►
From YouTube: 2022-11-17 meeting
Description
Open cncf-opentelemetry-meeting-3@cncf.io's Personal Meeting Room
A
A
B
C
D
C
B
B
Okay,
yeah
I
have
I,
have
I
have
this
for
you,
our
local
implementation
is
a
I.
B
Yeah
there's
missing
tests,
but
we
are
planning
to
support
to
support
logs
really
soon
in
Swift,
because
Bryce,
for
course,
an
opportunity
today,
because
I
saw
the
duties
has
been
implemented
it
and
okay,
well,
Advance,
okay,
okay,.
B
A
D
To
join,
listen,
yeah.
B
Yes,
in
the
document,
that
is
the
notes,
I
think
I,
don't
expect
more
people
to
join
at
least
price
is
not
joining,
and
from
that
time
someone
appears,
but
it
looks
like
we
are
proud
of
now.
So
let
me
connect
on
Saturday
foreign.
B
B
B
B
C
D
I'm
yeah,
okay,
so
I
created
an
Excel
sheet
listing.
You
know
a
bunch
of
things
at
the
bottom,
like
the
the
idea
is
that
we
we
want
to
kind
of
you
know,
get
together
and
standardize.
What
attributes
we
want
to
send
in
what
event
in
some
of
these,
you
know
could
be
spans,
but
I
think
we
want
to
use
a
general
term
events
so
that
we
can
set
the
expectations
with
the
back
ends
that
this
is
the
schema.
D
These
are
the
attributes
that
they
can
expect
right
and
we
are
doing
this
exercise
even
on
the
browser
side,
where
there
are
I.
Think
last
time,
I
showed
on
one
of
the
meetings
that
I
think
where
the
more
people
from
different
companies
you
know
we
got
together
and
did
this
exercise
and
I'm
thinking.
We
should
do
the
same
thing
and
there
is
one
more
Dimension
here,
though,
compared
to
browser
here
we
have
to
be
in
sync,
with
both
with
Android
as
well.
D
So
I'm,
going
to
pull
in
folks
from
I
mean
I'm,
going
to
request
folks
from
Android
side
as
well,
but
I
think
especially
the
the
from
the
Splunk,
because
the
there
is
an
Android
SDK
from
Splunk
today,
which
belongs
to
the
splunk's
repository
and
they
have
agreed
to
contribute
it
to.
You
know:
open
telemetries,
GitHub
organization,
but
but
the
timelines
the
given
is,
you
know
it
will
happen
only
sometime
early
next
year,
let's
say
Jan
or
Feb
after
the
holidays.
D
So
once
that
happens,
I
I
think
I'll
ask
you
know
them
to
you
know.
You
know
join
this
effort
or
even
you
know,
while
you
are
talking
about
this,
I'll
share
this
Excel
sheet
with
them.
But
if
you
go
to
the
splunk's
website,
they
have
already
listed
their
data
model.
So,
for
example
like
they
have
the
browser
web
applications,
iOS
applications
and
Android
applications,
and
each
of
them
they
have
a
section
called
data
model
and
in
the
data
model
they
have
listed.
D
D
I
I
want
to
do
a
similar
exercise,
but
but
I
want
to
make
sure
that
we
all
are
on
the
same
page.
With
respect
to
you
know
what
open
Telemetry
is
is
going
to
have.
D
So
you
know
vinod
I
I,
don't
know
you
work
exclusive
on
iOS
or
do
you.
C
D
E
D
Okay:
okay,
do
you
work
with
sirat
by
any
chance.
C
D
Yeah
yeah,
so
she
has
been
coordinating
a
few
things.
You
know
between
open
Telemetry
and
their
internal
folks,
so
I'll
reach
out
to
her
as
well
to
review
this
document.
So
I
have
put
a
few
things
in
each
of
these.
F
One
thing
actually
Santo
Stephen
here:
if
I
can
yeah
yeah
I,
don't
know
if
you
can
hear
me:
okay,
I
can
I
I,
don't
know
if
you
saw
my
note
about
the
resource
extension
that
I
got
working
yesterday
that
got
rid
of
the
unknown
name
thing.
That's
no
longer!
That's
not
there
anymore.
Oh.
