►
From YouTube: 2020-09-11 Java Auto-Instrumentation SIG
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
Oh,
you
know
it
looks
like
the
apocalypse,
but
and
it
is
the
apocalypse.
A
B
A
Upstairs
and
oh
man,
the
smoke
got
into
our
house
through
the
pretty
sure,
through
the
chimney,
flue.
B
A
Because
the
living
room
just
smells.
C
A
C
A
C
A
John
and
I
both
live
in
portland,
so
we're
commiserating
with
the
do
I
wanna.
Let's
just
look
out
the
window,
I
don't
think
you'll
be
able
to
no.
I
can't
really.
A
C
C
A
D
Yeah,
that
was
what
it
was
like
in
sydney
actually
in
summer
last
summer,.
D
Yeah,
so
there
were
fires
in
new
south
wales,
which
is
the
state,
that's
the
nissan,
and
yes,
it
was
all
over
and
yeah.
It
was
really
red
as
well.
D
I
don't
know
if
the
photos
have
photoshopped
some
of
the
ones
I've
seen,
but
the
red
in
sydney
wasn't
as
like,
saturated
as
the
red
that
I've
seen
in
some
of
the
photos
right
now
so.
B
A
B
Portland
is
it's
just
kind
of
grayish
brown
everywhere.
C
A
A
Let's
see
we
had
a
small
meeting
today,
a
small
number
of
people,
maybe
the
holidays,
not
sure
so
p1
issues
back
to
tracking
those
one
one
down
last
in
the
last
week,
13
to
go.
We
seem
to
be
kind
of
up
one
down
one
up
two
down
two,
let's
see
stuff
that
people
wanted
to
chat
about.
A
Sergey
was
the
wanted
to
chat
about
the
amount
of
like
you
turn
on
debug
level,
and
it's
like
you
go
from
nothing
to
everything
yep,
so
we
all
seem
to
agree
that
some
of
that
could
definitely
be
bumped
to
trace
logging
and
so
sergey
or
somebody.
The
thought
was
just
go
ahead
and
start
doing
that
and
then
at
some
point,
once
we
kind
of
get
a
feel
for
what
should
be
trace
versus
debug.
It
would
be
nice
to
have
a
doc
about
that.
C
Yeah,
if
this
is
controlled
by
some
flag,
that
could
also
solve
that
worn
on
instrumentation
failure
on
tests
issue
quite
nicely
also
as
long
as
those
things
are
like
I'm
guessing,
what's
going
to
be
less
is
left.
Is
that
the
instrumentation
failures
or
debug
everything
else
isn't,
in
which
case
it's
just
bumping
that
up
to
warn
during
tests.
B
A
C
C
So
I
guess
it's
not
so
awesome,
because
I
noticed
actually
that
now
that
we
I
mean
this
might
have
been
my
fault,
I'm
not
sure,
but
if
this
was
always
the
case,
but
since
our
tests
have
both
logged
back
and
self
f4j,
simple
log
back
wins
and
most
of
our
tests
don't
log
anything,
that's
really
annoying.
Actually
so.
C
Or
they
shouldn't
be
using
log
back.
Does
that.
E
C
C
C
C
I
don't
know
he
did
that
in
code,
okay,
yeah,
that
would
save
my
life
a
lot
yeah,
and
I
I
I
don't
remember
that
being
the
case
I
was
like.
I
couldn't
think
of
any
reason
that
would
have
changed
and
I
was
like
why.
Why
do
I
have
to
copy
this
log
back
xml?
And
that
was
happening
these
past
couple
of
weeks,
but
I
think
this
solves
it.
Yeah.
A
It
probably
looks
like
probably
me
package
yeah,
john,
all
our
package
and
module
renaming
isn't
only
causing
you
pain.
B
B
B
A
Yeah
really
cycles
are
never
great
speaking
of
release
cycles.
Honorable,
thank
you
for
making
zero
eight
and
nice
release,
notes
you
and
you
first
nikita
and
now
you're
raising
the
bar
further,
like
hey
yeah,.
