►
From YouTube: Community Meeting, October 4, 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
Foreign
welcome
this
is
the
kcp
community
meeting
October
4th
we've
got
a
agenda
here
that,
unfortunately,
was
just
created,
so
there's
not
really
anything
in
there.
But
if
you
do
have
anything
that
you
would
like
to
discuss,
please
feel
free
to
add
it
to
this
GitHub
issue
or
just
raise
your
hand
in
Google
meet
and
we'll
go
through
topics.
So
I
guess
first
call:
does
anybody
have
anything
you
want
to
chat
about?
That's
not
in
the
issue.
Yet.
A
A
No
all
right
well,
I
will
pick
on
Steve
here
for
a
minute,
Steve
K
and
just
highlight
that
he's
been
working
very,
very
hard
on
updating
the
generators
that
we
have
for
client
sets
informers
and
listers
to
make
all
of
them,
hopefully
properly
and
without
confusion.
A
Multi-Cluster
aware
he's
also
been
working
on
some
AST
based
tooling,
that
you
can
point
at
an
existing
GitHub
or
existing
repository
with
controller
logic
in
it,
and
it
will
do
somewhere
more
than
zero
percent
of
trying
to
automate
converting
your
existing
use
of
client
sets
to
make
them
cluster
aware.
So
that
is
in
Flight
right
now,
PRS
are
open
and
we're
getting
prowl
set
up
so
over
the
next
several
days.
A
Keep
your
eyes
on
that
and
we'll
be
converting
kcp
prop
upper
to
use
the
new
clients,
listers
and
informers
over
time,
and
we'll
update
the
controller
runtime
example
if
needed,
and
just
make
sure
that,
hopefully,
anybody
and
everybody
who's
looking
to
use
clients,
listeners
and
informers
and
point
them
at
kcp
for
a
multi-cluster
multi-work
space
setup
can
use
those
with
a
lot
less
confusion
than
we
have
today.
So,
thank
you,
Steve
and
everybody
who's
been
working
beforehand
on
those
generators,
definitely
really
appreciate
the
hard
work
there.
B
If
we
I
know,
somebody
was
looking
at
refactoring.
B
Edges,
what's
the
the
strumming
API,
we
always
use
manager,
manager
yeah.
So
if
anyone's
like
working
on
taking
an
existing
controller
and
like
making
a
kcp
aware
I'd
love
to
work
with
you
to
see
how
well
automated
refactoring
tools,
Work,
Work,
Out,
The
Kinks,
also,
if
you've
got
apis,
that
you're
generating
kcp,
aware
clients
for
same
thing
so
get
in
touch.
I'd
love
to
collaborate.
A
All
right
anybody
have
anything
you
want
to
chat
about
today
could
be
questions
about
kcp
or
kcp
features
things
that
you've
been
wondering
about.
Prs
that
you
want
to
get
some
Review
Time
on
any
topics
are
good.
If
not,
we
can
just
look
at
the
one
incoming
issue
that
we
have.
C
One
thing
Auntie.
C
Associate
example
and
okay
I
believe
it's
a
big
quotes
to
to
look
at
such
as
how
a
couple
of
improvements
in
control
runs
and
examples.
So
let's
be
useful
and
no
big
thing,
but
you
know,
makes
things
easier
for
people
wanting
to
to
develop
their
own
controller.
A
Okay,
cool:
let's
try
to
get
a
look
at
these.
Do
you
have
stuff
in
controller
runtime
as
well
or
just
the
example,
repo.
C
I
want
PR
against
the
Contour
runtime.
That's
it
for
updating
the
version
and
bringing
that
to
a
new
version,
but
yeah.
A
Okay,
thank
you.
We'll
I
will
try
to
find
some
time
to
glance
at
these
today
and
Steve
or
if
anybody
else
is
interested
in
reviewing
and
helping
out
would
love
some
additional
help
there.
So
thank
you
for
calling
those
to
our
attention.
A
A
Yeah,
so
Chris
is
asking
about
the
CEO,
which
is
stands
for
common
expression,
language
to
do.
