►
From YouTube: 2021-05-28 meeting
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
A
I
haven't
been
able
to
finalize
my
my
pr
and
send
that
out,
but
I
I'll
try
to
catch
up
later
today,
and
so
I
want
to
come
back
and
talk
about
the
the
schedule
and
and
see
the
progress
on
the
prototype.
I've
seen
some
like
progress
on
the
dominant
side
and
also,
I
I
think,
on
the
slack
channel
I've
seen
people
asking
the
java
part
so
want
to
understand.
What's
what's
the
current
progress,
let
me
share
my
desktop.
A
A
Okay,
we
can
start
so
so
one
thing
that
I
missed
is
I
I
was
unable
to
join
the
tuesday
spec
meeting,
so
it
seems
the
proportion
of
changing
the
the
timeline
was
not
discussed,
or
at
least
I'm
not
seeing
any
like
feedback
from
from
the
dog
I
shared.
So
did
anyone
join
the
tuesday
meeting
and
know?
What's
the
the
follow-up
there.
C
Let's
see,
I
joined
the
meeting
the
question:
this
is
the
first
question.
Nobody
really
objected,
I
think
carlos
read
it
feeling
a
little
uncomfortable
and
josh
has
a
pr
that
he's
put
in
well
actually
never
mind
I
the
answer
is,
it
was
read
out
and
nobody
said
much
at
all.
A
A
We
we
can
tell
people
like
we're
like
this
is
the
last
notice,
if
you
see
blocker,
just
called
otherwise
we're
going
like
I'm
going
to
change
the
schedule
and
I'm
working
to
execute
on
that,
and
I
have
a
question
for
for
the
tca
members.
I
think
josh
you're,
probably
the
only
tc
member
here
I
haven't
seen
bulk,
then
so
that
way
yeah.
So
do
you
have
any
visibility
on?
What's
going
to
be
the
next
spike
release.
C
A
Okay,
so
with
that,
I
I
want
to
see
what
are
the
things
that
you
believe
in
order
to
get
the
apis
like
released
as
experimental.
So
currently,
if
you
look
at
the
spec,
it's
putting
some
scary
thing
and
say:
hey
like
we're
like
doing
major
changes,
and
please
don't
do
anything
unless
we
communicate.
So
I
I
think
for
experiment.
The
reason
we
have
experimental
here
is
because
we
don't
have
a
definition
of
what's
before
experimental.
So
currently
it's
a
draft
version.
A
So
what
I
think
makes
sense
is
I
want
to
remove
this
entire
section,
so
we
only
have
status
experiment
and
remove
this
and
we
communicate
and
we
communicate
in
a
way
that
we
think
it's
ready
for
many
languages
to
do
more
experiment,
and
we
don't
expect
a
big
surprise
like
we're
going
to
scrub
everything
here
and-
and
this
is
what
I
want
to
do
in
number
one-
and
I
think
number
two
is:
I
want
to
catch
up
and
get
the
decision
tree
ready.
A
I
think
we
already
discussed
in
the
previous
pr
when
I
put
the
skeleton.
I
just
need
time
to
document
that,
and
that
should
be
fairly
straightforward,
and
I
want
that
because
without
that
I
I
think
people
reading
the
spec,
they
might
find
fun
a
lot
of
questions
and
the
time
I
spent
to
explain
that
probably
would
be
more
than
just
taking
the
time
to
document
that.
So
I'm
very
motivated
to
do
that,
and
the
third
thing
is
to
clarify
what
async
and
sync
means.
A
I
think
people
here
like
most
of
you
already
understand
that,
but
for
people
who
are
new,
especially
when
they
come
with
a
sync
async
like
in
the
maltese
writing
or
the
concurrent
programming
thing,
they
might
get
lost.
So
so
those
are
the
three
things
I
I
think
I
should
do
before
we
we
catch
up
the
release
trend
and-
and
I
think
it's
reasonable-
that
I
can
finish
the
two
pr's
very
quickly
like
today
or
tomorrow,
and-
and
I
don't
expect
a
big
surprise.
A
A
Yeah
yeah-
I'm
I'm
not
asking
for
anyone
here
to
write
this.
So
basically
those
three
items
are
small
things
and
I
I've
already
committed.
I
just
need
time
so
I'm
willing
to
catch
up
and
and
get
them
ready
before
and
now
this
friday
and
next
monday
it
will
be
holiday
site.
I
don't
expect
you
guys
to
review,
but
coming
back
on
tuesday,
we
can.
We
can
quickly
review
this
and
this
night
we're
ready
to
catch
the
release,
train
or
not.
It's
unfair.
A
John,
I
I
saw
you
also
agreed,
so
I
I
assume,
like
people,
are
okay.
So
as
long
as
I
can
solve
these
three
problems,
we're
ready
for
experimental
release,
okay,
so
that
means
I
I
might
I
might
prioritize
for
this.
I
I
have
some
draft
thing
here,
just
about
the
view
stuff,
so
probably
in
other
parts
of
this
I
I'll
work
on
the
view
after
I
finish
these
three
pr's
and
okay.
So
the
next
topic
is
about
the
the
prototype
update.
A
So
I
know
victor,
you
have
done
several
things
in
the
donut,
so
so
how
about
that?
You
give
a
update
to
other
folks.