C
Well
yeah.
I
just
copied
the
same
styles
arm
area
as
I
do
for
everything,
and
I
did
ask
them
like
do
they
do
anything
automated
and
it's
like.
No,
it's
all
manual,
and
so
I
was
like
yeah,
so
I
went
through
the
list
but
like
I
would
like
to
have
a
tool
that
at
least
collect
stuff
and
then
rewrite
the
wording.
So
it's
more
friendly
and
natural,
but
because.
B
There
are
people
who
believe
that,
like
I
know
that
this
has
been
talked
about
at
new
relic,
where
there's
a
format
you
can
use
for
all
your
commits
that
have.
F
C
A
B
A
How
did
you
get
the
list
of
names.
C
Yeah
just
check
out
all
the
portrait
icons.
This
part
I
haven't
mentioned
yet
and
then
going
through
the
issues
list
was
harder.
I
sort
of
went
through
them.
I
remember
a
lot
of
the
conversations,
so
that
also
is
more
based
on
like.
Was
there
anyone
not
commonly
in
this
repo
that
responded
to
this
comment?
I
think
I
remember
one
and
clicked
and
copied
the
name
and
said.
A
A
Wondering
how
you
got
that?
Because
some
of
these
are
not
pr
like
the
pr
ones
are
easy
to
pull.
But
the
issue.
A
All
right
well,
yeah,
I
mean
I
think
this
is
a
good
task
for
us
to
rotate,
which
means
it's
me
I'm
up
next,
I.
C
A
C
That
one
I
do
by
a
special
gradle
task.
We
don't
have
that
in
our
repo.
So
it's
a
bit
easier,
but
in
the
x-ray
sdk
one
I
have
a
preparation
test
which
just
does
a
regex
replacing
all
the
versions.
B
B
C
A
Yeah,
it
looks
like
we
grew
by
about
four
three
and
a
half
migs.
A
Make
a
note:
zero,
eight.
A
Yeah
we're
thinking
for
hotel
propagators,
especially
since
we
use
it
just
didn't,
want
it
to
get
changed
to
have
some
other
semantic
meaning
or
something
so
yeah.
I
volunteered.
If
to
open,
I
should
check
out.
I
don't
think
I
got
pinked.
A
A
Is
now
what
happened
with,
I
thought
that
the
we
were
pulling
x-ray
and
ot
tracer
out
of
the
sdk
repo.
Did
that
happen.
C
If
someone
deletes
them,
then
I'll
have
to
re-upload
them
somewhere
else,
but
I
my
whole
understanding
of
the
vendor
conversation
is
that
at
this
point
each
language
is
doing
their
thing
and
unless
there's
a
huge
reason
to
not
do
it,
I
don't
know
if
the
languages
are
going
to
delete
these.
That's
my
current
understanding.
I
don't
know.
If
john
has
any
thoughts
there.
C
B
C
A
Propagators
felt
like
reasonable
to
me
to
be
in
here.
Even
for
people
say
migrating
off
of
x-ray
and
on
to
you
know
something
else.
E
F
A
Oh
yeah
yeah.
I
just
saw
that
as
I
was
catching
up
on
notifications
and
yes,
I
I
approved.
B
Carlos
is
the
only
hold
out,
and
I
don't
I
think,
he's
worried.
So
this
is
what's
crazy.
He's
worried
that
you
would
be
maintainer
both
on
instrumentation
and
regular
java,
but
he's
a
maintainer
of
the
specs
and
javascript
he's
complaining,
yeah.
C
Pogba
actually
recommended.
I
should
nominate
myself
for
tc
then
I
was
like
I
don't
think
so.
B
C
For
me,
it's
sort
of
the
opposite.
Like
I
mean
I
still
work
at
amazon,
and
I
know
that
if
I
spend
100
of
my
time
on
open
telemetry
forever,
I
never
will
get
promoted.
