►
From YouTube: 2020-11-20 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
It
turns
the
title
of
our
block
actually
like
the
chat
window
is
very
dark
and
the
yeah
anyway.
Maybe
it's-
maybe
it's
mac
realizing
night
time.
B
A
B
C
A
B
B
B
A
The
you
can't
change
the
sound
with
pagerduty
for
a
while.
I
had
recorded
my
own
voice
as
my
my
page,
my
pager
alarm.
Yeah,
it
was
things
like
like
it
started
out,
really
quiet,
saying:
wake
up
wake
up
and
it
was
interesting.
It
totally
woke
me
up,
but
my
wife
was
not
bothered
by
it
at
all.
Awesome
yeah
like
for
some
reason.
My
hearing,
my
own
voice,
freaked
me
out
and
woke
me
up
and
my
wife
was
never
woke
up
at
all.
So.
B
A
I
was
like
so
confused
thanks
for
moving
that
tag
around,
I
was
yeah
trying
every
other
possible
solution
to
see
whether
I
can
hack
something
through,
but
I
in
the
end,
I
think,
unless.
B
A
B
Github
person
responded
and
they
said
they
can't
do
it
because
then
anyone
can
make
a
change
to
a
different
branch
and
it'll
go
through
like
the
checks
or
something,
but
that
didn't
really
make
sense
to
me,
because
our
workflow
is
only
enabled
on
master
anyways.
So
I
I
mean
I'm
sure
there
are
workflows
where
that
could
be
an
issue,
but
there
are
also
plenty
of
workflows
where
it's
not
an
issue,
and
so,
but
they
were
saying
yeah.
We
can't
do
it
because
otherwise
anyone
would
be
able
to
do
anything
to
repository
really
really.
B
A
B
A
Manual,
it's
not
the
end
of
the
world.
I
think
yeah,
it's
not
that
big
a
deal
as
long
as
we
just
put
kind
of
put
documentation
in
there
yep,
I
think
yeah,
it's
a
bummer.
It
can't
be
fully
automated
because
that
would
be
way
better
one
yeah
can
we
make
it
also
write
the
release,
notes.
That
would
be
even
better.
A
We
can't
I
mean
again,
but
it
was
actually
really
good.
My
maintaining
the
change
log
was
very
helpful.
I
just
I
just
was
able
to
basically
copy
and
paste
the
change
log
right
into
the
release,
notes,
and
I
was
done
really.
It
was
very
easy.
Okay,
I
also
have
been
adding
change
log
notes
into
the
commit
the
squash
commits
on
merges
just
to
see
how
it
would
work
it
works,
but
unless
we
all
do
it
consistently,
it
won't,
it
won't
be
yeah,
but
I
think
maintaining
it
annually
actually
is
working
fine.
A
B
B
A
So
all
right,
well,
is
that
a
is
there
a
pr
for
that
or
is
it
just
an.
B
A
B
Yep,
okay,
do
you
have
a
specific
question,
or
just
generally,
I
don't
know
anything.
A
B
Know
good
or
bad
at
least
the
harder
one
the
trace
from
bogdan
seems
to
be
reviewing
it.
So
that's
good,
I
think,
for
us,
it's
really
just
if
the
code
makes
sense
to
us,
we
can't
verify
the
actual
open
census
behavior
and
that's
fine,
but
we
can
assume
that
people
are
confirming
that
themselves
either
google
themselves
or
bugged
them.
Not
us
that's
been
my
stance
on
it,
but
this
is
sort
of
a
code
review.
A
A
Yeah,
let
me
take
a
look
by
the
way
I
reached
out
to
christian
oberon0
christian
neuemeler,
to
see
whether
he
was
interested
in
being
an
approver
because
he's
not
an
approver,
although
I
thought
would
have
sworn
that
he
was
an
approver,
but
I
never
got
never
got
a
response
from
him,
but
I
think
I
would
be
happy
to
have
him
be
an
approver
if
he
wants
to
be.
C
No
there's
no
well
he's
not
in
the
code
owners,
at
least
because
there's
no.
