►
From YouTube: Community Meeting, July 5, 2022
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
Hey
everybody
today
is
july
5th.
This
is
the
kcp
community
meeting,
I'm
going
to
go
ahead
and
paste
the
link
to
our
github
issue
for
today's
meeting
into
chat
and
if
you
have
any
items
you'd
like
to
add
to
the
agenda,
please
add
comments
to
this
issue
and
we
will
get
started
so
we'll
leave
the
incoming
issues
and
milestone
review
to
the
end
of
the
meeting
assuming
time
permits
and
go
into
the
first
one
from
stefan
about
chojan's
placement
pr.
So
stefan
you
want
to
talk
about
that.
A
little
bit.
B
B
It
has
a
namespace
selector,
so
you
can
choose
which
namespaces
it
covers.
You
can
have
multiple
placement
objects.
Each
can
select
a
location,
a
different
one.
Each
location
gets
a
cluster,
which
means
you
can
have
multiple
clusters
now
per
name
space.
So
everybody
who
is
working
on
those
multi-cluster
topics.
This
is
a
way
to
to
to
get
them
and
yeah.
This
default
placement
is
created,
but
you
can
delete
it,
it's
not
recreated,
so
it
will
remember
when
it
was
there
and
delete
it
and
there's
also
some
label
annotation.
B
I
forgot
described
in
the
api
to
avoid
the
default
placement.
So
if
you
have
githubs
or
anything
like
that,
and
you
don't
like
a
default
placement,
but
you
want
another
one
different
name:
that's
also
possible
so
play
with
it.
It
merged
it
has
entered
end
test.
Obviously,
but
beyond
that,
it's
not
much
tested,
I
guess,
but
it's
very
worse
to
explore.
A
Do
you,
you
know,
if
he's
planning
on
doing
any
sort
of
demo
to
show
it
off?
We
haven't.
A
C
D
D
But
I
still
credit
that
as
a
distinct
issue,
because
possibly
according
to
you
know,
timeline
and
deadlines,
we
might
want
to
accept
having
the
homework
spaces
implemented
and
then,
possibly,
you
know
defer
the
the
the
cleanup
of
of
this
code,
which
is
mainly
you
know
we
we
can
just
not
use
it
and
it
can
still
be
there
and
then
clean
clean
it
up
afterwards.
So
I
mean
it's
just
a
choice
we
can
do
either
to
include
that
in
the
current
milestone
or
to
put.
A
I'll
put
it
in
tbd
for
now
yeah.
Okay,
thank
you!
You
have
the
next
one
as
well
about
default.
Permissions.
D
Yes,
that's
a
bit
a
bit
quite
the
same,
an
outcome
of
the
homeworks
whole
personal
workspaces.
As
soon
as
we
have
my
you
know,
decide
to
migrate
and
have
everything
created
there
and
no
all
workspaces
anymore.
Then
we
have
to
think
about
adding
default
permissions.
So.
A
Thank
you.
So
we
had
the.
I
know
there
was
some
back
and
forth
on
cluster
workspace
type
name
and
capitalization
and
not
stefan
did
we
decide.
We
were
just
going
to
go
with
lowercase.
A
Let's
see
the
cluster
workspace.
Type's
name
is
expected
to
be
lowercase.
The
reference
to
it
is
expected
to
be
uppercase,
and
that
seemed
potentially
a
little
strange
that
there
is
a
mismatch
between
the
capitalization
of
the
name
of
the
type
and
how
you
refer
to
it.
I
know.
Stefan,
you
had
said
that
you
initially
were
trying
to
match
what
kubernetes
does
with
things
like.
A
I
don't
know
what
was
it
image
pull
policy
and
other
things
where
the
any
kind
of
constants
yeah
yeah
the
constants
are
expected
to
be
uppercase
for
the
first
letter
I
had
said
in
slack
last
week,
maybe
that
I
preferred
a
match.
So
if
the
type
name
is
lowercase,
the
reference
to
it
should
be
lowercase
as
well.
So
I
think
that's
what
we're
trying
to
decide
is
which
way
we
should
do
it.
B
And
I
I
commented
that
I
don't
want
cat
or
camera
case,
but
not
enforce
it
like.
If
you
have,
I
don't
know
continuous
category
also,
the
d
should
be
capital
as
well,
but
I
don't
want
that
users
should
write
ken
writes.
