►
From YouTube: Kubernetes SIG Service Catalog 2018-07-26 - Delete
Description
potential rubric for voting
Q&A
option winnowing and consolidation
A
A
Try
and
provide
a
fair
and
objective
way
to
evaluate
these
things.
I
was
thinking.
Maybe
we
could
come
up
with
some
kind
of
a
kind
of
rubric
that
asks
us
to
look
at
what
we
care
about
and
then
ask
us
how
much
we
care
about
it
and
I
wrote
some
cases
down
and
I'm
sorry
to
spring
this
on
you
at
the
last
minute
here,
but
you
know,
maybe
we
could
go
over
those
options
and
think
about
it.
A
B
Like
your
list,
I
think
those
are
all
very
important
things
we
need
to
find
out.
You
know
who
cares
about
what,
because
they're
all
very
good
topics,
but
at
some
point
I
do
think
we
should
look
through
the
other
document
and
a
list
of
options
that
people
put
forward
to
make
sure
people
understand
what
he
touching
actually
means
and
there's
no
confusion
about
them
sure
did
Gigi.
Should
we.
A
A
Since
we're
looking
at
this,
let's
just
discuss
this
rate
right
now.
The
my
rubric
is,
basically,
you
know
two
scales
for
each
for
each
option
should
be
you
know
how
well
does
this
achieve
the
objective
that
we
wrote
down
as
an
option
and
then
how
much
do
we
actually
care
about
solving
this
problem?
A
So
the
first
thing
I
wrote
down
was
is
implementable
with
CRD.
We
haven't
made
an
actual
decision
on
this
yet
and
the
decision
being
whether
or
not
to
go
to
CRTs
and
up
the
API
server.
It
seems
to
be
something
that
is
desired
on
multiple
fronts,
for
multiple
reasons,
so
it
seems
like
it's
a
fairly
important
objective
but
I
it's
it's
more
of
a
long-term
one
rather
than
a
right
now,
unless
you
know
a
CRD,
implementation
appears
out
of
nowhere
soon,.
A
So
I've
got
this
sort
of
behavior
thing
broken
up
and
do
a
couple
change
of
user
experience,
basically
being
how
much
it
if
we
do
provide
the
same
user
experience
to
an
end-user,
how
much
extra
wrapping
layers
and
how
much
does
the
client
library
have
to
do
versus
it's
a
standard?
It's
a
straight
pass
through
from
user
interaction
to
API
interaction,
Morgan.
B
My
my
question
or
comment
on
the
employ
of
the
CR
DS
is
for
the
most
part
it
does
sound
like
it
is
kind
of
important
thing
for
people
in
the
long
run
seems
like
a
fair
number
cool.
Do
you
want
to
head
that
direction
and
I?
Don't
know
Sarah
problem
with
that?
My
only
question
is:
do
we
have
a
sense
for
how
set
in
stone
CRD
support?
Is
you
know
so,
for
example,
right
now
we
know
that
some
of
the
options
won't
work
under
CR
these
because
they
don't
have
the
right
hook
points
for
it.
B
However,
I
don't
believe
that
it
means
those
whole
point
cannot
be
added
for
CRT
support,
because
CRT
support
is
still
sort
of
in
the
works
right.
It's
in
flux,
they're,
adding
features
all
the
time.
So
I
think
it's
important
to
understand
that
you
know
when
we
say.
Does
it
work
with
CR
D,
because
a
confusion
with
CR
DS
there's
two
different
answers:
there's
does
it
work
today
with
CR
DS,
and
is
it
possible
for
CR
DS
to
change
to
add
support
for
what
we
want
to
do
or
for
that
particular
option
right
now?
B
Of
I
I
still
need
to
send
up
a
note
to
the
coop
dead
mailing
list
to
see
if
there's
interest
in
falling
working
group.
But
yes,
if
the
working
group
does
get
formed,
then
yeah
we're
going
to
have
a
work
group
to
talk
about.
Do
you
want
to
support
these
use
cases
and
what's
the
possible
best
way
forward
to
support
them?
If.
C
Let's
say
that
like
we,
we
make
that
working
group
and
then
it's
it.
