►
From YouTube: 2020-12-18 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
Didn't
terry
pick
because
we
deleted
everything
right
away
or
we
couldn't
cherry
pick
off
master
because
yeah
we
had
already
merged
this
code
to
to
decouple
metrics
from
the
api
yeah.
I
don't
know
if
you
took
a
look
at
what
the
palms
look
like
with
the
with
the
o3
with
o13
one
they're
a
little
interesting,
so
so
the
api,
the
o13
one
api,
depends
on
the
metrics
o13
0
alpha.
A
A
B
A
A
A
That
will
still
have
the
same
problem.
Right
sdk
won't
depend
on
sdk
metrics.
It
will
depend
on
api
metrics
right.
A
B
A
It
was,
it
was
easy
enough
to
do,
but
it
was
funny
I
put
the
pr
in
originally
I'm
like.
Oh
crap
jason
can't
merge
my
pr
or.
A
A
A
I
even
like
I
could
not
even
get
it
when
you
depended
on
the
bomb
to
ignore
the
metric
stuff
from
the
bomb,
and
I
don't
know
whether
it's
a
gradle
bug
or
a
gradle
limitation,
but
I
couldn't
get
it
you
to
be
able
to.
If
you
use
the
bomb
to
not
not
include
one
of
the
bomb,
the
transitive
bomb
dependencies
right,
because
the
bomb
didn't
depend
on
it
directly,
it
depended
on
it
transitively
anyway.
I
couldn't
get
that
to
work.
A
The
bomb
still
has
the
metrics.
No,
the
bombs
doesn't
have
metrics,
but
api
depends
on
metrics.
A
B
B
A
B
A
An
hour
got
it
sorted
so
and
anyway,
jason
said
he's
like
he's
like
who
wrote
these
release
instructions.
These
are
amazing.
I'm
like!
Oh,
that's
all
wrong.
So
anyway,.
A
They're
they
were
like
they
were
exactly
except
for
that.
One
thing
where
you
weren't
allowed
to
have:
no
no
shaws,
it
worked
great,
so
I
mean
worked
like
it
was
the
smoothest
release
I've
ever
done
in
my
entire
life,
so
it
was
really
good.
B
A
Awesome,
I'm
trying
to
start
pushing
everybody
towards
github
discussions.
I
had
a
field
engineer
or
some
sort
of
sales
engineer
reach
out
to
me
today.
It's
like
hey.
Can
you
talk
to
this
customer
about
this
this?
This
I'm,
like,
I
don't
know
about
that
stuff,
but
maybe
they
could
post
a
discussion
on
the
instrumentation
repo
right,
so
he's
gonna
start
pointing
that
way.
Experimentation.
A
A
B
A
Yeah,
so
I
I
created
a
few
categories
in
the
in
the
regular
java
repo
like
api
and
sdk,
but
also
support.
I
created
a
new
one
called
support.
Yeah
you
can
see
here
specifically
if
people
have
support
questions.
D
Yeah
I
like
that
yeah
that
actually
reminds
me
that
I
have
not
been
on
getter
and
I
haven't
had
my
cab
open
in
getter.
So
hopefully.
D
Yeah
a
bunch
of
unanswered,
oh
john,
john's
still
there.
Thank
you,
john.
A
Any
answer
to,
though,
so
I
I
thought
I'd
leave
that
to
y'all.
D
Yeah
shoot
yeah.
I
normally
leave
that
getter
like
tab
open
all
the
time,
and
so
I
at
least
get
like
occasionally
pinged.
B
A
So
there
is
a
question
about
azure
in
their
trask,
so
I
definitely
had
no
idea
what
was
up
with
that.
So
it
might
be
a
good
one
for
you
this
year.
Okay,
okay,
that
one
from
willie
shernborn.
D
Oh,
I
think
this
person,
I
think
this
person
tweeted
tweeted
me.
B
D
A
Yeah,
that's
true
forcing
user
to
be
over
there.
I
did
update
the
readme,
the
readme
and
the
java
repo
to
say
we're
recommending
people
use
discussions,
might
waste
the
link
together
and
we
can
just
remove
the
link
together.
There's
a
build
badge
for
getter.
I
didn't
remove
it,
but
I
did
make
it
a
little
bit
less
prominent.
