►
From YouTube: WG API Expression Bi-Weekly Meeting for 20200901
Description
WG API Expression Bi-Weekly Meeting for 20200901
A
All
right
welcome
to
working
group,
api
expression,
it's
the
1st
of
september
2020
and
we
have
one
agenda
item
and
I
think
sometimes
you
edit
it.
C
C
And
we
run
into
this
quite
often
that
we
want
to
know
if
it's
largely
based
on
the
open
api,
and
we
often
run
into
the
situation
where
we
need
to
know
if
they,
if
by
just
looking
at
the
at
a
manifest
and
without
an
api
schema,
we
want
to
know,
is
this
resource
named,
is
not
just
things
like
being
able
to
add
a
namespace
to
a
resource
to
all
resources
if
they
are
namespaced.
C
A
Useful,
I
honestly
already
expected
that
to
be
in
there,
so
I'm
I'm
surprised
it's
not
yeah.
That's!
Where
does
that
information
currently
come
from?
I
think
it's
part
of
the.
C
The
only
way
we
can
find
that
now
is
by
using
the
discovery,
client
and
actually
look
it
up,
but
yeah
there's
nothing
in
the
namespace
or
in
the
open
api.
For
that.
D
A
Where
does
the
information
currently
come
from
if,
if
a
resource
is
namespace
or
not
like
on
the
internal
side,.
B
Yeah,
that's
a
good
question,
but
I
think
it's
mostly
coming
from
discovery.
I
I'm
curious
what
happens
if
one
specified
the
name
space
on
a
on
an
object
that
is
not
map
space?
Do
you
know,
martin.
D
B
Yeah,
I
I
suggested
that
to
mount
an
already,
but
my
concern
is
what,
if
you
send
an
object,
that
is
a
cluster-wide
cluster
and
it
has
the
namespace.
So
obviously,
you'd
only
include
increasing
the
in
the
path.
But
what,
if
it's
included
in
the
object?
What,
if
you're
trying
to
build
a
auto
completion
tool?
Is
it
going
to
suggest
that
you
specify
the
namespace
in
the
object?
B
B
D
Yeah,
that
would
be,
that
would
be
a
superior
typing
to
what
we
have
today.
I
think,
and
I'm
not
even
sure
I
would.
I
would
love
somebody
to
just
go
and
check
really
quick,
but
probably
not
blocking
for
this
meeting,
but
to
see
if
we
even
do
the
right
thing,
if
we
even
like
admit
an
error
back,
if
you
include
it.
B
Yeah,
let's
see,
let's
see
what
we
can,
let's
see
if
we
can
find
an
easy
way
to
remove
it
from
different
api.
A
A
Yeah,
okay,
so.
A
Okay,
so
that,
but
basically,
if
they,
what
was
it,
they
use
a
different
metadata
or
that
they,
at
least
in
the
open
up
api
specs.
They
don't
show
the
optical
meta
here,
that's
everywhere,.
D
Yeah
we'd
have
to
obviously
there's
going
to
be
wrinkles
in
this,
because
I
think
right
now
we
usually
shared
type
and
we
just
reference
it,
and
so
now
we
would
no
longer
be
able
to
reference
a
single
type
or
we'd
have
to
find
some
other
creative
alternative.
So
I'm
the
there
are
alternatives
here.
So
you'd
have
to
kind
of
like
go
through
them
and
figure
out
like
what
is
the
least
gross
way
to
fix
this.
D
Not
doing
that
would
mean
that
you
kind
of
have
to
like
mask
over
the
fact
that,
like
they're,
actually
the
same
type
in
open
api,
somehow
and
that's
also
a
little
a
little
weird-
I
I
don't.
Unfortunately,
I
don't
see
a
really
easy
way
to
to
tackle
this,
but
I
I'd
still
love
to
see
somebody
like
look
through
the
alternatives
and
see
which
one
is
at
least
the
least
gross.
A
D
Yeah
we
can
find
that
I
there's
got
there's
somewhere
in
the
native
types.
There's
some
there's
some
bit
that
gets
flipped.
It
has
to
exist,
so
we
can
we'll
definitely
be
able
to
find
it.
I
don't
know
like
how
like
we'll
be
able
to.
D
Yeah
that
was
yeah.
I
I
think
I
think,
sending
out
a
even
like
a
half
page
thing.
It
says
like
these
are
the
viable
options
in
the
design
space,
even
ones
that
we
know
we
don't
like
with
like.
