►
From YouTube: Kubernetes SIG CLI 20200826
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,
good
evening,
good
afternoon,
depending
on
where
you
are
today
is
august
26th-
and
this
is
one
of
our
bi-weekly
six
july
meetings.
My
name
is
mate
and
I'll
be
your
host.
Today
we
have
a
packed
agenda,
but
before
we
get
to
120
planning,
which
is
the
main
goal
of
our
today's
meeting,
a
couple
announcements
first
of
all
welcome
siam
siam
is
our
summer
intern
and
he
will
be
helping
us
with
improving
both
cube
cattle
and
customized
dogs.
A
So
welcome
aboard
siam
we're
super
thrilled
and
excited
to
have
you
here
with
us
and
we're
looking
forward
to
to
to
the
work
that
will
be
part
of
your
summer.
Internship.
A
You
want
to
add
couple
words
to
this
one
yeah.
Can
you
hear
me
guys
yeah
loud
and
clear.
B
Yeah
thanks
for
inviting
me,
I'm
really
excited
to
be
a
part
of
the
kubernetes
community
and
the
cncf
community
as
well,
and
the
thing
is
I'll,
be
contributing
primarily
to
docs
on
the
to
control
and
customize,
as
you
said
so
I'll,
be,
focusing
on
adding
more
examples,
creating
more
guides
to
make
it
more
accessible
to
the
end
user,
primarily
and
we're
also
planning
on
building
a
centralized
site
which
homes
both
tube
control
and
the
customized
docks,
so
that
people
don't
have
to
juggle
between
both
the
sites.
B
In
order
to
get
map
mapping
I
mean,
don't
they
don't
have
to
struggle
mapping
the
cube,
control
and
customize
stuff
and
that's
basically
the
goal
and
I'm
walking
along
with
phil
and
we're
kind
of
having
a
big
plan
posted
up,
and
we
are
planning
to
get
it
done
within
the
end
of
this
year.
Hopefully,
and
that's
that's
the
plan,
thank
you.
A
C
Yeah
he's
so
siam's
going
to
the
the
first
thing
they're
working
on
is
is
looking
at
the
unifying
the
cue
control
book
and
docs
and
the
customize
docs
to
have
like
a
single
page
that
kind
of
outlines
all
of
our
tools
and
provides
a
holistic
view
of
all
of
them,
and
so
he's
gonna,
convert
the
code,
control
locks
to
hugo
and
then
work
on
integrating
them
together.
C
So
if
that's
interesting
to
you,
reach
out
to
siam.
A
If
I
remember
you
are
using
cuda
already
in
the
main,
kubernetes
socks
is
that
right,
though,.
C
A
Okay,
another
important
announcement
is
that
119,
I
think
it
g8
earlier
today
or
is
about
to
be
ga
within
the
next
couple
of
minutes.
A
I
was
hoping
to
see
a
an
announcement
on
slack,
but
I
haven't
seen
one,
but
I'm
pretty
sure
that
I've
seen
someone
somewhere
and
definitely
the
docs
and
the
last
time
I
was
checking
the
sick
release.
They
were
yes,
26th
is
the
day
when
we
will
be
ga
119,
so
that
should
happen
today-ish.
A
Unfortunately,
I
don't
have
120
dates.
Yet
I've
heard
there
are
some
discussions
I
haven't
got
to
those
yet
and
I
was
just
checking
in
the
sick
release,
there's
120
directory
that
appeared
within
a
couple
of
past
days,
but
that
only
contains
the
the
release
team.
A
F
F
Sorry
to
interrupt
just
as
as
part
of
that
119
ga
I
put
at
the
bottom
of
the
topic
for
120
planning,
one
of
the
issues
that
jordan
and
the
release
team
had
brought
up
to
us
since
cli
about
the
ski
test,
failing
which
affected
the
the
release.
F
And
so
I
was
hoping
that
for
120
we
could
put
a
few
resources
into
maybe
getting
those
tests
back
on
on
track.
So
that
the
you
know
it
doesn't
get
in
the
way
of
the
the
rest
of
the
kubernetes
releases
and
and
another
a
big
thank
you
to
brian
personally,
for
for
actually
digging
into
to
that
issue
and
addressing
some
of
those
two
tests,
which
is
kind
of
usually
a
thankless
job,
but
really
appreciate
him.
Putting
effort
on
on
that
to
help
the
rest
of
the
communities
in
the
119.
F
G
You're
welcome
and
I'll
be
glad
to
help
with
any
other
work
on
that.