F
I
had
just
miscon
I
had
misconfigured
the
resource,
the
SDK
resource
extension
and
it
wasn't
being
added
once
that
was
included,
then
that
cleaned
up
the
on
you
know
the
name
and
the
namespace
fields.
Just
fine.
D
We'll
we'll
go
over
we'll
go
over
this.
You.
F
D
Okay,
yeah,
we
can
start
with
it
now
so
natural
and
vinod.
Do
you
have
any
questions
so
far?
I
think.
Do
you
guys
see
the
benefit
of
this
exercise.
B
A
C
D
So
so
yesterday
we
not
yesterday
a
few
days
ago,
I
think
you
know
we
went
through
some
of
these
resource
attributes
internally
within
our
company,
and
we
came
up
with
a
few
questions
and
we
wanted
to
discuss
here.
But
you
know
these
are
the
fields
except
this
device
model
name.
These
are
the
rest
of
the
fields
are
what
you
know
the
the
Swift
agent
can
send
today.
D
B
Device
I
mean.
E
E
B
B
B
A
Yeah
yeah
about.
B
The
model
name
and
modeling
the
fire
I,
don't
know
about
this.
Those
are
specific
thing.
I
think
we.
D
Both
of
them
Stephen.
F
D
F
Yeah
so
yeah,
my
note
to
you
yesterday
was
about
that
I
that
I
had
just
misconfigured
that
and
it's
that
the
SDK
resource
extension
is
what
adds
the
iOS
specific
data
to
the
resource
object.
F
I
was
using
just
the
default
resource
object
which
it
doesn't
know
anything
about
iOS
very
much,
but
once
I
got
the
proper
extension
configured,
then
we
picked
up
name,
namespace
version
device,
ID,
modeling,
again
I'm
still
running
on
the
simulator
until
Cisco
sorts
out,
our
Apple
account
so
model
name
in
my
case
showed
up,
as
you
know,
arm
64,
because
it's
a
M1
MacIntosh
and
it's
running.
F
B
Won't
it
won't,
but
it
does
not
very
very
limited
resource
but
yeah,
maybe
for
that
we
could
also
somehowat
the
simulator
somehow
so
I'll
use
something
more.
F
It
does
say
iOS,
it
does
recognize
it
as
running
on
running
iOS
on
an
arm
64
Mac.
So
from
that
you
can
deduce
it's
a
simulator,
because
there'd
be
no
other
arm64
Mac.
That
runs
iOS,
you
know
other
than
the
simulator,
and
it
does
recognize
itself
as
iOS
and
the
correct
version
of
of
iOS.
F
It
doesn't
Rec,
it
doesn't
call
it
Mac
OS
or
the
Mac
version.
No,
it
does
correctly
get
the
simulator's
iOS
version.
F
C
F
Yeah
yeah
sorry
we
were
just
it
was
between
the
time
we
had
the
meeting
where
we
discussed
this
Santosh
and
I,
and
the
rest
of
the
team
and
I
accidentally
had
a
had
a
minor
breakthrough
with
the
resource
object
yesterday.
So,
okay,
I
guess
I,
probably
didn't
distribute
that
information.
Well
enough
in
my
team.
Sorry
about
that
yeah!
No
problem!
No.
C
Problem,
great
okay,
yeah,
then.
C
D
Os
resource
provider,
great
okay,
then
moving
on
I
think
the
next
topic
is,
you
know
the
service.name
and
service.name
space.
These
two
attributes,
I
think,
looks
like
the
iOS
apps
have
a
display
name
and
a
bundle
ID
like
x,
a
DOT
P,
dot,
C
types
which
I
think
in
Android
apps
as
well.
They
call
it
as
a
package
name.
It
is
also
in
the
form
of
you,
know,
a
dot
b,
dot.
C
D
And
the
you
know,
One
requirement
that
I
want
to
put
forth
is
that
the
namespace
is
used
to
is
typically
on
the
back
end
in
a
namespace
is
a
grouping
of
multiple
services.
You
know
so
multiple
services
that
have
a
common
name
space.
You
know
they
get
grouped
together,
and
so
here
the
namespace
can
group
The,
IOS
app
and
the
Android
app.
D
So
between
IOS
and
Android
same
customers
app,
you
know
their
display
names
could
be
different.