You
know
say
it
say
it
like
takes
a
month
and
then
something
comes
out
of
it.
What
would
be
the
earliest,
like
kubernetes
release
that
could
have
the
changes
in
I'm
just
trying
to
get
like
a
best
case,
an
area
of
like
how
long
would
it
take
for
us
to
have
to
sit
and
wait
on
doing
the
CRE
migration?
While
we
wait
for
something
from
the
working
group
to
land
and
see
our
knees,
you
know
what
I
mean
yeah.
C
B
It's
possible
right,
I
mean
it
depends
on
the
will
of
the
community
right
if
the
bullet
case
and
say
hey.
These
are
super
important.
You
know,
and
it
only
requires
a
fine
line
code
change.
Then
yeah,
then
getting
them
on
thirteen.
They
look
at
it
and
say:
oh,
my
god.
We
need
to
really
think
long
and
hard
about
this
and
that
it's
gonna
take
a
couple.
Recycles
ish
I,
just
like
a
total
unknown
to
me.
Okay,.
C
Yeah
I
was
just
trying
to
give
or
like,
if
best-case,
we're
still
gonna
have
to
wait
months
and
months
and
months
like
it's
probably
worth
looking
at
both
sides
of
that
like
CRT
today,
we're
so
CRT
after
the
working
group
sounds
like
you
can't
just
look
at
one
or
the
other.
We
have
to
do
both
yeah
based.
D
On
our
discussions
with
seek
architecture,
I
don't
see
I
I,
don't
I
can
say
that
there
was
any
interested
in
like
producing
such
behavior
I
believe
so
I
I
would
say
that
it's
probably
safe
to
assume
that
it
will
never
make
it
to
theory,
because
it's
like
they're.
The
only
reason
why
we
can
can
implement
such
behavior
in
API
server
is
that
we
have
so
much
power
and
in
CI
I.
D
Don't
think
they
really
need
want
to
do
that,
so
they
already
have
all
the
all
kinds
of
like
admission,
web
hooks
and
other
kinds
of
like
integrating
intercepting
requests
and
so
on.
I
think
they
would
say
that
there
are
there
already
ways
of
customizing.
The
behavior
is
just
not
the
work,
not
the
one
proposed
there.
I'm.
B
Not
sure
you
can
really
say
that,
though
right
I
mean
at
one
point
in
time
all
the
various
hook
points
that
in
kubernetes
never
existed,
and
yet
they
were
added
slowly
over
time.
So
I
don't
think
it's
fair
to
say
that
it's
additional
hook
points
may
not
be
added
this
year.
Ds,
we
just
don't
know
no.
D
I
disagree
because
there
was
a
they
were.
There
was
support
for
admission
controller
for
for
a
long
time
and
like
the
only
problem
with
what
with
that,
was
that
if,
in
order
to
introduce
new
admission
controllers,
you
would
have
to
recompile
the
entire
communities.
So
then
they
introduced
this
admission
webhooks
for
both
API
servers,
NCS
as
well
and
for
CID
I,
don't
think
we're.
We
have
this
point
of
extension
where
we
would
like
we
would
hook
because
I
like
again
in
the
docs.
D
D
It's
like
it's
I,
I,
guess
it's
the
same
as
like
CID
will
never
support
anything,
but
the
etcd
is
a
back
story
and
they
made
a
conscious
decision
that,
like
even
though,
for
example,
Amazon
wanted
to
introduce
like
abstraction
where
to
support
dynamic
DB.
They
just
clearly
said
no.
We
just
want
to
use
all
the
EDC
d3
features
and
like
just
want
to
commit
to
using
etcd
all
the
way,
so
basically,
they
still
probably
able
to
hack
something
with
API
server.
D
C
Want
to
clarify
something:
it
was
my
understanding
that
what
we
were
going
to
put
up
to
the
sig
two-photon
is
what
to
do
in
the
intermediate
term.
Between
now
and
whatever
comes
out
of
the
working
group
and
whatever
comes
out,
the
working
group
is
what's
going
to
help
set
our
long-term
goals,
but
we're
voting
on
what
we
can
do
today.
Do
we
all
agree
on
that,
because
otherwise,
it
kind
of
changes.