You
can
even
put
like
fine
print
they're
like.
I
don't
think
this
is
an
acceptable
solution,
but
these
are
the
only
options
we
have
and
then
we
could.
D
People
on
this
call
or
other
people
that
are
interested
could
get
going
and
kind
of
weigh
in
on
like
other
details
and
thoughts,
because
I
think
there's
some
little
cheap
thinking.
That's
going
here.
A
D
Yeah
I
mean
even
like,
even
if
we
had
to
like,
like
one
option,
I'd
like
to
see
considered
in
the
variance
is
like,
should
we
just
do
like
an
x-kubernetes
open
api
thing
that
at
least
tells
you
what's
going
on?
If
you
care,
that's,
obviously
a
lot
less
clean
than
anton's
suggestion
of
actually
picking
the
fixing
the
type
system,
which
would
be
amazing,
but
you
know
kind
of,
like
figuring
out
what
the
options
are
and
then
figuring
out
how
tractable
those
different
things
are
and
whatever
they
would.
D
A
Okay,
is
there
anything
else
we
need
for
this
or
what?
How
how
do
you
want
to
do?
We
want
to
continue
in
that.
C
D
Yeah
so
we'll
keep
tracking
it.
It
would
be
fine
to
open
an
issue
too
if
anybody
doesn't
have
time
to
do
more
than
that.
E
B
There
is
an
issue
in
customize,
we
don't
have
one
and
that's
already
an
item
in
the
in
the
list
below.
A
Okay,
so
does
somebody
want
to
own
it
or
and
or
push
it
or
should
we
set
it
on
help
need
it
add
it
to
the
board.
A
B
Miss
and
I've
already
put
it
as
healthy,
okay,
okay,
so
I'm
gonna
put
a
link
to
the.
A
Notes
all
right
does
somebody
else
have
something
for
the
agenda,
or
should
we
continue
with
the
stairs
update.
A
B
A
A
B
E
B
A
All
right
built
in
declarative,
defaults.
A
B
A
Yeah,
okay,
yeah
everything
is
related
and
it
depends
on
each
other.
A
At
first
I
would
like
yeah
for
status
walking.
I
figured
out
how
to
fix
some
custom
resources
that
works.
Now.
A
I
have
one
remaining
issue
with
custom
resource
definitions,
because
the
current
test
tests
by
conflict
and
crt
status
gets
changed
behind
the
scenes
which
cause
the
conflict
in
the
test.
So
I
had
to
ignore
it
from
the
test
for
now,
until
I
find
a
better
way
to
test
it,
but
in
general,
all
types
have
implemented
all
types
of
cost
conflicts
so
far
have
implemented
status,
writing
and.
A
A
D
Yeah,
so
this
is
moving
along
where
jenny
and
I
spent
some
time
prototyping
late
last
week,
we're
gonna
spend
some
more
time.
This
week
there
was
an
interesting
thought
from
justin
santa
barbara
that
I
liked,
which
was:
if
we
did
it
right,
we
might
actually
be
able
to
have
the
same
interface
for
the
generated
go
types
that
we
use
normally
and
the
ones
that
we
need
to
use
to
build
manifests.
D
So
that's
that
was
a
comment
in
the
design
that
I'll
respond
to
today.
That
has
some
corner
cases
to
think
through,
but
yeah
we're
making
we're
making
a
little
bit
of
progress.
Currently,
the
goal
is
to
work
in
a
cube
builder.
A
feature
branch
then
use
that
as
a
way
to
get
feedback
from
a
community.
D
D
This
one
I
I
haven't,
made
real
progress
on
since
we
last
talked
because,
like
I
got
pulled
away
on
a
couple
other
things,
but
I
am
trying
super
hard
to
get
pulled
back
to
the
this
week.
I
really
want
to
get
a
example
of
how
this
might
look
if
we
put
the
tombstones
into
the
field
set
and
a
couple
other
things
so
hopefully
I'll
have
an
update
next
time.
A
B
A
A
D
I
don't
have
any
update
on
that
one
from
her.
I
don't
think
I
don't
think
that's
been
a
focus
right.
A
Now
this
one
should
be
merged
now
that
we
press
open
again.
D
Okay,
yeah
there
was
quite
a
there-
was
quite
a
elaborate
return
to
normal
process
going
on
for
the
main
master
branch.
Okay,.
