►
From YouTube: 2022-05-06 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).
D
B
A
E
B
A
E
F
I
don't
have
any
of
my
calendar
invites
or
anything
set
up
like
that's.
Why
I'm
late
on
the
new
on
my
new
work
laptop
so.
E
F
I
got
the
m1
the
new
m1
13
inch
macbook
pro
as
far
as
I
can
tell
it
has
infinite
battery
life
like
it's,
it's
been
just
plugged
into
my
usb
switch
and
it
hasn't
needed
any
more
power
than
that
like
I
don't
even
have
it
like
on
the
real
power,
brick
and
you
know
been
running
docker
and
idea
and
just
never
never
goes
below
100.
What's
your
wattage
on
that
that
charger,
it's
not
even
a
charger.
It's
just
a
cheapo
chinese
usb
switch
like
no
idea
like
powered.
F
F
A
I
actually
should
give
it
a
try.
I
did
I
I
gave
it
a
try
and
there
were
some.
There
were
some
hiccups,
but
they're,
not
they're,
not
insurmountable
hiccups.
A
I
had
an
m1
laptop
as
like
a
trial
because
they're
going
to
roll
them
out
across
our
company.
A
But
it
was,
it
was
really
obnoxious
because
it
was
just.
It
was
an
m1
air
instead
of
a
macbook
pro,
and
so
it
only
had
eight
gigs
of
ram
and
it
everything
was
memory
bound
trying
to
like
run
stuff
in
intellij.
It's
just
like
you
cannot
get
away
with
eight
gigs
of
ram
running
the
instrumentation
repository.
F
Yeah,
I've
only
got
16,
which
is
probably
barely
enough
to
what's
up
it
said
yeah
32
gigs
is
the
minimum
for
me
at
virta.
We
pretty
much
run
everything
in
remote
kubernetes,
and
so
we-
and
we
have
a
way
this
telepresence
thing-
to
hook
up
local
processes
to
remote
remote
kubernetes
and
make
it
act
like
it's
part
of
the
cluster.
F
B
Yeah,
no,
are
you
honorary
you're
talking
about
still
writing
stuff
locally,
just.
E
E
B
I
was
asking
what
people,
if
anybody
was
already
doing
this,
it's
a
popular
apm
feature
where
you
automatically
inject
your
javascript
monitoring
inside
of
your
server
framework.
B
It's
also
notoriously
problematic
inside
your
server
framework
like
for
node.js
or
something
yeah,
or
in
your
servlet
stream,
like
you're,
like
in
java
in
your
java
server,
yeah
inspect
the
you
know,
the
the
http
response
and
you
watch
for
the
head
tag
and
you
inject
the
javascript.
There.
E
B
Yeah
so
we're
we're
going
to
have
a
and
it's
it
was
proposed
as
an
intern
project
here.
So
I
just
wanted
to
make
sure
that
a
nobody
else
was
already
working
on
it
and
be
that
there
was
a
curious.
What
if
other
people
would
have
been
talking
about
this
already
in
it
jason
plum
mentioned
that
it
had
just
come
up.
Somebody
had
just
asked
about
it
this
morning
at
splunk,
but
there
were
no
current
plans
and
they
have
the
same
I'll.
Just.
E
B
Do
you
happen
to
know
jack
or
john
or
or
john
if
new
relic
does
injects
in
neti,
because
I
know
I
know
they
do,
or
at
least
they
used
to
in
servlets.
B
Whether
it
injects
the
the
browser,
rum
snippet,.
A
C
I
don't
remember
yeah,
I
don't
remember
I
mean
I
I
worked
on
initially.
I
was
on
the
java
support
java
agent
support
team,
so
you
know
trying
to
remember.
I
know
that
there
is
instrumentation
for
nettie
in
the
agent
but
as
far
as
the
the
rum
injection
part,
I'm
not
I'd
have
to
I'd
have
to
go.
Look.
I
don't
remember.
A
B
Not
in
open
telemetry,
but
so
what
we
would
probably
do
is
make
it
where
you
could
configure
the
snippet.
