►
From YouTube: 2023-02-15 meeting
Description
cncf-opentelemetry meeting-2's Personal Meeting Room
C
C
Okay,
I
think
we
can
get
started,
I,
reinstalled
rooms
or
I'm
still
like
starting
to
find
things.
Can
you
all
hear
me
like
cloud
and
clear
I've
had
some
audio
issues
as
well.
C
D
C
F
E
Okay,
so
since
I
on.
G
Phone,
let's
finish
the
initial
agenda
topics
and
I'll
figure
out
how
to
share
from
my
desktop
while
I'm
on
my
phone.
G
Can
one
of
you
share
the
screen
and
present
the
agenda
like
I
cannot
see
who
all
are,
and
maybe
like
Mothra
like
if
you
can
or
vishwesh
like
who
works
available
to
the
send.
Please
do
so.
G
Yeah
you
can
share
and
take
the
agenda
note,
so
everyone
can
see.
C
G
I'm
still
trying
to
parallely
see
how
can
I
present
when
I
do
the
demo.
So
it
should
be
good
now,
because
I.
G
Okay
yeah,
so
if
you
haven't
been
a
call,
please
add
your
name
to
the
entity
list,
so
we
have
a
very
small
agenda
followed
by
a
demo,
so
we
got
some
feedback
earlier
about
doing
some
demos
of
upcoming
features
or
features
which
were
recently
implemented.
G
So
we
will
try
to
spend
like
few
minutes
every
week,
just
to
show
something
so
this
week,
I'll
be
doing
a
quick
demo
on
exemplars
and
next
week
Alan
has
agreed
to
do
exponential
histogram,
which
is
another
feature
which
we
expect
to
come
in
the
next
couple
of
months.
But
yeah.
Let's
quickly
go
through
the
agenda
item,
so
I
have
very
small
set
of
updates.
It's
all
about
the
release,
so
so,
starting
with
1.4
stable
release.
G
We
ship
the
last
release
candidate
sometime
last
week,
so
we
haven't
updated
all
the
the
open,
telemetry.io
docs
so
hopefully
like
that
can
be
like
done
this
week
and
then
we'll
get
some
more
feedback
next
week.
If
we
don't
hear
anything
bad
about
it,
this
would
be
the
last
release
candidate
and
we'll
call
it
1.4
stable
by
next
Friday.
G
So
that's
the
first
update
and
last
focus
on
like
what
do
we
do
next
because,
like
1.4
was
supposed
to
come
out
like
last,
like
November
end
got
delayed
and
delayed
and
delayed.
So
this
time
we
are
trying
to
like
see
if
we
can
put
a
more
predictable
timeline.
So
we
have
a
list
of
items
which
we'll
be
covering
for
1.5.
The
list
is
very
short.
Unfortunately,
I
forget
I
added
it
already,
but
it
doesn't
look
like
added
so
be
sure.
G
So
it
contains
the
list
of
items
which
we
are
hoping
to
ship
in
1.5,
and
we
have
also
put
the
timeline
of
three
months
so
like.
If
we
are
shipping
stable
next
week,
three
months
from
next
week
would
be
approximately
May,
so
March
April
May,
if
I
put
yeah
I,
think
I
put
you
yeah
and
it
has
the
items
which
we
are
currently
aware
of,
which
will
be
done
in
1.5
timeline.
G
D
Cool,
so
the
only
no
on
1.4
I
had
I
I
missed
the
the
removal
of
the
convenience
methods
for
ad
exporter
and
a
few
other
bits
that
we'd
added
in
so
those
look
like
they've
been
removed,
which
caused
me
a
few
problems
with
doing
the
upgrade.
It's
not
the
end
of
the
world.
Just
annoying
spent
a
lot
of
time
on
those
methods.
So
if
we
can
try
and
get
those
convenience
methods
back
into
1.5,
then
that
would
be
great
yeah.
D
Because,
obviously,
he's
been
closer
to
that-
and
there's
might
be
some
reasons
why
we
didn't
so
I'll,
have
a
chat
with
him
offline
and
then
create
an
issue
for
which
ones
we
want
to
bring
back
in.
If
we
want
to
bring
some
back
in
okay.
F
G
We
already
have
a
feedback
about
the
resource
one,
so
that's
the
only
thing
which
I
am
aware
of,
but
if
you
have
more
yeah,
please
let
us
know
but
like
in
general,
like
adding
new
apis
relatively
easy.
You
want
to
be
like
careful
about
adding
things
which
we
are
not
like,
100
sure,
so
we
can
always
add
new
things
into
the
next
stable
release.