You
know
their
package
ID
or
the
bundle
ID
could
be
different,
but
the
namespace
should
be
hard
coded
to
something
common.
D
That's
something
that
the
customer
needs
to
overwrite.
You
know
when
instrumenting,
you
know
when,
when
integrating
this
instrumentation
with
their
apps,
so
this
the
the
default
instrumentation
should
leave
it
unset.
D
And
that,
if
you
agree
with
that,
then
that
you
know
opens
up
the
question
that
where
should
we
send
the
bundle
ID
right
either
either
we
something.
B
G
D
No,
no
I
think
it
is
a
namespace
is,
is,
is
typically
used
for
grouping
things,
so
things
that
belong
to
same
namespace
are
are,
are
displayed
together
or
grouped
together
in
the
UI.
D
B
B
D
And
that
that
we
won't
know
that
the
customer
only
can
provide
that
info,
so
so
this
should
be
yeah.
B
That
that
that's
something
that
I
mean
could
be
probably
done
at
the
back
end
side
to
relate
Services.
Somehow,
okay,.
B
I
I'm
I
hope
that
it's
with
this
this
one
sorry
I
mean
that's
something
that
you
could
also
configure
saying.
This
audio
is
up
with
this
name
is
related
with
this
other
Android
app
with
this
other
name
manually.
B
On
the
back
end
by
the
user
of
the
application,
that's
my
point:
instead
of
having
to
set
a
value
related
to
different,
if
because
they
are
different,
apps,
it's
I,
don't
get
this
very
well.
I
mean.
B
D
B
B
Mean-
and
it
looks
like
a
very
product
focusing
probably
for
your
up
Dynamics
product,
very,
very
focused
to
that
you.
B
I
think
I
know
that
this
is
just
my
personal
opinion.
As
you
know,
I
am
not
working
on
on
drama
stuff,
but
yeah
yeah.
But
this
simply
all
the
rest
looks
great.
This
one
is
habit
abstract.
For
me,
it
looks
like
moving
some
logic
from
our
platform
into
the
Tracer
side.
That
can
be
done
because
we
could
have
this
default
attributes
and
maybe
we
can
have
extra
attributes
depending
on
the
platform
on
the
language.
Sorry,
the
platform
and
the
different
implementers
of
the.
E
So
so
not
so
this
is
already
there.
This
attribute
is
already
there.
You
know
the
default
is
unknown
service,
yeah,
yeah.
B
Okay,
yeah,
but
the
rest
look
perfect
this
one
yeah
for
me.
So.
D
So
yeah
we
don't
have
to
decide
now.
I
I
just
wanted
to
give.
You
know
my
thought
process
and
what
I'll
do
is
I'll
create
an
issue
in
the
Swift
repo
and
let's
get
comments
from
more
people.
D
I
I
recommend
that
you
reach
out
to
your
product
managers,
internal
other
teams,
working
on
the
Android
side.
If,
if
there
is
any
and
and
then
you
know,
we
can
take
a
call.
C
D
Okay-
and
this
is
also
result
right
now-
we
need
something
to
indicate
the
hybrid
agent
type
and
we
don't
know
what
that
will
be
I.
Think.
F
Let
me
I,
don't
know
if
you
guys
are
familiar
with
the
nature
of
our
hybrid
apps,
but
we
have
apps
running
on
flutter
and
react
native
and
xamarin,
which
are
have
some
native
code
in
those
environments,
but
also
embed
both
the
IOS
and
Android
native
agents
as
well.
And
so
you
might
have
a
case
where
it's
the
flutter
agent
running
over
iOS
or
the
flutter
agent
running
over
Android.
F
B
Time
they
basically
built
on
top
of
existing
Swift
and
or
Android
stuff,
yeah,
okay,.
D
I
have
a
feeling
that
this
will
fit
well
under
the
service.
Namespace
I'll
put
it
closer
to
that,
but
we
need
to
figure
something
some.
We
need
to
see
if
we
can
fit
existing
attribute
into
this
or
we
need
to
introduce
something
so
we'll
see.
C
G
B
Probably
something
may
be
related
to
the
deployment
Target
somehow,
so,
in
the
same
way,
we
can
say
that
it's
an
iOS,
Android,
samarin
or
flutter,
so
something
like
that
might
make
sense,
because.
