►
From YouTube: 2022-06-28 meeting
Description
cncf-opentelemetry meeting-2's Personal Meeting Room
A
B
B
Well,
yeah,
that's
good.
D
B
Yeah,
you
know
it's
like
it's
it's
okay.
I
mean
it's
it's
weird.
It's
definitely
weird,
but
yeah.
It's
good
yeah!
It's!
You
know
supportive
team,
there's
this
guy
sam
super
unsupportive.
But
besides
that
it's
true,
it's
bad
yeah,
I'm
a
hostile,
co-worker
yeah.
He
lacks
empathy
and
that's
great.
B
I
don't
like
his
face,
but
yeah
yeah,
it's
been
yeah.
It's
good.
I
mean
my
my
partner's
still
on
leave
she's,
not
back
at
work
till
like
august.
So
you
know
that's
been
nice.
I
feel
like
makes
my
stuff
easier,
but
yeah
it's
it's.
B
Yeah
they're,
it's
like
it's
a
lot
of
work.
They're
like
it's
like
they
need
to
be
cared
for.
It's
crazy,
it's
so
crazy!
It
wasn't.
I
was
aware
of
this,
but
then
you're
like
doing
it.
You
know
yeah,
that's
a
lot
of
work,
yeah,
that's
good
yeah
and
then,
like
you,
know
stuff.
It's
like.
I
came
back
and
there's
all
these
like
rfcs
and
project
proposals
and
stuff,
and
I'm
like
I
don't
remember
how
software
works.
It's
been
two
months
like
I
had
all
these
aspirations.
B
E
Yeah,
I'm
still
reminding
myself
how
to
type
this
guy.
I.
F
B
I'm
very
lucky:
there's,
like
you,
have
all
these
moments
of
like
45
minutes
of
like
you
can
do
a
task
in
45
minutes,
so
you
have
to
sort
of
like
plan
your
day
like
d-day
like,
but
also
the
task
has
to
be
something
like
you're,
not
you
have
like
these
windows,
and
so
you
just
end
up
watching
like
a
lot
of
bad
tv,
because
that's
sort
of
the
only
thing
that
feel
like
that
you
can
just
like
brainlessly
do.
But
then,
if
you
need
to
drop
it,
it's
not
a
big
deal.
B
You
matter
or
any
of
you
or
your
colleagues
at
monitorama
today.
D
So
I
probably
gave
like
a
bad
summary
of
this
several
weeks
ago,
but
there's
the
exponential
histogram
there's
a
lot
of
work
that
went
into
it
and
then
some
people
from
prometheus
dropped
into
one
of
the
meetings
and
was
kind
of
upset
and
berating
us
for
our
implementation,
mainly
because
they
have
something
similar.
D
Whatever
you
would
convert
between
prometheus
and
hotel
and
bortel
chose
like
the
buckets
to
be
lower,
bound,
inclusive
and
I
think
prometheus
chose
them
to
be
upper
bound
inclusive,
and
I
think
the
problem
ends
up
being
like
if
you
convert
between
the
two
things
could
end
up
in
the
wrong
bucket.
D
So
I
believe
that
j
mcd
has
kind
of
in
order
to
be
a
team
player,
even
though
I
think
he
and
the
hotel
community
strongly
feel
that
hotel
or
that
lower
bounds
are
far
superior,
have
decided
to
go
with
the
upper
bounds.
Just
to
be
diplomatic
about
things.
B
A
D
There's
an
otop
partial
success.
Pr
we've
mentioned
this
a
couple
of
times.
I
think
you
know
right
now.
It's
just
you
get
a
an
accepted,
whatever
signal
records,
so
the
number
of
things
that
were
accepted
and
then
an
error
message,
and
I
know
folks,
wanted
like
more
detail.
D
I
think
to
maybe
even
be
able
to
possibly
retry
and
do
other
things,
but
I
think
there's
an
issue
that
was
created
for
that
people
were
at
least
happy
enough
with
what
is
here
to
to
go
forward
with
that
and
then
maybe
he
would
see
something
more
advanced
come
later
on.
D
D
Removing
the
direction
label,
basically,
I
think
it
instead
of
having
metrics
with
a
with
a
direction
like
in
and
out
it
has
kind
of
changed
the
just
change
the
metric
name,
the
in
and
out,
has
gone
into
the
metric
name
itself.
