►
From YouTube: 2022-03-29 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).
B
Yeah,
yep
and
honorag
is
practically
done
with
tuesday.
B
A
B
Oh
you're
there
I
just
oh
wrong.
I
just
dialed
you
on
slack.
C
A
I
can't
share
but
hold
on
a
second,
I
I
just
started
my
laptops,
like
I
still
don't
have
anything
that
I
should
have.
A
I
can
share
it's
okay,
I
mean
the
pr
is
not
that
big
anyway,
I
just
put
the
just
the
necessary
things
so
yeah.
The
issue
was
about
adding
version
to
each
and
every
instrumentation
library
and
generated
instrumentation.
We
produce
as
part
of
the
agent,
and
I
will
remember
the
issue
had
the
suggestion
of
using
a
file,
a
properties
file
for
that.
But
that's
what
I
thought
it.
It
would
make
sense
in
this
scenario.
A
A
A
Yeah
yeah,
it
will
be
version
line,
they
would
just
have
all
the
same
version,
and
but
in
that
case
it
doesn't
make
sense
to
time
to
tie
it
and
do
instrumentation
api,
because
we
have
telemetry
sdk
version
and
we
have
the.
A
A
Yeah
there
are
resource
attributes
that
just
identify
the
version
precisely
so
that
leaves
us
with
two
cases.
One
is
library
instrumentation
that
okay,
that
can
be
used
without
the
java
agent
or
possibly,
maybe
even
with
like
different
versions
of
library,
experimentations
but
yeah.
The
bond
files
should
should
prevent
that,
and
there
are
extensions
which
may
be
compiled
once
contain
some
custom
instrumentation
for
whatever,
and
they
can
have
different
versions
than
dividing
instrumentations.
A
But
then
again,
if
you
as
a
user,
the
agent
build
your
own
custom
extension.
You
can
use
this
set
version.
Instrumental
builder
method
to
you,
you
don't
have
to
you-
rely
on
the
file
so.
C
B
A
A
Yeah,
so
that's
a
good
idea
and
if
we
decide
to
go
with
this
approach,
I'll
just
refactor
it
to
multiple
files,
but
overall,
I
think
that
it's,
it's
probably
a
bit
complicated
and
the
only
thing
that
we're
getting
out
of
it
is
kind
of
like
a
backup,
the
the
fallback
of
the
version,
because
if
any
any
third
party
I
mean
not
us
wants
to
set
the
instrumentation
version,
they
have
a
live
builder
for
that,
so
it's
like
95
90,
just
for
our
own
usage
and
yeah.
A
B
So
what
about
a
a
simple
kind
of
naive
approach,
for
it
doesn't
cover
the
java
agent,
but
for
like
if
we
had
the
the
file
name
based
on
the
instrumentation
name,
so
they're
unique
and
just
read.
B
A
B
Can
we
for
that
the
instrumenter.
B
C
We
could
I
mean,
of
course,
it's
more
code,
but
it's
once
it's
there.
It's
not
that
hard
for
the
instrumentation
to
push
up
the
version
like
it
doesn't
have
to
be
read
in
the
instrumentation
you
play
could
be
read
in
the
instrumentation
telling
like
the
telemetry
builder,
whatever
we
always
have,
so
that's
possible.
That's
annoying,
but
it
can
be
pushed
up,
but
I
sort
of
also
wouldn't
mind
doing
that
only
if
someone
reported
an
issue
because
I
don't
expect
people
to
run
into
issues
for
this
class
loader
problem.
B
A
B
A
B
C
B
A
Okay,
then
so
I'll
refactor
that
and
change
that
to
use
multiple
files
and.
A
A
C
A
C
B
C
B
A
What
okay,
so
just
to
repeat
in
case
you,
I
didn't
miss
yeah.
How
are
you
doing
this?
Okay?
So
I
think
that
we
have
like
gradient
modules
that
have
more
than
one
instrumentation
module
inside,
but
I
think
they're
all
using
the
same
instrumentation
there's
like
several
bytecode
com
modification
components
that
end
up
being
the
same
instrumentation
like
an
instrumentation
concept.
B
A
That's
that
could
be
possible
I'll.
I
check
it
out,
but
yeah
it.
It
should
be
possible
right,
you'll
be
right.
I
kind
of
missed
it.
A
One
one
more
thing:
can
you
open
the
changed
files
and
then
helper
resources,
there's.
A
A
Forget
to
go
up
here
to
the
other
yeah
there's
five
filters
right,
there's
one
more
improvement
that
I
I
had
to
make
to
make
this
thing
work,
but
maybe
it's
useful
to
include
it
like
anyways,
currently
we're
in
our
helper
resources,
implementation
we're
just
studying
just
one
url.
So
in
case
you
have
an
instrument.
