►
From YouTube: 2022-12-08 meeting
Description
Instrumentation: Messaging
A
B
How's,
the
weather
down
there.
A
A
Know
is
that
normal
for
y'all,
not
usually
this
warm
yeah
I
mean
usually
there's
a
couple
Sprints
of
like
high
60s,
maybe
into
the
70s.
Maybe,
but
we
got
a
about
a
week
where
it
was
in
the
70s
most
days
and
it
didn't
get
below
like
65.
wow.
It's.
B
A
Yep,
it
happens,
how
about
you
how's
it
up.
There.
B
I
think
it's
a
little
warmer,
not
80,
but
it's
like
in
the
40s
now
and
rainy.
So
pretty
pretty
standard
for
up
here.
Pretty.
A
Gonna
go
see
my
folks
in
Virginia
in
like
a
week
so
I'm
like
oh
man,
it's
gonna,
be
it's
gonna,
be
cold
like
that
I
remember
those
days,
yeah.
A
B
The
Eastern
Seaboard
too,
like
it,
can
get
really
the
humidity
element
of
the
the
cold
really
sets
in
there.
Yep
yeah
I
used
to
live
in
New,
England
and
I.
Remember
it's
like
it's
cold
and
then
there's
like
just
brutally
cold.
A
Going
to
see
folks
yeah,
so
my
my
parents
live
up
there
so
say
hi
to
them.
Yeah.
B
I'm
in
that
category
of
that,
as
well,
cool
I
think
we're
seeing
this
past
I
think
we
have
Quorum
as
well.
I'll
start
sharing
my
screen
here.
If
you
haven't
already,
please
add
your
name
to
the
settings
list
and
if
everything
you
want
to
talk
about,
also
add
that
to
the
agenda
to
start
us
off,
I
wanted
to
kind
of
just
point
out
as
I'm
sure
it
doesn't
really
need
to
be
pointed
out,
but
this
time
of
year
a
lot
of
vacations
are
happening.
B
If
you
have
not
already,
please
update
the
vacation
calendars
for
the
hotel,
I
think
it's
like
an
Hotel
maintainers
vacation
calendar,
but
it's
been
said
that
it's
not
just
for
maintainers.
So
if
you're
an
active
contributor
in
the
open
television
space,
Please,
add
your
vacation
days
there
it's
not
required,
but
it
helps
I
think
understands
if
we
should
be
canceling
meetings
or
things
like
that,
so
yeah,
it's
be
much
appreciated
and
I
kind
of
on
with
that
I
wanted
to
ask.
B
If
there's
I
guess,
maybe
we
should
just
preemptively
look
at
like
what
we
should
be
canceling
in
the
next
few
weeks.
I
know
this
next
week,
I'm
out
Monday
through
Wednesday,
but
I
will
be
here
for
next
Thursday's
meeting
I.
Think
the
following
week
after
that
is
I.
B
Guess
Christmas
falls
on
a
fun
weekend,
yeah
or
Sunday,
but
I
didn't
know
like
I'm
guessing
the
attendance
of
the
22nd
meeting
is
probably
gonna,
be
pretty
low,
so
I
don't
know
if
we
wanted
to
cancel
that,
and
then
I
know
that
the
week
after
Christmas
the
last
week
of
the
year,
I'm
I'm
out
the
whole
week
so
that
one
maybe
also
want
to
cancel
I
think
last
year
we
did
both
of
those.
But
if
there's
any
opinions,
I'd
like
to
hear.
A
I
will
also
be
out
the
22nd
and
the
29th,
so
okay,
unless,
unless
somebody
else
wants
to
run
the
meeting,
I
would
be
up
for
canceling.
Those
too.
C
Yeah
I'm
good
with
that
I
think
I'll
be
in
the
office,
but
there's
plenty
of
other
things
to
do.
B
Yeah,
okay
and
then
David's
got
a
thumbs
up
as
well.
Okay,
cool
I
can
take
the
action
item
on
that.
B
Okay,
cool
I
think
then.
The
next
thing
I
had
on
the
agenda
was
to
go
over
the
just
PR's
and
things
to
get
us
to
the
metrics
beta.
Still
the
the
priority
I
think
here,
So.
Currently
what
we
have
in
progress,
this
unified,
metrics,
SDK,
caching,
I
guess-
has
been
put
on
the
back
burner
for
me,
but
we
could
probably
finish
this
up.
I
think
that
I
was
hesitant
to
try
to
finish
this
because
there's
another
PR's
that
are
probably
going
to
be
changing.
This
instrument
provider
code.
