►
From YouTube: 2021-11-09 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
C
Yeah,
I'm
good
doing
well,
that's
why?
How
are
you.
D
C
A
A
So
I've
I
never
read
the
books
actually,
but
I
think
I
think
it's
a
decent
adaptation
of
a
limited
primary
text
is
my,
but
I
have
to
read
the
primary
text
to
understand
that
to
really
get
the
depth,
but
I
think
cool,
cool,
visuals,
cool
cinematography
plot.
D
I
think
I
think
most
of
the
sci-fi
that's
been
coming
out
recently
is
sci-fi
eye
candy,
because
I
I
have
read
the
books,
they
are
problematic,
the
story
itself
and
characters
are
problematic.
I've
read
foundation
and
I'm
watching
apple
tvs.
Oh.
D
To
me,
I've
gotten
to
a
point
where
I'm
I'm
enjoying
hate
watching
it,
because
it's
like
sci-fi
eye
candy,
but
this
is
not
the
these,
aren't
the
characters
in
the
season,
the
story
in
the
books.
So
I'm
just
like
all
right,
you're
telling
something
you're
telling
a
different
story
where
you've
reused
names,
yeah.
A
Yeah,
it's
yeah
I've.
I
was
annoyed.
I
stopped
after
like
three
two
episodes
or
something
you
found.
B
Anyway,
cool
so
onto
something
even
worse,
we'll
go
ahead
and
start
the
the
yeah,
the
the
specs
like
recap,
and
go
from
there.
It
was
pretty
short
for
the
most
part.
B
The
sampling
pr
is
also
on
track
to
to
emerge
soon.
I
think.
B
B
Yeah,
I
guess
there's
now
this
areas
of
interest
part
of
the
open,
telemetry
community.
So
was
this
a
call
for
specifically
maintainers
and
approvers
so.
B
A
Yeah
I
know
jim's
been
working
on
some
sampling
stuff,
so
you
may
have
some
opinions
because
I
could
see
the
use
case
I
could
our
the
sampling
stuff
is
definitely
dynamic.
So
I
think
the
description
is
just
avoiding
that
by
not
including
all
the
dynamic
details
as
as
he's
written
it,
but
that's
interesting.
I
don't
think
it
could
hurt
if
it's
not
if
it's
not
frozen
and
could
be
changed,
it
could
provide
some
benefits.
A
I
guess
basically,
if
the
sampler
you
know
updates
according
to
I,
don't
know
some
api
call
and
then
it's
like
updates.
It's
sampling
value.
You
would
want
the
sampler
description
at
that
time
to
reflect
for
what's
the
returned
sampling
rate
at
that
at
the
time
which
might
be
different
in
five
or
ten
minutes
or
something.
B
B
Yeah,
I
didn't
feel
like
there
was
any
strong
opposition.
It
was
just
a
mention
that
hey
we
have
this
issue
and
if
anybody
knows
why
it
was
immutable,
please
chime
in
because
you
should
probably
change
it.
It
was
kind
of
impression
I
got.
B
More
metrics
discussions,
so
I
don't
know
how
useful
it
is
to
go
into
so
much
of
these
unless
we
are
actually
working
on
stuff.
I
think
that
there
was
kind
of
a
discussion
about.
B
Otlp
and
and
otlp
kind
of
has
this,
like
instrumentation
library,
name
stuff
on
you,
know
spans
and
metrics
kind
of
group
stuff
by
them,
and
I
think
that's
this.
This
main
discussion
here
was
about
and
kind
of
the
whole
reason
for
that
is
that
you
know
we
have
this
named
tracer
or
named
meter
situation
and
the
both
of
these
kind
of
help
prevent
like
conflicts
in
names
when
you
think
about
it.
B
So
if
you
have
like
a
named
tracer
and
you,
you
might
not
know
what
name
everybody
is
using
for
a
span
in
the
application,
but
if
you
know
the
names
that
your
instrumentation
library
is
using
for
spans
like
you're,
not
gonna
get
a
duplicate,
and
it's
kind
of
like
this
because
of
this
instrumentation
library
things.
You
can
always
scope
things
by
internet
instrumentation
library
and
get
like
a
unique
fan.
