►
From YouTube: 2020-08-07 Java 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).
A
B
A
So
I'm
still
confused
about
why
you
and
nikita
are
so
intensely
interested
in
having
this.
This
fan
data
thing
I'm
like
it's
like
four
clicks,
an
idea
and
you
can
make
a
rapper
it's
like.
B
B
A
A
A
Yeah,
so
I'm
taking
the
day
off
after
the
the
java
meeting
tomorrow
morning,
so
I
figure
I
should
still
probably
show
up
for
that.
B
Are
you
able
to
have
any
celebrations,
so
I'm
going
to
dinner
with
my
wife
tomorrow
night.
D
A
B
A
Saturday,
my
wine
buddies
and
I
are
having
a
little
party,
I'm
opening
up
a
bottle
of
1955
cote
d'ir
rouge,
which
I'm
very
excited
yeah.
So
that's.
A
B
C
A
B
C
Do
you
do
you
still
go?
Do
you
have
ac
in
your
place.
C
C
B
C
It
reminds
me
of
the
the
last
company
I
worked
for
was
based
out
of
houston,
texas,
and
so
it
was,
I
would
go
when
I
would
visit.
I
would
stay
in
a
hotel
near
nearby,
like
too
close
to
like
take
a
cab,
but
far
enough
that
and
and
a
perfectly
decent
walking
distance.
Except
you
know
the
houston
humidity
pretty
much
any
time
except
winter.
A
Have
you
had
a
chance
to
thoroughly
scour
the
new
relic
agent
code
base,
yet.
B
B
B
C
Talk
about
http
today
and
the
jsons
from
the
new
relic
agent
team,
because
I
was
like
ooh.
We
can
look
and
see
how
what
you
guys
did
and
they're
like.
D
C
B
B
Maybe
then
the
first
topic
is
one:
that's
come
to
mind
in
best
practices
of
writing
agents
or
at
least
defining
our
own
best
practices
like
one
is
not
using
private
apis
unless
absolutely
necessary.
Like
these
sort
of
practices,
I
don't
think
we
have
them
well
defined
yet,
but
they
probably
exist
in
our
minds
right
and
that
might
be
something
to.
B
So
that's
what
those
versions
do
tend
to
make
so,
and
I
don't
there
might
be
a
lot
of
like
when
there's
a
huge
amount
of
value
in
a
private.
You
should,
of
course,
do
it.
I
think,
but
of
course
I'm
sure
we
have
these
sort
of
practices
like
in
general
straight
towards
public
and
these
sort
of
things-
and
I
think
you
mentioned
the
static
import
thing
that
might
also
be
included
in
our
best
practices
guide
for
writing.
Open
telemetry
java.
B
C
C
C
And
because
it
broke
you.
B
A
I'm
wondering
I
haven't
got,
I
haven't
actually
looked
and
seen
what
we've
put
out
there.
I
wonder
whether
we
put
out
our
our
quasi-legal
mule,
like
mulesoft
instrumentation,
where
we
we
kind
of,
had
to
reverse
engineer
their
server,
to
figure
out
how
it
worked,
which
is
definitely
not
not
necessarily
a
hundred
percent
kosher.
D
A
C
Yeah
we
had
somebody
ask
they
wanted
to
use
our
apm
to
monitor
cassandra
and
they're
like
well.
Just
because
it's
running
java
doesn't.
A
If
you
look
at
it's
actually
kind
of
interesting
to
take
a
look
at
the
way
we
do
our
instrumentation,
we
have
this
thing
called
the
weaver,
which
kind
of
does
uses
annotations
to
do
like
if
you
annotate
the
code
and
then
the
the
weaver.
You
can
see
here
that
at
weave
annotation
on
line
21
yeah,
then
our
we
have
this
weaver
code
that
actually
looks
at
the
annotations
and
deals
with
like
matching
up
the
method
names.
So
right
here,
you
can
see
like
under
125.
That
get
connection
is
the
same.
A
A
D
A
C
Yeah,
I
was
just
trying
to
see
if
you,
if
it
was
doing
anything
postgres
specific
versus
just
jdbc
driver
like
do
you
have
to
in
you
don't
instrument
jdbc
directly.
You.