B
Sure
so
we
have
been
so
the.net
cjo
has
branched
out,
basically
the
metrics
portion
in
our
github
and
we've
been
doing
stuff
inside
that
branch
called
the
metrics
branch,
and
in
that
branch
we
have,
I
guess,
experimental
versions
of
the
api,
as
well
as
an
experimental
version
of
the
sdk
and
soon
we
know
that
the
net
team
has
also
been
working
and
has
gotten
their
net
api
approved
through
the
dot
net
community,
so
they're
in
the
midst
of
basically
pr
in
the
last
minute
of
changes
and
checking
that
in.
B
But
we
have
a
net
package
that
we
can
use
and
we
are
going
to
soon
be
able
to
flip
to
using
that,
and
we
could
actually,
for
example,
do
let's
say
the
grocery
example
that
riley
had
put
up
earlier
and
we're
going
to
try
that
with
it
with
an
sdk
experimental
sdk.
B
So
the
experimental
sdk
is
trying
to
follow
the
spec,
as
is
so.
We
are
basically
at
a
point
where
we're
following
the
pipelining
spec
that
riley
has
put
in
together
we're
kind
of
ready
to
try
and
experiment
with
the
view
aspect
as
soon
as
he
has
that
ready
as
well,
and
currently
the
export
is
really
nothing
more.
The
aggregator
and
export
is
really
nothing
more
than
just
output
to
the
command
line,
and
that's
it
so.
A
Okay,
thanks
victor,
so
so
it
seems
like
I
I'm
I'm
your
blogger
and
I
I
should
work
on
the
view
pr
at
least
to
unblock
your
work
and,
and
you
have
done
the
the
other
part
like
currently
you.
So
I
wonder,
based
on
your
your
prototype,
have
you
figured
out?
How
do
we
support
multiple
pipelines
like
how
do
we
support
one
yeah.
B
A
Like,
for
example,
I
have
a
meter
provider
and
I
I
have
two
meters
and
one
is
sending
something
like
temperature.
Another
one
is
sending
like
the
humidity
and
I
want
temperature
to
get
processed
in
one
way
and
sent
to
a
push
exporter.
While
the
humidity
is
using
a
pool
exporter.
So
those
things
have
you
figured
out
a
way.
If,
yes,
would
you
be
willing
to
take
part
of
the
sdk
spike
work,
because
I
have
the
skeleton
there?
We
can
divide
and
conquer
yeah.
B
Yeah,
so
so
so
number
one
I'm
happy
to
help
out
where
we're
needed
there
and
number
two.
We
were
hoping
that
the
view
api
will
actually
help
us
identify
how
to
potentially
branch
out
for
different,
potentially
export
frequency
per
se.
So
so
the
quick
answer
to
that
is
no.
I
don't
think
we
really.
B
You
know
figured
that
out.
We
know
that
there's
two
frequency
we
are
currently
trying
to.
We
just
have
specked
out
to
say:
hey,
we
have
measurements
coming
in
and
we
need
to
do
some
conversion
to
metrics
and
we
have
some
basic
aggregator,
but
the
actual
you
know
fine
details
of
how
to
convert
an
aggregated
value
into
a
metric
and
what
the
identity
of
a
metric
that
we
haven't
gotten
to
really
yet.
B
But
one
of
the
big
questions
there
is:
how
do
we
do
the
different
pipeline
because
we
have
at
we
know
of
at
least
two
frequency
right,
so
we
have
a
frequency
of
collecting
observations
and
then
we
have
frequency
of
exporting
and-
and
we
don't
know
right
now,
when
we
do
the
export
do
we
need
to
do
something
about
re-aggregating
the
things
that
we've
aggregated
when
we
do
collection?
So
there's
a
bunch
of
questions.
B
A
Sorry,
sorry
for
rambling,
so
so,
if
you
see
those
questions,
I
encourage
you
to
either
take
part
of
the
spike
and
say
hey.
I
want
to
work
on
this
section.
I
already
have
some
idea
in
some
pr
or
if
you
think
you
don't
have
a
good
idea,
and
you
want
to
rely
on
me
to
to
drive
this
file
an
issue
so
create
a
spec
issue
and
describe
what
like,
where
you're
blocked
and
what
you
expect
and
we
can.
We
can
do
the
group
triage.
A
B
Yeah
so
I'm
happy
to
take
on
whatever
portions
needed
as
necessary.
I
will
continue
to
you
know,
push
forward
on
trying
to
do
a
experimental
version
of
this
with
multiple
frequency
for.
A
Export
yeah,
so
if
you
see
any
blocker
or
any
part
of
the
spec
is
missing,
just
create
an
issue
on
the
on
the
spec
ripple
and
tag
them
with
matrix
and
api
sdk
and
okay,
you
just
assign,
if
you
don't
I'll
I'll
change
the
issue
every
one
or
two
days
so
like
I'll,
bring
them
together
here
for
people
to
discuss
and
split
the
work.
B
A
B
D
We
have
neither
progress
nor
plan,
I
mean
at
this
point.
We
obviously
have
the
old
api
and
sdk
working.
I
mean
as
well
as
it's
ever
worked,
but
we
have
no.
We
have
no
plan
or
no
product
no
progress
to
report
on
anything,
that's
new.
