►
From YouTube: 2022-12-14 meeting
Description
cncf-opentelemetry meeting-2's Personal Meeting Room
B
C
B
B
D
A
A
So
we
can
continue
from
yesterday,
foreign
I
made
further
changes
after
you
added
your
comments.
Last
night.
E
Yeah:
okay,
I,
just
just
reviewing
that
yeah,
okay.
A
So
I
had
a
few
questions
and
and
for
the
context
of
Martin
and
purvi
I
think
we
are
putting
down
all
the
events
in
in
one
place
in
the
form
of
a
markdown
file.
It's
going
to
show
up
like
this
just
for
reference,
so
we
know
what
events,
what
is
the
structure
of
the
different
types
of
spans
and
events
we
are
going
to
emit.
D
A
So
I
have
added
the
resources
to
just
just
for
now,
so
that
we
review
them.
So
we'll
we'll
see
how
this
you
know
evolves.
So
this
is
newly
added.
So
please
go
over
this
as
well.
A
So
I
had
a
few
questions
so
for
some
of
them
some
of
these
you
know
they
they
are
actually
not.
They
will
remain
as
spans.
Let's
say,
Ajax
right,
so
jacks.
A
All
of
this
are
going
to
be
span,
attributes
and
not
they
cannot
be.
They
cannot
go
into
an
event.data
attribute.
A
So
I
assume
you
know
we're
on
the
same
page
with
respect
to
that,
you
know.
If
so
then
I'll
go
ahead.
Make
some.
You
know,
text
adjustments
to
indicate
that
this
is
not.
Unlike
the
other
events,.
E
Yeah
I
think
there's
that's
a
good
point.
I
just
didn't
realize
that
we
probably
had
to
split
this
up
into
two
sections
or
something
like
that,
but
one
where
we
described
the
table
refers
to
the
data
attribute
that
you
have
to
build
instead
of
every
event
and
then
the
things
the
Ajax
and
timings
I
believe
are
there's
one
more
span.
E
A
Other
span
is
that
possibly
of
the
page
view
right.
E
So
we
probably
should
have
a
different
section
for
that
and
then
probably
you
know
somebody
to
that
section.
We're
able
to
find
those
those
events
that
are
going
to
be
represented
as
fans.
A
I
think
the
the
the
problem
will
be
that
you
know
the
an
Ajax
call
is
is
no
different
from
you
know.
Any
standard
HTTP
request
right.
It's
just
coming
from
the
browser,
so
you
know
in
general,
people
would
be
expecting
all
these
attributes
to
be
on
the
span
not
inside
a
span
event.
E
That's
this
particular
event
is
the
call
that
you
make
right.
It's
not
a
it's,
not
an
incoming
interest.
It's.
A
Yeah
yeah
outgoing
yeah,
so
now
even
outgoing
requests,
they're
called
client
requests.
So
there
are,
there
are
client
spans
and
server
spans
explicitly
called
out
in
the
spec,
so
this
will
be
a
client
span.
F
Still,
no,
that
spend
kind
client
is
not
an
Ajax
request.
I
I
bought
this
up
like
many
moons
ago
going.
Okay,
we
can
just
do
this
and
I
got
shattered
down
on
the
spec.
Beating
that
that
client
span
kind
is
not
a
rum.
Client's
bank
account
that's
a
client
of
the
server
request.
So
no
don't
don't
compute
that
that
client's.
G
A
Yeah
any
client,
HTTP
client,
as
long
as
it
is
the
one
that's
originating
the
request.
You
know
that
it
will
be
a
client
and
therefore
it's
bank
account
will
be
client.
A
It's
already
there
I
think
there
is
nothing
further
needed
on
the
spec
I
think
we.
We
need
to
confirm
to
this
spec.
C
F
B
A
So
this
spans
child
span
will
be
the
server
span.
A
Yeah
of
the
receiving
on
the
receiving
side,
you
know
the
server
will
also
create
a
span
and
they
will
Mark.
You
know
their
span
as
a
server
span
and
similarly,
when
that
service
makes
another
outbound
request
to
another
service,
then
then
that
the
first
service
and
I
will
be
creating
two
spans
one
to
receive
the
request
and
one
to
make
another
outbound
request.
So
for
the
outbound
request,
you
know
it
will
create
a
span
and
that
it
will
Mark
as
a
client
span.
