►
From YouTube: WG Component Standard 20190813
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
All
right,
good
morning,
everyone
welcome
to
the
tuesday
august
13th
working
group
component
standard
meeting.
We
had
a
great
meeting
last
week,
kind
of
brainstorming,
a
bunch
of
different
component
config
stuff
with
sig
cluster
life
cycle
and
some
of
the
api
machinery.
Folks,
you
can
see
the
notes
for
that
and
the
link
in
our
meeting
notes
and
the
youtube
recording
for
that
meeting
is
also
up
in
our
playlist.
A
So
we
have
kind
of
a
short
agenda
today.
I
guess
let's
just
get
started
with
that
alex.
If
you
want
to
jump
in
on
that.
B
Yeah
sure
so,
hi
everyone,
my
name,
is
alex
quite
new
contributor
to
the
project.
Lee
was
so
kind
to
give
me
a
little
on
contributor
onboarding
last
time,
and
he
suggested
that
I
pick
up
this
full
request,
which
was
initially
created
by
lucas,
so
I
basically
just
took
it
added
some
tests
made
it
all
green,
but
I
do
have
a
couple
questions
in
regards
of
how
this
all
ties
into
like
the
bigger
effort-
and
I
think
this
was
also
one
of
ross's
questions
in
the
comments.
B
B
What
has
been
discussed
in
this
regard
and
any
plans
on
you
know
how
this
functionality
is
to
be
supported
in
the
existing
serializers
or,
like
any
other
points
where
you
know
this
could
be
a
foundation
of
so
yeah.
Well,
I'm
still
trying
to
wrap
my
head
around.
All
the
you
know,
concepts
around
decoding
and
coding.
I
think
I'm
missing
a
couple
of
big
picture
points.
I
just
wanted
to
know
if
yeah
you
guys,
can
show
some
insight
on
that.
C
Yeah,
so
I
had
like
five
minutes
to
look
at
this
earlier
from
what
I
remember.
The
meta
factory
here
is
allowing
it's
currently
used
in
kubernetes,
so
this
patch
here
is
porting
it
over
into
an
api
machinery.
Runtime
library,
basically
still
can
be
used
in
other
places.
C
What
the
matter
factory
does
is
it
allows
you
to
figure
out
what
the
gvk's
are
for
a
stream
of
multiple
yammel
docks
and
basically
it's
a
primitive,
that's
necessary
if
you're
going
to
be
doing
things
in
the
style
of
kubernetes,
where
you
are
trying
to
point
out
a
stream
of
bytes
that
potentially
contain
multiple
things,
different
kinds
and
then
now
the
output
of
the
meta
factory
is
a
list
of
gbks
and
then
you
can
switch
on
it
conditionally
to
do
various.
You
know,
controller
component,
config
or
api
related
things.
C
Hopefully
that's
clear
and
then
so.
The
desire
here
is
to
move
the
meta
factory
from
a
place
where
only
kubityam
can
use
it
into
the
api
machinery
staging
dependency
and
there's
a
couple
benefits
to
that,
one
being
that
it's
no
longer
in
the
core
tree.
C
That's
not
part
of
the
court
tree
to
then
do
like
multi-dock
stuff
and
so
at
the
bottom
of
this
patch.
If
you
were
to
just
scroll
down
to
the
to
the
last
comment,
I
think.
C
Ross
here,
I
think,
was
the
person
who
asked
if
we
are
maybe
gonna
extract,
split
yaml
documents.
I
would
imagine
that
that
would
probably
go
into
component
base.
Do
you
have
any
thoughts
on
that
ross.
D
Yeah,
so
for
me
this
is
like
trying
to
build
from
bottom
way
up,
rather
than
top
down
approach,
and
basically
I
see
something
that
amounts
like
a
network
adapter
being
introduced
in
a
system
with
no
networking
layer.
Yet
so
it's
a
little
bit
unclear
to
me.
D
C
D
Yeah,
I
think
that
probably
I
want
to
see
more
things
like
split
the
ammo
documents
and
more
things
being
extracted
just
to
see
what
the
the
interface
of
the
thing
is
going
to
be
like
how
it's
supposed
to
be
used
by
third
parties.
C
D
A
C
Yeah,
I
think
you
know
it's
like.
C
We
should
probably
have
a
better
habit
of
that,
looking
at
lots
of
like
tools
out
in
user
land
right
with,
like
argo
and
flux
and
jenkins,
x
and
stuff,
they
basically
just
shell
out
to
coop
control,
because
it's
the
gold
standard
for.
