►
From YouTube: 2021-07-02 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
B
B
E
B
B
So
I
saw
honorary
that
you
you
thumbs
up
the
monday,
the
switch
to
tuesday
from.
A
I
might
be
more
likely
to
stay
up
for
that,
since
I
do
have
video
gaming
with
friends
on
monday
nights,
so.
A
D
B
All
right,
so
one
four
zero
target
a
week
from
tomorrow,
and
so
I
think
the
main
the
only
open
question.
Well,
there
were
two
open
questions.
One
is
the
parameter
annotation
stuff
from
halo
4..
B
So
the
the
agent
support
is
ready.
Mateish
was
just
asking
for
a
second
reviewer,
so
I
volunteered
to
review
by
monday,
so
that
we'll
know
basically,
if
it's
good,
we'll
pull
it
in
and
then
we'll
keep
the
annotations
in
the
sdk.
Otherwise,
if
we
really
think
it's
a
you
know
a
ways
off
which
I
doubt.
E
D
D
A
It
was
the
http
client.
D
A
E
A
D
A
It
almost
feels
like
we
need
some
sort
of.
We
need
to
create
a
new
utilities
project
that
has
like
this
kind
of
generally
useful
stuff.
The
clock
could
go
in
there,
for
example,
there's
all
sorts
of
good
stuff
that
could
end
up
in
the
generally
useful,
open,
telemetry
utilities
bucket.
B
E
D
A
B
A
A
E
D
D
B
D
A
D
D
An
ac
bureau
connection
did
you
know
like
it's
implemented
with
http
client
in
java
11
up
if
you're,
if
you've
ever
seen
the
stack,
traces,
yeah
yeah,
it
doesn't,
I'm
sure
they
disable
the
like.
I
doubt
it
supports
http
2.
that'd
be
the
one
difference
in
features.
Otherwise
it's
probably
the
same
stuff
yeah.
A
A
D
A
C
I
mean
I
did
see
that
call
out
on
that
on
that
issue
from
trask
he's
like
handmade
issue
like
ready-made
benchmarking
issue
jason.
Here
you
go
no,
so
this
is
kind
of
involved
in
the
main
reason
that
iguanodon
has
not
been
contributed
just
like
straight
up
yet
because
that
was
on
the
plan
we
talked
about,
like
I
don't
know
two
months
ago
or
something
so
there's
this
like
increased
interest
in
getting
some
some
sort
of.
I
don't
know
like
regression
is
kind
of
built
in,
but
it's
mostly
like
overhead
testing.
C
I
do
in
large
part
because
customers,
always
you
know,
come
come
with
the
statement
that
the
agent
introduces
overhead
and
I'm
like
yeah,
that's
what
it
does.
C
It
has
to
right
so
being
able
to
quantify
that
or
at
least
having
a
baseline
understanding
of
what
it
looks
like
for
a
representative
or
a
sample
application.
I
think,
is
useful
because
at
least
gets
the
dialogue
rolling
and
so
we're
figuring
out
exactly
what
we
want
to
do
right
now.
The
thought
is
to
put
it
in
into
contrib.
C
And
the
idea
is
that
it
would
like
run
nightly
and
it
would
eat
multiple
configurations
of
which
I
think
to
start
this,
probably
like
one
or
two:
it's
like
run
with
the
agent
and
no
agent
like
that's
a
no-brainer.
We
have
to
do
that
and
then
possibly
run
with
release
and
snapshot
for
comparisons
right.
So
you
know
snap,
like
latest
snapshot
versus,
let
latest
stable
release.
C
Those
are
the
initial
configs,
I'm
thinking
and
then
to
start.
You
know
the
you.
I
have
this
ui
stuff
in
iguanodon
and
it's
great
you
know,
but
none
of
us
are
really
javascript
programmers
or
react
programmers,
and
so
the
initial
thought
is
just
to
publish
the
data.
Just
have
the
csvs
out
there
and
then
allow
the
next
phase
to
sort
of
figure
out.
Ui.
D
B
Why
why
not
I
didn't
know
like
I
tend
to
leave?