A
So
so
let's
say
this
is
the
you
know
browser
you
know.
Let's
say
this
is
the
internet
and
you
know
the
server
will
first
accept
and
it
will
create
a
server
span
and
then
let's
say
this
is
the
network
boundary
and
then
it
it
needs
to
make
one
more
call
to
another
service
right.
So
this
is
the
browser.
This
is
the
service
one.
This
is
service
two,
then
this
will
be
a
client
sorry,
this
will
be
a
server
span.
This
will
be
a
client
span.
This
will
be
a
client
span.
C
B
G
C
A
Yeah
yeah
correct
Ajax
request.
No,
you,
you
can
put
all
that
in
in
an
event,
but
my
my
point
is:
it
will
be
confusing
because
it,
an
Ajax
request,
is
not
any
different
from
any
other
HTTP
clients
request
and
and
therefore
it's
it's
better.
We
confirm
to
what
everybody
is
using,
which
is
putting
all
these
attributes
on
this
path.
E
No,
so
I
think
the
confusion
might
be
back
when
we
were
discussing
these
events.
You
know
I,
think
you
go
to
the
timeline
spreadsheet,
that
we
have
yeah
I'll
open
that
we
had
proposed.
There
would
be
stands
the
next
star
span
and
under
HDR
timing,
and
you
know
two
separate
things:
there's
a
span
there's
somewhere
along
the
line,
we
kind
of
got
confused
and
we
converged
them.
That's
probably
where
the
questions
arising
right
now.
A
So
so,
if
you
see
you
know
this
color
coded
timeline
here,
I
I
have
mentioned
this.
As
you
know,
the
blue
ones
are
spans
and
the
orange
ones
are
events
yep,
so
the
xhr
and
fetch
are
are
going
to
be
spans
and
they
will
have
corresponding
timing
information
in
the
events-
and
you
know
these
we
said-
could
either
be
span,
events
or
log
events.
A
Can
still
do
that
you
we
can
still
do
that
where
you
know
the
xhr,
you
create
a
span
and
emit
it
before
waiting
for
the
timing,
information
and
when
the
timing
comes.
Timing
info
is
ready.
You
associate,
you
create
a
log
event
associated
with
this
span
and
the
limited.
E
Yeah,
so
sorry,
what
is
what
is
the
then?
We
need
to
Define
these
things
as
two
separate
things
right.
A
They
are
already
separate
like
if
you
see
I,
have
separated
Ajax
timing
from
Ajax.
E
Oh
okay,
then
we
need
to
keep
them
together.
So
that's
this
number
one.
It's
our
editorial
thing
and.
E
You
either
keep
them
together
or
you
know
you
decide.
The
spans
are
all
going
to
get
defined
in
a
separate
section.
You
know
we
group
them
together,
so
either
one
okay,
okay,
I,
think
that's
kind
of
what
you're
going
for
okay.
That
makes
sense,
but.
A
But
at
high
level
we
agree
that
an
Ajax
you
know
the
the
the
parameters
I
put
in
this
table.
They
will
be
on
the
span
not
in
a
in
a
span
event.
F
I
haven't
looked
at
the
thing,
I've
really
been
looking
at
this
morning.
I
agree:
there
should
be
a
spam,
but
I
was
also
going
on
the
Assumption.
There
would
be
a
a
span
and
then
the
two
events.
A
We
can,
we
can
still
do
it,
but
but
my
concern
is,
you
know
it
will
be
there.
A
A
So
let's
say
you
have
a
span,
you
know
you
can
attach
events
to
it.
Spanish.
E
A
C
So
I
in
my
mind,
I
think
it's
okay!
If
they're
on
both.
C
So
we
have
like
the
pretty
fine,
HTTP
HTTP
attributes
that
can
be
on
the
spam
they're
already
defined
in
the
semantic
conventions
right
now
right
and
the
event,
if,
like
the
resource
timing
event,
it
just
comes
from
the
browser
API
and
it's
gonna
have
some
Fields
duplicated,
but
but
it
should
stand
on
its
own.
In
my
opinion,.
A
A
Yeah
I'm
talking
about
the
core
attributes,
Martin
I'm,
talking
about.
C
A
B
C
A
So
we'll
go
with
this
nerve
and
if
you
still
feel
maybe
you
know
we
can
discuss
again
another
time.
