►
From YouTube: Kubernetes KubeBuilder Meeting 20201203
Description
KubeBuilder Meeting for 2020/12/03. See https://sigs.k8s.io/kubebuilder for more details.
A
All
right,
hello
and
welcome
to
the
q
builder
controller,
runtime
and
controller
tools
meeting
for
thursday
december
3rd
2020..
As
a
reminder,
this
is
being
recorded,
so
don't
say
anything
you
don't
want
to
be
recorded
for
all
posterity
and
uploaded
to
the
internet.
A
All
right
looks
like
we
have
a
few
things
on
the
agenda
today.
I
think
the
first
is
from
madeirio
about
go
module
support,
so
why
don't
you
take
it
away.
B
Okay,
yeah,
we
have
detected
that
we
weren't
fully
applying
go
modules
when
we
changed
to
version
2
of
kuby
builder
and
some
tools
weren't,
integrating
correctly
with
it,
for
example,
field
race.
The
other
day
that
packager
doesn't
understand
that
we
are
in
version
2.3.1.
B
So
I
wanted
to
ask
if
it
would
be
a
good
idea
to
change
the
change
needed
to
to
fulfill
with
person
with
code
modules
so
that
the
version
2
are
recognized
is
only
to
change
the
module
path
to
include
a
version
2
at
the
end,
and
there
is
a
there-
is
a
link
in
the
agenda
that
shows
which
would
be
the
the
change
required
and
we
could
apply
it
to
old
versions
of
q,
builder
q,
builder
2.3.1,
or
maybe
even
older
versions,
so
that
the
tools
recognize
those
versions,
and
I
wanted
to
know
which
is
the
position
of
the
rest
of
the
members
about
that.
A
A
We're
doing
it
for
version
three
anyway,
so
makes
sense
to
see
if
we
can
export
it
to
version
two,
because
previously
it
wasn't
really
go
guttable.
So
I
don't
think
it's
a
problem
of
breaking
import
paths
as
much.
I
think
there's
only
a
few
people
that
are
importing
it
directly
and
so
they'd
probably
be
fine,
but
I
think
we
probably
want
to
hear
from
the
operator
sdk
folks
on
that.
First.
A
B
D
Just
one
question:
do
you
really
need
to
care
about
what
was
released
about
it?
B
I
think
that
it
would
be
nice
to
to
have
like
a
a
dot,
as
at
version
2
release,
at
least
because
we
are
going
to
have
like
version
one
releases
which
are
recognized
without
anything
in
the
past
and
also
then
version
3
releases.
That
will
be
a
bit
strange,
but
it's
not
really
required.
If
that's
the
question,
because
version
3
has
already
that
fixed.
B
A
E
F
F
Yeah
like
if
we
just
did
a
version
three
release
like
is
it
that
we're
not
ready
to
do
a
version
three
release
and
then
that's
our
hesitation
there
right,
because,
as
soon
as
we
do
a
version
three
release,
it
should
be
go,
get
a
ball
for
version
three
and
just
version
two
I'll
not
be
go.
Get
a
ball
right
exactly.
F
And
so
then
the
so.
The
solution
is
to
just
change
the
paths
to
be
have
v2
in
them
for
the
v2
branch
right.
B
D
B
D
F
Is
that
release
v2
or
is
that
something
else.
B
B
C
C
Great,
can
you
all
see
that
yeah
yeah
next
week
120
is
supposed
to
be
released,
kubernetes,
120.
C
and
assuming
that
zero
seven
zero
gets
released
either
this
week
or
early
next
week,
I'd
like
to
ask
what
the
feasibility
is
of
merging
this
pr,
that's
currently
open,
which
bumps
kubernetes
to
1.28
controller
runtime
and
then
cutting
a
0
8
0
release
after
that.
C
Getting
kubernetes
releases
into
the
sdk
as
soon
as
possible
and
having
not
special
releases
but
considering
kubernetes
release
dates
in
hourly
cycle
so
that
we
can
have
a
release
pretty
quickly
after
kubernetes
has
been
released,
and
so
I'm
wondering
if
we
can
do
the
same
thing
in
controller
runtime
and
cut
like
a
kubernetes
release.
C
G
Okay,
it's
me
yeah.
I
was
actually
wondering
because
we
never
managed
to
do
a
0.7
release.
Actually,
we
only
have
if
us,
which
is
the
version
that
would
have
cube
1.19
if
it
would
make
sense
to
just
skip
having
a
release
based
on
cuban
19
entirely
and
instead
directly
go
to
120,
because
the
release
of
that
is
basically
just
a
couple
of
days
away.