Well,
I
I
think
I
know
why
you
hate
monorail,
you
hate
the
model
repo,
but
my
feeling
is
keeping
it.
You
know
keeping
it
close
to
the
code
that
it's
testing
is
beneficial.
Also.
C
Yeah,
I
guess
I
guess
my
leaning
toward
contrib
is.
Is
it's
probably
exactly
what
trust
like
you
know
me
well
enough
by
nowadays
you
know
my
feelings
on
monorepos
and
that's
why
it
lives
in
its
own
repo
right
now.
D
D
D
C
Okay,
I
will
go
back
to
the
team
with
this
idea,
because
I'm
not
opposed
to
it.
I
mean
the
the
repo
is
very
big
right
now
and
adding
stuff
to
it.
Another
consideration,
though,
is
also
this
use
of
the
page,
the
gh
pages
branch
to
to
have
any
sort
of
ui
and
results.
I
don't
think
we're
using
that
for
anything
right
now.
So
are
we
opposed
to
having
it's
probably
linked
a
little
bit
deeper,
but
are
we
opposed
to
having
any
sort
of
gh
pages
on
the
repo.
C
C
C
Yeah,
that's
fair.
You
know,
there's
also
that
effort,
that's
being
done
by
the
intern,
doing
the
open,
telemetry
to
open
tracing
comparison.
B
C
B
No,
the
that
project
is
fairly
well
scoped
to
specific
instrumentations
and
more
less
end-to-end
testing
and,
more
like
true
a
little
bit
more
true
micro,
benchmarking,
okay,.
C
C
E
C
B
A
Time
through
so
josh's,
you
know
we
all
know
working
on
rewriting
metrics
to
conform
to
new
apis
and
sdk.
A
You
know
some
of
it
prototype
spec,
so
the
conversion
of
the
s
or
the
api
to
use
attributes
instead
of
labels
that
got
merged
that
was
3352..
A
3353
introduces
exemplars
into
the
sdk's
data
model
and
maps
it
to
both
prometheus
and
otlp,
and
I
have
gone
over
that
and
it
looks
fine
for
the
most
part.
Josh
really
wanted
somebody
who
knows
prometheus
well
to
review
the
prometheus
mapping
because
he
wasn't.
He
thinks
that
what
he's
got
is
correct,
but
he
needed
somebody
who
was
an
expert
and
that's
certainly
not
me.
A
Yeah
the
metrics
meeting
went
off
into
the
deep
end
and
there
was
not
even
the
tiniest
bit
of
chance
to
ask
about
that.
It
was
almost
entirely
eaten
of
histograms.
Yes,.
A
Yeah
well,
the
popcorn
yeah.
I
needed
some
popcorn
at
lunchtime
the
fascinating
thing.
I've
I
thought,
and
we
I
don't
think
splunk
has
much
of
a
cat
in
this
fight
at
the
moment.
But
apparently
there
was
some
meaning
of
the
minds
of
vendors
around
metrics
around
the
open
metrics
and
whether
open
metrics
should
be
a
standard
and
apparently
every
vendor
every
apm
vendor
that
accepts
metrics
said
no.
We
don't
want
that.
We
don't
want
to
use
that.
A
We
want
to
use
something
simple
with
a
binary,
something
something
histogram
which
was
interesting
because
all
of
the
open,
telemetry
metrics
sig
members
really
think
we
should
just
go
with
open
histogram
and
like
the
fact
that
it's
kind
of
standard
and
one
size
fits
all.
B
A
It's
certainly
it's
basically
what
circonus
would
like
everyone
to
kind
of
sarcomas
has
come
in
and
said
the
the
hardest
part
with
getting
people
to
use
histograms
is
that
nobody
knows
how
to
configure
them
and
everyone
configures
them
differently,
and
two
people
in
the
same
company
will
configure
them
differently
and
then
there's
no
way
to
merge
them
and
it's
a
big
mess.
So
basically
their
philosophy
is
we've
developed
this
new
thing.
A
A
It
has
a
very
large
number
of
buckets,
but
you
can
store
it
very
efficiently,
blah
blah
blah.
I
don't
know
not
my.
I
don't
know
this
stuff.