So
previously
you
would
have
something
like
system.paging.operations
with
a
label
of
direction
with
values
within
and
out,
but
this
has
kind
of,
I
guess,
flattened
that
relationship
for
all
the
metrics
that
have
it.
D
Grpc
status
code
has
been
added
to
the
semantic
conventions
these
are
for
metrics.
So
if
it
is
a
grpc
call
things
for
matrix
yeah
so
for
any
grpc
call.
D
D
For
logs,
you
can
have
nested
attributes
where
you
can
have
like
a
map
of
those
values
you
can
have
nested
maps
and
otlp
kind
of
the.
The
definition
of
attribute
in
otlp
is
reused
for,
like
I
think,
span
attributes
the
thing
that
used
to
be
metric.
Labels
are
now,
I
guess,
metric
attributes
and
then
log
attributes
they
all
kind
of
use
the
same
definition,
and
they
all
allow
for
this
technically
in
the
proto-definition
allow
for
this
nested
map
representation.
D
But
specifically
this
was
added
for
logging
and
I
think
initially,
when
it
was
added,
it
was
added
with
the
intent
that
it
couldn't
use
everywhere,
but
some
people,
including
myself,
were
just
asking
if
this
made
any
sense
at
all
and
like
what
would
a
tracing
back
end
actually
end
up
doing
for
an
attribute
who
had
these
nested
map
values
just
most.
C
D
Back
ends,
don't
really
support
it.
You're
gonna
have
to
like
try
to
flatten
out
some
key
representation,
and
I
have
like
foolishly
tried
some
of
this
stuff
in
the
past,
which
is
why
I
was
so
against
it.
D
It's
like,
if
you
get,
I
don't
know
you
can
get
in
a
situation
where
you
have
an
array
and
nested
map
value
or
map
values
in
the
array,
and
it
just
gets
terrible
because
there's
not
a
lot
of
great
ways
to
flatten
that
you
end
up
with
like
an
index
in
the
key,
I'm
probably
going
off
the
rails
here.
Anyways.
There
were
reasons
why,
having
these
nested
values
don't
make
a
lot
of
sense,
but
I
think.
D
Nev
was
looking
at
the
js
implementation
and
thought
that,
at
least
for
spans
and
metrics
that
you
were
able
to
use
these
nested
values
it
turns
out.
I
don't
think
that
you
are
so.
He
was
trying
to
kind
of
update
the
spec
to
say
that
you
know
you
can't
you
can
use
nested
map
values
again,
basically
bringing
it
in
in
line
with
what
the
protos
technically
allow.
D
I
mean
I've
gone
too
far
to
the
weeds
on
this
one,
but
I
think
I
I
would
be
surprised
if
this
goes
through
without
a
lot
more
discussion.
Anyways.
F
The
only
thing
that
I
had
was
about
the
http
request
and
response
size,
not
really
a
question,
just
a
comment
which
was
just
that
we've
looked
into
doing
this
for
radius,
not
for
us.
I
don't
think
that
goes
over
http.
I
don't
know
what
they
have
some
kind
of
custom
whatever,
but
it's
cool
to
see
that
this
is
maybe
gonna
get
into
the
spec,
because
there
might
be
analogous
stuff
for
databases
and
what
have
you
because
we're
in
a
situation
where
it's
like?
F
Oh
we're,
interested
in
like
knowing
how
big
the
you
know,
keys,
are
or
how
big
are
the
yeah?
How
big
is
the
value
stored
at
a
specific
key
in
redis
and
like
want
to
build
reports
off
of
that
and
currently
we're
kind
of
home
brewing
our
own,
our
own,
like
solution
there
and
when
we
tried
to
when
we
thought
about
contributing
it
upstream,
it
was
like
well
like
is
this:
is
this
at
all
relevant
to
open
telemetry?
Or
is
this
just
us
anyway?
That's
off
topic.
F
I'm
now
one
of
those
people
that's
going
off
topic,
I'm
supposed
to
be
the
the
teto
spec
sig.
You
know
keep
time
and
check
guy,
so
that's
I'll
finish
that
there.
D
D
In
between
messaging
and
databases,
yeah
for
fun.
D
Yeah,
so
we
only
have
one
more
thing
to
get
through
and
it
is
hardware
metrics
semantic
conventions.
