►
From YouTube: Community Meeting, July 6, 2023
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
Okay,
we're
now
recording,
so
thank
you.
Everyone
for
joining
the
kcp
community
call
today,
just
as
a
brief
note,
we
have
a
code
of
conduct
for
the
project
and
that
also
obviously
extends
to
the
community
meeting
so
be
nice
and
abide
by
that
kind
of
conduct.
A
Okay,
let's
walk
through
the
commune
the
meeting
agenda.
The
first
agenda
point
is
from
Stefan
regarding
using
Google,
doc
and
I.
Think
we
had
a
poll
on
the
slack
Channel
about
this
as
well
right.
B
A
B
A
B
A
A
And
I
guess
we
will
try
to
okay,
we
will
try
to
do
Google,
Docs
and
I
will
figure
out
where
Sebastian
and
maybe
review
Stefan,
where
we
should
put
this.
So
we
have
a
location
to
store
it,
where
it's
accessible
to
everyone
and
where
it
won't
get
lost.
B
A
Thanks
give.
A
The
ball,
okay
and
I
just
realized.
Do
we
want
to
look
at
incoming
issues?
Because
we
haven't
done
this
in
the
last
call.
B
We
usually
do
that
at
the
end
as
a
last
topic,
if
there's
time,
okay,
then,
let's
keep
that
conditional,
yeah
and
I
think
we
could
or
we
should
try
it
ignore
all
the
TMC
ones,
but
there
are
many
so,
but
it
was
slow
progress
at
the
end.
We
will
go
through
everything
like
it.
Just
takes
time.
A
D
Yeah
so
I
just
have
a
simple
question,
because
I'm
trying
to
find
recently
and
I
realized
that
there
are
some
similarity
between
cool
bind
and
kcp
CPR,
exploring
API
binding
in
terms
of
the
functionalities
just
but
I'm
just
a
beginner,
so
I
just
want
to
learn
more
from
the
high
level.
B
B
And
in
the
kcp
workspace,
this
is
dedicated
to
a
cluster,
so
there
could
also
be
cluster
scope
resources.
That
was
the
idea
in
in
the
queue
pack
and
a
cluster
has
basically
accessed
to
a
namespace
name,
cluster
name
space
or
something,
whereas
The
Binding
information
is
stored
and
there's
a
request
API
to
create
further
namespaces,
but
there's
no
way
in
a
cube
back
end
to
have
cluster
scope
resources.
E
F
Can
I
I'm
having
a
little
trouble
understanding
a
question?
Maybe
I?
Maybe
I'll
try
to
answer
my
question
by
restating
what
you're
saying
tell
me
if
I've
got
it
right,
coup
bind
is
something
that
is
can
be
used
independent
of
kcp.
It's
a
mechanism
that
you
know
Works
in
a
plain
kubernetes
context
for
having
a
clean
separation
between
a
service
provider
and
multiple
service,
consumers
and
separation
between
the
service
consumers
and
it
can
be
defined,
used
exercised
in
plain
kubernetes,
with
no
kcp
in
sight
and
you're.
F
Also
talking
now
about
back-ends,
so
I
think
maybe
you're
thinking
something
along
the
lines
of
okay.
There
could
be
a
back
end
that
in
some
sense
backs
on
to
kcp,
API,
export
and
binding,
so
that
a
perhaps
a
a
multi-cluster
aware
controller,
a
service
provider
could
be
written
to
an
API
defined
by
Kube
bind
and
then
in
an
actual
deployment.
F
The
deployer
can
choose
whether
to
use
the
the
plane,
Coupe
back
end
or
the
kcp
back
end
and
in
either
case
the
consumers
and
the
provider,
don't
know
or
care
which
backend
is
being
used.
That
was
the
deployer's
choice
in
terms
of
how
to
assemble
their
system.
Have
I
got
that
right.
B
Yeah
perfect
summer,
we
please
write
it
down
and
put
it
somewhere.
So
a
few
words
about
the
back
end.
We
call
it
a
sample
back
end.
So
basically
it
implements
the
apis
which
are
provider
size,
but
it's
just
really
an
example.