A
D
A
Yeah,
we
still
don't
work.
Great
guys,
cost
like
it
doesn't
mean
empty.
D
Oh
sorry
about
that
which
one
we're
looking
at
yes,
so
we
have
continued
to
make
progress
on
this.
I
have
a
a
prototype
that
does
a
fix
up
past
so
that
we
deal
with
the
corner
case
of
what
happens
when
you
add
one
of
these
annotations
to
a
type
that
didn't
have
it
before,
and
you
have
a
bunch
of
data
in
the
old
format.
D
It
seems
to
work.
I
wanted
to
cover
the
reverse
case
of
when
you
go
to
turn
the
comic
back
off
again,
which
is
actually
a
little
trickier,
because
it's
kind
of
harder
to
identify
that
that's
happened.
So
I'm
going
to
try
and
get
that
sorted
out,
and
I
do
actually
need
to
write
a
little
mini
design
on
this,
because
I
think,
there's
enough
alternatives,
everybody's
going
to
see
that
so
we'll
probably
have
that
out
later
this
week,
but
yeah.
This
is
actually
one
of
the
things
I've
been
working,
the
most
on.
A
D
A
Open
a
page
doesn't
include
this
scope.
Now,
that's
the
one
we
just
talked
about.
Okay,
no
change
apply
still
persists.
I
haven't
seen
that
one
before.
B
A
B
B
B
But
even
if
you
specify
a
string,
the
string
might
change
because
it's
going
to
be
500
meters
or
something
like
this,
so
the
string
is
changed
by
the
appearance
of
the
process.
We
can't
detect
to
change
or
not
so
we're
always
paying
the
date,
which
is
then
triggering
an
entity
right
of
the
resource,
even
if
nothing
has
changed.
It's
also
bad
for
you,
it's
bad
for
many
reasons.
It's
we
also
have
about
that
quantities
are
of
type
string
rather
than
types
of
industry,
so
it
also
fits.
B
So
I
suspect
one
more
time
we
are
going
to
need
a
specific
flag
in
the
open
api
to
say
this
is
the
quantity
you
need
to
kind
of
engage.
A
B
D
B
The
change
would
happen.
Yeah.
That's
a
good
question
like
if
you're
using
clients
out
of
black
the
cube
canal
is
going
to
detect
the
change
and
send
the
patch
to
update
the
value.
B
B
So
you
never
have
like,
if,
like
even
if
you're,
not
changing
anything,
you
never
have
a
clean
div
and
also
you
can
have
and
relevant
conflicts
like,
because
maybe
you've
got
a
conflict
with
someone,
because
it
believes
you
tried
to
change
that.
Even
if
you
didn't.
A
B
It's
a
bug,
but
it
has
a
backwards
compatibility
issue
and
it
has
the
extra
conflicts
and
the
div
and
the
lcd
update.
I
mean
that's
enough
to
make
it
somewhat
important.
A
Backward
and
comfortable
namespace
webhook
nodo
is
on
it.
I
don't
know
what
it's
about.
F
Yeah,
this
is
about
in
server
side
apply.
We.
F
A
A
A
Okay,
okay,
yeah.
We
have
this
one
last
week
that
we
want
to
investigate
the
curricularization
code
and
find
opportunities
to
define
some
validation
in
the
schema
or
make
it
a
exposure
schema,
and
we
still
need
somebody
to
look
at
it
once
we
want
to
do
it.
We
want
to
do
it
and
yeah
I'm
not
getting
any
new
input
from
from
lucky
lucky.
A
I
don't
know
how
to
say
I
pinged
him
and
said
I
put
him
some
stuff
and
if
you
want
to
do
a
meeting,
but
I'm
not
getting
anything
so
I
would
expect
the
steering
people
to
get
towards
us
if
they
need
more
from
us.
A
B
B
A
Okay,
yeah
I'll
I'll
continue
with
the
standard,
wiping
and
hope
that's
done
soon,
and
I
can
do
something
else.
Okay,
I'm
looking
forward
to
you.
B
A
Depending
on
how
much
time
you
have
that's
the
problem,
yeah
right
now,
so
my
my
weekends
were
instead
of
working
for
my
saturdays
and
that's
almost
all
the
time
in
my
head,
so
I'm
I'm
I'm
checking
in
once.
I
have
something:
okay,
all
right,
then,
thanks
for
coming
and
have
a
great
remaining
day,
see
ya.