A
It
used
to
be
the
first
badge
underneath
open
telemetry
and
I
moved
it
to
the
last.
I
don't
know
I
don't
know
if
that
makes
much
of
a
difference,
but
I
didn't
want
to
quite
delete
it
quite
yet,
but.
A
B
D
So
there
were
a
couple
of
things
in
here
that
I
wanted
to
write
down,
because
this
seemed
useful
getter
start
directing
people
to
github
discussions.
D
D
B
D
B
B
A
B
A
A
B
A
B
D
Doesn't
sound
like
it's
a
like
a
vendor
exporter
in
the
sense
that
we
wouldn't
allow
it
in
any
of
our
repos,
like
that
the
vendor
has
to
host
it
but
yeah
from
a
priority
perspective.
A
A
B
A
D
A
D
B
B
D
B
B
I
know
a
lot
of
I
mean
I
think
kinesis
people
use
it
like
because
it
can
then
just
wrap
the
messages
to
lambda
as
with
everything
in
aws,
and
so
I
think
the
way
you
would
use
this
is
to
export
to
it
so
that
you
have
sort
of
a
message
queue
in
between
and
then
in
your
lambda.
You
would
receive
it
and
then
finally
exported
to
zipkin
or
whatever.
B
D
Jason,
we
heard
you
had
fun
with
with
versions
today.
C
Yeah,
it
was
not
boring
anyway.
Yeah
I
mean
it's.
C
I
guess:
what's
I
guess
why
it
was
challenging
is
because
it's
a
bit
of
a
hack
like,
and
I
think
it
was
always
intended
to
be
a
hack
right
like
we're,
using
the
version
to
carry
additional
an
additional
tag
and
it's
pretty
non-standard
to
apply
a
different
version
across
a
repo
right.
Some
projects
are
one,
some
are
the
other,
so
hopefully
we
have
it
all
short
up
now,
but
did
require
also
going
through
a
patch
cycle.
A
Yeah
I
mean
the
real.
The
real
giant
issue
is
that
nebula
is
super
opinionated
and
it's
really
like
they
have
emphatically
said
you
get
one
thing
for
the
whole
project
and
you
you
can't
customize
anything
like
you
can't
even
like
take
one
of
the
modules
in
your
project
and
say:
don't
apply
nebula
to
this
we'll
do
it
ourselves,
like
they.
A
They
close
explicitly
because
they
don't
want
to
support
it
jason
and
I
spent
a
bunch
of
time
playing
around
with
it
on
couldn't
get
anything
to
work
yesterday
and,
unfortunately,
the
way
that
the
plug-in
works
and
we
we
we
worked
really
hard
at
this.
It
only
evaluates
the
version
one
time
and
one
time
only
like
it
doesn't
ever
even
call
into
the
into
the
extension
for
subprojects
like
it
does
it
once
so.
A
C
A
A
D
I
don't
know
nikita
is
nikita.
Incredibly,
we
have
at
least
somebody
else,
who's
not
doesn't
hate
gradle.
D
Yeah
I
was
so
happy
when
nikita
first
joined
the
instrumentation
project
because
he
like
that
was
one
of
the
first
things
he
volunteered
to
take
on
was
the
great
old
stuff,
and
I
was
like.
C
So
I
kind
of
have
a
side
topic
that
I
didn't
write
down,
because
I
apparently
can't
be
bothered,
but
it's
kind
of
spill
over
from
something
I
mentioned
very
briefly
this
morning
and
this
morning
was
packed.
Full
of
talk
like
I
feel
like
it
was
very
well
attended.
But
how
are
you
all
feeling
I'm
mostly
looking
at
trask?
I
think
how
are?
C
C
C
D
And
just
reviewing
you
know
going
through
and
reviewing
all
right,
so
I
can
give
you
a
little
bit
of
background
on.
If
you
haven't
seen
this
doc.
C
D
So
yeah,
because
there
is
a
there's
way
too
many,
and
so
I
don't
think
we
you
know,
maybe
at
the
very
beginning
we
thought
we
might
get
to
all
of
them.
But
we
quickly
realized
that,
honestly
that
we
were
only
going
to
target
p1s
for
ga,
okay.
