►
From YouTube: Workspace refactoring discussion
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).
C
B
Just
go
ahead,
but
Safari
It's
tricky
to
do
that.
So
give
me
a
second.
A
C
B
C
B
B
Okay,
so,
but
independently
of
the
Sinker
and
TMC,
so
we
have
name
and
path,
and
the
most
important
rule
which
is
different
is
indexes
and
inform
Us
in
general
is
the
zq
is
always
a
name
there's
nothing
about
paths
by
default.
B
So
every
time
you
you
want
to
use
a
singer,
you
must
use
names
as
index
in
the
index
to
get
an
object
client
to
purpose
what
technically
support
they
support
the
path,
but
you
can
transform
a
name
to
a
past
just
with
the
past
particle,
and
there
are
some
cases.
Some
exceptions
and
TMC
is
also
affected
or
uses
that.
So
we
have
one
indexer
which
uses
an
annotation
which
is
optional.
It's
called
kcp
def
pass
it's
not
on
all
objects,
it's
just
on
those
which
need
to
reference
bypass
and
we
have
a.
B
So
if
you
need
to
look
for
objects
in
the
indexer
in
the
list
or
in
the
indexer
correctly,
because
the
list
of
course
doesn't
give
you
that,
but
there's
one
index
in
those
objects
which
need
it
where
you
can
look
up
by
path,
and
this
includes
API
exports.
Today
it
includes
cluster
workspace
types,
which
are
called
workspace
types
now
with
auto
cluster,
and
it
includes
the
location
if
I'm,
not
one,
maybe
I
can
start
sharing,
so
I'm,
not
sure
how
stable
this
is.
A
It
works
for
me,
but
I
have
you
on
a
big
screen.
So,
okay.
C
A
D
B
B
And
but
you
can,
every
everybody
who
creates
objects
can
add
The
annotation
and
it
automatically
filled
in
like
it's
a
key
empty
value
and
the
value
set.
B
That's
great
and
I
mean
you
will
notice,
like
exports
are
needed
because
binding
is
kind
of
reference.
Bypass
locations
are
needed
because
placements
have
a
reference
two
locations.
B
B
Yes
and
build
space
type,
everything
for
bookspace
type
is
by
path
like
in
in
the
workspace
it's
by
path
in
the
cluster
workspace
types
themselves,
so
it's
also
bypassed
and
with
those
three
we
get
to
Android,
so
our
apis
are
not
using
pass
so
much
so
very
good
anyway.
So
this
is
automatic.
So
just
add
the
sanitation,
annotation
kcp
Dev
pass
and
get
what
you
want.
D
B
I
see
what
else
is
important,
so
Andy.
If
you
have
questions
you
look
through
the
pr
just
yeah.
A
B
Yeah
we
have
the
big
separation
of
apis,
so
we
have
the
core
API
Group
and
in
that
I
put
the
Sharks
as
a
format.
Cluster
workspace
chart
has
been
moved
into
core
and
the
second
one,
obviously
is
a
logical
cluster,
and
this
is
the
biggest
change
of
the
whole
bunch.
Basically,
it
changes
the
data
structure
of
our
data
model,
so
in
every
logic
cluster,
which
I
mean
in
the
sense
of
a
density
partition
which
forms
a
cluster.
We
have
a
logic
cluster
with
a
Capital
Area
Capital
C
object.
B
So
this
object
is
created.
It's
visible
to
at
least
two
developers
like
controller
authors.
Where
will
know
this
object
because
it
shows
up
in
Virtual
workspaces,
for
example,
so
normally
user
will
not
use
that
like
they
will
use
the
old
workspace
object
so
see
I
plug
in,
for
example,
users.
It's
a
workspace
object
as
before,
but
every
time
you
create
a
workspace,
we
have
the
workspace
controller,
which
is
in
the
reconciler,
but.
B
Every
time
you
create
a
workspace,
this
code
will
choose
a
chart
and
create
I
mean
it
creates
a
name
which
is
just
random
here:
a
random
16
character
base,
36
object,
name,
not
object,
and
the
conflict
is
pretty
low
that
you
get
the
same
again
and
it
will
create
blindly
logical
cluster
objects
on
that
chart
with
that
name
not
with
his
name
Panda
that
Dodger
cluster,
the
name
is
always
cluster.
