►
From YouTube: 2022-06-13 meeting
Description
Open Telemetry Meeting 1's Personal Meeting Room
B
Hello,
how
are
you
today,
I'm
good
and
good
yourself,
I'm
I'm
doing
great
sun
is
shining
had
a
good
weekend.
Good
relaxing
things
are
feeling
great.
So
you
know
overall
feeling
good
about
things.
A
Cool
pierre.
B
I
am
out
of
canada
right
next
to
detroit
michigan
actually
and
super
excited
for
cubecon
this
year,
because
I
could
literally
drive
to
it
every
single
day,
but
it's
still
going
to
take
a
hotel
only
because
I
have
to
cross
the
border
to
to
do
it
and
crossing
the
border.
B
There's
time
and
weight
and
stuff
like
that.
That
I'd
like
to
avoid.
B
It
would,
I
would
be
crossing
either
the
detroit
windsor
tunnel
or
the
ambassador
bridge
to
get
the
cube
caught,
but
I've
because
twitter
was
telling
me
hotels
are
in
short
supply.
I
actually
booked
a
hotel
over
the
weekend
instead
really.
B
I
don't
know
I
tech
twitter
people
are
saying
hotels
are
in
short
supply.
I
was
like
really.
B
That's
what
I
thought
as
well.
I
was
just
like
you
know,
whatever
I'm
just
going
to
book,
and
you
know
if
y'all
want
to
tour
I'll,
show
you
where
to
hang
out
at
for
dinner
and
stuff
like
that,
it'll
be
all
walking
distance
which
will
be
nice
or
scooters
if
you're,
I'm
so
inclined
to
take
a
scooter
around.
C
B
C
C
C
Trying
to
get
back
in
the
u.s,
no
that
that's
where
they
put
all
the
real
jerks
I
was
trying
to
get
back
into.
I
went
up
to
speak
at
some
place
in
saguanae
up
in
quebec
like
a
couple
years
ago
and
going
through
there's
like
ways
they
wave
you
through.
It's
like
hey,
welcome,
eh
have
some
maple
syrup,
speak
frenchy,
and
then
I
come
back
and
like
guys,
kidded
out
in
body
armor
and
assault
rifles
and
dogs
and
asking
questions,
and
it's
like
oh
yeah
you're,
not
joking
about
any
of
this.
Are
you.
B
As
someone
who
lives
in
that
border
town,
I
have
no
comment.
I
concur
very
much
of
all
your
statements,
but
no
comment.
C
No
coupon
and
borders.
D
Okay
cool:
we
should
probably
cancel
our
our
meeting
next
week,
just
since
that's
on
hotel
community.
A
C
C
D
Yeah
definitely
definitely
useful
they're
kind
of
expensive
they're
250.
So
that's
a
good
deal
definitely
take
austin
up
on
that.
Okay,
so
free
the
agenda.
So
I
just
have
a
short
community
update
and
then
I
already
said
no
meeting
for
next
week
and
then
we
can
kind
of
have
open
items
giuliano.
I
know
james
has
a
couple
like
development
principles.
He
wants
to
talk
about,
but
it
seems
like
I
I
know
last
week
he
had
a
holiday.
He
wasn't
able
to
join
again.
D
So
do
would
you
like
to
cover
those
again
or
we
just
like
to
keep
pushing
that
off
a
little
bit
yeah?
I.
A
I
I
would
rather
keep
bushy
okay,
I
pinged
him
in
today,
but
he
just
came
back
from
holiday,
so
he
was
kind
of
busy
but
yeah
I'll
try
to
get
him
in
the
call
just
to
disclose
those
points
because
yeah,
it's
not
my
not
my
area
of
expertise,.
D
Yeah,
no
totally
understandable,
okay
and
so
for
the
quick
community
update.
So
I've
been
thinking,
we
need
a
bit
more
approvers
and
maintainers
on
a
project
just
to
ensure
that
pull
requests
are
getting
reviews
consistently
and
also
that
the
code
is
getting
merged.
