►
From YouTube: 2020-06-11 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).
A
B
A
C
B
D
B
D
D
B
Perspective,
I
I
I
don't,
spend
a
lot
of
face
time
on
the
repo.
To
be
honest,
I
check
it,
maybe
a
occasionally
so,
but
if
you
have
the
specific
PRS,
please
ping
me.
I
think
that's
probably
the
easiest
way
for
me
to
recognize
they.
You
need
help
my
github
notification.
This
is
like
way
too
backed
up.
I
need
to
like
yeah.
D
D
A
Yeah
I
can
give
some
feedback
on
that
and
I
know
it
just
sayin,
but,
like
you
know,
when
people
like
when
we
have
PRS
from
like
anuraag
or
Rashmi
and
Nikita
has
approved
them
I
just
take
a
you
know,
a
really
quick
look,
then
so
that
that
is
a
big
difference
between
you
know.
Knowing
that
somebody,
a
trusted
approver
has
taken
a
close
look,
then
we
don't
all
need
to
look
at
it.
D
Yeah,
so
I,
don't
I,
don't
think
we
need
more
or
more
approve
vote.
So
many
on
anything
with
PR.
That's
not
a
problem.
In
my
opinion,
one
approval
is
mostly
okay,
but
well.
We
have
four
official
approvals
like
Java
Java
repo
has
like
ten
of
them,
or
even
more
so
I
think
my
main
concern
me
that
can
we
can
we
somehow
get
more
groomers?
Oh,
oh
I
have
only
such
limited
amount
of
one.
I.
A
A
A
A
A
A
B
E
A
B
A
B
E
A
C
A
I
know
it's
a
lot
harder
for
vendors,
who
also
have
their
own
sort
of
own
agent,
already
an
art
or
art
replacing
their
existing
agent.
With
this
agent,
like
I,
know
it's
it's
easier
for
the
easiest
for
myself,
Nikita
and
I
think
on
Iraq,
probably
because
this
is
the
what
essentially,
what
we're
shifting
to
customers.
A
What
yeah
I
think
in
the
same
boat
is
Tyler
I'm,
not
sure
about
John
New
Relic
I'm,
not
sure.
If
there's
a
strategic
direction
there
I
won't
speak
for
them,
but
yeah
yeah
I'm,
not.
F
A
A
G
By
the
way,
we
also
have
an
engineer
here
in
at
except
who
need
a
lot
of
instrumentation
for
Java,
so
he
will
be
joining
little
by
little.
Now
that
we
are,
you
know
he
used
to
be
not
it's
not
a
bad
way.
It's
considered
a
he
used
to
help
with
a
special
agent.
Now,
since
we
are
changing
the
focus
on
their
side
too,
we
will
be
having
calm
little
boy,
people
here
as
well,
working.
A
D
A
You
know
everything
kind
of
GA
related
in
here,
and
we
were
talking
in
the
maintainer
z--
meeting
on
Monday
and
folks
were
saying
that
that
github
I
don't
really
have
too
much
I,
don't
have
any
experience
really
with
the
github
projects,
but
the
consensus
seems
to
be
that
the
projects
don't
tend
to
be
that
useful
when
they
get
a
lot
of
things
in
them,
because
you
can't
prioritize
or
so
the
recommendation
and
Andrew
is
drafting
sort
of
some
guidelines.
They're
all
optional,
but
I
mean
it
certainly
I
know.
A
So
when
you're
I
did
that
with
the
zero
four
zero
as
opposed
to
milestones
well,
you
can
see
I
still
I
mean
I
was
using
milestones
before
and
I
kind
of
liked
it
I
mean
made
sense
to
me.
It
was
a
nice
little
board
to
come
to
and
see
that.
That's
what
motivated
me
yesterday
to
to
carry
you
know
some
of
these
simpler
ones
that
I
know
we
were
wanted
to
prioritize
for
the
next
release
over
here.
It
seems
like
a
nice
view,
so
I
guess
the
questions
I
have
and
want
feedback
on.
A
D
D
Milestones
because
well,
I
I,
don't
see
how
can
we
currently
prioritized?
Do
we
need
these
tasks
in
the
next
milestone
or
the
one?
After
that,
we
we
can't
prioritize
what
do
we
need
before
GA,
but
but
before
that,
after
that,
probably
we
also
want
to
understand
what
the
minor
change
was,
the
main
major
change.
We
want
to
relieve
that
into
the
next
major
version,
because
that's
breaking
change
or
not
that
will
make
sense,
but
for
now
where
every
change
is
the
same,
doesn't
matter?
A
So
the
value
that
I
found
for
this
zero
four
zero
piece
was
tagging.
There
were
a
couple
of
issues,
I
agree
with
you
that
for
the
most
part,
you
know
we're
just
kind
of
rolling
fixing
stuff
adding
stuff,
but
sometimes
there's
things
that
people
have
other
people
have
dependencies
on
like
John.
You
know
really
wanted
this
external
metrics
exporter
so
that
he
could
proceed
with
New
Relic
support
for
that
this
one.