D
D
But
do
you
think
it's
time
for
us
to
get
out
a
0.4
of
the
instrumentation
packages
and
move
us
away
from
the
RC
monikers.
G
In
general,
like
I'm
in
supportive
of
that
idea
like
we
could
do
that,
it's
just
that,
like
we
do
not
have
like
a
list
of
items
to
be
completed,
because
if
it's
called
like
stable,
then
we
need
to
like
do
some
sanitary
test.
It's
been
there
for
like
quite
some
time,
but
it
also
had
a
lot
of
public
AP
changes
in
the
last
several
months.
So
we
just
need
to
like
set
aside
some
time
like
review
the
apis.
G
At
least
we
don't
want
to
ship
something
which
will
break
in
the
1.0
so,
but
that
I
said
like
it's,
something
which
we
should
definitely
be
doing
like.
Ideally,
along
with
1.5
timeline,.
D
You're
only
sorry
you're
you're
proposing
three
months
to
get
the
zero
points
out.
G
No,
that
3.64
1.5,
so
it
should
be
like
sometime
in
in,
like
some
type
between
that
time
frame,
it
could
be
earlier
right,
I,
don't
think
like.
We
definitely
need
three
months:
it's
it
should
be
much
shorter
and
the
list
is
only
three
instrumentation
packages
right:
the
SP
net,
core
HTTP,
client
and
SQL.
Those
are
the
only
ones
which
we
ship
from
the
main
repo.
D
It's
the
ones
that
are
dependent
on
Microsoft,
BCL
stuff,
so
anything
that's
tied
to
BCL.
Basically,
because
that's
the
expectation
that
those
ones
are
you
know
stable
and
people
can
use
them,
there's
an
expectation
that
redis
and
anything
that's
non,
Microsoft
related
that
those
ones
yeah.
Those
could
come
secondary,
the
ones
that
people
use
out
of
the
box,
the
ones
that
we
document
on
the
main
Docs.
F
G
G
Then
SQL,
so
one
the
only
thing
which
we
might
expect
to
be
part
of
the
main
repo,
but
not
it's
not
part,
is
the
asp.net
classic
like
do
you
have
any
say
on
that
one
or
you
don't
like
yeah.
D
So
I'm,
just
looking
I'm
just
looking
down
the
list
of
what
I
was
including
so
yeah
it
was.
It
was
basically
those
four.
Those
were
the
four
that
are
part
of
the
main
reason
yeah.
D
So
I've
got
like
so
the
way
I'm
doing
this
is
because
I
don't
think
it's
right
to
just
say:
let's
do
a
0.5,
because
there's
no
real
documentation
about
what
we
mean
by
an
RC.
What
we
mean
by
an
alpha,
what
we
mean
by
Beta,
so
I've
built
a
bit
of
a
document
that
I'm
running
past
Carter
Philip.
D
So
because,
obviously
he's
very
big
in
the
docks
and
what
have
you
around?
How
do
we
Define
releases,
because
what
we've
been
releasing
is
not
really
an
RC,
it
was
an
alpha
and
then
it
was
a
beta
and
then
it
was
then
an
alpha
again
now
it's
an
RC
and
I
think
we
need
to
Define
what
those
semantics
are
and
that's
what
that
whole
document
is
about.
D
G
D
So
it's.
How
do
we
get
something
that
makes
sense
to
the.net
people
and
talking
to
Java
people
and
the
go
people?
Not
none
of
this
that
we're
doing
right
now
makes
sense
to
them
so
yeah
when
I
was
talking
to
them
about
rrcs.
They
were
like
that.
Isn't
an
RC
dude,
so
yeah
I
think
we
just
need
to
Define
what
we
mean
by
an.
G
Rc
yeah
I
mean
right,
I
want
to
like
see
that
talk
before
I
comment
on
that,
but,
like
you
I
totally
agree,
we
did
not
Define.
The
only
statement
we
have
is
an
artist
more
stable
than
beta,
which
is
more
stable
than
Alpha,
but
that's
the
only
recommendation
which
we
have
in
our
dotnet
web
page.
Like
sorry,
the
the
there
is
a
dock
which
talks
about
versioning
and
the
only
mentioned
about
these
three
pre-release
identifiers
are,
they
are
just
like
one
is
better
than
another.
That's.
G
F
G
G
G
D
Right,
okay,
so
what
I
will
do,
then?
The
document
I've
got
I
will
put
that
as
a
PR
to
that.
So
you
can
comment
on
it.
Perfect
yeah
that
works
right.