D
B
Now
we
don't,
we
are
not
saying
this
is
an
IOS
app
or
an
Android
app
in
any
of
those,
maybe
just
because
the
OS
version,
but
okay.
G
B
Maybe
an
attribute
that
says:
what's
the
final
platform
that
we
are
targeting,
we
need
also
iOS
or
Android
or
Tamarin
or
flutter.
That's
what
I
mean?
Okay,
we
we
don't
have
a
proper
way
currently
to
say
this
is
Android
or
iOS
just
because
we
have
the
OS
name,
but
it
won't
make
sense.
If
we
I
mean
it
won't
be
the
final
answer:
if
we
have
flutter
on
top
so
probably
having
a
not
having
a
hybrid
stuff,
but.
B
You
know
that
includes
both
others
Android,
but
also
submarine
flutter
react
native
as
final
as
final
kind
platform
for
the
app
does
that
make
sense.
D
D
Yeah
so
I
think
this.
B
Or
maybe
the
final
native
iOS,
or
so
we
have
a
direct
attack
that
can
save
you,
which
platform
you
are
running
so
iOS
or
Android
directly.
You
know
what
I
mean
instead
of
having
to
no.
No,
what
I
say
is
that,
instead
of
saying
native,
say
iOS
Android,
so
the
final
task,
so
you
don't
have
to
take
two
tasks
to
know
what.
B
Because
you
are
not
you're
mean
because
you
don't
know
that
someone
is
gonna,
build
on
top
or
flutter
is
gonna
load.
E
E
No
I
think
what
he's
suggesting,
or
at
least
what
I
understood
is
like
OS,
you
have
a
platform,
DOT
type
and
platform
dot,
name
and
so
on.
That
makes
it
much
more
simpler
and
easier
to
understand.
E
B
D
Yeah,
so
let's
think
this
through
and
maybe
next
time
we
speak,
you
know
I
think,
let's
get
some
in
exchange
thoughts
further.
C
D
Yeah
but
but
I
I
I,
ideally
I
I
forgot.
What
is
the
possible
values
for
type
and
name?
But
let's
say
the
OS
name
is
iOS
or.
C
D
D
Yeah
then,
in
that
case,
we
just
need
to
enumerate
all
the
possibilities.
We
need
to
have
a
detailed.
A
D
So
that's
all
for
the
resource
attributes
now
moving
on
to
this
other
ones.
I
I,
just
put
you
know
the
you
know
these
attributes
looking
at
what
Splunk
has
okay
right
now
we
need
to
go
back
and
see
whether
there
are
any
additional
attributes
that
we
might
be
interested
in
like
one
one
thing
that
I
I
noticed
is
in
the
iOS
in
the
Splunk
case.
They
don't
have
an
error.
D
They
only
have
a
crash
report.
So
do
we
do
we
want
to
combine
them
into
one
or
do
you
want
to
keep
them
separate?
D
I
I
prefer
them
separate,
but
they
could
be
same
as
well.
So
I
want
to
hear
thoughts,
an.
C
D
B
D
B
Is
it's
yeah,
it's
a
final
message.
The
product
I
have
with
this
is
that
it's
only
reports,
one
for
it,
which
is
usually.
G
B
Enough
so
just
having
I
mean
this
would
have
a
an.
B
That
crashed,
so
you
have
the
state
of
all
of
them.
Sometimes
your
classes
come
from
different
threads.
So
having
just
one
thread,
ID
is
extremely
Limited.
D
But
what
do
you
want
from
all
threads
like
all
these
details,
the
where
the
stack
was
in
all
the
threats.
B
When
the
crash
happened,
yeah
that
would
okay,
that
would
be
the
okay.
That
crashed
could
be
the
trade
name
of
that
thread
that
crashed.
But
this
is
your
message
type.
The
stack
Trace
is
the
one
that
should
include
all
the
okay.
All
the
threats
in
the
tactics
could
be,
but
not
a
per
thread
stack
Trace,
but
a
fully
stack
Trace
that
could
work.
G
D
Yeah,
but
that
is
not
in
our
default
understanding,
because
if
I
see.
B
C
G
B
D
So
it
doesn't
tell
you
anything
about
you
know
which
thread
it
belongs
to
then
it
is
not
an
array
of
structures.