They,
you
have
two
instrumentations
that
add
the.
A
Spi
of
the
same
type,
like
separately
of
each
other
one,
will
override
another
and
because
we're
selling
just
one
url
to
just
one
sba
file.
So
maybe
it
would
be
useful
to
make
that
a
list.
So
if
you
have
like.
A
For
example,
if
we
have
the
built-in
java
agent
log
background
for
jspi,
and
then
somebody
makes
an
extension
so
that
it
copies
something
from
baggage
to
mdc
and
it
doesn't
work.
A
B
No,
I
make
sense
to
me-
I
mean
since
since
you've
written
and
tested
it
with
this,
and
it's
certainly
something
that
someday
will
happen.
A
B
I
had
a
question
for
both
of
you.
I
was
looking
over
and
sorry
for
dredging
up
more
comments
on
your
spec
tr
mateish,
which.
A
B
B
But
I
had
a
question.
I
didn't
comment
there
about
this
because
it
seemed
fine,
but
the
up
down
counters,
like
the
semantic
conventions,
will
define.
B
B
So,
for
example,
this
one
I
think
it
was
probably
incorrectly.
B
Called
an
async
counter,
we've
implemented
it
as
a
synchronous
counter,
because
we
want
to
you
know
each
request.
We
want
to
the
attributes.
We
want
to
tag
on
to
it,
but
in
general,
like
I,
was
reading
some
of
the
spec
sdk
spec
about
asynchronous
up
down
counter
or
or
no
synchronous
up
down
counters
and
the
spec
says
something
like
you
should
prefer
an
asynchronous
up
down
counter.
B
If
the
value
you
know
if
you
can
get
the
calculated
value
already
and
sometimes
especially
like
with
the
database
connection
pool
ones
in
some
cases,
you
would
count
you
would
calculate
it.
Synchronously,
like
the
active
database
connection,
you
might
calculate
it
synchronously
or
you
might.
The
database
connection
pool
might
give
you
those
metrics
already
and
you
would
just
get
it
asynchronously.
B
B
The
only
thing
I
could
think
of
was
like,
if
it
if
it,
if
we
didn't
spec
it
as
async
versus
sync,
then
like
views
might
be
like
there's
some
things
where,
like
views
take
into
consideration,
I
believe
async
versus
sync,
and
that
all
that
five-way
function
stuff
on
iraq,
that
jack's
been
talking
about
right.
Those
the
five
are
async
sync
histogram
async,
sync
up
down
counter.
B
A
C
B
So,
what's
the
difference
on
the
back
end,
because
I
thought
that
I
thought
they
both
exported.
They
both
had
the
same
export
format
like
at
the
otlp
layer.
Don't
they
convert.
A
C
B
B
B
B
Got
it
okay,
I'm
gonna
have
to
go
back
and
read
through
some
stuff,
but
that
yeah.
C
C
B
C
C
C
C
So
that
and
then
jmx,
I
guess,
is
a
good
example
again.
Maybe
that's
why
we
can't
spec
async
versus
sync,
because,
depending
on
what
you're
instrumenting,
you
might
only
be
able
to
use
this
asynchronous
approach,
but
we
would
still
prefer
to
use
the
synchronous
we
can.
So
that's,
probably
a
non
convention
specific,
just
best
practice
with
optometrics,
where
possibly
asynchronous
otherwise
use
asynchronous.
C
A
C
C
C
C
B
A
Issue
about
that
yeah.
I
also
noticed
that
there
were
already
meters
specified
as
instruments
specified
as
asynchronous
versus
synchronous.
So
I
just
followed
that
but
yeah
but
yeah
there's
a
deeper
story
behind
it.
It
seems.
B
C
C
C
C
C
C
B
Cool
well
thanks
for
waking
up
early
getting
on
daylight
savings.
We
missed
you.
A
B
Yeah,
I
just
feel
like
that
kind
of
stuff
has
so
much
weight
moment
like
it
takes
so
much
momentum.
Something
like
that.
So
I'm
not!
I
don't
have.
A
B
Yeah,
a
couple
of
the
some
of
the
states,
like
the
some
of
the
western
states,
have
signed
like
some
pact
to
drop
daylight
savings.
I
don't
remember
what
who
all
needs
to.
B
Like
last
week,
been
the
house
pass,
I
think
it
was
the
senate.
The
weird
thing.
B
Yeah
yeah
they're,
like
they
pulled
a
fast
one,
basically
and
like
slipped
it
in
when
there
were
hardly
anybody
there
paying
attention.
I
I
don't.
I
don't
understand
how
that
is
like,
but
even
possible.
B
I
guess
that's
why
we
have
checks
and
balances
that
it
still
has
to
go
to
the
house.
That's
true.