I'll!
Do
it
to
that
yeah
I
was
doing
something
completely
separate
so
but.
I
D
D
So
if
we've
got
an
actual
document
that
says
it,
then
you
know
we
can
just
point
to
that
and
just
go
yeah.
So
I'll
reread
that
and
I'll
put
my
my
document
that
I've
been
created
on
that
one.
F
Okay,
yeah.
G
Anything
else
to
be
discussed
on
terms
of
yeah
I
mean
I,
while
on
the
timeline
so
1.5,
and
then
we
have
1.6
plan
where
we'll
bring
back
all
the
loading
changes.
If
you
are
paying
attention
like
we
removed
all
the
recent
logging
related
stuff
from
the
main
branch,
they
are
now
sitting
on
a
separate
Branch.
So
we
expect
to
bring
that
back
by
1.60
that
it's
going
to
be
coming
even
after
1.5
I
guess
the
logging
stick
would
be
calling
the
loading
SDK
spec
as
stable
by
then.
G
G
D
That's
fine,
that's
fine!
We
can.
You
know
if
I
can
tell
people
that
I'm
talking
to
about
it,
that
we're
expecting
it
for
1.6
and
the
timeline
is
based
on
the
perspect
being
stable.
We
expect
it
to
be
around
this
time.
People
just
want,
you
know,
actual
information
that
they
can,
that
they
know
the
basis
of
the
information.
I
think
is
probably
the
best
way
to
put
it
so,
but
yeah
close
door
stuff.
That's
the
thing.
They've
got
problems
with.
F
G
So
that's
all
I
had
about
the
Milestones,
it's
already
updated
in
the
Milestone.
So
if
you
look
at
the
repo
I
have
created
One,
Note,
5
and
1.6,
so
1.5
already
has
released
and
1.6
the
issues
are
specifically
to
logging.
It's
tracked
as
a
separate
issue.
It
said
again
link
to
from
the
yeah
that
Milestone
35
would
be
a
good
proxy
to
read
what
is
required
for
1.6.
G
Basically,
the
things
which
are
required
for
logging
to
become
stable,
I
think
we
are
still
missing
new
things,
because
a
lot
of
changes
which
occurred
still
occurring
in
the
logging
specification
so
yeah.
So
that's
for
1.6.
D
Awesome,
can
we
just
acknowledge
that
the
disciplinary
for
Mike
being
light
will
be
next
week.
F
Okay,
any
other
topic
I
will.
G
Try
to
share
my
screen
and
see
if
I
can
do
a
quick
walkthrough
of
what
we
have
for
examplars.
Are
there
any
other
topics
which
needs
to
be
discussed?
Oh,
it
looks
like
plans
also
join
just
now.
D
Yeah,
that's
what
I'm
saying
you
just
sorry:
I
refer
to
his
mic,
but
yeah.
That
was
that
that
was
the
disciplinary
comment.
I
made
God
my
humor
is
British.
Humor
is
lost
on
you
all
sorry.
B
Mine's
I,
don't
know
where
it
really
fits
to
be
honest
with
you,
because
this
is
my
first
time
kind
of
joining,
but
it
has
to
do
with
the
service
name
and
coming
up
with
like
a
better
default
specifically
for
IIs
applications
and
I
created
a
couple
issues
out
there,
or
at
least
one
and
then
who
was
I
working
with.
Let
me
think
it
was.
F
B
It
was
Rob,
payak,
I,
think
oh.
B
Yeah,
yeah
and
and,
and
he
said
to
make
to
suggest
a
change
to
the
actual
specification,
so
I
wasn't
quite
sure
what
to
do
there,
but
I
did
open
up
a
a
ticket
with
that
as
well.
B
A
B
And
that's
the
issue
I'm
having
right
now
is
we
have
a
bunch
of
applications
tied
to
a
single
lap
pool
I
know
it's
not
ideal,
but
we
have
hundreds
of
them.
So
it's
not
really
possible
for
us
to
break
those
apart
and
you
know
we
have
like
different
environments
and
levels
and
things
this
is.
B
You
know
20
year,
old,
Legacy,
application
type
of
thing,
and
so
what
I
was
hoping
we
could
do
is
have
some
sort
of
a
a
service
name
that
corresponded
to
the
virtual
directory
instead
or
or
something
like
that,
and
if
that's
not
possible,
I'm
gonna
have
to
do
some.
Some
actual
code
changes
probably
to
see
if
that
will
even
work.
So,
oh.
