►
From YouTube: 2020-09-03 Java 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).
A
B
B
A
A
B
A
C
A
B
Yeah
I
chatted
with
him
earlier
today.
I
don't
know
whether
he'll
be
on
tonight
tonight
or
not
since
he's
not
here.
Do
you
want
to
talk
a
little
bit
about
that?
First,
pr
to
add
the
the
generic
key,
I
don't
know
if
you've
got
to
see
my.
B
Yeah
that
the
the
actual
the
issue
with
the
colliding
with
the
map
interface
is
because
the
the
there
the
the
live
span,
the
record
events
readable
span,
which
is
named
horribly,
uses
a
different
attributes.
Implementation
that
is
a
extension
of
hashmap.
D
B
But
as
I
commented,
I
hate
that
I
don't.
I
don't
think
you
should
ever
extend
hashmap.
I
think
it's
a
terrible,
terrible
idea.
Yeah
it
supposes
a
giant
api
that
you
probably
don't
actually
want
yep.
So
I
think
the
right
answer
here
is
actually
to
finally
there's
actually
an
issue
logged
about
a
long
time
ago,
but
to
rip
the
band-aid
off
and
stop
having
it
extend
hash
map
and
just
have
it
wrap
a
hash
map.
B
So
so
I
may,
I
actually
have
done
that
locally.
My
first
hope
was
actually
to
see
if
I
could
just
implement
that
with
a
with
the
actual
attributes
builder
itself.
B
But
the
the
interface
that
you
need
on
a
live
span
is
a
little
bit
more
complex
and
you
have
to
end
up
exposing
a
whole
bunch
of
weird
stuff.
On
the
builder.
A
B
A
B
Fine,
so
I
probably
should
just
put
that
into
this
pr
since
there's
a
very
little
code
change
actually
and
then
I
can
revert
all
of
the
get
value
to
just
be
get,
and
then
I
think
lines
of
code
wise
will
probably
be
about
the
same,
which
isn't
very
many
lines
of
code.
Anyway,
it's
a
fairly
small,
it's
a
fairly
small
image.
A
B
B
B
A
B
I
think,
having
having
this
interface
be
a
top
level
thing
first
level,
api
artifact,
rather
than
make
people
implement
that
kind
of
deep
in
weird
interface.
That's
that
has
all
the
types
on
it.
I
think
it's
a
better
api
anyway.
As
for
a
consumer.
D
D
D
D
Morning
and
asking
basically,
you
know
what
their
what
their
needs
are,
what
it
would
look
like
if
they
needed
to
maintain
a
fork,
and
she
pulled
in
also
bogdan
and
ted
since
they're,
both
on
the
also
on
the
governance
committee.
D
Let's
see
saying
yeah
basically
like
that,
the
that
the
usage
trends
around
java
7
don't
really
support
the
idea
for
building
java
7
leaving
out
new
features
of
java
8.
Her
question
was:
what
does
the
dynatrace
customer
profile
look
like
how
many
of
their
customers
are
running
java,
7.
and.
D
D
B
D
B
D
D
If
there's
one
vendor,
I
mean,
if
there's
one
vendor,
that
would
typically
be
what
would
happen
is
they
would
need
to
maintain
a
fork
and
the
only
problem
with
that
potentially
is
if
their
customers
are
using,
that
directly
also,
and
but
it's
not
clear
to
me
so
no
response
yet,
but
oh.
D
Yeah
he
joined
a
couple
of
the
very
early
java.
Instrumentation
meetings
is
the
only
place.
I
know
him
from.
B
Yeah
that
sounds
right.
Yeah.
D
Yeah
it'll
be
interesting
to
see
if
who
does
giovanni
does
christian
usually
come
to
the
friday
morning
java
sig
meetings.
D
B
Don't
think
I've
seen
him
there
in
a
while
giovanni
is
often
there
he's
there
more
often
than
christian.
D
Okay,
oh
yeah,
and
so
sarah
thought
that
it
made
it
was
a
good
target
to
she
thought
it
was
a
good
idea
to
escalate
it
bring
it
up
in
the
monday
maintainers
meeting,
so
she
just
suggested
a
week.
Well,
this
monday
is
labor
day
anyway,
but
also
give
that
would
give
a
week
or
so
to
potentially,
you
know
hear
back
from
eloise.
B
Get
all
the
get
all
the
cats
aligned
before
bringing
it
up
publicly.
B
D
That
we
would,
she
supported
that
idea
of
us
bringing
it
up
the
week
after
this
monday
in
the
maintainers
meeting,
as
sort
of
a
way
to
try
to
force
the
discussion
forward.