Maybe.
A
It's
part
of
the
fetched
timing,
I
think
yeah.
E
I
think
both
me
and
tutu
made
a
comment
there
I,
don't
know
we
should
call
this
Ajax.
It
should
be.
You
know
we
should
do
the
generic
name
for
that
yeah.
F
E
I,
don't
even
want
to
go
well,
I
know
what
you're
trying
to
do,
but
I
don't
think
we
want
to
go
Wikipedia
based
stuff,
but
it's
a
technology
that
we're
talking
about
right.
A
What
do
you
yeah
I
think
any
flame
is
fine,
I.
D
A
So,
let's
do
one
thing:
where
is
that
comment?
If
you
can
comment
here
somewhere,
I
think.
G
A
The
face
I
need
to
refresh
yeah
yeah.
So
how
did
I
get
a
link
to
this
yeah?
Maybe
others
you
know,
can
add
comments
and
then
we'll
we
can
decide.
F
F
B
E
A
Think
HTTP
request
might
be
better
because
tomorrow
there
might
be,
let's
say,
web
sockets
or
other
non-http
outgoing
requests.
E
We
should
probably
you
know,
be
able
to
use
the
same
thing.
I
don't
see,
they.
A
They
their
behavior
will
be
different,
though,
like
as
in.
E
E
A
You
know,
but
they're
going
to
be
long-running,
so
we
may
not
even
represent
them
as
fans
yeah.
C
D
A
I
think
the
the
point
is
we:
we
want
to
list
down
what
all
spans
and
events
or
browser
instrumentation
limits.
C
So
yeah
I
mean
so
I,
don't
want
to
like
Sidetrack
the
conversation
here,
but
I
was
wondering
like
if,
if
this,
if
this
document
should
be
just
the
events
schemas
for
browser
domain
and
then
maybe
maybe
we
have
a
separate
document,
that's
like
a
guide
for
implementing
browser.
Sdk
that
says
you
can.
This
is
where
you
would
use
advancing
display
would
use
spans.
A
Martin,
maybe
you
know
like
I
said
yesterday:
you
know
like
we're,
not
creating
this
as
a
PR
to
any
of
the
public
reports.
Yet
maybe
you
know,
let's
focus
on
the
content
now
and
and
then
you
know
after
the
holidays,
let's
come
back
and
and
get
a
plan
on
where
this
needs
to
actually
go.
Finally,.
A
I
have
a
few
more
things
in
interest
of
time.
A
Can
I
go
ahead
with
other
things?
Okay,
the
other
thing
is
similar
to
Ajax.
Even
for
the
page
view,
you
know
we
will
have
one
parent
span,
which
is
you
know
this
Pan,
the
page,
the
the
top
level
parent
span,
for
which
you
know
the
the
resource,
Fetch
and
other
things
you
know
could
be.
You
know
the
children
now
for
that
span.
You
know
we
could
literally
use
the
same
attributes
as
that
of
the
you
know
this
page
view
event
or
are
they
could
be?
A
You
know
fewer
than
them,
just
an
FYI
that
also
I
think
we
can
punt
it
for
now,
but
there
will
be
a
span
corresponding
to
this
page
view
with
the
same
attributes.
A
Please
continue
to
add
comments,
so
we'll
we'll
keep
working
on
this
until
we'll
resolve
all
the
comments.
So
that
way,
you
know
it's
easier
to
track
whether
we
are
done
or
not.
F
F
Let's
say
looking
at
you
know,
the
span
is
the
grouping
mechanism
and
then
the
back
ends
will
be
pulling
up
the
event
data
for
the
detail,
okay,
think
of
as
though
some
some
yeah,
so
some
some
will
be
the
same
and
some
might
be,
but
for
page
view
I
could
see
yeah
referrer,
URL,
the
user
consent,
and
the
type,
though,
would
probably
more
be
be
better
in
the
in
the
event
rather
than
on
the
spam.
But
we
could
talk
about
semantics
too.
F
There
will
be
some
people
who
are
already
using
this
and
putting
on
the
space.
So
we're
talking
about
different
instrumentations
in
different
implementations,
but
generally
I
think
keep
the
span
as
a
grouping
with
the
bare
minimum
context
and
the
event
is,
the
detail
would
be
the
way
I
would
view
it.
Yeah.