B
That
being
said,
it
may
make
sense
just
to
close
this
for
now,
because
it's
done
a
bunch
of
other
things
and
then
maybe
open
an
issue
to
add
caching
at
this
level
in
the
host
GA
phase
as
well.
Given
this
isn't
going
to
be
strictly
required
and
it
can
be
added
in
the
backwards
compatible
fashion.
C
B
Cool
I
will
I
think
address
that
after
the
meeting
then
next
up
was
this
Json
encoder
for
standard
metrics,
exporter,
I,
don't
know!
If
there's
any
progress
on
this,
it's
been
a
while
I.
Don't
have
any
time
to
pick
it
up
so
I'm
not
really
going
to
comment
on
this,
but
I
think
if
somebody
else
has
interest
in
this
I
do
think
that
there's
I
took
a
quick
look
at
this.
It's
not
easy.
B
No,
it's
really
not
easy.
It
may
mean
that
we
actually
may
need
to
adjust
some
of
the
structure
of
the
standard,
app
reader
or
exporter.
But
that
being
said,
I
don't
know
if
it's
I
think
that's
more
important
things
to
be
doing
for
the
beta
currently.
But
it's
it's
on
the
list.
If
you
have
thoughts
on
it,
I
think
add
a
comment
here.
Do.
C
C
B
B
So
I
wanted
to
have
a
clear
line
of
sight
on
resolving
this.
Given
our
existing
standard
addicts,
our
standard
out
exporter
for
traces
as
a
Json
feature,
as
well
as
the
previous
metrics
one
had
a
feature
for
this,
so
I
wanted
to
make
sure
we
were
able
to
implement
that
I.
Guess,
like
I
mean
I
I,
don't
know
if
this
has
to
be
complete
for
the
beta,
but
it
definitely
has
to
have
a
line
of
sight
if
we
want
to
release
the
standard
out
exporter
as
stable.
B
I
would
say
no
because
I'm,
like
99
sure
things
are
going
to
need
to
change.
Given
my
my
look
into
this
and
if
we
know
things
are
going
to
change,
it's
not
really
a
release
candidate.
B
B
And
no
and
I
mean
like
the
standard
ad
exporter,
I,
don't
even
know
if
his
is
a
requirement
of
the
specification,
it's
just
something
that
we've
had
and
it's
a
backwards
compatible
thing.
I
do
think
that,
like
having
the
ability
to
Json
and
code,
it
is
is
kind
of
a
nice
thing
for
sure.
B
Yeah
you're
right.
Sorry,
it's
one
of
us
as
I.
Looked
at
this
correct!
Yes,
so
I've
been
speaking
poorly
about
this,
then
that
it's,
it
is
I,
think
more
for
the
testing
and
the
generic
ability
to
Output
things.
So
that's
a
good
point.
B
Yeah
I
mean
the
ability
to
not
have
time
stamps.
I
think
is,
is
definitely
not
a
requirement,
however,
like
to
include
the
standard
out
metrics
exporter
in,
like
example,
tests
like
you
can't
have
it
output
time
stamps
to
be
useful
time.
I!
Guess
that's
really
where
it
comes
from
I
guess
then,
in
the
documentation
process
we
need
some
way
to
Output
metrics
in
some
way
that
are
I,
think
reliable
and
consistent
across
time.
B
B
I
I
really
like
them,
I,
think
it's
a
great
feature
but
yeah
like
if
you're
outputting,
the
metrics
that
you
generated
into
the
all-include
timestamps
they're
wrong.
Every
time
you
run
the
test
right,
so
yeah,
not
a
great
thing.
That
being
said,
like.
A
Well,
all
right
so
I've
looked
at
it
at
least
once
and
I
know,
there's
a
very
inefficient
way
that
we
can
do
this.
That
would
meet
the
requirements,
but
it's
not
something
that
you
would
want
to
use
in
production.
Of
course,
right.
A
Oh
just
taking
whatever
metrics
you
get
and
transliterating
it
into
a.
A
B
That's
yeah,
so
that's
kind
of
how
I
was
looking
at
it
as
well,
which
is
means
that,
anytime,
that
the
data
structure
changes.
The
exporter
needs
to
also
change
right,
because
otherwise
you
don't
know
where
the
time
stamps
are
and
and
I
think
that
was
like
so
and
so
then
I
was
looking
at
like
well.
B
Maybe
you
could
just
use
like
you
essentially
recreate
a
Json
encoder
right
so
dive
into
the
reflex
package
and
start
you
know
dynamically
pulling
up
types
and
then,
if
you
find
a
you
know
a
time
stamp
type,
then
replace
it
with
something.