Name,
no
matter
what
you
know,
things
are
happening
elsewhere
in
the
app
for
spans.
That's
convenient,
I
think,
for
metrics.
B
It's
actually
like
a
really
good
feature,
because
of
because
all
this
data
like
aggregates
and
you
don't
want
to
accidentally,
like
you
know,
start
aggregating
data
that
isn't
the
same
because
you
get
weird
stuff
happening.
B
Trying
to
like
map
other
systems
onto
otlp
does
is
not
always
seamless
because
they
don't
they
don't
necessarily
have
these
concepts,
so
I
think
they
were
trying
to
like
figure
out
how
to
resolve
that.
So
I
think
one
suggestion
was
that
if,
if
you're
using
something
like
prometheus,
perhaps
you
could
export
a
meter
name
as
a
label
on
a
metric
and.
A
B
C
B
But
yeah
so
cool
spans.
We
have
resource
fans
here
we
have
resource
metrics
and
it's
kind
of
the
same
same
deal,
and
I
think
the
other
thing
is
that
I
think
everything
is
attributes
now,
so
I
think
yeah
we're
sometimes
still
using
this
word
label.
B
A
That's
my
understanding,
too,
is
prometheus
language
kind
of
sneaks
into
some
of
the
discussions,
but
it
is
metric
attribute
where
it
ought
to
be,
or
people
agreed
it
was.
I
don't
know
if
I
got
updated.
Okay,
that
thank
you.
That's
not
super
specific
to
the
question
yeah.
I
think
it
would
be
important
to
have
some
sort
of
library
name
added
to
metrics.
A
I
don't
know
you
get
multiple
http
libraries
combined
all
into
one
metric
and
stuff
like
goofy
stuff
like
that
that
you
don't
really
want
to
do
so
cool
yeah
I
mean
this
is,
would
be
good
to
get
this
support
out
of
the
box
with
prometheus
stuff,
because
that'll
be
what
most
metrics
are
coming
in,
as
I
think.
D
Like
the
collector
host
metrics,
so
the
collector
scrapes
up
a
whole
bunch
of
stuff
and
differentiates
between
two
things
and
values
rather
than
than
attribute
names,
so
we
have
to
do
some
transforms
to
turn
them
into
like
this
ad.
This.
This
metric
coming
in,
has
an
attribute
of
like
cp1
versus
cpu2,
like
let's
just
throw
those
into
column
names,
and
they
have
a
value
so
that
we
could
squish
them
into
wider
events.
D
I
say
that
because
having
a
library
name
means
that
we
can
go,
oh
that
instrumentation
library
is
admit,
emitting
a
metric.
We
know
what
those
metric
shapes
are,
so
we
can
transform
them
into
higher.
C
B
Cool
yeah
this
this
does
validate
the
design.
I
will
admit
that,
like
when
named
tracers
appeared
on
the
scene,
I
was
confused
and
not
convinced
they
were
a
good
idea,
but
I
think
that's
a
natural
reaction
to
any
change
in
the
world.
It
should
be.
People
should
fear
change,
but
as
as
they've
been
around
like
it
does
seem
like
they
do
solve
a
lot
of
problems
at
the
minor
inconvenience
of
having
to
kind
of
name
your
tracer
when
you
get
it
but
yeah
at
first,
they
were
a
weird
thing.
D
Admittedly,
as
I
was
getting
no
open,
telemetry
having
to
name
a
tracer
like
to
get
a
tracer,
you
have
to
give
it
a
name,
an
optional
version,
I'm
like.
Why
would
I
do
that
and
you
know
yeah
a
couple
weeks,
maybe
a
month
in
I'm
like?
Oh,
oh,
okay,
know
the
providence
of
a
thing
that
was
emitted.
D
B
B
A
Yeah
yeah
we
used
I
like
scottsdale,
I,
like
john
spawn,
let's
see
I
I
can
talk
about
the
rspec
thing.
I'm
sure
people
have
opinions
on
that,
but
if
other
people
have
more,
I
know
rob
had
some
stuff.
He
was
bumping
up
against
last
week.
If
you,
if
there's
any,
I
don't
have
anything
burning.