C
What
we
want
to
like
evaluate
some
of
these
on,
like
the
possibility
versus
what
could
we
start
implementing
tomorrow
and
submit
a
PR
core,
so
I'm
hoping
we
can
say
we're
talking
about
just
a
short-term,
so
that
if
I
can't
guess
the
future,
if
things
change
later,
then
we
can
reevaluate
what
we
want
to
do.
Yeah.
B
I
I
think
in
general
I
agree
with
your
think:
Earl
I,
guess:
I'm
I'm
not
quite
clear
of
the
implications
of
this
statement,
though
right,
because
if
you're
saying
we
need
to
make
sure
we
go
as
a
short-term
solution
because
we're
living
in
the
world
we
have
today
right,
we
can't
assume
anything
in
the
future
and
I
get
that,
but
does
that
imply?
We
only
need
to
care
about
API
server,
or
does
that
mean
we
need
to
think
a
little
bit
longer
ahead
and
no,
they
were
probably
muda
charities
and
take
that
into
account.
C
Think
that
our
ability
to
move
to
CRT
is
within
our
control
and
I
believe
all
major
roadblocks
that
we
had
previously
to
consider
adopting
it
we
figured
out,
and
so
for
me,
it's
like
it's
not
a
huge
thing
out
of
our
control
to
say
we
as
a
stake
want
to
move
too
Ciardi
and
that's
something
we
like
to
do
in
the
near-term
future
versus
the
working
group
stuff,
which
unfortunately,
like
we
have
very
little
control
over.
We
can.
We
can
argue
our
case.
C
We
can
talk
to
other
cigs
and
people
who
have
somewhere
use
cases,
but
if,
if
cigar
CH
doesn't
agree
like
we
don't
have
much
to
do
about
it.
So
like
the
Sierra
D
feels
like
worth,
including
in
this
discussion,
because
it's
just
the
probability
is
a
lot
higher
that
we
can
just
do
it.
If
you
want
it,
I
got.
B
It
not
yeah
no
I,
agree
and
and
I
so
I
think
what
you're
saying
is.
The
current
state
of
things
is
probably
move
to
see
our
DS
and
that's
part
of
our
short
term.
This
short
term
decision
planning
thing
and
I'm.
Ok
with
that
decision,
I
just
wasn't
sure
when
you
start
talking
about
short
term
decisions,
it's
it's
going
to
see
our
DS
part
a
short
term
decision
process,
or
is
it
a
long
term
thing
you're
saying
I'm?
Okay
with
that
answer,
okay,.
C
Yeah
yeah,
that's
what
I'm
hoping
for
we,
if
we
all
agree
to
that.
That's
cool
I,
just
wanna,
make
sure
that,
like
when
people
they
understand
what
they're
voting
on
it'll
be.
You
know,
let's
just
say
that
what
we
could
do
and
like
the
next
two
releases
of
kubernetes,
does
that
help
us
focus
on
what
we
want
to
do
for
arabic.
I.
E
There's
another
I:
don't
know
how
to
I,
don't
know
where
to
put
this
or
worded
or
that
kind
of
thing.
So
I'll
just
say
it
straight
up:
I,
don't
know
if,
let's
say
I,
don't
agree
that
we
need
to
fulfill
the
full
purpose
that
is
set
out
in
the
problem.
Description,
I,
I
feel-
and
maybe
this
is
just
part
of
the
discussion
of
the
solutions.
E
I
feel
that
if
there
is
a
a
check
available
and
we
check
with
a
broker
about
doing
the
delete
and
then
you
know
it's
part
of
the
flow
I,
don't
know
that
I
care
about
the
time
between
a
pre-check
and
and
the
delete
and
the
broker.
You
know
trying
to
change
in
his
mind
and
saying
and
I
guess
the
reason
I'm
bringing
this
up
is
this
I
think
we're
going
to
enter
the
description.
A
A
E
A
A
You
know
how
we
block
delete
our
expected
status.
The
only
things
modified
by
you
know
the
controller
or
the
user.
That
kind
of
stuff
sort
of
this
is
kind
of,
maybe
part
of
it,
but
I
think
it's
a
little
outside
of
API
specifically
and
basically,
what
is
what
does
an
observer
expect
to
see
happen
and
are
we
violating
that?
