►
From YouTube: 2021-01-13 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
A
A
B
Would
yeah.
B
So
how's
how's
java
how's
java
life.
A
B
C
A
I
was,
I
was.
A
Debug
gradle
running
today,
it's
it's
madness
like
that.
The
code
that
code
is
crazy,.
C
C
B
C
C
A
Done
I
don't
think
so
I
mean
just
probably
a
few
lingering
change
log
updates,
but
I
could
probably
get
those
in
tomorrow.
A
B
A
C
B
B
C
A
C
C
A
A
C
A
A
C
A
With
the
global
metrics,
whatever
it's
called
instance
and
somehow
reverse
the
dependency,
so
it
depends
on.
A
A
C
Also,
what
yeah,
so
what
ted
wants
is
to
make
sure
the
transition
from
rc
to
ga
is
easy.
I
don't
think
he
cares
that
much
about
alpha
right.
There
just
need
to
be
have
that
time
where
we
have
an
rc
of
some
sort
that
gives
them
time
to
migrate
right.
I
don't
know
how
important
this
is.
I
mean
they
could
just
migrate
to
ga
if
they
have
to
migrate.
Rsc
anyways.
C
A
B
C
B
What's,
what's
the
I
mean
the
downside
to
splitting
out
all
the
having
different
globals
for
each
signal.
C
A
C
Yeah
yeah,
I
mean,
I
think
that
is
he
mentioned.
He's
gonna
update
his
doc
for
more
detail
on
not
having
problems
when
updating
from
alpha
rc
or
whatever
it
is
to
ga.
We
might
still
comment.
I
I
mean
I'm
even
okay,
with
just
having
the
depleted
global
meter.
There
forever.
I
mean
it's
one
class,
even
if
there's
some
craft,
it's
not
the
end
of
the
world.
A
C
B
C
Exactly
yeah,
they
would
just
delegate
to
open
tomorrow
and
get
the
meter
provider
and
yeah
every
library
is
going
to
have
depleted
stuff
at
some
point.
So
it's
not
good
yep
that
bad
so
that,
like
again,
it
was
more
theoretical
for
me
to
remove
it,
but
on
the
flips
I
don't
care
that
much.
We
don't
even
have
to
really
remove
it.
It's
just.
A
Right
exactly
you
could
just
it
could
just
delegate
everything
over
to
well,
I
mean
there's,
there's
some
methods
on
there
like
the
s
like
the
the
set
like
the
set
would
be
end
up
being
a
little
weird,
I
think,
like
what
will
we
do
with
the
set?
A
A
A
C
Yeah,
let's
figure
something
yeah
I
mean
the
set,
is
only
there
for
use
from
the
sdk.
I
think
so.
Hopefully
you
can
figure.
We
can
even
the
worst
case
actually
just
comes
to
mind
and
five
seconds
is
to
check
the
stack
trace,
if
not
from
sdk
your
own
exceptions,
stuff
like
that,
do
not
allow
users
to
do
it,
but
that
okay,
I
think
we
have
a
reasonable
story
around
that.
A
I
mean
they're
they're
fooling
themselves,
but
and
I
think
yeah,
I
don't
think
it's
a
big
deal
to
leave
a
hundred
to
maintain
that's
directly
deprecated
lines
right.
B
There
was
also
the
proposal
of
in
the
rc
removing
it
in
the
rc
once
we
know
that
it's
going
to
go
in
that's.
B
Hey,
I
was
just
remembering
I
had
a
question
for
y'all
leighton
from
the
python
group
was
asking
me
about
why
we
have
resource
in
the
span
like
why
not
if
it's
you
know
if
it's
immutable,
why
not
just
add
it
into
the
exporter
like
not
carry
the
extra
memory
around
everywhere.
B
C
B
I
was
just
trying
to
think
of,
I
mean,
I
know
it's.
C
Earlier
and
then
currently,
it
is
possible
to
share
an
exporter
between
two
tracers
that
have
different
resources,
this
code
wise,
whether
it's
something
we
want
to
support.
