►
From YouTube: WG Component Standard Office Hours 20200310
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
Okay,
good
morning,
everyone,
this
is
the
tuesday
march
10th
working
group
component
standard
office
hours.
This
is
our
hour
for
folks
to
drop
in
ask
questions,
get
help
with
what
they're
working
on.
B
B
Yeah-
and
they
did
mention
the
bot
mentioned
about
a
kubernetes
api
review.
So
do
I
need
to
create
a
cap.
A
We
yeah,
we
need
to
add
a
release,
note
to
your
pr
description
too
I'll.
Do
it
so
that
you
can
just
see
what
the
format
is.
B
A
Every
time
a
release
goes
out
like
the
first
draft
for
the
release.
Notes
gets
auto-generated
from
these
on
every
pr,
and
then
people
go
through
and
edit
it
to
clean
it
up.
B
B
A
So
if
you
go
into
edit
your
the
original
comment.
A
Yeah
you'll
see
it's,
it's
like
it's
like
the
code
quotes,
but
then
it
also
has
release
note
as
the
tag
after
the
first
one.
B
That
is
cool,
so
all
right,
so
once
this
gets,
I
got
an
ltm.
I
need
to
get
a
another
approval
right
from.
B
And
I
am
while
this
is
going
on-
I'm
gonna
edit,
the
document,
the
initial
contribution
document
that
you
put
together
with
everything.
A
Oh
actually,
do
you
want
to
squash
your
commits
too,
before
we.
A
B
All
right
and
then
I
need
to
again
once
I
push.
I
need
to
get
the
run
the
test,
and
then
I
need
to
get
the
approvals
again.
B
B
And
the
next
one,
I'm
thinking
that
a
couple
people
reached
out
to
me
about
this
one
and
I
did
give
them
some
materials
and
some
of
them
are
in
india
another
place
in
the
world.
They're
not
able
to
attend
the
meeting.
Someone
go
reach
out
to
them
and
say:
hey
want
to
smash
for
like
this
is
almost
complete.
So
I'm
gonna
give
this
as
the
baseline
or
like
take
a
look
at
it,
and
these
are
all
the
documentations
and.
A
That's
great,
if
you
can,
if
you're
interested
in
kind
of
helping
them
ramp
up
and
like
make
some
more
progress
on
this.
That
would
be
really
helpful.
B
Definitely
I
I
would
be
happy
to
help
as
much
as
I
know.
I
would
be
happy
to
pass
don
and
that's
how
I'm
gonna
learn
more
and-
and
I
will
come
back
for
questions.
B
C
B
B
Thank
you
mike
I'm
gonna
drop
out
now
I'm
gonna
run
to
another
meeting.
I
just
stopped
by
because
I
needed
to
ask
this
question.
I
told
my
colleague
that
okay
just
give
me
10
minutes
I'll,
be
back
so.
A
Hayley,
how
are
you
pretty
well
was
joe
able
to
swing
by.
I
haven't
seen
joe,
yet
savita
dropped
in
really
quick
just
for
a
couple
things.
A
C
A
Okay,
which
looks
good
she's,
the
only
thing
left
is
just
to
squash
the
commits,
I
think,
that's
ready,
so
you
can
actually
get
moving
on
the
rest.
Now
she
said
she
there's
a
few
people
in
india
too,
that
reached
out
to
her
who
want
to
work
on
it,
but
can't
come
to
the
meetings
so
she's
helping
them.
Oh
you're,
wrapped
up
yeah.
C
Wow,
that's
a
that's
a
great
responsibility
to
take
on.
So
thank
you.
Yeah!
That's
really
exciting!
Pretty
awesome!
Yeah!
It's
really
neat
to
see
how
people
like
so
quickly
well
past
the
torch.
I
think
that's
pretty
pretty
common.
Actually
not
it's
an
uncommon
thing
for
people
to
take
on,
but
it
seems
like
folks
who
come
here
to
contribute,
see
how
things
are
done
and
then
start
helping
others
accordingly,
yeah.
It's
really
good
because.
A
A
C
A
C
Yeah
you
and
I
haven't
chatted
in
a
while.
I've
been
working
behind
the
scenes
with
chris
hein,
who
oh,
I
didn't
even
mention
this
in
the
project
management
thing.