G
A
F
Put
some
time
in
on
it
as
well,
and
I
apologize
that
that
I
wasn't.
A
Yeah
well
put
together
before
we
move
on
to
120
planning.
I
would
like
to
welcome
new
faces.
I
know
that
paul
is
new
because
he
he
spoke
up
before
we
started
the
recording
that
he's
new,
so
welcome
paul.
A
H
Sure
you
probably
know
my
colleague
bailey
hayes.
I
think
she's
joined
a
couple
of
times,
she's
one
of
our
young
leaders
and
she's
off
being
brainwashed
in
how
to
be
a
good
one
by
hr
and
things
like
that.
It's
a
real
honor
actually
for
her
to
be
selected
into
that
program,
but
still
I'm
her
boss,
I'm
all
guy
who's.
You
know
been
doing
this
for
a
really
long
time
since
he
works
on
punctuards.
H
I'm
also
the
troublemaker
of
all
of
the
changes
that
happened
in
customize,
3.8
and
3.81,
where
the
switch
over
to
a
different
yaml.
Sorry,
someone
else
is
calling
me
on
there
we
go.
H
They
introduced
some
some
changes
that
our
we've
got
a
pretty
big
investment
in
customized,
and
we
want
it
to
work
from
month
to
month
because
we're
deploying
to
lots
of
customers
and
we
want
their
customizers
to
be
stable
across
month
monthly
releases
of
our
software.
So
I'm
joining
in
to
see
how
it
works
and
how
I
can
contribute.
H
I
did
try
to
start
debugging
on
a
plugin,
but
on
my
mac
I
cannot
get
the
plugin
and
the
main
thing
to
be
the
same
version
of
go,
and
so
my
debugger
just
gives
me
the
finger,
so
I
don't
mean
to
derail
the
meeting,
but
if
anyone
knows
how
that
works,
I'd
appreciate
a
little
tip
sheet
after
the
meeting
or
offline.
Somehow.
So
thanks
for
letting
me
talk.
A
Yeah
definitely
worth
reaching
out
to
at
this
at
slack
and
writing
out
writing
down
your
problem.
I'm
pretty
sure
that
others
that
are
working
on
max
specifically
will
be
able
to
help
you
develop
the
problem.
I
I
personally
am
I'm
not
a
mac
user,
so
I
won't
be
able
to
help.
I
remember
that
phil
was
the
mad
guy,
not
sure
if
sean
is,
I
I
it's
hard
for
me
to
say
who,
who
else.
E
H
I
do
my
handles
horn,
polish,
so
we've
talked
a
little
bit
there.
I
will
get
up
with
you
on
the
slack
channel
thanks.
A
Perfect,
do
we
have
anyone
else
who
is
new
to
our
call
that
would
like
to
introduce
themselves.
I
Hello,
my
name
is
rohit.
I
am
thank
you.
This
is
the
first
time
I
am
attending
this
meeting
and
I
am
totally
new
to
this
con
kubernetes
repository
amla,
looking
forward
to
contribute
code
documentation
almost
everything,
but
my
main
focus
is
contributing
code
and
learning
more
about
it.
A
A
A
A
There
is
a
big
hope,
because
six
scheduling
started
and
convinced
the
sick
architecture,
and
especially
the
part
that
is
responsible
for
code
organization
to
have
a
library.
A
Currently,
the
library
will
be
called
will
be
called
component
helpers
if
I
remember
correctly
I'll
make
sure
to
link
the
pr
after
the
call
to
the
agenda
once
that
that
new
repository
with
basically
library
utilities
will
be
created,
we
will
be
able
to
move
our
auth
shared
code
which,
which
we
share
between
auth
reconcile
and
the
api
server,
and
as
soon
as
that
happens,
we
will
be
then
able
to
to
move
cube
cuddle
out
of
the
main
kubernetes
repository,
which
is
something
that
we've
been
working
on
for
for
quite
a
while
already.
A
So
that
is
our
top
priority,
and
everything
looks
like
120
will
be
the
place
where
we
will
have
completed
the
move
and
then,
once
the
move
is
completed,
we
will
start
discussing
the
process
how
to
release
cube
cuddle,
how
to
build
it
and
release
it
outside
of
the
main
kubernetes
repository,
which
will
be
a
another.
A
Big
battle
for
us
another
one
that
is
happening,
I
think
for
two
releases
already-
is
cube.