I
am
studiously
waiting
until
something
is
actually
released
in
the
spec
because
having
been
bitten
too
many
times
now,
but
moving
target
like
I'm,
not
starting
anything
until
there's
something
like
released.
A
Yeah,
you
already
gave
a
lot
of
like
guidance
and
feedback
here
so
yeah
you
started
like
far
far
ago,
so
so
just
questions
so
like
you
have
a
strong
willing
to
push
the
spike
to
get
get
to
the
stable
state
as
early
as
possible.
So
I
want
to
have
a
clear
expectation
from
you.
A
So,
for
example,
if
we
release
the
spike
as
experimental
and
we
remove
all
the
scary
things
here,
do
you
think
that
gives
you
this
enough
signal
or
motivation
that
you
can
have
a
little
bit
of
confidence
and
work
on
the
metrics
or
you're
busy
with
other
stuff
and,
if
not,
then,
for
example,
we're
making
this
to
like
feature
freeze.
Would
that
be
the
signal
so.
D
So
I
think
the
feature
freeze
would
definitely
be
a
good
signal,
but
also
I'm
working
on
other
stuff.
I
don't
really
have
time
to
be
working
on
this
right
now,
so
if
we,
if
there
were
someone
who
could
come
in
and
help
do
that
in
the
java
system,
that
would
be
very,
very
helpful
because
I
don't
have
time
to
I
don't
know.
I
just
don't
have
time
to
work
on
it.
D
Splunk
is
having
me
work
on
other
stuff
right
now
so,
and
I
will
bring
this
up
at
our
next
sig
meeting,
which
well,
I
guess,
I'm
meeting
with
at
least
one
maintainer
in
a
couple
hours,
so
yeah
I'll
chat
with
anarag,
but
I
think
I
think
so
far
the
three
maintainers
don't
have
time
to
work
on
it.
So
we're
happy,
I
think,
we'd
be
happy
to
review
from
someone
else
did,
but
we
don't
have
time
to
be
building
it
ourselves.
Right
now,.
A
What's
the
expectation
for
the
spike
group
like,
for
example,
people
can
say:
hey
we're
not
going
to
do
anything
unless
you,
you
move
the
spec
too
stable
or
all
start
once
you
have
experimental
or
feature
freeze
and
then
what's
the
what's,
the
like
the
possible
level
of
investment
like
who's
going
to
work
on
this,
and
so
we
can
probably
figure
out
like
how
we
can
how
we
can
push
this
forward
or
the
answer
could
be
hey
like
java
is,
is
super
busy
with
other
stuff
and
we're
unfunded.
A
D
D
He
has
also
hasn't
been
at
the
sig
meetings,
so
I
think
he's
also
been
working
on
other
things,
so
our
sig
meetings
are
very,
very
sparsely
attended.
At
this
point
I
mean,
I
think,
there's
like
usually
three
people,
so
very
few
people
are
interested
in
working
on
java
lots
of
people
like
to
use
it
but
nobody's
interested
in
working
on
moving
things
forward,
except
on
instrumentation
and
there's
a
lot
of
lots
and
lots
of
work
in
instrumentation,
but
we
do
not
have
really
anybody.
D
Well,
it's
a
little
different
than
that,
because
we
have.
We
have
a
1.0
sdk
that
is
very
solid
and
being
used
extensively
and
is
tons
and
tons
of
instrumentation
is
being
written
against
it,
like
the
instrumentation
project,
extremely
active,
no
problem
with
no
problem
with
contributions
on
that
side.
It's
just
there's
very
few
people
are
interested
in
working
on
like
the
apis
and
sdk
itself.
They
want
to
write
instrumentation.
D
I
see
okay
good
to
know
so,
and
I
was
waiting
to
kind
of
bring
this
up
at
the
maintainers
meeting
until
it
was
clear
that
we
had
a
go
on
metrics,
so
I've
been
I've
been
holding
off
and
waiting,
because
there's
no
reason
to
get
people
involved.
If
there's
something
to
work
on
right
and
originally
joss
earth
was
like,
he
started
working
on
some
prototyping
stuff
and
then
he
hasn't
touched
anything
for
a
long
time.
So.
A
Yeah
he
he's
very
busy
with
his
other
stuff,
like
he
will
spend
a
lot
of
time
on
c,
plus,
plus
and
java,
and
but
now
he
probably
got
bitten
by
other
parties.
I
think,
like
google,
microsoft,
those
big
companies
probably
have
some
similar
business
rhythm
and
in
this
in
these
two
weeks,
I
I
wouldn't
expect
him
to
be
able
to
work
on
this.
A
This
is
like
the
like.
The
two
sides,
like
the
spike,
has
to
be
there
to
help
the
ic
to
move
forward
and
the
sdk
or
the
language
six
has
to
move
forward
to
give
enough
feedback
for
the
spike.
Otherwise,
we'll
we'll
just
do
that
for
one
language,
and
we
won't
be
able
to
have
that
confidence
that
the
spec
is
going
to
work.
So
so
thanks
john
and
I'm
moving
to
python,
I
I'm
not
sure
if
we
have
python
like
maintainers
or
provers
here.