I
just
kind
of
try
to
absorb
what
I
can
learn
anyway,
but
they
they
think
this
is
kind
of
the
one
size
fits
all.
It
requires
zero
configuration,
you
just
turn
it
on
and
it
works
and
they
they
think
and
at
least
especially
having
been
a
vendor
of
histograms.
Essentially,
they
think
this
has
worked
out
great
because
the
customers
don't
have
to
configure
anything.
A
B
I
think
the
other
people
like
the
well
my
experience
was
with
the
header
histogram
hcr
histogram,
so
it
was
actually
very
intere
in
intrigued
me
when
gil
tamiya,
your
team
joined
the
conversation
and
that's
just
a
pure
exponential
yeah,
and
I
think
that's
what
the
I
don't
know.
The
d
there's
the
dd
sketch
and
then
dinah
dyna
historic
histogram.
A
And
apparently
there's
a
new
udd
sketch,
which
is
some
academics,
have
gone
and
made
dd
sketch
something
more
compact,
or
I
don't
know
something
like
that
anyway,
interesting
yeah.
This
seems
like
it's
the
interestingly
the
most
controversial
aside
from
the
open,
metrics
stuff,
which
I
think
was
also
pretty
controversial.
This
has
definitely
been
the
most
controversial,
like
vendor
colliding
with
technical
committee
thing
that
I've
seen
in
open
telemetry
so
far.
A
That's
what
I
was
saying
along
with,
along
with
the
open
histogram
thing-
I
mean
open.
Metrics
thing
excuse
me,
but
this
one
is
even
more
contentious.
I
think
everyone
pretty
much
got
on
board
that
we
need
to
support
prometheus
and
so
like
as
a
first
class
citizen,
because
it
is
the
most
common
open
state,
but
not
really
a
standard
but
open
metrics
platform.
A
D
E
B
So
it's
amazing
to
me
how
how
many
people
are
know
so
much
about
instagram
yeah.
Well,
like
I
I
mean
I
kind
of
get
it
because,
like
if
you
didn't
have
like
when
I
needed
a
histogram,
you
know
I
went
and
I
was
like
okay
there's
this
hdr
histogram,
it's
written
by
somebody
who
I
truly
trust,
and
I
was
like
okay,
throw
it
in.
You
know,
move.
B
You
know
don't
really
worry
about
all
the
the
micro
details
but,
like
I
can
see
how
you
know,
dina
trace
has
been
in
business
for
so
long
that
you
know
clearly
they
had
had
to
do
all
of
that
from
scratch,
work
from
scratch
across
lots
of
different
languages,
and
so.
A
A
I
don't
want
to
have
to
re
like
change
the
on
disk
format
for
all
of
new
relics,
metrics
if
we've
already
got
it
in
place,
so
I'm
pretty
sure
that
he's
there's
a
little
bit
of
that
kind
of
coloring
his
his
gigantic
treatise
that
he
wrote.
A
B
B
A
D
A
D
A
D
D
A
Do
it
I'm
not
gonna?
I
have
like
kind
of
like
the
mechanic
for
my
car
right
like
I
don't
want
to
know
how
it
works.
I
don't
want
to
do
it.
I
roughly
I
have
a
rough
idea
of
how
a
car
works.
I
mean
I
have
a
pretty
good
idea
of
car
works,
but
I
want
the
mechanic
to
deal
with
fixing
it
same
thing
with
histograms.
I'm
like
I'm
glad.
Somebody
cares
enough.
A
So
anyway,
it
was
a
little
bit
of
a
side
note,
and
then
we
talked
a
little
bit
about
if
there's
any
ways
that
we
could
split
up,
prs,
yeah
pr
3333
in
a
way
that
make
it
more
consumable.
A
I
was
suggesting
that
the
instrument
renames
might
be
the
next
good
part
in
the
api
and
josh,
was
very
resistant
because
he
said
it
was
going
to
break
everything.
I
don't
really
exactly
understand
why,
although
it
sounded
like
a
lot
of
it
had
to
do
with
attributes
in
the
sdk,
maybe,
but
also
there's
a
whole
bunch
like
all
of
the
sdk,
is
really
interwoven
and
tangled
up
with
each
other
and
the
model,
and
so
he's
going
to
try
to
kind
of
tease
that
apart
and
make
it
a
little
bit
more
componenty.