You
want
to
inject
so
vendors.
Could
you
know
splunk
could
in
fact
inject
their
rum
snippet.
We
could
inject
our
snippet.
That
kind
of
thing
for.
E
B
Yeah,
so
what
do
we
have?
We
have
I
mean
servlets
nettie.
E
E
Yeah,
so
I
think
it
would
be
a
some
more
instrumentation,
especially
if
it's
not
midi,
but
more
so
I
think
like
it
would
be
in
the
instrumentation
api
having
some
response
setter
type
of
abstraction,
something
like
that
and
then
how
that's
applied
into
like.
We
would
figure
out
how
to
apply
it.
But
we
need
that
abstraction.
I
guess
similar
to
the
request,
getters
or
whatever,
that
we
have
right
now.
E
A
It's
different
like
because
you
want
to
take
the
response
like
body
and
you
know
kind
of
parse
it
for
its
html
and
walk
the
dom
and
inject
the
snippet.
At
a
particular
point.
E
Yeah,
so
whether
it
makes
sense
to
use
the
same
abstraction
for
both-
I
don't
know,
but
they're
both
modifying
the
response,
so
you
can
imagine
them
still
being
the
same
customization
if
it
works.
But
it's
true
that
the
response
hijacking
is
much
more
complicated
than
this
injecting
headers.
So
we
might
not
want
to
play
them,
but
at
least
they
are
sort
of
in
the
same
boat.
E
B
So
we
talked
a
decent
amount
about
the
tests
and
how
to
best
test,
metrics
and
because
delta
has
its
problem.
Delta.
Has
those
problems
and
cumulative
has
its
problems?
Delta
has
its
problems
because
we
get
them
in
chunks,
so
we
sort
of
have
to
maybe
aggregate
them
in
the
test
or
have
more
flexible
assertions,
since
they
could
be
split.
B
B
A
Maybe
even
the
scope
is
wider
than
that,
so
the
instruments
that
are
collected
from
one
test
to
another,
the
basically
the
state-
is
polluted,
and
so,
if
you
have
cumulative
metrics-
and
you
know,
you're
the
tenth
test
to
run
in
a
class,
you
have
the
cumulative
metrics
you're
interested
in
plus
all
the
ones
that
have
run
for
other
tests
in
the
same
class.
A
So
that's
not
good
and
then
the
other
issue
is
that
the
way
that
we
get
metrics
right
now
is
we
have
the
periodic
metric
reader
set
up
with
a
really
small
collection
interval.
I
think
it's
like
on
the
order
of
like
10,
milliseconds
or
something,
and
so
we
just
rely
on
that,
like
cumulative
metrics
being
collected
constantly
every
10
milliseconds
and
then
you
know
on
the
test
side,
you
know
we're
we're
gonna.
We're
saying.
Does
one
of
these
many
many
many
like
cumulative
metrics
that
are
largely
duplicative?
A
So
you
don't
have
the
pollution
from
other
tests
and
also
so
that
the
you
have
cumulative
temporality
instead
of
delta
cumulative
is
nice
because
you
know,
as
trask
mentioned,
you
don't
have
to
worry
about
having
measurements,
get
split
across
collections
and
then
having
to
aggregate
them
and
assert
against
them
in
your
test
itself.
So
those
are
kind
of
the
two
conditions
that
would
represent
like
the
a
good
outcome.
A
B
B
E
A
That's
what
we
were
thinking
of
and
so
lori
was
going
to
prototype,
that
with
using
reflection
to
accomplish
the
same
thing
and
just
see
if
it's
like
viable,
there's
one
or
two
spots.
I
think
that
you'd
have
to
worry
about
in
the
agent
for
like
clearing
the
the
storages
and
yeah.
I
I
think
it's
doable.