It
took
me
a
minute
to
kind
of
understand
what
they
were
talking
about,
but
basically
the
the
problem
people
were
having
with
this
is
that
you
end
up
with
a
lot
of
these
metrics
that
are
kind
of
interesting.
D
Here's
an
example:
it's
like
a
up
down
counter
where
you
have
a
label
set
of
okay,
degraded
and
failed,
and
you
would
expect
only
one
of
them
to
be
one
and
the
rest
of
them
to
be
zero,
and
there
were
a
lot
of
those
showing
up
in
here
and
I
think
the
metrics
folks
were
saying
they
wanted
something
called
a
state
set.
D
It's
like
some
new
newfangled
thing
where
it
is.
Basically,
I
don't
know
some
first-class
representation
of
this
sort
of
situation,
so
you
don't
have
to
kind
of
manually.
Do
it
with
an
up-down
counter
and
some
you
know
validations
yourself
to
kind
of
take.
Take
the
take
the
hard
work
out
of
it
and
I
think
one
of
the
reasons
why
they
like
that
is
that
they
could
really
make
the
representation
compact
on
the
wire.
D
Where,
like
anything
in
the
state
set,
you
only
send
your
one
and
then
your
everything
else
is
implied
to
be
zero,
but
I
think
ultimately
everyone
was
like.
Well,
we
don't
have
this
thing
and
everybody's
too
easy
to
work
on
it,
so
I
think
they
were
going
to
try
to
find
a
way
to
see
this
thing
through,
but
the
the
conversation
on
this
thing
was
quite
lively
with
120
comments.
E
D
F
It's
good
to
like
just
know
what
the
weeds
are.
You
know
and
it's
horrifying
to
begin
to
understand
what
you're
talking
about
that's,
really
I've.
I've
gone.
I've
changed
over
to
the
other
side.
Now
I'm
like
oh
yeah,
the
histogram
bucket
thresholds,
like
I
totally
that
makes
sense
to
argue
about
with
the
prometheus
people.
D
F
I
got
some
general
thumbs
up.
I
I
don't
know
I
I
have
like
an
itch
to
just
kind
of
standardize,
our
rubocops
and
just
have
a
robocop
dot
yaml
in
the
root
of
the
repo
and
then
just
like
periodically
copy
it
over
or
even
just
like
sim
link
it
in
ci
or
something
like
that.
F
Any
objections
to
doing
that,
because
it's
kind
of
going
to
be
I'm
not
sure
that
I'll
actually
do
it
in
the
short
term.
But
it's
kind
of
like
mindless
work
that
can
be
achieved
when
my
brain's
not
really
working.
So
it
seems
appealing.
Would
people
be
okay
with
consolidating
onto
one
rubocop
file
per
repo
and
then
just
dealing
with
the
fallout
in
a
bunch
of
prs
fine
by.
A
F
Right,
no,
I
know
I
bet
there.
I
bet
there
is,
I
don't
exactly
know.
What's
going
on
on
the
file
system
during
ci
runs
but
yeah,
I
think
referencing,
a
parent
would
be
the
ideal.
I'm
not
exactly
sure
why
there
is
one
robocop
per
gym,
but
anyway,
that'll
be
some
boring,
tedious
work
for
you
know
the
mid-afternoon.
When
I
should
be
asleep.
A
E
A
D
E
A
The
one
the
one
to
test
is
because
we
have
a
rule
that
we
just
disable
like
certain
things,
because
it's
like
you
have
a
test
and
like
if
you
have
described
block
it'll,
be
like
that's
too
long,
bro
and
you're
like,
but
it's
a
describe
block
and
here's
all
my
examples.
They're
like
we
don't
care
that
method
is
too
long,
and
so
it
gets
disabled
for
tests
and
then
I
think
for
the
non-test
extension.
A
D
There
we
go
there's
at
least
some
precedent,
possibly
some
opportunities
to
name
things
a
little
better,
but.
A
F
D
Yeah
I
mean
I
feel
like
some
of
the
lack
of
sharing
could
have
just
happened.
You
know
from
the
early
days
of
cutting
and
pasting
your
scaffolding,
and
probably
just
was
in
oversight
in
the
process,
but
there
could
be
some
better
reason.
D
F
I
had
just
sort
of
a
general
question:
arielle
had
an
issue
in
the
contrib
repo
and
people
don't
care.
I
don't
have
to.