B
B
A
B
A
Tomorrow
to
add
him:
let's
do
it
yeah
he
if
he
thumbs
downs,
then
I
guess
he
won't
be,
but
I'm
totally
happy
to
have
him
be
improver.
I
don't
always
agree
with
him,
but
that's
okay,
right.
A
C
A
C
Yeah,
I
I
I
was
gonna
clear.
Nikita
has
a
change
requested
on
one
of
the
instrumentation
pr's
and
I
was
gonna.
C
But
I
was
like
you
know
I'll
I'll,
just
ask
him,
you
know
to.
A
C
It
but
bogdan
is
a
different
story.
A
I
work
with
both
of
them,
so
I
can
allegedly
I
can
go
and
exert
some
influence.
B
C
Oh
nikita
will
reply
yeah
he's
just
asleep
yeah
all
right,
so
I
will
I'll
comment
on
the
pr.
How
about
and
ask.
A
C
On
the
pr,
if
he
will
remove
his
block
point
him
to
this
comment,.
A
A
C
A
B
C
Yeah,
but
I
will
do
it
again
here
and
I'll
specifically
ask
him
if
he
can.
You
know
clear
if
he's
good
with,
can
clear
his
block
on
this
and
then
and
then
then
it's
in
your
hands,
john.
A
C
A
Yeah,
I
don't
know,
maybe
the
maybe
the
the
title
of
this
particular
issue.
The
only
exposed
interfaces
is
maybe
a
little
strong.
B
Yeah
I
mean,
but
at
the
same
time,
once
we
get
spam
context,
that's
an
improvement,
but
ideally
for
instrumentation,
like
even
our
labels,
copying
it's
quite
expensive.
There's
a
lot
of
labels
and
everything
is
painful.
Really
the
span
context
happens
to
be
the
most
common
one.
I
think
so.
That's
still
a
good
one.
B
I
think
users
generally
reference
expand
cosmetics
in
their
code
for
a
lot
like,
I
think,
that's
more
common.
The
only
problem
happens
when
a
user
references
it
in
their
code.
Otherwise
it's
not
a
problem
right,
so
I
think
the
classes.
A
B
C
To
me
the
difference,
so
I'm
not
as
worried
about
attribute
attribute
keys
and
labels
because
those
shouldn't
be
created
so
often
so
we
can,
you
know
we
can
bridge
them
once,
whereas
the
span
context
is
more
important
because
it's
afresh
each
trace.
A
B
B
A
A
C
Where
it
will
be
essentially
users,
it's
not
byte
code
compatible,
but
as
long
as
they
recompile,
they
wouldn't
have
to
make
source
code
changes.
A
C
C
I'll
make
a
comment
about
that
span:
context
labels
attribute
yep
cool.
A
C
B
B
Yeah
doctor
have
also
like
it's
understandable
what
they've
done,
but
it's
so
weird
how
like
everything,
seems
to
be
going
wrong
this
month,
though,
like
did
they
just
plan
it,
like
maybe
been
trained,
docker
hub
got
together
and
it's
like.
Oh
we're,
just
going
to
ruin
all
the
premium
users
in
november.
A
C
C
B
B
B
B
B
C
I
honestly
I'm
very
happy
with
the
lazy
approach
I
think
it's
very
painful
to
have
to
I
don't.
Like
the
you
know,
the
two
phase.
B
C
We
took
the
latest
dependency
tests
to
nightly,
so
that's
the
those
are
the
ones
that
search
maven
central
for
the
latest
releases
of
each
library
and
those
would
just
break
like
sporadically,
because
akka
would
release
a
new
version
and
then
some
contributor
would
submit
a
pr
and
their
pr
would
fail
on
akka,
and
they
would
be
like
what
the
heck.
A
C
Yeah,
so
that's
what
we
moved
to
nightly,
but
now
we
have
it
everything
on
merge
to
master,
which
is
both
good
and
problematic,
because
we
were
we
had
it
on.
We
have
it
creating
a
new
issue
in
the
repo
if
the
nightly
build
fails,
but
the
nightly
build
is
now
emerging
on
master.
So
so.
C
B
C
Yeah
well
once
nikita
recovers
from
bin
tray
disaster
land,
I'm
sure
he'll,
poke.