We
knew
we
needed
to
do
for
this
release
because
it
was
added
in
the
zero-five-zero
Java
SDK.
A
This
was
a
priority.
This
was
a
priority
because
it
was
blocking
on
Iraq
and
AWS.
Work.
Sort
of
you
know
having
at
least
having
the
ability
for
people
like.
If
people
ask
hey,
I
really
want
this.
I
can
tell
them.
Okay,
add
it
to
the
next
release,
and
that
gives
us
visibility.
So
we
don't
forget
that
that.
D
D
D
So
the
second
type
it
is
like
this,
so
where
you
have
my
high
priority
low
priority
enclosed
because
well
canton
style,
doesn't
make
sense
in
my
opinion.
But
if
we
can,
we
can
use
just
milestones.
But
if
you
want
to
distinguish
between
like
high
priority
things
and
low
priority
things,
especially
before
GA,
that
may
make
sense.
If
we
see
that
we
cannot
make
it
happen,
that
we
cannot
solve
all
issues,
then
we
can
like
split
them
into
high
priority
low
priority,
then
that
that
project
probably
makes
more
sense.
D
A
D
D
A
A
E
E
B
Here's
a
thought
on
the
a
day
to
dog:
we
don't
really
assign
things
to
milestones
beforehand.
We
we
do
it
when
the
when
the
PR
gets
merged.
So
this
is
more
like
for
after
the
fact
of,
like
hey,
I
want
to
figure
out
which
release
was
a
specific
PR
visible
in,
and
this
is
probably
okay
to
deal
with
on
the
project
side,
but
having
it
to
happen
like
automatically,
is
really
nice
and
not
have
to
worry
about
it.
B
A
A
E
A
Yeah
I
haven't
looked
I
just
saw
the
issue
over
email.
I
haven't
looked
at
it
closely,
but
yeah
I'm
excited
to
check
that
out.
Also
it's
basically
what
anuraag
has
been
doing
some
work
to
how
to
wrap
the
open
telemetry
agent
in
a
vendor
sort
of
vendor
package
and
I've
done
some
I'm
doing
that
internally.
But
my
way
is
super
happy
and
he
and
I
chatted
last
week
about
that
I
told
him.
Don't
look
at
what
I've
done
and
so
I'm
excited
to
see
what
he's
I
think
he's
got
some
good
ideas.
There.
D
G
A
D
D
D
D
If
you
are
interested,
you
can
read
pull
request
description,
and
so
the
main
artifact
is
Java
agent,
plus
all
exporters
and
Carlos
Lee
requested
that
we
would
publish
the
agent
without
any
exporters
as
before,
so
that
vendors
can
take
it
and
combine
it
and
whatnot.
So
so,
currently,
we
publish
two
artifacts
one
with
all
agents,
with
all
exporters
included
these
one
and
one
without
any
exporters.
This
one.
B
A
D
D
Don't
ask
me
currently,
but
well,
that
should
be
the
correct
size,
we're
all
exporters
included,
okay
I.
So
this
is
sustain
reason,
small
accidents
and
even
they
believe
well.
I
cannot
answer
that
right
now,
if
you
take
a
look
right
but
but
the
type,
so
that's
the
correct
time.
So
these
are
like
go
come
from
the
same
deployment.
D
So
these
are
correct
types
707,
so
they're
like
comes
from
the
same
deployment,
and
so
these
so
these
we
are
going
to
deploy
when
we
release
next
version,
so
we
are
going
to
deploy
them
here
to
betray
and
I
believe
trust.
You
have
to
click
a
button
over
here
which
says,
publish
to
maven
central
as
well
or
leap,
to
make
sure
that
has
to
be
done
once
and
then
I
believe.
All
all
our
religious
will
be
automatically
synced
to
maven
central
as
well
and
they
are
in
Jason
enter
anyway.
A
G
D
D
Or
another
option:
another
option
is
to
have
open
telemetry
out
or
don't
jar
without
any
explorers,
there's
an
open
channel
out
all
which
has
all
Italy
all
these
portals
included.
My
idea
was
here.
The
question
is
what
we
like
propose
like
default
and
default
in
my
opinion,
should
be
one
we've
exported.
So
that's
why
without
any
classifiers?
For
me,
if
we've.
C
D
E
D
D
D
D
The
cocoa
from
from
the
other
point
of
view,
if
users
download
that
artifact
to
just
to
really
instrument
the
application,
doesn't
make
any
difference,
they
will
use
that
kernel,
oht
file,
whatnot
and
so
and
if
they
want
to
use
to
rebound
or
something,
then
they
probably
want
just
agent
without
exporters.
And
then
the
Gradle
amazing
dependency
will
will
be
a
little
bit
easier
without
any
classes.
So
probably
that
having
all
classifier
is
not
that
bad.
A
E
A
G
A
A
D
D
Between
different
implementations
and
have
extracted
that,
like
HTTP
server
tracer,
which
is
somewhat
similar
but
somewhat
different
to
those
tracer
that
Tyler
was
proposing
back
at
the
time,
and
so
that's
competency
in
the
master.
Bra
branch.