E
Are
we
talking
about
names
that
appear
in
go
source
or
in.
B
E
B
E
A
Yeah,
so
cluster
workspace,
the
api
type
has
spec
dot
dot.
Okay
name
is
a
the
the
name
field
contains
a
string
and
it
is
a
reference
to
a
type.
The
type
is
a
separate
resource
instance.
It's
a
cluster
workspace
type.
It
has
a
name,
and
so
the
discussion
here
is
when
you
set
spec
type
name,
should
it
be
lower?
Should
the
the
casing
match
the
casing
of
the
name
of
the
cluster
workspace
type
or
should.
A
G
What
we
think
and
so
I'd
say
almost
like
if,
if
there's
good
ux
for
like
which
types
can
I
choose
whatever
that
outputs,
you
know
if
we
can
make
that
consistent
with
the
way?
That
is
the
case.
That
should
be
fine.
E
G
The
user
does
not
necessarily
have
an
act
like
our
back
to
read
their
workspace
object
from
the
parent,
nor
do
they
have
necessarily
permission
to
read
the
cluster
workspace
type
that
their
workspace
object.
References
even
if
they
did,
though
the
allowed
children
would
be
references
and
would
have
uppercase
names,
and
if
they
were
to
do
that,
they
would
see
you
know
like
a
self-consistent
list
there
right.
E
E
G
We
need
one
anyway,
and
the
references
in
their
parent
type
are
going
to
be
uppercased.
You
know
if
references
are
uppercase,
then
the
references
there
will
also
be
uppercased.
I
mean
in
general.
I
agree,
I
think,
the
number
of
times
I've
messed
this
up
has
driven
me
absolutely
insane.
It
would
be
much
simpler
if
they
were
all
lowercase,
but
all
of
that
being
said,
we
need
a
ux
regardless,
because
users
are
not
going
to
have
that
privilege.
So.
B
D
Well,
I
mean
upper
case
with
only
one
upper
you
know
case.
One
character,
uppercase
is
just
I
mean,
doesn't
bring
anything
being
able
to
really
completely
camel
case,
as
as
the
user
wants
just
gives
more
naming
power,
I
mean
or
naming
flexibility
for
the
user
when
you
want
things
like
my
organization
that
does
this
and
that
and
then
it's
quite
cool
to
be
able
to
communicate
that.
But
maybe
it's
just
you
know
a
corner
case
that
we
don't
want
to
allow,
because
having
me.
D
D
User
wants
to
if
a
user
wants
to
to
create
its
own,
his
own
type.
He
might
you
know
like
being
able
to
camel
case
the
name
he's
giving
I
mean
for
the
the
externally
visible
type.
I
mean
I
mean
from
his
point
of
view,
not
the
the
cube
object
itself.
D
Yeah
but
but
on
the
other
hand,
if
you
can
put
hyphens
inside
the
name,
you
know,
then
it's
just
a
question
of
preferences
about
the
case.
A
A
Okay,
I'm
going
to
put
this
in
tbd.
I
A
Next
up,
this
was
just
the
reminder
that
we
shouldn't
have
admission
plugins,
look
at
versions.
I
A
A
A
A
It's
to
use
a
non-existent
workspace.
I
haven't
tried
to
validate
or
reproduce
this,
but
assuming
it
it
is
valid.
It's
not
good.
B
A
B
We
don't
authorize
first
accessories
exactly
but
know
you
in
the
future
in
the
very
near
future.
Everybody
in
front
of
the
front
proxy
cannot
be
system
master
anymore,
which
means
you
must
really
be
on
the
sharp
level
to
exploit
that,
and
it's
intentional
that
we
support
it
like
we
have
use
cases
like
shadow
cds,
for
example,.
I
I
A
Cube
control
version
produces
an
error.
I
thought
we
oh
right,
so
this
one
was
saying
like
I
just
I
followed
the
getting
started,
which
says
that
you
should
use
go
run
and
when
you
use
go
run
and
then
you
try
to
run
cube
control
version,
you
don't
have
any
dld
flags
in
there
for
the
version
information.
A
B
Why
do
we
use
this?
Is
there
some
default.
G
Zero
from
1970,
why
are
we
saying,
go
run
like
what's
the
utility
there.