A
C
A
A
Exactly
or
parsing
specific
flavors
of
sql
or
whatever.
C
Oh
or
even
labeling,
yeah
labeling.
We
have
that
semantic
convention
now
in
database,
yeah
yeah
system
right,
db.system.
A
D
A
D
D
C
Now
do
you
know
if
these
stack
or
if
they're
independent.
C
A
It
will
depend,
it
depends
on
whether
the
signature
like
like
the
10.00
may
just
be
adding
a
new
method
to
the
instrumentation
right.
I
mean
I
don't
know
in
particular
in
this
case,
but
or.
A
A
B
B
A
It
was
all
originally
written
by
saxon
saxon
dobbin,
who
came
with
lou
from
from
wiley,
so.
C
A
C
The
the
aspect-
jay
yeah.
C
Oh
yeah,
no,
if
we
let
this.
D
A
B
C
Well,
the
difference,
the
the
big
difference
that
pops
out
there
is
the
the
join
point
delegate,
which
you
can't,
which
bite
buddy
doesn't
do
bite
buddy
you
do
before
and
after
methods
separate
and
you
have
to
pass.
You
have
to
pass
your
local
variables
across
there
if.
C
Versus
doing
that
join
point
thing,
which
is.
A
C
A
C
Oh
yeah,
the
bite
buddy
does
that
when
you
do
suppress,
throwable
or
whatever,
and
then
it'll
delegate
that
to
your
handler
like
if
you
cause
a
problem,
that's
what
you
mean
like
if
there's
a
error,
in
your
exception,
thrown
by
your
instrumentation.
A
It
will
it
does
that,
and
it
then
will
handle.
I
mean
it
properly
handles
if
the
calling
code
throws
an
exception
to
make
sure
that
that
gets
properly
propagated,
etc.
But
the
weaver
does
all
that
work.
Yeah,
so
it'll
it'll
wrap
your
the
euro,
our
own,
the
instrumentation
code
in
the
try
in
try
catches
and
in
and
the
actual
original
code
as
well.
C
C
A
C
Oh,
that's
cool!
That's
really
good
to
see
like
because
this
is
a
good
data
point
for
us
like,
because
you
know
somebody
came
somebody
came
along
and
was
like.
Oh,
we
should
have
apache
http
2o.
C
C
I
think
also
the
okay
http
client
was
added
after
the
fact,
the
2x
one
like
we
had
the
3o
and
later
and
then
somebody
added
okay.
I
can't
do
the
alphabet
there.
We
go
yeah
this
one
got
added
later,
but
yeah
so.
A
That's
a
so
a
lot
of
that
stuff.
When
we
got
little
one-off
requests,
our
professional
services
would
go
and
build
the
instrumentation
custom
for
the
client,
and
I
think
that
stuff
is
all
going
to
be
open
source
pretty
soon
as
well.
So
all
of
this
kind
of
super
hacked
together
pro
services
instrumentation
that
has
no
tests
and
probably
only
works
for
one
customer
interesting,
but
I
think
that
stuff
is
eventually
going
to
be
open
source
as
well.
C
D
C
It's
possible,
but
right
now
we're
thinking
to
keep
the
sort
of
custom
stuff.
C
The
hooks
limited
to
say
not
trying
to
dynamically
load
stuff,
but
if
we
can
make
it
if
we
can
make
it
easy
enough
to
build
your
own
custom
distribution,
you
know
where
we
have
some
gradle
templates,
and
you
know
you
just
plug
in
your
instrumentation
here
and
then
you
get
your
own
java
agent.
That's
nice!.
B
Talk
about
disco
today,
so
disco
is
this
agent
framework,
whatever
that
aws
is
developing
and
they
actually
took
the
exact
opposite
approach
where,
instead
of
having
distros
they're
doing
a
totally
plug-in
based
approach
where
the
agent
itself
has
and
then
you
there's
only
the
agent,
then
you
have
to
provide
it
a
folder
with
a
bunch
of
other
plug-in
jars
inside
it
and
it
loads.
Those
yeah.
A
B
C
Yeah
it
is,
it
is
easier,
for
I
mean
it
allows
users
to
upgrade
the
agent
more
easily
just
drop
in
a
new
version
of
the
agent.