Cuddle
debug,
it's
one
of
the
two
alpha
commands
that
we're
trying
to
introduce
debug
is
being
mostly
worked
on
by
liv
urban,
with
another
person
helping
him.
We
currently
have
support
for
affirmative
containers,
debugging
as
a
copying
through
copying
the
pods.
A
A
If,
at
any
point
in
time,
you
have
any
questions,
feel
free
to
interrupt
me
I'll,
be
talking
through
the
ones
that
I
know
that
I
remember
keep
cuddle
events
we
put
together.
Well,
we
japan,
should
put
together
a
proposal.
He
punches
part
of
the
sick
user
experience.
He
put
together
a
proposal
about
two
releases
ago
or
something
like
that.
It
got
merged,
but
unfortunately,
somehow
we
never
got
to
actually
implementing
it.
He
started
just
before
119,
but
that
did
not
get.
A
Suffering
time
to
be
finished
for
119,
so
I'm
very
hopeful
for
120
to
be
able
to
introduce
the
keeper
cattle
events
command.
There's
the
work
of
servers.
I
describe
I've
been
talking
with
clayton
because
he
has
he
already
figured
it
out,
but
the
problem
is
that
he
has
it
currently
in
his
head.
A
I
need
to
reach
out
to
him
and
once
again
to
put
to
put
the
thoughts
into
a
document
or
enhancements.
Ideally,
and
then
we
can
start
the
implementation
phase.
A
The
next
one
is
coming
from
a
proposal
from
from
phil
from
quite
a
while,
where
we
will
be
sending
the
comment
metrics
through
cubecattle
headers,
so
that
the
server
can
collect
those
metrics
and
we
will
have
more
much
more
insights
into
what
are
the
most
used
comments.
Unfortunately,
that
did
not
happen.
A
The
next
one,
which
is
hiding
server
side,
applied
data
in
cube,
cuddle
edit
that
actually
made
it
in
119,
so
cube
cutter
edit,
should
not
show
the
managed
fields.
There
are
still
discussions
about
whether
we
want
to
show
the
managed
fields
and
with
with
different
outputs,
such
as
json
or
yam,
but
from
the
dose
discussion.
Is
that,
since
both
json
and
yam
are
meant
to
provide.
A
So
I'm
gonna
remove
that
one,
because
that's
completed
cube
color
edit
enrich
with
open
api
data,
which
is
basically
having
some
kind
of
hints
inside
of
edit
coming
from
open
api
data.
The
open
api
data
is
something
that
you
can
see
when
you
invoke
cube
cattle,
explain
on
a
particular
field.
A
Keep
cuddle's
q
warning
whenever
version
is
2
or
higher
we
have.
If
I
remember
correctly,
we
have
two
warnings.
Currently
there
is
one
warning
about
beta
and
that
was
implemented
by
jordan.
I
can't
remember
sean.
Did
you
get
a
chance
to
to
finish
that
skew
warning
when
the
version
is
two
or
higher,
or
that
was
brian,
that
implemented
that
one.
J
Hey
mache
for
that,
edit
with
open
api
data.
Is
there
an
issue
open
for
that
already.
J
A
Perfect
thanks
a
lot
eddie
you're
saying
yourself
and
curry:
yup,
okay,
perfect
separate
cobra
from
options.
I
think
there
were
some.
Oh
there's
an
uber
issue
from
sean
and
people
started
working
on
it,
but
I'm
worried
that
I
missed
couple
reviews
on
those.
So
I
think
the
only
problem
is
pushing
those
through
reviews.
F
There's
there's
been
one
effort
so
far
with
apply
to
to
basically
try
to
get
some
tracer
bullet
and
see
what
actually
works
to
try
to
separate
these
two
and
make
these
just
to
give
everybody
else
a
little
bit
of
background.
The
reason
we're
doing
this
is
to
try
to
make
coop
cuddles
code
more
reusable
and
easy
to
import
without
pulling
in
the
cli
specific
stuff,
so
that,
if
it's
used
in
a
controller,
you
don't
have
this
unneeded
and
unnecessary
cli
cobra
stuff.
F
So
we've
tried
it
with
one
and
we
actually
haven't
found
a
overarching
way
that
we
want
to
approach
all
of
these
yet
but
yeah.
That
sounds
like
a
great
thing:
to
put
resources
on
to
what
and
and
hopefully
we'll
all
keep
in
touch
in
this
effort,
so
that
we
can.
We
can
do
it
in
a
consistent
manner
across
all
the
commands.
J
I
think
we
should
merge
that
one
and
the
one
below
it
the
complete,
validate,
run
stuff.