C
C
D
D
Would
be
best
because
I
think
we
have
a
couple.
I
think
honorable
created
a
couple
that
were
not
required
for
ga,
but
but.
D
Okay:
okay,
oh,
but
we're.
B
Doing
this
and
this
one
actually,
I
do
want
to
change
it
to
record
for
g8,
because
we've
had
a
couple
of
bugs
because
of
those
tests
not
catching
it.
So
this
is
actually
a
pretty
important
issue,
though,
of
course,
matthias
being
awesome.
Like
his
muzzle
thing,
I
think
would
have
him
also,
but
I
also
had
to
add
the
disclaimer
to
my
testing
document.
I
want
to
go
to
that.
C
Yeah,
if
I
can
just
also
just
do
another
non-sequitur
speaking
of
testing,
I
guess
there's
the
it's
also
test
related.
So
there's
the
in.
I
think
that
flapping
test
was
a
dupe
of
another
issue
and
I
just
put
in
the
pr
for
it
so
see.
D
B
D
D
Yeah,
but
they
may
not
all
be
things
that
we
need
to
do
for
yeah,
totally
ga
yep.
D
D
Ga10
yeah,
so
honorable
this
morning
would
talk
through
the
java
agent
1-0
plan
that
we
discussed
on
tuesday
wednesday
asked
about
sort
of.
If
anybody
had
so
I'm
going
to
open
an
issue,
a
tracking
issue,
sort
of
documenting
what
we
consider
part
of
the
agent
stability
contract
being
like
configuration
settings,
spam
attributes.
D
Logging,
I
think
the
login
exporter
makes
sense
to
me.
I
like
I
liked
this
potentially
having
a
simpler.
I
mean
there
was
a
discussion
about
potentially
having
a
simpler,
just
debug
exporter
for
local
local
work,
which
wouldn't
be
under
a
stability
guarantee.
D
And
then
the
distro
I
added
here
the
distro
the
apis
will
not
will
still
be
alpha
also
at
that
time,
so
the
java
agent,
spi
and
api,
so
really
the
only
thing
that
so
yeah,
if
there's
anything
else
that
you
think
of
that,
should
be
under
that
stability.
B
B
D
So
there's
I
mean
there's
kind
of
a
lot
of
things
that,
like
span
processor
like
if
somebody
wants
to
do
a
spam
processor
that
requires
the
spi.
B
D
Oh,
I
know
which
oh
I
see
the
the
hotel.
B
D
Important
does
it
so
we
have,
I
forget,
so
we
never
moved
that
metric.
The
metric
at
or
the
span
exporter
factory
is.
B
B
I
mean
there's
no
dependency
hell
possibility.
So
in
that
sense,
even
if
it's
alpha,
it's
not
a
big
deal,
we
just
tell
people
and
write
in
custom
exporters
to
use
it
and
don't
worry,
because
you
can
keep
up
with
the
changes
right,
something
that's
like
it
wouldn't
cause
a
problem
with
the
user's
app.
So
it's
not
that
big
of
a
video
if
it's
alpha,
so
maybe
it's
okay,
yeah.
B
D
Okay,
yeah
we'd
had
I
thought
they
had
a
distro,
because
we've
had
a
somebody
opened
an
issue.
B
B
D
I
do
I
have
it
attached
to
my
calendar
item,
so
I
can
pull
that
up.
D
D
B
B
D
Oh
I
know
I
was
editing
here.
Stackdriver
is
using
this.
D
D
Yeah
and
then
pablo
asks
about
the
the
agent
debug
logging
that
it
how
it
only
goes
to
console
currently
and
asking
about
plans
to
go
to
support
files
and
nikita
said
that's
part
of
the
vendor.
Distro
support
issue
that.
B
D
D
Actually,
the
log4j
simple
does
support
lot
files.
I
should.
D
And,
let's
see
I
had,
I
wanted
matej
to
go
over
his
cool
things.
I
remember
that
we
had
never
shared
with
this
with
the
sig
group,
the
helper
class
name
stuff-
that
he
did
also
of
removing
all
of
that
and
then
talked
about
the
helper
resources
and
how
that
unlocks
kind
of
new
instrumentation
possibilities
for
us
via
spis,
which
is
cool,
and
there
was
one
question
about
assert.