But
you
essentially
need
to
Traverse
the
data
tree
at
that
point
and
do
exactly
what
the
Json
encoder
is
doing,
which
is
like
I
mean
it
seems
problematic,
not
only
from
like
a
performance
standpoint
which
again
like
is
not
necessarily
critical
as
I.
A
C
B
C
B
I
mean
I,
don't
know
if
that's
the
case,
I
think
it's
worth
pursuing.
Something
like
that,
like
I'm,
not
opposed
to
to
running
this
through
some
sort
of
stream
parsing
after
the
fact.
A
B
Which
one
are
we
going
to
go
with
is
what
I
was
what
I'm
saying
like,
but
do
you
really
want
to
do
that
or
do
we
want
to
take
this
and
say
like
hey
like
this
is
outputting
Json,
but
we
want
to
go
back
to
you
know
like
what
you're
saying
also
with
like
the
recreate
the
data
structures.
That's
what
we
had
before
and
we
output
it
in
our
like
own
custom
format
like
do
we
want
to
continue
to
do
that
or
do
we
want
to
keep
this
Json
coding
and
then
still
like?
A
Of
either
the
the
encoder
interface
or
or
the
metric
data
apis
right
well,.
B
I
think
it
would
that's
what
I'm
saying
it's
like
the
encoder
interface
would
change
if
you
say
like
well
we're
going
to
put
this
in
our
own
data
structures
anyways,
which
is
what
we
were
doing
so
we're
going
to
Output
it
like
as
a
straight
format.
So
our
encoder
interface
doesn't
need
to
be
there
anymore,
so
get
rid
of
it,
but
I
mean
how
would
you
do?
How
would
you
write
an
encoder
if
it
needs
to
like
encode
to
a
private
data
type
like.
A
B
Well,
yeah,
but
that's
what
I'm
saying
like
we
expose
an
option
to
provide
your
own
encoder.
But
if
a
user
provides
an
encoder
external
to
the
package,
they
need
the
internal
data
type
to
be
encoding
too
so
like
that,
doesn't
make
any
sense
anymore
to
be
providing
an
external
encoder
that
they
can't
actually
encode
to.
B
B
Okay
with
that,
but
like
I,
think
it
just
needs
to
get
thought
through
like
as
well
as
similar
with,
like
Anthony
here
like
if
we
can
viably
do
this
with
like
parsing
timestamps
here
or
regex,
then
like
yeah.
Maybe
there's
no
changes
to
the
API
that
are
needed,
but,
like
somebody
needs
to
come
in
here
and
have
like
a
plan
before
I.
Think
we
could
take
this
out
is
what
I'm
saying.
B
Currently,
fat
sheep
9146
is
assigned
to
this,
but
if
Aaron
or
Anthony
you
want
to
come
in
and
just
like
I'm
saying,
like
not
even
a
full
proof
of
concept,
but
like
just
understanding
like
the
components
and
giving
some
sort
of
design
proposal
here,
it
would
be
I
think
enough
to
to
validate
whether
this
should
be
done
or
it
needs
to
be
moved
to
different
Milestone.
Does
that
make
sense?
B
Okay,
yeah
I
mean
I'm
happy
to
take
another.
Look
at
this
as
well,
but
I've
got
some
other
things.
I
wanted
to
keep
working
on
here
in
the
beta
project
first,
and
with
that,
a
lot
of
this
has
to
do
again
with
the
API
non-compliance
issues.
So
I've
got
two
issues
with
PRS
associated
with
them
open,
although
one
of
them
is
included
in
this
Milestone
or
this
projects,
one
is
for
the
creation
of
asynchronous
instruments,
not
accepting
callbacks
and
being
able
to
define
a
callback
on
creation.
B
This
has
this
PR
here,
which
kind
of
balloons
a
little
bit
because
it
really
involves
two
things.
One
of
the
issues
with
adding
a
callback
at
this
level
was
that
it
was
scoped
by
type
based
on
the
structure
of
the
Callback
and
wanting
it
to
conform.
The
specification
allowing
restriction
on
instruments
that
it
actually
is
updating
for
that
split
all
the
configuration
values
into
the
component
packages.
I.
Imagine
if
we
in
the
future
remove
these
packages.
B
There'll
be
separate
option
types,
but
I,
don't
think
like
600
lines
of
this
is
just
changing
instrument
with
description
to
sync
float,
64,
with
the
description
kind
of
changes
in
this
which
I
yeah.
It's
a
it's
tough
to
look
at
that
number
and
then
come
in
here,
but
I
think
that
this
is
kind
of
like
the
key
there's
just
this
new
callback
type
in
this
one
is
in
the
async
464
package.
