
as
an
attribute
on
every
span?
That's
emitted
or
you
know
just
one
set
the
resource
level.
That
could
be
one
consideration
to
think
about.
A
D
D
B
A
D
Okay,
okay,
so
this
is
one
topic
the
there
are.
D
You
mentioned
about
metrics,
so
I
think
one
approach
that
some
of
us
on
the
browser
are
taking
is
that
we
are
saying
that
we
will
not
generate
metrics
flight
from
the
browser
agent,
the
reason
being
anyway,
you
need
to
aggregate
the
metrics
from
multiple
users
on
the
back
end,
so
you,
and
also
since
now
there
is
a
whole
bunch
of
agent
code
gets
repeated.
You
know,
gets
the
the
agent
size
increases.
So
if
you
have
the
whole
SDK
Force
traces
for
spans
for
logs
and
for
metrics.
D
So
for
that
purpose,
we're
thinking
that
we'll
just
stick
to
spans
and
logs
and
not
do
metrics.
Because
metric
has
a
data
point,
you
know
you
could
just
send
it
as
an
event.
A
Yeah
I
mean
for
for
some
for
some
values.
It
makes
sense
okay,
but
if
you
want
to
really
get
they
did
I
mean
data
during
a
period
of
time.
A
I
I
think
that
if
you
want
to
yeah
integrate
that
values
on
a
period
of
time,
I
think
metrics
I
mean
do
the
work
for
you,
but
yeah
I,
probably
metrics.
Also
metrics
are
a
bit
outdated.
We
have
users
of
it,
but
yeah
the
the
API
might
change.
Okay,
yeah.
B
B
D
Okay,
okay
and
and
then
for
the
customer,
you
would
still
show
it
as
in
an
aggregate
form
across
all
users
or
do
you
they
also
see
individually.
B
You
know
I
I,
honestly,
don't
really
know
how
that
how
The
Collector
would
handle
it?
Okay,
okay,
yeah,
because
because
it
would,
you
know,
I
guess
the
dimensions
on
the
metric
would
kind
of
have
to
be
defined
and
that
sort
of
thing.
So
it's
kind
of
hard
to
figure
out:
okay,
okay,
because
we
yeah,
we
applied
the
same
resource
to
the
metric
data
as
we
do
to
the
to
the
spans.
So
yeah
I,
don't
really
know
how
they're
handled
on
the
back
end.
D
Okay,
okay,
so
yeah,
then,
if
there
are
valid
use,
cases
to
are
rather
strong
reasons
to
you
know,
generate
metrics
from
the
client
and
maybe
on
the
iOS
I
think
it's
the
bundle
size
is
not
a
big
deal
right.
It's
done
only
once
yeah
yeah
I.
A
I
will
also
say
that
it's
extremely
small,
the
footprint
of
adding
metrics,
can
be
in
a
low
case.
Kilobytes
number
I
I,
don't
know,
but
I
mean
it's
just
compiled
code
that
it
doesn't
have
lots
of
them.
It
doesn't
have
external
dependencies.
So,
okay.
A
I
would
say
that
it's
not
yeah
okay,
I
mean
some
exporters
can
take
much
more
memory
like
the
otlp
exporter
can
has
a
lot
of
dependencies
that
adds
memory,
but
specifically
the
metrics
one.
I
would
say
that
it's
very,
very,
very
small
footprint
for
it.
Okay,
I
I,
have
not
tested
it,
but.
D
D
B
D
Yeah
I
think
in
the
browser
every
time
you
access
I,
think
it
gets
downloaded.
All
the
all
the
JavaScript
gets
downloaded
so.
C
E
So
Santosh
I'm
wondering
you
need
a
per
user
Matrix
even
on
browser
right.
So.
D
E
D
Mean
basically
yeah,
they
are
no
I,
think
they
all
those
web
vitals.
You
know
they're
they're.
Definitely
you
know
they
are
considered
metrics
right,
except
that
you
know.
If
we
were
to
represent
them
as
metrics,
then
you
have
to
include
the
whole
metric
apis
metric
SDK,
so
it
increases
the
bundle
size
and
already
you
know
there
is
a
you
know.
Somehow.
The
design
of
open
telemetry
you
know
has
led
to
in
a
large
sizes
of.
C
D
Yeah,
in
fact
my
my
next
topic
is,
is
one
more
related
to
the
side
effects
of
open
Telemetry,
which
is
like
now.
There
is
a
you
know,
spans
logs
and
metrics.
You
know
there
are
three
distinct
signals.
You
know
they
are
sent
in
parallel
now
the
number
of
connections,
the
exporters
you
know
each
of
these
have
different
exporters
and
the
the
otlp
apis.
You
know
the
otlp
protocol
has
defined
three
different
endpoints
for
these.