It's
just
single
structures.
Okay,.
D
Yeah
no,
but
if
you
want
for
all
threats,
then
then
I
think
we
need
to.
We
need
to
figure
out.
B
Another
mechanism
I
mean
our
provides
a
string
with
cross
dump.
That
includes
all
the
threads
that
that's
usually
what
is
sent
to
the
back
end
and
symbolicated.
So
we
can
just
add
that
to
the
stack,
Trace
and
Escape,
it
think.
What's.
B
Maybe
it's
Java
thing
or
Android
thing.
B
D
Means
you
know
here
there
is
one.
D
B
D
I
think
this
looks
like
at
this
for
python,
because
yeah.
C
B
G
B
D
Yeah
I
mean
I,
don't
want
to
go
through
these
things.
You
know
now,
but
something
that
we
could
work
on
offline
review
them
and
and
then
come
back
yeah
and
we
can.
B
Only
another
thing
that
that
brings
me
is
the
cross
address.
We
already
have
the
stock
Trace
the
address
there
is.
D
I
took
it
from
here
address
of
defaulty
instruction.
B
B
That
I
don't
know,
what's
the
what's
the
usage
for
that,
but
if
you
have
the
statue
you
have
the
address.
Definitely:
okay,
yeah.
Let's
continue
yeah
yeah.
E
Actually,
the
the
structures
of
the
other
two
you
were,
you
guys
were
talking
about
stack
Trace.
It
seems
like
it
should
be
an
array
right
because
especially
not
even
just
on
the
device,
but
even
on
the
back
end,
you
could
have
you
know
a
bunch
of
stock
traces
bubbling
up
so.
B
E
Yeah
so
so
the
wherever
is
I
actually
thinking
about
this
was,
with
the
spring.
The
spring
boot
I
think
the
newer
version
of
springboard
they
have.
E
You
know
the
exception
and
then
the
stack
traces
that
come
up
when
they
they
collected
an
array,
because
you
need
that
to
debug
an
issue
because
just
having
the
top
fund
doesn't
help
you,
because
the
problem
is
somewhere
else.
So
I
think
this
should
be
an
array
which
cannot.
F
E
B
So
it
might
be
also
done
on
the
on
the
back
end,
probably
with
different
Logic
for
its
language
or
its
platform,
but
yeah.
But.
E
I
think
I
think
if
you
do
not
have
let's
say
you
know
instrumentation
in
one
of
the
components,
but
you
still
have
a
exception.
D
So
so
one
method
I
can
think
of
is
that
there
could
this
could
happen.
C
You
know,
as
an
array
yeah,
that
certain
modes.
B
B
Yeah
that
that
would
that
also
makes
sense,
I
and
also
allows
having
smaller
tags
when
I
I,
don't
know
depends
on
its
platform.
But
maybe
size
of
the
attacks
can
be
very
big
if
we
send
all
the
stock
trades
in
just
one
one
attribute,
let's
say.
B
D
In
in
the
case
of
an
error,
this
is
just
an
unhandled
exception.
You
you
want
to
only
report,
you
know
just
one
error
right
or
even
in
this
case,
you
want
all.
B
Yeah,
if
you
you
can't,
maybe
you
can
have
a
logic
error
in
your
app
and
you
can't
report
that
as
a
logic
error,
but
that
will
I,
don't
know
if
that
could
be
called
an
exception.
Is.
D
Is
this
error
required
as
Stephen?
Do
we
know.
F
I
mean
yes,
we
are,
we
have
it
in
the
CSUS
product.
You
know
the
ability
to
report
there,
optionally
include
and
you,
and
we
we've
made
it
optional
as
to
whether
you
want
to
include
the
stack
trades
or
not,
they
can
get
very
large
and
in
certain
errors
it
was
like,
if
it's
a
logic,
error
some
sort
of
thing
that
shouldn't
happen.
Maybe
you
don't
think
it
would
be
useful,
so
you
can
exclude
it
because
they
get
very
large.
F
D
And
Does
it
include
the
stack
trace
for
each
thread
or
just
the
thread
that
you
know
where
the
crash
the
error
occurred.
D
F
Connective
include
all
yes.