Of
course,
that's
one
thing,
but
the
spec
doesn't
say
anything
related
to
that
we've
had
this
discussion,
like,
should
exporters
be
allowed
to
be
used
with
multiple
providers.
D
B
And
here
does:
does
this
have
a
reference
back
to
the
provider?
I
guess
the.
C
C
C
B
C
D
C
C
Cool,
I
was
sort
of
thinking
of
this
in
my
head
when
I
had
the
idea
of
having
resource.
Only
in
one
location
like
a
straight
span
has
a
references
tracer,
which
refers
to
trace
provider
with
reference
to
open
telemetry,
which
has
the
resource,
and
I
was
thinking
we
just
go
back
and
then
everyone
would
somehow
have
its
parent
reference
to
get
these
things.
C
B
B
A
C
B
Cool
thanks
yeah
I
was,
I
was
looking
and
I
couldn't
really
think
of.
I
thought.
Maybe
there
was
a
reason
like
immutability
or
something
so
good
I'll,
pass
that
info
back
to
leighton.
I
think
somebody
was
proposing
doing
that
over
there
in
python
and
he
was
looking
at
he
said.
Both.Net
and
java
both
had
the
the
like
this,
so
he
was
second
guessing
himself.
A
C
A
A
C
A
I
don't
think
I
had
anything
else.
Oh
the
only
thing
I
wanted.
I
wanted
if
there
was
any
other
feedback
on
the
my
that
zipkin
example,
where
I'm
starting
to
make
things
a
little
more
separated.
I
wasn't
super
happy
with
the
name
of
the.
What
did
I
call
it
example?
A
Yeah
the
example
configuration
yeah
yeah.
It
seemed
okay
to
me
too.
I
just
couldn't
think
of
anything
better,
but
I
do
like
separating
the
sdk
stuff
from
the
app
and
having
it
be.
Like
a
simple
call,
I
was
bothered
a
lot
by
all
these
examples.
I
have
to
keep
an
sdk
reference
around
and
an
api
reference
around
in
the
example
class,
and
it
just
felt
like
it
was
giving
people
a
terrible
example
of
how
they
should
structure
the
code.
A
B
Yeah
yeah
that
will
help
also
for
the
people
with
the
agent.
I
mean
the
more
that
that
delineation
of
sdk,
config
and
api
stuff.
D
Yeah,
so
this
did
remind.
C
Me,
like
hey,
I
don't
think
this
is
controlled
by
the
spec,
but
one
thing
that's
always
bothered
me
is
that
users
will
often
have
to
expose
the
sdk
tracer
management,
for
example,
as
a
beam
or
a
force
flush
I
mean.
Maybe
if
they
don't
need
force
flush,
it
doesn't
matter
well
shut
down,
they're,
probably
going
to
want
to
do
shut.
C
Or
you
just
as
soon
as
you
create
it,
you
add
a
shutdown
hook
or
something
yeah.
I
think
that's
also
a
common
way
yeah.
So
I
wonder
if
we
wanted
to
separate,
I
mean
again
last
when
I
had
a
meeting.
I
don't,
I
think,
john,
you
were
there
right,
the
meeting
with
ted
and
you
and
like
we
were
discussing
force
flush
and
the
idea
was
just
that
we
don't
want
people
to
use
force
flush.
C
I
hate
it
as
much
as
possible,
but
I
don't
want
people
to
expose
force
flush
to
their
apps
and
shut
down
at
the
same
time,
if
we
had
a
way
to
do
that
which
might
just
be
separating
these
interfaces.
A
Yeah
I've
been
thinking
even
slightly
differently.
A
And
that
is,
it
would
be
really
cool.
I
mean
we've.
We've
got
kind
of
on
halfway
there
with
the
tracer
management,
but
it'd
be
really
cool.
If,
like,
there
was
just
no
way
for
you
to
use
your
sdk
for
getting
access
to
tracers
and
you
had
to
ask
the
sdk
for
an
implementation
of
open
telemetry
that
you
could
then
use.
