►
From YouTube: 2021-11-12 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
A
A
A
A
All
you
need
is
you
have
a
batch
of
metric
and
how
you
define
the
metric,
and
the
matching
point
is
up
to
your
decision,
because
each
language
will
be
very
different
and
this
is
mainly
driven
by
performance,
and
then
there
are
folks,
especially
from
the
javascript
comments.
You
can
give
some
concrete
example
that
helps
us
to
make
the
decision.
A
If
the
the
trace
is
sampled
and
if
you
only
have
matrix,
there's
no
trace,
then
exemplar
would
simply
become
a
no
out,
and
there
shouldn't
be
any
performance.
Implication
well.
Sigil
proposed
in
this
pr.
That
example
should
always
be
off
by
default,
and
you
have
to
enable
that
so
josh
doesn't
think.
That's
the
good
way,
and
my
gut
feeling
is
probably
like.
I
would
more
be
aligned
with
josh.
C
I
think
it
should
be
off
by
default
because
the
so
as
far
as
I
know
as
of
today,
only
one
back-end
supports
examples
which
is
prometheus
now
the
example
are
supports
in
prometheus
is
seen
like
feature
preview
or
I
don't
know
what
is
the?
What
is
it
right?
What
is
the
terminology
that
they
are
using,
but
you
need
to
explicitly
enable
in
the
it's
like
this.
I
don't
know
experimental
feature,
so
it's
it's
disabled
by
default,
so
even
if
auto
enables
it
by
default.
A
C
I
have
a
quick
question,
though,
about
exemplars,
because
you
mentioned
that
if
tracing
isn't
present,
then
examples
will
be
off.
Is
there
a
way
for
auto
to
support
like
getting
the
examples
from
like
whatever
source
like
not
necessarily
from
water
tracing,
but
from
like
a
third-party
tracing
system
that
users
are
using.
A
A
If
I
want
to
get
more
perspective,
I
like
the
the
reason
we
keep
this
pr
there
and
it's
a
little
bit
stuck
is
I
I
can't
see
two
different
parties
and
they
all
have
very
clear
explanation
of
why
they
think
that's
the
case.
One
is
more
looking
forward.
Thinking
like
most
of
the
people
would
find
it
useful,
but
when
you
look
at
today,
not
many
systems
would
benefit
from
it.
So
most
likely
you'll
collect
the
data
and
then
you'll
drop
the
date
on
the
floor,
which
means
you
burn
cpu
cycle
without
getting
additional
benefit.
A
A
The
the
ask
is
very
simple:
so
in
the
current
sdk
spec
we
define
multiple
aggregation
and
we
have
an
aggregation
called
non,
which
means
you
don't
do
anything
besides.
Just
stop
the
data
on
the
floor,
so
cgo
suggested
that
we
try
to
be
explicit
and
and
change
that
to
job.
I
think
the
the
rationale
is
non
has
has
some
meaning
in
certain
language.
So
when
people
see
an
enum
or
some
value
called
noun,
it
might
already
be
taken
in
certain
language
which
makes
it
harder.
A
A
The
thing
here
is,
if
you
have
like,
if
you
have
a
library
that
is
already
instrumented,
they
use
different
things
like
the
counter
asynchronous
gauge
those
things-
and
you
know
you
want
all
of
the
metrics
from
that
library,
but
there
are
certain
things
you
want
to
remove.
Then
what
you
can
do.
Is
you
specify
a
view
saying
I'm
going
to
take
this
temperature
from
the
room
and
the
aggregation
is
numb
or
job?
Okay?
A
A
C
A
I
want
everything
or
you
can
see.
I
want
everything
that
name
has
a
temperature
in
it,
but
for
this
particular
room
temperature,
it's
just
too
noisy.
So
I
I'd
rather
to
drop
the
data
and
the
current
spec
is
documented
that
you
got
to
put
a
non
type.
So
the
suggestion
is
try
to
make
it
more
explicit.
A
And
it
seems
it's
a
minor
thing
and
people
don't
feel
so
strong
about
it,
but
I
personally
think
that
this
will
bring
some
clarity.
B
C
A
D
A
A
So
there's
a
long
discussion
and
I
I
personally
feel
we
don't
have
enough
column,
but
I
want
to
show
you
a
quick
demo,
so
the
proposal
from
this
pr
is
very
simple.
So
when
you
export
to
permissive
exporter,
you
should
have
an
option
to
say
I
want
to
have
the
meter
name
and
version
as
extra
dimension,
which
is
a
label
to
premises.
A
I'll
show
you
something
like
this.
So
so,
basically
you
have
something
like
the
library
and
version
you
see.
The
library
is
a
where's,
the
library
the
library
is
test
meter.
C
A
A
C
Sorry,
I
I
completely
forgot
that
auto.
Has
this
weird,
like
intermediate
step,
called
the
meter.
A
A
The
name
and
also
you
can
see
the
library
is
probably
fine,
but
I
think
if
you
put
the
version
there,
it
has
a
high
chance
to
conflict
with
some
existing
stuff
and
also
I
I
think
there
are
cases
where
you
only
have
library
and
don't
have
version,
and
the
current
proposal
doesn't
clarify
what
you
should
do
there
and
there's
also
a
debate
whether
this
should
be
owned
by
default
and
people
can
opt
off
or
it
should
be
out
out
by
default,
it's
very
related
to
the
next
pr.