C
C
We
talked
most
of
the
time
about
the
build
problems.
We
pavel
wasn't
happy
about
not
publishing
the
java
agent
instrumentations
separately,
so
I
asked
if
he
would
help
basically
with
the
issues
that
we're
having,
and
he
said
yes
and
so
we
started
going
down
that
route,
but
also
we
were
kind
of
tucked
about
an
artifactory.
C
B
B
C
There
was
also
pavel
mentioned.
He
had
one
some
colleague
who
was
using
this
makito
ship.
It
ship
kit,
which
is
another
kind
of
publishing,
plug-in,
sounds.
C
Yeah
when
he
said
it
on
the
call
I
was
like
come
again,
the
kick
so.
Yes,
he
put
it
in
the
chat
window
so
that
I
could
pull
it
up.
A
By
the
way,
I
added
you
two
to
the
the
attendees
for
today's
this
evening's
meeting.
C
B
C
Yeah,
I
mean
so
it's
obviously
technically
feasible
like
and
then.
A
C
Shadow
to
to
filter
out
what
you
want,
but
he
seemed
at
least
I
mean
I
just
that.
That's
why
I
kind
of
pushed
back
as
far
as
well.
If
you
want
this,
can
you
help
us
troubleshoot
our
build
problems,
so
we
went
as
far
as
I
would
set
him
up
with
the
bin
tray
access
so
that
he
can
troubleshoot
that
and
I'll
give
him
right
access
to
the
repo
so
that
he
can
work
on
that.
C
C
So
basically
cloud,
you
know
a
hosted.
Our
own
artifactory
instance
hosted
nikita
said
that
that's
certainly
more
robust
than
dintre,
but
he
was
going
to
so
he's
he's.
Taking
the
immediate,
immediate
action
over
the
next
couple
days,
he's
going
to
discuss
with
his
jfrog
friend
and
get
back
to
us
basically
on.
If
he
thinks
we
should
go
this
route
or
if
we
should
keep
going
down
the
bin
tray
route.
C
Yeah,
I
don't
care,
I
mean
I'm
happy
to
you
know
to
push
through
the
the
funding
for
it.
I
can.
I
can
expense
it
for
you
know
and
then,
at
the
same
time
start
I
know
how
to
go
down
that
cncf
path.
Now
I
just
want.
A
B
C
So
so
yeah,
so
nikita
is
going
to
get
back
to
us
early
next
week
if
and
then
we'll
either
go
this
route
and
I'll
set
up,
set
it
up
and
start
and
I'll
expense.
It
initially
and
work
with
cncf
to
do
that
or
if
there's,
if
we
want
to
keep
working
on
the
bin
tray
different
gradle
plug-ins.
Something
like
that.
Then
I'll
set
pavel
up
with
permissions
that
he
needs
to
troubleshoot
that
so
I
don't
know.
C
If
there's,
if
there's
no
urgency
I'd,
I'm
sure
none
of
us
really
want
to
try
to
release
zero
11
until
this
is
resolved.
C
B
What
about
11
we're
still
working
on
right
now?
He
got
a
catch
release
of
the
previous
versions,
but
now
we
have
odot
tinder
to
have
published,
but
it's
only
half
published.
C
Yeah
yeah
and
honestly,
I
I'm
honestly
totally
fine
with
that.
Like
you
know,
that's
just
it's
in
a
bad
place
right
now,
yeah.
C
B
B
C
Until
then,
I
don't
want
people
wasting,
you
know
a
whole
day
trying
to
get
ben.
Try
to
publish
we've
got
more,
we've
got
better
things
to
be
doing.
A
Oh
yeah,
I
think
my
my
twitter
post
from
in
the
channel
earlier
today
sums
it
up.
Well,
though,.
A
A
Well,
I
mean
the
big
one
is:
is
expo
86
in
vancouver,
but
the
and
then
the
other
one
is
like
a
riff
on
the
obama
hope
poster
except
it
has
what's
his
name,
the
bb4
he's
a
trap
guy,
it's
a
trap.
