►
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
D
Is
this
I
heard
this
was
a
is
a
light
step
holiday,
andrew.
B
Yeah,
this
is
a
it's,
not
a
local
or
or
government
holiday.
D
B
John
watson
light
stamp
holiday,
we're
like
that
guy
he's
his
own
holiday,
so
we're
going.
C
Happy
birthday
then
yeah
happy
birthday,
yeah,
it's
the
big.
I
turned
50
today.
E
D
Oh,
my
wife
and
I
are
going
out
to
dinner
tonight,
there's
a
restaurant
that
has
outdoor
seating
that
we
like
that.
We
haven't
been
out
to
dinner
in
so
man
so
long
for,
like
let's
go,
do
it
and
then
tomorrow
my
wine
buddies
and
I
are
getting
together-
we're
gonna,
drink
some
old
old
wine.
I
have
a
bottle
of
1955
cote
d'iro
that
I'm
gonna
open
up.
C
B
A
C
E
D
E
E
E
D
D
B
Just
an
fyi,
but
actually
the
real
meat
of
it
comes
with
carlos's
item
that
we'll
talk
about
later
that
the
spec
trace
items
are
the
scoped
focus
of
top
priority
for
ga
and
all
those
spectra
sizes
the
p1
issues.
So
of
like
something
89
issues,
that's
required
for
ga
40,
something
in
spec
and
then
12
that
are
p1
for
specs.
That's
what
we're
focusing
on
first.
D
D
Especially
if
logan
isn't
here
to
chime
in
on
the
next
item,
which
is
about
async
export,
changing
the
exporters
to
be
async
carlos,
you
have
reservations.
E
Yeah
I
mean
I
am
not
opposed
to
that,
but
my
initial
feeling
is
that
it's
something
we
don't
need
and
we're.
We
were
trying
to
solve
a
problem
that
we
don't
have,
but
given
the
fact
that
I
don't
remember
the
name
of
the
author
of
this
pr
here,
he
well
he
worked
more
along
along
and
to
to
have
some
some
more
like
code
for
actual
showing
what
he,
how
he
envisions
this
as
an
improvement,
I'm
planning
to
go
and
do
a
review
this
weekend.
E
D
It
is
sdk,
sdk
change,
so
it's
not
and
it's
actually
like.
If
someone
is
writing
an
exporter
and
they
have
to
change
it,
they
can
change
it
with
like
one
line
of
code.
It's
not
going
to
be
optimal,
but
it's
a
very
it's
a
very
simple
change
from
the
consumer
side
or
from
the
user
side,
but
yeah,
and
it's
not
something
I'm
spending
any
time
on.
It's
all
just
contributed.
So
that's
good.
E
Yeah,
if
you
ask
me,
that's
something
that
we
could
do
after
ga,
but
since
somebody's
trying
this
out,
we
can
go
and
review
that.
D
Yeah
yeah,
I
think
I
think
I
don't
think
it's
needed,
but
I
don't
also
don't
think
it's.
I
don't
think
it's
a
bad
idea.
The
author
of
that
is
he's
one
of
the
authors
of
one
of
the
async
or
has
worked
in
one
of
the
async
scala
frameworks
and
I
think
he's
trying
to
integrate
it
into
some
sort
of
scala
exporter,
and
it
was
going
to
be
very
helpful
for
that.
So
I
think
you
know.
E
D
E
D
Yeah,
so
the
bug-
and
we
were
just
chatting
about
this
pr
that
is
put
in
by
an
external
contributor-
about
making
the
exporters
async
and
it
will
help
so
someone
who's
a
contributor
or
maintainer
on
one
of
the
one
of
the
scala
async
frameworks,
and
it
was
going
to
be
something
very
helpful
for
them
and
that's
not
something
we're
spending
a
lot
of
time
on.
D
So
I
think
it's
probably
not
a
bad
idea
doesn't
actually
harm
anything
so
just
need
to
figure
out
if
we
want
to
move
forward
with
it
or
not.
D
Option,
I
think
it
would
be
well,
you
wouldn't
have
to
do
it
asynchronous,
because
you
can
immediately
return
a
future
like
a
few.
You
can
just
return.
Have
it
be
synchronous
and
return
the
future
if
you
want
to,
although
it's
probably
not
optimal,
yeah,
it's
using
the
java.
F
D
Yep
yeah,
exactly
I
don't
think
it's
a
big
deal.
He's
done
a
lot
of
good
work
and
honorable
pitched
in
and
helped
out
to
give
it
like
change
the
grpc
or
the
oh
otlp,
one
of
the
exporters.
