►
From YouTube: SIG Cluster Lifecycle - Cluster Addons 20200428
A
A
B
Yes,
I
think
there's
some
sort
of
weird
more
complicated
to
my
mind,
matching
system
where
CN
CF,
as
is
a
sort
of
sponsoring,
or
has
to
say
how
many
projects
they
want
slots
for,
and
then
the
google
Summer
of
Code
program
tells
them.
You
can
have
a
different
or
a
potentially
different
number
and
then
I
guess
CN
CF
says
all
right
of
given
we
have
n
slots,
we
want
to
do
these
projects
and
then
then
it
finally
comes
to
us
as
the
sort
of
mentors.
B
C
B
B
Is
the
only
thing
we
can
do?
We
can
click
on
want
to
mentor
and
yes,
a
that's
what
we
can
do
yes,
but
we
can't
click
on,
don't
want
to
mentor.
So
it's
a
little
like
I.
Don't
know
that
there's
any
visible
difference
between
the
only
thing
we
can
do
is
+1
I,
don't
think,
there's
any
minus
one
or
zero.
B
Click
on
it,
hi
I
have
done
that
I
think.
Yes,
thank
you,
yeah
yeah.
If
there
are
others
as
well,
then
that
was
a
good
thing
to
do,
but
yeah
it's
I,
think
other
organizations
have,
for
example,
exported
spreadsheets
and
constructed
their
own
voting
schemes.
Externally,
CN
CF
has
not
done
that.
To
my
knowledge.
A
B
A
A
A
B
D
D
D
C
B
C
D
Because
we
can
post
that
as
well
and
the
google
group
putting
that
kind
of
thing,
yes,
for
the
main
sake
to
see
on
all
the
email
updates
sure.
Well,
we
already
got
a
great
start
on
the
outline,
so
I
will
start
translating
that
into
some
slides
and
then
ask
for
edits.
Give
the
document
up
somewhere
already.
It
is
inside
of
our
meeting
notes.
Oh
several
weeks
ago.
D
A
E
Yeah
just
thought
it
may
be
a
small
revelant
to
mention,
because
we
do
come
come
back
to
the
topic
of
templating
every
now
and
again
and
I
don't
know.
What's
the
view
that
this
group
broadly
accepts
its
present.
Aside
from
you
know,
we
shouldn't
single
out
on
a
particular
template,
Ian
or
you
know,
force
use
of
certain
tools
and
users
or
anything
of
that
sort
of.
If
there
is
any
and
any
thoughts
on
what
puts
our
recommendation
on
best
practice,
do
we
have
one.
B
I
think
we
we
generally
are
trying
to
not
sure
I'm
trying
to
take
more
of
a
case-by-case
basis
right
to
like
I,
think.
We
think
that
some
places
where
we
historically
use
templating
are
better
done
using
like
llamo
transforms
rather
directly
or
in
an
operator
mm-hmm
but
yeah
that
we
know
that
doesn't
encompass
everything.
I
think
mm-hmm.
F
E
E
E
I'm,
just
a
thinking
well
generally,
like
there's,
there's
a
lot.
One
can
do
with
you
so
been
trying
to
solve
one
particular
problem
and
realized,
maybe
maybe
simple
somewhat
opinionated,
rapper
and
type
of
queue
is
actually
of
some
interest
to
those
who
may
be
interested
to
use
queue
and
I.
Thought
too
is
kind
cool
and
it's
nice
that
it
integrates
well
as
a
go
library.
E
You
know
go
language
somewhat,
I
mean
he.
You
can
convert
directly
back
from
cue
types
into
into
go
types
and
such
things
right,
I
had
a
brief
look
at
you
know.
I
forget
is
that
star,
Larco
Skylark.
Now
the
the
python-based
interpreter
Python,
asked
interpreter
that
is
now
actually
integrating
this
customized,
which
was
kind
of
curious.
The
the
thing
about
that
is
that
you
kind
of
have
to
whatever
you
you
create
in
your
starlight
program.
E
You
you
have
to
convert
back
to
your
go
types
and
such
things,
and
you
can't
really
just
go
straight
to
Jason
from
there.
So
yeah
I
found
that
a
little
bit
a
bit
of
a
challenge
and
I
learned
that
with
cue,
you
can
actually
evaluate
UQ
program
and
and
get
back
Jason
and
who
has
some
interesting
integrations
most
kubernetes
types
as
well.