C
I
could
speak
that
I
know
diego
from
lightstep
has
been
doing
a
prototype,
he's
been
looking
at
spec
and
reading
it
carefully,
and
I
see
his
sort
of
reports
every
couple
days,
but
I
don't
know
anymore
or
the
state
of
like
what
branch
he's
using
or
anything
like
that.
He
has
approached
me
with
some
questions,
about
conceptual
models
and,
and
I've
answered
some
questions
for
him,
so
I
know
he's
doing
doing
it.
The
work,
I'm
not
sure
I
have
anything
more.
I
can
answer,
though.
Okay.
A
Yeah,
thank
you
that
that
also
aligned
with
my
understanding,
so
he
reached
out
to
me
asking
a
lot
of
questions,
but
so
far
I
think
I
answered
all
of
them
at
least
from
the
api
part,
and
so
far
he
seems
to
be
happy.
So
my
understanding
is
he's
not
blocked
and
on
the
ick.
I
know
we're
we're
still
a
little
bit
far
from
from
where
he
can
really
do
the
like.
A
lot
of
work.
C
Yeah
and
I'll
be
able
to
help
when
when
he
does
begin,
I
just
responded
to
your
earlier
statement.
I
think
we've
always
had
this
problem
where
there's
not
quite
enough
attention
or
developers,
actually
writing
metric
sdks
here,
and
I
think
a
lot
of
it's
had
to
do
with
just
tracing
isn't
ready.
Yet
so
java
can't
get
anyone's
attention
on
metrics
while
tracing's
not
quite
finished,
I
think
is
what
I
heard
and
I
think
it's
the
same
with
the
other
languages
that
are
not
making
progress
on
metrics.
A
C
A
Lisa
is
not
reporting
to
me
now,
so
I
have
to
find
something
to
motivate
him
or
you
guys
can
try
to
motivate
him
as
well
I'll
I'll
poke
him
and
also
for
for
ghosts.
My
understanding
from
josh
is
like
you.
You
have
the
willing
to
work
on
this,
but
at
least
not
in
the
next
couple
weeks
or
months,
because
you
have
other
priority
right.
C
Yes,
lightstep
has
I
mean
I
I've
always
got
this
tc
commitment,
which
I'm
always
falling
behind
on
and
this
work
on
the
data
model.
I've
been
trying
to
help
as
much
as
I
can
so
hoping
that
that's
kind
of
wrapping
up
a
bit
and
that
we
don't
need
to
be
writing
specs
and
stuff
as
much.
I
do
really
hope
to
be
involved
in
the
go
sdk
since
I,
since
I'm
proud
of
what
we
have
there.
It
just
has
a
lot
of
renaming
to
be
done
and
then,
of
course,
there's
questions
about.
C
It
has
a
batch
api.
Should
we
remove
it,
and
you
know
that
kind
of
question,
but
lightstep
has
formally
decided
to
make
make
a
project.
That's
going
to
be
called
the
open,
telemetry
project
and
I'll
be
on
that
and
we
hope
to
make
sure
metrics
happens
this
year.
Okay,.
A
And-
and
you
guys
saw
my
my
post
on
the
spike
and
the
maintainers
channel
where
github
like
I-
I
worked
with
them
and
and
they
are
now
citing
open
telemetry
as
their
direction
and
they
actually
contribute
to
open
time
to
goal
and
and
ruby.
So
currently,
I
think
they're
making
a
lot
of
contribution
ruby,
but
not
a
lot
on
goal,
but
I
believe
they're
motivated.
So
if,
if
you
can
give
some
guidance
and
help
them,
I
I
guess
I
can.
I
can
probably
motivate
some
developers
from
from
from
the
github
team.
C
That
sounds
good.
We
have
a
light.
Stepper
working
on
the
hotel
go
to
help
them
get
to
the
1.0.
Essentially,
gustavo
is
doing
that
and
once
he's
done,
I
think
he's
going
to
come
step
back
and
I
might
step
in
at
that
point.
A
A
Okay,
so
that
covers,
and
just
before
we
move
to
jonathan's
question
anyone
know
like
any
other
language,
six
that
want
to
make
progress,
because
I
I
remember
I
heard
from
ruby
like
in
the
spec
meeting
when
I
asked
hey,
we
need
some
language
to
work
with
this
bag
say
to
with
a
prototype
I
heard
from
ruby,
but
I
never
heard
like
they're
they're,
saying,
hey
riley,
I'm
going
to
reach
out
to
you
offline,
but
then
I
never
heard
about
like
heard
from
that.
A
To
I
don't:
okay,
so
I'll
I'll
poke
the
ruby
folks
and-
and
I
believe,
if
we
can,
we
can
catch
the
release
trend
that
probably
will
give
more
motivation.
Okay,
so
we're
going
to
move
to
jonathan
your
turn.
E
Yeah
I
have
a
couple
of
questions
about
like
the
api
and
the
timeline
asl.
The
first
one
is
only
apigee,
so
the
current
ac
instruments.
They
say
that
the
callback
function
is
responsible
for
returning
the
measurements
and
the
measurement
object
is
basically
all
the
measurement
entity
that
consists
of
two
things.
One
is
the
raw
value,
the
measurement
itself
itself
and
the
other
one
is
the
attributes
or
tags.
E
So
does
this
mean
that
users,
when
they
want
to
have
a,
I
don't
know,
a
asynchronous
counter?