I
think
this
is
a
little
little
looser
thing
personally,
but
last
thing:
I
added
based
on
the
con.
The
running
commentary
here
was
basically
and
I'm
short
term.
A
A
B
Yeah
I
don't
need
to
read
it
to
you
guys,
I
guess.
The
question
is
where,
where
is
a
disagreement
on
the
problem
statement?
I.
C
E
E
Don't
maybe
this?
Maybe
this
is
just
best
left
to
the
vote.
My
feeling
is
that
with
the
pre-hook,
you
know
with
a
pre-check
if
things
change
between
a
pre-check
in
the
actual
deletion.
That
that's
that's
fine,
that's
you
know
that's
a
kind
of
an
expected
edge
case
that
results
in
a
wedged
problem.
Oh
a
wedged
resource
that
that
needs
to
be
dealt
with
right
now,.
B
D
I
have
raised
some
concern
with
the
problem
statement
on
my
side
and
I.
Think
that
so,
like
this
statement
implies
that
community
sees
has
incorrect,
behavior,
which
is
not
compliant
with
always
be
respect
and
I,
think
that
this
is
not
necessarily
true,
because,
like
communities
has
their
own
like
design,
documentation
and
so
on,
and
there
is
like
a
community's
API
style
and
like
how
everything
works,
how
should
you
integrate
with
giving
this
system
and
so
on?
And
so
this
catalog
has
a
like
a
base
over
or
C
ad
it
doesn't
matter.
D
Is
it
should
be
like
communities
compliant
in
that
matter?
So
I
believe
that
the
problem
we
actually
have
now
is
that
we
have
a
service.
Catalog
server,
which
is
e
G,
which
is
qunit,
is
compliant,
and
we
have
always
this
tag
which
has
a
corner
case
which
is
not
like
compiler,
like
not
hundred
percent
compatible
with
this
cuny's
compliance,
and
this
is
the
problem
area
we
have
here.
D
Is
that
like
when
the
user
issues
a
delete
in
service
instance,
for
example,
in
units
user,
doesn't
do
anything
wrong
and
if
community
is
like
authorization,
I
mention
who
will
cook
validation
whatever
if
it
bypasses
all
these
checks,
and
it
make
makes
you
makes
it
to
the
etcd
change?
There
is
nothing
wrong
with
that.
The
problem
is
that,
when
the
look
as
a
controller
tries
to
sink
two
systems,
one
is
communities
as
a
platform,
and
then
the
other
is
always
be
broker
and
there
is
like
definitely
a
mismatch.
D
There
is
a
corner
case
where
one
picture
of
the
world
doesn't
match
the
other,
and
for
me,
that's
the
best
problem
statement.
What
do
we
do
when
we
have
a
mismatch
between
two
girls
and
the
police
problem
statement?
Clearly
says
that
broker
is
the
right
one
and
community
is
the
Rockland
and
communities
should
have
never
actually
allowed
this
division.
But
to
me
it's
it's
like
we.
We
already
have
like
J
written
a
solution
that,
like
you,
can
prevent
this
by
this
and
that
which
is
also
communities
compliant
and
I.
I.
D
B
I
don't
think
table
statement
as
I
says:
one
side
is
initially
right
or
wrong.
It's
more
stating
the
state
of
the
world
right
in
the
sense
that
kubernetes
does
solution.
Timestamp
the
service
broker
can
saying
no
and
you're
right.
The
problem
is,
we
have
two
different
views
of
the
world
and
we're
trying
to
reconcile
those
III
but
I.
Don't
think
the
problem
statement
says
who's
right
or
who's
wrong.
D
D
Just
say
is
like
here
is
my
view
of
what
like
desired.
State
is,
and
it
has
a
division.
Timestamp
and
I
will
try
and
make
it
happen
in
the
real
world.
If
it
can't
make
it
happen,
there's
it
doesn't
mean
that
something
in
communities
is
wrong
or
it
should
be
rolled
back
or
anything.
There
is
no
such
thing
as
a
rollback
and
I.
Don't
know
for
me.
That's
all
right,
I,
don't
know.