So
I
don't
want
to
lock
myself
into
that
road
just
java.
I
think
I
can
manage
time
wise,
but
if
I
add
a
tc
to
that,
I
don't
see
myself
having
a
career
at
amazon.
C
Just
do
something
other
people
find
cool
and
sdks
are
not
cool.
I
don't
know
what,
like
even
at
google.
It
was
similar
like
if
you
work
on
the
client
side,
you
don't
get
promoted
past
a
certain
point,
just
not
there's
no
hard
rule.
It's
just
sort
of
what
people's
perceptions
are
of
the
complexity
of
that
work
and
whatnot,
and
they
assume
it's
not
very
complex
and
you
can't
become
a
senior
principal
engineer
or
something
based
on
that.
C
E
C
C
C
A
A
No,
I
no
I
I
will
I
I
I
owe
the
I
fessed
up
to
not
to
owing
the
spec
a
pr
one
of
these
days,
so.
C
C
Yeah,
I'm
a
lazy
guy.
Someone
tries
because,
like
I
don't
know,.
B
Well,
I
I
luckily
decided
to
trust
but
verify
and
then
discovered
it
wasn't
working
at
all
and
yeah
anyway,
and
then
in
the
process
discovered
that
the
the
readme
had
not
yet
been
updated
for
the
debug.
The
logging,
like
the
logging
logo,
also,
and
then
I
was
that
was
the
part
that
actually
triggered.
C
C
Whether
I
should
have
just
like
committed
that,
rather
even
sending
a
pr
like
I
did
it
right
away,
because
I
knew
it's
so
important,
because
I
also
get
that
debug
flag
wrong
on
our
snapshot
builds
every
time
because
I
copy
paste
from
there
but
yeah
that.
Maybe
I
should
just
committed
that,
rather
than
sending
a
pr
for
it,.
B
Well,
this
next
release
is
going
to
break
all
sorts
of
stuff
for
you
all
so
yeah
are
you
working
on
snapshots
again,
so
you
you
get
to.
B
B
B
I
have
the
full
pr
now
in
for
that
changing
to
the
typed
keys.
B
It
actually
really
all
I
really
care
about
is
because
it
most
of
that
pr
is
just
enormous
test
changes,
because
tests
create
attributes
like
crazy,
so
I
had
to
change
enormous
amounts
of
test
code,
but
the
actual
api
service
change
is
pretty
darn
small.
So
maybe,
if
you
could
just
like,
take
a
look
at
the
api
surface
yeah
before
and
like
put
any
comments
before
tomorrow
morning,
because
bob
didn't
put
a
comment
on
it
then
he's
like
well,
I
see
why
you
might
like
this.
I
don't
like
it.
Let's
talk
about
it!
No,
no!
B
B
Just
the
api
stuff
is
the
important
stuff
yep
and
I
figured
out
a
way
by
the
way
to
get
to
do
the
consumer,
the
attribute
consumer
with
a
so
it's
still
a
functional
interface,
one
method
and
it
works.
B
So
after
all
that
debate
back
and
forth
on
the
original,
whether
we
do
the
eight-way
interface
or
the
single
one,
I
got
one
that
works
with
the
single
one
and
I'm
reasonably
happy
with
it.
C
B
C
Is
it
just
about
java
agent
versus
auto
at
this
point?
Is
the
actual
structure
settled
or.
A
Yeah
during
the
build-
that's
not
that
that
should
be
straightforward,
ish
gradle
wise.
The
other
thing
that
I
was
a
little
concerned
about
is
the
tests
the
bootstrap,
because
we
have
have
the
bootstrap.
We
set
the
bootstrap
class
path.
A
A
C
C
A
Yeah,
I
guess
I
would
I
mean,
as
sort
of
like
it's
separate,
because
it's
a
separate,
separate
target
audience.
C
A
C
A
C
C
C
C
A
A
A
E
C
A
C
C
A
Yeah,
I
think
I
think
I
had
leaned
towards
auto
at
that
time,
as
we
had.