B
B
And
then
it
doesn't
airbag
also
has
a
check
that
the
user
has
access
web
access
to
the
Slash
non-resource,
and
if
this
is
the
case,
the
request
is
permitted.
Are
we
not
committed,
but
that's
the
individual
check.
Of
course
there
are
more
checks
like
the
normal,
airbag
delegate,
also
either,
which
is
called
here.
No,
not
here,
there's
a
Code,
your
problem,
yeah
yeah.
B
With
it
so
again,
this
one
is
the
most
important
reconciler.
Here's
a
function
which
creates
the
logic
cluster
object.
B
A
B
Api
taste
and
yeah
so
anyway.
So
this
one
points
to
workspace
and
basically
it
tells
it's
a
logical
cluster
reconcile
which
also
exists
and
workspace
deletion.
Everything
that
this
logic
clusters
owned
by
that
workspace
object.
So
the
lifecycle
is
connected
and
they
are
bi-directional
finalizers
to
make
sure
their
life
cycle
matches,
but
very
important.
Those
objects
like
the
workspace
entire
cluster
can
be
on
different
charts
or
different
regions
or
whatever
that's
why
it's
a
little
tricky
to
like
yeah.
B
Maybe
I'll
show
you
that
if
you
go
here
to
the
reconciler,
they
will
find.
B
There's
a
special
client,
a
client
which
is
not
a
look
at
client,
but
it's
yeah
this
one
here.
It's
a
logical
cluster
admin,
client
and
this
client
uses
a
user
which
is
restricted.
It's
not
system
master,
so
it's
a
restricted
user
which
can
create
logical
clusters
and
personalization
will
skip
the
check.
I
showed
before,
like
you,
can
trade
luxury
cluster
object
without
the
logical
cluster
object
existing
right?
So
you
need
at
least
one
client
which
can
do
that,
and
this
is
a
client
and
this
will.
B
B
B
Let
me
check
what
this
is.
Oh,
this
is
our
old
user
owner,
so
the
user
would
just
use
my
initialization.
What
else
do
we
have?
We
annotate
the
the
type
here.
B
And
we
put
the
past,
which
we
have
seen
before
so
basically,
this
past
annotation
is
maintained
by
copying
it
from
The
Logical
cluster
object
down
to
the
sub
logic,
cluster
object.
So
every
of
those
objects,
including
the
API
exports
I,
mean
the
logic.
Cluster
object
has
basically
the
source
of
tools
for
the
path
and
when
an
API
export
or
a
location
or
what
was
the
third
one
workspace
type
when
those
are
created.
Submission
takes
the
value
from
here
from
the
logic
from
The
Logical
cluster,
the
past
annotation.
B
A
I
have
a
question
about.
We
talked
about
how
kcp
core
in
the
future
would
just
have
logical
clusters
and
not
paths
and
that's
a.
D
B
This
is
is
a
higher
level
implementation,
so
in
a
core
project
which
doesn't
use
tenancy,
like
all
our
workspace
stuff,
there
is
a
path
as
a
type,
but
it's
always
just
just
cluster
name.
A
B
B
Yeah,
that's
not
your
cluster
object.
I
think
we
see.
B
So,
division
of
workspaces,
so
I
hope
I
find
the
right
to
laces
here
here,
probably
so,
when
you
create
personal,
when
you
delete
the
workspace
pathological
cluster,
but
the
user-facing
workspace
object,
this
code
will
run.
B
So
what
it
does,
it
will
also
create
delete,
The
Logical
cluster
object
like
when
the
deletion
timestamp
is
set
on
the
workspace.
This
is
propagated
down
and
it
again
I
think
uses
this
logic
class
document
client.
B
B
Izer,
so
if
you
go
into
the
cluster
workspace
deletion
codes,
what
does
the
workspace
logical
cluster
relation
code
is
called
now
it's
here.
This
has
finalizing
method
like
when
everything
is
deleted.
B
B
And
the
cluster
hold
bindings
and
holds
are
deleted
and
at
the
very
end
before
actually
before
the
luxury
cluster
is
deleted,
we
did
its
owner,
probably
removes
the
finalizer
from
someone,
so
it
says
no,
no,
it's
not
your
cluster
like
this
course
workspace
or
cross-cluster
reference.
We
looked
at
before
the
finalizer
STD
comes
out
again.
This
can
be
cross
region.