D
B
You're
talking
about
a
particular
solution
which
is
different
than
whether
the
problem
statement
is
right
or
wrong,
because
if
the
false
statement
is
saying
he
that,
if
there's
something
in
the
problem
statement
that
is
incorrect,
let's
get
that
fixed,
but
I
haven't
heard
anybody
say
what's
incorrect
about
it,
just
that
their
solution
is
the
right
one,
even
though
someone
says
no,
it's
not
I
focus
just
on
the
problem
statement.
It's
a
bouquet
wedding
here
is
incorrect,
so
we
can
fix
that
first.
So.
D
As
undoing
something
like
I
know
that,
like
for
update
case,
which
it's
I
think
it's
written
somewhere
below
that
for
updates,
we
can
undo
our
edits,
but
it's
basically
like
there
is
no
such
thing
as
undo
it.
Just
users
will
apply
the
previous
feature
of
the
world.
It
has
a
remembered
somewhere.
There
is
no
such
thing
as
reversing
role,
rolling
back
or
like.
There
is
no
history
of
changes.
Nothing
like
that.
D
There
is
no
chemita
on
transaction
right
and
for
me
at
least
the
statement
is
like
it's
confusing
and
also,
if
you
scroll
down
the
main
reason
why
I
actually
started
commenting
on
this
problem
statement
was
that
probably
it
was
Morgan
who
said
that,
like
it
doesn't
fix
the
problem.
For
me,
it's
like
this
is
not
an
option
and
I
disagree
with
it
main
reason
being
that
I
I
see
the
problem
from
the
different
perspective
right.
D
We
all
understand
that
there
is
a
problem
of
getting
stuck
where
you
can
do
anything
because
broker
won't
allow
their
tuition
and
unities
won't
allow
undoing
the
deletion.
For
me,
I
think
everyone
agrees
with
that,
but
when
we
say
that
like
communities
should
be
able
to
undo
the
update
for
me,
it's
too
too
strict
state
too
strict
requirement,
because
it
already
implies
some
implementation,
details
or
API
change
in
details.
So
I.
B
B
It's
just
if
you
stayed
in
the
current
behavior
of
what
we're
seeing
it's
done.
This
like
proposing
a
solution.
Anything
like
that,
it's
not
if
they
need
to
imply,
we
have
to
support
undo
cuz,
I,
actually,
I
don't
want
to
be
will
undo.
It
was
just
saying
if
we
do
allow
it
to
set
the
timestamp,
then
in
essence,
conceptually
we're
looking
doing
undo,
but
obviously
you
can't
because
that
has
a
whole
bunch
of
negative
ramifications,
so
help
me
reword
that,
then,
if
that
freezing
is
I,
think.
D
It's
again
already
talking
about
implementation.
Details
like
if
cube
does
list,
and
we
need
to
do
that.
It's
already
like
drug
driven
for
me,
it's
like
here's,
the
problem
we
got
stuck
now.
We
have
many
options
and
one
like
I
have
actually
added
zero
option,
which
is
do
nothing
does
document
and
web
people
fix
the
problem
right.
D
It
doesn't
like
answer
this
question
at
all.
How
do
you
do
this
undo?
Update
is
just
like
going
takes
one
system
or
another
system
or
delete
the
object
and
create
a
new
one.
It's
just
like
we
got
stuck
and
we
need
to
get
out
of
this
state
to
healthy
state
back.
How
we
do
this,
why
I'm
doing
redoing
proceeding
with
division?
It's
already
implementation
specific,
might
be.
A
A
A
I
mean
the
deletion
timestamp,
the
focus
on
that
is
driven
by
that's
what
people
watch.
That's
what
sets
everything
else
in
kubernetes
world
into
action.
That's
what
starts
the
finalizer
is
operating,
etc
and
so
forth.
So,
while
we
can
consider
it
implementation
detail,
in
my
opinion,
it's
it's
its
internal
details
that
people
are
watching,
even
though
they
probably
shouldn't
be
so
I,
don't
know
how
they
make
that
balance
and
write
that
down,
but
I.
D
Think,
actually,
the
last
item
in
this
problem
statement
makes
it
clear
that
this
is
the
problem
we
have,
and
this
is
why
we
get
stuck,
and
this
is
why
there
is
no
easy
way
to
get
out
of
it.