I
guess.
A
A
E
D
E
A
B
We
should
have
done
what
you
were.
You
suggested
like
six
months
more
than
that.
E
A
E
B
Yeah,
no,
no
in
the
thing
where
we
suffer
is
we
have
the
open,
telemetry
metrics
bridge
on
the
in
the
instrumentation.
B
D
D
A
A
Yeah
so
anyway,
josh
is
going
to
continue
crunching
on
it.
Now
that
jakob
or
kuba,
as
he
goes
by
now
that
kuba
is
going
to
be,
is
an
approver
I've.
I'm
trying
to
get
him
sick
on
some
metrics,
pr's
and
yeah
he'll
start
digging
in.
D
D
A
D
I
painted
the
aws
distro
and
there
were
no
again
the
api
didn't
change,
so
the
pr
build
was
fine
and
then
our
integration
test
on
main
field,
because
the
label
stopped
getting
filled
in
on
the
center
cloudwatch.
This
is
integration
test,
so
whatever
was
in
cloudwatch,
and
so
I'm
guessing
it's
because
of
an
incompatible
skin
collector.
C
D
C
D
D
B
Java
17
is
going
to
be
a
big
deal
in
a
couple
of
months,
we're
going
to
get
a
lot
of
questions
about
it
and
so
seems
like
at
minimum,
and
I'm
not
quite
sure
if
nikita
volunteered
to
do
this,
he
said
that
somebody
splunk
had
tried
to
do
this
upgrade
previously
and
didn't
get
very
far.
D
D
A
D
B
Look
at
the
up:
they
had
to
upgrade
spock
two
zero
migration
guide.
The
migration
guide
did
not
look
that
horrible,
but
then
you.
A
D
A
A
B
D
I'm
not
gonna
push
super
hard
on
it,
at
least
just
to
compare
and
see
what
people
think
it's
obviously
more
verbose,
but
if
it
gives
us
refactoring
and
the
test
run
a
bit
faster,
I
personally
might
still
prefer
it.
But
of
course
we
need
to
see
what
other
people
think
it
wasn't
hard
to
do.
The
migration.
D
B
Yeah
yeah
I'd
love
to
see
it.
I
I
think
we
have
at
least
three
out
of
four
trainers
who
are
open
to,
but
we
have
to.
We
have
one
that's
pushing
hard
one.
That's
holding
holding
strong
and
two
that
are
open
open
to.
B
B
Yeah
yeah
I
mean
I
would
love
to
get
to
our
tests
to
java
17.
D
D
D
E
A
D
D
B
D
B
D
A
B
A
D
D
That
will
be
cool
to
do
eventually
and
then
all
of
our
repos
will
be
doing
the
exact
same
thing
like
finally,
now
instrumentation
requests
finding
the
same
as
sdk.
I
think
almost
all
the
like
kotlin
build
files
and
everything
it
seems
like
we've
caught
up.
So
now
is
a
good
time
to
possibly
share
the
code.
B
C
I
mean
we
will,
we
will
accept
prs
in
the
splunk
repo
for
the
same
yeah.
You
know
get
bored.
B
Yeah,
I
was
just
updating
hours.
I
just
pulled
in
all
the
latest
stuff
from
that
on
her
eye.
Well,
we
have
that.
I
have
that
fork
that
horrible
fork,
and
so
I'm
updating,
but
it
forces
me
to
actually
like
understand
the
stuff
that
I
rubber
stamped
approved
earlier.
B
Or
you
still
have
your
own
modules
that
aren't.
No,
I
just
today
was
updating
them
to
kotlin.
D
B
C
A
E
C
B
C
Yes,
derp
a
good
point,
yeah
anyway,
just
an
observation,
so
I
think
we'll
hear
that
come
up
at
some
point
again.
B
We
still
need,
I
think
first
prior,
have
to
be
sampling.
Example.