E
E
E
B
Cool
one
question:
one
thing
topic
we
wanted
to
bring
up
was
the
sdk
auto
configure
module
if
there
were
any
plans
any
I
mean
I
kind
of
know
in
that
check
with
jack
that
we
haven't
really
discussed
this
recently,
but
wanted
to
get
your
thoughts
on
our
again
john
on
the
path
to
the
auto
configure
module
being
stable,
and
it's
not
blocking
us
at
this
point,
but
I'd
say
maybe
we're
you
know
in
the
order
of
three
to
six
months
away
from
being
able
to
make
the
extent
java
agent
extension
stable,
and
there
was
just
one
in
the
sdk
auto
configure
module
we're
really
only
using
one
class
out
of
there.
B
B
And
we
pass
this
in
primarily,
it
seems
like
we
passed
this
in
to
get
the
resource,
so
it's
not
entirely
clear
that
we
need
the
needed.
F
F
A
One
thing
I
brought
up
earlier
was:
you
know
how
we
handle
unstable
dependencies
in
the
open
in
the
sdk
module
that
we
have.
You
know
like
we
have
a
dependency
on
the
metrics
sdk,
even
that
isn't
stable,
but
our
sdk
is
stable
because
we
have
an
implementation
dependency
on
the
metrics
sdk
and
so
like
the
assertion
we're
making
by
doing
that
is
that
the
parts
of
the
metrics
sdk
that
we
actually
expose
in
our
api
are
stable
enough,
that
they're
not
going
to
change
so
the
sdk
meter
provider,
like
we
can't
change
that.
A
You
know,
even
though
that
artifact's
unstable
so
like
in
a
similar
way,
the
agent
extension
could
become
stable
with
an
unstable
dependency
in
the
you
know,
the
auto
configure
module.
You
know
if
we
have
confidence
that
this
auto
configured
sdk
class
does
not
change.
E
B
A
B
A
F
E
F
A
And
logs
are
now
exposed
as
well
and
logs
aren't
particularly
close
to
being
stabilized
at
the
sdk
level.
But
that's
not
a
problem
because,
for
the
same
reason
why
it's
not
a
problem
that
our
stable
sdk
artifact
has
an
implemented
implementation
dependency
on
the
log
sdk.
A
A
A
I
think
it's
good
to
think
about,
and
you
know,
if
there's
nothing
blocking
us,
I
think
we
should
pursue
it
because
you
know
it
is
something
that
we
want
folks
to
use
and
it
it
really
improves
the
usability
of
the
sdk,
and
I
mean
we
could
linger
in
the
alpha
status
for
like
a
long
time.
But
you
know
I'm
not
sure
how
much
we
buy
from
that
we're
just
kind
of
hurting
users
by
not
stabilizing
it.
B
B
Some
hairy
bite,
buddy
bite
could
hackery
it's
actually
not
so
horrible,
but
it
it's
kind
of
con
like
it's
not
really
clear
the
benefit
I
mean.
We
know
that
looking
up
resources
is
a
bottleneck.
We
don't
know
how
much
of
a
bottleneck
looking
up.
Annotation
type
resources
is,
but
I
think
our
we,
the
consensus,
was
just
to
go
ahead
and
go
forward
with
this.
B
B
You
know
something
that
the
operator,
I
think
he
ran
into
this
and
the
operator
is
already
kind
of
doing
something.
So
who
is
just
curious?
What
kind
of
you
know
support
from
sdk
authors?
There
was
for
this
idea
like
in
terms
of
was
this
good
timing.
I
think
primarily
that
was
his
question.
Not
I
think
everybody
knows
that
we
want
to
do
this,
but
whether
this
was
a
good
timing
or
whether
it
would
be
a
distraction
from
like
metrics
and
logs
work.
B
Yeah-
and
I
showed
him
our
as
far
as
motivation
for
this
I'll
check,
our
ever
growing-
we're
up
to
22
thumbs
up
on
this
one.
Our
by
far,
our
most
popular
issue
is
essentially
blocked
on
having
a
configuration
format
that
could
that
we
could
more
rich
than
that.
We
could
model
something
like
this.