Chris
has
a
proposal
and
an
example
pr
open
for
getting
component
config
for
the
manager
into
controller
runtime.
C
I'm
not
sure
I
parsed
that
right,
yeah,
controller
runtime
creates
controllers,
it's
used
in
cube
builder,
and
it
has
a
thing
called.
The
manager
that
manages
all
of
the
control
and
the
manager
has
a
bunch
of
config
options
that
are
currently
either
they
hard
default,
or
they
have
a
few
flags,
and
basically
chris,
got
really
excited
last
kubecon
about
some
of
the
things
that
we
were
doing
with
component
config
and
with
config
custom
resources.
C
Some
of
like
how
these
ideas
overlap.
So
he
and
I
have
been
working
on
the
talk
for
kubecon
and
that
has
promoted
some
of
his
priorities,
so
he's
working
on
getting
component
config
into
controller
runtime
and
a
usable
in
coupe
builder.
C
A
Awesome,
that's
really
good.
Do
you
know
if
alex
is
looking
at
any
of
that
stuff
that
they're
doing,
because
that's
pretty
much
the
same
spaces
like
you
know,
controller
manager
is
not
running
on
that
framework,
but
it's
the
same
problem.
Space.
C
It
is
the
same
problem
space
I
probably
should
inform
alex
about
that
work.
I'll,
see,
see
him
on
the
pr
and
then
dm
him
the
link
for
chris's
proposal,
as
well
as
his
example.
C
We
have
this
thing
called
github
controller
that
we're
trying
to
improve
a
little
bit
for
github
c
use
cases
and
stuff
that
he
is
supporting
the
controller
runtime
work
too,
to
have
an
example
result
for
the
work
so
cool.
C
Cool
to
promote
that
yeah
into
something
internal,
but
basically
I
think
chris
just
built
an
interface
instead
with
a
bunch
of
getters
and
setters.
So
the
opponent
config
ended
up
implementing
the
interface
or
something
like
that,
and
then
he
has
a
step
for
using
the
interface
to
populate
the
existing
config
struct,
I'm
not
sure
from
the
design
perspective,
it's
totally
congruent
with
what
we've
been
doing,
but
it
it
does
walk
talk
and
act
like
a
component
config.
C
C
C
What
did
looking
for
some
feedback
proposal
with
controller
runtime
vince
just
replied?
I'm
gonna
mention.
C
C
I
I
did
just
mention
you
in
slack,
and
that
has
a
link
to
his
proposal
and
then
the
proposal
has
the
issues
with
the
documentation,
as
well
as
the
link
to
the
github
controller
example.
A
C
A
C
I
it's
not,
I
think
chris
just
went
for
the
thing
that
was
gonna,
be
least
breaking
right.
I
don't
think
that
the
the
existing
config
is
non-serializable,
but
I
it
probably
would
change
shape
because
it
was
not
being
like
it
did
not
have
you
know,
json
tags
and
all
that-
and
I
don't
know
if
it
was
intended.
You
know
really
obviously
doesn't
have
like
type
meta.
A
Unless
the
manager
sounds
like
this
manager
is
like
the
object
that
runs
at
runtime,
and
these
are
all
the
things
that
go
into
it.
A
Yeah
manager
configuration
is
really
the
is
really
the
component
config
at
least
an
interface
around
a
component
config,
and
all
this
other
stuff
is.
C
Yeah
there
is,
if
you
go
to
package
v1
alpha.
One
he's
got
a
stub
of
default
controller
configuration
which
has
a
spec
field
as
well
as
a
type
meta.
C
C
Because
the
the
trick
is
yeah,
this
is
probably
not
clear
since
it's
part
of
controller
run
time,
which
has
a
bunch
of
code
gen.
The
idea
is
like
to
make
it
easy
to
generate
different
component
configs,
where
people
could
actually
extend
these
things
for
their
own
use.
C
I
don't
know
if
that
kind
of
generation
and
composition
or
that
specific
generation
versus
having
like
a
multi-kind
composition
model
would
be
more
useful,
but
that's
you
know
up
for
this
is
what
chris
has
proposed
and
he's
put
work
into.
It.