D
You
know
you
you
don't
send
them
or
you
know
one
API
call
you
send
them
over
three
independent
API
calls
that
you
know
compared
to
you
know
our
existing
products.
Now
we
are
making
three
times
more
number.
Although
there
is
a
batching,
you
I,
I'm,
assuming
we
we
do.
We
batch
the
expands
before
they
are
sent
out,
like
let's
say,
instead
of
sending
as
they
are
generated,
you
send
once
every
few
seconds,
so
you
collect
them,
but
there
are
still.
D
D
If
we
can
have
like
a
unified
exporter,
can
we
have
like
one
endpoint
that
X
that
you
know
accepts
all
in
one
together,
so
there
was
a
good
amount
of
resistance
because
it
has
implications
on
The
Collector
and
it
has
to
change
in
a
lot
of
places.
I
mean
a
lot
of
places,
they
have
to
add
the
support.
D
So
instead
what
was
suggested
to
try
it?
It's
not
clear.
If
that's
going
to
work,
is
you
know
if
you
use
grpc,
then
you
could
share
the
grpc
connection
across
these
three
exporters.
D
So
you
so
you
use
you
basically
create
a
grpc
connection
and
then
configure
each
of
these
three
exporters
to
use
the
same
connection.
So
so
they
all
go
on
the
same
connection.
So
so.
C
A
You
guys
have
use
your
PC
today,
yeah
in
fact,
for
otlp
our
default
exporter
is
grpc.
Okay,
our
Json
exporter,
let's
say
Json
exporter,
it's
just
for
traces.
I
would
say
we
don't
have
the
full
support
for
though
I
know
we
have
for
now.
We
only
have
erpc
for
exporting
everything
in
our
DLP
and
we
have
Json
supported,
but
just
for
response
I
would
say
so.
D
When
you
say
Json,
you're,
you're
saying
that
it
is,
it
is
going
on
HTTP.
Is
it
not
grp.
D
Okay,
portable
for
grpc-
and
in
that
case
you
know
we
could
do
whatever's
say
that
you
could
share
the
you
could
somehow
modify
our
exporters
to
to
share
the
connection.
D
It
only
in
terms
of
number
of
TCP
connections
it,
it's
just
you
know,
one
connection,
so
the
connection
overhead,
at
least
that
reduces,
but
the
the
number
of
messages
that
the
payload
size
is
still
the
same,
because
you.
A
D
Connections,
I
think
you
know
once
we
recognize
the
problem,
then
I
think
we
can
work
out,
but
but
I
I
feel,
like
you
know
now.
If
we
are
including
metrics,
then
it's
going
to
be
like
three
times
the
number
of
connections
yeah,
so
the
number
of
connections
at
the
back
ends
will
have
to
handle
will
be
like
a
lot.
D
Okay,
that's
all
I
wanted
to
discuss
on
the
events.
I
think,
if
not
next
week,
I'll
come
the
following
week,
but
maybe
yeah.
Let's
discuss
that.
Sometimes.
C
A
Yeah,
okay,
so
for
the
rest,
yeah,
let's
meet
in
actions.
Let's
continue
with
the
topics.
Let's
meet
in
actions.
I
will
go
quick.
Make
me
not
part
of
a
pentelementary
group.
That
was
my
topic.
Yeah
I
finally
checked
it,
but
as
I
put
in
the
channel,
probably
you
saw
it
be
not.
A
You
have
to
create
yourself
a
PR
in
a
in
a
in
a
repository
and
mentioned
Bryce
and
me,
so
we
can
say
that
you,
you
are
in
with
I,
don't
know
exactly
the
words
that
says
they're
in
the
in
the
topic
with
that.
It
should
be
done
after
that.
Yes,
we
have
just
to
change
during
the
in
the
code
owners
file
and
it
should
work
automatically
for
for
approving
or
for
doing
things.
A
You
definitely
covered
them
all
from
what
I
saw
just
create
the
pr
I
I
would
say,
but
if
you
want
to
review
but
I
think
just
know,
okay
for
the
rest,
Swift
documentation,
yeah
I
had
to
change
the
I.
Didn't
do
it.
Sorry
I
completely
forgot
it.
A
C
Yeah,
what.
E
So
the
thing
is
when
you
use
a
plugin,
the
plugin
looks
for
a
swift
as
well
Swift
Distribution.
A
E
A
A
Plugin
work
lately
work
with
a
new
Swiss
package
plugins,
so
they
they
have
I
have
seen
that
there
was
one
explicitly
written
for
creating
the
protobuf
for
processing
the
protoba,
so
I
didn't
check
more
yes,
so.
E
What
happens
is
yeah
when
you
run
that
the
plug-in
process
it
looks
for
Swift.
You
know
this
fifth
binary,
so.