Okay,
I
think,
that's
I,
think
it's
strong
enough
to
like
is
basically
give
gives
the
context
right.
This
is
what
we
currently
have
and
we
need
to
change
something
to
to
get
out
of
this
situation.
Okay,.
A
A
A
I
think
some
of
the
some
of
they
just
just
recreated
is
is
problematic
in
terms
of
a
we
have
to
know
what
to
recreate
and
it's
gone
and
be
when
you
delete
some
of
these
things,
they
recursively
delete
everything,
making
it
not
a
recovery
case.
So
much
as
you
know,
backup
and
restore
a
case,
because
there's
not
going
to
be
there's
gonna
it
could
it
could
wipe
out
a
bunch
of
things.
So
that's
that's
that
problem
and
then
you
know
in
terms
of
your
user
view
they
may
be
from
the
brokers
perspective.
B
I,
just
I
just
don't
think
it
solves
a
problem
to
be
honest,
I
think
it
leaves
your
your
Kirby's
model
in
a
state
where
the
things
are
basically
unusable,
and
that
means
that
a
user
who
shouldn't
have
been
allowed
to
perform.
Basically,
this
action
whatsoever
has
now
moved
the
system
into
a
state
where
things
are
basically
busted
and
I.
Don't
want.
D
To
say
that
sorry,
so
when
you
say
that
user
shouldn't
have
been
allowed
to
the
users
request,
has
passed
the
community
authorization
and
has
passed
all
the
validation
has
passed
all
the
admission
controls
right.
So
from
the
community
perspective,
there
is
nothing
that
would
say
that
hey
would
have.
Actually
there
was
some
bug
in
the
system
or
some
breach
which
was
used
by
a
user
right,
but
that's
why
it.
A
D
A
It
wasn't
explicitly
I
was
implicitly
loud,
it
was,
it
was
yes,
the
thing
went
in
the
fact
that
all
the
rules
don't
align
with
the
backend
system.
It's
not
I
might
not
have
access
to
the
rules
of
the
broker.
That's
a
you
know,
you're
not
allowed
to
do
this
and
that
the
other
thing
that's
flow
I
feel
like
the
focus
everybody
keeps
saying.
Is
you
will
just
make
everything
synced
up
all
the
time
is
not
implausible
solution
on
its
face
cream.
If
Ashley.
This
is
not
a
thing
that
we
can
say.
A
C
Do
you
think
it
would
help
us
if
we
kind
of
avoided
your
avoided,
requiring
the
solution
to
solve
the
particular
problem
as
it's
been
worded
and
and
focus
simply
on
answering
questions
about
the
various
options
that
we
have
just
so
we
can
ask
questions
and
get
along,
so
people
can
think
about
stuff
more
without
requiring
that
we
agree
on
the
problem
statement
and
whether
or
not
solutions
solve
the
problem.
Because
I
have
some
questions,
I'd
like
to
ask
and
get
answered,
yeah.
E
I
think
what
seems
to
me
we're
just
going
to
kind
of
we're
all
in
specific
camps,
we're
going
to
argue
it
and
we're
not
going
to
really
achieve
anything
versus
we
all
know.
The
end
goal
is
to
arrive
at
a
community
vote
and
move
forward
with
the
results
from
that.
So
I
think
questions.
Are
we
on
agree
on
at
least
hit
the
questions?
First,.
A
C
Yeah
I
I
had
a
couple
questions.
If
it's
okay,
I
was
hoping
I
just
asked
them
before.
I
have
to
drop
off
the
call
for
I
had
a
suggestion
about
the
the
counter
for
like
a
delete.
Counter
and
I
read
the
responses.
C
The
comments
on
that
and
I'm
still
a
little
confused
and
I
was
just
hoping
to
kind
of
hear
it
worded
a
little
differently
so
that
that
I
can
understand
why
the
suggestion
about
the
delete
counter
wasn't
helpful
was
for
the
first
option,
because
there
was
questions
about
like
delete
requests
and
the
fact
that
it
was
a
boolean
flag
and
like
naturally,
that
brought
the
question
of
like
rolling
back
the
boolean
flag
and
I
was
just
curious,
like
changing
it
from
a
bool
to
a
counter,
regardless
of
the
name.