It's
nothing
more,
but
it's
purely
Cube
based,
it's
not
kcp
where
so,
basically,
the
idea
is
to
to
give
the
semantics
to
the
coupon
provider
type
objects,
but
speaking
to
a
kcp
cluster
and
creating
workspaces
on
demand
binding
on
demand
all
those
things.
B
F
Okay,
now
this
may
be
a
little
more
detailed
than
the
agenda
suggests,
but
I'm
not
quite
sure,
I
understand
when
I
and
I
did
when
I
read
the
coupon
website
it.
You
know
and
watched
your
your
kubecon
presentation,
I
didn't
quite
clearly
get
that
separation
between
a
coupe
bind
interface
that
providers
and
consumers
use
versus
the
back
end.
F
The
deployers
choose
in
in
particular
what
I'm
seeing
is
that
the
provider
knows
the
structure
of
workspace,
I'm,
sorry,
namespaces
and
how
they
correspond
to
Consumers
and
that's
an
answer
that
works
for
namespaced
resources.
I,
don't
understand
what
the
answer
is
for
a
provider
that
wants
to
deal
with
clusterscoped
resources.
B
B
This
could
also
be
it's
just
not
built
it's
just.
The
design
was
with
kcp
in
mind,
basically
hoping
that
will
extend
explicitly.
F
You
may
have
had
a
typo
in
what
you
just
said.
You
said
that
design
was
with
kcp
in
line
hoping
it
would
extend
to
kcp.
B
F
Right
so
now,
we've
surfaced,
maybe
a
point
that
I
would
like
to
be
really
clear
on.
So
you
suggest
that
there
should
be
a
discussion
of
these
details
somewhere
where
right.
There
is
no.
B
A
Okay,
great
Yoon
is
your
initial
question.
Is
that
answered
yeah
I.
A
Okay,
great
then,
let's
move
on
to
the
next
topic,
which
happens
to
be
contributing
the
CD
puller
to
cube,
binds
and
an
appropriately
modified
API
Jen
mic.
This
is
again
referring
to
coupon,
so
I'm
not
sure
how
much
of
its
would
go
to
that
dedicated
meeting.
But
do
we
want
to.
F
Well,
there's
overlap
right
because
So
currently
there's
this:
let's
start
with
the
crd
puller
right.
This
is
a
thing
that
exists
in
kcp.
It
seems
to
be
a
fairly
General
utility
and
who
bind
is,
is
another
I?
Think,
like
you
know,
my
understanding
is
kind
of
in
some
sense.
Logically,
larger
Community,
since
kcp
bind
is
intended
to
work
with
both
k,
Sorry
cubine,
intended
to
work
with
both
kcp
and
plain
Cube.
B
I'm
not
sure
coupon
is
the
right
place.
We
could
put
it
into
one
of
the
Orcs
either
I
mean
cubind
or
kcp
difference
all
Repository,
it's
pretty
independent
right,
but
I
see.
It
is
indeed
like
that
yeah.
F
G
F
I
guess
in
the
first
place,
my
question
is
really
to
this
group.
What
is
your
vision
of
what
you
want
to
do
with
it?
I
will
say
from
an
outside
perspective
that
I
think
it
is
something
that
is
useful
outside
the
context
of
kcp.
So
I
would
like
to
suggest
that
you
know
it
should
have
so.
F
First
off
people
should
be
able
to
use
it
independently
of
kcp
I
think
they
are
frankly,
because
it
is
just
a
command
line
tool,
but
that's
you
know
suggests
to
me
that
it
should
have
some
home
that
makes
its
broader
applicability
clear,
maybe
has
a
broader
Community
kind
of
embedding.
You
know,
I'm,
not
sure
you
know
that's
kind
of
the
question
I'm
trying
to
bring
here.
F
Let
me
also
ask
or
suggest
there
is
a
kubernetes
Sig
that
what
what
is
what
Sig
is
in
charge
of
custom
resource
definitions?
Maybe
they
should
own
it.
D
F
D
B
F
F
B
And
don't
disagree,
I
think
is
the
right
place,
I'm
just
fearing
this
captioning
about
scope
and
philosophy
and
can
of
worms
and
stuff.