Right.
A
I
mean
I
think
we
should
probably
have
the
0.7
release
with
one
119,
just
like
people
a
lot
of
times,
there's
there's
a
segment
of
users
that
very
much
like
to
have
particular
versions
matching
for
whatever
reasons,
sometimes
there's
changes
in
kubernetes.
Sometimes
it
revolves
around
policy
or
whatever.
So
I
think
it's
probably
a
good
idea
to
have
a
119
release.
A
I
think
just
at
this
point
the
only
thing
blocking
0.7
and
also
yeah.
The
the
main
reason
we
haven't
released
0.7
yet
is
was
an
incompatible
change
that
we
haven't
quite
gotten
a
fixed
merged
for
yet
just
due
to
everything.
A
H
I
think
eric
has
a
pr
for
the
web
book
change
that
we
need
to
make.
I
just
haven't
had
time
to
review
it
yet
before
it
really.
H
Because
we
didn't
get
any
update
on
the
old
one,
I
think
all
right
can
you.
A
It
it
is
because
it
like
it,
it
removed
some
stuff.
We
were
depending
on
or
1.19
removed,
some
machinery
we
were
depending
on,
and
so
we
have
to
fix
this
somehow,
like
otherwise
there's
a
bit
of
functionality
that
just
goes
away
in
controller
on
time,
and
that
would
be
not
great
for
a
release.
D
E
D
I
doesn't
know
the
bug
I
just
it
was
my
wonder.
H
C
Yeah
cool,
so
it
sounds
like
we
are
okay
with
cutting
zero,
eight
zero,
assuming
everything
goes
well
that
pr
that's
currently
open
for
merging
the
120
bump
and
that
there
are
no
crazy
changes
that
need
to
be
made
pretty
soon
afterwards.
That's
correct.
C
B
All
right
back
to
adirio,
I
think
the
next
topic
in
the
hinda
is
pretty
long.
I
don't
know
how
much
time
does
field
need
for
the
demo.
If
he
doesn't
need
that
much
time
we
could
go
first
to
the
demo.
He
wants
to
show
and
then
go
for
the
other
one.
What
do
you
think
phil.
F
I
can
keep
it
short,
maybe
I
it
really
depends
on
how
many
questions
there
are,
but
there
probably
won't
be
that
many
since
it's
a
follow-up,
so
let
me
see
what
I
can
do
here.
F
Sorry,
just
sharing
my
stuff
here.
E
E
E
F
All
right,
so
so,
the
a
couple
weeks
ago,
I
showed
just
kind
of
the
like
a
the
idea
of
creating
this,
like
config
gen
program,
which
generates
the
config
for
you,
and
so
the
idea
there
was
like
you
can
in
the
project
file.
Have
this
this.
F
You
know
spec
that
says,
okay,
what
components
do
you
want
to
generate
and
what's
the
image
and
what's
the
project
name
and
stuff
like
that,
and
then
it
does
and
then
it
generates
the
config
for
you
rather
so
that
you
can
kind
of
specify
like
the
experience
is,
if
you
just
want
to,
you
know
enable
certain
things
you
can
just
set
them
in
the
project
file
and
it
sets
the
stuff
up
for
you
and
so
there's
a
question
about
well,
how
do
I
think,
how
do
plugins
work
and
then
how
do
how
does
this
fit
in
with
customizing
that
sort
of
stuff?
F
So
this
is
written
as
both
a
standalone
and
as
a
customized
plugin.
It
works
in
both
the
same
binary
works
in
both,
and
so
this
is
just
a
demo
of
what
it
would
like.
Look
like
with
customize.
F
So,
in
this
case,
you're
running
as
a
generator,
for
instance,
and
then
so
that
same
sort
of
the
same
thing
that
goes
in
the
project
file
I
put
in
here
with
a
couple
more
features,
so
I
have
like
a
dev
and
and
a
prod
version,
and
so,
if
you
do
customize
build,
you
know
with
enabling
plugins
here
you
can
see
that
that
it
builds.
It
runs
it
as
a
customized
plugin,
and
so
this
means
that
the
questions
around
like
well.
What?
F
If
the
user
wants
to
you,
know
patch
this
thing
or
have
the
same
level
of
customization?
They
do
with
customize.
They
do
and
it's
very
familiar
how
they
do
it
right.
They
would
just
add
patches
here
or
that
sort
of
thing
and
then
and
then
the
question
of
like
how
do
we
do
plugins?