D
Not
stuck,
I
have
not
dug
deeply
enough
yet
to
like
figure
out
what
was
going
on,
that
the
app
is
a
it's
another
open
source
rails.
App
has
other
things
in
it.
Like
sentry
and
app
signal
already,
I
turned
off
those
gems.
D
It
does
not
appear
that
they
are
eating
the
exception,
such
in
a
way
that
the
rack
instrumentations
in
span
call
in
the
middleware
wouldn't
see
the
exception.
I
I
have
it
on
my
plan,
maybe
today,
to
try
to
replicate
the
issue
in
a
much
simpler
rails,
app
to
before.
F
What
you're
saying
is
that
it's
not
reaching
record
exception.
D
No,
I
only
did
collector
or
otlp
exporter
to
a
collector,
but
I'll
that'll
be
my
next
step.
I'll.
Try
a
console
cool,
so
yeah,
I'm
not
opening
a
bug
yet
could
totally
be
me,
but
I'll.
Try
that
next.
A
D
When
sentry
is
enabled,
sentry
sees
the
exception,
but
when
I
disabled
century
it
didn't
change
any
behavior.
So
I
don't
know
I'll
I'll
dig
in
deeper
before
I
open
any
bugs.
No.
A
Worries
I
just
wanted
to
make
sure
you
had
time
appreciate.
G
A
Yeah
I'm
around.
If
you
need
anything
yeah,
I
could
talk
about
the
r
spec
stuff.
I
reviewed
the
r
spec
pr.
I
think
it's
pretty
much
good
to
merge
like
there
is
some
small
feedback
like
one
like
some
of
the
attribute
names
ought
to
get.
I
think,
prefixed,
you
know
the
same
way.
We
do
like
messaging
dot
whatever
or
some
of
them
are
a
bit.
Some
of
the
attribute
names
span.
A
Attributes
are
like
kind
of
arbitrary,
so
it
just
feels
like
a
good
housekeeping
thing,
because
we're
sort
of
like
inventing
specification
here
or
to
say
like
well,
let's
prefix
this
stuff,
because
the
the
names
that's
themselves
aren't
super
descriptive.
The
only
other
thing
and
like
there's.
No,
this
is
sort
of
like
not
established
anywhere,
except
like
I
was
looking
at.
A
Like
jonah
dugan
wrote
like
a
go
test
instrumentation
and
like
I
was
looking
at
the
datadog
stuff
like
setting
this
step
like
what
determines
when
you
would
set
a
span
status
of
error,
for
you
know
an
r
spec
test,
and
so,
like
I
kind
of
interpret
it,
it
looks
like
the
prior
is
like.
If
the
test
fails,
then
you
say
this:
the
r-spec
span
is
an
error
or
it
has
a
status
of
error,
and
you
know
you
provide
the
description
of
what
I
forget.
A
There's
like
some
message
you
can
but
like,
but
that's
sort
of
like
not.
I
just
made
that
up.
That's
like
I
could
see
that
being
a
you
know
you
in
my
mind
the
reason
you
have
our
expected
situation
is
you
want
to
see
like
okay,
I
ran
my
test
suite
how
many
failed
tests
do.
I
have
and
the
best
way
the
canonical
way
to
sort
of
check
that
in
like
the
open,
otlp
context
is
like
check
status
so
like
so,
you
know.
A
However,
many
traces
are
like
errors
in
equals
could
be
one
to
one
with
like.
However,
many
traces
are
like
failed
tests,
but
like
that's
not
defined
anywhere,
it
just
sort
of
felt
like
good
practice,
so
that
was
my
only
real
feedback
like
that
was
the
only
thing
missing.
A
I
thought
everything
else
looked
fine
and
I
sort
of
feel
like
we
should,
because
it's
like
experimental,
we
should
just
sort
of
like
get
it
out
in
the
wild
and-
and
you
know
whatever,
and
if
people
want
to
like
give
feedback
around
our
conventions
on
whatever
we're
missing
some
attribute
like
cool
but
yeah.
I
don't
know
so
that
was.
I
finally
took
a
look
sorry
for
the
delay
with
this
stuff,
but
yeah
it
was
good.
Thank
you.
