►
From YouTube: WG API Expression Bi-Weekly Meeting for 20210511
Description
WG API Expression Bi-Weekly Meeting for 20210511
A
A
A
All
right,
that
means
no,
I
suspect,
let's
hop
to
the
sha
requirements,
really
quick.
We
seem
to
be
good
on
all
the
features
and
bugs
which
is
great.
We
are
hitting
the
feature.
Phrase
feature
enhancements
freeze
this
week.
I
think
so.
I
think
we
need
to
update
our
cap.
I
read
a
comment
about
that
jeffrey.
Were
you
updating?
A
A
Oh,
that's
already
merged,
never
mind,
hey
antoine,
I
was
just
going
over
the
ga
stuff.
Again,
I
think
we
have
to
update
the
cap.
I
read
an
email
earlier.
A
B
I
think
in
the
message
they
said:
it's
fine,
I
think
if
we
let
it
this
way,
it's
not.
I
don't
think
any
of
that
would
be
upon
finance
meant
please
face.
A
Good
then
immutability
say
you
want
to
talk
about
that.
C
Yeah,
I
have
updated
some
expertise,
design
talk
and
I
think
how
narrowed
on
the
use
cases
that
I
think
we
want
to
support
and
trying
to
give
a
proposal
to
how
to
support
it.
So
the
key
differentiate
difference,
I
guess,
is
that
we
probably
want
to
treat
the
collection
a
little
bit
differently
than
other
fields.
C
When,
when
talk
about
what
it
means
about
the
immutability
so
yeah,
I
I
will
think
I
think
the
next
step
for
me
is
then
to
like
keep
your
uterus
on
this
dog
and
share
out
for
some
feedback.
C
And
if
you
have
some
comments,
yeah,
please
take
a
look
yeah
that
can
be
all
from
for
me.
Thank
you.
A
Okay,
oh
yeah,
I'll,
open
it
as
well.
Okay,
this
one-
I
I
don't
think
is
happening
anytime
soon,
with
none
of
the
relevant
people
showing.
B
Kevin
is
trying
to
think
about
decent
channels,
so
there's
a
possibility
that
kevin's
going
to
do
something
with
that.
We.
A
Yeah
discount
as
well,
I
keep
having
more
discussions
about
improving
the
customer
resources,
I'm
involved
with
here
now
and
how
they
grew
over
time
and
there
was
no
best
practices
before
and
what
we
should
change
and
what
the
best
practice
would
be.
Actually
I
had
a
discussion
today
that
kind
of
matched
the
email
about
best
practice
for
references
that
came
in
handy,
and
so
I
I
wouldn't
mind
helping
on
that
with
the
news
found
problems.
E
Yeah,
I'm
having
conversations
with
some
of
the
names
on
this
list
like
this
week
to
try
to
see
kind
of
just
what
what
the
kind
of
minimal
mvp
of
some
some
sort
of
you
know.
Better
conventions
are
and
we'll
keep
you
updated
on
any
progress
from
that.
A
That's
good,
if
you
yeah,
if
you
want
to
invite
me
or
anything,
always
feel
free
to
drop
me
a
meal
all
right
defaults
in
open
api,
oh
yeah,
I.
E
B
Yeah
we
need,
we
need
to
revamp
this
dog
so
much
because
I
mean
right
now
it
can
we're
kind
of
in
the
middle
in
the
team.
So
I
propose
some
changes
internally
at
google
and
so
what
kind
of
restructuring
to
know
who's
going
to
do
what,
and
so
I
don't
think
most
of
this
makes
any
sense
at
this
point.
A
B
The
default
tag
is
being
worked
on
by
yash,
on
that
we
talked
to
on
slack,
and
so
it's
gonna
be
challenging,
but
it's
it's.
Someone
is
working
on
that.
A
B
Let's
try
and
and
work
on
the
dark
as
we
talk.
A
For
the
that's
not
on
the
list
right
now,
I've
created
a
cat
pr
for
the
examples,
but
it's
very
shallow.
So
far
I
saw
the
comments
and
started
replying
to
a
few,
I'm
not
very
positive
on
getting
into
implementable
merged
until
what
is
it
end
of
this.
D
F
B
E
Sure
yeah
I've
been
putting
together
a
small
dock
around
some
ci
cicd
best
practices
for
ssa.
That
was
getting
feedback
from
antoine
before
sharing
with
some
users,
just
as
a
a
starting
point
for
discussions
around
how
to
use
ssa
in.
In
that
context,
and
as
previously
mentioned
going
to
be
chatting
with
mostly
eric
tune
and
some
others
around
kind
of
the
how
to
best
standardize,
some
or
kind
of
what
the
first
steps
are
for
trying
to
tackle
the
api
conventions
and
principles.
E
Kind
of
standardization
question,
but
haven't
really
spent
too
much
time
on
that,
yet
so
unsure
what
that's
gonna
lead
to.
But
those
discussions
will
be
happening
soon.
B
A
Sure,
but
can
I
mention
to
anton
a
few
days
ago,
I've
been
in
a
pretty
active
conversation
about
how
people
would
use
apply
in
git
ops
workflows.