C
So
yeah,
when
you,
when
you
look
at
the
default
controller,
configuration
section
and
then
see
you
know
how
he
mentions,
this
would
allow
a
controller
author
to
use
a
struct
with
any
config
that
supports
interface
compliance
structure.
So
the
kind
here
is
foobar
controller
configuration
right
and
then
the
spec
has
a
section
that
has
all
of
these
runtime
configurations
manager
interface.
C
And
so
this
is
a
bit
funky
right,
because
controller
runtime,
combined
with
coonbuilder,
you
know,
allows
people
to
create
these
bespoke
command-line
apis,
which
mostly
look
like
each
other
right.
They
have
all
these
shared
kind
of
origins
and
the
code
gen
produces
the
same
flags
hydrates
the
same
configs,
all
that
so
it's
a
this
is
an
interesting.
A
A
A
A
Sounds
like
we
can
expose
a
meta,
runtime,
config
type
that
can
be
embedded
similar
to
meta
v1
like
type
meta
and
object,
meta.
C
A
On
this
was
oh,
so
I
had
two
two
questions.
One
was
like:
where
does
it
live
is
met
at
the
right
place,
which
I'm
commenting
on
on
the
pr.
A
C
Alpha
one,
I
do
think
it
would
be
fine
and
he
he
doesn't
say
v1
he
he
says
alpha
here,
so
that
would
is
there
an
alpha,
v1
v1
alpha
one
that
seems
like
it
wouldn't
follow
inside
of
api
machinery,
because
we
would
be
into
v2
alpha.
Wouldn't
we.
A
Some
cases
you
can
have
like
a
we
can
add
at
a
kind.
I
don't
know
it's
it's.
I
don't
know
how
common
that
is,
because
you
can
always,
you
can
add
fields
and
kinds
to
existing
api
versions.
A
C
See
it's
a
little
weird,
a
v1
directory
inside
of
kubernetes
api
machinery
package,
api.
C
So
this
this
is
not
versioned.
This
is
this
is
internal
api
machine
that
you
build
code
on
part
of
the
leaky,
the
leaky
part
of
api
machinery,
versioning.
C
And
yeah
there's
commits
in
here
from
you
know,
four
months
ago,
interfaces
to
meta
v1,
how?
Why
is
that
version?
Let's
see
this
commit.
A
C
C
A
Yeah
this,
this
runtime
config
is
just
a
like
sort
of
default
config
for
controllers,
like
it's
the
base,
config,
that
every
controller
is
expected
to
need.
A
The
thing
that's
sort
of
interesting
about
this
is
is
controller,
run
times
the
separate
like
projects
inside
kubernetes,
so
putting
a
lot
of
these
options
in
the
core
repo.
Instead
of
in
the
separate
projects,
repo
doesn't
really
feel
aligned
with
the
pre-existing
layering
well.
C
We
have
the
staging
repo
with
this
api
machinery
portion
as
a
future
breakout
yeah,
but.
A
C
And
actually,
my
first
intuition
as
to
where
that
should
be
imported
from
is
actually
a
controller,
runtime
repo.
So
right.
A
C
C
A
C
A
It
is
yeah,
so
it
could
be
useful
there.
I
think
it
would
be
okay
to
have
something
there
and
then
have
controller
runtime
depend
on
that.
Maybe
the
other
angle,
the
negative
angle,
is
like
we
haven't
adopted
controller
runtime
as
the
way
we
write
controllers
in
the
core
project.
A
Right,
so
if
we
sometime
in
the
future
did
that
the
easiest
thing
would
be
to
just
then
depend
on
controller
runtime
for
it,
which,
once
it's
stable,
wouldn't
break.
Anything
it'd
be
like
a
little
bit
of
extra
complexity
like
due
to
you
know
historical,
like
development,
but
it's
not
the
end
of
the
world,
but
if
we
do
it
now,
it's
sort
of
a
leading
decision
that
I
don't
know
if
we
should
be
making.
A
C
A
C
Maybe
alias
it,
I
would
just
I
just
wanna
and
I
I
would
also
be
quite
keen
actually
on
looking
at
some
sub
packaging
techniques
for
go
modules
for
component
base,
because
our
our
repo
does
not
have
to
be
our
unit
of
packaging
with
things
like
go
module
proxies
now.
Oh
that's
interesting.
A
It's
still,
how
do
you
you're
running
if
you're
doing
separate
packages
in
there?