I
am
waiting
for
Giovanni
semantics,
pants
to
start
using
them
as
well,
and
so
what
I
think?
What
we
have
to
do
with
Tyler
is
to
understand
how
our
two
different,
somewhat
different
approaches,
how
they
different
and
what?
What
the
best
pieces
to
to
pick
from
both.
B
B
The
long-term
picture,
so
in
our
code
base
like
you,
one
can
work,
we
can
like
be
a
lot
more
restrictive
and
be
careful
with
code
reviews
and
all
that
kind
of
stuff,
but
I
think
it
really
comes
down
to
encouraging
people
that
are
not
as
familiar
with
the
API
to
do
the
right
thing
by
default,
and
so
that's
one
of
the
things
that
I
really
tried
to
do
in
my
approach
is
like
expose
only
the
things
that
I
want
people
to
do
by
default.
So,
like
I,
tried
to
be
very
careful
about
like.
B
Specifically,
for
example,
I
had
a
distinction
between
having
logic
on
the
span
versus
logic
on
the
trace,
and
that
was
specifically
because
the
the
the
tracer
doesn't
need
to
be
called
on.
Everything
so
I
mean
I.
Think
in
my
example,
you
could
still
call
span
dot
end,
but
at
least
there
was
a
way
to
like
guard
against
it.
D
D
B
We
it
does
so
so.
The
trade-off
here
having
everything
on
tracer
I
think
is,
is
simpler
for
the
development
of
those
classes,
because
it's
just
in
one
class,
you
don't
have
to
really
define
things.
You
don't
have
to
define
the
relationship
between
the
span
and
the
tracer
and
ensure
that
those
are
compatible.
D
One
of
my
goals
was
exactly
is
correctly.
If
you
want
to
ride
some
instrumentation,
either
manual
or
automated,
doesn't
matter.
You
have
to
know
specific
that
you
have
to
call
these
methods
in
that
order,
creates
fun
on
request
on
connection
blah
blah
blah
I
want
to
remove
that
knowledge,
so
I
fully
support
her
yeah
and.
A
B
Think
that
is
my
biggest
concern,
and
so,
if,
if
we
wanted
to
to
handle
it
this
way,
we
could
say
that
tracer
dot
start
span
returns
a
wrapped
span
that
throws
an
exception
on
end
there
and
that
if
you
call
tracer
not
end
with
the
span
as
the
argument
that
the
tracer
can
unwrap
that
and
call
it
so
like
there's
ways
that
we
could
be
defensive
in
encouraging
the
right
behavior,
but
it
still
feels
a
little
less
intuitive.
To
me.
B
A
B
H
B
A
All
right
so
I
think
we've
identified
a
good
potential
issue
that
we
would
need
to
resolve
for
using
introducing
instrumentations
fans,
but
I
think
that
same
question
will
have
to.
It
seems
like
that
will
have
to
be
answered
for
semantics
bands
anyway,
and
then
we
can
probably
piggyback
on
that
solution.
Hopefully.
D
A
A
B
Yeah,
so
the
other
thing
that
I,
so
one
concern
I,
have
with
Nikita's
your
your
current
solution
of
everything
in
the
tracer
is
one.
Do
we
want
to
have
a
hierarchy
of
like
similar
to
what
we
have
with
the
decorators
of
like
a
base
and
then
a
client-server
client?
You
sound
like
right
now,
it's
just
a
single
class.
So
there
might,
you
might
want
to
explore
before
you
get
too
far
into
the
implementation
of
that,
how
that
class
hierarchy
looks
like
because
it
can
uncover
some
interesting
problems.
B
D
Absolutely
so
currently
in
my
third
branch,
there
is
already
like
small
hierarchy.
It's
like
three
plus,
but
you
not
so
absolutely
correct.
Yeah
I
I
have
made
the
simplest
thing
possible
and
one
of
the
next
step
we
used
to
try
to
bring
several
different
instrumentation
on
the
same
bandwagon,
starting
from
some
client
implementation.
Some
database
implementation
then
certainly
that
here
are
he
and
in
a
structure
of
those
traces
will
change.
My
main
hope
is
that,
like
public
API
will
remain
the
same
from
the
instrumentation.
D
A
A
Real
quick,
the
save
I
was
at
schema
on
this
PR
cuz.
If
everybody's
not
using
the
save
plug-in,
then
you
end
up
like
I.
Did
this
PR
yesterday
and
you
know
it
Auto
adds
all
these
finals
to
everything
and
you
end
up
with
you
know
a
bunch
of
discs
in
the
PR
that
aren't
really
related
to
the
PR.
A
So
it's
kind
of
asking
do
you
know?
Do
we
want
to
abandon
that
and
just
not
anybody
use
it?
Do
we
want
to
it's
hard
if
there's
not
a
way
to
enforce
it
and
I
don't
want
to
be
you
know,
super
I
mean
unless
we
kind
of
enforce
it,
I
don't
want
people
to
submit
PRS
and
us
be,
like,
oh,
add,
final
hero.
Add
final
hero.
Add
final
here,
so
I
don't
know
any
thoughts.
I.