B
Or
maybe
it's
the
other
way
around
I,
don't
remember
what
what
I
implemented
at
the
end
could
be
that
the
workspace
I
think
it
checks
the
technology
cluster
once
in
a
while
there's
a
logic
which
does
that,
and
maybe
that's
what
we
have
seen
before.
Actually.
C
B
We
don't,
but
we
could
have
some
okay,
it's
not
implemented.
If
we
have
some,
we
would
likely
either
we
wouldn't
put
the
owner
or
we
would
have
another
kind
of
ownership
which
is
not
linking
the
life
cycle.
We
can
think
about
that.
But
it's
implementing
I
mean
it's
feasible
to
implement.
That's
not
done.
B
So
all
the
the
controllers,
like
garbage
collection
and
quota,
which
used
to
watch
workspaces.
Of
course
they
now
watch
as
well
share
cluster
and
the
change
is
pretty,
was
pretty
mechanical.
B
Here
and
get
location,
I
hope
this
is
an
example.
So
again
we
have
name
and
parts
right
here.
In
this
case,
I
chose
New
Paths,
because
selected
location
is
a
possible,
so
I
need
to
get
location
function.
This
shouldn't
be
a
client
right.
We
never
want
client
access
and
controllers,
so
it
must
be
based
on
some
indexer
or
at
least
on
some
informal
and
this
function,
and
you
see,
if
you
look
in
into
those
Getters
and
in
general,
those
abstractions
for
the
client
calls
or
informal
calls
most
of
them
use
cluster
name.
B
Some
of
them
use
the
path,
and
yes,
actually,
this
must
be
powers,
of
course,
and
so
I
default
to
name
but
very
few,
and
this
is
connected,
of
course,
to
this
orientation.
This
kcp
Dev
pass
annotation,
they
need
pass,
get
location
is
such
case,
and
let
me
see
yeah
and
this
code,
you
see
in
a
couple
of
places.
A
B
B
E
To
see
A,
New,
Path,
I,
guess
I,
don't
know
wherever
you
were
before.
Oh.
B
Okay,
you
mean
this
one.
Yes,
so
we
haven't
yet
changed
those
past
fields
to
use
logical
as
the
paths.
Okay,
reason.
D
B
We
have
a
private
fields,
we
have
a
private
field
called
value
and
for
reasons
we
want
it
for
reasons,
different
discussion,
and
if
you
replace
it
here
with
luxury
cluster
path,
then
what
was
it
deep,
equal
or
something
breaks.
A
Yeah
and
we
actually
like,
if
we
Port
the
there-
was
a
change
that
went
into
go
I,
don't
know
sometime
in
the
past
couple
of
years.
That
would
fix
that
issue
for
us
without
having
to
make
any
changes.
So
we
could
update
the
forked
copy
of
D
deep,
equal
in
kubernetes
and
not
have
any
problems.
B
I
guess
we
have
to
ask
for
that
which
are
open
but
unfinished.
So
it's
not
super
critical.
It's
just
Cosmetics
it's
nice
to
have.
If
you
can
do
that,
maybe
you
could
extremely.
B
Wrong
yeah,
yes,
and
we
have-
and
we
have
this
pattern
here
like
15
times.
Of
course,
we
can
have
at
least
a
type
which
is
an
yes
of
stringing
and
then
get
rid
of
that.
That's
work
to
be
done
so
PR's
welcome
yeah.
What
else
is
interesting.
B
I
mean
authorization,
obviously
there's
more
to
that.
So
I
already
went
through
the
content
authorizer,
which
checks
it's
not
your
cluster
existence,
the
top
level
authorizer
is
gone.
It's
replaced
by
something
called
the
required
groups,
also
either
I
think
Andy.
We
talked
about
that
in
Detroit,
and
so
you
can
put
an
annotation
I'm
not
from
this
one
here
on
a
logical
cluster
objects
and
whenever
you
create
a
sub
workspace,
a
sub
logical
cluster.
This
one
is
copied
over.
B
So
if
you
change
the
root
logic,
cluster,
there's
not
copy
open,
like
there's
no
update,
but
if
you
go
to
that,
if
you
want
to
something
which
I
mentioned
implicitly,
if
I
didn't
call
out
the
root
workspace,
the
only
thing
which
is
special
about
this
is
that
its
name
is
equal,
so
the
cluster
name
of
hoot
is
equal
to
the
to
the
workspace
path
like
both
food,
but
that's
basically
the
only
thing
which
is
special
about
that.