C
So
I
could
see
us
getting
there
yeah,
I
think,
for
ga
probably
will
just.
C
Let's
see
yes,
yes,
so
we're
down
to
23
p1
issues,
I'm
only
tracking
p1
issues,
so
we
had
made
good
progress
in
the
last
week.
I
don't
know
we'll
see
there
were
a
bunch
of
easy
ones.
So
I
I
closed
several
easy
ones,
just
because
I
wanted
to
make
nikita
happy
because
he
sounded
really
stressed
the
week
before
about
the
the
ga,
the
p1
issues.
D
C
A
When
you
all
get
through
those
23,
you
can
start
looking
at
the
needed
for
ga
and
the
the
sdk
and
the
api,
because
we
still
got
a
ton.
C
Because
I
don't
know
that
at
least
that
helped
us
a
lot,
because
if
you
look
at
required
for
ga
there's
91
and
that
that's
just
not
going
to
happen-
and
I
think
it
is
it's
fine
to
still
have
it
labeled,
like
I
don't
know,
maybe
required
is
not.
Maybe
we
should
relabel
desired
for
ga.
C
Yeah
there
you
go
targeted
for
ga,
but
then
you
know
I
mean
honestly,
oh
we're
at
20.,
you
just
merged
that
one
yeah
under
eight
yep,
I'm
like
that's,
that's
how
we
got
down
to
20.,
so
yeah
I
mean
these
are
like
you
know
like
cannot
go
ga
without
it
and
so
that
and
it
feels
like
a
it's
a
more
enjoyable
number
to
look
at.
So
I
highly
recommend
doing
that
in
the
sdk
repo.
Also.
C
A
I'm
getting
to
be
I'm
having
to
be
the
bad
guy,
as
maintainer
like
people
are
like.
Hey,
completely
reorganize
your
api.
This
way,
I'm
like
no.
B
C
A
C
Yep
did
bogdan
have
to
mention
anything
about
the
scope
listener.
A
We
we
I'm
trying
to
remember
what
I
mean.
I
have
only
been
chatting
with
him.
I
just
poked
him
yesterday
saying
hey,
you
said
you
were
going
to
look
at
this.
A
I
don't
think
he's
been
doing
anything
he's
been
focused
almost
100
on
trying
to
get
the
metrics
protocol
otlp
protocol
into
shape
that
we
can
start
getting
it
close
to
stable.
B
D
A
I
know
in
the
metrics
group
metrics
meeting
we
had
this
morning,
rogen
showed
up
late
and
said
that
the
technical
committee
he
had
just
gotten
out
of
the
meeting
and
the
technical
committee
was
trying
to
figure
out
how
to
start
moving
moving
the
spec
along
faster
yeah,
certainly
and
they're,
getting
a
lot
of
pressure
from
like
I
know,
lightstep
leadership
is
getting
super
antsy
and
starting
to
lean
on
lean
on
people.
B
A
A
B
A
B
A
You're
continuing
beating
the
drum
of
saying
tc
members
should
not
be
maintainers
is
appreciated.
C
Did
you
hear
from
did
morgan
ever
get
back
about
the
google?
He
said
somebody
from
google
may
be
able
to
join.
A
C
Yeah
we
need
to,
we
need
to
figure
out,
get
somebody
on
the
somebody
else
on
the
maintainer
path.
There.
A
Well,
he
was
saying
he's
bogdan
was
saying
he's
going
to
try
to
get
nikita
on
that
path,
but
I
don't
know
I
don't
know
how
that's
going
to
work.
A
C
Part
of
my
objection
to
merging
the
two
groups
was
just
that
it's
the
ideal
state
would
be
to
have
three
people.
You
know
three
feels
like
a
really
nice
number
maintainer
number.
On
the
on
the
instrumentation
side.
I
mean,
I
guess
we
actually
had
technically
have
four
but
tyler's
not
too
involved.
A
B
A
B
A
C
Well,
you
know
what
helped
me
on
that.
John
was
the
once
I
had
this
p1
issue,
the
tracked
yeah
I
was
just
like
yeah.
I
think
I
closed
six
out
of
these
and
you
know
most
of
them
were
little.