If
someone
goes
on
vacation,
no
issues
are
occurring,
so
I
asked
if
pierre
would
potentially
be
interested
in
being
in
a
maintainer,
but
I
also
wanted
to
run
it
by
the
group
first.
So
I
think
pierre
has
done
a
great
job
on
the
honeycomb
demo.
D
You
know
clearly
has
a
great
understanding
of
how
it
works.
I'm
gonna
also
bring
a
lot
of
value
to
the
maintainer
team,
but
of
course
this
isn't
a
dictatorship.
So
so
what
do
y'all
think.
D
Welcome
pierre,
we
really
appreciate
your
support
and,
if
anyone's
either,
I
guess
watching
this
recording
or
also
live
and
wants
to
be
an
approver
or
something
like
that
or
also
a
maintainer
feel
free
to
reach
out
to
me
and
we'll
go
ahead
and
get
that
conversation
started.
We
love
the
support
and
help
in
any
way.
We
can.
A
Just
a
quick
question
carter.
Sorry,
just
just
quick
question:
if
someone
is
watching
like
me
and
and
that
doesn't
know
the
difference
between
an
approver
and
a
maintainer,
what
would
be
the
the
main
differences
there.
D
C
C
C
C
Chat,
there's
like
specific
things
about
it
that
I
think
armin
just
posted
but
yeah
broadly.
C
There's
a
higher
expectation
of
like
how
much
time
you
should
be
able
to
put
into
it.
I
do
think
having
an
extra
maintainer
or
two
is
gonna,
be
good.
I
really
think
our
one
of
our
bigger,
so
I'm
just
let
me
look
real
quick,
more
approvers
would
all
would
actually,
I
think,
unblock
us
as
well,
because
really
there's
effectively
only
two
maintainers
right
now,
that's
myself
and
giuliano
right,
no
carter.
A
C
But
if
someone
wanted
to
be
an
approver
or
specifically
that
wanted
to
help
like
do
more
pm
sort
of
stuff
on
this,
I'm
trying
to,
but
it's
a
lot
to
pm
this
and
also
the
com
sig
and
also
do
my
normal
job.
So
if
someone,
if
anyone
was
interested
in
being
in
a
mostly
for
the
recording,
but
if
you're
listening
to
this-
and
you
want
to
just
kind
of
help
with
issue
triage
and
sort
of
helping
the
project,
you
know
the
work
of
the
work
and
get
done
more
smoothly.
E
C
Yeah
that'd
be
great
yeah,
just
let's
talk
about
it
on
slack
and
we
can
yeah
do
some
buttons.
D
Awesome
thanks
miko
yeah.
I
was
thinking
about
there's
this
one
pretty
active.
I
think
chinese
developer
on
github
too,
on
the
demo
so
far.
I
think
about
asking
him
to
be
an
improver
just
because
I
would
get
more
geos
involved
as
well,
which
I
thought
might
be
valuable.
D
Okay.
So
let's
talk
maybe
about
kind
of
release.
Time
frame,
I
know
we're
not
like
operating
on
a
hard
release.
Cadence
like
we
didn't
necessarily
would
be
in
an
enterprise
I
was
thinking
probably
getting
like
complete
trace
coverage
was
my
initial
thought
of
maybe
where
we
should
place
most
of
our
emphasis,
but
I
know
we
don't
even
have
like
metric
support
today,
which
is
also
kind
of
considered
one
of
the
two
key
pillars.
D
C
C
D
C
D
And
cubecon
is
in
october
right
right.
C
C
If
I
was
a
betting
man,
I
would
say
that
it's
going
to
take
till
cubecon
before
a
lot
of
the
metrics
sdks
are
like
super
ready
like
across
a
broad
spectrum
of
languages.
So
maybe,
if
we
think
about
our
first,
you
know
kind
of
releases
being
like
okay.