I
don't
remember
which
one
I
think
otlp
exporter,
it's
actually
really
async.
So
I
think
it's
probably
a
good
change.
F
F
So
in
the
collector,
maybe
it's
not
the
case,
but
in
the
collector,
because
we
have
like
we
are
like
a
sync:
we
we
get
data
from
a
bunch
of
things
and
we
have
this
exporter
as
the
only
way
to
export
things.
We
we
sometimes
have
to
queue
before,
because
the
exporter
is
not
fast
enough
to
catch
up
with
the
with
the
amount
of
data
that
are
coming
so
sorry
think
about.
If,
if
that's
something
that
you
may
want
to
to
implement,
how
would
you
implement
with
this
model?
F
It's
it's
just
an
idea
of.
If
this
is
possible,
then
yeah.
D
Cool
all
righty:
let's
move
on
to
talk
talk
about
the
char
sequence
idea,
so
I
saw
your
I
looked
through
your
comments
and
they
all
seemed
totally
reasonable.
Do
we
wanna?
So
I
am
I'd,
be
happy
to
try
to
take
what
I
have
written
and
make
sure
the
internals
are
always
storing
us
along
and
see
how
that
looks
so
that
we
know
that
it
will
be
immutable
and
unchangeable.
F
F
So
because
of
that,
we
we
have
to
to
make
these
things
immutable
and
so
far
I
saw
these
two
options.
We
either
stored
things
in
the
span
context
as
longs,
which
are
immutable
because
we
force
them
to
be
immutable
or
we
use
strings
which
are,
by
definition,
immutable
guaranteed
by
jvm.
Both
of
these
options
are
fine.
I
had
a
question
about
the
char
sequence:
subsequence
stuff.
Did
you
see
that
one?
I
think
it's
important
in
order
to
determine
the
api
boundaries,
and
if
that's
the
case
that
I
was
believing
should
be
the
case.
F
If
that's
the
case,
then
the
api
should
look
like
what
I
propose
more
like
that
gives
us
freedom
to
change
the
the
implementation
in
the
back
and
in
the
back
we
can
start
with
longs,
and
if
we,
if
we
see
that
we
do
too
many
conversions
to
string
at
one
point,
we
can
change
internally
to
store
directly
streams.
So
both
are
valid
options
for
us.
F
F
Initially,
I
thought
that
that
accepting
bytes
and
charge
sequence
would
allow
us
to
change
the
size
of
these
ideas,
but
actually
that's
not
true.
The
reason
why
is
not
true
is
because
we
have
a
contract
and
we
have
some
some
things
to
say.
We
expect
this
to
be
16
characters,
even
though
api,
wise
or
or
the
the
the
boundary
allows
you
to
to
say
from
now
on
will
accept
24.
F
F
The
the,
as
you
said,
the
biggest
problem
for
me
right
now
is
the
mutability
part.
Once
that
is
solved,
I
I
don't
really
care
which
version
we
go
as
long
as
in
my
mind,
changing
the
story
to
be
longs
or
string
should
be,
should
not
require
the
api
changes.
If
that's
what
we
achieve,
then
we
are
in
a
super
good
shape
that
we
can.
We
can
play
with
whatever
we
choose.
D
Yeah
and
I
think
your
question
about
whether
what
the
jvm
does
when
you
call
two
string
on
a
char
sequence,
that
is
a
string,
it
does
just
return
itself.
I
know
that
that's
definitely
the
case,
and
so
I
think
if
we
do
an
internal
storage
as
string
when
they
give
us
a
char
sequence
in
almost
every
case
it
will
work.
Just
it
will
work
just
fine
or
we
could
actually
ask
we
could
even
do
it
like
a
type
check
and
see.
D
F
Cases
when
the
string
is
already
16
characters
versus
the
string
is
not
already
16
characters
and
I
think
it
does
different
things
based
on,
because
if
I
give
you
a
buffer,
I
mean
in
my
proposal
in
my
proposal,
I
am
giving
you
a
charge
sequence
and
an
offset
where
in
this
buffer,
you
find
the
trace
id
and
my
mind
was
like.
Okay,
we
get
a
only
one
string
for
the
for
the
trace
for
the
trace
context
in
the
w3c.
E
F
F
F
The
other
option
is
to
have
a
trace
id
in
the
sdk
and
pass
that
and
and
just
just
say
that
hey
this
is
the
class
that
we
pass.
It's
not
exposing
the
api,
it's
just
an
sdk
interface
or
whatever
your
final
class.