C
B
A
C
A
A
C
C
The
problem
is
that
our
java
agent
instrumentation
is
not
it
requires.
We
didn't
want
to
pull
in
the
three
meg
oshi
jar,
yeah,
and
so
our
java
agent
instrumentation
is
kind
of
wonky.
C
And
cool
yeah
the
problem
is
that
they
are
the
main
problem.
I
tried
to
get
make
this
clear
for
carlos.
Is
that
they're?
You
know
they
used
to
have
seva
to
do
all
of
the
java
tricky
java
agent
stuff
and
they
don't
now
and
so
they're
struggling
just
you
know,
with
class
loaders
and
agent
java
agent
stuff.
A
Got
it
well,
I
have
a
my
my
issue
is
more
that
I
I
feel
like
we
need
a
place
for
contrib
instrumentation
and
I
don't
think
contrib
instrumentation
belongs
in
the
main
java
repo,
and
I
don't
know
if
you
all
want
to
own
contribute
instrumentation
in
your
current
repo.
But
this
feels
like
it's
a
separate
like
it's
a
separate
class
of
thing
that
might
need
its
own
home.
C
I
think
we
would
want
oshie
as
a
not
not
often
contributed,
because
that
is
implementing
the
systemetric
spec
stuff.
A
B
A
A
A
C
C
We
spent
you
know
we
spent
half
an
hour
last
week
with
with
with
them
sergey
trying
to
help
him
through
this.
We
spent
half
an
hour
on
tuesday
trying
to
help
them
with
this.
A
I
just
feel
like
this
should
not
it
should
just
be.
I
mean
it
only
uses
the
api,
it
could
be
a
standalone
module,
you
can
live
in
the
instrumentation
repo
be
published
separately.
People
can
pull
it
in
if
they
want
and
away
they
go
like.
What's
I
don't
know,
I
guess
I'm
not
clear
on
what
the
objection
is.
That's.
B
C
Know
what
I
yeah.
C
C
C
Not
for
the
library
instrumentation,
I
don't.
Actually
I
don't.
I
don't
like
the
java
agent
instrumentation.
I
don't
think
that
I
think
what
we
could
do
better
for
the
java
agent
instrumentation
is
have
an
optional.
C
I
mean
I
know
we
already
have
a
dash
all,
but
a
dash
all
plus
a
dash,
all
all
really
all
that
also
bundles
oshie.
You
know
that
because
just
from
a
used
user
perspective
like
if
they
do
want
system
metrics,
I'm
still
up,
I
still
I'm
a
fan
of
it.
Being
you
know,
just
the
single
java
agent.
B
But
I
mean
one
issue
is,
I
think,
we're
using
oshi
for
things
we
can
get
from
the
jpm
like
I.
What
the
jvm
provides
has
always
been
good
enough
for
me.
I've
never
needed
this
washi.
I
didn't
even
know
about
the
library,
and
so
I
think
what
we
want
is
for
our
job
agent
to
bundle
in
the
nonoshi
stuff
and
that's.
A
But
I
think
the
like
the
the
runtime
metrics
that
are
currently
in
the
java
repo
they're
very,
very,
very
basic,
and
I
think
that
if
they're
an
instrumentation
repo,
it's
a
really
good
place
to
then
actually
build
them
up
into
something.
That's
more
useful
and
right
now,
they're
super
basic.
They
have
very
very
little
in
there,
but
we
could
always
build
on
it.
B
Yeah
because
for
me
also
just
in
principle-
I
mean-
I
know
it's
a
spec.
I
personally
would
never
want
my
apps
to
be
reporting
all
the
system
and
always
use
containers
like
system.
Metrics
is
sort
of
a
thing
where
you're
running
one
app
per
vm.
I
think
that's
where
it's
useful
in.
In
any
other
case,
it's
not
even
useful
to
have
every
app
exporting
the
same
numbers
if
they're
running
in
containers,
so
this
still
like.
B
A
Know
there's
another
there's
another
potential
reason
why
people
want
this
stuff.
I
think
in
that
and
I've
just
been
starting
hearing.
People
say
they
say
these
words
and
I
don't
have
any.
I
don't.