I
mean
I
just
went
through
and
I
cherry-picked
all
the
super
easy
ones,
because
I'm
like
gotta
gotta
gotta
get
that
number
down.
B
A
A
Are
actually
just
open
bugs
that
no
one's
been
able
to
reproduce
that
I
kind
of
wanted
and
the
purpose
there
was
that
I
labeled
those
as
required
for
gas
like
we
either
need
to
decide.
This
is
a
real
thing
or
not
and
close
it
like
it
doesn't
necessarily
have
to
be
fixed.
If
nobody
can
reproduce
it,
we
could
just
could
just
end
up
closing
it.
C
You
know
what
has
been
working
well
for
me,
both
in
the
application
insights,
our
microsoft
public
repo,
when
people
submit
issues
there,
instead
of
going
through
our
like
paid
support
and
same
in
open
telemetry
instrumentation
is
I
ask
them
to
provide
a
repro
in
a
separate
github
repository?
You
know,
publish
a
github
repo
that
has
a
reproduction.
B
B
No,
it's
github.
I
think
it's
by
github.
It's
really
I
didn't
know
there
was
such
a
good
collection
of
content
about
how
to
run
open
source
projects.
I
definitely
recommend
I
mean
I
would
recommend
just
every
single
maintainer
read
this,
but
at
least
for
now
this
it
was
a
good
read,
including
like
the
best
practices
for
maintainers.
They
to
talk
about
that
sort
of
topic.
Also.
B
B
B
B
B
C
And
keep
the
conversation
from,
I
am
yeah,
I
I
know
you
had
that
you
made
that
point
last
night
honorary.
C
B
C
B
A
B
C
Yeah,
I
would
love
to.
I
mean
this
would
be
something
that
would
be
good
to
like
share
present
talk
about
in
the
monday
maintainers
meeting,
which
is
conveniently
scheduled
in
the
middle
of
your
night.
C
Will
I
will
morgan
has
mentioned
a
couple
of
times
that
since
as
they
bring
google
people
on
to
this
project,
they're
going
to
be
australia-based,
so
there's
some.
I
think
so
he
mentioned
that
around
because
I
had
said
that
on.
I
don't
know
a
couple
weeks
ago,
a
maintainers
meeting
that
we
now
have
some
a
maintainer
who
is.
B
C
Yeah,
so
I
was
explaining,
so
carlos
was
wondering
why
we
can't
use
java
util
logging,
while
we
use
slf4j-
and
I
don't
know
if
you're
you
had
known
onorak,
but
it's
it's
kind
of
a
java
agent
problem
in
general
of
at
startup.
We
can't
touch
the
log
manager
too
early,
otherwise.
Well,
the
main
problem
is
wildfly
and
jboss
will
fail
to
start
because
they
expect
to
be
able
to
feel
a
big.
B
B
C
Do
it
just
like
boom
because
they
expect
to
pro
they
pro
programmatically
set
up
the
log.
B
C
And
you've
touched
it
too
early,
so
you've
instead,
you've
started
it
so.
A
We
ran
into
something
interesting
similar
to
this
with
log4j,
and
I
think
it
was
j9
ibm's
j9
jvm
that
they
actually
do
logging
as
a
part
of
their
class
loading
and
like
they
log
they
log
during
class
loading
in
in
j9,
which
caused
all
sorts
of
crazy
problems.
When
you
try
to
hijack
logging
for
the
like,
we
were
trying
to
hijack
logging
and
take
over
logging
for
our
agent.
So
we
had
our
own
agent
logging
and
then
the
j9
thing
would
be
trying
to
use
that
logging.
A
While
we
were
like
in
the
midst
of
the
hijacking-
and
it
was
a
big
mess
and
would
we'd
end
up
with,
like
I
think,
the
like,
the
application
logs
would
actually
end
up
in
our
agent
logs
or
vice
versa.
I
don't
remember
one
of
the
two,
but
it
was
a
big
mess
like
they're
like
we
turned
on
agent
version
whatever
and
all
of
our
logs
disappeared
in
the
agent
logs
rather
than
in
the
actual
logs,
so
yeah
that
was
fun.
C
Yeah,
I
think
carlos
was
just
was
asking
about
the
pair
this
era.