A
Yeah,
it
also
supports
multi-doctorate,
so
there's
like,
I
should
probably
should
have
said
this
earlier,
but
we
should
make
sure
we're
not
duplicating
work.
It's
already
been
done.
C
B
One
question
so
I
was
like
digging
through
the
code
base
in
api
machinery
and
there's
like
package
util,
yemo
decoder.go
and
there's
already
like
new
document
decoder,
which
takes
a
demo
document
function.
So
I
was
wondering
how
this
or
if
this
is
used
at
any
place
or
how
this
can
be
used.
A
Yeah,
okay,
so
yeah.
Let's
take
a
look
at
that
stuff
alex.
Maybe
you
want
to
do
that
and
see
if
it's
basically
the
same
as
what
we
have
in
the
q,
a
dm
thing
we're
trying
to
patch
in
or
if
there
are
ergonomic
differences
where
it
would
make
sense
to
make
some
changes
and
just
you
know,
help
figure
out
what
the
direction
is
here.
C
Well
and
yeah
actually
help
me
understand
the
usage
a
little
bit
because
the
meta
factory
it
returns.
The
signature
was
a
list
of
gvk's
right
and
are
the
associated
bytes
for
each
gvk
returned
with
it,
or
is
that
just
like
meta
information
about
it
that
says
hey?
If
you
you
know
want
to
look
at
this,
then
you
can
expect
that
there's
going
to
be
three
of
these
things.
Two
of
these
things.
A
B
C
So
yeah
just
to
make
sure
taking
the
notes
so
that
we're
not
losing
what
the
action
item
here
is.
So
basically.
C
In
the
note-taking
section,
I'm
writing
goal
is
to
provide
multi-doc
api
machinery
helpers
in
staging
instead
of
core,
and
then
we
would
like
to
review
other
multi-doc
support
in.
C
Existing
code
other
than
kubernetes
some
examples
being
the
coupe
ctl
multi-doc
support
the
customized
stuff.
C
C
In
yeah
decoder
by
proxy.
A
C
B
So
far
I
mean
I'm
still
figuring
it
out
right,
but
I
haven't
encountered
like
a
major
roadblock
in
that
journey.
I
would
say
just
for
like
understanding
why
it's
or
like
what
we
want
to
achieve
is
basically
we
give
this
map
or,
like
some
meta
factory,
to
a
serializer
right
and
the
serializer
would
be
able
to
parse
multi
dark
demo
document.
Is
that
correct.
C
So
yeah,
at
the
end
of
the
day
like
we,
we
want
to
be
able
to
like
get
a
a
strict
serializer
like
multi,
yaml,
doc
or
multi-document
implementation
that
a
component
can
use.
So
if
you're
setting
up
a
new
component
or
as
we're
retrofitting
all
of
the
old
components
that
you
would
be
able
to.
A
B
So
you
got
like
in
api
machinery,
you
got
like
runtime,
serializers
right,
and
so
such
a
serializer
would
be
able
to
parse
or
serialize
a
motor
multi-dog
yaml
file.
Is
that
correct,
like
this
is
what
we
want
to
achieve.
C
C
It's.
I
think
it's
a
really
leaky
abstraction.
I
at
some
point
it
would
be
interesting
if
we
could
just
flatten
kodak
factory
and
serializers.
B
Yeah
yeah,
I
was,
I
was
confused
because
I
think
most
like
the
bulk
logic
of
the
serializers
is
on
the
json.
So
you
have
like
a
json
serializer,
but
then
you
set
an
option
that
is
able
to
pass
and
you
pass
it
another
factory
and
then,
but
you
also
have
like
a
serializer
yamo,
which
only
has
like
two
functions.
So
it's
like
yeah.
I
wasn't
quite
sure
how
this
all
plays
together.
There.
C
And
yeah,
making
sense
of
how
these
things
are
related
has
been
difficult
for
me
personally
in
the
saga
of
this
next
bullet
point,
which
is
that
kodak
factory
options
did
get
merged.
I
just
saw
apparently
like
a
few
days
ago.
I
don't
even
know
how
I
missed
that,
but
yeah.
So
we
have
the
ability
to
actually
use
the
native
api
machinery
libs
now
to
do
strict
serializing
through
codec
factory.
C
But
so
I
don't
know
how
codec
factory
relates
to
multi-doc
support
at
the
moment,
because
if
the
multi-doc
stuff
is
in
the
decoder,
then
your
your
two
abstractions
down
from
codec
factory,
which
means
we
probably
the
meta
factory,
does
it
take
a
bite
stream?
C
I
don't
know
anyway,
so
the
codec
factor
is
going
to
do
all
the
defaulting
stuff,
which
is
great,
but
now
we
need
to
figure
out
how
to
marry
some
of
these
ideas.