A
B
A
I
I
can
add
another
entry,
then
for
the
spam.
F
By
identified
that
I
expand
may
contain
and
then
I
like
Martin's
suggestion,
we
say
these
are
the
events
and
then
we
have
another
another
overarching
one
that
I
think
he
describes
the
timeline
spreadsheet.
That
we've
got
to
say
this
is
how
they
would
appear.
A
Yeah
so
so,
essentially,
I'll
create
one
more
object
here
and
then
put
fewer
attributes
there.
Foreign.
F
A
F
Because
yeah,
when
I
first
started
looking
this
this
morning,
albeit
I'd,
not
long
been
I've
been
awake,
I've
got
a
little
confused
with
all
the
resources
at
the
top
yeah.
A
Yeah
yeah
I
I
put
it
just
for
the
purpose
of
you
know
reviewing
them,
but
yeah
yeah,
we'll
we'll
love
to
see
how
to.
A
And
the
next
point
on
my
list
was
the
exception
so
exception,
even
in
the
case
of
spans.
It
is
currently
held
as
a
espan
event,
but
the
current
convention
is
that
all
of
these
exception
dot
stack,
Trace
message
type.
They
go,
as
you
know,
the
top
level
attributes
on
the
event,
but
so
they
cannot
go
into
an
event.data.
So
again,
this
will
have
to
deviate
yeah.
F
C
Okay,
even
though
they're
the
exception
attributes
are
used
on
event
or
sorry
on
spans
right
now,.
A
A
span
Martin,
it
is
currently
inside
a
span
event.
C
A
A
But
my
my
point
was
that
what
we're
saying
is
that
with
event.data,
we
are
generally
saying
that
all
events
have
a
standard
structure,
whether
they
are
span
event
or
a
log
event
and.
A
F
I
guess
what
we
could
propose,
which
was
sort
of
my
original
PR
in
terms
of
defining
me
a
event.data,
is
go
out
and
say
all
events
will
have
event.data.
F
Knowing
that
exception
is
going
to
cause
a
lot
of
discussion
and
see
if
we
can
Define
all
events
to
look
the
same,
because
that
would
be
ideal
for
a
back-end
to
say
an
event.
This
event
is
an
event.
F
F
Because
really
an
event
should
always
look
the
same,
so
the
back
ends
can
always
look
at
it
and
store
it
the
same
but
yeah
it's
at
the
very
worst.
We
could
say
an
exception
has
event.data
and
inside
of
that
that
has
fields
which
have
the
exceptional
prefix,
which
is
not
optimal,
but
it
would
still
then
conform
to
a
mapping
in
the
back
end
and
still
generally
conform
to
somatic
conventions.
A
Yeah,
okay,
so
if
you're,
okay
with
that
deviation,
I'll
highlight
it
here.
A
A
Yeah
so
we
said
I'll
add
the
prefix
I
missed
it,
so
there
are
going
to
be
some
new.
So
all
these
attributes
they
have
to
get
registered
in
this
pack
repo,
because
these
are
not
in
the
event.data
and
only
the
fields
inside
the
event.data.
We
we
said
we
don't
have
to.
A
Them
but
all
the
top
level
attributes
they
have
to
go
into
the
spec
repo
yeah.
F
A
So
please
comment
on.
You
know
how
you
know
these
names
should
be
so
so
I
can
I
can
create
a
PR
with
like
when
we
are
ready.
You
know,
I
can
I
can
create
a
Spectra
like,
for
example,
the
session
ID
I'm
wondering
since
it's
applicable
to
mobile
apps
as
well.
You
know
it
should
not
have
a
browser
prefix
so.
A
Yeah
so
I'm
thinking
to
keep
this
second
level
page
so
browser.page
dot.
You
know,
you
know
various
things,
so
this
could
be
browser
dot.
So
so
T2
had
a
comment
like
what
exactly
the
screen,
width
and
Screen
height
referred
to.
I
I
took
the
liberty
of
using
the
term
window
width.
But
what
exactly
is
it
yeah.
E
I
think
that's
what
did
you
also
pointed
out
there?
Okay.
A
So
is
it
okay,
if
I
rename
this
to
browser.page
dot.
E
A
Yes,
I'll
give
context,
so
the
reason
we
have
this
here
is
for
the
single
page
applications.