You've
got
all
the
new
services.
You've
got
all
the
service
swaps
in
you've
got
the
feature
flag
so
that
we
can
do
some
fault
injection.
C
It
deploys
the
kubernetes
and
then
it's
getting
metrics
from
kubernetes
and
it's
also
tracing
kubernetes.
If
you're
on
124,
then.
D
D
I
know
gary
said
he
wouldn't
be
able
to
look
at
the
rust
example
for
a
month
or
so,
but
I'm
not
expecting
that
to
like
last
until
september,
and
I
think
php
and
ruby
are
in
progress
and
then
wallet
on
c
plus
plus
shouldn't
have
an
issue
knocking
something
out
so
yeah,
I
would
say,
probably
sometime
in
august,
we'll
have
all
the
services,
if
not
july,
just
depending
on
how
things
shake
out.
Yeah.
C
Because
at
some
point,
this
switches
from
being
like
at
some
point,
the
kind
of
the
the
work
switches
a
little
bit
from
being
just
like
okay,
we
need
to
do
these
big
replacements
to
we
need
to
go
through
and
be
like
kind
of
tiki
tacky
like
okay.
What
is
the
actual
telemetry
supposed
to
look
like?
What's
the
model
here?
The
other
thing
that
I
would
probably
prioritize,
for
you
know
v1
or
for
kubecon
at
least
is
the
front
end
work
with
at
least
rebranding
it
a
little.
C
A
Quick
questions
yeah
go
ahead.
Sorry.
B
B
C
And
it's
pretty
easy
to
do
it's
just
it's
not
super
interesting.
I
don't
think
there's
a
reason
why
we
couldn't
actually
go
and
do
that
as
part
of
this,
but.
D
C
B
B
Do
we
expect
to
rebrand
the
the
demo
protofile,
because
it's
hipster
shopping
all
the
traces
everywhere.
A
C
I
think,
as
long
as
it
should
oh
yeah,
I'm
that
to
me
is
a
clean
up
thing
for
a
later
and
would
only
really
become
super
important
when
we,
I
think,
actually
become
very
important
in
one
at
the
point
where
we
do
a
front
like
we
kind
of
rework
the
front
end
and
have
a
spa
or
something
so
that
we
can
because
my
mind,
if
you're
going
to
do
it,
then
you
need
to
actually
like
kind
of
go
hog
wild
and
have
like
web
workers,
and
you
need
to
do
all
that
happy
horse.
So.
B
C
Yeah,
like
we
can
change
it
when
we
get
like
the
name,
the
the
coat
of
paint
can
come
first,
you
know
like
nobody's
gonna,
it's
not
like
we're
trying
to
obfuscate
where
this
came
from
or
anything.
I
think
we
just.
I
think
my
aesthetic
senses
would
be
pleased
if
we
could
show
this
and
it
would
looked
very
open,
telemetry
ish
or
it
was
something
that
was
like
visually
distinct
as
being
something
that
the
community
that
the
hotel
community
made
versus
like
we
just
filed
the
serial
numbers
off
of
hipster
shop.
D
F
F
Yeah
it
was
like
it
was
pretty
straightforward
to
do
those
changes
and
I
don't
think
it'd
be
an
issue
so
once
we
have
just
the
one
version
of
the
profile
that
gets
copied
into
all
the
services
by
the
by
the
docker
compose,
I
don't
think
changing
like
the
hipster
shop
name
in
there
would
be
like
much
of
a
change
some
languages.
I
think
they
do
use
it
like
c-sharp.
I
believe
I
had
to
change
like
one
of
the
name
spaces
or
something,
but
it
was
like
very
minor.
C
C
Because
I
think
that
if
we
had,
if
we
could
go
to
the
program
committee
and
be
like
hey
look,
we
want
five
minutes
or
whatever
to
show
like
here's.