B
C
No
I'm
just
from
gathering
generating
logs
from
the
agent
for
debugging
purposes,
and
so
there
were
a
couple
things
that
we
don't
so
we're
not
documenting
the
different
properties
like
there's
more
properties
than
just
this.
C
So
this
is
always
doc
and
the
fact
that
you
have
to
shade
those
property
names
and
then,
ideally,
then,
instead
of
having
to
put
all
of
those
on
the
command
line,
how
the
external
properties
file
and
tyler
mentioned
that
they
have
somebody
who
just
started
working
on
a
datadog,
a
replacement
for
the
slf
for
jay,
simple
impul,
so
their
own
slf
4j
implementation.
That
is
not
as
powerful
as
logback,
but
as
at
least
small.
And
so
that's
something.
Potentially
we
could
use
in
the
future.
A
C
You
actually
do
because
we're
piping
that,
because
we
shade
the
we
shade,
the
open,
telemetry
sdk,
we
shade
the
its
use
of
java
util
logger
to
we
do,
because
that
gives
us
the
same
problem
at
startup
right,
because
we.
C
Sdk,
which
touches
the
java
util
lager
yeah.
So
do
you
remember
that
the
junit
four
test
that
you
just
converted
recently
on.
C
Something
that
looks
exactly
like
the
java
util
lager
and
at
and
we
shade,
so
we
should
be
able
to
find
in
gradle
files
with
shade
we
shared
it
all
over
the
place
yeah.
So
we
rewrite
the
lager
so.
C
D
A
C
C
A
A
A
Yeah
we're
usually
using
jul
because
we
don't
we
want
to
have
well
except
for
some
reason,
jrpc
context,
don't
want
to
have
any
dependencies.
C
A
Did
either
of
you
ever
follow
hani
suleiman
in
his
bio
blog
back
in
the
early
2000s?
No,
he
was,
he
was
hilarious.
He
was,
he
was
a
java
guy,
but
he
he
had.
This
thing
called
the
bio
blog,
which
would
basically
just
spewing
bile
about
everything,
and
it
was
hilarious.
It
was
just
done.
A
B
C
They
don't
tend
to,
I
cc
tyler
on
issues
that
were
like
bugs
that
I
know
affect
them
also,
and
so
he
does.
I
have
seen
him
back
port.
Those.
D
B
D
C
B
C
So
yes,
so
this
was
all
right.
When
we
started
the
call
I
was,
I
wrote
a,
I
wrote
a
smoke
test
or
I
added
to
our
smoke
test
here,
the
a
width
span
so
that
I
can
we
can
prevent
this
from
happening
again
and
but
it's
like
and
I
fixed
the
shading,
but
it
still
wasn't
generating
span,
and
so
I
just
was
looking
at
the
log
file
and
so
it's
getting
muzzled.
C
D
C
I
think
it's
muzzling
the
whole
that
whole
instrumentation
right.
C
Yeah
and
so
the
problem
is
missing,
method
check
this
out
missing
method
kind,
so
it
was
the
kind
it
was
when
we
introduced.
The
kind
attribute
is
when
this
broke,
and
the
reason
it
broke
is
because
we're
looking
for
we're
shading
the
span.kind
right,
but
it
really
it
should
that
should
be
user
land
span
kind
here,
because
the
annotation
lives
in
user
land.
So
I
haven't
figured
out
what
to
do
about
it
yet,
but.
C
Because
we
don't
shade
until
the
end.
B
C
Possibly
was
that
when
we
added
the
kind.
C
C
C
Yeah
yeah
yeah
we're
getting
some
good
good
user
feedback.
B
C
B
C
B
B
C
C
C
C
So
what
I
was
thinking,
maybe
osu
was
that
they're
in
the
shim
they
have
their
own
span.
Implementation.
A
C
B
B
B
B
D
C
C
And
yeah
we're
getting
that's
why
I'm
so
excited
to
have
the
new
relic.
I
really
want
to
join
forces
with
the
the
new
relic
agent
team
and
it's
I
really
would
like.
I'm
really
hoping
that
tyler
can
pull
the
data
dog
folks
in
because
yeah,
it's
a
there's,
there's
little
tricks
and
it's
it's
great
to
have
a
community
of
people
to
chat
about
them.