So
for
a
single
page
application,
the
the
URL
could
change
without
your
SDK
to
reinitialize.
F
So
they
would
just
then
send
a
new
page
view
event
which
would
have
the
new
screen
with
the
mic.
Yes,.
A
A
We
could
we
will
do
that,
but
I
think
the
intent
of
putting
this
here
was.
Let's
say
for
other
events:
let's
say
user
interaction
are
the
Ajax.
F
But
for
your
example
of
an
Ajax
request,
it
is
the
same
for
button
clicks
and
all
the
rest
completely
understandable.
A
So
the
the
problem
is
that
the
referrer
I
think
there
is
a
referral
policy
or
something
right.
So
even
the
referrer
may
not
always
be
available,
whereas
you
could
always
query
the
current
window
location
and
then
you
know
send
it
here,
which.
F
G
F
What
the
refer
is
for
an
Ajax
request,
because
you
are
the
page,
you
are
the
hosting
page.
What
you're
talking
about
with
the
policy
is
whether
the
referrer
Executive
Center
is
that
as
a
header,
that's
the
back
end
to
consume.
A
A
It
simplifies
if
we
move
it
to
the
individual
events,
because
then
we
don't
have
to
worry
about.
You
know
this.
E
A
No,
no
I'm,
just
saying
that
you
know
this
varying
resources
is,
is
anyway
at
this
moment
a
long
shot
right.
So
if
any
of
these
you
know
move
out
of
here,.
A
You
know
that
you
know
that
will
help
us
make
faster
progress,
but
then,
if
we
were
to
you
know
see,
if
it
is
the
right
thing,
then
I
would
say
the
right
thing
is
right
place
for
it
is
here.
E
I
I
did
some
not
to
the
question
the
the
namespaces
right.
There's,
multiple
Parts,
no.
A
No
I
think
nerves.
Point
is
you
know
this?
This
is
this
field.
This
attribute
is
better
moved
into
the
individual
events
and
and.
E
Not
I'm
not
asking
you
about
that.
No!
Let's
take
another
example
where
you
had
browser.pherson
ID
or
something
like
that
right.
So
why
is
a
page
dot
equation
idy?
Why
did
that
browser
impression
idea
of
browser,
a
great
impression
I
did
now?
Why
do
we
need
to
have
the
sub
ninja
spaces,
whatever
I
I
learned
from
training
property,
okay,
dot
and
instead
of
browser
dot?
This
impression
idea
browser
that
page
and
personality
I
thought
you
had
a
you
know,
reasoning
for
that
I
just
didn't
get
it.
F
And
I
guess
looking
back
at
the
word
document,
these
are
all
under
the
browser
and
comma
browser
common
resource,
which
some
of
these,
where
we
were
going
through
the
spreadsheet
was
the
case
of
these
are
just
common
fields.
That
would
also
be
on
the
event
not
that
they
would
all
be
attributes.
F
In
the
in
the
word
doc,
sorry,
hello,
doc
so
like
in
here
we've
got
like
the
spreadsheet.
You
just
had
there.
The
screen,
myth
and
height
were
actually
defined
as
values
for
the
event
so
effectively
like
the
varying
I.
Think
Kirby
had
a
bunch
for
in
the
spreadsheet,
as
well
I
think
Martin,
where
effectively
there's
just
a
bunch
of
common
attributes.
Well
they're
attributes
today,
but
common
attributes
that
are
always
present
in
an
event.
D
F
A
So
so
I
think
we
we
are
going
back
and
forth
on
this.
So
I
think
it
is
okay,
but
let
me
clarify
what
I
mean
I
think
you
know
prior
to
open
Telemetry.
Each
of
us,
you
know,
did
not
have
a
concept
of
a
resource
if
I'm,
if
I'm,
not
wrong
and
and
therefore
you
know,
we
all
used
to
you-
know
duplicate
these
things
in
in
each
event.
A
But
now,
with
the
with
the
way
the
resource
is
modeled
as
a
separate
common
object
in
the
payload.
You
know
we
have
an
opportunity
to
move
the
repeating
Fields.
You
know
out
into
this
resource,
I.
F
I
agree,
and-
and
it's
really
the
varying
attributes
from
talking
about
here,
because
the
varying
attributes
are
the
problematic
ones,
for
example
the
screen
width
and
height.
It's
a
case
of
someone
could
resize
the
browser
between
events
before
the
batch
has
gone
out.