C
A
A
A
C
C
A
A
And
the
same
with
the
the
main
yeah.
C
C
A
This
is
you're
tired
of
our
15,
but
the
bike
shed
needs
to
hold
three
bikes
and
be
blue.
C
A
Time:
okay,
I,
like
it
okay,
important
question,
plural
or
singular.
C
Like
I
think
it
used
to
be
singer
and
became
plural
at
some
point.
If
I.
C
A
All
right,
let
it
be
so.
A
B
So
my
question
is:
it
was
my
my
brain
got
rattled
around
when
you
were
talking
about
sbi,
so
we
have
an
issue,
an
open
issue
in
the
java.
I
guess
it's
really
part
of
the
sdk
about
osgi
and
osgi
and
spi
really
don't
work.
They
don't
work
well
together
at
all,
and
so
there's
a
user
who's
using
some
app
server.
B
I
don't
remember
which
one
and
they
built
an
sbi
bundle
and
it
doesn't.
It
doesn't
work.
The
whole
obfuscated
thing
blows
up
and
I
think
it
just
doesn't.
It
does
like
the
sbi
system
doesn't
work
when
you're
in
an
osgi
bundle.
So
we
have
an
open,
a
very
old
open
issue,
jib
to
build
osgi
friendly
configuration
something
so
I
don't
really
know
sgi
at
all,
but
we
have
an
open.
We
have
an
open
issue
about
osgi
and
I
was
just
curious
whether
anyone
knew
anything.
C
C
B
Right
this
is
the
the
issue
where
they're
broken
running
on
yeah
and
then
there's
an
old,
an
old
or
an
older
issue
asking
for
osgi
support.
So
I
know
nothing
at
all
about
osgi,
yeah.
C
B
A
C
A
No
well,
if
it
is
so,
it
is,
but
in
in
our
own
class
loader
right.
So
we
we
load
we're
sort
of
underneath
even
osgi
right,
so
osgi
doesn't
muck
around
in
our
stuff
and
what
we'd
have
to
do.
Is
instrument
class
loaders
to
force
all
class
loaders
osgi
class
loaders
to
look
to
our
bootstrap
classes
and.
C
C
C
A
C
C
B
C
It
got
it
yeah
yep
here
it
is.
F
A
Well,
there
was
a
any
thoughts
on
the
java
7
stuff.
A
Oh
sorry,
you
have
to
read
from
bottom
up.
C
C
A
Yeah,
so
that's
what
sarah
asked,
if
would
be
so
sarah
emailed
heloise,
eloise
and
asked
asked
said
we
talked
through
a
few
options
or
asked
how
many
of
their
customers
are
running.
Java
7,
we
talked
through
a
few
options,
including
forking,
the
forking
now
and
leaving
a
java
7
version
for
dynatrace
to
maintain.
A
So
I
think
we
just
need
so
alois
said
that
they
do
need
java,
7
support
and
said
something
about
having
a
dedicated
fork
sounds
like
a
good
approach.
Other
companies
like
zalando
and
most
likely
ibm
would
be
interested
as
well.
A
And
bogdan
asked,
can
you
tell
me
if
the
you
need
to
support
manual
instrumentation
for
this
or
just
auto
instrumentation
using
the
agent
by
manual
I
mean
somebody
going
and
manually
instrumenting
their
java
7
app
with
open
telemetry.
A
A
And
ted
asked
is
it
this
was
just
today.
I
didn't
not
in
my
pace
ted
asks.
Is
it
possible
for
dynatrace
to
support
java
7
manual
instrumentation
using
a
legacy
tracer
agent?
This
is
the
plan
at
lightstep
and
new
relic.
F
C
B
C
A
It
sounds
like
there's
not
like
yeah
a
bit
of
I'm
not
super
duper
clear
about
that.
It
is
like
the
api
and
the
sdk
that
they
would
be.