It
sounds
like
that.
C
D
D
C
Do
you
think
that
it
would
be
helpful
to
update
the
option
to
use
that
instead
of
the
boolean,
so
that
people
don't
have
to
dig
into
that
question
each
time
like
do
we
think
that
that
would
be
a
useful
change
to
the
option?
Just
is
it
when
people
are
voting
on
them,
like
you,
don't
have
like
six
options.
Hopefully
we
have
four.
You
know
so
like
we're
not
voting
between
the
boolean
and
the
counter
or
anything
like
that.
C
A
A
A
A
C
D
C
D
Just
that,
actually,
okay,
with
with
the
option
eight,
you
can
also
once
you
do
provision
in
the
instance.
For
example,
you
can
flip
the
flag
back
and
provision
it
again
without
division
doesn't
make
sense,
so
saying
exists,
external,
it
falls.
Then
circle
goes
and
do
provisions
the
instance.
Then
go
user
goes
back
and
says:
hey
exist,
externally,
true
versions,
again
and
so
on.
You
can
do
that
like
as
many
times
as
you
want
without
ok,
ok,
so
so
it
is
a
different
option.
I
think
it's.
It
has
more
like
more
code.
C
Setting
a
flag,
but
some
people
like
the
two-stage
aspect
of
eighths
and
other
people,
like
the
simplicity
of
one.
We
made
enough
we're
like
eight
and
one
trashed
the
vote,
and
neither
one
comes
out
as
a
clear
winner.
Even
though
maybe
there
was
consensus
about
the
general
approach
so
before
I
go
I,
just
curious,
like
does
this
mean,
maybe
I
should
change
how
we
put
the
phone
up
or
well.
B
I,
if,
if
it's
what
you're
saying
is,
it
sounds
like
that
some
people
may
be
interested
in
some
sort
of
merge
into
more
than
one
of
these
and,
if
so,
I'd
like
to
have
those
discussions
someplace
on
how
much
dr.
sure
the
best
way
to
do
that.
But
if,
as
a
result
of
that,
we
can
put
up
another
proposal
that
crosses
off
two
or
more
of
these
that
are
in
the
doc.
I
think
that'd
be
positive.
Progress
and
I
think
we
should
that
before
book.
D
There
was
also
like
DPR
opened
by
Morgan,
which
is
option
one
I
believe,
and
it
has
many
comments
on
changing
it
to
this
exist
externally.
Implementation
I
believe
because
it
was
mentioned
by
Mike,
Cuba
I,
believe
in
the
issue
in
the
comments.
So
at
least
two
or
three
people
were
complaining
that,
like
you,
should
go
and
fix
it
to
you
to
switch
to
exist
externally
first
and
then
we
can
discuss
it.
So
that's
why
I
did
this
option
cuz?
It
was
some
disagreement
on
how
to
actually
approach
this.
D
So
my
point
is:
whoever
proposes
the
option.
One
should
first
agree
with
others
that
option
eight
is
not
more
preferable
than
the
option
one,
but
I
am
personally
fine
with
merging
two,
because
that's
where
like
just
if
we
choose
either
option
one
or
option
eight
I
believe
there
will
be
people
who
will
like
try
to
flip
this
from
option
up
one
to
eight
or
rightwise
versa,
because
they
are
very
similar
to
each
other
from
the
problem
they're
trying
to
solve.
But
the
implementation
implications
are
different.
B
What's
interesting
is
if
we
could
find
some
way
to
merge
more
of
these
and
say
you
know,
I,
don't
I
know
where
the
Wonder
you
can
merge
those
back,
but
let's
say
for
a
moment
actually
could
be
merged,
but
as
a
result
of
that,
there
may
be
different
variants
or
different
options
within
this
scope
of
that
merge
proposal.
That
might
be
a
way
to
reduce
the
proposals
in
front
of
us.
D
B
B
$0.99,
we
start
if
we
talking
about
merging
in
some
of
these,
if
someone
could
enumerate
sort
of
what
are
the
key
points
of
the
merge
result
and
then
what
are
the?