E
A
It
is
true,
so,
like
you
know,
if
you
have
a
deployment
that
depends
on
a
config
map
where
you
have
your
yaml
config
and
you
update
the
config
map
like,
if
that
doesn't
that
doesn't
trigger
a
redeployment
of
all
the
pods
in
the
your
deployment
and
so
you're,
stuck
with
your
old
config,
you
have
to
like
manually,
like
kind
of
delete,
everything
and
redeploy.
E
B
A
I
think
would
be
traditional.
I
think
that
would
give
me
enough
time.
I
I
got
pulled
into
some
stuff
last
week,
and
so
I've
been
trying
to
be
heads
down
on
a
variety
of
metrics
sdk
issues.
I
think
I
can
get
them
done
by
next
week.
A
A
So
there
are
a
couple
things
that
are
sticking
out
so
right
now
our
plan
is
to
well
so
there's
your
like
any
any
finishing
test
touches
on
the
the
testing
module,
because
the
plan
is
to
stabilize
that.
So
I'm
not
sure
how
far
along
you
are
on
transitioning
that
stuff
over
and
then
there's
a
rather
big
bug
that
I
found
and
which
pr
is
it
fix.
Metric
four
four
three
six
yeah
fix
fix
metric
sdk
when
multiple
readers
are
present.
A
Okay,
so
I
got
to
remember
that
old
minimum
collection
interval-
I
I
got
to
the
bottom
of
that
and
that
that
was
like
more
insidious
than
I
originally
thought.
I
thought
it
was
like
relatively
benign
in
terms
of
what
it
was
doing,
but
the
kind
of
design
of
the
sdk
and
the
storage,
I
think
is.
A
I
think
it
has
like
a
pretty
major
conceptual
flaw
in
that
it
doesn't
work
when
when
multiple
readers
are
present
and
this
pr
fixes
that
and
the
root
of
the
issue
is
around
the
the
fact
that
our
pattern
is
to
have
one
storage
per
view
per
instrument.
A
So
if
you
have
an
instrument
like,
let's
call
it
foo
and
you
have
one
view
for
it-
you
have
one
storage
despite
having
potentially
two
or
more
readers,
and
so
like
one
storage,
is
somehow
responsible
for
managing
the
state
of
of
of
many
readers
and
it
actually
doesn't
work.
So
if
you
try
to
have
multiple
readers,
one
of
them
will
read
the
values
for
for
your
for
your
instrument,
foo,
and
it
will
reset
the
state
such
that,
like
your
additional
readers,
actually
don't
see
those
measurements.
A
So
it's
like
it's
it's
wrong
and
there's
the
fbr
fixes
that
I.
A
Let's
see
it,
it
might
not
work
for
cumulative
two
because
it
might
reset
the
state
in
the
way,
so
the
way
that
the
way
that
the
storages
work
is
everything
is
maintained
in
this,
like
kind
of
intermediate
storage
layer
as
as
deltas,
and
then
those
deltas
are
taken
and
merged
with,
like
previous
cumulative
results
like
to
compute
the
current
cumulative,
and
so
the
problem
is,
is
that
if
you,
if,
if
one
reader
reads
those
deltas
and
then
resets
them
and
another
reader,
comes
along,
that's
cumulative
and
tries
to
read
those
deltas
and
tries
to
merge
them
with
its
its
its
current
like
cumulative
state?
A
It
might
not
have
any
values
to
merge.
A
F
Did
you
did
you
chat
with
josh
about
this?
F
A
Yeah-
and
so
I
I'll
I'll
ping
him
on
that,
I
know
he's
tagged
on
all
the
pr's
because
he's
an
improver
but
I'll
I'll
alert
him,
and
so
like
this
change
that
I
make
it
paves
the
way
to
do
several
things.
Actually,
so
it
fixes
this
kind
of
conceptual
bug
that
I
I
claim
is
present
and
then
trask
had
opened
this
issue,
maybe
a
week
or
two
ago
about
the
delta
intervals
being
wrong.