So.
G
But
I
think
I
would
be
okay
with
a
separate
repo.
What
time,
what
it
is
repository
explosion,
because
maintaining
multiple
repositories
requires
time,
rebasing
quiet
time
compared
to
having
a
different
mono
repo.
G
H
I
Your
hand
up
yes,
just
just
a
question
about,
would
the
Shema
compatibility
utility
fall
into
the
same
type
of
case,
because
there
is
also
this.
You
know,
utility
which
is
used
in
one
of
the
is
the
common
lines
which
boils
down
to
comparing
to
Chisholm
schemas.
I
The
name
is
Shema
compat,
I
think
in
package
package.
This
is
used
in
a
number
of
places
too,
to.
F
Okay,
so
I'm
not
hearing
a
lot
of
decisiveness,
so
I
will
volunteer
to
races
with
the
API
Machinery
Sig
and
and
see
what
their
reaction
is.
If
it's
favorable
you
know,
we
can
go
that
way.
If
it's
unfavorable,
we
can
go
differently.
G
Yeah,
it
makes
sense
I
think
at
least
we
can
try
to
make
sure
that
each
and
every
of
those
tools
go
install
compatible
within
kcp
mono
repo,
so
it's
easy
to
consume
from
outside,
and
if
we
see
that
there
is
a
why
the
music
can
always
pull
it
out
to
the
external
repo
I'm.
Just
conscious.
A
Yeah
but
I
mean
this
seems
to
be
a
reasonable
because
of
action
and
I'm,
bringing
it
up
with
music
app
Pi
machinery
would
probably
at
least
also
got
people's
interest
in
general
right,
even
if
they
don't
want
to
adopt
it.
Maybe
it's
still
it
still
like
brings
some
feedback
to
us.
Hey
other
would,
like
others
would
like
to
use
this,
so
it
would
make
sense
to
move
this
to
its
own
repository
at
the
very
least.
F
Okay,
I
think
that's
as
far
as
we
can
go
here
on
that
one.
The
other
part
of
my
question
now
was
about
this
API
gin.
Actually,
that
may
have
been
based
on
a
misunderstanding
in
coup.
Bind
there
are,
does
the
provider
have
to
do
API
gen,
or
does
the
provider
simply
provide
the
custom
resource
definition.
F
Right
in
in
kcp
there's
this
API
gen
that
the
provider
that
the
provider
has
to
do
to
produce
these
the
special
schema,
API
resource
schema
objects
and
my
question
first
question
should
be
in
coup
bind:
does
the
provider
also
have
to
generate
special
schema
objects,
or
is
that
automatic
and
the
provider
only
provides
the
custom
resource
definition.
F
F
B
We're
my
gut
feeling
is
Leaf
set
in
kcp,
because
occupy
is
just
too
simple.
It's
very
trivia,
API
resource
schema
is
a
little
different.
B
I
I
think
this
is
even
done
by
the
back
end.
If
I
remember
right
something
to
to
look
into
the
code.
F
B
H
A
Foreign
I
think
that
concludes
this
part
of
the
agenda.
Yes,
okay,
then,
let's
move
on
I
have
a
brief
thing,
but
I
would
like
to
to
show
off
and
get
some
feedback
on
and
that
feedback
can
be
async
on
the
community
Channel.
A
It
doesn't
need
to
be
now
but
I
after
some
discussions
on
slack
I
looked
into
building
a
new
kcp.io
website,
because
the
current
one
we're
using
is
built
with
website
builder,
it's
difficult
to
maintain,
and
there
was
the
wish
to
have
something
that
looks
a
bit
more
modern
and
I
looked
into
that
and
I
came
up
with
fpr.
That
is
linked
in
the
meeting
notes
and
I
can
give
a
quick
demo
of
this
is
like
pure
Hugo.
So
it's
not
like
any
third-party
tools,
and
this
is
how
it
could
look
like.
A
A
G
A
Okay,
great
I
mean
still
like
looking
for,
like
feedback
seems
to
be
over
overly
positive,
so
I
think
I
will
just
continue
with
building
this
it's
kind
of
happening
on
a
low
priority,
because
it's
obviously
not
like
writing
kcp
code
or
something
like
that.