Oh
yeah
pavel
I
mean
matej
mentioned
to
everyone
just
to
look
at
the
dot
like.
D
Was
discussing
like
I
would
like
I
kind
of
would
like
to
see
us
use
a
cert
inverse
more
just
because
it
gives
me
the
confidence
that
earlier
versions,
we
won't
accidentally
apply
our
instrumentation
to
earlier
versions
and
break
something,
so
it
does
feel
like.
Maybe
somebody
mentioned
possibly,
it
should
be
the
default
that
seemed
like
a
potentially
good.
I
don't
know
how
that
works
with
the
I
don't.
C
And
then
relic,
validator
also
has
that
as
the
default
just
for
what
it's
worth,
I
think
they
do
that
as
well.
D
And
then
nikita
I
had
I
had
to
queued
up
these
topics
for
different
people
to
talk
about,
because
there
was
some
good
stuff
that's
gone
in.
I
think
the
server
handler
stuff
is
as
much
as
I
hate
having
the
having
to
you
know
have
different
instrumentation
for
each
of
these
application
servers
is
actually
pretty
nice,
especially
the
use
case
of
our
span,
our
status
code
being
wrong
on
exceptions,
because
we
don't
know
what
the
server
is
going
to
do
with
it.
D
Call
it
error
handler
or
something
like
that,
so
that
I
had
pointed
I
don't
know
if
you
had
seen
in
one
of
those
the
tomcat
one
I'd
asked
if
that
resolved
the
that
issue
that
frank
spitalski
had
opened
a
long
time
ago
and
that
it
I
gotta.
B
D
I
gotta
learn
all
your
new
teammates
names,
john
and
jason,
there's
so
many
of
them
in
the
instrumentation
repo.
All
of
a
sudden.
D
Vladimir
okay,
vladimir
saur
yeah,
so
he
had
confirmed
that
that
it
resolves
that
that
issue
of
capturing
the
right
status
code.
So
that's
a
pretty
big
win
and
jason
plum,
created
a
tracking
issue
for
us
to
add
this
new
support
into
the
readme.
D
D
Yeah
and
then
this
is
another.
I
think
this
is
a
great
thing
that
is
going
in
lately
from
the
the
plumber
folks
also
and
lumping
nikita
in
there.
D
Our
smoke
test
matrix
is
growing.
Oh
yeah,
the
I
hadn't
seen
at
this
morning's
meeting,
because
I
had
a
reviewed
everything
that
happened
the
night
before,
but
your
honorary.
I
really
like
your
safety.
B
B
D
D
Yeah,
like
testing
against
ibm,
jvm
and
yeah,
and
even
not
necessarily
like
instrumenting
all
the
app
servers,
but
at
least
testing
on
all
the
app
servers
is
super
awesome.
D
We
did
talk
about
the
load,
the
overhead,
the
load,
testing,
stuff
pr,
and
so
the
do.
You
have
thoughts
on
what
to
have.
You
looked
at
the
numbers
coming
from
them.
D
Is
it
doing
like?
Is
it
produce
running
the
same
test
with
and
without
the
java
agent.
D
B
B
Okay
and
that's
what
they're
doing
on
the
python
repo.
I
don't
know
what
we
want
to
do,
but
there's
many
options
for
what
we
can
do.
I
think.
B
Exactly
yeah,
like,
I
think,
definitely
if
we
can
just
we
want
them
to
automatically
run
at
probably
weekly
or
something
those
are
planned.
Initially,
if
I
remember
correctly,
and
if
we
can
publish
numbers
with
each
release,
just
somewhere
sort
of
nice
to
have
those
to
refer
back
to.
D
Because
I
was
sharing
sort
of
my
experience
with
benchmarking
glow
root
in
the
past.
D
I've
like
actually
I
have
this
box
here
that
I
bought
just
to
like
have
this
isolated,
because
I
was
trying
to
I
used
aws
originally
to
like
spin
up
a
whole
bunch
of
cloud
instances,
and
it
was
just
the
variability
run
to
run
is
so
much,
but
even
on
a
standalone
box,
the
variability
is
very
high,
and
so
you
know
when
your
overhead
is,
you
know
under
a
percent.