I
don't
know
if
you
need
an
interface
or
a
final
class,
because
it's
in
the
sdk
and
past
that
I
think
it's
sorry.
F
D
A
F
So
we
have
to
lock
it
somehow
at
one
point:
that's
why
that's
why
all
these
trade-offs,
maybe
maybe
we
create
our
own
class,
to
give
us
flexibility
about
the
internal
implementation
of
that
thing,
but
still
user
sees
it
as.
D
F
I
was
just
pinging
nikita
to
explain
to
me
again
I'm
still
failing
to
understand
some
of
the
shading
things,
but
he
will
explain
probably
next
week
so.
D
Cool
alrighty:
I
will
take
a
look
at
that
all
of
that
next
week
and
keep
playing
with
the
ideas
all
right,
carlos
triage.
F
Yeah
sorry
before
this
one
last
thing,
I
am
all
about
removing
the
ids
classes
and
store
everything
into
spam
context.
It's
a
matter
of
how
do
we
put
things
together?
How
do
we
massage
the
api
and
stuff,
but
I'm
all
about
doing
that?
I
think
it's
overall
we
we
get.
We
should
get
a
lot
of
improvements
if
we,
if
we
just
simply
blindly
move
these
longs
and
create
the
right
apis
to
to
to
access
the
ids
from
the
span
context
directly.
F
We
should
get
a
lot
of
improvements
because
we
no
longer
unbox
and
box
things
in
the
in
the
api.
E
Yeah
two
things:
the
first
is
that
kind
of
obvious
is
that
you
know,
based
on
the
tough
love
that
we
got
last
monday
in
the
mountaineers
meeting,
I
will
be
trying
to
spend
more
cycles
in
specification
trying
to
help
there,
so
I
will
be
reviewing
less
and
less
things
well,
not
like
less
and
less,
but
in
comparison
listings
on
the
java
repo.
So
I
will
be
trying
to
rely
on
you,
john.
I
will
be
trying
to
review
things
that
are
important
like
the
charge,
sequence,
changes
or
well.
E
You
know
the
ideas
that
we
just
talked
about
and
the
async
exporters,
so
that
that's
for
your
information-
and
this
is
of
course
just
to
just
to
clarify
that
I'm
planning
to
spend
more
cycles-
and
the
second
item
is
related
to
that,
and
is
that,
as
morgan
mentioned
also
last
monday,
we
are
trying
to
set
a
date.
E
D
E
I
can
make
wednesday
work
perfect,
sweet
and,
by
the
way,
john,
since
I
will
be
well
not
not
mostly
gone,
but
not
spending
so
many
cycles,
I
can
probably
have
sergey
who
already
you
know,
created
a
pair
of
small
pr's
help
here.
Given
you
know
it,
we
will
see
based
on
the
items
that
we
get
that's
important.
D
Yeah-
and
I
know
anurag
is
starting
to
do
a
lot
more
reviews
as
well
and
giovanni
as
well,
so
the
reviews
are
helping
it's
good,
investing.
D
F
F
D
I
actually
did
have
one
more
issue:
somebody
we
had
an
external
bug
report
that
I
don't
remember
the
number
off
my
head.
I
can
look
it
up
that
the
counters,
when
you
bind
the
labels
to
a
counter
the
bound
counter
you
get,
even
if
you
make
no
recordings
on
it
all
you
don't
do
any
increment
or
any
add
on
it.
We
still
will
report
a
zero
in
the
to
the
exporter
every
every
collection
cycle.
D
We
don't
do
that
when
it
isn't
bound
because
we
save
off
that
aggregator.
When
we
do
the
binding,
we
save
off
the
aggregator
until
it's
completely
unbound
and
there's
no
more
references
to
it
yeah,
and
so
I
think
this
is
the
behavior
right
now.
The
real
question
is:
is
that
okay?
Is
that
a
bug?
Is
it
a
just
happens
to
be
a
weird?
It
is
idiosyncrasy
about
bound
counters,
I'm
not
sure.
F
To
be
honest,
I
think
I
try
to
to
avoid
that.
It
means
that
it's
a
bug
I
just
double
check.
If
I
added
I
will,
I
will
double
check
after
this
meeting,
but
I
I
remember
that
I
added
is:
is
change
an
atomic
boolean
there
that
that
was
flipped
whenever
I
make
a
change,
but
let
me
let
me
double
check
if,
if
that's
the
case,
it's
it's
a.
D
F
It's
not
it's,
not!