Then
it
would
just
return
itself,
except
if
you
can't
just
return
itself,
because
it
has
to
do
like
it
has
to
actually
like
have
some
wrapper
around
itself.
A
I
don't
know
it
just
it
continually
bothers
me
that
it's
so
easy
for
people
to
end
up
just
keeping
the
sdk
around
and
using
it
for
instantiation.
I
haven't
come
up
with
a
good
solution
to
that.
A
C
C
A
C
C
D
B
Makes
sense
to
me
the
shutdown,
because
you
do
typically
like,
like
wherever
your
your
app
server
life
cycle
or
whatever
the
higher
level
life
cycle,
is
that
when
you
create,
you
typically
have
some
way
to
register
shutdown
at
that
time,.
D
A
That's
an
that's!
Odd,
odd
vestige
hanging
around.
C
B
Yep,
did
you,
what
did
you
think
honoraga
about
your
idea
of
a
flushable,
open,
telemetry
or
flushable
tracer,
or
you
know
something
like
that
as
an
extent
as
an
extension.
C
That's
right,
hey
that
would
be
extra
motivation
to
go
with
that
yeah
to
do
some
weird
down
casting
to
get
flushed
using
an
extension
rather
than
having
a
whole
interface.
For
I,
like
I'm
good
with
that.
C
A
B
Yeah,
I
liked
the
extension
idea.
Just
from
I
mean
it
didn't
doesn't
feel
like
the
spec
is
gonna
change,
and,
and
it
is
something
that
only
a
small
portion
of
people
need
so
to
be
able
to
point
them
to
a
extension
that
gives.
C
B
And
then
also
we
could
we
could
bridge
that
for
the
agent
we
could
bridge
the
extension,
which
would
be
nice.
A
A
D
A
But
yeah
that's
a
good
idea.
If
we
could
get
that
in,
I
guess
I
was
really
hoping
we
could
release
candidate
at
the
end
of
the
month.
A
A
A
A
A
A
All
right
cool!
Well,
I
will,
I
will
check
on
the
release
in
the
morning
and
see
how
things
look.
Yeah,
release
notes
should
be
like
the
what's
in
the
changelog
should
be
pretty
close
to
release,
notes
ready
just
a
couple
extra
little
bits
that
happened
since
4
o'clock
today.
A
D
C
B
This
was
I
I
noticed.
There
was
also
the
spec
issue
talking
about
resources
as
api.
A
A
C
C
A
Yeah
we
taught
he
and
I
talked
about
that
for
a
while,
and
I
think
he
gave
it
a
try
and
it
ended
up
being
a
little
bit
tricky
because
I
don't
remember
why
I
think
he's
you
can
take.
I
mean
you
can
take
a
look
at
the
splunk
agent.
B
For
logging,
I
mean,
if
you
want
to
put
them
in
if
you
want
to
encode
it
in
the
output,
I
mean
you
need
to
get
it
in
to
like
system
properties
or
mdc
or
something
that.
A
I
think
honorable
saying-
and
this
is
what
was
actually
my
suggestion
as
well-
the
vladimir
was
you
could
create
a
little
bridge
class
in
the
bootstrap
that
could
get
the
date,
get
the
information
and
expose
it
to
any
anybody
who
needed
it.
A
B
It
oh
its
property,
but
it
is
the
java
util
properties
like
if
you
do
system
get
properties,
but
I
think
you're
right
that
there's
nothing
guarding.
I
don't
know.
A
D
A
D
A
A
little
tricky
to
figure
out
when
you
can
get
a
handle
on
if
you're
in
the
bootstrap.
How
do
you
get
a
handle
to
another
class
loader's
resource,
though
right,
because
that
lives
on
that
often
the
agent
the
agent
would
have
like
the
agent
class
loader
land
would
have
to
push
it
into
the
bootstrap
right,
yeah
yeah.
I
suggested
this
as
well.
He
looked
into
it.
I
think.
Whatever
he
came
up
with,
it
was
a
little
bit
simpler.
C
A
C
B
Would
honor,
how
would
you
would,
how
would
you
get
it
from
the
map
and
the
bootstrap
class
loader
into
the
log
statement.