Api,
conversions
and
I,
put
together
a
demo
which
I
will
grab
a
link
hold.
D
On
a
second
hold
us
off
of.
A
A
All
right
I'll
pull
this
up
later,
because
it's
being
slow,
so
we
are
working
on
a
prototype
to
be
able
to
do
conversions
of
API
resource
schemas
that
are
part
of
API
exports.
A
So
if
you're,
an
API
service
provider
and
you're
used
to
working
with
crds,
where
you
can
do
conversion
web
hooks
to
convert
between
versions,
V1,
beta1
and
B1
beta
2
and
whatnot,
we
have
a
statically
compilable
and
no
web
hooks
needed
approach
to
doing
this
and
I
have
some
changes
to
our
Fork
of
kubernetes
and
some
changes
to
kcp
to
make
this
doable.
A
A
So
if
you
take
a
look,
we
talk
about.
How
are
we
going
to
do
conversions,
so
here's
a
sample,
API
resource
schema.
It
looks
very
much
like
a
crd.
So
if
you've
never
looked
at
an
API
resource
schema
before
but
you're
familiar
with
crds,
it
should
look
very
familiar.
You've
got
an
API,
Group,
you've
got
the
names,
scope
and
then
individual
API
versions
with
their
schemas
and
you
indicate
what's
served
what's
stored
and
what
sub
resource
sub
resources
are
available.
A
So
what
I
ended
up
doing
is
something
that
looks
like
this.
For
the
time
being,
so
you
create
a
separate,
a
separate
type
called
API
conversion
and
in
its
spec
you
just
list
going
from
V1
to
V2
and
from
V2
back
to
V1,
and
you
can
list
out
individual
common
expression,
language
rules.
So
it
actually
sorry.
A
It
looks
more
like
this,
where
you
specify
the
version
name
for
where
you're
coming
from,
and
then
you
tell
it
where
you're
going
to
and
you
don't
need
to
include
a
version
name,
and
so
this
is
able
to
translate
between
spec.firstname
and
spec.name.first
and
the
same
thing
for
last
name
and
this
works,
but
there's
a
lot
of
conversion
options
that
we
may
need
to
support.
A
So
what
happens
if
I
have
a
single
value
in
one
version
and
I
need
to
go
to
multi-value
or
what
happens
if
I
need
to
go
through
a
list
of
things
and
move
them
somewhere
else
and
maybe
change
the
field
name.
So
if
you
all
have
examples
of
crd,
conversions
that
are
not
listed
here
would
love
to
get
feedback
on
what
sort
of
conversions
you
need,
and
hopefully
we
can
figure
out
a
way
to
make
that
happen.
A
But
it
does
work
and
the
the
code
is
inflate,
so
I'm
continuing
to
to
make
some
changes
and
there's
I
have
a
to-do
list
of
things
that
I
am
working
on.
A
So
if
you're
interested
in
the
pr
it's
2105-
and
so
just
things
that
are
still
to
do
is
an
admission
plug-in
that
validates
that
your
CEO
compiles
and
is
valid
when
the
resource
is
created,
making
sure
that
you're
that
it
can
convert
via
a
hub
or
as
Steve,
was
suggesting
just
that
everything
is
convertible
to
everything
else
in
some
sort
of
some
sort
of
traversal.
A
So
there's
a
bunch
of
stuff
here,
if
you're
interested
in
looking
at
the
pr,
please
do
and
yeah
that
was
a
short
summary
of
what
I've
been
working
on
there.
A
Any
questions
on
that
here,
okay,
so
I,
see
the
next
hand.
Is
it
apoorva
did
I
say
that
right.
E
Yeah
yeah
go
ahead
so
yeah.
Actually,
this
is
regarding
the
pr
I
was
just
trying
my
hands
and
you
might
speak.
I
might
be
aware
of
it,
so
this
is
like
regarding
the
label,
the
value
of
label
exceeding
63
characters
when
we
sync
the
workloads
in
a
workspace.