A
Time
of
the
collection
interval
is
the
start
time
of
the
application,
not
the
start
time
of,
like
the
last
time
that
reader
read
those
metrics,
and
so
this
issue
kind
of
in
a
in
a
way,
makes
it
a
lot
simpler
to
to
solve
that
as
well-
and
I
I
have
this
is
in
a
draft
pr
right
now,
because
it
builds
often
the
the
previous
pr.
A
So
you
know
I
don't
want
to
draw
too
much
attention
to
it
yet,
but
I
think
that's
like
an
important
thing
to
get
in
as
well
and
there's
one
more
thing.
A
Don't
know
if
I
should
talk
about
it
right
now,
because
I
don't
have
a
pr
open
for
it.
I
have
it
locally,
but
so
the
the
metrics
sdk
says
that
metric
readers
have
the
ability
to
influence
the
default
aggregation.
A
A
Histogram
aggregation,
the
the
sdk
says
that
specification
says
that
you
should
be
able
to
when
you,
when
you
implement
a
reader
or
when
you
provide
a
reader,
you
should
be
able
to
change
the
default
aggregation,
and
so
the
idea
would
be
that
you
know
I
could
say
that
all
histogram
instruments
should
be
exponential
histograms
instead
of
explicit
bucket
histograms,
and
you
know
this
is
actually
it
sort
of
overlaps
with
views,
but
it's
kind
of
distinct
and
important
for
kind
of
a
nuanced
way
reason,
but
as
effectively
what
this
change
does
is
it
or
what
this
part
of
the
sdk
implies
is
that
each
reader
has
its
own
set
of
views
so,
rather
than
having
one
global
set
of
views
that
is
shared
across
all
your
instruments,
each
reader
would
have
its
own
distinct
set
of
views
that
it's
configured
for
this.
B
A
F
So
what
you
know
one,
since
I
think
the
changes
that
you're
talking
about
here
aside
from
the
you
know
an
enhancement
to
the
reader
interface,
so
you
can
specify
temporality
or
not
temporality,
but
default
aggregations
or
whatever
we
could
release
without
these
fixes-
and
just
say
here
are
some
known
issues
like
you
know.
We
only
support
a
single
reader,
for
example,
because,
like
your
that
that
pr
that
you
put
in
there
doesn't
change
any
of
the
apis
of
the
sdk
right,
it's
an
internal
thing
and
everything
does
work.
F
Fine
at
the
moment
with
single
reader
right,
that's
correct,
so
we
could
do
a
release
and
say
it's
stable.
Then
that
means
api
stable.
It
doesn't
mean
there
are
no
problems,
I
mean,
there's
nothing
we're
going
to
change,
which
means
the
apis
are
stable.
We
could
release
and
just
say,
no
an
issue.
You
can
only
have
one
reader
at
the
moment,
a
known
issue.
Well,
I
guess
the
if
you
only
have
one
if
you
can
only
have
one
reader.
Well,
I
guess
known
issue.
It's
not.
We
have
an
unempl.
Some
unimplemented
features
right.
F
A
So
I
think,
that's
totally
valid.
There
are
two
pr's
open
or
two
unreviewed
prs
that
are
open
that
do
make
minor
changes
to
the
actual
api
that
we
would
want
to
consider.
A
A
So
I
agree
with
that
issue
that
trask
opened
and
that
pr
reflects
that
and
then
another
one
is
4438,
and
you
know
what
this
does
is
this
kind
of
addresses
the
the
problem
that
we've
been
talking
about,
which
is
like
you
know,
do
we
need
these
helper
functions
on
your
metric
exporter
to
help
you
assist
with
always
cumulative
and
delta,
preferred
and
so
kind
of
my
approach
to
that,
and
you
could
take
a
look
at
the
pr,
but
essentially
you
know
we
have
this.
A
This
function
that
you
pass
around,
so
you
can
figure
your
your
your
metric
exporter
with
a
function
which
says
for
this
instrument
type
return,
this
aggregation
temporality.