That's
been
open
for
a
while.
We
still
aren't
in
agreement
on
like
what
belongs
where
so
I
I
really
think
we
just
all
need
to
come
together
and
like
dissect
a
command
and
tweak
it
to
where
we
all
want
it,
especially
if
we're
separating
the
command
the
cobra
logic
out
of
it.
So.
A
I
was
about
to
talk
about
the
other
one,
which
is
the
complete
validate
run.
I
remember
that
I
was
looking
at
it
eddie
last
time
a
couple
weeks
back
and
I
probably
still
owe
you
a
review.
A
A
Get
it
completed
and
then
and
then
definitely
we
can
always
tweak
it
as
we
need
as
we
find
necessary.
But
at
least
we
will
have
a
a
pointer
for
newcomers
and
other
people.
So
yeah.
Can
you
actually
ping
me
with
the
link
to
that
pr?
J
Yeah
now
that
sounds
great,
so
the
reason
I
brought
it
up
was
because
part
of
this
was
119.
We
had
originally
wanted
to
rework
all
the
commands
to
have
that
same
model.
So
if
we're
gonna
be
separating
cobra
out
of
them
anyways,
we
should
just
all
do
the
one
refactor
together,
it'd
probably
make
a
lot
more
sense
than
doing
two
passes
on
all
of
it.
A
Yeah
yeah,
I
had,
I
see,
no
objections
on
that
one,
although
probably
since
you
have
pretty
much
completed
the
first
one,
you
can
get
that
one
merge
and
then
work
another
pr,
but
I
it's
hard
for
me
to
say
anything
about
merging
because
I
don't
know
the
date
for
120,
so
I
don't
know
when
we
will
be
able
to
merge
whether
it's
it's
worthy
to
wait,
a
little
bit
more
and
get
the
rest
work.
A
Although,
if
I
remember
correctly,
the
conflict
that
you
were
working
on
was
pretty
simple,
so
separation
of
the
cobra
logic
should
be
pretty
straightforward
and
simple
to
do
I'll.
Leave
it
up
to
you
whether
you
want
to
do
it
within
that
pr
or
have
a
follow-up,
I'm
open,
I'm
open
and
I'll
merge,
whichever
whichever
option
you
want.
J
A
Okay,
so
the
last
item
for
120
is
the
skew
test
once
again
brian.
Thank
you
very
much
for
your
work
on
the
tests
I
still
have.
I
also
have
that
that
particular
issue
for
my
list
for
tomorrow
to
review
where
we
are
with
this
one,
how
far
we
should
be
investing
in
this
one,
and
definitely
that
should
be
a
super
high
priority
for
us
for
120
to
get
that
one
cleaned
at
this
point
in
time.
Does
anyone
have
it
yeah
go
ahead.
F
Sorry
to
interrupt
so
so
I
just
wanted
to
point
out,
in
addition
to
some
of
this
q
test
failing
there
were
some
flaky
tests
as
well,
which
were
hammering
the
119
going
to
ga,
and
I
know
that
that's
an
especially
thankless
job
to
try
to
clean
up
these
flaky
tests,
but
I
think
it's
been
a
while,
since
we've
actually
put
resources
and
effort
into
our
our
testing,
you
know
into
the
coop
cuddle
tests
and
you
know.
F
Maybe
we
should
make
sure
that
that's
got
some
focus
for
the
this
coming
up.
120
release
yeah,
definitely.
J
I
was
talking
to
hash
dan
yesterday.
I
think
he
was
he
under
sig
release
or
sick
testing.
He's
the
one
who's
been
hounding
us
for
a
flaky
test
for
a
while
awesome
dude.
He
said,
he's
totally
open
to
help
folks
who
want
to
get
started
with
testing
more
and
he
talked
about
doing
a
live
stream
even
to
walk
through
it
for
new
contributors.
So
maybe
we
can
sync
with
dan
on
that.
A
There's
actually,
I
can't
remember
where,
where
I
saw
it,
I
was
going
through
a
lot
of
my
cues
in
the
slack
and
emails
there's
a
recording
from
sick
testing
and
a
document
from
jordan
there's
a
probably
about
35-minute
walk
through
from
jordan
about
how
to
nail
and
solve
problems,
especially
the
tricky
ones,
I'll
try
to
find.
A
A
Even
for
myself,
when
I
was,
was
solving
different
problems
with
units
integration
and
e-to-e's.
I
found
it
very
super
helpful.