So
if
that
matches
what
you're
doing
with
the
ci
stuff,
maybe
I
can
share
with
you
the
discussions
they
had
the
problems
they
had
the
customers.
We,
I
saw
yeah.
E
Definitely
yeah,
I'm
gonna
follow
up
with
with
anton
after
this
and
see.
If
see
what,
because
I've
been
out
for
the
past
week,
see
what
the
status
is
and
then
I'll
reach
back
out
to
you
and
kind
of
compare
notes
on
that.
A
B
Yeah
yeah,
I
think
it
makes
a
lot
of
sense.
We
we
had
a
few
discussions
and
so
we're
kind
of
switching
back
to
ssa,
but
we
had
a
few
discussions
about
how
to
document
ssa
and
I
think
the
vision
eventually
was
that
that's
documenting
how
this
works
and
there's
explaining
and
I'm
not
possibly
not
using
the
word
document
but
explain
how
to
use
it,
and
so
how
to
use
it
is,
has
different
aspects
of
that
right.
You
can
want
to
write
a
controller.
B
You
can
want
to
use
it
in
a
ci
cd
system
or
you
can
want
to
use
it
as
a
cube,
color
user,
and
so
I
was
trying
to
separate
these
stories
and
I
think
explaining
how
to
use
it
in
the
documentation
is
not
the
best
and
so
having
blog
posts
or
like
stories
of
how
you
should
do.
That
seems
interesting
to
me
and
so
yeah
we're
trying
to
have
maybe
a
blog
post
or
whatever
about
this
ci
cd
system
story
with
solar
side
of
life.
A
Yeah
yeah
and,
like
I
don't
know
when
the
schedule
for
us
that
is
or
if
we
like,
who
probably
talked
to,
I
don't
know,
sully
or
somebody
about
the
idea
of
doing
a
breakout
session
for
the
next
cube
con,
where
people
can
bring
their
operator
or
like.
We
show
an
operator
that
we
convert
sounds
like
a
good
intro
on
how
to
allow
people.
A
A
Okay
eddie:
do
you
want
to
talk
about
anything.
G
No,
I'm
just
I'll
give
your
kept
a
review
too,
and
I'm
really
interested
in
that.
So
we'll
see
where
it
goes.
A
Awesome
things
anton.
B
B
Yeah,
and
so
I
was
even
having
this
idea
right
now-
that's
something
we
discussed
with
sully
before,
but
you
know
I
don't
know
if
you've
used
skip
got
all
described
before,
but
it's
extremely
hard-coded,
and
when
you
do
a
girl
describe
nodes,
it's
capable
it's
capable
of
actually
showing
you.
The
pods
that
are
associated
with
with
each
individual
node,
and
so
something
that
would
be
nice
is
to
give
the
ability
to
people
to
have
a
girl
described
for
their
own
types.
B
The
problem
is
there:
is
this
cost
field
dependency
between
between
these?
And
so?
If
we
had
this
references,
maybe
we
would
have
the
ability
to
say:
okay,
look.
My
note
is
actually
made
of
pods
and
when
I
want
to
pin
the
note,
I
want
to
also
pin
the
the
parts
and
whatever
characteristics
of
these
parts,
and
so
for
that
we
need
the
cost
references
and
we
need
to
be
pretty
smart,
that
it
it's
programmatically,
usable
and
so
that's
kind
of
a
long-term
vision.
B
I
think
it's
a
long-term
vision
to
do
that,
but
that's
kind
of
the
direction
I
want
to
go
so
well,
and
so
the
references
are
actually
super
useful
in
so
many
different
contexts.
A
G
A
A
And
that's
like
feels
like
it's
similar,
like
it's
all
referenced
somehow,
but
you
can't
use
owner
references
for
it
because
then
stuff
gets
garbage
collected
and
everything.
So
I
don't
know
if
that
matches
that
to
some
extent,
but
the
reference
itself
like
how
objects
are
referenced
is
already
a
problem
right
now.
So
yeah.
G
Yeah,
I
just
so
mache
told
me
the
other
week
he
said
yo
someone
swapped
out
some
of
the
open
api
libraries
that
we're
using
we
swapped
out
to
go
open
api
for
something
else,
or
you
know
anything
about
that.
B
G
No,
no,
that's
what
it
was
so
do
do
we
have
any
do
we
have
any
like
ties
to
using
gnostic,
because
gnostic
is
still
used
in
the
cli,
runtime
libraries
and
stuff.
G
B
Yeah,
I
think
I
would
love
to
understand
this
plan
better,
so
for
context.
Here's
what
we're
doing!
We
have
the
internal
in
kubernetes.
We
have
this
json
representation
of
the
open
api
which
is
done
by
this
cube,
open
api,
open
api
model
that
I
mentioned
just
before,
and
sometimes
when
we
want
to
solve
it
to
people
and
especially
google,
we
transform
it
into
a
portable
version.
B
The
portable
version
is
not
made
of
the
same
structure
as
the
as
the
json
one,
so
we
have
to
transform
it
into
a
gnostic
object
that
we
then
solidize
into
portable
cube.