What
open
telemetry
should
be,
or
here
here's
how
it's
like
hey.
You
want
to
get
and
start
with
telemetry
here's
this
demo
app.
You
know
like
that's
something
that
we
could
probably
sneak
into
a
keynote
slot
or
part
of
a
keynote
slot.
D
B
D
D
Are
we
saying
that
all
of
these
requirements
should
be
kind
of
a
v1
release?
I
know
we
already
kind
of
called
out
that,
like
intra
process
context,
propagation,
probably
isn't
a
v1
requirement.
Are
there
any
others
where
we
think
you
know
that's
not
necessary
to
have
initially
or
maybe
even
not
ready
at
the
moment,
I'll
zoom
in
a
little
bit
too,
and
when
we
say
automatic
instrumentation,
this
would
be
from
the
sidecar
collector
correct,
awesome.
A
The
collector
will
only
work
when
we,
the
sidecar,
will
only
work
when
we
have
the
kubernetes
approach.
Yeah.
C
Yeah
yeah,
no
I'm
thinking
so
like
so,
for
example,
if
for
a
java
service,
then
the
preferred
way
to
initialize
from
a
demo
perspective.
In
my
mind,
the
preferred
way
to
like
initialize
hotel
in
a
java
process
or
a
java
service
is
to
use
the
java
agent
right
like
for
go.
There
is
no
agent,
so
this
would
be
not
useful
for
go.
D
Okay
should
definitely
maybe
call
out
those
where
they're
unnecessary
but
I'll
go
ahead
and
just
check
this
off
for
now
and
then
maybe.
B
B
Yeah
that
you
could
just
roll
with
the
the
python
wrapper.
C
F
C
C
D
C
C
D
C
Capture
the
current
context
and
extend,
I
mean
that
should
be
pretty
straightforward
in
every
language,
because
the
co
you're,
always
assuming
everything,
is
right.
You're
always
going
to
have
the
span
context
started
by
the
grpc,
client
or
grpc
server,
in
whatever
the
grpc
context
is,
and
though
the
grpc
contexts.
C
Have
to
be
passed
into
their
handler,
so
in
the
handler
you
can
get
the
context
out
and
then,
if
you,
I
guess,
wanted
to
put
it
there
or
you
wanted
to
like
pass
it
down
again.
You
would
need
to
extend
the
context
yeah,
but
I
started
to
see
more
people
having
questions
like
on
slack
about
using
hotel
context
in
general.
B
So
you
mean
things
like
just
get
spam
from
context
would
be
what
we
you're,
calling
that
there
right.
C
Yeah,
it
gets
banned
from
context
and
then
proper
context
handling
so
that
you
don't
like
clobber
the
existing
span
and
properly
closing
the
new
span
and
da
da
da.
B
The
front
end,
we
pulled
the
current
span
from
context
inside
the
grpc
handler,
so
we
could
add
more
attributes
to
it.
So
that's
why
I
was
asking
that
question
specifically.
You
know
what
did
we
call
that
trace
dot
span
from
context.
C
C
I
actually
don't
have
an
opinion
but
in
my
mind,
there's
two
routes:
there's
do
what
you
did
pierre
right,
which
is
take
the
grab
the
current
span
in
the
context
and
then
inject
additional
attributes
in
it
or
grab
that
span
context
to
create
a
new
span.
To
encapsulate
the
new
work
being
done
and
then
add
those
attributes
to
that.
B
C
B
C
C
C
B
Yeah
yeah
for
sure
we
could
spin
up
new
spans
and
wrap
each
handler
with
an
additional
spam,
but
they
kind
of
get
one
from
automatic
instrumentation
and
I
feel
like
we're
kind
of
just
repeating
ourselves.
After
a
while
yeah.
C
I
mean,
I
would
say
the
the
only
reason
you
would
ever
need
to
do.