B
B
A
B
D
A
So
it
was,
we
had
a
funny
conversa
vlad
and
I
had
a
funny
conversation
because
he
started
asking
these
questions.
I'm
like
this
is
definitely
not
something
we
would
support
officially
as
maintainers
and
he's
like.
Oh,
I
know
I'm
just
hacking,
I'm
like
all
right.
If
we're
in
hacking
mode,
then
I
will
just
take
the
gloves
off
and
yeah
yeah
exactly
yeah.
C
I
remember
once
I
think
I
had
a
lot
of
weirdness,
because
I
was
copying
all
the
gradle
system
properties
into
an
app
system
properties
automatically,
so
that
you
can
control
your
apps
using
the
gradle
command
line
and
then
that
would
from
intellij
debug
mode
that
had
some
collision
or
something
and
caused
craziness
for
a
week
or
something
until
I
figured
that
out.
So
ever
since
then,
I've
been
scared
of
system
properties,
yeah
yeah,.
B
B
Do
you
do
either
you
have
like
inclination
on
where
that's
gonna
go
as
far
as
instrumentation
land,
because
it
does
seem
like
something
that.
B
Some
people
will
want
like
from
I
don't
know,
instrumenting
I'm
trying
to
think
of
how
that
would
work
and
how
we
would
bridge.
I
mean.
B
B
C
C
B
You
had
showed
us
a
cool
way
that
people
were
launching
an
agent
in
maine
at
the
beginning
of
maine
was
that
there
was
some.
C
D
C
B
I
mean
we
can
re-transform
classes,
we
can't
add
fields,
but
we
fall
back
to
that
map.
Okay,
so
for
the
most
part
we
can.
D
B
Except
in
cases
I
remember
because
pavo
was
asking
about
the
dynamic
attach
yep
and
he
was
pointing
out
that
something
like.
D
C
Well,
I
mean,
ideally,
we
don't
have
to
go
too
crazy,
like
if
there's
some
possibly
novel
way
of
bridging
configuration
in
from
an
app
into
the
agent
like,
even
if
it
means
anytime.
We
find
that
we
just
re-initialize
the
sdk.
I
mean
that's,
not
a
huge
deal.
I
think
worst
cases
you
missed
a
span
or
something,
but
maybe
that's
expected
in
such
a
situation
anyways.
D
B
C
C
Yeah
I
mean
I
use
even
not
in
spring,
but
I
always
use
that
pattern
like
application
alpha
for
dev
and
application
fraud
in
the
same
binary
that
can't
work
with
agent
right
now.
So
if
there
was
a
way,
I
don't
think
this
is
any
high
priority
for
now,
but
this
would
also
solve
the
resource,
because
any
resource
could
also
just
be
configured
using
that
mechanism.
B
C
It
could
just
be
based
on
strings
for
the
most
part,
I
think
so
it
might
to
miss
things
like
custom
sample
implementation
or
whatever.
C
B
To
trying
to
bridge
to
something
existing,
I
think
that
that
feels
like
it
could
work.
Yeah
the
I
guess,
the
main,
the
the
the
complete
the
except
in
app
servers
with
multiple
war
files.
That's
always
our
downfall
on
the
agent
side
right
where
we're.
C
B
A
C
A
Is
not
correct,
what
we're
going
to
do
is
there
is
going
to
be
at
least
as
far
as
I
understand.
I
think
this
is
what
is
currently
written
into
the
pr
that
the
sdk,
maybe
on
the
resource
class,
as
an
extra
method,
will
expose
a
new
resource,
which
is
just
the
things
that
are
all
the
defaults.
Basically
has
the
defaults
in
it.
A
Right
but
get
default
right
now,
yeah
it
does
all
the
providers
and
stuff.
This
will
just
be
something
that
returns
well
right
now,
just
the
service
name.
But
if
we
add
instance
id
is
the
other
thing
that
people
are
that
josh,
especially
mcdonald,
is
thinking
needs
to
be
also
a
mandatory
resource
attribute
for
metrics,
especially
that
that'll
be
like
there's
just
another
accessor
accessor,
which
will
give
you
that
thing.