E
Next
to
their
value
function
or
the
value
provider
function,
they
always
like
need
to
wrap
this
into
another
one
which
will
basically
report
a
measurement
instead
of
a
real
value.
E
Not
necessarily,
I
think,
what
can
we
or
can
we
figure
out
something
that
will
basically
ease
this
so
that
david
said
hey.
This
is
the
static
list
of
tags
that
I
always
want
to
provide,
or
this
is
my
I
don't
know
the
function
that
we
provide
the
tags
or
supplier,
that
we
provide
the
tags,
and
this
is
my
value
function
or
something
that
will
basically
make
this
like
less
boilerplate.
A
I
I
see
so
so
number
one.
I
I
think
we
give
certain
flexibility,
we're
saying
certain
languages
can
decide
that
they
want
to
return
a
list
of
the
measurements
or
they
can
have
some
like
placeholder
object
and
use
that
to
call
some
function
to
report
individual
data
and
the
only
suggestion
here
we
put
is
we
don't
want
people
to
report
duplicate
values
in
a
single
invocation
callback
like
the
callback
regarding
your
question
so
jonathan,
I
I
think
it's
very
similar.
It's
not
specific
about
the
asynchronous
one
right.
A
E
And
it
is
related,
yes
yeah,
so
that
that
is
a
connected
problem,
but
I
believe
it
is
like
less
apparent
if
you
are
using
a
synchronous
instrument.
If
it's
synchronous,
then
you
can
basically
just
store
these
static
things
in
a
variable
and
just
throw
it
in.
But
if
it
is
asynchronous,
then
you
have
a
function
provided
by.
I
don't
know
a
third-party
library
or
or
whatever
language
sdk
itself,
and
you
need
to
wrap
this
function
into
another
function,
which
will
like
provide
the
text
and
that
will
make
it
a
little
bit
complicated.
E
Yeah,
so
that
you
don't
so
what
you
what
you
are
doing
here,
let's
say
that,
like
in
the
indy
python
example
like
in
the
pf
callback,
let's
say
that
you
are
getting
those
values:
eight,
three,
seven,
seven,
one
zero
something
from
a
function
which
is
provided
by
a
library
to
you.
In
that
case,
you
need
to
wrap
this
callback
right.
E
Sorry,
you
wrap
that
function
into
a
callback
which
will
basically
like
call
the
function
that
is
provided
by
the
sdk,
which
will
provided
the
value
and
you
attach
the
the
attributes.
So
you
know
always
need
to
do
this.
This
wrapping.
My
question
is:
can
we
like
give
some
helper
that
you
don't
need
to
do
this?
You
just
need
to
hey.
I
want
to
like
use
this
function
to
to
report
the
value
and
by
the
way,
these
are
my
tags.
A
Yeah,
I
guess
this
can
be
built
on
top
of
the
this
functionality.
So
exactly
yes,
so
let
me
ask
you
this
question:
do
you
think
this
is
a
performance
consideration?
You
think
we
want
to
give
people
something
that
they
don't
have
to
repeatedly
give
us
the
same
combination
of
attributes-
or
this
is
more
a
convenience.
Ask.
E
A
Cool,
so
my
suggestion
would
be
this:
like
anything,
convenient
api
can
be
added
on
top
of
using
api.
We
look
for
trading
like
we
also
have
a
workstream
originally
we're
trying
to
make
some
convenient
api
for
the
trade
tracer
like
tracer
provider
setup.
But
later
it
is
okay
number
one.
This
is
not
going
to
block
the
core
api,
because
convenient
api
can
always
build
on
top
of
things.
Unless
you're
saying
this
is
a
performance
issue
which
is
totally
different
right,
you
have
to
build
it
and
number
two
is
convenient
api.
A
We
we
probably
need
to
get
some
feedback
from
the
customers
instead
of
us
trying
to
imagine
things
so
so
I
I
suggest
that
you
create
a
an
item
like
github
issue
on
the
spike
repo
track,
the
scenario
and
you
can
assign
to
me
and
make
the
similar
tagging
I
already
put
in
the
in
this
one
and
describe
the
scenario,
and
we
can
try
to
get
the
feedback,
but
I
number
one.
I
don't
think
this
blocking
issue
number
two.
I
think
convenience.
We
need
more
evidence
because
it's
a
little
bit
subjective.
A
I
want
to
see
like
whether
there's
enough
interest
and
and
number
three,
I
think
this
convenient
api
can
be
always
added
at
any
time
without
having
to
break
the
the
underlying
layers
so
similar
like
tracing.
I
would
say
we
release
the
tracing
as
stable
and
we
can
come
back
at
additional
convenient
api
without
having
to
get
blocked.
So
so
this
is
a
nice
suggestion.
A
It's
a
it's.
A
relatively
lower
part,
compare
comparing
to
our
current
things
in
the
plate
and
and
we
should
gather
feedback
from
the
like
the
broader
audience,
one
issue
and
assigned
to
me
I'll
I'll
put
it.
I
like
after
we
have
the
experimental
thing
I
would
expect
more
language
sex
would
be
motivated
or
we'll
be
less
scared
to
make
progress,
and
we
can
leverage
that
to
to
see
if
we
can
gather
those
feedback,
cool.
E
Okay,
my
next
question
is
like
my
all-time
favorite.