That,
though,
is
if
you
had
to
do
well,
you
had
to
like
drill
down
context
and
you
didn't
have
a
context
object
already,
but
you
could
always
I'm
in
languages
where
you
don't
have
like
conventional,
like
if
you're
in
something
other
than
go
right.
You
don't
have
a
convention
for
this.
C
So
the
handler
function,
the
handler
function,
has
my
request
context
and
that's
where
the
auto
instrumentation
span
is
living
and
then
I
call
another
method
and
I
don't
pass
anything.
I
don't
pass
a
context
argument
to
that
child
function.
I
just
call
it
one
of
our
arguments,
then
I
would
have
to
rewrite
the
method
signature
to
include
the
context
or
I
would
need
to
just
create
a
new
argument
for
the
span
contacts
to
go
in.
I
think
I
don't
know
what
the.
B
Sure
I
it's,
you
could
argue
that
went
both
ways
yeah
other
languages,
though
it's
different.
I
think
I
want
to
show
both
the
different
approaches.
B
B
C
B
As
you
can
see,
especially
like,
I
think
it's
the
checkout
one
or
the
place
order
handler,
that
would
probably
make
more
sense
where
we
could
spin
up
a
few
more
spans
on
there,
because
it
does
a
little.
C
C
D
B
D
B
If
we're
looking
for
hey,
let's,
let's
stick
to
best
practices
as
well
as
showing
as
many
different
examples.
We
can
check
out
services
right
for
spinning
up
spans
and
it's
in
go
so
we'll
have
both
examples
covered
very
well
of
adding
in
context
and
spinning
up
new
ones.
The
other
languages
are
ripe
for
both
as
well,
where
all
the
other
services
would
be
right
for
all
of
them
as
well.
B
So
if
the
goal
here
is
to
show
the
example
in
both
doing
it
both
ways
in
per
per
language,
we
would
be
able
to
cover
that.
I
think
that
kind
of
pulling.
C
Catalog
is
going
to
get
some
new
stuff
anyway,
because
it's
going
to
it's
going
to
read
from
the
data
store
from
the
product
store.
B
C
Right
yeah:
well,
we
have
every
language.
I
thought
right
now.
C
C
I
mean,
I
think
it's
I
think
yeah
so
check
out
is
the
one
that
does
a
lot
of
the
more
interesting
stuff.
So
that
would
probably
be
where
I
would
put
more
of
the
custom
things
in.
I
can
see
like
a
v2
here
where
front
end
gets
renamed
a
router,
and
then
we
have
like
more
explicit
clients
that
are
in
browser
js
or
are
running
in
some
weird
scenario.
C
D
Then
going
back
to
the
issue
real
quick,
I
think
the
only
thing
two
things
we
haven't
really
talked
about
in
terms
of
requirements
for
baggage
and
then
trace,
linking,
I
think
off
the
top
of
my
head.
I
don't
remember
what
trace
linking
is
to
be
honest
links.
C
So
like
one
could
be
like
a
security
boundary
like
trusted
or
untrusted
to
trusted
so
like
if
we
had
a
good
example,
would
actually
be
if,
when
we
come
back
in
and
we
do
like
an
actual
front
end
with
like
kind
of
deeper
instrumentation-
and
you
have
a
you-
know
more
fully
feature
a
fleshed
out.
Front-End
trace
then,
rather
than
making
the
back
end
work
a
child
of
that,
then
the
back
end,
work
that
was
requested
by
the
front
end
would
become
trace
links
the
others.
D
C
Instead
of
creating
explicit
parent-child
relationships
between
all
of
those
between
the
consumer
side
or
multiple
consumers,
then
I
use
trace
links
instead,
so
that
each
of
my
explicit
producer,
each
explicit
action
of
a
consumer
is
kind
of
its
own
trace,
linked
back
to
whatever,
I
think
is
the
the
right,
for
you
know
the
right
place
right
so
do
I
write
something
once
and
then
read
it
five
times
through
workers
and
do
stuff,
then?