Otherwise,
the
woodwork
space
is
like
anything
else
and.
B
But
basically
we
can
have
arbitrary,
many
yeah,
basically
the
roots
in
the
forest,
so
many
hierarchies
many
trees
in
the
forest
do.
A
B
Yeah,
so
if
we-
and
this
is
I-
think
into
your
favorites
a
homework
spaces,
everybody
who
remembers
that
code,
it's
mostly
gone
so
the
Handler
here
after
checking
it's
the
request
which
is
against
Foods.
It
looks
up
so
it
derives
the
home
cluster
name
from
the
username
via
some
hashing.
So
it's
deterministic.
B
D
B
Moment,
there's
no
paths,
no
kcp
Dev
path
in
in
private
chat,
I
sent
and
the
an
idea
for
a
task
for
for
business
and
and
then
like.
B
B
But
at
the
moment
it's
just
chords,
This,
Is,
Random,
16
character,
base,
36
value.
B
Yeah
and
does
this
yeah
so
who
Tome
is
a
workspace
type?
Let
me
find
that
so
the
workspace
type
concept
is
part
of
tendency.
It
won't
be
part
of
core.
B
B
Let's
see
it,
oh
it's
in
settings
here
the
face
is
also
part
of
logical
cluster,
so
the
phases
scheduling,
initializing,
ready
and
initializers
must
be
empty
until
it
goes
from
initializing
too
ready.
So
all
this
this
logic
is
part
of
core,
but
it's
actually
controller
which
does
initialization
and
which
does
so
virtual
workspace,
which
makes
this
thing
visible
to
to
users
or
to
develop
us.
That's
part
of,
and
also
tenancy
extension.
B
Not
sure
what
you
mean
there
was
an
annotation.
Do
you
mean
that.
B
B
Yeah,
it
cannot,
because
this
is
I
mean
it
wasn't
annotation
before
it's
a
first
season.
The
second
reason:
it's
in
tenancy,
it's
a
tenancy
concept.
Okay,.
E
B
E
Was
expecting
to
I
was
expecting
to
expose
tendency,
intent
through
objects
and
not
annotations.
C
B
A
B
B
If
this
workspace
is
created
in
a
in
a
home
workspace
or
in
a
woodwork
space,
there
is
no
type
because
there
is
no
workspace
right,
there's
no
parent
anymore,
but
we
still
need
to
know
the
type.
That's
why
we
put
it
not
on
the
parent,
but
we
copy
the
type
basically
from
parents
onto
the
luxury
cluster
as
an
annotation
or
in
the
case
of
the
homework
space.
B
A
B
Yes
and
no
I
mean
in
the
into
oid
code,
we
had
this
packet
hierarchy
yeah
and
we
could
look
into
the
bucket
workspace.
The
buff
yeah
look
on
the
workspace
object
for
our
home,
find
the
type
this
doesn't
exist
anymore
and
what
you
say:
Andy
yeah
now
there's
nothing
about
us,
so
we
have
to
look
somewhere
locally.
B
E
B
B
E
Well,
I:
guess:
what
are
the
other
places
where
the
tenancy
code
has
to
be
privileged.
E
B
I
mean-
and
we
just
have
to
do
that,
like
you
notice,
when
you
split
this
this
kind
of
code,
it's
like
a
big
spaghetti
where
everything
is
connected
to
everything
and
you
it's
at
some
point.
You
notice,
when
something
just
doesn't
work
like
the
past
concept.
We
try
to
keep
it
out
of
core,
but
it
didn't
work
out.
So
now
we
have
the
past
concept,
the
abstract
one
in
core,
but
the
permutation
outside
these
kind
of
compromises
we
have
to
do,
and
here
I'm,
not
I,
don't
know
okay,
sure
place
for
exploration.
B
B
B
B
So
this
is
a
place
where
we
can
start
to
split
who
are
non-core
stuff-
oh
yeah,
possession
that
might
be
interesting.
Oh.
B
B
D
A
B
B
Yeah,
oh
yeah,
maybe
something
about
virtual
workspaces
in
general.
You
know
they
have
this
so
formally
they
had
a
what's
the
name
API
domain
key
or
something
like
this
part
of
the
URL.
B
This
is
never
a
pass.