The
timer
really
have
one
for
the
for
the
initial
version
of
the
api
and
how
does
it
fit
into
the
into
the
timer?
If
you
will
have.
A
Yeah,
so
it
seems
to
me
like,
based
on
my
understanding,
I
I
think
this
is
my
personal
belief
that
timer
is
not
a
critical
thing.
It's
a
it's
a
convenience
thing
and
it
can
be
added,
and
and
this
it's
a
very
common
scenario
and
from
my
read
from
bolton,
it
seems
he
he
doesn't
have
a
strong
saying
that
that
we
must
have
timer.
It
seems
john,
probably
coming
from
the
micrometer
background
or
the
java
background
like
like
he
mentioned
the
timer
several
times,
and
especially
when
he
mentioned
you
want
to
time
something.
A
C
I'm
moderately
a
fan
of
it.
I
think
it's
a
good
plan.
I
mean
the
most
of
our
instrumentation
packages.
Have
problems
with
clocks
which
clock
you're
going
to
use?
How
much
precision
does
it
have
what's
the
type
it
returns
and
asking
the
user
to
do
it
right
just
invites
mistakes?
I
think
that's
the
reason
why
it
is,
though,
I
think
just
like
somatic
sugar
is
what
people
say
about.
It's
like
just
a
wrapper
around
something
like
a
histogram,
but
you
know
the
units
and
you
don't
ever
make
a
mistake.
A
Yeah
so
my
challenge
with
the
with
timers
in
some
languages.
I
know
we
have
the
time
span
concept,
so
people
don't
have
to
put
the
unit
and
they're
always
safe,
because
people
always
mess
up.
If
you
just
give
them
a
number
they're
going
to
pick
different
units
and
for
some
languages
they
just
don't
have
that.
So
if
we
try
to
enforce
the
time
concept,
I
I
don't
know.
For
example,
if
you
ask
me
like
cc,
plus
plus,
I
know
there
there's
state
time,
they're
also
like
some
time
like
structure,
so
it
seems
to
make
sense.
A
E
To
me,
like
the
the
question
is
more
like
less
about
like
the
the
units
and
more
about
like
how
can
you
measure
duration,
and
that
is
usually,
I
believe,
it's
it's.
It
must
be
easy,
but
I
think
most
of
the
people
they
are
struggling
with
it,
and
I
have
seen
this
a
lot
of
time.
Then
people
are
using
the
wall
clock
not
not
a.
C
E
Yeah,
and
by
the
way
like
then
the
so
we
had
a
python
example
and
also
a
java
example
for
the
timer
just
fyi
both
are
wrong.
They
are
using
the
whole
clock.
So
yes,
even
the
example
that
we
that
that
that
were
given
to
the
users
are
has
this
this
very
issue
and
I
believe
the
best
so
the
biggest
benefit
of
having
a
timer
is
this
that
we
will
basically
remove
this.
This
huge
pit
hole
from
the
way
of
the
users.
A
E
The
this
specification
repo
contains
a
issue
and
also
a
github
discussion.
I
believe
about
the
timer.
A
Yeah,
so
the
the
thing
I'm
like
the
reason,
I'm
I'm
a
little
bit
hesitant
about
time,
is
because
in
c
parts
past
I've
seen
I've
seen
a
lot
of
discussion.
Whenever
people
talk
about
the
world
clock
about
monotony
of
the
clock
and
when
it
comes
to
like
ntp
or
just
the
time,
it's
just
hard,
and
my
worry
is.
If
we
start
that
topic,
we
probably
will
take
forever.
B
C
A
B
A
B
Have
a
instrument,
yet
you,
you
said
riley,
that
this
is
a
convenience.
So
if
it's
a
convenience,
then
that
implies
we
could
just
write
like
a
best
practice
guide
to
say:
hey
if
someone's
looking
at
doing
some
timer.
This
is
what
you
would
do
to
map
it
to
one
of
our.
You
know
up
down
counters
or
a
gauge
or
whatever
it
is,
and
you
know
so,
maybe
maybe
it's
more
of
that
if
we
know
that.
A
Yeah
and,
for
example,
like
in
download
questions
like
hey,
this
is
a
like
a
function
signature,
and
can
you
just
change
that
parameter
to
a
time
span,
instead
of
putting
something
like
an
integer
and
put
a
comment,
this
is
milliseconds
and
I've
seen
a
lot
of
issues.
So
currently,
we've
been
very
successful
on
pushing
back
on
that,
because
people
both
time
spent
there.
A
There
are
many
other
issues
that
he
introduced
so
for
this
one
I
I
guess
we
probably
can
take
two
languages
to
do
the
prototype
like
if,
if
java
already
have
micrometer
or
a
lot
of
examples,
I
believe
we
can
do
that
and
and
for
for
downloading,
do
an
experiment
and
see
that
that's
worth
enough
victor.
What
is
my
worry
is
like
we
start
the
spec.
Without
the
prototype,
we
probably
don't
have
a
clear
understanding.
What
what
are
the
scenarios,
because
here
it
seems
like.
E
I
I
have
another
scenarios,
so
I
I
have
another
example:
why,
like
a
timer
like
other
than
being
able
to
to
use
a
always
monotonic
clock
and
then
and
get
the
cat,
get
the
measuring
the
duration
right?
It
is
testing.