Currently,
we
include
all
that's
kind
of
one
of
the
issues
we
occasionally
have
with
certain
applications
that
can
be
hundreds
of
thousands
and
sometimes
our
customers.
You
have
accidentally
used
it
almost
as
a
logging
mechanism
and
it
gets
out
of
hand
in
terms
of
data
in
the
back
end,
but
more
operational
issue.
B
F
B
F
If
it's,
if
it's
not
nil,
then
we'll
we'll
send
it
in
an
automatic
fashion
and
and
unhandled
classes,
there's
a
manual
API.
They
can
use
whether
without
stack
trains.
B
F
Yeah
the
trick
is
the
trick
is
to
make
sure
that
people
don't
are
trying
to
encourage
them,
not
to
abuse
them
in
your
API
and
send
you
know:
50
000,
bytes
on
every
type
of
event
they're,
if
they're
trying
to
treat
it
like
a
not
so
much
an
error
and
or
an
aberrant
case,
but
just
some
sort
of
reporting,
some
sort
of
issue
that
ends
up
being
more
like
something
that
should
just
be
a
log
event
or
something.
D
F
B
D
B
A
Error
laws
with
attributes.
B
D
Let
me
ask
one
question:
there:
I
think:
that's
that's
interesting!
You
you
talk
about
it.
Basically,
the
logging
API
that
we're
introducing
do
in
in
my
mind,
you
know
that
API
is
used
only
to
create
events,
but
are
you
also
thinking
of
using
that
API
to
ship?
The
you
know,
application
logs
to
the
to
the
you
know
back
end.
B
D
B
I
mean
yeah.
B
Use
the
user
can
lock
things
to
know
about
the
state
of
the
app
I
mean
it's
opportunity,
you
provide
them.
Yes,
they
will
have
in
their
server.
Okay,.
C
B
And
they
can
also
lock
errors
currently
directly
log
error,
and
so
it
appears
as
a
narrower
in
the
logs
of
the
app.
So
you
can
that
that's
the
kind
of
thing
we
are
not
I
mean
the
the
local
apis
would
support
that.
The
Finish.
D
Yeah
there
is
a
subtle
difference,
though
I
think
what
we
could
do
is
we
could
have
support
for
both
and
you
know,
leave
it
to
the
customer
to
you
know
either
have
both
or
one
and
the
the
difference
subtle
difference
that
I
can
think
of.
Is
the
default
logging
API
in
in
let's
say
in
Swift
iOS
side,
you
know
whatever
language
you
use,
Objective
C
or
whatever
the
default
logging
API.
Will
they
will
they
log
all
of
these
things
that
we
need
so.
G
D
I
think
you
need
to
check
so
if
if
all
of
them
are
not
available
and
then
you
need
to
explicitly
query
them
and
then
you
know
lock
them,
then
you
know
the
customer
can
enable
their
event,
but
otherwise
I
I
agree.
It's
redundant
yeah.
B
D
Secondly,
the
customer
also
needs
to
decide
the
volume,
because,
if
you
just
enable
shipping
of
all
the
logs
there
are,
there
are
two
downsides
now
one
is
the
end.
User
will
be.
B
D
B
From
the
instrumentation,
it
makes
all
sense
to
appear
here
because
it's
the
somehow,
if
the
the
the
monitoring
software,
the
one
who
reports
that
something
wrong
happened
in
the
same
way
that
it
reports
that
across
happened
in
in
comparison
with
the
user,
trying
to
show
an
error
on
the
platform
that
should
probably
be
allowed
itself.
That's
but
yeah
we
can.
We
can
also
think
more
about
it
and
and.
B
Yeah
I,
I,
yeah
I,
would
say
yes,
I
mean
we
are
currently
supporting
doing
that
with
a
span
events
exporting
them
as
logs,
for
example,
yeah
to
the
logs
I
I
I
am
actually
doing
that.
Okay,
all
the
events
are
written
to
the
logs
and
you
also
have
a
reference
to.
B
In
the
log,
correct
correct,
you
can
reference
in
the
back
end
to
the
span
that
created
the
law
and
the
error
and
that
stuff
yeah.
C
Yeah
yeah,
okay,.