So
I
will
occasionally
check
in
but
I
think
I'm
making
progress
on
this
because,
honestly,
it's
it's
fun
in
a
way,
so
I
hope.
Maybe
in
two
weeks
I
have
like
the
whole
website
converted,
but
I
can't
promise
anything
yeah
MJ.
G
A
Yeah,
so
we
we
could
adapt
that
and
basically
have
a
shared
theme
that
powers
both
the
landing
page
with
the
blog
and
the
documentation,
although
the
documentation
would,
of
course
be
more
work.
So
the
question
is:
do
we
want
to
change
everything
at
the
same
time
or
do
we
move
bits
and
pieces.
B
A
Okay,
cool
I
don't
want
to
spend
too
much
time
on
that.
You
can
check
out
the
pr.
You
can
run
the
website
yourself
with
Hugo
if
you
have
node.js
and
npm,
but
okay,
eventually.
E
Yeah
pretty
much
whatever
is
written
there,
so
I
wanted
to
talk
about
revising
to
q127,
and
if
we
have
a
timeline
in
Milan,
when
we
want
to
do
it
so
I'm
interested
in
trying
it
out
this
time,
probably
that
I
could
put
someone
and
then
learning
how
to
do
it.
A
A
A
And
as
you
see,
there
are
some
PRS
here,
but
I
don't
believe
that
they're
all
done,
they
probably
need
to
be
views,
so
I
think
the
timeline
on
that
is
as
soon
as
we
go
back
to
reviewing
these
PRS
and
finishing
them.
I
think
that's!
What's
going
on
right
now,.
C
E
A
Very
cool,
yeah
I
think
it
makes
sense
to
coordinate
on
Slack,
see,
request
of
left
of
things
if
there's
something
to
to
take
over
and.
C
E
Yeah,
so
I
just
wanted
to
check
back
on
the
progress
where
we
landed
at
about
moving
the
DMC
really
judicials
to
its
own
repo.
Some
other
had
an
automated
approach
that
he
posted
in
slack,
but
that
requires
all
the
DMC
related
issues
to
be
labeled
with,
like
an
area
transfer
multi-cluster
label,
someone
had
to
check
in
how
we
planned
to
triage
our
issues
that
we
have
on
the
kcp
report
today.
A
I
believe
we
already
have
plenty
of
issues
labeled
with
area
multi-cloud,
right,
foreign
cluster,
but
it's
a
transparent,
multi-class,
select,
TNC,
okay,
so
I
guess
the
first
step
would
be
to
move
those
that
already
are
labeled
appropriately
and
then
see
what
is
left
over
and
if
something
was
not
classified.
Mj.
G
G
If
it
changes,
some
of
those
issues
might
be
obsolete,
but
for
the
sake
of
cleaning
the
main
repository
I
think
we
can
just
start
moving
those
there
and
after
that
code
will
follow
it's
not
what
I
put
television,
but
if
people
want
to
get
rid
of
those
and
many
people
like
we
can
do
that
now.
I
suspect
the
TMC
code
will
land
in
the
next
two
weeks.
There.
A
A
I
Yeah
just
a
question
about
ongoing
peers.
There
were
I
mean
when
I
stopped
when
I
had
to
stop
working
on
that
there
were
a
number
of
ongoing
peers
which
some
of
which
were
I
think
at
least,
if
TMC
has
some
future,
some
of
which
could
be.
You
know
interesting
to
continue
or
finish
so.
My
question
was:
what's
the
plan
for
these
ongoing
peers?
G
Honestly,
it's
very
hard
to
answer
this,
because
currently
TMC
code
changes
quite
a
lot
it.
Basically,
it
still
uses
the
Builds
on
top
of
the
kcp
core.
It's
still
not
yet
a
plug-in,
but
in
example,
there
is
no
sharding
code
yet
because
I
had
to
remove
it
because
of
some
Sinker
issues
and
Fiona
wants
and
I
kind
of
have
a
feeling
that
these
PRS
will
need
to
be.