So
therefore,
by
changing
the
resources
on
those-
and
you
end
up
with
it
with
a
different
group
of
events,
so
you
want
to
like
separate
them
where,
if
that
was
just
inside
of
yeah,.
A
Yeah,
so
are
you
suggesting
we
do
that?
Only
for
you
know
these
two
are
in
general.
We,
you
know
not
worry
about
varying
attributes
at
all.
F
Well,
this
comes
back
to
you.
Yeah
there's
advantages
of
disadvantages
either
way,
so
I
think
it
is
an
event
by
event
basis
whether
we
want
to
need
these
in
the
event
versus
in
a
resource.
I
know
I'm
still
not
quite
awake
this
morning
to
come
up
with
any
concrete
examples.
A
So
why
don't
you
add
as
a
comment
here
and
have
in
in
the
pr
that
way?
We,
if
you
want
to
think
about
it,
you
know
we'll
leave
this
topic
open,
yeah,.
A
Okay,
I'll
I'll,
take
an
action.
Attempt
to
you
know
come
back
with
either
with
a
you
know,
good
reason
for
this
intermediate
term
page
or
I'll,
remove
it.
B
A
That's
it
from
my
list.
We
can
go
over
if
your
comments
or,
if
others
have
any
topics.
A
Yeah
I'll
address
some
of
these
comments.
F
E
Like
what
you
know,
one
thing
that
being
sure
I'll
get
on
the
same
page
about
a
nice
and
I,
see
I.
Guess
in
my
in
my
head.
You
know
going
back
and
forth.
You
know,
depending
on
you
know
what
we're
discussing
and
stuff
you
know
these
definitions
right,
I
think
we
need
to
be
clear
on
what
are
we
talking
about?
Are
we
talking
about
the
events
as
they
are
yeah
over
the
wire
are:
are
they
as
as
they
are
going
to
be
at
the
receiving
end
once
they
are
addressed?
E
If
you
know
what
I
mean
right,
I
think
they
are
charified
that
really,
you
know
concretely
benefited
it's
been
easier.
Sometimes
I
I
see
myself
just
thinking,
oh,
when
I'm
consuming
my
page
view,
I
need
to
have
all
these
things,
but
if
it's
in
resource,
how
does
it
show
up
and
all
these
kind
of
things
so
I,
I
I,
think
I
realized?
You
know
most
of
them.
You
do
talk
about.
How
are
we
representing
these
events?
Over
The,
Wire.
A
Yeah
actually
on
the
event.data
I
had
that
question
too
so,
for
example,
Ram
not
from
Nev
I
think
currently
in
all
the
attributes.
You
know
they
have
like
this
right.
Let's
say
you
know
name
and
then
under
it
you
know
we'll
have
a
type
string
and
then
value
is
4
right.
It's
something
like
this.
F
If
this
daughter's
attributes,
so
so,
in
fact
the
every
attribute
on
the
wire
has
its
name,
its
type
and
its
value
as
separate
subfields
as
a
decent
Lobby.
If
you
like
one
potential
advantage
of
consuming
Cloud
events,
data
is
that
we
could
actually
say.
Data
is
actually
just
now
A
straight
map
and
they're
not
attribute
objects
at
all.
It's
just
a
a
straight
name:
bad
repair
and
we
Define
in
the
convention
what
the
type
is.
A
Okay,
so
yeah
I
think
in
the
event.data.
So
how
do
you
define
that,
though,
so.
F
We
can't
do
that
today,
which
is
why
I
think
if
we
just
gave
a
simple
pseudo
code,
Type
representation
or
we
call
it
pseudocode,
but
I
think
we
just
represent
as
a
nice,
simple
Json
blog
for
visualization,
knowing
that
the
stuffing
event.dato,
even
though
we
say
it's
like
Name
colon
value,
it
may
end
up
becoming
a
full
attribute,
but
just
from
a
visualization
perspective
we
could
say:
event.data
has
let
me
type
something
up.
Then
I'll
drop
it
into
the
chat,
we're
trying
to
solidify
it
might
make
it
even
easier.
A
F
A
F
Okay,
so
we've
got
element
with
thought
Okay,
so
it
might
look
something.
A
B
A
So
do
you
know
why
you
know
the
the
type
was
exclusively
added
because
of
some
programming
languages
have
some.