A
A
C
Yeah
I
wanna
I
gotta
reach
out
at
some
point
to
the
elastic
apm
folks
again
see
if
I
can
get
there
get
them
interested
because
that
they
seem
like
the
last
like
at
least
open
source
agent.
That
is
not
that
is
still
off
on
their
own
sort
of.
B
A
A
Yeah
and
now
that
the
now
that
our
java
agent
is
finally
open
source-
because
I
mean
our
team
has
been
spent-
the
last
like
two
and
a
half
months-
getting
it
ready,
so
it
can
be
open
source
working
with
legal
and
working
through
all
of
that
stuff.
They're
gonna
actually
have
time
to
start
jumping
into
this
stuff.
Now.
A
She
manages
that
team
and
she
manages
like
me
and
tyler
a
couple
other
people
so
tyler
young.
C
A
C
A
One
of
the
maintaining
he's
one
of
the
maintainers
of
the
go
yeah
open,
telemetry,
no,
she
well.
She
manages
the
java
agent
team
and
then
our
open
instrumentation
team,
which
has
a
couple
front-end
developers
who
are
mostly
javascript
people,
so
so
she
has
like
nine
people
reporting
to
her.
I
think
something
like
that.
So.
C
C
A
C
Good
yeah,
thanks
for
showing
up
john,
that
was
all
that.
C
B
C
We're
into
oh
yeah:
this
was
not.
This
was
me
starting
to
review
it
the
other
day
and
being
wrong,
so
we're
just
gonna
delete
that
and
move
on
j
unit
j
unit
j
in
it
same
more
wrong,
oh
yeah!
So
these
I
mean
this
needs
a
fresh
class
loader
stuff.
C
Yeah,
I
I
mean
we
can
come
back
to
that.
It's
not.
I
was
trying
to
think
if
there
was
like
on
annotations.
B
B
B
C
So
maybe
just
create
an
issue
to
revisit
this
just
so
we
don't
lose
track.
B
C
C
B
C
B
B
B
B
B
B
Yeah-
let's
play
said
the
http,
so
here
like
this,
just
blew
my
mind
because
I
mean
it
is
true
that
header
names
aren't
supposed
to
have
spaces
in
them
and
plays
in
play,
does
validate
that
so
for
those
tests
to
be
passing
right
now
makes
no
sense
to
me
and
with
my
change
they
stopped
passing
as
I
would
expect
and
same
with
the
other
exceptions.
So
this
I'm
not
sure
what's
going
on,
but.
C
B
C
D
B
B
C
Come
back
to
it
at
some
point:
yeah,
but
label
it
on
p2
contributor
experience.
C
B
B
B
B
C
C
D
C
Yeah,
so
that's
why
I
make
that's
why
I
have
tyler
hit
the
merge
button
and
also
at
this
point,
they've
all
signed
the
cla
for
open
telemetry,
and
so
the
idea
is
it.
It
comes
under
their
contributor
license
agreement
plus
they
made
the
initial
donation.
But
what.
C
This
my
issue,
I
will.
B
B
Thing
happened
out
comment.
The
same
thing
happened
in
the
collector,
where
we
were
sort
of
copying
code
in
from
the
x-ray
daemon
for
the
new
x-ray
receiver
and
anytime.
I
copy
code
from
a
different
pause,
or
I
always
copy
the
copyright
statement,
and
I
know
we
don't
do
that
for
datadog.
I
wonder
if
that's
special
or
if
I'm
just
being
over
cautious
or.
C
B
C
C
C
Pr's
that
I
don't.
C
C
Cool,
what's
failing
here,
the
build
has
been
failing
a
bunch
just
and
I've
had
to
hit
retry
there's
been
like.
Oh,
oh
wow.
This
is
a.
C
Oh,
no,
that's
a
different,
that's
not
that's!
When
you're,
I
don't
even
know
where.
C
Yeah
and-
and
you
know
what's
weird,
is
I
checked,
it
seems
to
still
be
working
in
the
datadog
repo
because
their
builds
I
went.
I
went
and
looked
today
yesterday
when
I
was
digging
into
it.
They
hadn't
run
any
builds.