E
So
when
you
want
to
test
like
this
scenario
that,
for
example,
you
are
measuring
your
own
code
or
you
are
measuring
like
a
a
third-party
code-
and
you
want
to
test
this
using
the
system,
clock
will
will
give
you
a
lot
of
struggles
in
your
tests.
But
if
you
have
a,
I
don't
know
like
a
clock,
object
or
something
that
you
can
mock
or
fake.
Then
in
your
tests
you
can
basically
simulate
different
scenarios
and
it
without
the
timer.
E
Basically,
the
users
need
to
write
that
as
well
for
themselves
and
I'm
not
sure
anyone
who
wants
to
do
that,
though
it's
it's
very,
very
simple,
but
usually
people
don't
do
this.
This
extra
step.
D
D
That's
provided
in
our
test
artifact,
but
if
there
are
people
who
are
going
to
want
to
write
instrumentation
and
want
to
be
able
to
test
what
the
instrumentation
is
doing,
what
it's
supposed
to
do
are
they
going
to
also
need
to
have
a
clock,
abstraction
or
some
sort
of
clock
interface
or
something
in
the
api
that
they
can
then
use
for
testing
for
that
sort
of
testing
purpose.
D
It's
a
question.
I
don't
know
don't
know
the
answer,
but
it
does
make
me
wonder
a
little
bit.
E
Either
in
the
api
or
or
like
well,
I
believe
the
only
way
to
like
enforce
it
in
the
sdk
is
putting
it
into
the
api,
but.
E
Be
the
that
would
be
the
the
goal
that
you
have
clock:
abstraction
which
can
give
you
the
four
clock
or
a
synchronous
or
sorry
like
a
monotonic
clock
and
in
your
tests.
You
can
basically
fake
that
that
method,
which
will
which
will
provide
you
with
a
monotonic
time.
D
A
It's
very
tricky
I
I
know
that
for
for
tracing
like
we
have
the
similar
thing,
but
for
tracing
we
never
mentioned
the
the
tracer
sdk
has
to
give
people
of
like
a
fake
timer
or
something
like
they
can.
They
can
inject
their
own
timer,
so
so
I
figure
jonathan
can
probably
tell
like
why
I'm
a
little
bit
hesitant
to
do
this.
I
figure
this
is
some
place
there
like
whenever
it
comes
with
time
or
some
basic
things.
There
are
a
lot
of
questions
and
people
have
strong
opinions.
A
A
B
C
That
we
have
a
units
field,
so
I
think
we'd
just
use
the
normal
mechanism
to
declare
units.
I
really
think
the
reason
why
jonathan's
asking
this
is
that
there
are
lots
of
ways
of
in
which
measuring
time
is
hard
for
myself
and
go.
I
needed
to
write
a
special
helper
library
so
that
I
could
have
a
one-liner.
I
want
a
one-liner
timing
expression
and
I
can't
do
it
even
with
a
span.
I
need
two
lines
of
code
because
of
the
context
variable
so
to
get
one-liner.
I
had
to
do
something
special.
A
Yeah,
so
so,
currently,
I'm
I'm,
I'm
still
very
hesitant
to
to
put
this
in
our
scope,
but
jonathan,
if
you,
if
you
do
see
a
value
and
you
want
to
push
this
forward.
My
suggestion
is
you
clarify
the
scenario
and
we
can
we
can
work
together
to
to
define
a
scope.
I
I
think
most
likely
we're
going
to
say
there
are
a
lot
of
interesting
topics
and
questions
here.
A
We
see
the
value,
but
these
are
the
three
or
two
things
we're
going
to
do
and
the
other
things
are
not
in
the
scope,
and
we
can
have
a
very
clear
statement
like
that
we
can.
We
can
try
to
come
up
with
the
estimation
like
how
much
effort
we
need
and
based
on
that
we
can
determine
the
time
everyone
have
here.
Are
we
going
to
make
it
or
not?
A
And
if
it's
going
to
risk
an
entire
thing,
then
I
think
it
makes
sense
to
make
that
an
additive
change
so
because
you
guys
already
see
like
in
the
in
the
epi
spike.
I
think
tigran
added
something,
and-
and
he
mentioned
this
one
since
one
that
means
like
if
people
use
the
previous
version
or
they
stick
with
spike
version
1.0,
then
they
don't
have
to
do
that.
I
believe
timer
can
also
take
this
approach.
It
doesn't
have
to
be
in
the
first
release,
but
if
we
can
make
it
awesome,
we
cannot.
A
A
Qualification,
I
would
prefer
use
old
tab,
because
I
want
that
old
tab
to
be
the
single
place
where
we
have
all
the
scenarios
and
then
people
have
question
like
hey
like
I
read
the
api
spike
and
something
is
not
clear.
I
hope
it
makes
our
job
easier
by
telling
them
hey.
If
you
want
to
have
a
concrete
scenario,
you
go
to
this
old
tab,
it
gives
you
the
entrance
and
you
can
even
build
when
you
build
the
api
sdk,
you
follow
the
spike.
A
You
got
a
little
bit
stuck
as
long
as
you
follow
the
old
tab.
It
can
give
you
an
idea.
What
are
we
trying
to
achieve
so
imagine
put
something
here
like
a
library.