C
Okay,
so
I
have
my
producer
trace
and
then
my
consumers
all
link
back
to
that
one
in
terms
of
actually
viewing
these
trace
links.
I
think
honeycomb
might
be
the
only
place
that
lets
you
do
that
right
now,
I'm
not
sure
if
jager
supports
it
or
actually
no
jager
does
support
trace
links.
I
don't
remember
how
it
visualizes
them,
but
they
jager
does
support
trace
links
but
yeah.
It's
really.
D
D
B
I
I
could
do
a
thing
in
the
front
end,
so
everything
coming
in,
we
would
just
kill
the
trace
or
and
we'd
start
a
new
trace
instead
and
and
that
new
trace
would
be
linked
to
the
existing
one.
But
that.
B
C
This
would
probably
if
we
were
going
to
do
that,
I
would
actually
say
that
should
be
feature
flagged,
because
I
don't
know
how
other
analysis
systems
would
work
with
it.
I
think
that
would
make
a
lot
of
sense
to
feature
flag,
though
I
think
it
makes
sense
to
have
it
and
I
think
it
would
make
sense
to
put
it
behind
a
flag
so
that
people,
because
what
I
don't
know
what
I
don't
want
to.
C
What
I
would
rather
avoid,
is
getting
a
situation
where
we're
trying
to
kind
of
sell
this
to
the
to
all
of
our
wonderful
employers
and
run
a
situation
where,
because
you
know,
we've
put
trace
links
at
sort
of
a
really
important
part
of
the
system,
the
traces
all
look
weird,
because
our
back
ends
don't
understand,
trace
links.
B
C
D
Okay,
cool
and
I
think
the
the
last
item
is
baggage
and
of
course,
I
think
we
want
to
be
fairly
careful
here-
maybe
not
doing
it
in
every
language,
since
we're
just
constantly
going
to
be
adding
a
bunch
of
attributes
to
every
trace.
We're
worried
about
that.
If
we
just
add
like
11
baggage
attributes,
essentially.
C
B
Yeah,
I
think
we
were
just
going
to
pop
in
session
id.
C
C
D
C
B
C
C
Yeah,
that
should
be
pretty
trivial
and
yeah
that
flows
everywhere,
but
you
know
we're
only
adding
two
things
so.
D
Yeah,
let's
look
at
this
real,
quick
baggage
just
being
set
and
read
appropriately:
okay,
yeah
that
shouldn't
be
an
issue
then
telemetry
isn't
being
sampled
or
filtered.
I
think
that's
already
kind
of
just
a
completed
requirement
throughout
the
app.
I
think
I
removed
any
explicit
sampling
inventions,
so
we
should
just
be
able
to
set
that
at
the
collector
and
move
on
there
unless
anybody
implements
it
in
the
meantime.
D
Okay.
Well,
I
think
we're
directly
at
time,
but
we've
made
some
good
progress.
I
think
we
can
maybe
spend
our
next
week.
Maybe
reviewing
the
metric
space
and
talking
about
how
advanced
you
want
those
capabilities
to
be
in
time
for
kubecon
or
what
feature
set
we're
looking
for.
But
besides
that,
I
think
we're
in
a
good
spot
right
now
and
I'll
open
up
a
pr
to
add
miko
and
pierre
as
an
approver
and
maintainer
respectively,
and
then.
C
D
Probably
release
a
blog
post
sometime
to
announce
the
release
of,
or
you
know,
kind
of
the
start
of
the
community
demo
effort.
Would
we
potentially
want
to
do
that
before
the
community
day
or
like
around
the.
D
D
Okay,
cool
well,
thanks
for
joining
everyone
feel
free,
like
I
said
anyone
watching,
recording
to
reach
out
to
me
or
austin
or
whoever,
if
you're
interested
in
helping
out
as
an
improver
or
a
maintainer.
We
love
this
system,
but
otherwise
have
a
lovely
day.