It's
always
the
cluster
name.
B
And
maybe
yeah.
This
is
also
interesting,
so
your
team
found
out
today
is
at
the
CLI,
was
broken
and
just
want
to
briefly
explain
so.
Basically,
this
case
was
broken,
and
so
is
it
yeah.
So
the
URL,
which
is
in
the
workspace
object
like
in
the
status
URL.
Everybody
knows
that
this
was
used
by
the
CLI
just
copied
into
the
group
config.
B
So
we
don't
take,
surpass
directly
and
that's
why
what
we're
doing
here
and
actually
I'm
not
this
is
a
swan.
It
doesn't
hurt,
but
it's
wrong
so
before
here
was
UL
magic
and
this
broke.
So
now
we
just
take
the
workspace
name,
which
was
just
created
in
this
case.
It's
a
create
function
here
so
CI
and
this
is
passed
to
the
use
command
and
that's
why
it
works
and
when
we
do
dot
dot,
we
remove
one
element
from
the
path
and
yeah:
that's
how
it
works.
B
B
Name:
go
yeah
yeah,
oh
it's
Pascal!
Actually,
so
Steve
had
added
this
comment
and
Lucas
took
it
over
into
his
PR.
So
if
I,
if
you
have
a
hierarchy
like
that
here
at
the
end,
there's
a
cluster
names,
those
numbers,
any
realities
are
a
bit
longer
because
you
want
some
to
be
unique.
B
But
in
any
case
this
is
a
hierarchy.
Example,
and
you
have
many
valid
paths
right
so
for
the
invoice
here
for
this
one.
It's
the
longest
pass,
because
the
canonical
nickel
tops
of
this
workspace
oops,
but
you
can
also
go
to
Just
management
and
go
for
management
into
the
sub
workspace
and
well
maybe
it's
here
from
accounting
to
the
subject
space,
and
maybe
we
have
examples
here,
yeah
that
one.
This
is
also
a
valid
path,
so
for
one
proxy
can
translate
that.
B
But
it's
not
canonical
it's
not
something
the
system
understands
it,
but
you
cannot
reference
it,
for
example,
because
our
indexes
won't
have
it
and
it's
not
something
the
user
will
use.
B
D
A
A
B
Yes
or
no,
we
do
so
again
Andy's
and
Ben's
task.
The
idea
is
to
to
give
user
homework
spaces,
something
like
an
IAS
path
like
there
should
be
user
prolong
ND
instead
of
some
base,
36
string,
which
we
would
put
that
we
would
put
that
on.
The
kcp
dev
pass
annotation
of
the
homework
space
of
the
Home
Logic
cluster
right
and
in
that
moment
maybe
the
longest
path
is
actually
taking
this
past
annotation
space.
E
Can
we
quickly
mention
so
I
guess
when
you
said
that
there
were
no
aliases
I
took
that
to
mean
it's
a
one-to-one
relationship
between
workspaces
and
logical
clusters?
Is
that
not
true
foreign.
B
B
Zero
yeah
there's
no
one
to
end
I
think
we
briefly
chatted
about
that.
I,
don't
think
that
soon
link
will
influence
the
phone
proxy
and
we
set
what
the
valid
path
is.
I
think
our
assembling
will
just
be
something
of
fun
and
will
understand
like
the
CLI
will
understand,
but
at
the
end
you
must
reference.go
clusters
and
workspace
names.
A
B
Yeah,
but
what
I'm
talking
about
is
something
completely
different:
it's
not
about
Isaac,
it's
about
giving
a
parentless
a
logical
cluster,
the
canonical
path,
which
is
a
cluster
name,
there's
two
different
things.
Yes,
it
was
also
different.
I.
Think
that's
why
Canada
passes
it's
not
just
the
longest
past,
which
starts
which
it
was
a
logical
cluster
name,
it's
a
little
bit
different.
B
If
we
do
that,
okay
anyway,
so
let
me
finally
go
to
the
front
part
scene,
and
maybe
I
start
here
so
I
split
up
the
phone
proxy
code
into
the
actual
command
and
the
index,
and
it's
still
very
naively
implemented
it's
implemented
by
our
inform
us
against
all
shots.
We
will
change
that,
of
course,
eventually.
C
B
What
is
important
this
thing?
It
watches
logical
clusters,
it
watches
workspaces
and
charts,
and
by
that
you
can
reconstruct
it's
a
hierarchy
right,
you
can
imagine.