It's
a
useful,
instrumentation
and
yeah.
C
A
There's
good
directions
in
there
on
like
how
to
set
up
a
separate
pipeline.
Basically,
so
it
doesn't
interfere
with
your
existing.
You
know
so
it's
sort
of
like
either.
Let's
say
you
have
instrumentation,
you
want
to
run
and
check
correctness
of
like
your
r
spec
spans
won't
interfere
with
those
anyway
yeah.
I
think
it's
good
stuff.
It
would
be
nice
to
see.
A
I
I
tried
to
take
like
a
little
survey
of
like
what
the
broader
community
is
doing
around
trace
tracing
for
like
test
libraries
and
besides,
like
jana's
thing,
which
is
just
like
an
open
source
that
you
know
she
didn't
like
make
a
repo
in
the
context
of
inflammatory,
like
she
just
like
made
a
personal
replay
like
there's
no
real,
like
I
couldn't
find
anything
else
in
other
languages.
I
didn't
dig
super
deeply,
but
yeah.
A
It
would
be
nice
to
see
specification
get
formed
but,
like
I,
don't
really
want
to
be
the
person
to
do
it,
because
it's
a
lot
of
work
and
yeah.
So
anyway,
that
was
my
respect
feedback.
F
I'm
really
curious
about
the
schema
url
stuff
because
that's
impacting
like
that,
might
impact
the
direction
going
in
the
future
and
kind
of
like
what
my
I
thought.
Thoughts
are
about
the
like
evolving,
the
semantic
inventions
gem
in
some
way
right
because
we
have
we're
generating
them
as
like
constants
right
now,
but
not
really
leveraging
them.
F
You
know,
we've
had
some
I'm
chatting
about
this.
I
think
I
just
have
to
get
myself
more
involved
in
the
conversation
upstream,
but
trying
to
understand
like
how
do
we,
including
like,
if
there's
only
one
notion
of
one
schema,
does
that
schema
represent
all
versions
of
auto
instrumentation
semantic
conventions
and
then,
as
far
as
like
our
extensions
to
them,
so
in
a
sense
like
any
library
instrumentation
that
we
write
we're
generating
attributes
in
some
cases
that
aren't
part
of,
like
the
official
semantic
conventions,
also
right
right
and
so
like.
F
How
do
we
what's
our
play?
There?
Do
we
say
that
a
for
a
specific
version
of
semcom
all
of
these
gems,
like
the
each
gem,
know
what
version
of
semcom
that's
using
it's
it's
it's
a
little
yeah
a
little
ambiguous
to
me
versus
like
say
like
when
you
look
at
the
golang
implementation.
F
A
You
can
optionally
specify
a
schema
url
that
which
is
that
it
adheres
to
okay,
you
don't
have
to,
and
that
doesn't
have
to
be
in
lockstep
with
your
your
sdk,
like
your
resource
schema
convention,
your
schema
url
that
can
be
different.
D
A
I
was
like,
but
there
isn't
really
any
and,
and
the
idea
is
like
there
will
be
some
abstract
process
for
like
merging
this
stuff.
If,
like
let's
say
our
rails,
instrumentation
is
based
on
an
older
schema
and
then
our
you
know
api
or
resource.
Excuse
me
urls,
like
the
latest
one.
There
will
be
some
process
for,
like
you
know,
doing
the
translation
to
the
latest
one.
You
know,
but
that's
all
arbitrary
and
hasn't
been
done
anywhere
like.
F
A
Can
it
can
be
so
like
all
schema,
url
points
to
is
like
a
file
of
yaml
and
it's
like
translations
from
a
tribute
name
to
a
tribute
name
that
should
be
based
on
it.
It
can
be
the
like
of
otlp
semantic
convention
version.
It
doesn't
have
to
be.
It
can
be
like
some
arbitrary,
like
github
semantic
conventions
you
can
and
you
can
define
it
for
a
particular
instrumentation
or
for
like
your
resource.
Basically,
and
I'm
sorry,
I'm
talking
in
circles,
I
guess
it's.
A
And
then
there's
no
real.
The
problem
is
like
there's
no
real
agreement
or
process
around
like
well.