C
D
D
In
the
in
the
repo
that
I
haven't
checked
in
like
24
hours
now,
I'm
feeling
totally
out
of
date.
C
B
D
B
A
A
A
B
B
D
Cameron,
oh
yes,
yes,
cameron
joined
from
the
new
relic
android
team,
so
we
had
a
good
chat
about
what
new
relic
does
for
android
support
and
it's
interesting.
So
they
are.
D
They
just
support,
auto
instrumentation
in
on
android,
and
they
support
it
via
compile-time
byte
code
instrumentation.
D
So
they
instrument
and
primarily
the
instrument,
I'm
sure
we
could
dig
up
the
public
docs
but
cameron
said
primarily
everything
they're
just
capturing.
I
mean
network
traffic
via
okay
http.
D
D
D
Overhead
of
coming
through
all
the
the
every
byte
code
instruction
so
yeah,
I
thought
that
was
quite
interesting
and
made
me
kind
of
step
back
to.
You
know
that
idea
of
waiting
until
we
have
a
vendor
who
you
know,
has
a
direction
that
they
want
for.
C
D
For
their
customers,
because
that
is
a
very
different
direction
than
the
one
we
were
pondering
of
providing
library
instrumentation
for
ok,
http,
for
example,
I
mean
we
could
provide
library
instrumentation
for
http
url
connection.
Also.
D
Cameron's
thoughts
on
sort
of
the
manual
instrumentation
was
that
if
they
had,
if
customers
had
to
do
that,
then
they
wouldn't
have
any
customers,
because
they
all
all
the
customers,
like
the
the
just
apply
it
at
in
during
the
automatic.
During
the
compile
step,
I'm
sure
there's,
you
know
I'm
sure,
there's
people
out
there
who
would
really,
who
would
like
library,
instrumentation
just
like
in
normal
server
java
server
stuff.
D
I
just
wonder
if
the
vendors
who
are
driving
the
project
are
that
interested
in
that,
since
it's
more
of
since
that
doesn't
seem
to
be
where
the
customers,
the
majority
of
customers,
are.
A
A
A
D
D
C
A
B
Okay,
I
wonder
I
wonder
if
you
could
use
the
c
plus
use,
like
c
plus
plus
auto
instrumentation,
in
some
crazy
way
with
crawl.
I
don't
know
enough
about
how
girl
works.
A
B
A
It's
the
funny
thing
that
can
go
like
every
time
I
talk
to
someone
in
from
go.
They
still
use
the
word
auto
instrumentation
for
the
concept
of
library,
instrumentation
that
we
have
like
it's
one
entry
point
to
automatically
instrument
that
library,
so
they
still
call
that
auto
instrumentation.
If
you
work
in
the
go
world,
that's.
B
B
B
B
D
So
I'm
not
sure
where
that
I
mean
I
guess
this.
My
my
feeling
after
this
was
more
to
the
oh
and
cameron
was
great
offered
to
oh.
He
said
he
would
continue
joining
the
meetings
yeah.
So
that's
definitely
somebody
we
can
chat
with.
Let
me
see
what
other
interesting
things
he
said
they,
you
know,
they
register
crash
detection.
D
Uncaught
exception
handlers.
Some
of
these
they
don't
do.
This
was
just
sort
of
ideas
of
things
to
do.
They
do
capture
disk
usage
and
memory.
D
Interested
in
that
yeah
android
has
been
increasing
security
and
limiting
access
to
some
of
that
data.
D
Yeah,
just
that
I
mean
it
seemed
very
targeted
to
oh,
I
asked
about
distributed
tracing.
He
said
that
they're
not
capturing
it.
Currently
they
had
it,
but
then
they
had
disabled
it
for
some
reason
but
a
long
time
ago,
but
he
expects
that
they
will
bring
that
back.
Sometime
soon,.
A
A
A
A
Thought
of
that,
but,
like
one
thing
that
just
came
to
mind
was
how
the
nice
thing
about
libra
instrumentation
is:
it
allows
different
types
of
weaving
right,
either
use
bite,
buddy
like
ours
does
or
if
you
can't
use
bite
buddy
for
some
reason
you
can
still
use
something
else
also,
so
it's
not
necessarily
in
the
user's
app.
It
could
still
be
an
abstraction
for
a
different
type
of
agent.
That's
not
our
agent.
D
A
D
Yeah,
okay,
but
I
yeah
I
mean
it
is
a
still
in
our
I
mean
I
still
do
like
the
breaking
out
of
the
library
instrumentation
and
then
our
auto
instrumentation
is
like
you
say:
it's
just
bind
yes,.
A
D
A
D
D
Oh,
I
asked
about
this.