Every
cluster
is
a
who
would
of
some
kind
of
of
a
tree
in
the
forest,
and
then
you
can
append
the
workspaces
which
are
just
name
segments
in
the
past,
and
by
that
you
can
reconstruct
the
whole
hierarchy,
and
this
is
what
is
done
here.
B
I
won't
go
into
the
exact
details,
basically
those
few
Maps
here
they
give
you
the
data
to
to
Traverse
the
hierarchy
and
they
find
everything
we
need
like.
We
can
find
the
longest
path
in
hierarchy
by
that
we
can
get.
The
longest
is
electronic
deposit.
If
you
want
to
and
again
if
you
want
to
name
the
homework
space
versus
pass,
which
is
not
the
cluster
name,
you
would
have
to
add
that
here,
so
the
font
proxy
has
to
know
that
there's
this
concept,
this
index
is
used
in
two
places.
B
It's
used
in
the
front
proxy.
So
if
we
go
here
just
go
to
the
yeah
yeah,
that's
our
phone
box
implementation,
especially
the
old
code,
just
I
accepted
the
index
logic.
Out
of
that
mostly
unchanged-
and
here
you
see
also
the
informers
I
talked
about-
and
the
second
place
is
something
I
call
a
local
proxy.
B
This
is
basically
the
case
when
you
run
kcp
start
there's
no
fault
proxy,
so
we
have
to
have
this
logic
in
the
kcp
instance
in
the
shot.
B
If
you
are
running
just
one
chart
the
information
of
this
index
here,
the
local
proxy
is
complete
and
we
set
the
local
proxy
can
do
everything
we
need.
It's
always
resolving
every
every
pass
which
exists
in
the
system.
So
it's
replacing.
It's
inform
proxy.
B
A
I
think
this
is
a
good
intro.
I
had
made
it
through
I'd,
say
somewhere
between
a
third
and
a
half
of
all
the
files
on
Friday
which
I'm
gonna
have
to
go
and
retract
in
the
new
PR
yeah
like
the
index
code,
was
something
that
I
had
skipped
over
because
I
it
was
just
late
and
I
didn't
want
to
try
and
use
too
much
brain
power
on
Friday
late
afternoon,
but
yeah.
It
makes
sense.
So
I'll
keep
looking
at
it
today
and
tomorrow.
B
How
much
do
I
have
something
I
forgot,
yeah
I
mean
you
will
notice
when
reviewing
a
number
of
things
I
had
to
speed
up
so
the
former
cluster
workspace
reconciler
is
now
part
of
the
logic.
Cluster
reconcile
are
part
of
the
workspace
sequence
Island,
the
same
thing
you're,
seeing
here
for
admission,
maybe
of
a
member
of
our
place,
transitional
restrictions
that
have
been
in
the
cluster
workspace
submission
before
same
thing
for
a
few
other
things
here,
you
see
the
initializing
checks
about
yeah.
B
Spec
so
there's
something
called
directly
deletable,
so
imagine
you,
you
create
a
new
tree
in
the
forest,
so
the
prime
example
is
an
organization
a
tenant.
We
won't
put
that
Underwood
anymore.
We
will
have
another
three
in
the
followers
forehead
for
Deutsche
Bank
for
whatever
and
then
the
question
is,
how
can
you
delete
as
a
user?
B
This
organization
and
one
idea
was
okay:
I
can
create
I
can
delete
The,
Logical
cluster
object
that
should
be,
and
if
I
do
that
and
the
whole
hierarchy
goes
away,
and
this
is
the
idea
that
you
can
do
that.
Of
course,
there's
a
permission
question
around
that,
but
this
use
case
will
come
up
in
some
way.
B
B
Initially,
yes,
yes,
if
you
want
to
do
it
that
way,
I
mean
you
can
yeah
something
some
other
API
to
to
trigger
delete.
If
that
work,
the
creation
of
an
organization
is
something
which
is
not
implemented
in
kcp.
This
would
be
something
we
would
do
externally,
basically
similar
to
The
Home
Group
space
example.
That's
a
similar
Direction,
but
in
our
implementation
we
would
create
an
organization
by
creating
the
logic
cluster
and
you
would
make
use
of
this
required
group
authorization.
So
you
would
add
some
some
tenant
group
as
a
prereq
to
enter
key.