How
do
you
merge
these
things
if
they're
different
or
if
they're,
where
do
you
merge
them,
hasn't
really
been
clearly
defined?
I
think
they're
starting
and
go
what
happened?
Is
they
realize
that
you
can
basically
merging
resources
of
two
different
semantic
conventions?
A
A
What
would
happen
is
there's
a
merge
error
on
that
resource
and
you
lose
everything
like
you.
Don't
get.
You
basically
lose
all
the
attributes
instead
of
just
like
losing
just
the
schema
and
I'm
not
doing
a
very
good
job
here.
Basically,
what
I'm
saying
is
they've
sort
of,
I
think
now
circle
back
and
realize
there
hasn't
been
enough.
A
They
haven't
done
a
good
enough
job,
defining
some
of
this
behavior
and
my
understanding
is
like
aside
from
us,
implementing
support
so
that,
like
it's
available,
it's
like
an
argument.
If
you
want
to
find
a
resource,
and
like
it's
added
in
like
the
protos,
you
know
like
when
we
do
export
like
we
really
shouldn't
do
too
much
here
until
there's
more
to
find,
because.
F
I
appreciate
you
walking
me
through
this,
and
I
see
that
matt
brought
up
on
this
screen
the
protos.
So
clearing
up
the
ambiguity
for
me
is
that
each
instrumentation
can
have
its
own
version
of
the
semantic
conventions
each
resource
can
have
its
own.
The
resource
can
have
its
own
version
of
the
semantic
conventions
and.
F
The
the
is
going
to
sound,
really
silly,
but
any
manual
spans
are
considered
a
name
tracer,
which
is
an
internal
instrumentation,
internal
instrumentation
library.
So
each
then
internal
span
set
can
have
its
own
schema.
A
The
schema
url
gets
passed
to
a
tracer
provider.
Oh
I'm
not
doing
a
great
job,
remembering.
F
It's
okay,
would
you
mind
scrolling
around
a
little
bit
matt
if
the
and
I
apologize
if
this
is
not
what
people
want
to
talk
about?
No.
A
It's
good.
The
other
thing
to
bring
up
is
like,
if
you
let's
say
we
have
some
non
semantic
convention
attributes
like
I
don't
know.
Let's
our
spec
is
a
good
example
right.
We
want
to
say
like
well.
How
do
we
like
version
this
stuff
and
like
say
it
relate
you
know
like
or
let's
say
you
have
some
honeycomb
specific.
You
know
semantic.
A
There's,
no
real
way
to
do
like
parentage
or
sort
of
like
root,
schema
urls,
and
so
like
the
honeycomb
schema
url
or,
like
the
you
know,
shopify
one
there's
no
way
to
define
that
it's
like
a
super
set
of
a
particular
version
of
like
hotels,
schema
url.
So
it's
yeah.
Basically,
unless
it's
like
there's
no
real
clear
way
to
do
that.
Stuff.
D
This
would
be
why
we
have
just
like
you
tell
us
what
your
columns
are,
we'll
we'll
derive
it
from
the
data
that
you
send
us
and
if
you
get
some
weird
queries,
because
some
spans
come
in
calling
calling
a
field
one
thing
and
others
fans
come
in,
calling
that
what
is
in
essence
the
same
data
but
as
a
different
column.
Name
we're
like
cool.
D
A
A
To
understand
so,
if
you
do
choose
to
review,
I
would
say
review
it,
comparing
it
to
what
python's
done
it
should
be
one-to-one.
I
am,
I
think,
there's
some
minor
open
questions
around
like
are
google
or
gcp
resource
detector?
Like
do
we
want
to
define
technically
that
should
define
a
schema
url?
A
Yes,
I
believe,
but
it's
a
little
bit
unclear
to
me
whether,
if
are
doing
so,
would
become
in
practice
very
problematic
if
you
know,
because
it
would,
in
effect
create
merge
resource,
merge
errors.
If
any
user
wants
to
then
define
their
own
schema
url,
I
believe
which
feels
like
something
they
wouldn't
expect
to
happen.
A
Even
if
it's
technically
correct,
I
don't
know
so
yeah,
I'm
kind
of
like
fine
to
let
this
sit
for
a
while.