A
G
A
Yeah,
I
mean
we
aren't
producing
binary,
build
assets,
so
we
have
another
issue
for
that.
E
Right
so
I
agree
with
steve.
If
you
know,
if
the
answer
is
to
have
the
you
know,
tester
build
their
own
just
say
so.
I
A
Yes,
I
still
haven't
had
a
chance
to
read
through
the
this
full
thing,
but
was
this
about
being
able
to
update
a
workspace
af
as
the
type
evolves?
Oh,
no!
No!
It's
different!
It's
about.
B
B
G
B
I
A
So
I
have
seen
several
issues
in
ci
where
it
doesn't
seem
to
really
matter
what
the
test
case
is,
but
it
starts.
Kcp
kcp
is
up
and
live
and
ready
for
some
period
of
time
and
then
at
some
point,
the
go
routine
that
we
have
running
in
the
background
as
part
of
the
end-to-end
framework
gets
a
failure
on
the
reddy's
endpoint
and
it's
ncd
related,
and
you
can
see
that
the
health
check
failed
from
a
context
deadline
exceeded.
B
E
E
Right
today
it's
not
I
I
was
hoping
to
do
it
for
last
milestone.
I
missed
that.
So
my
plan
is
the
next
milestone.
You
know
get
it
working
workspace
aware,
but
today
it's
not
helping
you.
G
A
E
B
I
think
this
is
a
false
claim.
I
mean
we
see
that
locked
line,
but
I
think
so
so
it's
about
the
requests
which
come
from
the
syncer.
They
connect
to
the
virtual
workspace
and
if
you
see
there
there's
a
long
line
so
it
starts
with
clusters,
something
and
then
slash
services.
B
A
B
B
B
A
Yeah
all
right:
well,
it
doesn't
sound
super
important,
I
mean
it,
we
should
fix
it,
but
it
doesn't
sound
like
it's
a
functional
issue.
A
Okay,
make
code,
gen,
fails
and
you
fixed
it
by
running,
make
build
or
go
build.
E
B
G
I
was
going
to
say,
if
you're
not
putting
in
the
flags,
it
should
be
fine,
but
you
actually,
you
need
to
get
flag
for
the
generator.
B
We
could
we
could
write
it
in.
A
So
steve,
do
you
want
to
take
this
one
and
and
run
with
it?
What
do
you
think
in
here
sure.
I
Okay,
don't.
B
So
I
tried
and
I
could
confirm
it.
A
B
A
D
Okay
yeah:
this
is
yes,
sir,
go
ahead.
Yes,
I
was
about
to
say
that
this
is,
if
I'm
not
mistaken,
virtual
workspace
related
right.
G
G
There
was
a
bunch
of
verbs
that
were
not
supported
and
it
would
panic,
but
then
the
oh
you're,
seeing
the
panic
was
just
deregistering
them,
but
actually
making
it
function.
I
think
we
are
still
open.
It
should
be
linked
at
the
bottom.
C
Yeah,
so
now
that
we
we
added
a
fix
for
adding
the
update
operation
is
merged,
but
but
when
so
doing,
an
update
operation
via
any
client
go
works,
but
when
we
try
to
do
a
cube
cattle
apply
or
create
on
the
api
export.
So
this
is
something
specific
to
cube
cattle.
C
There
is
an
issue
which
is
weird
because
when
I
try
to
run
the
the
curl
command
with
the
url
yeah
with
the
swagger
open
api
url,
it
seems
to
work
individually.
C
D
C
D
It
might
be
because
open
api
is
missing
from
the
api
export
virtual
workspace,
and
so,
if
you
there
is
no,
there
is
a
command
line
argument
in
cube
ctl
to
not
check
open
api.
When
trying
to
do
an
update.
I
don't
remember
exactly
the,
but
if
you.
F
B
D
A
H
B
Because
those
endpoints
are
not
meant
for
cube,
cuddle
clients
or
anything
like
that,
it's
not
worth
to
invest
the
time.
Those
are
not
conformant
endpoints.
D
D
Yeah,
I
mean
in
fact,
even
even
in
the
crd,
like
handler
in
the
dynamic
virtual
workspace
implementation.
D
At
some
point
we
calculate
the
open
api
you
know
individually,
but
but
it's
the
open
api
publishing
with
all
the
merging
based
on
on
you
know
the
merging
of
of
the
various
types
and
all
this
which
is
not
managed.