F
How
do
plugins
work
differently
and
kind
of
my
recommendation
would
have
been
like
would
be
kind
of
do.
A
plugins
is
actually
as
a
transformer
right,
and
so,
if
you
have
like
operator
sdk
specific,
like
plugins,
you
could
do
a
transformer
and
it
would
allow
you
to
to
have
a
lot
more
flexibility
with
what
you
do
by
not
having
to
to
work
around.
F
What's
there
and
uncommon
things
you
can,
you
can
delete
resources,
for
instance,
or
add
resources
completely
and
then
just
a
quick
demo
of
like
some
features,
I've
added
since
then
I
have
a
dev
and
a
prod
version,
the
dev
version.
I
have
kind
of
this.
This
pop
version
that
version
I've
added
this
development
options
to
generate
certs.
F
So
so
this
is
kind
of
a
pool.
I
think
where
we
can
automatically
generate
the
certificate,
as
part
of
you
know,
creating
the
configuration
in
the
plug-in,
and
so
you
can
wire
in
the
ca
bundle
in
all
the
places
it
goes
and
then
and
then
create
the
secret
and
wire
in
the
secret.
F
So
that's
one
nice
capability
for
development.
We
can
just
generate
a
certificate,
that's
now
or
long,
and
then
users
don't
have
to
figure
out
how
to
install
certificate
manager
or
something
when
trying
to
like
develop
their
web
hook
right
and
then.
The
second
thing
I
added
was
a
component
config
option,
so
you
can
specify
your.
F
You
can
specify
your
your
component
config
in
a
file
and
then
reference
it,
and
then
it
will
go
ahead
and
generate
the
config
map
and
then
also
wire
this
into
the
appropriate
places.
So
in
this
case
manager
config
you
can
see,
then
it
gets
wired
into
the
deployment
and
again
this
is
in
that
vein
of
like
what
is
a
consumer
or,
as
user
of
who
builder
do
I
want.
I
just
want
to
give
it
a
component,
config
file
path
and
then
have
it
figure
out
how
that
gets.
F
You
know
wired
in
rather
than
uncommenting,
for
your
four
different
lines
in
a
customization
animal
to
make
it
happen.
F
So
that's
that's
my
demo.
You
can
use
it
as
cut.
You
can
use
it
in
as
a
customized
function
now
and
so
patches
and
everything
else,
just
kind
of
works,
the
way
it
does
in
customize
and
a
couple
more
features.
D
A
I
I
I
think
it's
in
the
last
recording,
which
I
just
haven't
uploaded,
because
I've
been
a
little
bit
behind
on
uploading.
All
of
those.
F
Sure
I
can
do
like
when
I
think
this
is
getting
quote
relatively
close
to.
I
think
what
would
be
available
as
an
opt-in
possibility.
So
I
think
the
the
the
modules
thing
was
a
blocker
because
it
breaks
the
libraries
I'm
using,
but
once
that's
fixed.
I
can
probably
get
this
in
the
state
where
it's
ready
as
a
proposal
that
could
be
accepted,
and
as
part
of
that
I
can
do
a
like
20
minute.
Walkthrough
of
all
the
various
aspects
of
it
as
a
standalone
thing.
D
F
Yeah,
how
you
extend
it
I
can,
I
can
do
it
as
part
of
the
demo.
I
can
actually
write
my
own
extension
to
do
some.
You
know
something
some
toy
thing,
but
that
would
delete
a
resource,
add
a
resource
and
modify
a
resource
in
place
or
something
like
that,
and
then
I
I
can
check
that
in
as
an
example
for
an
extension
as
part
of
the
pr.
A
You
right
looks
awesome
so
far.
I
look
forward
to
it
all
right
all
right
back
to
adiria
for
the
last
thing
on
the
agenda.
I
think.
B
Yeah,
okay:
this
is
going
to
be
a
bit
longer,
though
we
have
been
discussing
lately
about
the
project
file
and
which
data
should
we
store
there,
and
there
were
some
issues
raised
lately
about
trying
to
store
everything
needed
to
recreate
the
project
in
a
future
just
by
having
the
project
file
comfy
and
there's
an
also
another
aspect
that
we
have
talked
several
times
that
we
want
to
add.
But
we
don't
get
to
it
yet
that
it's
about
adding
external
resources
and
using
them
to
generate
controllers
for
them
or
we
can't
actually
do
for.