B
C
C
Yeah
for
what
it's
worth
a
colleague
of
mine
at
new
relic
had
the
exact
same
experience
like
with
numbers
varying
and
yeah.
It's
exactly
like
driving.
B
D
B
I've
been
telling
people
on
my
team.
I
was
like
these
sdk
tests.
The
most
important
thing
is
to
catch
disasters
like
if
they
have
a
spike,
we'd,
probably
notice
it
and
that's
a
good
safeguard
to
have,
but
otherwise
comparing
the
numbers
to
each
other.
That's
not
going
to
help
with
the
numbers
we're
talking
about.
Definitely
so
that's
probably
how
it's
like
this.
The
most
important
thing
here
is
just
to
make
sure
we
don't
introduce
a
huge
regression
blow
it
up.
Yeah,
yeah,
yeah.
D
Yeah,
because
if
we,
you
know
all
of
a
sudden
overhead's
15
yeah,
we
wouldn't
notice
that
that
would
yeah
jason.
It
looks
like
so
on
this
one.
What
is
batik
was
mentioning.
Oh,
how
do
you
say
it
patek?
Thank
you
that
he
was
also
planning
to
get
memory
data.
D
Well,
that's
run
time
memory
data
we
were
also
talking
about
sort
of
the
oh.
I
guess.
If
we
have
run
time
memory
data
we
can
compare
with
and
without
the
agent.
C
That
312
that's
linked
up
there.
I
think
that
that's
like
a
really
old
one
and
concerned
primarily
with
memory,
and
I
think
they
were
looking
at
two
different
things,
right,
jmh
and
then
also
pet
clinic,
so
like
something
with
jmh
to
kind
of
get
like
two
different
types
of
memory:
interest.
B
B
A
D
Oh
yeah,
this
is
the
this
is
the
box.
I
have
sitting
down
near
me
still
to
this
day.
C
Yeah,
so
there's
a
couple
of
things
that
I
thought
would
be
when
I
was
looking
at
that
312
so
funny
how
low
that
number
is
3
12..
I
was
thinking
that
maybe
there's
an
opportunity
there
to
also
use
jfr
to
analyze
some
of
the
memory
allocation
rate
and
to
see
if
allocation
rate
is
impacted
at
all
too,
but
I
mean
it's
gonna
require
somebody
to
to
carve
out
some
time
to
do
that.
Work
and
I
don't
know.
D
Yeah,
I
will
say
that
during
I
spent
a
lot
of
time
tuning
to
get
these
numbers,
and
that
was
my
go-to
tool
was
jfr
with
the
the
async
profiling.
D
Async
stack,
trace,
stuff
yep,
that's
very,
very
good
tool,
but
but
it
takes
a
lot
of
time
to
dive
through
that
data,
and
I
actually
wrote
processors
to
process
the
output
of
that
to
parse
out
the
wherever
it
was.
You
know,
found
a
stack
that
had
the
agent
code
running
at
that
time,
since
it's
already.
B
D
Such
a
small
portion
of
the
overall
app
so
that,
then
you
can
look
at
that
narrow
profile
but
yeah.
It's
time-consuming
stuff
for
sure.
A
D
B
D
Sure
yeah,
I
can
definitely
write
down
sort
of
my
the
is
batek
looking
at
profiling.
It
also
not
just
running
the
numbers.
Is
he
interested
in
in
profiling?
It
and
tuning.
B
I
think
he
would
be
interested
in
having
the
hardness
for
us
to
be
able
to
debug
issues.
I
don't
know
if
he
would
go
as
far
as
tuning
things,
but.
D
D
Yeah
next
week,
so
we're
gonna
cancel
both
thursday
meetings
next
week,
but
keep
the
tuesday
meeting
on
the
calendar.
D
I
need
to
do
that.
I
always
need
to
go
back
and
I
don't
know:
do
it
up
the.
B
D
And
then,
yes,
a
brief
update
from
john
about
how
zero
thirteen
zero
breaks
all
of
the
sdk
in
the
same
way,
zero.
Ten
zero
broke
all
of
the
api.
B
A
It's
we
didn't
repackage
very
much,
it's
not
quite
as
bad,
but
we
did
rename
all
the
core
classes
so.
B
A
Yeah,
I
really
want
to
go
into
the
examples.