F
We
were
thinking
about
like
some
kind
of
guide
for
style
guide
for
writing
instrumentation,
because
there's
you
know
certain
stuff
that
we
kind
of
implicitly
think
and
it
might
be
good
to
like
lay
it
out
in
a
document
and
read
me
of
some
kind
or
contributing
guide.
A
This
has
to
be
performance,
mindful
right,
like
so
avoid
unnecessary
string
allocations,
don't
just
like
do
merge
without
a
bang
or
compact,
without
a
bang,
because
you'll
duplicate,
whatever
element
you're
iterating
on
in
the
early
days,
you'd
see
like
a
method
or,
like
a
maybe
say,
like
a
40
line,
block
a
code
get
blown
out
into
like
15
methods,
and
it's
like
kind
of,
like
prefer
simplicity,
right
things
like
that.
I
think
are
really
really
important.
F
A
Yeah
there's
other
ones
like
we,
we
use
pre-pending
for
our
preferred
flavor
of
monkey
matching
versus
like
method
aliasing
intermingling.
The
two
is
a
recipe
for
disaster,
so
maybe
in
a
future
future
world
we
support
both
but
like
right
now,
we've
made
the
opinionated
choice
of
being
like
we
prepend
and
it
doesn't
come
up
too
often,
but
sometimes
people
will
be
like.
Oh
I
heard
instrumentation.
They
might
either
be
used
to
using
method,
aliasing
or
they're
using
reference
material.
That
does
that
and
then
we
say:
hey,
no,
don't
do
that
and
so,
like.
A
D
Those
are
all
good
ideas,
I
would
add,
like
one
of
one
of
the
big
ones
is
that
instrumentation,
including
instrumentation,
should
not
result,
in
any
exceptions
being
raised,
that
yeah,
in
any
new
exceptions
being
raised
in
any
in
any
scenarios,
so
either
like
explicitly
an
explicit
raise
or
something
implicitly
just
because
you
mess
up
and
those
are
ours.
A
Yeah,
like
I
think
it's
worth
like
yeah
like
that
is,
it
does
actually
make
sense
to.
I
know
it
was
like
kind
of
teasing
there,
but
it
does
make
sense
to
add
that
and
pointing
a
link
to
the
open,
telemetry
error.
Handling
headlines
like
open
telemetry
is
not
supposed
to
have
like
imploding
versus
exploding
like
if
open
telemetry
fails.
It
fails
within
open
telemetry.
A
It
doesn't
leak
out
and
blow
up
your
app
that
you're
instrumenting
and
like
as
well
for
most
of
us
or
people
who
operate
code
that
runs
on
a
bunch
of
services.
We
don't
own
being
really
strict
about
that,
has
made
deploying
it
too
many
services
a
lot
more
comfortable
right,
like
we
can
roll
out
of
change
to
like
hundreds
of
apps
at
shopify
across
hundreds
of
deployments
or
thousands
of
deployments,
and
it's
like
because
we've
been
really
strict
about
following
the
implode
versus
explode
guideline.
It's
like
it's.
E
A
So
I
think,
like
there
needs
to
be
like
probably
like
a
distinguish
between
like
the
hard
rules
and
the
like
the
guidelines.
Right,
like
your
stuff
doesn't
blow
up,
your
stuff
is
performant,
and
then
it's
like
the
software
guidelines.
It's
like
your
stuff
is
easy
to
read.
You
have
not
used
100
methods
for
your
100
lines
of
code.
A
D
I
think
it's
a
good
idea
to
have
this
document
and
I
think
we
can
all
kind
of
maybe
add
some
suggestions
as
as
we
introduce
and
review
it,
but
I
feel
like
it's
one
of
those
things
where
like
if
we
do
get
into
like
some
longer
more
challenging
conversations
on
a
pr
and
kind
of
like
arrive
at
aha.
This
is
really
something
we
should
just
codify
and
not
have
to
debate
in
the
future
and
just
kind
of
point
to
the
guide.
D
Then
we
can
kind
of
you
know,
keep
it
up
to
date
and
improve
it,
as
things
become
more
more
apparent.
A
Yeah,
I
think,
like
a
nice
addition
to
something
like
this
and
I'm
like
just
kind
of
talking,
because
I
don't
know
if
I'll
actually
do
it,
but
providing
scaffolding
for
people
to
pretty
easily
like
profile
their
contribution
like
if
they
had
like
a
significant
change.