E
E
Just
for
the
record
I
sort
of
referenced,
that
was
the
epp-
are
that
added
start
like
support,
and
anyway
it's
it's.
It's
interesting
because
I
mean
I
thought
of
using
customized
for
I.
Think
recently
and
I
found
it
kind
of
hard
that
there's
only
certain
things
that
you
can
do
like.
For
example,
you
can
insert
the
name
prefix,
but
you
can't
override
the
name
as
a
whole,
so
you
have
to
have
some
base
name
and
what,
if
I,
don't
have
one?
E
B
B
I
just
want
to
say,
like
I
something
I've
learned
about
customize
is
like
one
of
the
things
that
I
think
they've
been
working
on
for
I
guess
the
past.
One
of
the
big
names
in
requests
here
is
exactly
the
feedback
you
gave,
and
so
things
like
name
prefix,
are
actually
defined
now,
as
under
the
covers
as
external
as
transform
that
everything
is
a
transform
and
in
theory
everything
is
pluggable
and
I'm.
E
C
B
A
project
path
or
as
I
think
go
plug-in
libraries
and
I'm
guessing,
but
I,
don't
know
that
this
star
luck
is,
though,
is
intended
to
be
a
third
and
affordable
way
of
implementing
them.
So
you
could
implement
so
customized
already
has
the
way
that
you
could
implement
your
name,
replace
transform
today,
you'd
have
to
write
essentially
a
binary,
copilot
and.
C
D
I
mean
my
so
much
more
useful,
interesting
yeah.
E
Yeah,
that
is
quite
interesting,
yeah
but
I
mean
I,
had
some
interesting
findings
with
askew
as
well,
because
my
years
case
would
be
to
embed
something
in
in
in
the
CLI.
That
should
be
fairly
small
and
I
go
on
to
introduce
something
that
is
obviously
for
in
my
natural
preference
would
fall
with
something
like
typescript,
otherwise
that
anyway,
so
what
yeah
I'm
happy
to
share
a
few
things
later
on
some
one.
Some
once
I
have
some
some
code
that
that
actually
is
presentable.
Yeah
and.
E
Simpler
syntax,
let's
say
and
I'm
a
native
implementation,
the
kind
of
Li
the
canonical
implementation
is
they
even
go
on
and
it's
easily
callable
as
you
go
library
and
I've
been
able
to
do
a
couple
of
things
and
yeah
and
I
mean
like
once.
One
thing
I
found
is
that
you
know
once
you
write
a
program
like
that,
we
fundamentally
in
queue.
E
D
Could
certainly
see
that
even
outside
of
the
NGO
world,
because
it's
nice,
that
it
is
a
library
you
can
use,
also
set
up
a
init
container
and
then
have
a
config
volume.
That
kind
of
thing
yeah
check
it
into
a
repo
home,
yeah
pre-process
it
you
could
do
all
sorts
of
pipeline
tech
things
in
the
same
way
that
Jason
at
that
kind
of
thing,
yeah.
D
C
D
The
most
interesting
part
about
how
this
fits
into
the
total
overview
of
all
the
work
streams
is
that
is,
in
the
packaging.
An
application
bit
by
application.
I
mean
like
coop,
CTL,
apply
kind
of
workflow,
and
we
we've
jumped
on
to
like
okay,
well,
coop,
see
tail
kind
of
supports
the
customize
a
plier
in
a
way,
and
then
it
has
these
other
arbitrary
ways
to
source
things
that
feel
like
packages
like
the
HTTP
source.
D
Extending
coops
et
isn't
really
our
current
mission
right
so
using
what's
already
in.
There
is
sensible,
despite
the
caveats
and
arguments
there
around
that
area,
but
we
can
build
things
directly
around
what
is
currently
supported
in
coops
ETL,
for
which
there
are
a
couple
mechanisms,
and
then
we
have
the
proposal
for
supporting
OCI
package.
D
Nice
yeah
I
just
was
talking
about
the
ITER
lifecycle
manager
is.
F
D
F
D
D
D
F
This
may
be
a
good
time
to
bring
up
that
that
bundle
cap
is
still
available
for
view
and
that
this
is
what
this
is.
It's
something
to
address
is
to
define
just
a
generic
packaging
format
for
essentially
stuff
that
results
in
llamó
that
you
can
then
associate
with
various
tools
that
know
how
to
unpack
them.