You
will
not
be
able
to
lift
and
shift
them
to
the
new
repositor,
so
I
think
we
just
need
to
look
PR
by
PR
and
see.
G
So
maybe
we
need
to
do
the
something
similar
with
the
PRS
and
see
what
we
can
basic
level
take
over
and
redo
relays
and
that
never
Repository
once
the
code
lines
sure
okay,
it's
like
there
is
a
lot
of
things
in
here
now,
I,
don't
think
we
even
have
a
like
established
part
and
how
we're
gonna
work.
Will
this
be
part
of
this
meeting
or
we're
gonna
separate
one
or
some
things
like.
C
G
I
Yeah
anyway,
feel
free
to
reach
out
and
to
me,
if,
even
if
I'm
not
very
present
on
the
community
or
for
anything
related
to
TMC
and
peers
and.
B
Yeah
those
who
try
the
rebates
I
will
notice
there's
a
special
control
plane
repository
package,
so
it's
package,
General
control,
plane,
I,
think
something
like
that
in
our
Fork
and
it's
a
copy
of
parts
of
the
cube
API
server
and
the
effort
here
which
I
described
here
as
the
several
steps
are
basically
to
get
rid
of
that
package.
It's
pretty
big
and
when
you
rebase,
you
basically
compare
change
for
change.
B
What
has
changed
since
126
and
which
lines
and
copies
them
over
to
the
to
our
Fork
package
and
the
goal
here
is
to
get
rid
of
that.
So
there
was
an
enhancement
which
landed
for
next
cue
police.
It's
about
generic
control,
plane
staging
repository,
which
actually
is
not
super
critical
for
us.
So
we
can
we
for
cube
or
we
render
Cube
anyway,
so
the
main
repository
so
mono
repo.
So
we
don't
depend
on
staging
repository,
but
anyway
it's
nice
to
have,
and
it
gives
us
confidence
and
Upstream
that
the
direction
is
okay.
B
And
if
you
go
back
to
the
other
steps,
some
of
them
landed
already
so
basically
I'm
splitting
yeah
the
cube
API
server
code,
there's
a
command
CMD,
Cube
API
server
in
upstream
and
is
CMD
control,
plane
which
actually
yeah
sorry,
not
CME,
package
control
plane,
which
is
actually
not
control,
plane,
it's
Google,
API
server
and
that
one
yeah
and
I'm
splitting
it
up.
So
there's
an
API
server
package
already,
which
I
created,
but
everything
below,
like
so
the
instance
and
the
server
services
or
whatever
it
is
there.
B
That's
a
big
blob
of
yeah,
most
spaghetti
code,
so
really
really
bad
and
I'm
trying
to
get
rid
of
those,
and
so
one
step
I
did
already,
which
landed
the
options
which
you
see
there
are
those
are
generic
already.
So
there
are
options
in
the
in
the
CMD
package
and
there
are
those
in
the
package,
control
claim
options
and
the
data
they
are
generic
and
the
former
ones
they
are
for
two
API
servers.
So
as
an
example,
what
I
do
I
remove
codes
from
the
generic
part
which
talks
about
Services
Services?
B
Don't
exist
in
kcp,
there's
something
about
IPS
and
endpoints
and
all
those
stuff
so
I
leave
that
incubatory
as
well,
but
everything
which
is
generic
like
content
map
Secrets
these
kind
of
things
I
moved
to
it's
legendary
part
options
are
done
the
instance.
There
is
not
done
so
I'm
removing
stuffs
there
at
the
moment
or
speeding
them
up
cleaning
things
up
and
that
my
progress
so
looks
good.
B
B
And
maybe
just
last
comment
the
bigger
context.
We
will
have
changes
in
our
Fork
around
logical
clusters,
so
those
will
I
don't
think
they
will
go
Upstream
every
part
of
it,
but
by
far
all
of
them.
But
although
emails
should
be
focused
on
those
changes
not
on
the
generic
generic
Plumbing
of
NECA
ISO.
A
A
A
Thank
you
that
doesn't
seem
to
be
the
case.
So
I'll
just
put
these
meeting
notes
on
here.