B
Yeah
yeah,
so
if
I
change
it
via
like
the
web
config,
which
is
what
I
would
normally
do,
that
only
works
per
app
pool
because
of
the
way
that
it's
set
up
right
now
in
the
auto
instrumentation
piece,
I.
Think
and.
D
Yeah,
so
your
problem
is
they
also
instrumentation
side.
Then
the
auto
instrumentation
side
is
setting
up
the
tricks
as
built
in
he's,
setting
up
open
Telemetry
based
on
the
config
that
you've
set
in
the
web
config.
So
that's
not
really
a
specification
thing
or
even
on
a
hotel
thing,
I'm
afraid
really.
Okay,.
D
D
So
as
far
as
the
library
that
we're
talking
about
today,
it
just
takes
that
service
name
and
provides
it
over
doesn't
provide
us
with
a
service
name,
I'm,
not
sure
why
it's
coming
over
here.
It
should
really
be
the
auto
instrumentation,
but
if
the
Oster
instrumentation
can't
solve
it,
yeah
yeah,
you
have
the
option
to
actually
put
hotel
in
manually.
B
I
do
and
I
can
it's
just
it's
just
a
bit
more
lift
I
was
hoping
the
auto
instrumentation
would
kind
of
would
do,
would
do
what
I
needed
to,
but
the
it's
it's
possible
to
do
it
manually,
I
guess
I
was
another
question,
I
have
and
sorry
if
this
is
not
the
right
spot,
but
is
there
a
way
to
do
both
the
auto
and
the
manual
at
the
same
time,
or
is
it
just
one
or
the
other.
G
D
What
I
would
say
is
I
will
offer
some
of
my
time.
I'll
dump
my
details
in
the
channel.
If
you
want
to
have
a
chat
offline.
D
F
G
Looks
like
we
have
one
more
Topic
in
legendary.
So
can
you.
F
Open
that
PR
right
yeah.
J
Yeah
so
I
changed
this
to
not
add
anything
to
the
options
class
really,
but
to
add
like
a
new
project.
J
Method,
that's
being
added
just
like
configure
the
persistent
storage
Factory
which
will
like
later
be
created
and
then
used
by
the
otlp
exporter.
Just
was
looking
for
people's
like
comments
about
this
way
of
doing
the
configuration
for
the
the
persistent
storage
people
did
seem
to
be
like.
Okay,
let's
put
in
a
different
package,
so
I
think
that's
in
the
right
going
in
the
right
way,
but
the
just.
If
anybody
has
any
comments
about
like.
G
J
Yeah,
so
if
you
could
you
go
to
the
otlp
exporter
options
class,
you
can
see
the
what
was
added
to
that
I.
Think
it's
just
a
yeah.
Okay.
G
E
G
G
G
Likely
to
see,
but
just
be
clear,
like
we
won't
be
merging
it
for
1.4
for
PS
reasons,
because
we
we
want
to
like
stabilize
it
so
only
like
dog
changes
or
any
other
change
which
are
like
not
going
to
affect
the
core
packages
would
be
like
merged
until
we
ship
1.4.
So
that's
at
least
a
week
out
so
I
mean
even.
I
E
G
All
the
metrics
work
into
a
separate
Branch
just
to
not
affect
the
stability.
So,
if
you
want
like,
we
can
put
that
like,
if
you
want
like
really
move
very
quickly,
you
can
ask
one
of
the
maintainers
to
create
a
separate,
Branch,
main
hyphen
odlp
and
then,
like
start
merging
things
into
it.
So
you
don't
need
to
like
wait
for
the
whole
thing
to
work
like
just
create
like
really
small
PRS,
like
one
PR,
creating
the
new
project,
one
PR
exposing
the
option,
then
the
next
one
doing
the
thing.
G
So
it
will
be
like
a
lot
easier
to
get
people's
attraction
to
review.
G
G
H
So
and
also
I
think
like
what
whatever
changes
Michael
has
here.
I
mean
if,
if
this
looks
good,
I
would
like
to
get
this
merged
and
then
follow
up
on
the
other
Suggestions
by
Blanche
in
the
follow-up
PRS.
If
this
looks
good
enough,
if
not,
then
like
definitely
we'll
go
ahead
and
Implement
those
other
suggestions
as
well,
so.
G
G
So
you
have
an
easy
time
getting
things
merged
into
that
plants
because
we
have,
we
have
to
protect
the
main
branch
now
yeah.
So
anything
in
specific
you
want
to
ask
now-
or
apart
from
asking
everyone
to
review.