F
Oh
okay,
you
mean
for
the
protocol,
that's
this
forever,
so
yeah
like
for
client.
We
don't
need
it,
we
can
Define
it.
We
can
say
okay,
this
is
what
it
is,
but
from
a
product
perspective,
they
wanted
to
be
able
to
definitively
have
it
transformed
and
I.
Think
it's
just
the
product
implementation
to
to
have
the
generalization
of
the
types
like
we
use
product
and
identity.
I
didn't
like
it
for
the
same
reason,
because
there
was
extra
stuff
you
didn't
need
for
some
languages.
F
F
Part
of
this
would
come
out
like
if,
if
there
was
a
schemer
associated
with
this
like
if
we
say
that
we
have
Cloud
events
and
we
eventually
have
a
event.data
schema,
that
would
then
negate
the
purpose
of
the
protobuf
attribute
today,
because
that
the
schema
could
Define
the
type
and
therefore
the
back
ends
could
map
a
string
to
an
end
value
or
an
end
value
to
a
string
depending
on
what
the
schema
was.
But
that's
not
what
we
are
today.
A
Okay,
does
this
need?
Any
discussion
with
these
spec
folks
are?
Is
there's
no
confusion
on
this
correct,
but
today,
in
the
log
record,
does
it
say
that
it's
a
nested,
a
bag
of
attributes,
or
is
it
a
just
an
asset
object
so.
F
This
is
where
there's
a
slight
difference,
so
in
the
spec
it
is
it's
a
map
of
a
screen
to
any
which
is
exactly
what
I've
just
put
on
the
screen.
Okay
in
protobuf,
it's
a
map
of
a
string
to
any
value
and
any
value
must
be
yeah
a
string
of
whatever,
and
so
you
end
up
with
an
attribute
so
as
part
of
on
The
Wire
over
protobot.
F
It
actually
would
become
attribute
values.
A
F
F
A
Actually,
this
is
a
blocker.
I
was
hoping
that
we
could.
You
know
not
wait
for
that
event
or
data.
You
know
idea
to
be
accepted
in
the
in
the
spec,
because
it's
already
supported
by
the
log,
but
if
the,
if
the
protobuf
you
know,
doesn't
support
what
we
want,
then,
then
it's
a
non-starter
well.
F
A
F
I
think
you're,
probably
fine
with
collectors,
are
in
the
log
sync
today
with
tigrin.
That's
probably
the
we'll
probably
get
an
answer
from
that:
okay,
because
the
collector
needs
to
know
how
to
group
things,
especially
when
it
comes
to
like
metrics
and
probably
less
so.
It's
been
came
in
as
part
of
spans.
A
Yeah
I
I'll
add
examples
for,
for
each
of
these
yeah
and
I'll
Rama
will
also
the
exceptions.
Examples
are
Java
examples,
JavaScript
examples,
sure.
E
A
Oh
yeah,
the
the
further
it
goes
back
to
the
you
know
this
so
page
view
also
has
a
HTTP
URL,
and
my
varying
attributes
also
has
a
URL,
so
I
prefer
it
like
I.
Think
you
you're
going
to
come
back
on
this,
so
we
will
decide
based
on
that,
we
have.
This
comment
open
yeah.
F
There's
a
complication
there
where
we
start
talking
about
the
the
timings
and
we
say
the
timings
are
coming
directly
from
the
w3c
browser,
because
then
we
could
have
complex
but
yeah.
How
do
you
think
about
that?
A
bit
more
okay,.
E
Got
it
yeah,
I
understand,
but
I
think
the
the
thinking
was
that
they
were
good
documented
here
and
then
without
a
link
that
prompt
would
prompt
us
to
go
Define
it
and
you
know,
work
on
defining
them,
like
you're
doing
for
browser.
E
Exactly
so
the
the
the
flow
in
my
you
know,
my
expectation
is,
we
did
discuss
as
a
group
and
we
documented
things
in
their
dog
that
state
which
transfer
largely
to
the
pr,
and
then
we
can
continue
with
the
pr.
Let's
go
there.
E
It
cannot
be
a
you
know,
that's
basically
it
so,
whatever
we
captured
in
the
dark
largely
shows
up
in
the
pr.
These
are
missing
fields
from
the
dark.
That's
my
point
sure.