A
Did
I
just
start,
I
skipped
the
previous
plane
seems
interesting.
Oh.
C
D
They
have
any
kind
of
because,
like
the
agent,
the
new
relic
java
agent
has
an
api
sort
of
like
a
small
version
of
open
telemetry
api
for
sending
custom.
D
For
android
they
do
not-
which
I
thought
was
interesting-
that
there
really
hasn't
been.
I
find
it
interesting
that
there
hasn't
been
demand
for
that.
He
said
the
one
thing
they
do
support
is
a
trace,
annotation
yeah,
and
so
they
will.
What.
D
That's
a
good
question:
what
would
it
do
generator
a
request
like
because
it
wouldn't
be
a
there's
not
like
a
it?
Would
I
assume
some
kind
of
a
span,
an
internal
span?
It
would
generate.
A
D
That
they
that
they're,
it
kind
of
shows
the
lack
of
maturity
on
of
android
develop
like
of
the
whole
concept
of
tracing
distributed
tracing
on
the
mobile
side.
A
D
A
It's
too
bad,
like
I
remember
the
first
time
using
browser
is
so
awesome
to
have
traces
starting
from
the
browser,
because
then
you
really
get
to
see
how
the
network
affects
here,
because
actually,
even
at
google,
like
our
traces
for
the
google
maps
app,
we
did
start
them
from
the
app.
So
you
can
actually
see
the
network
impact
on
your
traces.
Also
and
that's
awesome
so
yeah
yeah
yeah.
D
Yeah,
so
it
didn't
seem
so
they
they're
just
doing
right
there.
They
are
doing
targeting
the
byte
code
to
java
7.
D
D
Yeah
yeah
he's
did
mention
that
that
that
some
you
mentioned
that
some
customers
or
several
customers
are
yeah
on
older,
not
necessarily
older
android
api
levels,
but
older
built
android
build
stack.
D
A
A
Is
that
a
thing
that's
too
bad?
Well,
I
don't.
D
A
A
D
A
D
Yeah
yep
so
light
step.
What
did
you
mention
here?
So
they
who
said
that
they
have
all
they
have
a
bunch
of
avid
customers
already
on
the
auto
instrumentation
and
so
they've
been
blocked,
but
they've
been
blocked
by
these
two
problems.
D
D
D
So
our
instrumentation
interop
story
is
broken,
so
we
agreed
for
regressions
prior
to
ga
that
we
would
do
branches
and
that
so
they
were,
you
know
basically
like
how
can
we
help
so
the
I
think
the
idea
is.
We
would
create
a
branch
they
could
submit
prs
to
that
brand
cherry
pick
prs
to
that
branch
and
then
we
could
release
from
that
branch.
A
D
Yeah
yeah-
and
I
think
nikita
mentioned
that
he's
got
to
do
something
with
make
just
make
sure
that
the
patch
stuff
works
off
of
the
build
off
of
the
patch
branch.
A
C
A
A
A
D
Yeah,
I
I
kind
of
talked
through
just
for
people
in
general,
that
for
ga
we've
sort
of
cut
the
idea
of
dynamic
stuff
and.
A
A
D
A
A
D
D
A
A
D
Oh
yeah
john
asked
about
the
the
pr
to
the
java
repo
deleting.
A
D
Is
there
any
difference?
Really,
though,
of
I
mean
he
just
needs
to
point
to
the
agent
tooling
now
and.
A
D
From
you
know,
we're
pre-ga
the
as.
D
A
Got
it
yeah
cool
so
in
the
repo
yeah.
D
A
A
D
Yes
to
construct,
you
probably
do
new
names.
A
D
A
A
A
Code
so
right,
that's
that's
like
that's
the
more
high
level
comment
that
I
don't
know
if
I
want
that
qualified
name
cache
ever.
Definitely.
I
don't
want
it
here
because
it
seems
like
overkill
and
if
it's
not
here,
we
don't
have
to
make
it
public
right
away,
which
is
nice
if
we
keep
it
in
the
base
decorator,
but
even
just
as
a
general
concept.
I'm
not
a
huge
fan
of
it.
D
A
D
That
that
that
definitely
rang
a
bell
for
me,
so
no
we'll!
I
will
we'll
we'll
address
that
before
we
merge
that,
but
yeah
yeah,
this
looks
great,
you
can
I'll.
I
won't
merge
it
in
case.
You
want
to.
A
A
D
This
also
there's
another
muzzle
issue
that
this
might
be
even
more
related
to.
C
C
C
A
D
A
A
D
D
I
I
was
was
reorged
this
morning.
D
D
It
is,
I
mean
it's
it.