However,
now
we
can
take
a
look
at
incoming
issues,
I'm,
not
sure
if
there
was
much,
but
we
can
take
a
look
and
start
that
tradition
again,
I
guess
and
actually
I
am
a
little
bit
okay.
These
are
all
new,
so
chef
and
you
need
to
guide
me
I
think
a
bit
through
this
process.
So
the
idea
was
that
we
go
over
all
of
them
right.
B
Yeah
and
we
we
try
to
keep
it
short,
so
quickly
clarify
what
it
is.
Basically,
it
was
about
labeling,
so
move
them
out
of
the
status
new
okay,
and
we
just
went
one
by
one
for
10
minutes
or
so.
B
H
Not
sure
I've
I'm
not
sure,
that's
accurate,
there's
another
I
guess
actually
it
did
show
up
in
TMC,
but.
F
Really
it's
a
it's
about
core,
a
core
function.
Well,
that
sort
of
isn't
yeah
I,
guess
it
actually
is
showing
up
in
TMC,
okay,
yeah
I,
guess
you
can
call
this
TMC
that.
B
I
would
if
there's
some
core
functionality,
yeah
I'm,
not
even
sure
I,
think
you
can.
You
could
build
that
and
we
have
to
extend
the
framework
a
bit
but
a
little.
B
C
H
F
The
compute
resources
that
TMC
defines
you
know
the
the
r
back
that
it
was
chosen
to
to
put
in
the
exporting
workspace
does
not
support
a
patching
sub
resources.
Neither
status
nor
scale.
H
B
F
A
Yeah
I
think
that
would
make
more
sense.
Okay,
improved
dark
comments,
package
reconcile
record
namespace.
C
B
B
Yeah
we
changed
that
we
discussed
it
earlier.
We
moved
some
over,
but
this
is
more
like
in
the
next
days.
Some
people
will
do
that
for
now
we
just
label
them,
so
the
automation
will
see
the
label
and
move
it
over.
A
Okay,
perks,
API
export
identity
validation
is
incorrect.
A
Mj,
do
you
think
is
this
or
TMC
issue.
A
G
G
A
Yeah,
okay
seems
to
me
that
this
should
go
into
API
exports.
Now
as
an
area.
C
A
So
maybe
we
should
add
those
amateur
if
those
are
defined
in
the
info
Repro
they
might
be
so
I
can
take
a
look
and
see
if
we
can
add
some
priority
labels
yeah.
C
A
A
I'll
I'll
keep
this
in
new,
so
we
don't
forget
to
label
is
important.
Yes,
docs,
pointing
to
Sinker
image.
A
Yeah
I
mean
I,
guess
this
is
like
TMC
documentation
in
a
way,
but
I'm,
not
sure
if
we
are
going
to
publish
GMC
documentation
separately
or
if
it's
also
going
to
live
on
Docs
kcpio.
So.
B
F
Isn't
this
a
generic
issue?
You
know,
since
you
can't
time
travel
and
if
you
anticipate
you've
got
something
that's
broken
for
a
while.
But
you
know
if
the
release
is
an
immutable
thing,
it
can't
reference
itself
no.
A
I
mean
I,
guess
it
could
be
updated
in
the
kind
of
like
release
commits.
That's
all
I've
seen
it
done.
Yeah.
G
A
E
A
A
G
B
A
B
I
think
the
confusion
is
that
they
haven't
been
soup
cutter,
create
workspace
like
it
couldn't
customize.
What
create
workspace
means,
so
we
had
a
had
to
add
another
command
for
create
I
think
this
has
changed.
Upstream
I
think
machine
told
me
that
you
can
put
a
command
now,
which
is
special
I
mean
which
customizes
and
create
sub
command.
B
A
A
A
One
or
two
more
I
guess
sync
resources:
State
should
be
pending
an
API
resource.
Import
is
missing,
sync
targets.
A
Okay,
I
think
we
are
done
for
today,
since
it's
very
close
to
the
ends,
so
thank
you.
Everyone
for
attending
have
a
great
day
and
see
you
all
in
two
weeks.
I
guess!
Thank
you
bye-bye.
Thank
you.