I
don't
have
any
evidence
citation
needed,
but
that
the
collector,
like
originally
was
planned
to
to
provide
system
metrics
out
of
the
box.
If
you
were
running
it
like
locally
as
an
agent.
B
A
Terminology
that
I
hate,
but
that
they're
the
the
current
state
of
things,
is
they're
not
ever
planning
on
doing
that.
B
B
A
A
B
B
A
A
C
A
B
B
C
Mean
fever
generates
about
the
host
system
script
from
clark
when
the
collector.
B
B
C
B
C
Yeah,
I
wonder
if
I
should
potentially
just
pull
out
the
java
agent
instrumentation,
because
it's
already
kind
of
wonky
how
we
had
that
in
the
pre-main
we
have
some
like
detection
and
then
we
could,
even
you
know,
have
some
instructions,
for
we
could
add
it
to
nikita's
example.
I
mean.
B
B
C
But
how
like
that's
not
very
discover,
I
mean
it's
kind
of
weird,
I
think
from
a
user
perspective
like
first,
I
don't
think
very
many
users
have
oshi
on
their
class
path.
C
So
if
they're,
if
they
want
these
metrics,
it's
more,
you
know
hey
here,
here's
the
java
agent.
That
gives
me
my
metrics
as
opposed
to
having
to
put
oshi
on
there
in
their
app
like
as
a
dependency.
C
I
did
the
whole
like,
I
mean
I
kind
of
liked,
and
maybe
we
just
didn't
document
I
mean
I,
the
dash
putting
it
on
the
command
line.
Class
path
seems,
I
mean
yeah
I
mean
that
seemed.
I
thought
that
was
not
so
bad.
C
B
B
C
B
C
Right
right,
that
kind
of
does
feel
like
as
our
open
telemetry
like,
maybe
that
the
java
agent
one
could
live
off
in
the
contrib
land
and
we'd
keep
the
library
one.
C
B
C
I
I
I've
been
too
many:
microsoft
people
there's
a
whole
like
a
whole
sibling
team
that
does
that
kind
of
agent
and
they
get
so
mad
at
me
when
I
use
the
term
agent
that
they've
like
be
beat
it
out
of
me,
like
nobody
at
microsoft,
understands
when
I
say
java
agent,
they
all
everybody
thinks
agent
agent
right
out
of
prac
agent.
C
In
fact,
my
last
manager
went
so
far
as
to
like
rebrand
what
we
were
doing
in
ipa
in
process
agent,
just
to
like
help,
people
get
their
minds
around
that.
No,
it's
not
an
out
of
process
agent.
It's
an
in-process
agent.
B
A
A
A
A
I
know
it's
funny
like
john
fly
was
like
yeah
you're
super
familiar
with
the
agent
code
base
right
and
I'm
like,
I
literally,
haven't
cloned
it
ever.
B
A
Turned
out
this
week
that
I
forked
it
like
three
months
ago
and
then
I
like
pulled
down
my
fork
and
I
tried
to
build
it
and
it
didn't
work
because
it
was
depending
on
080
snapshots
that
don't
exist
anymore.
So
all
right
very
I.
B
B
That's
true:
okay,.
C
You
you
know
where
to
phone
a
friend.
A
C
And
then
I
did
ask
about
the
builder
agent
builder
idea
that
you
had
honorag
and.
C
I
think
that
the
only
thing
that
I
think
is
going
to
be
tricky
is
the
class
loader.
Where
do
we
put
it,
and
how
do
we
do
that?
C
I
was
just
reviewing
pavel
had
put
in
a
pr
here
and
he
actually
as
kind
of
a
side
effect
or
just
a
refactoring
in
here
that
he
did.
He
pushed
a
bunch
of
stuff
from
agent
initializer
in
the
which
is
in
the
bootstrap
class
loader
to
the
agent
installer
in
the
agent
class
loader,
which
I
think
is
good
for
the
for
the
agent
builder
thing.
If
we
can
just
like
get
everything
sort
of
pushed
down
into
the
agent
class
loader,
then
we.
C
B
C
So
I
think
yeah
I
I'm
I'm
optimistic
about
that.