A
Yeah,
so
one
one
thing
we
talked
about
is
in
open
telemetry.
We
have
a
concept
called
unit
but
impromises
based
on
matters,
and
it
seems
like
normally
people
would
recommend
that
you,
you
put
the
unit
as
a
as
a
post
fix.
Yes,
so
the
current
prototype
I'm
working
out.
I
just
take
the
unit,
so
it's
instead
of
the
unit
sum
I
make
the
unit
underscore
not
name
under
support
unit
and
underscore
count,
and
there
are
many
like
details
like
this-
we
try
to
clarify
without
this
pr
being
merged.
A
I
I
talked
with
john
sturridge
and
both
of
us.
I
think
we
won't
be
able
to
have
a
stable
version
of
permissions
and
which
is
probably
fine,
but
given
promises.
Exporter
is
pretty
important.
I
would
say
it's
probably
the
number
one
important
matrix
explorer
for
the
open
timesheet
project.
So
we
want
to
get
more
opinion
here.
A
So
take
a
look
on
these
two
pr's
they're
very
related.
C
A
And-
and
this
is
what
promises
has
in
official
documents
now-
I
know
there
are
the
the
duration
seconds,
then
there's
buckets
a
little
bit
confusing
and
and
dmacd
actually
talked
with
the
the
promises
working
group
yesterday
and
he
put
a
nice
summary
there.
So
I
suggest
you
also
take
a
look
there.
There's
a
lot
of
history.
A
Okay,
the
last
thing
issue
triage,
so
there's
no
additional
issue
since
the
last
meeting.
Now
guys-
that's
all
so.
To
summarize,
I
I
think
those
are
the
smaller
things
if
you
can
give
a
comment,
give
your
support
or
you're
giving
a
thumbnail
saying.
No,
this
is
not
the
direction.
I
would
appreciate
that
and
for
the
for
the
remaining
two
issues,
not
the
prs
here
they
are
very
important
topic
and-
and
still
many
things
are
unclear,
so
I'm
doing
the
prototype.
A
C
Do
you
have
any
like
target
date
for
that
for
the
marking
the
sdk
stable.
A
I
I
think,
given
the
permissions
part,
we
have
two
options:
either
we
mark
the
isdk
as
stable,
but
the
premises
exporter
is
not
stable,
which
I
I
think
will
be
less
useful
because
like
why.
Why
would
people
use
a
stable
version
of
icky
with
no
stabilized
color
or
just
give
them
a
e-memory
or
console
exporter
right?
So
I
really
hope
that
we
don't
give
us
excuse
just
to
just
to
postpone
the
premises.
I
I
want
us
to
to
take
some
effort
to
hammer
out
the
premises,
exporter
and
announce
a
stable
release
on
both
of
them.
A
That's
my
hope
and
I'll
try
to
drive
that
as
hard
as
I
can,
but
in
case
we
think
premises
will
take
more
time.
For
example,
we
think
there
are
additional
things
that
will
take
several
months
for
us
to
debate.
Then
I
think
it's
probably
fine
for
us
to
get
the
sdk
market
stable,
so
that
gives
the
the
language
maintainers
more
confidence,
meanwhile,
making
the
premises
exporter
as
experimental,
and
we
can
still
ship
the
pre-release
version
of
the
explorer
by
telling
people
the
interface
is
unlikely
to
change,
but
the
behavior
might
change.
A
Yeah
so
so
short
answer
is
I
I
still
hope
that
we
can
have
the
promises
and
the
isdk
spec
shape
has
stable
on
the
same
day,
but
in
case
we
hear
strong,
pushback
saying
the
premises.
Spec
is
not
going
to
be
stable,
and
not
this
month
or
end
of
this
year
or
even
next
january,
then
we'll
have
to
make
a
decision
to
separate
them.
B
B
Sorry,
just
sorry
is
this:
is
this
mating
acceptable
to
discuss
the
metric
semantic
convention,
or
is
that
is
that
is
there
another
meeting
for
that.
A
B
Yes,
yeah,
I
think
it's
just
on
at
like
3am
or
something
for
me.
Oh,
so
the
time
wouldn't
work.
A
B
Right,
yeah
yeah,
I
I
just
I
did
have
a
few
questions
about
the
the
metric
semantic
invention,
because
I'm
doing
some
work
on
it,
but
I'll
put
it
in
the
slack
and
we
can
just
talk
about
it.
Asynchronously,
okay,.
A
So
so
my
my
suggestion,
if
you
go
to
the
semantic
convention
sick,
you
probably
will
be
able
to
find
the
the
meeting
notes
there.
My.
A
You
leave
a
note
there
saying
you
have
a
challenge
joining
that
due
to
a
time
zone
issue,
you
don't
have
to
join
the
meeting,
but
as
long
as
you
leave
a
topic
there,
you
mention
what
time
zone
you
are,
I
I
think
they
will
ask
around
and
this
seems
there's
enough
column.
People
might
think
to
have
the
the
rotation
similar
like
this
meeting,
like
we
have
the
morning
and
the
afternoon
meeting
rotate
every
week,
just
to
cover
different
time
zone.