B
But
the
thing
is
that
we
have
been
like
adding
additional
fields
to
the
project,
config
files
to
the
sections
where
the
resources
are
listed
and
it's
starting
to
get
a
bit
messy
and
right
now
we
have
like
two
different
representations
of
the
resource,
the
one
that
we
use
internally,
that
is
in
the
package
model
resource
and
the
one
that
we
use
for
the
configuration
file
and
there
is
like
having
two
different
models:
one
for
the
configuration
file
and
other
for
the
for
the
internal
use,
for
something
that
is
the
same,
doesn't
seem
to
be
a
very
good
decision
and
and
will
probably
be
probably
be
problematic
when,
when
we
went
to
the
second
phase
of
the
of
the
plugins
system
to
the
plugin
phase
two,
so
one
of
the
issues
I
raised
was
to
try
to
have
a
single
view
of
of
the
of
the
resource
and
use
it
for
both
storing
it
in
the
config
file
and
also
use
it
internally.
B
B
B
Yeah
to
this
one,
so
basically
the
the
test
project
files
will
be
something
like
this.
That
will
add,
like
the
person
used
for
the
api,
is
if
it's
namespace.
This
is
info
that
we
don't
have
currently
if
the
controller
was
deployed
also
which
webhooks
will
were
scaffolded
to
and
which
person
of
the
webhook.
B
A
B
Right
now,
no
okay,
so
the
idea
was
to
to
apply
something
like
this
model.
The
group
diversity
and
the
kind
are
the
only
things
that
are
currently
being
stored.
If
I'm
not
mistaken,
maybe
pluralistic
plural
is
also
been
stored.
D
B
B
I
think
it's
quite
mature
and
we
may
be
able
to
get
it
merged
before
version
3..
So
we
won't
need
a
version
four,
but
I
don't
know
what
the.
B
What
do
you
think
about
this?
If
we
are
rushing
a
bit
and
we
may
delay
the
version
three
or
and
you
prefer
to
go
for
a
version
4
or
if
we
should
just
add
some
fields
from
for
now
that
adding
fields
is
not
a
problem
and
redesign
it
a
bit
later
for
for
version
4
or
I
don't
know
or
or
not,
added
them
at
all
that
are.
That
are
the
three
options.
Basically,
so
I
would
like
to
to
have
some
feedback
from
you
guys.
A
And
just
to
be
clear
for
the
people
who
anybody
who
doesn't
quite
have
context
as
well,
that
would
be
ver
specifically
version.
Four
of
the
project
file,
like
adirio,
said,
which
doesn't
necessarily
mean
we
need
to
release
version
four
of
cube
builder.
As
long
as
we
keep
support
for
project
version
three,
so
we
could.
A
We
could
release
support
for
version
four
format
in,
like
you
know,
3.1
or
3.2
or
whatever,
since
we've
we've
decoupled
the
project
file
format
from
the
general
project
format,
we
can
make
those
changes
without
as
long
as
we
don't
remove
version
three,
it's
not
a
breaking
change,
so
we
can
play
around
with
stuff
like
this.
D
But
if
he,
if
here
we
change,
for
example,
one
of
the
association
was
move
the
crg
versus
for
inside
of
the
api.
So
would
you
not
easy
consider
a
break
change
because
you
are
moving
around
things?
For
example,
if
we
release
it
now,
the
victory
we
have
the
crg
version.
Indeed,
the
web
hook
version
in
the
route.
D
D
Then
we
can
rerun
the
commands
and
you
can
create
a
project
from
the
scratch
probably
to
to
do
things
like
a
creature
tool
to
help
them
to
migrate
in
the
next
for
the
next
plugins
for
the
next
versions,
because
it
is
a
common
complaint
for
the
sdk
project
to
users
like
it's
too
hard
to
upgrade
it's
very
hard
to
take
a
delicious
version,
so
something
like
to
make
their
life
easier.
D
But
if
you
don't
have
the
data
that
they
put
to
create
a
project,
it's
like
impossible
to
to
try
to
so
tooling
that.
But
I
fought
in
something
very
simple
engineering.
A
lot
of
suggestions
engineers
raise.
D
B
Just
to
add
a
a
few
comments.
Currently,
we
are
having
we
have
like
the
version,
2
project
file
and
the
version
3
alpha.
So
we
are,
we
don't
have
a
a
version,
3
stable
project
version,
so
actually
we
could
add
or
modify
fields
without
breaking
backwards
compatibility
if
we
do
it
versus
before
we,
we
release
version
three
that
that.