It's
like
did
like
did
what
you
do,
allocate
a
bunch
more
memory
like.
Does
it
suck
right
like
not
suck
but
like?
Are
you
missing
something
but
like,
instead
of
just
like
trying
to
have
to
lean
entirely
on
reviewers
to
try
to
like
rock
it
some
some
of
it?
A
It's
obvious?
Sometimes
it's
not
obvious,
like
I
did
when
I
did
that,
like
refactor
with
the
reddest
thing
like
I
was
like
oh
wow,
like
I
made
a
big
improvement,
I
had
set
up
this
little
like
script.
To
do
and
it's
like.
Can
we
provide
something
like
that?
That's
a
little
bit
more
canned
and
a
little
bit
more
generic,
and
then
now
I'm
doing
the
tangent
thing
it's
like.
A
Should
we
be
baking
that
into
all
of
our
test
cases,
for
for
all
these
things,
that,
like
you,
run
like
the
heat,
profiler
and
there's
no
regressions
here
like
we
were
allocating
this
much
before
and
we're
either
doing
less
or
the
same
right
so
that
we
don't
have
to
lean
on
us
being
like?
Did
you
do
it?
Is
it
good
like
it's
like?
A
No,
like
tests
failed
and
like
it,
and
there
might
be
perfectly
valid
reasons
where
it's
like
yeah,
like
the
footprint
did
increase,
but
like
we
think
that
this
is
like
an
acceptable
trade-off,
because
this
is
a
valuable
feature,
so
yeah
I'll
hit
the
end
of
this
tangent
now.
But
I
think
that
would
be
a
good
like
addendum
to
that.
D
D
Right
there
above
the
couches
now
this
whole
thing
that
robert
was
just
talking
about
reminds
me
of
ruby
bench
and
I've.
Always
I've
always
wanted
this.
I've
always
wanted
this
for,
like
like
a
variation
of
this
for
instrumentation
and
generally,
like
some
commit
benchmarks.
I
don't
know
something
like
this.
D
And
yeah
so
there's
like
a
test
active
record,
my
sequel
to
destroy,
and
this
runs
on
like
every
commit,
and
these
are
the
iterations
per
second.
You
can
kind
of
see
that
occasionally
you
do
get
these
variations
and
some
regression
was
introduced
here.
They
fixed
it
and
then
you
can
kind
of
see
like
all
of
the
objects
allocated,
and
it
looks
like
the
that
has
been
trending
down,
which
is
pretty
awesome,
and
it's
like
a
simple
script
like
this.
D
So
for
instrumentation,
I've
always
kind
of
wanted
to
see
like
for
active
record,
destroy
or
yeah
my
sequel
destroy,
for
example
like
what
does
that
look
like
what
does
the
profile
look
like
without
instrumentation?
What
does
it
look
like
with
instrumentation
and
then,
if
it's
a
new
commit,
you
know
what
those
look
like
between
commits,
but
I
feel
like
this.
D
D
B
B
There
was
a
lot
of.
There
was
a
lot
of
noise,
but
I
think
what
robert
is
saying
is
correct,
and
maybe
just
generally
we
might
want
to
have
something
in
there,
like
you
know,
provide
evidence
like
the
more
evidence
the
better
here,
so
whether
that's
like
you're,
including
profile,
is
awesome,
whether
that's
a
screenshot
of
like
here's.
What
this
trace
looks
like
in
a
trace
viewer,
here's
how
you
know
like
here's
or
like
a
lot
of
times
like
from
shopify.
It's
like
well,
we've
run
this
in
prod
or
at
planet
scale.
B
So
like
we
have
confidence
so,
like
you
know
so
stuff
like
that,
I
think
it
would
be
nice,
like.
Maybe
we
have
something
in
there
like,
we
feel
we
know
we
may
ask
for
these,
please
be
prepared
to
provide
them,
and
maybe
we
can
point
them
to
tools
to
generate
that
stuff.
Whether
that's
like.
Oh
here's,
you
know,
here's
an
open
source
trace,
viewer,
here's
a
profile
or
whatever,
but
yeah
it
would
be.
I
would
love
if
this
ruby,
something
like
ruby
bench
was
it's
like
the
yeah.
I
think
it's
awesome.
B
I
just
I'm
not
gonna.
Do
the
work
so
yeah.