We
should
probably
at
some
point
we
probably
will
need
to
get
emerged
to
like
keep
up
with
spec.
F
Resource
spans,
that
message
is
a
collection.
Is
it
is
it's
wrapping?
It's
wrapping.
The
collection
of
interest
of
those
instrumentation
spans
has
a
single
resource,
has
a
schema
and
that
schema
corresponds
to
the
resource,
not
to
the
library.
The
library
span
itself
in
its
wrapper
message
includes
a
schema
url
with
its
list
of
spans
that
all
correspond
to
the
instrumentation
library
resource
data
and
where
are
those
where's
that
resource.
A
C
A
A
A
F
My
eyes,
like
sort
of
vague
understanding,
because
theoretically
right
you
can
from
the
client
side-
and
you
said
this
already-
I'm
just
reiterating
it.
So
I
make
sure
I'm
I
understand
it
correctly.
As
as
a
client,
you
know
my
rails,
app
is
emitting
version
1.7
and
for
simplification
purposes,
everything
is
1.7
and
it
sends
it
out,
but
the
downstream
we
really
want
2.0.
F
So
we
configure
our
collectors
to
say
go
from
1.7
to
2.0,
because
this
is
my
target.
Schema
output
and
the
collector's
processor
pipeline
will
have
a
attribute,
schema
processor
or
a
schema
attribute,
processor
or
whatever,
and
that
does
the
transformations
and
amidst
version
2.0,
and
then
you
look
in
honeycomb
and
then
you
scan
around
in
there
and
you're
like
oh
everything's
in
2.0,
but
honeycomb
also
can
be
like.
Oh
as
I'm
ingesting
things,
I'm
going
to
do
some
other
transformations
and
special
stuff.
F
D
A
Future
is
they
don't
is
not
every
you
know
the
collector
isn't
can't
be
assumed
as,
like
a
you
know,
there's
folks
here
who
represent
companies
that
encourage
people
to
not
use
a
collector
in
some
use
cases
and
so
like.
I
think
they
want
to
be
able
to
have
that
processing
like
we
talked
about
last
week
within
profs
within
app,
which
is
a
performance
penalty,
but
I
think
they
do
want
that
and
then
also
there's
ambiguity
around
like
well.
A
What,
if
you
want
to
what,
if
you've
defined
your
schema
at
you
know
whatever
in
your
column,
encoded
back
in
like
at
1.5
and
now
like,
and
you
know
it's
already
been
set.
You
have
all
your
dashboarding
on
1.5
and
you're.
You
know
and
you're
indexing
and
then
like,
but
the
scheme
has
moved
on
to
like
1.8.
A
You
can't
there's
no
way
to
like
walk
it
back
down,
there's
no
way
to
do
like
backward
compatibility
right
now,
which
is
problematic
because
you
know
at
some
point
you
have
to
block
your
you
want
to
be
able
to,
like
you
know,
know
what
columns
you
have
in
your
back
end,
maybe
anyway.
So
I
think,
there's
still
lots
of
like
whole
gaps
in
how
this
can
be
used
practically
speaking
to
like
provide
value
for
people.
F
C
F
It's
just
more
about
like
our
implementation
and
what
we
can
do
to
obstruct
to
to
mitigate
like
having
the
instrumentations
become
complicated
as
how
they
choose
what
name
they
want
for
a
particular
thing.
F
If
we're
ultimately
wanting
to
like,
like
eight
instrumentation,
you
know
whatever
rack
once
one
seven
versus
two
three
or
something
like
that.
It's
like
what
do
do
those
does
our
instrumentation
semantic
versioning
map
to
a
semantic
convention,
or
is
that
something
that
we
want
holistically
to
be
able
to
support
across
all
instrumentations?
A
Yeah,
I
think
that's
right.
The
semantic
convention
conversion
would
be
included
in
the
schema
url
file
that
the
instrumentation
schema
url
would
point
to.
D
A
A
You
know
this
random
attribute
name,
go
update,
you
know
like
they
just
want
to
be
able
to
stay
like
this
is
the
at
the
point
in
time
we
added
this
stuff
like
here
is
what
the
spec
said,
and
we
will
assume
those
like
most
libraries
will
be
out
of
we'll
lose
lockstep
from
you
know
what
the
latest
and
greatest
version
details
are
and
whatever
like
and
that's
fine
as
lo,
but
by
allowing
them
to
specify.