G
B
Mean
if
we
merge
it
before
the
first
alpha
of
the
person
three
release,
but
it
does
mean
that
we
should
do
it
before
the
first,
a
stable
version
of
q,
builder
version
3.
B
or
the
other.
The
other
option
would
be
a
try
to
make
whatever
right
now
is
in
version
3,
alpha
project
version
into
a
stable
product
version
and
a
leaf
for
version
4.
The
changes
that
we
are
proposing
right
now,
basically,
the
other.
The
only
two
changes
that
are
a
a
breaking
chains
against
the
version
3
alpha,
not
against
the
version
2,
because
none
of
them
is
breaking
chains,
were
against
the
version.
Two
are
a
this
version
field
over
here
that
it
was
in
the
root
object
instead
of
inside
the.
B
D
Maybe
maybe
wrong,
but
I
believe
that
she,
the
expectation
the
problem,
is
that
the
concern
that
she
maybe
the
others,
king
reeves,
is
like.
We
would
like
to
have
the
three
gold
plugins
table
as
soon
that
we
bump
the
supports
to
1.20
the
the
kubernetes
api.
She's
is
the
reason
why
k-
maybe
it's
not
possible,
then
I'll
cut
those
before
the.
B
Actually,
the
changing
the
project
file
only
in
the
project
file
is
mostly
done
by
this
by
this
pull
request
by
camilla.
She
she
adds.
I,
I
think
she
doesn't
add
all
the
fields
that
you
see
that
you
have
seen
in
in
that
image,
that
I
will
update
with
the
with
a
new
name.
D
B
D
B
B
B
I
removed
basically,
I
removed
the
gvk,
which
was
being
used
to
to
have
like
the
second
representation
of
the
resource,
and
I
made
a
the
config
file.
Was
this
here
directly
have
a
directly
import
the
resource
class?
So,
with
this
pull
request,
we
will
be
able
to
only
have
one
representation
that
I
think
that
it's
what
it
really
will
raise
problems
in
a
future,
but
this
requires
to
change
the
project
file
because
there's
too
much
information
inside
the
resource
type.
D
I
think
if
we
try
to,
I
think
he
okay,
I
agree
that
you
just
cannot
be
a
blocker
because
he
the
expectation
for
don't
break
the
expectation,
but
she,
if
he
will
try
to
do
that
before
the
coach,
maybe
the
strategy,
the
the
best
strategy
is
like.
We
try
to
manage
what
changes
the
project
file
layout,
which
is
something
which
can
cause
break
chains
engineers
improvements
in
the
codes.
It
should
be
easier
to
keep
him
attained
and
to
grow
this
stuff.
A
Okay,
all
to
to
clarify
the
delta,
the
like
the
externally
visible
delta,
from
from
camilla's
changes
to
your
changes.
Those
are
just
additions
of
fields
right,
yes,.
A
A
We
might
just
want
to
make
it
one
one
field,
but
like
we
assuming
that's
the
case,
we
could
merge
camilla's
pr
now
and
then
we
could
merge
your
pr
afterwards
and
since
your
pr
wouldn't
be
a
breaking
change
the
project
file.
Since
it's
only
adding
a
new
optional
field
to
the
project
file,
we
could
we
could
do
that
later.
B
D
D
B
I
agree,
but
I
I
wanted
to
to
note
it
that
for
the
developers
like
operator
sdk
or
like
that
it
would
be
a
small
they.
They
may
not
be
importing
that
class
at
all.
They
will
probably
not
be
importing
that
class
at
all,
but
it
could
be
breaking
in
that
sense
from
the
library
point
of
view,
but
from
the
break
point
in
the
view,
it
wouldn't
be
broken,
break
a
break
in
change.
I
A
C
B
B
I
don't
know
if
I
I
think
I
show
you
an
example
of
how
the
project
file
would
look
like
right.
D
D
A
B
Is
that
right
now
we
have
version
v1
so
as
they
are
in
the
root
field,
we
have
web
hook
version
and
crd
version,
but
after
that
it
would
be
inside
an
object.
So
I
my
suggestion
was
to
remove
crd
because
it's
no
longer
needed
because
there's
no
conflict,
it's
ap
version
and
the
same
with
workbooks,
because
it's
going
to
be
webcook's
person.
D
But
I
I
think
it's
more
intuitive
if
you
became
the
same
name,
because
it's
the
name
of
the
argument.
I
know
that
you
should
not
do
the
model
basis
in
the
ux,
but
if
to
keep
him
on,
then
maybe
using
the
same
name.