A
So
we
might
need
to
figure
out
we're
gonna
figure
out
naming
for
that
like
get
empty
with
defaults,
or
something
like
that
or
maybe
with
defaults
or
something
if
there's
something
empty,
with
fallbacks
or
anyway.
Something
like
that
and
then
any
exporter
that
needs
to
have
a
service
name
can
just
use
that
thing
to
merge
into
the
merch
or
yeah.
I
think
that
was
the
idea
and
then
into
order
to
make
murder
work
right.
You
have.
A
Yeah,
so
I
think
what
that
means
is
that
I'll
close
both
of
my
outstanding
prs
with
the
two
resource
default
options
and
do
a
third
one
that'll
do
this
weekend
and
then
we'll
need
to
rip
out
the
the
service
names
out
of
the
jager
and
zipkin
exporter,
configs,
so
cool
anyway,
I'll
probably
as
soon
as
that
gets
merged.
I'll
pick
that
up
and
crank
that
out
pretty
fast.
Hopefully,
that
will
only
be
the
next
one.
So.
B
Are
we
going
to
get
a
service
name
environment
variable.
B
A
This
particular
pr
isn't
isn't
adding
that.
B
D
C
B
C
It's
an
improvement,
yeah,
sorry
to
jump
back
and
forth.
I
remember
one
question
with
metrics,
mainly
triggered
by
seeing
that
refactor
to
have
them
the
meter
provider
implement
metric
producer
or
whatever,
and
as
I
wrote
this
configuration
stuff,
I
still,
I
really
hope,
there's
a
way
to
make
it
fluent
like.
Have
you
ever
thought
of?
Is
there
no
way
to
make
the
sdk
meter
provider
builder?
Then
three
point
everything
rather
than
having
two
entry
points:
it's
just
not
possible.
C
A
I
suspect
my
gut
tells
me
that
a
bunch
of
this
is
going
to
change.
Okay,
because
there's
still
a
desire
to
have
the
ability
to
have
both
push
and
export
in
the
same
sdk,
and
I
think
it's
going
to
require
some
significant
rework
to
that
part
of
things.
C
A
A
I'm
going
to
miss
the
middle
of
it
because
we're
having
our
holiday
celebration
for
my
team
that
splunk,
so
the
second
half
may
be
a
little
tipsy.
Let's
just
put
it
that
way,.
A
Yeah,
I
do
not
have
high
hopes
for
quality
outcome
from
that
meeting,
but
we'll
see
what
happens.
C
A
C
C
C
D
C
D
A
Yeah,
then,
you
get
to
implement
micrometer
in
every
language
in
php
you're
going
to
have
to
implement
micrometer
and
php.
B
I
have
I
have
a
new
php
friend
from
new
relic
johannes.
D
A
A
Are
you,
are
you
working
closely
with
him
at
all.
A
He
everyone's.
B
Yeah
yeah
yeah
from.
C
A
B
A
A
B
B
Trying
to
blink
but
yeah
for
some
reason
I
thought
he
said
somewhere
in
portland.
I
don't
know.
B
Honorary
just
wanted
to
on
this
one
I
mean
we'll
we'll
definitely
want
to
check
with
nikita,
because
I
thought
that
this
I
remember
having
the
same
question
as
you
did
like
it
seems
like
overkill
and
so
yeah.
So
I
had
asked
a
bunch
of
questions
on
this
one
where
they
wanted
to
be
able
to
exclude
specific
methods.
D
B
And
here
I
think
I
finally
unders
got
them
like
it's
over
instrumenting,
so
they
have
customers
who
have
over
instrumented
using
the
trace
annotation,
which
was
the
old
data
dog.
Now
it's
with
turned
into
with
span.
C
B
Yeah,
let
let's
leave
it
there
and
get
we'll.
B
C
B
B
Okay,
because
it
should
be
because
I
thought
I
fixed
all-
of
the
properties.