It
shouldn't
be
in
the
counter.
It's
in
the
bounds
things
every
time
when
you
do
any
so
so
all
the
logic
is,
is
independent
of
counters
or
anything.
It's
just
like.
I
have
a
bound
instrument
and
then-
and
then,
if
I
don't
remember
if,
if
I
was
worried
about
performance
and
did
not
put
that
in
but
every
time
when
you
do
a
record,
you
have
to
flip
a
bit
or
something
to
say
change
and
every
time
when
you
do
collector
you've
unflipped
that
to
unchange.
F
You
you
have
to
keep
them
in
the
map,
because
that
person
still
has
a
pointer
to
this,
and
I
have
to
keep
them
in
the
map
until
they
call
them
bound.
So
now
now
it's
to
determine
if
that
entry
in
the
map
had
a
change
for
things
that
are
not
unbounded,
I
simply
remove
them
from
the
map.
If
they
appear
again,
they
have
a
change.
But
for
these
things
I
I
think
I
was
thinking
to
add
this
bullen
that
I'm
flipping
every
time.
F
Maybe
there
is
a
bug,
but
I
think
that
was
the
idea
and
it's
not
generic
to
counters.
It
will
be
to
every
instrument
if
it's
happening,
it's
happening
to
all
the
synchronous
instruments.
There
is
no
special
logic
for
for
counters.
D
Yeah,
I
was
just
thinking
that
the
counter
store.
Well,
I
I
don't
see
that
atomic
boolean.
I
think
you
didn't
end
up
putting
that
in,
but
I
think
we
could
probably
put
that
in
without
a
big
deal.
The
only
difference
I
think
the
difference
with
counters
is
they
store
everything
in
just
an
atomic
long,
and
so
you
can't
tell
whether
it's
changed
or
not
so
yeah.
We
need
to
have
that.
We,
I
think
the
other
ones
like,
for
example,
the
the
min
max
sum
count.
D
F
It's
not
the
counter
is
the
aggregator,
the
sum
that
does
not
yeah,
but
I
I
think
I
think
what
I
tried
to
do
was
to
decouple
the
aggregator
with
the
map
and
the
changes
in
the
map
with
this
with
the
bachelor.
What
we
call
so
in
that
bachelor
the
bachelor
should
know
if
entry
was
alternated
or
not,
and
as
I
said,
I
am
pretty
confident
I
try
to
put
that.
F
I
I
added
it
in
the
in
the
goal,
because
I
made
the
same
change
in
the
goal.
I
don't
know
if
you
know
but
yeah
we
start.
I
did
the
same
idea
in
the
go
with
that
reference
counter
magic
thing,
but
I
don't.
I
don't
remember
if
I
put
a
bullet.
D
F
We
have
two
options.
We
either
say
working
as
intended
for
performance
reasons,
which
I
don't
believe
it's
it's
true.
I
you
can
prove
me
that
doing
the
atomic
thing.
It's
it's
good
enough,
because
you
don't
you,
don't
you
don't
need
to
read
the
the
value,
so
you
just
do
a
set
true,
all
the
time
blindly
and
yeah,
and
you
just
on
the
on
the
right.
You
do
a
swap
to
to
false
all
the
time
and
yeah.
D
E
D
D
F
Document
that
it
may
happen
in
this
case
on
general
cases
shouldn't
happen,
but
it
may
happen.
Probably
that
was
the
reason
why
I
didn't
add
it,
because
I
was
thinking
I
either
do
it
correctly,
which
I
cannot
without
having
a
real
performance
problem,
or
we
can
do
a
best
effort
on
this
and
say
hey.
You
know
what
sometimes
we
may
we
know
about
this.
Sometimes
you
may
get
zero.
D
D
E
E
So
I
was
wondering
if
that
can
be
considered
as
a
potential
thing
for
a
patch
release.
What
do
you
mean?
Is
it?
Let
me
let
me
look
for
the
actual
pr
in
a
second.
I
know
that
we
are
over
time,
but
sorry
I
totally.
I
was
totally
meant
to
do
that,
nine
to
twelve
times
we
have
until
8
45,
officially
right,
but
still
we
were
kind
of
trying
to
well
yeah,
never
mind
there.
You
are
so
it's
1518,
but
I'm
gonna
paste
it
here.
F
E
D
E
F
By
the
way,
there
are
instructions
how
to
cherry
pick
and
stuff
in
the
release.
I
think
I
added
them.
Please
follow
that
because
it
keeps
links
between
pr's,
fantastic
yeah,
so.