So
I
I
I
think
we
should
name
that
function
with
a
functional
interface
and
I
think
that
cleans
up
things
in
a
variety
of
ways-
and
you
know
that's
something
to
think
about,
because
that
would
impact
our
api
compatibility.
E
F
F
A
So
then,
I
think
the
the
the
ones
that
we
would
want
to
look
at
are
the
two
minor
changes
that
affect
api,
potentially,
which
are
four
four:
three
nine
and
four
four
three
eight
and
then
any
any
other
finishing
touches
on
the
metrics
testing
stuff.
So.
E
E
A
A
B
B
Yeah
I
started
the
meeting.
I
couldn't
feel
anything.
I
couldn't
move
my
lower
lip
but,
as
you
can
see
like
it's
actually
moving
while
I
talk
now.
C
So
we
bravely
cracked
the
beer
open
with
the
fear
of
it
dribbling
out.
B
C
F
Cla
in
the
on
the
pr
okay,
and
that,
should
that
should
kick
it
into
gear.
C
Okay,
but
assuming
the
rebasing
works
and
all
that,
basically,
all
this
pr
does
is
just
add
in
a
builder
option
to
the
exporter,
so
you
could
export
it
as
json
instead
of
proto,
and
this
was
something
that
I
wanted
to
do
for
testing.
C
On
the
consumer
side,
we
use
the
sdk
and
our
functional
tests
to
send
the
data
and
tran.
We
have
our
consumers
on
our
on
our
end
that
we'll
then
we'll
translate
it,
but
the
tests
are
written
in
java
and
so
being
able
to
use
have
an
option
to
send
it
as
json
to
test
that
would
be
very
convenient.
Yeah,
I'm
not
sure
where
that
came
from
the.
B
B
This
a
couple
weeks
ago,
when
this
started
popping
up
everywhere.
F
C
C
Okay,
but
otherwise
I
know
that
just
from
talking
with
jack,
I
know
that
it
sounds
like
this.
Maybe
has
come
up
a
couple
of
times
or
has
maybe
come
up
before,
so
I'm
not
sure
what
the
where
those
conversations
have
led,
but
it
seemed
like
a
small
amount
of
code
change
just
for
at
least
making
testing
more
convenient.
If
not,
there
may
be,
I'm
speculating,
but
there
may
be
other.
You
know.
C
Whoever
who
knows
some
others
might
have
use
cases
for
being
able
to
export
as
json
so
yeah.
I
just
I
just
put
it
up
and
see
what
you
all
thought.
E
E
Specifically,
why
do
you
want
to
use
json
here
instead
of
photo,
though?
Okay,
so
don't
touch
that.
C
It's
it's
just
for
testing,
so
yeah
we're
we're
testing
we're
implementing
the
json
http
json
protocol,
support
on
our
consumer
side
and
our
functional
tests
to
exercise
that
path.
Just
happened
to
be
written
in
java.
So
having.
C
A
way
to
export
things
as
json
to
test
that
path
is
just
convenient
because
all
of
our
other
tests
use
these
exporters
and
and
so
just
being
able
to
to
use
it.
That
basically
just
says
export
it
I
mean
everything.
Was
there
other
except,
like
all
of
the
functionality
is?
Is
there
it's
just?
There
was
no
way
to
express
that.
So
I,
this
pr
just
adds
a
simple
way
to
express
that.
B
C
A
Via
browser
yeah,
we
want
to
be
able
to
accept
all
the
different
forms
of
otlp,
and
so
you
know
we
we
test
for
grpc
http,
protobuf,
http
json.
I
understand.
E
I
think
if
we
could,
as
jack
is
very
familiar
with
their
reflection
to
avoid
api's
strategy,
it
might
be
good
to
not
make
this
a
public
api
to
make
it
an
internal
one,
that
we
acknowledge.
That
uralic
is
using,
at
least
for
now,
because,
like
people
have
asked
for
json
thinking,
it's
better
than
protobuf,
even
though
it
almost
never
is
so.
I.