You'll
be.
You
know
easier
to
know.
Okay,
it
is
the
webbing
book
where
some
cheese
is
the
syringe
person.
You
know.
A
A
So
it's
a
little
bit
of
an
overloaded
term.
I
it's
not
the
end
of
the
world.
If
we
go
with
this
to
me,
though
I
just
before
we
go
down
into
the
the
nitty-gritty,
though,
do
people
have
comments
on
the
general.
C
Yeah,
I'm
personally
fine
with
all
this
as
long
as
both
camilla's
and
adrian's
prs
line
up,
because
I
know
that
there
was
a
lot
of
back
and
forth
about
how
this
should
look
and
that
adrian's
pr
is
yeah
not
not
breaking,
except
in
that
one
edge
case.
D
Yeah,
we
probably
until
modding-
I
I
think
I
we
have
been
spoken-
a
lot
adria
angie.
We
tried
to
reach
like
a
consensus
in
the
names
and
everything
that
she
it's
good
for
v3,
because
I've
been.
I
thought
that
it
is
was
the
best
approach.
You
know.
If
you
read,
we
came
a
consensus,
so
probably
shape
enough
for
the
community
as
well,
so
but
yeah.
I
think
we
make
a
lot
of
package
efforts,
but
we
are
closer
to
closing
the
captur.
B
I
think
that,
right
now
the
only
discussion
was
in
diversion
names
that
I
just
raised
and
the
other
thing
that
we
had
like
that
we
were
having
debating
about
it
is,
if
the
ones
that
they
propose
are
really
needed
or
aren't
needed
to
to
detect
external
types
and
that's
a
problematic
question,
because
right
now
we
don't
allow
users
to
define
external
types.
D
Okay,
my
my
opinion
I
was
sexy.
My
intention
was
like
he
started
to
do
this
step
by
step.
You
know,
let's
track
the
info,
that
is
inputted
and
after
we
think,
if
we
should
redesign
or
not
like
one
step
per
time,
I
was
like
resonating
because
he
his
idea
of
he
supports
external
types.
It's
really
nice.
D
We
need
to
address
that,
but
we
don't
have
the
helper
to
do
that
yet,
so
I
thought
that
maybe
that
field
should
be
better
and
when
we
have
the
helper,
however,
has
one
one
way
has
one
one
way
like
here.
If
we
change
the
the
search
of
the
comments
has
one
scenario
where
maybe
we
would
need
to
choose
data
before
the
victory
to
try
to
recreate
the
project
from
the
scratch.
D
B
D
B
But
I
think
that
that's
the
only
point
that
we
that
we
have
to
agree
on
before
this
well
before
mine
gets
marred.
I
think
this
one
is
pretty
close
to
getting
married.
I
think
we
have
gone
like
through
several
yeah
look
103
comments
in
in
this
full
request,
so
I
I
think
it
was
pretty
close
to
getting
merged
this.
One
just
needed
like
the
approval
of
the
of
the
new
design
and
whatever
you
your
feedback
was
in.
In
that
sense,.
A
All
right,
so
we
only
have
two
minutes
left,
so
I
think
the
tldr
of
that
discussion
was,
I
think,
it's
let's,
let's
get
camilla's
merged,
let's
we
we
might
be
able
to
get
the
the
additional
pr
merged
as
well.
A
It
would
probably
be
a
good
idea,
considering
the
corner
cases
just
raised,
or
at
the
very
least
you
know,
get
get
the
externally
facing
changes
for
like
external
types
merged
in
a
second
follow-up
pr
or
something
just
to
avoid
user-facing
change
changes
coming
in
but
yeah,
so
I
think,
get
camille
is
merged
and
then
we
can
figure
out
the
rest
later.
But
it's
probably
a
good
idea
to
get
the
external
merge
based
on
the
feedback
that
or
the
the
corner
case
that
you
just
listed.
A
I
B
Yeah
the
idea
well,
eric
was
talking
about
the
releases
of
controller
runtime,
which
is
not
the
same
project,
so
it
shouldn't
affect
that
one,
but
it
would
affect
the
release
of
the
version
3
of
qbuilder,
but
that's
not
as
close
as
the
other
ones.
So
I
think
that
we
will
be
able
to
get
this
merged
before
before
the
release
and
also
we
still
have
to
go
through
the
alpha
or
maybe
even
better
process
for
the
version.
B
Three,
we
haven't
still
released
an
alpha
version,
so
I
I'm
pretty
sure
we
will
be
able
to
get
this
merged
before.