B
B
B
A
D
B
B
E
D
B
D
B
It's
a
pause,
it's
not
it's,
not
the
name.
That's
a
point,
and
cluster
name
is
a
cluster
Name
by
type.
E
E
I,
don't
know
if
you
guys
did
this,
but
in
the
generated
client
stuff,
there's
panics
everywhere.
If
you
try
to
scope
a
list
or
scope
a
client
to
a
particular
cluster
and
you
pass
it
to
wildcard,
because
the
previous
type
definitions
we
had
allowed
that.
So
that's
a
good
change.
We
can
just
delete
all
that
Panic
code.
B
A
The
only
place
is
like
this:
this
method
is
very
deterministic
about
when
wildcards
are
allowed
versus
not
and
there's
only
two
methods
to
get
for
identity,
wild
card
and
get
for
wild
card
partial
metadata
that
allow
wild
cards.
B
B
Okay,
maybe
take
a
look.
I,
don't
know
the
context
here,
but
this
is
it's:
it's
a
wrong
value.
The
star
is
not
allowed
for
name.
A
A
E
I'm,
even
looking
that's
a
line
line,
69.
yeah,
so
the
like
the
equivalent
implementation
of
this
method
for
all
of
the
generated
listeners
will
Panic.
If
name
is
wild
card.
Oh.
A
B
Okay,
then,
then
all
objects,
then
that's
the
type
breakage
at
the
moment
I
mean
it
works
because
we
don't
validate.
A
B
B
Yeah,
that's
good
I!
Think
oh
yeah,
maybe
this
one!
So
there's
always
another
thing.
If
you
look
on
I,
don't
know
which
one.
B
So
it's
a
moment
what
I'm
doing
when
you
create
a
workspace
fixture.
Whatever
turn
is
a
cluster
name,
so
it's
a
low
level
cluster
name.
If
you
want
to
the
past,
you
have
to
take
the
second
function
here.
I
didn't
come
up
with
the
better
names
and
object,
and
then
you
can
get
the
cluster
and
the
path.
B
A
Not
super
happy
with
this
hectoring,
the
the
cluster
field
in
status.
What
was
that.
B
The
workspace
schedule
now
is
a
reconcile
that
I
showed
earlier.
That
puts
the
cluster
name
into
the
workspace.status.
B
B
B
A
This
is
the
if
I
create
a
workspace
called
Foo,
it
is
backed
by
a
logical
cluster
called
a736
whatever
and
that's
what's
stored
in
the
status.
D
A
Backup
or
store
won't,
but
you
know
Bolero
backup
and
restore
wood,
so
yeah.
If,
if
there's
something
that
needs
to
survive,
non-ncd,
backup
and
restore,
it's
got
to
be
inspect.
A
B
I
was
just
confirming
I
mean
it's
a
fun
proxy
will
know
it,
but
there's
local
information
would
not.
D
A
Yeah
I
think
after
we
break
I,
will
look
at
the
API
extensions
code
and
make
sure
that
we've
got
the
wild
cards.
Working
right
and
I
can
move
that
to
spec
the
cluster
field
to
spec
and
I'll.
Probably
do
another
commit
that'll
collapse
down
all
the
indexer
by
cluster
name
and
path
usage
to
a
helper.
A
B
A
Honestly,
if
the,
if
the
tests
are
solidly
green,
I'm,
fine
merging
any
time
we
want
to,
if
they're,
flaky,
I'd
rather
not
but
I
mean
most
of
us
are
either
out
for
the
rest
of
the
year
or
almost
out
for
the
rest
of
the
year.
So
you
know
if
we
merge
it
today
or
we
wait
until
January,
3rd
or
4th
I,
don't
think
it
makes
much
of
a
difference.
B
Yeah,
but
basically
much
early,
luckily,
based
in
the
past,
like
those
who
remember
that,
there's
no
perfect
on
the
first
try,
then
you
will
I
mean
there
are
always
small
things.
You
try
to
improve
friends.
D
B
A
All
right,
well,
I,
think
the
the
review
principle
should
be
if
something
is
really
bad
to
merge.
Let's
fix
it.
Otherwise,
let's
just
collect
a
bunch
of
follow-ups,
yeah.
B
B
And
I
haven't
done
this
kcpio
change
right
intentionally
because
it
makes
you
know
even
harder.
I
think.