D
Yeah
I
think
we'll
we'll
discuss
internally
too
whether
shipping
of
application
logs
is
that
of
you,
know,
value
to
us
as
well
will
will
see,
but
from
from
when
Bryce
you
know
created,
you
know
these
PRS.
What
did
he
have
in
mind?
What
did
you
guys
have
in
mind?
Is
it
to
ship
the
application
logs
looks
like
it
right
from
what
you
are
saying.
D
That
is
the
tricky
part
right.
I
think
you
need
to
be.
You
know
careful.
Let
me
bring
up
what
I
mean.
D
I
think
for.
B
D
No
either
ways
I
think
there
is
one
important
you
know
Point
here
that
open
Telemetry
logging
API
is
not
meant
to
be
directly
used
by
the
end
users,
so
the
applications.
When
people
build
applications,
they
will
have
their.
You
know
built-in
loggers.
They
will
use
the
the
their
framework
loggers.
D
You
know
to
create
logs,
so
you
need
to
somehow
you
know
tap
into
that.
You
know
source
of
logs
and
then
ship
them.
But
if
you
are
expecting
the
customers
to
use
your
open,
Telemetry
logging
API
to
to
create
logs,
then
then
that's
a
different.
B
G
D
You
tap
into
the
application
logs.
B
B
You
can
also
one
maybe
to
provide.
We
provide
our
promise,
Decay
users.
We
provide
a
way
for
them
to
write
laws
with.
B
C
D
Yeah,
okay,
so
I
think
we
are
short
of
time.
So
I
just
want
to
say
a
couple
things
about.
You
know
some
updates
to
the
event
API
specification,
because
you
added
you
know
this
PR
right,
I
think
it.
D
So
if
you
see
in
this
logger
you
have
this
emit
event
Builder
and
all
that,
so
the
event
specification
in
open
Telemetry
is
still
evolving.
D
There
are
a
couple
of
changes
now
the
the
event
creation
methods
are
taken
out
of
the
logger
API
and,
and
then
there
is
a
there
is
a
separate.
There
is
a
logs,
only
API
and
on
top
of
it
there
is
one
more
layer
called
event
API,
which
internally
uses
the
log
API.
B
D
It's
built
that
way,
so
take
a
look
at
you
know
this
link
and
the
second
change
is,
you
know,
you
know
this
is
a
second
change
where
the
event
domain
will
go
as
an
attribute
in
the
log
record
earlier
it
it
was
written
in
such
a.
The
spec
was
written
in
such
a
way
that
it
goes
to
the
instrumentation
scope,
attribute
yeah,
just
an
FYI
and
and
this
spec
is
still
evolving,
so
expect
more
changes
in
the
you
know.
Coming.
D
Yeah
but
but
the
logs
part
itself
is,
is
quite
relatively
quite
stable,
so
you
can
proceed
with
the
logs
implementation.
The
way
you
know
your
PR
is
I
think
you
can
proceed
with
that.
Yeah.
B
B
D
Yeah,
so
what
happened
was
that
the
events
part
of
the
specification
is
is
turning
out
to
be
a
little
controversial.
D
C
D
B
The
the
events
is
built
as
a
spec
on
top
of
the
logs
yeah.
Basically
so
that
spec
is
what
is
still
being
on
on
on
experimental.
E
C
E
E
B
B
We
are
let
it
in
go
so
yeah.
B
Some
yeah,
yes
reviewing
last
meeting
actions.
That's
what's
missing
from
the
meeting.
We
are
almost
out
of
time,
but
basically.
B
Yeah
I
think
that
This
was
done,
they
read
me
and
the
documentation
is
done.
The
context
map
we
we
were
leaking
in
the
active
context
manager.
We
were
leaking
some
memory,
that's
now
fixed
in
appear,
and
we
have
the
pr
for
logs
API
and
the
logs
SDK
that
it's
I
would
say.
You're
more
finished
I
was
doing
some
review
just
before
the
meeting.
B
It
looks
really
good,
just
missing
some
tests,
and
but
we
are
probably
merging
it
really
quickly,
at
least
to
have
there
in
the
library
as
an
experimental
thing,
so
people
can
play
and
also
evolve
it
if
it
doesn't
work
as
expected,
yeah.
Basically,
that's
all
I
was
planning
to
release
a
version
with
the
leak
fixed.