This
returns
an
864
or
I'm.
Sorry
at
float,
64.
B
value
based
on
some
sort
of
attribute
set,
and
that
is
called
indirectly.
So
essentially,
what
this
allows
is
the
SDK
to
ensure
that
the
registered
instrument
that
is
registered
for
the
singular
callback
is
only
updating
that
singular
instrument
is
the
the
purpose
here
so
yeah.
This
actually
updates.
Not
only
the
API
but
the
no
app
implementation
of
the
global
implementation
and
the
sdks
as
well.
B
Additional
to
that
there's
this
asynchronous
instruments
with
multiple
instrument:
callbacks
need
to
be
able
to
unregister.
We
talked
about
set
as
a
PR
for
this
Zone
as
well.
It's
a
little
bit.
B
Which
one
wasn't
in
the
right?
Sorry
just
trying
to
make
sure
we
have
these
in
the
right
projects.
B
Okay,
cool
sorry,
so
this
adds
a
new
registration
type.
Similarity
kind
of
like
discussed
in
this
issue
itself,
that
registration
type
is
returns
from
the
register.
Callback
method,
and
let
me
see
if
I
can
just
pull
this
up.
It's
a
little
easier
to
read:
here's
the
registration
type,
the
registration
type.
Has
this
unregister
method.
You
can
kind
of
go
into
the
issue
to
find
out
some
of
the
design
choices
of
this
API.
B
There
was
a
design
choice
to
ask
or
call
this
on
register
ER,
similar
to
other,
like
go
interface
types
given
there's
only
one
method,
however,
I
wanted
to
keep
this
a
little
bit
more
generic
in
case.
We
do
want
to
add
more
methods
to
this
in
the
future
and
give
us
some
somewhere
of
a
identifier
of
what
it's
actually
representing
so
I
called
it
registration.
B
B
B
This
is
all
handled
by
tracking
all
of
the
callbacks
that
are
registered
here
as
linked
lists
as
a
way
to
be
able
to
remove
them
and
also
not
have
an
unbounded
amount
of
memory
being
used
like
in
a
slice
keeping
indexes
of
them
and
keeping
them
ordered,
which
is
not
strictly
required,
but
I
wanted
to
try
to
make
things,
reproducible
and
so
I
didn't
want
to
keep
them
in
a
map,
so
welcome
to
suggestions
or
critiques
on
this
one
as
well,
but
yeah.
B
Those
are
the
two
things
that
I
had
currently
added
and
are
in
progress
here
in
the
metrics
beta
there's
also
a
few
more
issues
that
we
needed
to
address
for
the
metrics.
Ga
I
guess
in
add
this
to
the
list,
there's
a
lot
of
verification.
Obviously
we
should
probably
try
to
prioritize
some
of
this
stuff
because
these
issues
are
coming
from
that
verification.
B
The
asynchronous
instrument
API,
does
not
follow
the
hotel
specification
for
recommended,
naming,
doesn't
use
observable.
There
must
not
be
an
API
for
creating
instruments
other
than
a
meter.
This
is
dealing
with
our
instrument
provider
API,
the
redesign
of
the
asynchronous
instrument,
registration
API
via
the
meter.
This
is
one
I
just
added
two
days
ago.
Currently
we
take
instruments
and
and
callback
one
of
the
things
that
was
kind
of
like
identified
in
this
other
issue.
We
just
looked
at
when
we
split
out
the
instrument
configuration
I'm.
B
Sorry,
this
isn't
the
right
one
I,
don't
think.
No,
it
is
this
callback
design
that
was
added
here
was
something
that
maybe
we
could
also
try
to
look
into
here,
where
this
registration
of
a
particular
callback
type
could
be
something
where
it
returns.
Observations
and
those
observations
could
be
explicitly,
I
think
evaluated
to
be
a
part
of
the
SDK
and
the
part
of
the
meter
that
they
came
by.
Currently,
these
are
things
we're
not
doing
that
the
specification
requires
so
there's
a
proposal
here.
B
I
think
that
we
want
to
take
a
look
at
this
to
consider
this
so
I
added
it
as
well,
but
other
than
that
I
think
these
are
like
pretty
important.
These
two
are
definitely
important,
because
I
think
we
need
either
a
good
reason
for
why
we're
not
following
this
recommendation,
or
we
need
to
fix
this
API
and
consistency,
but
other
than
that
we
have
a
lot
of
verification
tasks
for
the
API.
B
We
probably
want
to
prioritize
as
well
I'm,
not
saying
it
blocks
the
beta,
but
I
am
saying
that
this
beta
may
not
mean
much.