A
Okay
with
that,
we
could
fork
that
they
would,
you
know,
maintain
obviously
have
to
maintain
that
java
7
fork
and
then
the
main
project
could
move
forward
on
java
8..
C
A
Okay,
so
I
don't
know
who
will
from
dinah
trace.
I
could
ask
on
this
chain
just
very
sort
of
explicitly.
B
Haven't
seen
anyone
from
dina
trace
involved
in
metrics
at
all,
okay,
I'm
trying
to
think
giovanni
hasn't
been
in
there
and
christian
is
the
other
main
contributor
and
neither
of
them
have
been
in
metrics
at
all.
B
A
E
C
So
now,
when
spinning
it
seems
to
load
from
the
parents
or
something
if
you
look
at
the
test,
failure
see
which
test
is
failing,
yeah
and
I
can't
think
of
anything
I
did
unless
maybe
there's
some.
Actually.
I
didn't
try
searching
for
the
old
name
and
seeing
if
there's
a
string
that
I
had
to
update
that
sounds
like
it
could
have
been.
C
C
Yes,
so
our
boot,
like
our
test
class
path,
has
all
the
exporters
or
something
like
that,
and
so
here,
if
it's
loading
more
than
one,
I
guess
it's
not
all
of
them,
it's
two
of
them,
but
if
it's
only
more
than
one
that
means
I
got
it
from
the
test
class
path,
which
is
the
parent
of
the
of
something.
I
guess
that's.
C
A
No,
so
you
had
made
a
comment
to
maybe
it
was
matthias
about
the
helper
class
name.
A
It
and
yeah,
so
I
was
pointing
so
unfortunately
like
I
so
I
had.
I
thought
that
this
was
going
to
be
a
great
thing,
yeah
right,
but
after
testing,
even
with
the
fully
qualified
name,
intellij.
C
C
C
A
Mapper
class
loader.
A
For
john's
for
john's
exporter,
jar.
C
C
About
that
yeah,
so,
like
I
one
problem
with
this
pr,
I
don't
know
if
anyone
would
have
noticed.
If
I
didn't
say
it
right
now,
but
like
the
unit
test
doesn't
test
anything
anymore,
because
at
least
until
this
also
sort
of
depends
on
building
an
agent
jar
for
our
tests.
C
Until
then,
the
I
can't
think
of
any
way
to
hide
the
resource
from
the
tests
like
it
would
be
on
the
class
path
by
being
part
of
the
main
source
set,
and
I
don't
think
there's
any
way.
At
least
I
couldn't
come
up
with
any
way
to
not
have
that
resource
available
in
the
test
class
path
as
well
and
okay.
A
Yeah,
honestly,
for
my
experience
with
java
agents,.
E
C
A
Yeah,
I
kind
of
wonder
sometimes
about
you
know
the
the
unit
tests
that
we
have
for.
Like
the
tracers
I
mean
there's
we
have
so
many
integration
tests
that
test
all
of
that
stuff.
C
Yep
yeah
yeah,
okay,
the
other
there's
only
one
point
undecided
point
here
is
like:
should
we
dedupe
our
resources
versus?
What's
in
there
class
loader
or
not,
or
I
probably
d-dupe,
but
it
is
quite
expensive
to
do
that
service
order
shouldn't
be
on
the
hotpath
anyway.
So
it's
probably
okay,
anyways.
A
So
remind
me
so
so.
C
A
So
you're
thinking
like
in
the
case
of
where
one
class
load
where
it's
like
chained
and
a
class
loader
calls
another
class
loader.
So
we
might
inject
it
downstream
and
then
inject
it
again.
C
C
C
The
content-
that's
so
I'm
doing
that
here,
there's
apps
so
again
and
the
main
I
didn't
do
this
at
first.
I
added
it
because
there
was
no
way
to
get
the
test
to
not
fail.
If
I
didn't
do
because
it's
on
the
test
class
path.