A
That's
that's
a
good
point
too,
like
pictures
of
like
what
does
this
trace
like
actually
look
like
it's
like,
I
could
read
the
output
but
like
reading
raw
span.
Data
is
like
kind
of
brutal
like
it's
fine,
if
it's
one
span,
but
then
it's
like
you
have
two
and
you're
like
okay,
but
then
you
add
ten,
and
it's
like.
I
don't
wanna
do
this,
but
I
can
read
a
picture
pretty
quick,
so
it's
like
it
would
be
nice
and
we
don't
have
to
lean
into
like.
A
Here's
how
you
point
your
local
thingy
at
here's
like
a
script
that
works
and
like
you
can
actually
just
like
prove
it
like
you
say
this
works,
show
it
like
give
us
a
picture
yeah,
and
it's
also
just
like
really
informative
for
people
who
are,
I
think
it's
valuable
for
people
who
are
experimenting
with
it,
because
this
is
the
sort
of
thing
we
do
internally
when
someone
is
new
to
tracing
like
hey.
A
Here's
this
like
internal
thing,
we
have
just
like
run
the
boot
up
command
and
you
can
play
around
with
like
your
changes
and
see
them
before
you
push
them
to
production.
That's
like
it
really
encourages
discovery
and,
like
adoption
so
like
if
it
works
internally,
I
imagine
we
could
probably
have
similar
effects
externally
right,
which
is
good
for
us,
because
we
obviously
won't
be
able
to
use
this
stuff.
D
Yeah,
this
is
probably
a
a
pipe
dream
unless
we
get
plenty
of
time
and
lots
of
staff
yeah
having
having
some
pointers
or
guidelines
around
things
in
terms
of
being
able
to
provide
some
kind
of.
D
A
I
think
there's
like
the
first
pass
of
this
is
just
like
providing
an
example
script
where
people
can
just
like,
add
their
code
and
run
it
locally
and
they
can
share
screenshots
and
then,
like
step.
Two,
in
my
mind,
is
like
make
it
part
of
like
a
generic
test
that
we
can
copy
and
paste
into
instrumentation,
maybe
like
high
level
integration
thing,
and
it's
like
you,
wouldn't
provide
nice
pictures
or
anything.
A
But
it
would
just
like
rip
out
like
a
summary
line
from
the
profile
and
compare
it
against
like
some
hard-coded
value
like
it
should
be
less
than
or
like
it
should
be
within
a
range
of
this
number.
So
like
people
run
ci
and
those
changes
get
introduced,
it'll
like
automatically
check
it
and
then
like
the
very
golden
version,
is
what
you're
you've
got
in
front
of
us,
and
that
would
be
like
pretty
wicked,
especially
like
the
comparison
of
like
uninstrumented
and
with
instrumentation
like
that
is
really
cool.
F
We
could
do
I
mean
if
we
have
like
a
local.
If
we
have
an
easy
local
setup,
we
could
just
write
a
script
that
reruns
the
same
stuff
with
the
environment,
variable
flipped
on
and
off
for
a
particular
instrumentation,
and
that
could
get
us
part
of
the
way
there
without
having
to
integrate
ruby
bench.
But
we
could
just
have
it
as
part
of
the
pr
checklist
I
mean
I
hate
pr
checklist,
so
maybe
not,
but.
B
My
dreams,
if
I
have
to
run,
we
have
like
a
really
long
checklist,
but
then,
like
the
ninth
thing
on
the
checklist,
is
just
like
literally
check
this
box
and
we'll
send
you
three
dollars
or
something
and
see
if
people
do
like
one
of
those
like
I
used
to
have
a
teacher
who
would
like
do
stuff
like
that
anyway,
I'm
yeah,
the
sugar's
kicking
in
from
this
lunch.
C
A
A
But
I
like
look
at
my
phone
and
there's
this
brown
sludgy
substance
and
I'm
like
I
gotta,
go
wash
my
hands
because
he
sat
on
me
and
my
phone
because
he
doesn't
like
riding
in
the
car,
so
he
just
planted
himself
firmly
like
on
top
of
me,
because
in
car
rides-
and
I
was
driving
away
from
my
dad-
which
he
doesn't
like
so
there's
a
whole
thing.
A
So
I'm
gonna
drop
off
and
wash
everything
a
balance
is
required
by
the
way.
B
B
I
I
don't
have
anything
else
for
sure.
Do
my
best,
everyone
cheers.