E
C
Interestingly
enough,
I
I
don't
actually
need
it
to
be
on
the
otlp
http
metric
exporter
builder,
I
I
internally,
our
tests
use
the
directly
the
ok
http
exporter.
The
only
reason
I
added
it.
There
is
because
there
weren't
any
direct
tests
against
the
okay
http
exporter
that
I
could
find,
and
so
I
just
piped
it
through
to
make
the
testing
more
compatible
with
the
existing
tests.
But
I
don't
have
a
strictness
need
for
it
to
be
all
the
way
through
to
the
public
api.
A
So
so
on,
android
is
talking
about
like
a
reflection
api.
So
we
we
use
this
trick
elsewhere.
So
we
have
internal
kind
of
experimental
support
for
adding
retry
logic
and
and
so
there's
an
a
a
public
api
in
an
internal
package
that
uses
reflection
to
flip
on
retry
behavior.
So.
C
A
So
then
you
would
have
a
package
private
method
on
the
okay
http
metric
exporter
builder,
so
the
you
know
that
so
it
it
would
still
be
on
the
that
that
public
class,
that's
like
the
top
level
one.
Oh.
A
Yeah
otlp
http
metric
exporter
builder,
and
you
know
that
that'd
be
useful,
because
then
we
could
still
have
our
tests,
like
you
know,
validate
it
if
we
wanted
to-
and
you
wouldn't
have
to
jump
through
extra
hoops
to
to
test
that
that
code
works
as
intended,
but
you
could
still
turn
it
on
with
reflection.
A
A
C
A
C
That
yeah,
that
that's
fine
with
me
like
I
said
I-
I
only
added
it
to
that
class
just
to
make
testing
more
convenient
for
the
project,
but
I
I
don't
need
it
to
be
on
that
as
long
as
because,
like
honestly
as
long
as
it's
on
the
okay
http
one,
that's
the
that's
the
thing
we
use
directly
anyway.
So
if
I
mean
it
can
be
here
as
package
private
and
then
still
have
the
tests
and
then
when
I
use
it
for
our
functional
tests,
I
don't
even
need
to
worry.
B
Yeah,
I
was
wondering
why
why
not
just
add
it
on
the
ok
http
one
then,
since
that
was
an
internal
class
anyway,.
C
Yeah
and
it
is
on
there,
the
the
only
reason
I
I
wired
it
through
is
because
the
existing
tests
did
not
test
the
okay
http.
Oh.
E
B
E
F
A
Encoded
is
a
pain
like
sometimes
it's
nice
to
turn
on
clear
text
like
payloads,
and
so
you
know
if,
if
you're,
if
you
have
logs,
enabled
and
any
sort
of
proxy
or
layers
like
that,
you
can
just
see
the
payloads
directly
instead
of
having
to
deal
with
the
binary
payloads.
So
it
is
nice
for
kind
of
testing
and
debugging.
F
E
A
Yeah,
if,
if
they
have
green
check,
marks
on
them,
I'll,
merge
them
and
and
write
up
the
release,
notes-
and
you
know
we'll
give
the
the
new
release
flow.
Another
another
roll
and
oh.
E
B
A
I'm
leaving
town
tomorrow
at
like
4
p.m,
ish
or
like
3,
30
pm-ish,
and
so
I
have
to
get
it
done
in
the
morning
or
it's
not
gonna
happen
trying
to
wonder
if
it's
just
like
safer
to
wait
till
monday
or
I
can
do
it
over
the
weekend
too.
E
A
B
C
Did
sorry,
did
anybody
catch
what
john
said
about
how
to
re-trigger
the
easy
cla,
so
I
thought
he
said:
slash
easy,
cla
and
I've
I've
I've
done
both.
E
E
B
B
Id,
I
think,
if
you
look,
I
think
and
see
how
it's
not.
The
comments
aren't
linked
to
your
picture
here.
B
E
E
C
Okay,
I'll
I'll
I'll,
try
and
figure
that
out.
B
C
And
yeah
and
then
I'll
update
it
to
not
be
public.