If
we
come
back,
we
say
like
this:
API
actually
needs
to
change
again,
so
just
kind
of
pointing
that
out.
B
D
If
there
isn't
anything
else,
I
did
spend
a
little
bit
of
time
over
the
past
few
days,
working
on
the
what
it
would
look
like
to
implement.
The
bridge,
spec
and
I
had
I
was
hoping
to
get
just
some
high
level
feedback
on
two
different
approaches.
I
was
looking
at
so
I
have
those
two
proof
of
Concepts.
D
D
So
this
would
be
just
a
new
one
versus
it
is
possible
to
make
them
share,
but
I
think
there
are
some
drawbacks.
B
D
So
that's
part
of
it
and
the
other
part
is
that
there's
special
requirements
around
registering
sdks
that
aren't
that
don't
apply
to
General
producers,
which
is
that
if
you
register
an
SDK
twice,
it
shouldn't
be
double
collected
and
you're
only
allowed
to
register
a
single
SDK
with
a
reader.
So
that
makes
me
lean
towards
the
separate
interface
piece.
B
Yeah
agreed,
so
if
you
double
register
a
metric
producer
or
a
a
producer
twice,
it
can
just
double.
That's
on
you
essentially,
like
you,
screwed
up
like
you're
gonna,
get
double
yeah.
Okay,
I.
D
Definitely
not
I
was
just
trying
to
I
was
gonna
leave
a
comment
somewhere
saying:
hey,
take
a
look
at
these
interfaces,
but
I
figured.
We
could
just
discuss
it
here
as
just
as
easily
yeah.
B
B
B
B
Okay,
cool
so
I
think
with
that
we've
got
through
the
agenda.
Well,
I
I,
guess
maybe
just
to
back
up
a
little
bit
on
the
metrics
SDK
beta.
Is
there
anything
else
in
both
Center
we're
going
to
talk
about
before
we
move
on.
B
Okay,
then
yeah
I
think
there's
nothing
else
on
the
agenda.
We
can
pause
here.
David
came
up
with
a
great
thing
to
ask
up
for
any
other
issues
that
people
want
to
talk
about.
C
Just
a
quick
call
out
that
I
just
came
from
the
first
meeting
of
the
new
operator
Sig
the
open,
Telemetry
collector
operator
Sig,
and
there
was
a
bunch
of
discussion
there
about
starting
to
use
some
of
the
build
tools
that
we've
created
for
this
repo
and
The
Collector.
So
those
are
spreading.
Their
reach
is
growing
foreign.
B
B
Yeah,
that's
actually
positive.
I
I
do
notice,
there's
a
few
new
Utilities
in
there.
That
are
probably
things
we
could
probably
add
here,
one
of
them
to
change
log
generator.
It's
kind
of
cool,
still
learning
that
one
though.
C
Yeah,
that
was
that
was
one
of
the
tools
that
came
up
there
specifically
was
chain
talk
generator
and
how
to
make
sure
that
people
could
identify
what
components
were
changing.
I
think
we
could
gain
benefit
there.
Although
we
like
I,
said
to
the
to
the
operator
who
said
you
know
the
The
Collector
contributors
were,
the
real
problem
was
because
you
couldn't
go
30
seconds
without
having
to
update
your
PR
to
fix
the
change
log
conflict
but
yeah
it's
yeah
it.
C
B
Yeah
I
think,
if
there's
a
lot
of
there's
a
lot
of
value,
is
things
like
that
kind
of
on
the
collector
crossover
Anthony?
There
was
a
question
on
the
ownership
of
one
of
these
new
Auto
instrumentation
packages
that
was
being
added
to
the
contrib
repo
there's
a
user.
The
user
is
the
owner
of
the
package
trying
to
be
added
and
we
were
under
that
new
ownership
model
trying
to
add
them
into
code
owners.
B
Is
there
some
sort
of
requirement
for
an
owner
in
The
Collector
contrib
like
they
need
saying
that
it
doesn't
have
right
access,
but
I,
don't
know
the.
C
C
B
Cool
I
think
with
that
anybody
use
open
Summit
you
go
or
open
summitery
the
past
week.
That's
of
any
interest.
D
B
D
B
Okay,
cool,
then
I
think
if
that's
gonna
be
the
case,
we
could
probably
get
back
28
minutes
and
I'm
sure
there's
a
bunch
of
stuff
to
work
on.
So
that
sounds
good
thanks.
Everyone
for
joining
we'll
see
you
next
week,
but
after
that
we'll
be
done
for
the
year,
but
yeah
same
place
same
time,
but.