A
C
A
A
E
Does
it
does,
but
where
does
this
build
happen,
though
I
mean
that
is
so.
It
is
happening
on
the
Mac.
It's.
A
Like
yeah
it
will,
it
will
happen
on
the
on
the
Mac
building
it,
so
it
will
happen
on
the
CI.
So
you
can
declare
I
think
that
that's
that
that
I
didn't
check
that
plugin
specifically,
but
you
can
declare
basically
the
inputs
and
the
outputs.
So
if
one
of
the
input
changes
from
the
last
time,
it
generates
the
the
three
files
and
adds
it
to
the
compilation
so
that
that
could
work
as
an
option
just
to
build
that
ourselves
and
control
the
way
they
are
generated.
A
Okay,
we
will
also
have
to
I,
will
also
keep
the
protoba
files
ourselves,
the
ones
that
we
want
to
support,
not
linking
with
the
not
not
creating
a
git
surprise
or
something
like
that
with
the
official
protobuf,
because
we
wouldn't
be
able
to
to
keep
the
version
probably
correctly.
So
we
we
are
aware
when
things
change,
but
that
could
work
for
improving
that
a
bit
yeah,
okay,
so
and
about
the
release
I.
As
you
know,
I
released
a
new
version
with
the
change
with
the
protograph
and
it
was
working
right,
yep,
right,
okay,.
C
B
Was
it
wasn't
sending
data
because
I
was
accidentally
leaving
a
trace
open
that
was
gabbling
everything
up
and
never
ending.
A
No
problem
yeah
and
you
have
created
this
topic-
these
more
dogs
think
you
created
this.
A
B
So
it
talks
about
the
nsurl
instrumentation,
the
signpost
stuff.
What
else
is
the
other
one,
our
our
the
the
resource
extension
as
well.
A
Great
yeah
we'll
take
a
look
but
yeah.
It
looks
really
cool
thanks
for
that
I
also
yeah
I
will
talk.
Yeah
I
will
talk
about
that
later.
I
added
the
okay
new
new
issues
for
the
project.
There
were
some
support
things
that
I
have
been
handling
this
week
with
people
asking
questions
about
yeah
one
about
The,
Objective
C.
It
was
a
very
old
project.
It
was
here
like
create
an
object.
Is
he
friendly
literary
someone
asked
about
it.
A
And
also
this
one-
these
are
the
this.
This
was
we
created
this
issue
as
a
possibility
for
helping
people
as
a
first
good
first
issue,
but
no
one
has
taken
that
so
this
this,
this
user
asks
for
an
ATA,
for
it.
A
Yeah,
yes,
he
asked
for
an
update
and
I
say
no
update.
If
you
are
interested
in
doing
it
or
just
using
Associates
that
I
mean
I
can
I
think
that
she
and
she
said
that
it
was
for
using
and
ask
for
an
ADA
I
told
that
there
was
no
Ada
and
there
was
no
interest
so
but
and
I
just
created
this
example.
That's
just
right:
wrapping
the
the
hotel
span
and
Tracer
and
things
so
it
should
more
or
less
work
from
an
Objective
C,
but.
B
A
This
this
works
and
compiles,
so
it
was
based
on
some
work.
I
did
long
ago,
so
just
changed.
Something
yeah
is
even
has
the
header
injection
for
you,
wow,
so
yeah,
but
but
yeah
it
was
just
introducing
just
some
old
code
I
had
so
that
one
thing
and
she
also
asked
about
if
it
was
possible
to
to
link
as
a
sub
module,
so
I
can
imagine,
is
yeah.
C
A
I
I
would
say
that
I
am
doing
this.
This
thing,
let
me
I
am
using
something
like
this.
B
Yeah
I
only
see
I
only
see
the
browser.
A
Yeah
but
I
cannot
send
oh
I
can
yeah
just
a
chat.
Yeah
I
am
using
with
something
like
this
with
the
implementation.
Only
okay,
that's
really
important.
That
makes
I
mean
you
can
only
use
that
when
you
are
not
exporting
anything
that
it's
open,
Telemetry
related.
So
that
means
it's
like
a
private
import,
so
you
are
not
exposing
any
open
Telemetry
outside
your
framework.
Okay,.
A
Cool,
that's
yeah,
so
you
can
link
with
that
without
problems,
because
it
just
links
inside
your
library
and
doesn't
expose
any
any
library
right.
A
So
yeah
it
sits
here
to
to
for
that.
So
it's
not
very
documented
option,
but
for
me
it
works
really
well.
So
yes,
that's
a
so
yeah
and
I
think
that
also
I
I
don't
know
if
they
are
LinkedIn
as
a
sub
module.
There's
plank
people
for
switch
I,
don't
know
how
they
are
doing
it.