Developer
is
struggling
because
they
they
struggle
with
the
wall,
clock
versus
the
monotonic
clock
and
we-
and
he
just
want
to
have
one
single
way,
but
he
doesn't
have
to
even
worry
about
the
timing.
The
api
will
just
take
care
of
the
timing
for
him
by
default.
A
A
The
original
timeline,
in
my
mind,
is
I
want
to
get
it
done
in
this
week,
but
looks
like
I.
I
got
screwed
up
by
other
stuff,
so
so
I'll
I'll
catch
up
after
I
finish
the
these
three
things,
so
I
would
say
next
week
and
hint
I
will
say
right
after
view,
so
I
believe
hint
is
almost
the
same
as
the
view
except
for
is
part
of
the
api
and
for
hint.
A
I
think
the
difference
is,
if
you
get
give
certain
hint
sdk
start
to
use
that,
and
I
think
it's
saying
I
don't
understand
that
hint
there
might
be
a
versioning
issue
or
there
might
be
vendor
specific
thing.
I
think
that
isdk
is
okay
to
ignore
that,
but
I
think
he
cannot
ignore
the
thing
people
put
on
the
video.
For
example,
if
you
put
something
in
the
in
the
video
saying,
hey
I'm
going
to
take
a
counter
instrument,
and
I
don't
want
to
report
a
counter.
I
want
to
report
something
else.
A
I
give
me
a
different
aggregation
or
I
want
this
dimension.
I
want
this
attribute,
but
the
ick
is
saying:
no.
That
instrument
does
not
have
that
attribute
at
all
like
or
that
aggregation
is
not
something.
I'm
aware
of.
I
have
never
seen
that.
I
think
the
ick
should
give
the
user
error
in
in
certain
way,
but
if
it's
a
hint
right,
I
think
they
probably
will
decide.
Okay
I'll
just
ignore
that,
for
example,
people
can
have
a
counter
in
their
library,
hdb
library
they
want
to
give
a
hint.
A
I
want
to
use
this
awesome
aggregation
or
the
sketch
and
icky
just
decided
due
to
the
memory
limitation.
I
cannot
have
this
support
or
I
don't
support
this
algorithm,
so
I'm
going
to
give
you
the
fallback
behavior,
so
I
want
to
give
them
some
flexibility
there.
So
the
ict
might
decide
to
fill
the
user
if
the
hint
cannot
be
fulfilled
or
it
can
try
to
try
to
satisfy
the
hint
as
much
as
possible,
but
not
try
to
100,
like
coverage,
make
sense.
E
A
Yeah,
if
you
want
to
know
something
like
it's,
it's
from
open,
sensors,
okay,
cool,
an
open
system
is
used
by
a
lot
of
folks,
especially
the
gcp
customers
in
production.
So
so
at
least
you
know
it's
working,
end-to-end
and
one
interesting
thing
in
open
sensors
is
the
the
api
is
just
one
api.
You
report
arbitrary.
Well,
they
don't
have
semantics
there
and-
and
the
view
is
the
the
thing
that
gives
all
the
semantic-
and
I
I
think
bogdan
probably
took
the
learning
there
and
he
started
to
prefer
that
api
should
have
some
strong
semantic.
E
Okay,
and
both
of
these,
like
the
same
timeline,
will
apply
to
these
as
as
anything
else,
for
example,
the
the
instruments
api.
Sorry,
what's
your
so
right
now
we
have.
We
have
the
proposals
that
the
that
the
the
api
will
be
in
feature
freeze
in
end
of
july.
I
guess
so
the
hint
api
and
the
view
sdk
has
this
same
restrictions
or
timeline
right,
yeah,
okay,.
A
D
Speaking
of
timelines
in
java
jonathan,
you
you're
a
java
person
and
interested
in
metrics,
would
you
like
to
help
build
it.
E
D
A
Yeah
so
far,
I
I
think,
john,
that's
a
perfect
question,
so
I
asked
jonathan
the
similar
thing.
My
understanding
is,
he
has
a
priority
to
make
sure
like
the
like
the
product
like
spring
and
those
products.
They
can
use
open,
telemetry
api
in
the
future
and
and
he's
working
on
those
projects,
but
I
believe
jonathan,
you
might
know
some
java
developers
who
who
might
want
to
come
and
help
out
here
on
the
I
mean
on
the
core.
Isdk
doesn't
have
to
be
matrix.
A
Cool
yeah,
so
I
I
probably
can
get
that
github
folks,
helping
out
on
goal
and
ruby,
but
for
java
I
I
think
microsoft
is
in
a
world
spot
where
I
don't
know,
there's
a
team
working
on
java,
but
they're
working
on
the
service
hosting
those
stuff
nobody's.
On
the
isd
case.
Well,
I
mean
we've
got
drafts
right.
D
A
Task
is
spending
a
lot
of
time
on
the
auto
instrumentation.
Oh
yeah,
okay
time
is
quite
limited.
Maybe
maybe
you
can.
E
Reach
out,
like
I,
as
far
as
I
remember,
like
microsoft,
is
like
enlarging
their
their
java
team
and
and
on
the
open
source
side
right
now,
and
I
believe
the
leader
of
the
team
is
julian
dubois,
I'm.
E
Pronunciation
is
right,
but
you
can
you
can
talk
to
him.
I
guess.