H
No
I
think
so,
like
whatever
Michael
has,
if
it
as
I
said
like
if
it's
good
enough
I
think
we'll
like
go
ahead
and
merge
this
one
and
then
iterate
on
the
other
suggestions
in
the
follow-up
yeah.
So
that's
the
only
ask
since
it's
blocked
since
long
time,
so
I
just
want
to
do
this
moving
forward.
Yeah
Michael.
Do
you
mind
updating
that,
in
the
description
that
it's
going
to
the
metrics
Branch
instead
of
yeah
I
changed
the
title,
all
right?
Okay?
Okay,
thank
you.
G
I
will
attempt
to
share
my
screen
now
and
still
keep
audio
from
my
phone,
so
we
share.
Thank
you.
Let
me
try
to
share
it
now
sure.
E
I
E
E
G
G
Yeah,
so
if
you
are
like
totally
unfamiliar
with
what
an
Exemplar
is
it's
one
of
the
features
which
we
are
plugging
for
Matrix,
it's
a
specific
metric
point
which
which
allows
you
to
give
exemplars
exemplars,
give
like
specific
data
points
which
can
be
used
for
navigating
from
Matrix
to
other
signals.
It's
best
explained
with
examples,
so
we'll
we'll
see
like
what
examples
are
in
a
moment.
G
So
what
we
are
trying
to
do
in
this
report
is
like
we
are
trying
to
add
the
support
for
recording
exemplars
into
the
dotnet
SDK
itself,
as
in
the
open,
telemetry.net
SDK.
So
we
expect
the
user
to
do
like
no
action
like
exemplars
would
be
like
collected
automatically.
G
G
So
before
going
to
the
details,
I
will
quickly
show
what
it
would
look
like.
So
how
I
have
just
few
minutes
ago?
I
ran
all
the
required
tools.
I
need
so
I
have
a
Docker
file
which
I'll
go
through
later
so
I
have
collector
running.
I
have
Tempo,
where
I'm,
storing
all
the
traces
and
I
have
Prometheus,
where
I
store,
metrics
and
I'm
using
refinance
the
UI.
So
I
have
this
Docker
infra
already
running
in
my
local
machine
and
on
top
of
that
I
have
the
demo
application
which
we
have
in
our
report.
G
So
this
is
the
example,
obviously
how
it
did
not
modify
it,
except
like
one
thing
like
this
is
a
one-liner
which
enables
Exemplar.
So
we
won't
like
describe
detail
of
what
these
lines
do,
but
in
general,
like
this
synapse,
which
is
instrumented
with
open,
Telemetry
traces
and
metrics
and
has
X
simpler
enabled.
So
it's
running
right
now
and
I'm
running
a
very
small
Lotus
to
it.
So,
let's
quickly
go
to
what
it
would
look
like
with
example.
G
So
this
is
my
graphina
dashboard,
which
is
running
in
my
local
in
the
docker
container,
it's
configured
to
query
Matrix
from
Prometheus
and
it's
configured
to
query
traces
from
Tempo.
So
we
can
do
it
like
independently
I
can
query
Matrix
and
traces
independently,
but
the
real
gist
of
this
demo
is
I,
am
able
to
query
Matrix
here
and
since
I
have
enabled
the
like
forget.
The
chart
like
I,
haven't
written
the
actual
query,
so
just
focus
on
the
part
where
it
it
shows
like
that
special
Diamond
Mark
thing.
So
that's
the
example.
G
So
if
you
look
at
this
would
be
a
actual
chart
and
you
would
see
Matrix
examples
popping
up
here
and
there,
so
the
one
which
I
am
showing
right
now
is
an
example
of
an
example.
So
if
you
look
at
it
like
it's,
it's
a
specific
data
point
like
us,
because
otherwise
what
you
get
from
Matrix
is
an
aggregated
thing.
So
you'll
never
get
to
see
the
row
metric
value
because
you'll
always
get
the
aggregation,
but
with
X
and
plus
you
will
get
the
row
value
the
actual
value
which
you
emitted.
G
You
never
know
the
exact
time,
and
so
xmlers
can
be
used
for
like
knowing
the
extra
time
and
value
the
real
value
of
using
X
and
plus
is
not
knowing
that
the
real
value
is
all
about
how
to
navigate
from
Matrix
to
Tracer.
So
that's
what
I'm,
showing
next
to
one
of
the
things
which
we
put
in
xmlr,
is
the
context
on
which
the
Matrix
was
reported.