But
for
me
it's
it's
used.
It's
been
linked
inside
with
that,
okay,
and
that's
about
the
new
issues
that
were
removed
this
week
and
peers
yeah.
There
were
some
peers,
also
PRS,
yeah
yeah.
A
We
have
two
here
that
I
one
of
them
I
wanted
to
talk
to
you.
This
is
about
he
wanted
to
set
the
parent
in
your
session.
Instrumentation.
A
To
change
the
name,
so
basically
he
wanted
to
change.
The
parent
and
I
told
him
to
make
something
more
generic,
so
you
can
pass
the
spam
Builder
and
modify
whatever
you
want
inside
the
Callback,
so
he
finally
implemented
this
and
yeah
I'm,
Aina
and
I
approved
it.
I,
don't
know
if
you
think
it's
a
good
approach
or
not
yeah.
A
Yeah,
it's
pretty
basic.
What
allows
to
yeah
to
be
created
before
when.
A
B
A
Useful
okay,
so
I
made
I
approved
it,
but
I
I
wanted
to
check
with
you
if
it
make
made
sense.
So
we
can
so
we
can
merge
that
sure
you
want
yeah,
okay,.
C
B
C
A
Okay,
so
I
will
add
this
one:
okay
and
the
other.
What
was
actually
funny.
A
It
was
about,
he
said
that
he
had
a
crush
with
the
instrumentation
Library
info
that
was
being
rotated
at
runtime.
B
C
A
Access
to
the
dictionary,
keeping
all
the
tracers
who
has
been
accessed
a
lot
of
time
so
yeah
they
created
this.
That
in
theory,
was
fixing
that,
but
it
looks
like
he's
still
having
the
same
issue
in
the
same
point
at
the
gate
here
with
a
new
thing
so
yeah
we,
we
open
Telemetry
1.0.
That
was
the
last
one
and
this
he's
using
a
Splunk
on
top
of
a
step
and
Telemetry.
But
it's
getting
with
this
one.
So.
B
A
A
Yeah
so
I
have
asked
him
because
he
didn't
show
the
proper
access
I
mean
he
didn't
say
the
proper
access
here.
So
I
would
like
to
see
the
the
the
the
the
treasure
provider
SDK
get
of
both
threads
just
to
check
what.
A
On
how
yeah
that's
what
it
was
trying
to
to
know
so
yeah?
It
was
done
quite
long
ago.
So
I,
don't
know.
This
is
Avita
strength.
So
I
asked
for
more
information,
yeah.
B
How
is
he
running
that
that
access
thing,
maybe
we
can
just
try
to
reproduce
it.
C
A
The
person
is
0.8
that
should
be
linking
with
a
newer
version,
but
I
think
so
that
that's
strange
yeah
and
the
thing
is
that
he
wanted
to
yeah.
It
was.
It
was
funny
he
wanted
to
to,
instead
of
using
the
the
own
object.
That
has.
He
said
that
creating
a
string
from
the
just
for
comparisons
was
avoiding
this
cross,
so
the
but.
A
B
B
A
Yeah,
that's
what
yeah
that's
right
here.
It
is
I
know
not
here,
but
yeah.
A
But
the
thing
is
that
the,
but
that
the
the
threat
sanitizer
is
saying
that
it's
in
the
in
the.
B
A
E
But
no
no
I
I
would
think
they
would
have.
You
know
immediately
picked
up
the
new
version
right,
usually.
A
C
A
A
That's
right:
yeah,
okay,
so
for
I,
just
have
no
more
topics
from
my
site
either.
E
So
so
so
the
the
events
of
that
that
Santosh
is
talking
about
is
that
something
that
they
are
adding
into
this
spec
yeah.
B
A
So
yeah
they
are,
they
are
building
events
on
top
of
logs.
They
are
adding
attributes
to
logs,
so
they
will
be
recognized
as
events
by
the
back
ends
and
yeah
and
use
that
for
for
yeah
for
noticing
or
for
certain
things
about
the
things
running
in
a
generic
way.
A
Some
events
we
had
in
the
past.
Let
me
some
Cisco
people
also
connecting
right.
Yeah,
Stephen
Mickelson
came
to
some
meetings
in
the
past.
He
was
saying
that
he
was
starting
that
week
with
open
telemetry
and
when
I
think
that
he
actually
told
that
had
started.
He
stopped
coming
so
I,
don't
know
if.
A
Yeah,
that's
right:
yeah
I,
don't
know
what
yeah
they
have
people,
also
for
that
in
Cisco,
Dynamics
working
only
yeah
I,
don't
know
with
these
big
companies
Maybe
they
they
have
different
people
working
on
the
same
thing,
yeah.
A
Okay,
all
right
so
have
a
nice
weekend.
Yeah.