A
A
Which
I
think
he
would
be
the
only
member
of
and
I'm
not
even
sure
anyway-
I'm
not
even
I'm
not
sure
what
to
do
about
it,
and
we
could
certainly
do
that.
But
I
feel
like
it's
a
little
overkill
for
getting
one
person.
B
Yeah,
if
the
problem
is
the
lack
of
visibility
to
breaking
changes,
we
can
try
to
have
some
pattern
making
sure
our
title
here.
Our
title
has
the
word
breaking
chain
inside
it.
If
that
helps
I'd,
be
okay
with
it
but
like
I
would
rather
this
get
solved
by
watching,
because
that's
sort
of
the
reason
that
feature
exists
on
github.
A
I
just
wanted
to
get
some
other
opinions,
so
the
other
thing
I
wanted
to
bring
up
was
so
the
metrics
sig
today
spent
some
time
and
actually
on
the
java
ins.
The
java
regular
javascript
yesterday
morning
spent
some
time
talking
to
carlos
and
andrew
andershu
and
josh,
and
it
sounds
like
everybody
is
very
much
in
favor
of
java
doing
a
metrics
sdk
rewrite.
A
I
am
waiting
for
someone
to
give
a
thumbs
up
on
that
issue
honorary.
If
you
want
to
do
that,
that
would
be
great
and
then,
if
carlos-
and
I
I
think
it
sounded
like
josh-
was
going
to
pester
carlos
to
give
a
thumbs
up
on
that
issue
as
well,
and
then
I
can
then
I
would
go
ahead
and
put
in
a
pr
for
my
steel
thread
with
one
instrument,
so
it
sounded
like
everybody
like
it
sounded
like
the
people.
I
don't
know
there,
it
is
there's
a
thumbs
up.
A
A
A
A
really,
I
mean
an
excellent
pro
excellent,
excellent
java
guy
it'd
be
great
if
we
could
get
him
involved
so
anyway,
and
and
a
little
bird
told
me
that
somebody
who
I'm
very
good
friends
with
here
in
portland
left
new
relic
last
week
and
might
be
also
available
soon
to
help
working
work
on
this
stuff.
More.
C
We
need
to
distribute
the
the
the
the
load
here.
B
A
A
B
A
A
On
lean
on
onerog
to
actually
get
his
affiliation
correct
and
then
maybe
he's
fine.
B
A
C
Yeah
not
really.
I
didn't
really
understand
why
anybody
cares.
A
B
A
B
A
Yep
anyway
yeah,
so
so
it
sounds,
like
I
mean
it's
possible.
New
relic
will
put
some
put
some
resources
into
helping
me
rewrite
metrics
and
it
sounded
like
the
metrics
sagan
general
was
in
favor
of
it
so
yeah.
I
think
I
want
to
go.
C
B
I
don't
think
so,
yeah
I
I'm
I'm
slowly
just
iterating
on
the
testing
change
thing
I
I
mean
yesterday,
like
I
saw
the
nikita's
experience
I
was
like.
Oh
my
gosh.
I
should
look
at
the
build
a
bit,
so
I
try
to
put
in
some
defunct.
Like
I
looked
at
the
build
scan
for
this
race.
Pr
build
timeout
thing,
nothing
stood
out
right
away
though
it's
like
it's
just.
I
guess
the
build
cache
is
probably
not
working
correctly.
C
B
C
Thinking
that,
because
I
want
to
so
if
we
merge
this.
C
B
C
C
B
C
Cool
can
you
because
I'm
I'm
just
trying
to
clear
out
pr's
here?
Can
you
review
this
one.
B
I
guess
so
the
groovy
thing
so
I
did
want
to
actually
just
touch
on
it.
I
noticed
it
because
I
changed
computers,
but
if
we're
not
using
spotlist
for
groovy,
oh,
I.
B
C
C
B
B
C
B
C
C
B
B
B
C
B
B
A
C
C
Gets
me
is
that
extra
space
below
the
license
header
exactly
I
don't
know.
C
B
A
B
C
C
C
B
Let's
see
so,
I
noticed
my
reload
mechanism
works
for
almost
everything,
except
for
enabling
disabled
instrumentation.