Like
you
know,
this
is
at
the
time
here's
what
a
snapshot
of
like
what
we're
looking
at
was.
A
D
I
have
a
semi-random
horse
to
possibly
start
beating
on
go.
D
D
It
are
baked
in
metrics
emitters,
particularly
in
jvm
around
the
jvm
metrics,
like
garbage
collection
and
heap
size,
and
all
that
jazz
is
that
I'm
I'm
slightly
intrigued
at
adding
auto
instrumentation
of
the
ruby
virtual
machine
and
a
little
thing
that
is
either
opt
in
or
opt
out.
If
you,
if
you
do
auto,
instrumentation,
all
or
instrumentation,
all
maybe
it's
you've
opted
in,
but
it
would
just
pull
on
some
interval
interesting,
ruby,
virtual
machine
metrics.
I
have
found
that
useful
in
the
past.
F
A
Yeah
I
mean,
I
think,
that
stuff's
great
I'd
like
to
see,
I
think
it's
just.
I
don't
think
anyone
would
say
no
to
it.
We
just
kind
of
need
a
metrics
api.
D
A
F
F
Yeah
so,
but
this
is,
this
would
be
sdk
specific
to
us
and
we
could.
E
A
Oh
really
until
matrix
is
my
understanding
is
until
metrics
is
stable,
not
to
switch.
I
would,
and
I
would
call
that.
F
A
I
would
it
would
emit,
I
mean
it,
would
it
would
have
a
you,
would
have
to
pass
in
your
your
own
satsvi,
client
or
prometheus
client,
or
whatever
is
my
sort
of
understanding?
Are
you
all
trying
to
noodle
away
to
have
this
thing
before
we
put
no.
D
D
F
A
A
D
I
said
I've
I've
written
a
a
really
janky
dirt
ball,
one
in
in
a
rails,
initializer
and
some
threads,
but
having
it
better
baked
with
other
smarter
people
reviewing,
it
would
be
delightful
and
yeah.
It
would,
as
as
matt
has
been
browsing
around
while
we've
been
nerd
sniping
each
other.
Yes,
all
of
these
semantic
inventions
are
all
happily.
B
I
don't
know
if
all
defined,
but
there's
some
placeholders,
where
you
know
people
have
been
thinking
about
some
of
these
things
yeah
so
yeah,
it
seems
like
host
metrics
are
maybe
somewhat
defined.
The
more
of
the
runtime
gc
stuff
is
I'm
totally
cribbing
off
of
jbm
metrics
coming
out
of
the
java
agent
yeah,
there's
at
least
a
runtime
prefix
that
we
know
about
so
so.
There's
that.
F
Well,
bring
up
the
open,
pull
request
right
now,
maybe
it'll
refresh
our
memories
or
any
new
issues
that
might
have
come
up.
B
Yeah-
and
I
guess
on
that
note-
I've
heard
that
francis
and
robert
are
vacationing.
I
can
only
assume
together
talking
about
open
telemetry
without
us.
A
I
think
so
yeah
it's.
I
think
it's
minus
some
very
crazy
number
of
degrees,
so
I
hope
rob
would
still
weren't
warm,
but
I
don't
know
where
he
is.
F
B
And
strengths-
I
was
just
gonna
say
in
their
absence
like
if,
if
you
need
anything
for
me
for
me
aggressively
ping
me
and
don't
feel
bad
about
it,
and
I
will
I
will
step
in
and
help
help
out
with
anything
and
yeah.
I
hope
I
didn't
hurt
anybody's
feelings
with
that
joke.
B
B
A
There's
the
kafka
instrumentation
that
I've
been
meaning
to
review.
That's
from
our
colleague
andy,
I'm
pretty.
E
A
It's
good
to
go.
I
think
rob
had
done
a
pretty
fair
review,
but
I'm
gonna
give
it
a
pass.
It
should
be
useful,
the
other
stuff.
I
know
we
merged
tim's
work
on
the
active
support
stuff.