What
are
the
things
that
are
to
be
determined?
So,
for
example,
if
I
mark
correctly,
the
aspect
of
number
eight
is
to
phase
delete,
I,
don't
think
number
one.
It
was
a
two-phase
delete
and
that's
why
I'm
not
quite
sure
where
they
can
be
merged
or
not,
but
I'd
like
to
understand
those
high-level
points,
in
particular
from
the
users
perspective.
C
D
A
B
C
B
So
let
me
give
you
a
concrete
example:
let's
say
we
take
a
road
and
we
end
up
picking
a
solution
that
some
people
thinks
solves
the
problem,
but
for
other
people,
because
it
doesn't
actually
solve
the
problem
they
were
trying
to
address.
They
may
actually
think
doing
that
solution
is
worse
than
doing
nothing
at
all,
and
so
we've
now
adopted
a
solution
that
no
one
is
happy
with
and
I
think
that's
partially,
because
we're
not
an
agreement
on
really.
B
What
is
what
it
is
we're
trying
to
solve
here
and
some
people
keep
referring
to
things
as
edge
cases
and
to
me
they're
not
edge
cases.
They
are
the
heart
of
the
problem
we're
trying
to
solve,
and
so
that's
different
that
disagreements
there.
It's
something
I'd
like
to
see
bring
resolved.
Maybe
we
can't-
and
it's
just
so
it
just
boils
down
to
bickering
and
that's
fine,
but
at
least
I
like
to
try
to
see
if
we
can
get
over
the
hump
and
make
sure
we're
all
in
same
page
about
what
it
is.
C
Morgan
you're
awesome
I
put
a
suggestion
under
the
problem
statement.
I
did
edit,
it
I
just
put
it
in
as
a
suggestion.
I
don't
know
if
it
helps
or
not
and
if
it
doesn't
help
just
ignore
it.
I
don't
know.
If
this
kind
of
address
is
what
you're
talking
about
at
all
Doug
I
was
trying
to
like
come
up
with
something
that
would
move
our
problem
statement.
A
little
farther
away
from
technical
details
and
I
was
like.
Maybe
it's
helpful
to
just
talk
about
the
end
user
experience
and
not
talk
about
the
internals.
B
C
C
B
C
B
C
There's
all
sorts
of
ways
to
get
to
that
I
was
just
trying
to
place
that
make
it
a
little
shorter,
so
you're
all
cool
with
that.
Maybe
just
click
accept
on
it.
I
don't
know
I'm,
just
hoping.
It
helps
because
I
feel
like
the
bickering
is
something
we've
been
doing
for
months
so,
like
I'm
sure
I'm
like
if
we
meet
tomorrow-
and
we
keep
talking
about
the
problem
statement
like
it'll,
keep
going.
C
Alright
I
have
to
drop
if
we're
meeting
tomorrow,
I'll.
Just
just.
Let
me
know
and
I'm
happy
to
show
up:
okay,
okay,
yeah.
A
A
B
So
I'm
gonna
need
to
shop
in
a
sector
because
I
need
to
run
a
phone
call,
but
I
do
want
to
have
a
phone
call
tomorrow.
If
nothing
else
discuss
things
like,
for
example,
Jay.
You
brought
up
the
question
of.
Why
doesn't
a
a
pre-check
type
of
thing,
a
new
API,
no
service
broker.
Api
saw
the
problem
stuff
and
I'd
like
to
discuss
that,
not
necessarily
to
convince
you
of
why
great
right,
but
but
solely
make
you
understand
why
I
don't
think
that
that's
possible.
Okay,.
B
G
E
A
Sounds
good,
okay
I'll!
Do
that
attempt
to
clean
up
a
one
I
will
attempt
to
resolve
one
in
eight
I'm,
not
sure
how
successful
I
will
be
given
the
divergence
in
implementation
backgrounds,
but
that
just
may
mean
there's
there's
two
separate
two
separate
options
but
I
think
I.
Think
we've
winnowed,
you
windowed
things
down
pretty
good
and
there's
only
three
or
four
or
five
I
guess
real
ones,
which
seems
like
a
better
better
place
to
be
in
then
we
started
so
yeah.