E
No
I
was
just
like
a
little
blur
on
the
next
steps
after
our
last
discussion
on
this
one
like
what
should
be
doing
like
if
those
labels
are
not
being
referenced
in
the
source
code.
So
what
could
be
done.
A
Right,
thank
you
for
following
up
on
this
sorry
I.
So
the
the
labels
here
I
think
I'll
probably
defer
to
David.
A
So
we
look
through
or
I,
look
through
the
what's
going
on
here
and
basically
we're
generating
labels
that,
as
best
I,
can
tell
aren't
used
by
anything
so
David.
Any
thoughts
on
this.
A
F
Ago,
yes,
it
seems
that
yeah,
it's
not
currently
used,
probably
something
relatively
to
Legacy
or
so
I'd.
Rather
awesome
wait.
G
G
A
F
The
type
of
shoes
you
are
mentioning,
maybe
would
push
us
removing
those,
because
it
seems
that
it
may
give
users,
you
know
IDs
or
other
feeling
that
they
can
get
the
workspace
name
and
and
play
with
that
and
filter
on
these
in
ways
that
were
not
expected
in
in
the
in
the
whole.
A
I
mean
I:
guess:
are
you
trying
to
you're
trying
to
label
select
on
logical
cluster,
no.
G
D
Okay,
discussed
and
Community
call
today,
let's
release.
F
As
for
the
sync
Target,
it
seems
that
we,
precisely
even
for
scheduling,
chose
to
use.
A
Okay,
so
my
suggestion
is
just
to
remove
them.
D
A
D
E
A
Right,
thank
you
all
right.
Next
up,
we
have
a
discussion
about
extending
the
server
in
code
to
let
you
add,
additional
handlers.
A
G
D
Okay,
go
ahead
step
on
yeah
I
commented
on
I,
think
I
mean
there's
a
natural
way
would
not
be
to
use
options
but
to
go
through
the
server
get
to
the
generic
API
server
object
and
there's
something
unprotected
in
class
or
some
some
field
like
that.
That
would
be
the
natural
place
to
add
more
additional
handlers
in
front
of
the
chain
so
completely
independent.
G
D
D
A
Okay,
yeah
take
a
look
and
just
report
back.
If
you
don't
have
access
to
it,
because
it's
private.
A
Yeah
this
one,
so
you
see
this
in
the
logs
there's
this
cluster
authentication
truss
controller,
which
comes
from
kubernetes
that
keeps
trying
to
do
things
with
a
config
map
which
actually
stores
the
ca
certificate.
That
external
servers
like
web
hook.
Servers
should
be
using
to
verify
the
incoming
request
from
the
API
server,
and
if
this
is
not
populated,
then
they
have
to
get
the
content
from
somewhere
else.
A
So
this
does
need
to
be
fixed
for
anybody
who
wants
to
use
this
particular
config
map
which
we're
not
currently
populating
right
now,
but
we
should
be
so.
B
Is
this
turn
on
the
controller
by
giving
it
clients
and
informers
and
whatnot
for
root,
logical
cluster,
or
is
this.
A
That's
a
good
question:
I
think
there
only
needs
to
be
one
of
these
for
the
entire
kcp
instance.
So
I
think
this
is
a
thing
like
doing
what
you
just
said
and
then
telling
controller
authors
or
sorry
webhook
authors
or
providing
a
utility
function
where
it
can
get
this
for
them,
but
yeah
there
should
just
be
one
of
them:
I
think
putting
it
in
a
really
makes
sense.
A
H
Andy,
it's
Andy
I'm,
just
letting
you
know,
I
put
an
issue
in
for
the
next
community
meeting
for
October
11th
I
created
one
and
put
us
in
there
for
Edge
workload,
distribution,
okay,
some
work
that
we're
doing
internally.
Our
demo
is
not
ready
for
this
week,
but
it
will
be
for
next
awesome.
Thank.
H
A
All
right
well
good
to
see
everybody
see
you
next
week
have
a
good.