So
I
don't
so
that's
awesome.
I'm
curious
about
what
the
active
support
stuff.
G
It
so
I
would
like
to
shower
praise.
Tim
can
speak
sure,
yeah,
it's
merged.
It
basically
allows
you
to
derive
spans
from
active
support
notifications,
yes,
and
it's
there
and
then
in
rail
seven,
it's
better.
We
don't
have
to
do
a
bunch
of
hacky
stuff
because
they
need
the
exception
handling
around
active
support
notification
subscribers.
So
it's
gonna
be
even
better
theo.
F
G
You
mean
this,
which
version?
Are
you
talking
about.
F
So
wonder
in
rails,
yeah
in
in
real
seven
we've
been
doing
some
experimentation
with
active
support
notifications
and
what
we're
seeing
in
some
of
our
profiles
is
that
the
logging
subscriber
in
that
pipeline
is
really
slow
and
I
haven't
done
any
more
digging
into
it.
I
was
just
curious
if
you
all
had
seen
that
when
you
were
doing
your
testing
of
action
view
and
active
support.
F
A
A
A
D
So
this
act
of
support,
I'm
I'm
only
just
seeing
the
the
pr
is
there
now
like
an
active
support.
Gem
instrumentation
gem
yeah,
correct.
C
G
You
click
through
it
there's
an
example
of
how
it
works
so.
G
Use
this
today,
yeah,
I
think
you'll
have
to
do
some
shenanigans
in
your
jam
file
because
we
haven't
released.
I
am
I'm
not
opposed
to
shenanigans
just.
D
Better
be
cutting
or
release,
because
I,
because
I
have
no
powers
in
this
repo
other
than
yeah
no
power.
A
F
B
A
F
What
it
doesn't
say
is
if
it's
comma,
separated
or
white
space
separated,
but
you
would
fill
this
form
out
and
click
run
workflow
and
then
it'll
open
a
pr
that
tr
that's
getting
you
ready
to
do
a
tag
and
then
the
approvers
will
have
the
ability
to
say
like
yes,
this
will
move
forward
and
then
they
take
another
step,
something
that
I
don't
have
the
permissions
to
do
and
then
from
there.
I
think
it's
going
to
be
it.
E
F
A
A
D
Yeah,
that's
cool.
I
will
note
that
as
a
even
though
I'm
a
member
of
the
open,
telemetry
org,
I
do
not
have
permission
to
see
that
button
on
the.
F
That's
okay
and
then
you
can
go
ahead
and
click
the
the
the
don't
do
that
don't
do
that
click!
The
link
where
the
the
title
is
so
that
you
can
follow
the
actual
and
I'm
doing
a
demo
of
the
github
product,
github
actions,
the
ci
cd.
F
C
F
F
You
shall
not
see
anything
in
here
matthew
where,
because
it
hasn't
released
it
yet
so
I
was.
C
C
H
B
Yeah,
I
wasn't
sure
if
we
were
starting
off
at
like
a
slightly
higher
somewhat
more
consistent
number.
But
it
seems
like.
D
As
an
as
a
semi
outside
observer,
it's
like
the
the
instrumentations
that
have
been
coming
up
through
the
ranks
towards
the
api
and
sdk
going
1.0
all
had
like
matching
version
numbers
with
the
api
and
the
sdk,
and
now
like
the
new
instrumentations.
Are
I
don't
mind
loading.
F
D
My
initial
desire
was
to
like
see
if
the
github
actions
could
fan
out.
All
of
the
maybe
I
know
toys
is
the
thing
that's
doing
the
purred.
D
E
D
A
F
D
D
That
so
that
each
thing
could
get
run
in
parallel
and
and
then
when
you're,
looking
at
github
actions
and
like
what
failed
you,
don't
you
aren't
scrolling
through
a
thousands
of
blind
log
of
of
toy's
output.
It's
the
job
that
was
named
ruby27
open,
telemetry
instrumentation.
Active
support.
Failed,
oh
cool.
D
B
I
do
have
to
skedaddle,
but
if,
if
somebody
would
like
to
review
this
I'd
be
happy
to
continue
to
push
this
release
out
as
long
as
we're
happy
with
it.