E
B
But
now
we're
all
gonna
just
point
them
at
x-ray
and
say:
hey,
go,
go,
have
a
x-ray
account
and
just
use
it
for
dynamic
sampling
and
send
keep
sending
all
your
back-end
data
to
splunk
or.
B
A
Jason,
hey.
Actually,
I
should
tell
you
trask
that
if
you're
interested
at
12
15
tomorrow
we're
meeting
at
a
taqueria
on
northeast.
A
B
Anyway,
my
calendar,
you
said
12
15.,.
A
A
It
is
called
the
sahu
I
I'll
send
it
to
you
I'll
paste
it
in.
I
can't
you
probably
I
don't
know
how
to
pronounce
it,
so
you
probably
won't
be
able
to
spell
it.
A
E
A
Yeah
yeah
sure
anyway,
yeah
12
15
tomorrow
we're
going
to
try
to
meet
for
some
try
out
this
talk
area
that
nobody
ever
no
none
of
us
have
ever
been
to
so
yeah.
I
just
wanted
to
call
out
that
my
past
three
four
months
of
work
we
finally
had
release
0.1
today
got
it
publishing.
A
We
definitely
are
interested
in
figuring
out
how
to
upstream
our
instrumentation
to
open
telemetry,
but
there's
no
like
there's,
there's
no
specification
or
anything
for
this
at
the
moment.
So
I
mean
you
can
dump
it
in
country
yeah
right
now.
Things
are
very
tight.
Well,
there
are,
I
guess,
there's
some
instrumentation
parts
that
could
be
used
by
everybody.
A
Wait
that
doesn't
say.
That's!
Oh!
Oh,
that's
funny,
you're
looking
at
the
tree
for
you're,
looking
at
the
readme
when
that
tag
got
done,
but
then
I
updated
the
reading
afterwards.
So
that's
not
actually
not
the
latest
readme
so
anyway,
but
yeah-
and
this
is
this-
it
basically
just
generates
spans
and
sends
them
to
zip
via
zipking
at
the
moment,
because
that's
what
our
rom
back
end
accepts
directly
from
the
phone
yep
directly.
A
D
A
Prototyping
we
have
a
so
we're
we're
sending
the
data
to
the
browser.
We
have
a
browser,
run
product
right
now
and
they're
currently
working
on
working
on
the
and
we
send
the
data
to
the
same
place.
The
ui
is
not
ready
for
mobile,
like
it's
still
very
browser-centric,
but
they're
working
on
that
right
now.
In
the
back
end,
they
still
want
to
generate
metrics
that
are
mobile-specific
and
they
don't
have
that
done
yet.
But
basically
this
gener
just
generates
zipkin
spams.
At
the
moment,
yeah
hi.
D
A
Yeah,
why
I
think
that
when
they
started
working
on
it,
the
otlp
in
the
browser
was
not
ready
for
prime
time
and.
E
A
Can
the
zip
the
browser-based
zipkin
was,
was
battle
tested
and
everyone
knew
was
going
to
work
really
well
so.
D
A
Anyway,
so
yeah
anyone,
I
guess,
could
use
this
plug
it
into
their
app
and
plug
in
the
different
back
end.
I
mean
I've
actually
just
sent
it
to
a
local
zipkin
installation
to
you
know
when
I
don't
want
to
have
to
go
and
read
stuff
out
of
the
ui
and
send
it
into
our
vpn
etc.
A
I
mean
the
code
will
take
that
off
token
and
append
it
pre-pen
append
an
attribute.
So
as
long
as
your
back-end
doesn't
have
any
problem
with
that,
it
will
work.
Just
fine
it'll
append
a
query
parameter
for
the
auth
when
it
goes,
it
does
like
question
mark
off
equals
and
then,
whatever
the
token
is,
but
as
long
as
your
back
end
is
is,
will
accept
it.
Then
it
works.
Just
fine.
A
Yep
yeah
this
just
uses
the
stock
sdk.
It
doesn't
do
anything
fancy.
It
would
mean
there's
a
special
appender
like
a
special
spam
processor
that
does
a
bunch
of
kind
of
probably
splunk
rum,
specific
attributes.
It
also
does
things
like,
unfortunately,
zip.