G
So
if
you
are
recording
a
metric,
for
example,
if
you
are
reporting
your
latency
and
assuming
that
you
have
a
distributed,
Trace
backing
that,
so
you
have
a
trace
area
and
spanning
so
the
XML
would
record
it
as
this
and
grafana
as
a
UI
like
it
knows
how
to
enable
excellence
and
navigate
from
one
signal
to
another.
So
here
I'm
going
to
click
query
with
Tempo,
which
takes
me
instantly
to
that
particular
Trace
in
my
tracing
back
end,
so
I
use,
Tempo,
I.
Think,
like
other
tools,
Like
Jaeger
also
supports
this.
G
Tempo
was
just
easier
to
get
started.
So
I
just
picked
that
so
now
we
can
see
like
the
metric,
which
I
the
xmlr
will
say,
click
it
was
recorded
in
the
context
of
this
one.
So
how
it
helps
is
like
when
you
have
a
x-pipe
in
your
latency,
you
can
easily
go
to
the
spike
in
your
metric
dashboard
and
click
on
an
Exemplar
which
will
take
you
to
the
Natural
Trace,
which
which
was
part
of
that
Spike
like
it
would
be
a
place
which
would
contain.
G
Let's
say
your
like
typical
latency
is
like
around
100,
but
then
you
get
one
something
which
is
like
way
above
normal.
So
you
want
to
know
like
what's
causing
it
so
matrix
by
default.
Will
just
tell
you
that
something
is
wrong.
You
wouldn't
know
like
what
or
where
so
by
using
xlrs,
you
are
able
to
navigate
from
Matrix
to
traces
and
from
traces.
You
know
exactly
which
component
is
misbehaving
and
then
like
two
way
more
narrower
focused
investigation.
G
I
did
not
do
the
make
it
part
of
the
remote,
but
it's
obvious
that
if
I
can
correlate
from
Matrix
to
traces,
then
you
can
correlate
to
low
successful,
because
logs
are
also
stamped
with
the
correlation
information.
So
I
should
be
I
mean
I
did
not
enable
Loki
or
any
other
loading
backend,
but
I
would
have
been
able
to
query
with
the
loading
backend
as
well.
So
you
can
see
all
the
logs
which
were
recorded
in
that
same
context
and
when
that
metric
was
recorded.
G
So
that's
the
crypto
model.
The
main
reason
I
wanted
to
show
this
is
It's
like
none
of
the
main
APM
vendors,
including
my
own
employer,
is
supporting
exemplars
at
this
point.
So
there
is
not
a
lot
of
interest
in
shipping.
This
feature
so
I'm
trying
to
like
demonstrate
features.
So
people
know
what
this
feature
is.
What
are
the
scenarios
which
it
can
enable
source
to
just
like
build
education
and
help
people
understand
the
value
we
need
eventually
with
the
goal
of
like
shipping.
G
This
part
of
the
SDK
and
I
think
there
is
also
a
requirement
from
the
specification.
The
specification
for
xmlers
are
currently
not
stable.
The
main
reason
is
only
Java
has
implemented
it,
so
they
are
looking
for
more
languages
to
implement
so
that
they
can
mark
the
spec
a
stable
and
unfortunately,
it's
a
chicken
onion
problem,
because
many
languages
are
saying
we're
not
implementing
it
because
the
spec
is
not
stable.
So
we
have
volunteered.
At
least
I
have
say:
I
have
volunteered
to
implement
this
in.net
to
help
the
specification
move
forward.
G
So
this
demo
is
like
checked
into
the
report.
So
if
you
go
to
the
repo
under
main
hyphen
Matrix
Branch
go
to
the
docs
metric
exampler
you'll
be
able
to
see
all
the
things
which
you
need
to
run.
This
thing,
I
I
quickly
show
what
those
things
which
I,
which
I
spawned
up
so
I,
have
my
demo
application,
which
is
the
same
application
which
we
have
in
the
repo.
Then
I
have
configured
that
application
to
export
metrics
and
logs
and
traces
to
this
collector.
G
So
I
need
a
collector
running
here
and
this
collector
is
running
with
this
configuration.
So
what
that
configuration
is
the
interesting
thing
here
is
from
The
Collector
traces
are
being
exported
further
to
Tempo
and
Matrix
are
being
exposed
further
to
Prometheus
again
like
we
use
some
settings
so
that
metrics
are
sorry
one
second
yeah.
So
we
use
some
settings
here
just
to
make
sure
like
the
metrics
are
exposed
in
the
open
metric
standard,
because
that's
a
format
which
Community
is
understands
which
can
support
exemplars.
G
So
the
X
component
is
like
Tempo,
which
is
I,
mean
Tempo.