How
do
you
do
releases?
Because
you
we're
still
version
tagging
and
get
yeah
that
that
gets
weird,
because
that
still
covers
the
repo.
C
The
so,
for
instance,
with
the
this
runtime
config
living
in
controller
runtime
and
then
say
that
the
controller
manager,
you
know
eventually
start
to
look
really
similar
to
it
and
we
decide
that
we
want
to.
You
know,
converge
those
apis
into
the
same
one.
C
Then
at
that
point
it
might
go
into
component
base.
If
the
community
has
not
adopted
controller
runtime
right,
because
you
you
would
just
you
know,
looking
to
avoid
somebody
having
to
potentially
vendor
or
even
just
you
know,
download
and
build
controller
runtime
in
order
to
just
get
that
type
yeah.
I.
C
Okay
yeah,
so
the
unfortunate
side
of
that
is
that
you
often
have
to
look
to
find
a
place
to
land
code
because
it's
not
trivial
to
create
new
repositories.
Yes,
but
in.
A
C
Yes,
yeah,
and
we
were
just
really
speaking
in
hypotheticals
in
the
future,
but
it
totally
makes
sense
just
in
there.
Okay.
C
C
You
still
get
versioning
for
all
of
them,
but
it's
some
of
them
may
bump
with
no
changes,
which
is
the
same
exact
thing.
But
then
you
can
still
use
the
unit
of
package.
C
C
Yes,
but
basically,
what
I'm
getting
at
is
all
of
those
get
the
same
tag
bumps,
regardless
of
whether
or
not
the
chat
the
code
is.
B
C
Oh
okay,
right!
So
in
the
same
way,
if
you
had
multiple
sub
packages
in
a
single
repository
right,
you
could
just
version
it
as
one
thing,
but
then
you
can
still
use
the
individual
packages.
A
Yeah,
I
liked
it
wouldn't
really
work
for
us
both
because
we
write
in
a
language
that
doesn't
support
it
and
because
I
think
our
project
is
the
way
our
project
treats.
Versioning
is
more
complex
than
just
like
the
interfaces
and
code,
but
I
really
liked
elm's
automatic
semver.
A
C
Be
pretty
wild,
I
don't
I
don't
know
what
tools
are
available
to
do
that
kind
of
analysis.
Oh
yeah.
C
C
C
A
I
don't
really
hang
out
on
twitter,
no,
not
a
big
social
media
person.
I'll
just
read
it
to
you.
It's
it's
pretty
fair.
C
C
C
Yeah
and
like
it's,
it's
kind
of
funny
later
down
in
this
thread,
alex
ellis
was
like
yeah.
There
was
this
thing
that
I
worked
with
in
the
inlets
operator
and
stefan
told
me
about
changes
to
codec
factory
that
one
of
his
colleagues
made,
and
he
was
kind
of
talking
about
me.
But
then
I
was
thinking
and
I
was
like
I
didn't-
make
any
backwards,
incompatible
changes
to
codec
factory,
and
so
I
I
looked
it
into
it
and
I
didn't,
but
somebody
else
did,
and
so
I,
like,
I
worked
all
this.
C
I
put
in
all
this
work
to
like
not
break
downstream
users
and
then,
in
the
same
exact
release
we
had
changes
like
I
might
as
well
have
just
not
done
that
work
right
since
the
the
dependency
broke
users
anyway,
right
for
a
different
portion
of
the
of
the
interface
so
alex
was
using
a
raw
codec
factory,
something
with
no
conversions
basically
and
they
removed
a
global
that
was
containing
a
member
or
a
struct
and
moved
it
to
a
method
where
you
have
to
initialize.
It.
C
C
Clients
yeah,
because
all
of
that
stuff,
the
only
interface,
is
the
rest
api
right.
A
C
A
Yeah
now,
there's
no
like
formal
control
that
enforces
that.
So
that's
maybe
something
to
explore
and
there's
obviously
like
all
the
downgrade
issues
once
you
start
using
it.
C
C
C
Yeah,
it's
30's
thanks
for
sitting
around
and
sharing
with
me
about
all
this
stuff
and
thanks
for
reading
through
the
proposal,
chris
is
really
going
to
appreciate
your
feedback.
There
yeah.