Can
the
zip
code
exporter
doesn't
take
anything
off
the
resource
except
the
service
name,
so
we
have
to
put
a
bunch
of
resource
stuff
just
directly
onto
the
spans
at
the
moment,
which
is
one
of
the
big
limitations
of
zika.
A
Unfortunately,
and
zipkin
also
does
not
have
a
rich
event
or
log
or
whatever
it's
called
model.
It
just
has
a
name
thing.
It
doesn't
have
like.
You
can't
put
the
stack
traces
there,
so
we
have
to
do
the
stack
traces
differently
and
so
there's
a
bunch
of
little
workarounds
to
make
the
data
flow,
be
it
through
zipkin
at
the
moment,
cool
yeah
anyway.
B
A
A
D
A
A
I
have
there's
documentation
to
be
written
still,
but
yeah
it
works
does
crash
reporting.
A
Everyone
was
very
impressed
that
you
could
just
add
a
unhandled
exception
or
whatever
it's
called
the
thread
on
on
caught
exception
handling,
and
I
do
I
do
a
force
flush
on
the
I
create
a
like
a
span
that
just
has
the
stack
trace
in
it
and
force
flush,
the
sdk
and
it
as
far
as
I.
A
It
always
makes
it
out
of
the
out
of
the
app
before
android
brings
it
down
like
I've.
Never
I've
never
seen
it
not
work,
which
is
everyone
I
mean
like
on
ios.
There's,
basically
no
way
to
do
that
at
all,
but
android
it
worked
perfectly
so
yeah.
We
actually
had
an
internal
internal
demo
where
one
of
our
internal,
like,
I
think
the
people
who
run
the
signalfx
android
app
were
trying
it
out
or
it
wasn't
a
demo.
A
B
Have
you
tested,
do
you
have
a
old
phone
on
you
to
test
with
no
I've
been
just
using
the
simulator
yeah.
A
I
have
my
I
have
my
pixel
3.
I
mean
it's
a
couple
years
old,
but
it's
still
running
the
latest
and
latest
release
so
yeah
I
was
I
tested
on
some
in
the
simulator
you
can
down.
You
can
run
this
little
android
api
level,
22
flip
phone.
A
I
think
it's
like
this
teeny
teeny,
tiny
little
thing
and
the
little
sample
app.
That's
in
that
repo
like
doesn't
even
render
correctly,
because
the
screen
is
too
tiny,
but
it
still
works,
so
that
was
good.
D
A
A
There
was
one
other
thing
that
I
put
in,
I
thought
I
put
in
some
other
android
specific
hack
a
couple
months
ago
that
fell
out
of
this.
I
don't
remember
what
it
was,
though.
A
B
B
B
With
the
simulator
pretty
good.
A
A
Eventually
your
app
may
work,
but
the
actual
os
stops
doing
network
connectivity
for
some
reason,
like
just
the
network
dies
and
then
nothing
works,
and
you
can't
the
ui
become
the
ui
not
of
the
your
app,
but
the
ui
of
the
actual
operating
system
becomes
unresponsive
and
it's
a
big
mess,
and
this
week
the
simulator
crashes
every
time
I
start
it
up
now.
I
don't
know
whether
what
happened,
but
I
can't
run
it
at
all
at
the
moment,
so
I
may
have
to
reinstall
the
sdk
from
scratch.
A
I'm
not
really
sure,
but
I
will
say
this.
This
is
the
big
thing
in
this
past
week,
I've
spent
just
trying
to
get
publishing
demand
central
to
work
all
of
the
convenience
gradle
plugins
do
not
work
with
the
android
library
plugin,
like
it.
The
android
library
plugin
has
a
bunch
of
opinions
that
break
all
like
the
nexus.
A
Publishing
plugin
doesn't
work
the
what's
that
the
one
from
netflix
the
nebula
plug-in,
doesn't
work
like
basically
and
there's
even
a
standard
like
not
standard,
but
there's
a
gradle
plug-in
from
this
guy
called
vanit
vanik
tech
that
does
all
the
publishing
and
stuff,
but
it
doesn't
work
with
standard
signing
like
you
can't
get.