Here
you
don't
need
anything
special
it.
The
tempo
itself
is
just
a
store
for
places,
it
doesn't
know
anything
about
exam
class
and
then
obviously
Prometheus,
which
stores
a
matrix
and
I
had
to
like
enable
some
extra
flag
so
that
it
Prometheus
supports
Exemplar.
But
it's
an
opening
thing
so
I
had
to
enable
these
flags
to
enable
Prometheus
to
support
this.
And
finally,
in
grafana
we
have
some
settings,
enabled
the
navigation
which
I
showed
like
navigating
from
one
signal
to
another.
G
So
how
I'm
studying
grafana
with
this
configuration,
which
has
enabled
both
Prometheus
and
tempo
as
two
data
sources
with
this,
is
the
magic
which
allows
us
to
navigate
from
xmlr
in
Prometheus
to
temporal
using
the
key
Trace
ID.
So
these
are
like
all
the
things
I
mean
if
you
are
trying
it
out
the
docker
file
and
all
the
configure
part
of
the
report
itself.
So
you
just
run
it
get
a
feel
of
how
things
would
look
like
and
like.
G
If
you
have
any
questions,
you
can
come
back
to
this
thing
or
even
the
specification,
because
the
specification
is
still
not
stable.
So
you
we
may
have
like
some
like
rough
things
which
are
not
yet
figured
out
in
terms
of
implementation,
like
I,
do
have
like
in
the
main
Matrix
Branch,
because
what
I
showed
was
like
real
that
it's
not
like
dummy
data.
G
However,
the
implementation
is
like
very
limited
I
only
added
like
in
X
and
plus
into
histogram,
because
histogram
each
one
metric
where
xmlers
can
be
of
like
most
valuable,
can
be
used
elsewhere,
but
it
it's
most
valuable
in
histograms.
So
that's
why
I
picked
histogram
to
start
with.
So
right
now
I'm
in
the
process
of
like
evaluating
performance,
it's
in
general.
The
performance
is
going
to
be
like
little
bit
slower
when
you
enable
X
and
plus,
because
we
have
to
do
like
a
lot
more
processing.
G
So
you
have
to
store
exemplars
serialize
it.
You
have
to
copy
things
from
the
pretty
much
all
the
things
which
we
do
in
the
Matrix
TK.
We
have
to
do
it
like,
like
repeatedly,
depending
on
whether
an
Exemplar
is
going
to
be
stored
or
not,
but
those
are
implementation.
Details
hopefully
like
once
the
implementation
is
done,
we'll
do
a
walkthrough
of
the
implementation
detail
as
well.
So
today,
I'll
just
focus
I
just
focused
on
how
the
end-to-end
story
looks
like
once.
G
The
whole
thing
is
done:
yeah,
that's
pretty
much
all
I
wanted
to
share.
Anyone
has
questions
or
if
you
want
to
contribute,
features
or
anything
to
excel
class
like
reach
out
to
me.
There.
I
G
Much
like
to
contribute
right
now,
because
I'm
figuring
out,
where
to
put
or
where
are
the
right
entry
points
to
attach
xmlers
into.
But
if
there
are
more
people
who
are
willing
to
help,
we
should
be
able
to
find
work
streams
so
like
what
can
happen
in
the
parallel.
G
Its
employers
I
mean
I'm,
not
an
expert
but
I
hope
I
can
answer
in
basic
questions.
If
not,
we
can
go
back
to
the
topic
synthetic
which
I
think
we
are
almost.
I
G
Not
have
anything
left
in
the
agenda,
so
that's
all
I
had
planned
if
no
other
topics,
let's
end
a
few
minutes
early
and
see
you
all
next
week
so
by
next
week,
maybe
two
three
days
after
the
next
meeting
we
expect
to
ship
the
one
road
Force
table.
D
The
only
other
point
I
wanted
to
bring
up
was
timing
of
the
meetings
we
taught
last
week
about
the
fact
that
there
was
switching
time
zones
because
of
somebody
else.
D
Yeah
two
little
things
because
of
somebody
else,
I'm
interested
to
know
you
know:
do
we
have
options
that
are
yes.
D
G
Yeah
definitely
do
that,
because
the
reason
why
we
did
this
automating
was
want
to
accommodate
like
people
from
Asia
and
then
one
of
the
medianists
who
is
no
longer
with
the
like
this
community.
So
he
used
to
be
requesting
specifically
for
4
PMI.
So
that's
why
we
have
the
four
pair
one.
So
if
we
do.