B
Names
to
be
more.
C
C
B
Oh
and
the
constant
is
still
that,
but
yes,
the
name
is
now
this
very
long
thing,
but
at
least
it's
like
scoped
to.
C
C
C
Wondering
really
that
being
said,
it
might
have
like
I've
definitely
got
an
impression
from
seeing
the
prs
from
spunk
folks
and
from
hearing
people
say
that
splunk
is
very
expensive.
Is
that
I'm
imagining
all
their
customers
are
able
to
see
monolithic,
apps
that
are
running
on
java
e
and
all
this
stuff?
So
I'm
sure
they
see
a
lot
of
weird
code.
So
maybe
that's
just
what
it
is.
B
Yeah,
I
think
that
is,
I
mean
the
sort
of
the
the
bread
and
butter
for
the
the
apm
companies
like
new
relic
and
yeah,
or
the
big
just
big
java
shop
java
shops,
enterprise
shops,
yeah.
B
B
Yeah
I
was
just
explaining
to
a
pm
this
week
that,
because
we
had
some
some
customer
that
was
asking
for
instrumentation
of
some
library
and
like
we
don't
even
have
that
in
open
telemetry.
Like
you
know,
I
don't
really
you
know.
If
none
of
splunk's
customers
need
it,
then
it's
not
very
likely
that
you
know
it's
really
that
critical
yep
so.
C
B
It's
it's
very
good
to
have
splunk
on
board
with.
C
B
D
B
Thinking
for
powell,
potentially,
that
would
give
him
some
more
ability
to
help
with
releases
if
he
wanted
that.
C
B
Yeah,
I
think,
those
anyway
yeah
I'll.
I
was
gonna,
wait
for
nikita
to
come
back
yeah.
C
B
Yeah
yeah,
I
wonder
what
like
is
there
something
we
can
ask
like?
I
know
we've
kind
of
discussed
in
the
past
for
maintainer
time.
The
like
I
mean
it's
quite
consuming.
D
B
C
C
B
I
was
just
curious
what
their
pr
totals
are.
C
B
Yeah,
I
kind
of
noticed
that
on
the
java
repo
there's
a
lot
of
there's
a
lot
of
not
acting.
C
D
B
C
B
Yeah,
no,
that
that's
actually
I
like
that.
It's
a
good
thing
to
to
plan
for.
B
Cool
anything
else
you
wanted
to
anything
else.
On
your
mind,.
B
C
C
B
I
know
the
so.
The
problem
is
that
the
team-
my
when
my
when
I
switched
got
my
manager
left
like
what,
like
in
september
or
something
and
the
person
who
left
with
him
was
were
they
were
the
ones
who
were
really
interested
in
that
they
were
trying
to
drive,
auto
instrumentation,
in.net
and
so
before
they
had
left,
though
they
had
gotten
the
net
core
team
interested
in
bytecode,
instrumentation
and
potentially
building.
D
B
C
Dot
net
instrumentation
that
repo
has
almost
no
development,
and
so
but
apparently
someone
else
is
doing
it
and
I
don't
know-
and
so
this
list
is
to
then
help
net
instrumentation
frame.
What
features
are
important
for
agents
regardless
of
whether
it's
java.net?
So
that's
the
background
for
this
yeah.
B
I
remember
yeah,
there
was
some
discussion
with
so
my
the
person
who
left
who
was
talking.
He
was
talking
with
the
splunk
splunk
and
datadog
and
new
relic
in
the
net,
auto
instrumentation
group,
and
they
were
all
at
that
point
trying
to
decide
on
the
foundation
of
like
which,
which
companies
agent
they
would
start
from,
and
so
from
what
I
heard.
There
were
two
votes
for
datadog
because
splunk
it
was
with
signalfx.
B
They
were
already
had
already
forked
datadog's
agent
stuff,
so
it
sounded
like
it
was
going
in
that
direction.
D
B
C
D
B
For
a
day,
just
in
case,
you
want
to
take
a
look
at
them
unless
they're
a
couple
trivial
ones.
I
emerged
this
morning.