I
started
updating
the
examples,
which
is
how
I
found
out
13
but
and
remove
all
of
the
global
the
usages
of
the
global
in
them.
That's
going
to
that'll
take
a
little
bit
of
time
to
sort
out
that
and
it
might
actually
uncover
some
issues
with
how
we
do
things
like.
I
think
already
it's
a
little
like
it's
super
unclear
even
to
me
how
I'm
supposed
to
do
this,
so
we
definitely
have
some
have
some
work
to
me.
B
B
B
A
Was
it
still,
oh
yeah?
That's
why
I
didn't
get
in
because
it
was
waiting
if
he
doesn't
respond
on
that
I'll,
just
override
and
run
yeah.
B
A
B
My
plan
is
like
once
that's
in
then
going
through
all
of
our
unit
tests
stop
using
the
globals
and
unit
tests,
replace
them
all
with
builder,
and
their
own
test,
like
having
tests
using
static.
Global
state
is
just
bad
anyways
right,
it's
confusing
how
they
interact
with
each
other.
So
I'm
looking
forward
to
that,
but
I
need
my
pr
to
be
emerged
yeah.
I.
B
B
A
B
A
Handles
all
the
auto
configuration
as
long
as
it's
really
easy
for
what
am
I
thinking
of
yeah
as
long
as
we
make
sure
that
when
you
get
access
to
the
global
as
an
instrumentation
author
that
it
does
the
no
op
thing
that
it's
supposed
to
be?
I
think
that's
the
you
know,
that's
just
the
main
thing.
B
And
and
using
spi
to
find
implementations,
and
so
my
thinking
is
that
this
rapper
should
implement
all
the
environment
variables
and
open
from
respect
rather
than
having
the
sdk
do
so
so
that
the
sdk
is
a
programmatic
thing,
and
so
even
the
backspan
processor
settings
like
I
feel
as
if,
instead
of
having
magic
in
the
sdk,
keep
it
all
in
this
configuration
layers
that
spring.
If
they
don't
want
to
use
it,
they
don't
have
to.
A
A
B
A
Really
hard
for
me
to
know
who's
like
relying
on
spi
the
spi,
for
if
anybody
yeah,
I
just
don't
know
if
we
were
to
break
it,
would
people
start
yelling,
it's
it's
I
I
don't.
Even
I
have
no
clue
might
be
worth
asking.
You
know
who
probably
would
have
an
opinion
is
christian,
so
oberon,
zero,
zero
and
yeah.
B
D
B
B
Yeah,
so
that's
a
good
call
out.
I
know
probably
using
this
so
like
I'm
also
wondering
whether
in
the
agent
then
we
would
use
this
auto
configuration
artifact
and
remove
all
of
our
code.
This
is
related
to
my
issue
or
some
discussion.
I've
had
a
while
ago,
where
there's
some
magic
in
the
sdk
there's
some
magic
in
the
agent.
It
should
all
be
in
one
of
them,
and
maybe
it's
just
the
time
to
make
that
happen.
C
A
All
right,
I'm
gonna,
I'm
gonna
fly
but
good,
hopefully
chat
online
and
see
you.
A
B
B
D
You
on
monday,
if
I
don't
have
it
done
on
monday
or
on
monday
morning
or
on
sunday
night
I
will-
I
will
pass
along
whatever
is
not
done
to.
B
D
B
D
Which
was
really
helpful,
I
think.
B
D
That
the
merge
conflicts
have
not
been
a
problem
and
I
did
build
out.
D
D
D
D
Mind
all
right,
yeah
anything
else
came
up.
D
B
B
D
B
D
D
I'm
good
with
that
yeah.
I
was
wondering
if
at
first
I
thought,
maybe
the
you
know
how
in
our
in
the
instrumentation
repo
you
started
the
staging
repo
like
in
the
beginning.
B
B
I
don't
know
if
that
works
for
snapshots.
That's
the
problem.
If
it
does
that'd
be
great,
maybe
I'll
look
into
if
it's
possible.
B
B
D
But
I
I
agree,
the
I
don't
think
that's
going
to
be
a
big
problem
for
us
and
and
if
somebody
really
need
wants
to
pull
it
in
earlier,
they
can
deal
with
that
mess.
Yeah.