It's
yeah,
there's
some
other
aspects
that
are
supposed
to
be
like
that.
I'm
supposed
to
watch
over
also,
but
I
had
an
agreement
with
my
prior
manager
that,
like
60
like
I
was,
I
was
trying
to
target
60
of
my
time
for
open
telemetry.
D
A
A
A
D
Right
right,
yeah,
we
could
ping
zenmoto
just
because
he
was
doing
he
seemed
to
be
contributing
some
logging
stuff
to
the
sdk.
A
A
Work
I
sent
a
link
in
psych.
Actually,
maybe
you
can
open
that
up
yeah
to
my
it's,
not
a
it's,
not
a
pr,
because
it's
still
too
early
stage
to
even
be
draft.
I
sent
you
a
compare
link.
Let
me
see
what.
D
D
D
D
C
A
D
A
So
this
one
I
put
on
the
bootstrap
class
path
so
that,
if
odd
instrumentation,
the
agent
class
loader
needs
to
access
or
resource
like
the
on
initialization
time,
the
agent
would
store
that
resource
into
this
holder,
so
that
during
instrumentation
I
can
read
it
from.
There
is
the
approach
I
thought
of.
A
D
D
A
Not
going
to
work
for
log
back,
I
think
it's
more.
If
we
want
to
change
the
users
one
for
now,
I
would
start
with
just
adding
it
if
it's
not
there.
I
think
for
the
log
for
j1
actually
is
what
I
had
in
mind.
While
I
was
working
on
this,
that's
service
loader.
So
if
we
can
just
add
the
service
loader
manifest
file,
oh
okay,.
D
A
D
A
D
A
D
Yeah
I
mean
some.
A
D
A
A
D
I
don't
think
this
exists,
possibly
in
our
code
base.
This
may
have
been
a
hack
that
I
did
over
here.
Let's
see.
A
D
A
I
now
I
see
where
the
boots-
okay,
like
I
was
like-
I
was
looking
at
this
like
you're
reading
from
the
agent
cluster
directly.
How
does
that
work?
Okay,
because
there's
a
reference
to
the
bootstrap
class
later
so
yeah?
Okay,
now
I
understand
this
code,
better
okay,
yeah,
and
that
makes
me
a
bit
more
comfortable
with
my
approach.
It
seems
similar
to
this
one,
so
hopefully
it'll
work,
cool.
D
Yeah
yeah,
that
would
that
will
open
up
some
nice
instrumentation
options
for
us.
D
D
You
can
stack
yeah,
I
have,
I
think,
there's
I
assume
that
bite
buddy
has
a
like
a
priority.
D
A
A
A
A
A
D
Yeah,
could
you.
A
D
A
D
A
D
Yeah
it
may
I
don't
know
if
this
helps
or
not,
but
the
going
to
the
log
back
instrumentation
the
old
logback
instrumentation
that
we
deleted,
but
that
will
bring
we'll
bring
this
back
once
we
have
a
logging
api
to
push
to,
instead
of
turning
them
into
spans.
D
The
way
that
I
was
before-
and
in
that
case
I
mean
it
may
be
the
same,
you
could
use
the
same
thing
I
mean
this
is.
D
Of
a
proven,
instrumentation
point
for
capturing
all
of
the
logs
logopender
stuff.
D
Idea,
what
is
it
logger,
dot
cola,
penders.
A
D
A
D
A
D
A
C
A
Yeah,
okay,
that's
cool
like
I
especially
love
like
log
back
in
the
latest
log4j.
If
we
support
those,
that's
pretty
good
coverage,
I
think
ids
are
important,
despite
not
really
being
ga
priority
tasks
or
whatever.
I
guess.
D
Yeah,
if
you
don't
mind
pinging
just
to
if
we
can
align
these,
if
they
have
thoughts
already
on
these
well,
so
that
we
can.
A
A
D
A
Exactly
and
then.
D
A
D
Yeah,
I
don't
know
because
so
technically,
if
does
arm
area,
include
a
notice
file.
D
I
think
in
that
case
we
don't
do
we
have
a
notice
file.
We
do
not
have
a
notice
file.
I
think
we're
supposed
to
have
a
notice
file
if
we're
using.
D
Project
that
we're
pulling
in
does
have
a
notice
file,
so
you
know
what
we
have
issue.
I
will
just
link
this
to
there.
It's
not
it's
a
p1
task
already
for
required
for
ga.
A
D
A
D
D
A
D
C
A
D
For
catching
me
up
on
the
latest
goings
on
yes,
I
will
try
to
carve
out
I'm
working
on
working
on
figuring
out
how
to
carve
out
more
time.