C
C
So
things
like
the
domain
names,
you
know,
follow
our
expected
format
and
if
the
rule
is
broken
there,
that
there's
an
actual
exception.
That's
put
into
the
test
code
for
that
api.
C
It's
I
on
some
of
them.
I
linked
in
the
tracking
issue.
Lucas
has
had
those
patches
open
since,
like
march,
I
think.
C
They
they
were
not
the
codex
factory
options
thing,
but
the
test-
lib
stuff,
yeah,
okay
and
yeah-
the
the
test
lib
stuff-
will
need
to
be
restructured
a
little
bit,
but
in
general,
most
of
what's
there
is
there.
I
have
a
bunch
of
like
branches
that
I've
been
working
on
that
I
can
probably
revive
so
this
should
be
coming
down.
I
find
soon
but
yeah.
B
Are
there
any
work
items
which
should.
C
C
We
need
a
test
library
that
just
exposes
some
like
a
runner
and
some
common
checks
that
are
parameterized
and
then,
when
we
instantiate
the
scheme,
then
the
apis
are
available
and
we
can
check
to
see
hey.
Do
these
apis
conform
to
what
we
expected
and
then
we'll
have
a
patch
for
each
component.
Api.
C
C
Yeah,
I
mean
honestly,
if,
if
you
can
figure
out
what
multi-doc
is
looking
like
for
all
of
those
bullet
points,
that
sounds
like
a
pretty
big
task.
B
All
right
well
I'll
dive
into
that
and
get
those
connection
points
working
on.
C
Yeah
is
if
that's
something
that
you're
looking
to
do,
which
I
think
we'd
all
be
very
glad
to
read
the
summary
links
to
the
independent
pieces
of
code
that
are
doing
that
right
now.
So
then
we
can
talk
about
okay.
Well,
it
looks
like
this
meta
factory
thing.
If
something
won't
emerge
and
we
know
we
can
consume
it
and
code
in
codec
factory
like
this
or
whatever,
then
that
would
be
excellent.
B
Yeah
I'll
look
into
it.
There
might
be
some
some
newbie
questions
coming
up,
so
I'm
prepared.
C
And
then
I
guess
we've
got
a
few
more
minutes
mike
did
do
you
know
what
the
status
of
the
coop
was?
It
kublet
legacy
flag
patch.
A
But
yeah
we
need
to
follow
up
on
that
and
then
I
also
wanted
to
ask
quick
ross
if
he
needed
any
help
getting
that
q
proxy
cap
in
I
seemed
like
the
discussion
had
kind
of
stalled
on
that,
but
I
I
want
to
merge
at
least
some
version
of
that
and
get
another
alpha
version
out
with
that
split
refactoring.
D
Yeah
so
I
actually
got
most
of
the
prototype
of
the
first
alpha.
That's
the
like
mode,
consolidation
done,
and
but
probably
this
is
going
to
merge,
at
least
in
the
the
next
cycle.
A
D
Yeah,
okay,
so
I'll
probably
just
put
on
hold
the
second
like
the
last
two
steps.
Just
leave
the
first,
the
mod
consolidation
one
and
merge
like
this,
and
if
we
decide
to
like
introduce
reintroduce
these
steps
in
like
alpha
three
or
alpha
four,
we
can
always
do
that.
C
Yeah
and
then,
as
far
as
the
okay
notes,
notes,
notes,
taking
notes.
A
C
C
Okay
and
then,
when
justin
was
chatting
about
the
optional
fields
thing,
do
we
want
to
prioritize
that
immediately
or
are
we
going.
A
C
C
Okay,
yeah,
I
mean.
Obviously
we
can't
I
I
would
say
we
probably
can't
change
like
the
format
of
of
structs,
even
if
the
apis
don't
change
right
without
a
new
version,
so
yeah.
A
A
C
So
we'll
get
the
the
test
lib
stuff
going
hopefully,
and
by
next
meeting
we
might
have
an
update
on
the
legacy
flag
patch
to
kublet
and
then
alex
is
going
to
go,
try
to
figure
out
what
our
approach
should
be
for
the
multi-dock
stuff
and
hopefully
we
can
keep
the
code
as
proposed
and
then
yeah
and
then
we'll
get
the
coop
proxy
stuff
going
more
whoa,
so
ross,
you're
you're
working
on
that
right,
yep,
all
right,
cool
yeah!
Any
any
last
comments.
C
Cool
well,
we
saw
one
of
our
guests
had
to
drop
other
than
that
thanks
everyone
for
joining
and,
let's
start
yeah.
Thank
you.