B
E
E
I
would
I
would
try
and
change
it
to
optional.
If
you
can.
A
Now
it
is,
it
is
optional.
Okay,
but
conditionally
required
is
is,
is
very,
you
know
is
very
loose
currently,
like
you
know
only
this
taxes,
but
but
there
is
no
machine
like
you
know,
you
can't
check
the
conditionality.
You
know
programmatically.
A
No,
no,
there
is,
there
is
no
solution
to
it.
I
think
people
are
okay
with
that,
but
Tito
I
have.
You
know,
comment
on
on
this
statement
that
the
status
code
should
be
zero.
We
actually
are
relying
on
the
spec
saying
that
if
there
was
no
HTTP
Response
Center
received,
then
their
attribute
should
not
even
be
present.
F
A
G
A
Yeah
yeah
I
think
that
I
think
then
either
we,
whether
we
fix
that
or
we
fix
the
spec,
because
we
are
relying
on
it
to
identify
the
the
number
of
network
errors.
A
A
A
Did
a
reference
Ram,
the
the
original
description
comes
from
the
Petro's
original
defined
and
you
can
override
it.
So
if
you
require
like
an
override,
but
even
in
our
case
it
has
to
be
a
full
HTTP
request.
A
You
point
out
like
what
is
missing
in
the
like:
did
you
mean
for
HTTP
URL
or
for
in
general
I.
E
Think
in
general
also
so,
if
you're,
if
you're
relying
on
you,
know
the
standard
definitions
that
come
from
semantic
conventions
defined
elsewhere
and
that
does
not
quite
match
what
we've
captured
in
the
talk,
we
should
augment
it.
E
Yeah
and
we
had
some
reviews
around
sanitization
and
things
yeah
I,
don't
think
we
need
to
go
through
this
as
a
group
here,
but
you
know
just
double
checking
you
know
seeing
you
know,
reservatively
matched,
if
not
yeah,.
A
No,
the
the
the
default
specs
command
does
have
only
the
you.
F
What
do
you
want
so
in
this
case
with
the
username
password
that
that's
the
standard
way
if
you're
talking
to
a
server
that
pass
credentials
in
the
clear
but
yeah
if
you're
an
app
you
may
have
you
know
pii
data
sitting
on
the
URL.
F
It
doesn't
that's
the
trick
so
either
like
the
way
we
handle.
That
is
either.
We
just
knock
it
back
to
the
domain
by
default,
or
we
actually
have
hooks
to
allow
people
to
go
and
sanitize
it
out.
So.
E
Yeah
exactly
I
was
going
to
say,
you
know,
we
should
say
generally
should
be
a
standardized
URL
and
we
allow
a
whole
bench
rotation
you
know
will
not
know,
for
example,
if
I'm
writing
my
application
and
I
know
there's
going
to
be
a
question
called
which
might
contain
or
something
like
that.
F
Yeah
log
in
when
you
do
it
do
a
a
remote
sign
in
it
may
actually
pass
the
the
email
address
both
for
Google
and
for
Microsoft
back.
It's
part
of
the
navigation.
A
Yeah,
so
this
will
be
done
through
the
processors
right.
Your
processors
can
alter
the
data
exported,
so
the
customers
will
have
to
add
processors
to
the
exporter
pipeline
to
to
modify
the
content.
A
All
right,
but
but
when
you
say
you
you
do
it,
you
know
you,
you
enable
the
customers
to
do
it
right,
yeah,.
G
Okay,
since
obviously
we
don't
know-
and
of
course,
everyone
has
different
jet
Precision,
for
example,
some
consider
numeric
IDs
also
to
be
yeah.
A
Oh
yeah,
it's
already:
okay,
okay,
all
right
so
I'll
make
the
changes
and
do
you
want
to
meet
on
Friday
to
close
on
this
before
you
all
when
we
all
go
on
holiday.
A
B
E
Many
open
comments
and
stuff
I
think
I
I
have
an
appointment
in
the
morning
Friday,
but
others
can
be
available.
Sure.
A
We
can,
we
can
just
have
a
short
one
to
to
close
any
thoughts.
Maybe
some
of
us
whoever
can
make
it.
F
Yep
I
guess:
let's
do
it
on
that's
the
thing
called
my
brain
would
be
causing.
B
F
I
believe
forgotten.
The
name
of
this
thing.