Does
that
sound
expected,
or
I
mean.
B
C
B
Okay,
it
was
only
this
enabled
thing
that
didn't
work.
Otherwise
it's
working.
Fine
surprisingly,
I
thought
it'd
be
harder,
but
it
just
worked
like
so
I
noticed
so.
The
approach
I'm
taking
is
probably
different
than
what
we
had
before,
which
might
actually
solve
our
fork.
Every
issue
like
I
don't
know,
but
you
know
some
of
our
tests.
We
have
like
that
fork
every
to
make
sure
it's.
B
C
So
the
one
place
I've
saw
the
fork
every
recently
was
some
place
where
we
it
was
actually
enabling,
in
one
of
the
instrumentations
kind
of
like
you
were
doing
there.
B
B
I
know
if
you
can
see
it,
but
I
noticed
reset
instrumentation.
B
C
So
I
think
that
install
method
part
when
byte
buddy
when
we
install
byte
buddy,
we
tell
it
to
re-transform
all
classes
that
have
already
been
loaded,
so
it
loops
through.
It
goes.
B
B
C
C
B
I
couldn't
figure
out
how
to
use
that
approach
when
our
agent
is
running
out
of
process,
so
I
used
reset,
which
actually
sounds
like
resetting
instrumentation,
and
this
my
understanding
was
so
I
guess
what
you're
seeing
is.
Maybe
I
didn't
even
call
reset
here
just
reinstalling
bike,
but
you
would
have.
C
So
what
would
be
nice
with
this
java
agent?
Can
we
get
to
the
point
where
we
don't
ever
retransform
classes.
C
Okay,
okay,
so
that's
where
matash
and
I
were
having
a
conversation
on
an
issue
about.
C
Oh
yeah,
oh
sure,.
C
So
this
came
out
of
some
discussions
and
some
reviews,
but
the
idea
was
to
have
config
return.
These
sort
of
config
wrapper,
config
value
objects,
yeah,
and
then
we
could
during
test
we
could
have
both.
We
could
have
a
dynamic
implementation
where
they're
updatable
at
runtime
and
a
version
where
they're
not.
B
B
Okay,
so
to
understand
your
explanation,
so
I'm
calling
this
dot
reset,
but
it
sounds
like
I
didn't
even
need
to
do
that.
If
you
just
reinstalled
by
buddy,
it
starts
from
the
base
classes,
it
doesn't
start
from
the
transform
classes
yeah.
So
it's
just
a
bit
of
extra
work
to
on
like
so
don't
reset,
I'm
assuming
under,
like
it
brings
the
base
classes
back
in.
B
B
B
B
Yeah,
it
works
fine
in
almost
all
of
our
tests,
but
this
flag
was
not
working
with
a
big
config
only
this
one.
So
I
moved
it
into
the
test
and
this
one
is
not
like
the
enabled
stuff.
This
one,
I
think,
is
supposed
to
work,
but
maybe
because
it's
the
bootstrap
class
loader
you're
not
able
to
re-set
it
or
something
or
I
don't
know.
B
C
C
Oh
okay,
would
I.
B
C
C
B
B
We
have
the
hotel,
trace
executors
yeah
this
one.
I
couldn't
get
it
to
pick
up
with
resetting.
C
The
change
I
see
here-
oh
yeah,
I
remember
seeing
this.
C
A
C
Do
the
config
value
yeah,
because.
B
C
B
C
Then
yeah
we'll
we'll
just
doc
dock
it
in
there
and
then,
when
we
hit
that
do
that
other
issue,
then
all
of
that
should
fall
away.
C
C
B
B
C
B
C
Yeah,
I
think
so
yep
I
just
don't
know
how
it's
why
it's
working
currently,
unless
we're
doing
that
for
every
or
we're
getting
lucky,
sometimes
the
ordering,
sometimes
with
that
I've
had
problems
where
the
ordering
like
it'll
depend.
A
A
C
B
B
C
B
B
Well,
let's
not
dig
too
much
into
this
stuff
because,
as
you
said,
yeah,
it's
not
really
worth
it.
It's.