Coral
uses
the
gnostic
version,
the
portable
version,
because
it's
much
faster
to
decode
downloads,
the
gnostic
russian
portable
version
decodes
into
a
gnostic
thing
and
then
transforms
into
a
portal
models,
which
is
something
I
invented
to
do
the
validation
and
some
other
things.
G
G
G
B
B
Yes,
I'm
if
you
can,
if
you
can,
maybe
maybe
we
can
do
this
online,
whatever
online
means
on
on
slack,
describe
what
you're
trying
to
do
and.
B
Jeff
is
gonna
mention
what
he's
working
on
and
then
we
can
talk
about
it.
H
Yeah
sure
so,
right
now
I'm
looking
into
sort
of
some
of
the
stuff
we
will
be
for
me
to
do
for
supporting
open
api
v3.
So
right
now
I
have
a
brief
document.
Probably
like
a
cap
is
coming
soon.
G
B
No,
I
I
think
so.
Open
api
has
the
type
definition
and
then
the
type
definition
typically
is
sort
of
like
a
chess
on
schema,
but
they've
made
their
own
changes
to
that.
So
they
started
from
a
specific
version
of
chess
on
schema
and
then
added
their
own
stuff.
On
top
of
that,
so
I'm
not
sure
I
can
say
it's
compatible
or
not.
Probably
not
it's
not
exactly
the
same
thing.
A
G
Yeah,
so
the
context
behind
all
this
was,
I
was
working
on
a
to
enable
kubernetes
cluster,
auto
completion.
You
can
take
the
yama
language
server
from
red
hat
and
you
can
give
it
a
json
schema.
We
went
through
this
antoine,
you
can
give
it
a
json
schema
representation
of
the
spec
and
it
can
do
cluster
aware
auto
completion
with
crds
and
everything.
G
The
problem
is
that
we
have
json
schema
internally
and
we
do
that
conversion
to
open
api
and
then
back
and
forth
right.
So,
at
the
end
of
the
day
it
it
needs
the
json
schema
and
I
don't
think
the
red
hat
team
was
looking
to
to
make
the
language
server
work
with
open
api,
because
I
don't
think
they
ever
replied
to
our
issues
on
the
tickets
and
I
emailed
them
so,
okay,
yeah!
That's.
B
That's
confusing,
I
don't
know
why
they
would
not
want
to
address
quantities,
and
I
I
wish
we
could
work
with
them
to
make
sure
we
have
a
reasonable
solution
to
that
problem.
A
B
Yeah
we
have
a
product
internally
at
google
called
cloud
code
yeah
that
does
a
lot
of
these
and
it
kind
of
uses
the
same
technology,
but
I
mean
it's
understandable.
It
is
it's
made
specifically
to
target
kubernetes
so
and
I
think
that's
yeah.
Thank
you,
eddie.
That's
definitely
a
use
case,
though.
G
Yeah
well,
openshift
does
the
same
thing
actually
open
shift
in
their
front
end.
They
do
the
same
exact
conversion
where
they
add
the
one
of
field
and
then
munch
the
open
api
spec
into
json
schema
and
that's
how
openshift
does
it
with
their
clients?
G
B
Yeah,
it's
confusing
why
we
can't
conversion
that,
in
the
open
source.
G
So
but
but
the
reason
I
the
the
open
api
v3,
I
think
that's
actually
more
compatible
so
okay
and
that
that
solves
my
problem.
So.
B
Yeah,
let's
look
at
it
specifically,
so
what
definitely
is
true
is
that
open
api
exp,
open
api
visitor
exposes
more
information
than
the
open
api
v2
and
we
are
reducing
the
scope
when
we
convert
the
v3
into
the
v2
to
make
your
cut
all
happy,
and
so
that's
why
I'm
also
trying
to
drive
some
changes
through
the
coil
validation
and
any
and
I'll
probably
go
to
talk
to
coop
cuddle
and
see
cli
about
that.
B
But
I'm
trying
to
find
someone
to
work
on
this,
but
I
will
go
to
six
years
and
I'll
trust.
This
specifically.
B
B
Jeffy's
so
from
the
internal
discussion
we've
had
so
far,
it's
probably
gonna
happen
with
the
v3.
People
are
not
excited
about
us
changing
v2
at
this
point,
especially
if
we're
not
interested
in
maintaining
that
for
the
long
run.
So
as
we
create
v3
we're
probably
going
to
introduce
this
change
at
the
same
time,.
B
A
A
B
A
I,
like
the
check
marks
on
the
ga
document,
yeah.
B
A
A
A
A
I
agree
that
they
I
well
no.
I
don't
agree
that
there
is
that
we
don't
have
a
reason
to
drop
him.
There
is
enough
people
complaining,
but
also
agree
that
it's
not
it
might
not
be
enough.
I
just
feel
bad
about
like
telling
people
yeah.
If
you
come
with
good
enough
reasons,
we
might
do
it.
It
kind
of
feels
like
stalling.
I.
B
Mean
it's
also
121
the
change
that
omits
it.
I
think,
and
so
once
most
people
have
migrated
to
121,
maybe
we'll
stop
hearing
about
it.