C
A
C
E
A
C
C
A
C
A
D
I
didn't
get
too
much
time,
but
I
did
take
a
brief
look
at
the
ticket
that
you
the
issue
that
you
sent
me
yeah.
I
think
it's.
I
was
actually
just
looking
at
the
like
augmenting
it.
So
I
think
the
there's
it's
saying
that
the
sorry
I'm
just
bringing
up
the
issue.
D
D
D
Yeah
cool
yeah
and
then
just
basically.
C
Like
all
of
our
properties
start
with
hotel,
so
that's
the
common
thing
and
then,
after
that
the
name
isn't
as
important.
I
think
okay
and
hotel
just
all
in
logist,
so
the
properties
are
hotel,
dot
and
then
some
like
prefix
dot
name,
that's
your
format
and
then
the
environment
variable
is
hotel,
underscore
like
converting
dot
separators
to
underscore
separators
with
uppercase.
A
Yeah,
the
convention
is
for
the
system,
properties,
hotel,
lowercase
and
then
do
we
do
mix
case
after
that
or
dots
for
everything
or
varies.
C
D
Sorry
so
hotel
and
then.
C
C
A
C
A
So
we
don't
do
and
we
actually
don't
do
this.
This
is
all
coming
from
the
sdk.
Now
the
there's
that
abstract
configuration.
A
Oh,
how
does
what
about
the
resource
piercer
batch
span
processor
this
this?
Does
it
also.
A
C
Yep
it's
using
conflict
builder,
so
so
yeah.
If
you
can
look
at
this
class,
called
trace
config,
that's
a
good
one.
That
has
an
example
of
how
you
might
define
it
has
a
lot
of
fields.
You'll
only
have
one
but
it'll
give
a
pattern
of
how
you
can
define
the
properties:
okay,
cool
yeah
and
yeah.
It's
a
bit
weird
for
trace
config
to
be
using
this
export
config
builder,
but
we
already
have
that
pattern,
so
you
don't
have
to
worry
about
it.
D
Also,
you
were
talking
about
sbis
and
osgis.
What
what
are
those.
A
Yeah
so
spi
is
stands
for
service
provider
interface
and
what
in
the
java
world
would
it?
Basically,
let's
see,
am
I
sharing
my
whole
screen?
I
think
I
am
cool,
so
we
can
probably
he's
the
best
to
go
to
some
code.
A
And
telemetry,
so
the
open,
telemetry
class
that
you
use
to
like
get
tracer
get
meter
all
that
good
stuff
when
it
instantiates
the
when
it
starts
up
it
loads
these
from
the
spi
for
this
and
how.
A
Yeah
in
the
java
world,
what
that
means
is
calling
this
serviceloader.load
and
it
takes
this
class
and
what
it
does
is.
It
looks
for
a
special
class
and
doing
a
special
file
on
your
resources.
C
A
A
A
So
if
you
ever
crack
open
a
jar
file-
and
you
see
meta-imp
services
directory
and
some
files
under
there-
this
is
a
the
class
name.
Is
it
always
an
interface?
It
can.
A
That
these
and
it's
telling
you
some
specific
subclass
to
instantiate
for
that
and
there's
two
that's
interesting,
so
you
get
back,
I
guess
service.
I
didn't
even
realize
that
this
gives
you
back
a
list
of
implementations.
A
C
A
Oh,
oh,
I
see
yes,
okay,
yeah
so
yeah,
so
in
my
world,
normally
it's
always
just
been
one
class.
A
A
A
Okay
and
then
osgi,
nobody
knows
what
that.
C
A
About
how
I
yeah,
I
guess,
maybe
the
the
the
way
that
maybe
the
easiest
or
a
way
of
thinking
of
osgi,
that
I
know
the
osgi
people
don't
like
is
that
java
9,
with
the
the
java
modules
system
introduced
in
java
9,
was
sort
of
is
sort
of
to
address
the
same
things
that
osgi
is
sort
of
this
module
platform.
A
E
C
Kidding,
okay,
I
have
to
get
going
cool
yeah.