I
don't
know.
Anyway,
it's
been
a
huge
mess.
A
D
A
It's
a
dot,
a
aar
and
not
a
gender
right,
and
I
don't
think
okay
hasn't
publishes
an
aar.
No.
But
we
wanted
to
publish
an
aar
because
we
actually
want
to
put
the
permissions
that
you're
going
to
need
into
the
library
manifest
which,
like
you're,
going
to
need
to
get
have
network
permission
in
order
to
use.
B
A
Yeah
there's
some
stuff
like
that
that
you
want
to
put
into
your
android
archive
manifest
in
order
so
that
the
application
will
get
it
to
know
that
it
also
has
to
have
it
yeah
and
there's
also
getting
like
a
version
number
into
an
android
archive
was
super
painful,
like
it's
crazy,
android
resources
rather
than
standard
anything
anyway.
No,
it
was
it's
been
an
adventure.
A
A
E
A
E
E
E
B
Dude,
because
open
jdk
adopt
open
jdk
doesn't
have
images
for
16.,
which
I
think
is
really
weird
but
open.
Jdk
is
this:
is
the
official
like
docker,
hub
or
docker
maintained.
E
B
And
they
have
seven,
they
also
have
17
early
access
releases.
So
I
have
a
feeling
this
might
be
the
better
one
to
go
with
in
the
future.
B
The
only
downside
is
they
don't
have
they
don't
have
open
j9
images
which
adopt
has
so
that's
why
I
had
to
split
out
and
have
one
set
with
hotspot
and
one
set
with
open
j9.
B
Yeah
this
well
so
the
way
that
the
windows,
so
the
tomcat
images,
are
from
real
I've
learned
all
about
this,
so
the
linux
tomcat
images
are
from
real
docker
hub,
tomcat
images,
the
windows
ones,
tomcat
doesn't
publish
windows,
one
we
build
it
based.
We
merge,
we
download
image,
we
start
from
a
windows
base
image
and
then
we
pull
in
adopt
or
open.
B
Okay,
I
don't
think
there's
really.
I
have
some
abandoned
prs.
I
should
revisit.
Oh,
this
is
the
one
I'm
gonna
review.
Oh
this.
B
Yeah
mateo
shapr
matesh
reviewed
it.
He
just
asked
for
a
second
because
he
said
it
was
pretty
complex
for
just
adding
yeah.
So
if
you're
interested
in
looking
at
it,
I'm
happy
to
or
go
for
it.
D
E
D
B
D
E
C
B
The
next
day
or
two
I
will
review
it
for
sure
yeah,
that's
about
it.
B
Can
we
get
to
the
bottom
of
extension?
Oh
yeah
nikita
brought
up
the
api
stability,
and
so
we
were
kind
of
talking
about
what
we
can
do
to.
Oh.
I
have
an
action
item
here.
B
Matej
mentioned
that
the
extension
api
stuff
he's
pretty
much
done
with
the
changes
he
wanted
to
make.
So
he
said
that
was
in
pretty
good
shape.
The
other
two,
of
course,
are
not
and
so
didn't.
I
write
an
action
item
for
myself
here.
B
No
okay:
I
will
to
open
a
tracking
issue
where
we
can
like
track
packages
or
classes
that.
B
B
E
B
B
And
then
we
pointed
john
to
the
new
http
attribute
extractors
for
content
length.
Oh
yes,
I
saw
you
were
having
some
discussions
about
instrumentation
api
is
continuing.
E
D
E
D
B
B
Now,
did
you
talk
about
that
in
the
in
the
spec
meeting
this
week?
Yep
cool
so
are
people
joining
you're
getting
some
people
joining
the
this
yeah?
It
looks
like
ted.
D
D
D
D
No,
I
I
forgot
to
share
that.
I
think
my
next
now
that
I
finally
finished
this
clockman
stuff,
which
no
one
really
needed
to
do,
but
I
felt
like
doing
this
so
going
to
get
back
to
library
instrumentations.
B
E
D
D
I
don't
know
I
actually
haven't
looked
at
it.
I
don't
know
what
change
at
http.