G
D
D
It's
more
at
the
moment,
I
think
if
you're
doing
the
earlier
one
and
nobody
for
now,
if
the
early
one
isn't
needed
for
anybody
in
particular,
I
can't
make
that
one
anyway,
but
if
it
was
at
midnight,
then
I
could
make
it
I
might
not
make
all
of
them,
because
it
is
midnight
my
time
but
I'm
at
least
I
at
least
don't
have
a
clash
other
than
clashing
with
sleep
so
but
yeah.
If
we
can
work
on
some
other
times.
That
would
be
amazing.
For
me,
okay,.
G
Yeah
I
think
if
you
propose
like
few
timings,
which
you
are
comfortable,
let's
see
if
others
are
okay
with
that,
and
then
we
can
do
a
quick
sharing,
because
there
are
many
six
who
change
their
timelines.
I
mean
meeting
time
based
on
other
people
preferences,
so
yeah.
We
should
be
able.
D
To
do
that,
but
yeah
I'm
not
all
asking
people
to
change,
but
you
know
at
the
very
least
if
we
can
kill
the
the
earlier
meeting
the
boys
at
11AM.
Your
time.
G
The
only
challenge
is
like
yeah:
we
just
have
to
make
sure
the
calendar
has
free
slots,
because
we
have
a
limit
on
like
the
number
of
Zoom
calls.
So
as
long
as
it's
not
clashing
with
any
of
this
thing,
and
as
long
as
it's
not
conflicting,
the
other
maintenance
as
well
I
should
be
usually
free,
because
I
have
the
most
free
calendar
of
all
yeah
yeah.
Can
you
can
you
like
put
that
into
the
like
the
slack
channel,
so
everyone
can
take
a
look
at
it
and
respond
if
you're?
Okay
with
it.
D
C
D
On
being
here
every
week,
I
am
planning
on
being
part
of
this
for
the
long
term,
so.
G
Foreign
first
thing,
the
first
part,
which
is
to
propose
moving
all
the
meetings
to
the
four
o'clock
one.
So
I
can
do
that
part
and
see
if
everyone
agrees.
The
second
part
about
like
totally
changing
the
time.
Please
propose
some
timings
and
then
we
can
vote
in
the
slack
Channel
and
based
on
that,
we
can
pick
a
alternate
time.
E
Absolutely
awesome:
okay,.
G
Yeah
anything
else
to
be
discussed
so
Martin
like
you,
are
you
like
planning
to
update
the
I
o
docs
or?
Is
that
something
like
you
are
counting
on
Carter,
because
I
did
see
some
PRS.
D
Happy
to
take
that
on
as
a
task
between
me
and
Philip,
so
me
and
Philip
work
together
so
but
yeah
me
and
Philip
will
do
the
the
main
I
o
docs
around.
You
know
I'm
happy
now
that
we're
this
close,
so
I
will
I
will
update
based
on
what
we
have.
I
D
What
I've
done
so
far
I
mean
it's
useful
for
you
to
know
Blanche,
but
I
went
through
and
tried
to
rebuild
our
distro
that
we
have,
which
is
the
convenience
distro
for
honeycomb,
based
off
1.4
notice,
that
the
ad
exporter
had
been
removed
and
I've
made
it
work
and
it
works
pretty
well
with
the
new
configure
methods
and
stuff
like
that,
just
basing
everything
on
the
dependency
injection
everything
just
seems
to
work.
D
I'm,
not.
You
know,
I'm
disappointed
that
we
missed
the
the
convenience
method,
but
other
than
that.
It's
all
pretty
nice,
so
I'm,
not
that
worried
about
it.
So
yeah
all
good.
For
me.
F
Okay,
yeah.
F
D
We
want
to
add
back
in
and
create
a
issue
for
that,
so
that
we
can
discuss
bringing
them
back
in,
but
I
said,
I'll
have
a
chat
with
you
offline,
because
I'm
assuming
there's
reasons
why
some
of
them
were
removed.
Yeah.
F
Sounds
great
yeah
yeah
thanks,
everyone
I
think.
G
Yeah
so
next
week,
like
like
I
mentioned
at
the
beginning,
we
hope
to
do
like
some
demo
of
something.
So
if
you
have
like
you're
working
on
something
you
want
to
demote
to
the
community,
you
need
like
just
let
us
know
in
dislike
or
like
message:
they
will
arrange
it
next
week.
Alan
will
be
doing
exponential
histogram
again,
it
will
be
very
quick
demo.
So,
if
you're
interested,
please
make
sure
you
attend
that
or
watch
the
record.