D
But
at
some
point
individually,
when
you
access
the
the
the
the
resource,
then
the
open
api
is
is
calculated
just
for
for
a
given
object
as
exactly
as
it
is
for
the
crd
handler.
But
it's
that's
emerging
as
submerging
as
messy
in
cube,
yeah
yeah
sure,
but
but
why
not
create
an
issue
and
having
something
you
know
which
is
not
planned,
and
if
someone
wants
to
come
with
help
quantity,
yeah.
B
G
B
B
D
And,
probably,
is
it
also
related
to
the
fact
that
it
might
be?
We
might
be
in
a
better
shape
to
support
this
when
the
switch
to
open
api
v3
is
down
because
it's
per
type,
isn't
it
yeah.
B
D
I
A
B
A
All
right
so.
A
A
So
hopefully
we
can
get
that
done
this
week.
Stefan
did
we
in
terms
of
like
turning
off
wild
card
informers.
I
swear.
I
think
we
can.
A
Quota,
I
need
to
do
a
little
bit
more
work.
I
think
hopefully,
now
that
some
flakiness
is
gone.
Hopefully
I
can
make
some
more
progress
there.
A
A
Cluster
workspace
type
take
two
steven,
stefan
and
I
chatted
earlier
today.
I
think
we
have
a
little
bit
left
to
finalize
in
terms
of
if
we
want
it
to
be
a
core
controller
or
not,
but
once
we
have
that
finalized,
it
shouldn't
take
too
much
longer
to
close
out.
D
Sorry,
I
was
about
to
say
about
homework
spaces
that
I
should
obviously
update
the
epic
because
two
tasks
on
18-
this
is
not
really
relevant.
It's
I
mean
the
the
the
main
the
only
remaining
peer
now
is
really,
after
all,
the
preparatory
ones
related
to
too
many
fixes
in
the
virtual
workspace
workspaces
virtual
workspace.
Now
the
only
peer
is
really
the
implementation
of
the
homework
space
handler
and
should
be
ready
for
review.
I
think
tonight.
I
A
Task
list
in
here
and
get
all
the
pr's
linkedin
thanks,
so
sean
if
you're
still
here,
I
know
I
got
a
partial
review
in
on
your
pr
and
I
think
steve
and
stefan
had
as
well.
How
are
things
looking
in
terms
of
at
least
what
you've
got
coded.
F
I
think
I'm
gonna
throw
it
over
to
stefan
and
steve,
but
I
I'm
working
through
the
last
round
of
reviews.
Now
it
seems
like
what
I
have
on
my
my
on
my
disk
right
now.
Things
are
working
I'm
through,
like
sixty
percent
of
the
latest
run
of
reviews,
I
even
picked
up
steve's
knit
thanks
for
that
require
empty
with
the
diff
thing
steve
that
was
a
nice
little
thing
to
learn,
so
I'm
I'm
there.
I
think
the
the
big
one
that's
still
outstanding
is
probably
stefan's
comment.
I
F
The
api
export
we're
building
the
api
definition
set
stuff
that
he
had
questions
about,
but
it
seems
to
be
working
so
yeah.
I
guess
the
question
is.
F
B
A
I
F
Yeah
I
bought
that
one
that
one's
fixed,
so
that
one
should
be
working
now.
It's
now
just
the
virtual
workspace
building
questions.
I
think
that
I'm
working
through
okay.
A
F
Like
I
said
like
it's
kind
of
up
to
y'all
on
when
you're
trying
to
cut
and
what
you
want
to
accept
or
not
accept,
because
I'm
just
trying
to
keep
up
with
reviews
at
this
point,
based
on
what
I'm
seeing
and
the
test
cases
it's
working
as
we
kind
of
have
talked
about,
so
I'm
just
kind
of
waiting
on
what
we're
talking
about
and,
like
you
know,
what
can
we
follow
up
on?
What
do
we
need
in
this
one
and
everything
else?
A
B
A
B
B
A
A
Yeah
I
can
put
in
a
remy
or
the
contributing
guide
if
anybody's
got
specific
questions
feel
free
to
follow
up,
but
I'll
see.
If
I
can
get
a
pr
open
just
to
update
the
doc.
A
Okay,
any
last-minute
topics.