The
insights
that
jordan
had
there
with
regards
to
how
to
stress
tests,
units
and
similar.
It's
definitely
worth
about
35
minutes
of
your
time,
especially
if
you're
interested.
A
Oh
yeah,
video
is
here:
okay,
yeah
and
that's
pretty
yeah.
It's
fresh.
It
was
presented
at
meeting
yesterday
so
yeah.
It's
definitely
definitely
worth
your
time,
especially
if
you're
into
debugging
those
flakes
so
yeah.
H
A
So
there
are
two:
there
are
two
steps
required.
One
is
customize
needs
to
clean
up
a
little
bit
of
their
own
dependencies.
If
I
remember
correctly
and
jeff
was
pretty
much
ready
with
it,
and
I
hope
I'm
pretty
sure
that
jeff
can
speak
much
better
than
I
do
about
that
part
and
on
our.
H
C
H
E
A
Okay,
perfect,
if
you
want
to
help
with
any
of
the
pieces,
feel
free
to
reach
out
to
the
people
named
right
next
to
the
to
the
attempts
either
or
myself
or
phil
or
sean,
and
we
will
be
able
to
point
you
to
the
right
people
if,
if
you
want
to
help
with
with
some,
the
next
topic
is
from
eddie.
A
I
think
we
talked
about
it
several
times
during
blocker
bugs
but
I'll
all
let
eddie
speak
about
it.
J
Yeah
I
just
we
talked
last
meeting
about
having
a
prfaq
with
just
like
consistent
answers,
and
so
we
have
historical
stuff,
that's
not
in
all
your
brains.
So
if
you
could
help
me
by
abstracting
some
of
your
your,
I
don't
want
to
say
tribal
knowledge,
anymore,
your
your
historical
knowledge
and
context
into
the
stock.
That
would
be
great.
I
will
keep
pestering
you
until
you
do
this.
Please.
A
I
promised
that
one,
probably
during
my
last
call
and
didn't
deliver
it
either.
So,
okay,
let
me
write
it
down
I'll
make
sure
to
do
it
tomorrow.
It's
well
running
a
little
bit
light
over
here.
It's
alpha
six
already.
C
I
did
suggest
a
fact
made
for
the
new
dark
side,
though
so
I
should
get
some
points
for
that.
J
Yeah,
no,
that's!
We
can
include
this
forever.
I
figured
this
would
just
go
in
the
repo
under
a
markdown
dock
and
we
can
link
to
it
in
like
the
pr
template,
but
we
we
got
an
issue
open
yesterday
for
confirmation
when
using
all
again-
and
this
is
a
historical
one-
that
we
said
no
to
right
so
like
if
we
can
just
get
together
a
copy
paste
like
here's,
a
reason
why
no
like
and
we
can
hand
off
the
run
book
to
other
people
too,
for
triage.
C
A
What
we
talked
about
is,
we
will
start
with
this,
because
it
will
be
easier
to
collaborate
for
starters,
and
then
we
will
just
put
that
one
into
the
main
kubernete
cube
cuddle
repo,
and
we
will
have
a
at
the
root
of
keep
cuddle.
This
way,
you
know
when
you
will
be
looking,
and
we
probably
should
link
the
fact
right
next
to
contributing
so
that
people
know
where
to
look
for
this,
but
yeah
it'll
be.
This
is
a
in
a
visible
place.
A
There,
okay,
perfect:
are
there
any
other
topics
that
people
want
to
bring
up
before
we
move
to.
A
E
Which
also
lives
in
the
customize
repository
and
there
are
currently
in
the
code
base
for
customized
customized
uses
these
interfaces
and
behind
the
interfaces
are
different.
Implementations,
implementations
that
are
based
on
the
code
in
api
machinery
and
implementations
that
are
based
on
the
code
in
the
kml
library.
E
The
two
issues
linked
from
the
notes
here
are
the
describe
the
final
work
necessary.
The
final
implementations
needed
to
finish
the
break
with
the
api
api
machinery
dependence
once
those
two
issues
are
closed
and
they're
both
labeled
help
on
it
by
the
way.
That's
why
I'm
here
once
those
two
issues
are
closed.
E
The
way
is
clear
to
ship
a
version
of
customize
that
does
not
depend
on
api
machinery
at
all,
and
once
we've
established
that's
pretty
much,
that's
no
more
buggy
than
the
version
that
does
depend
on
kubernetes.
E
Once
we've
established
that,
then
we
are
clear
to
delete
the
api
machinery
independence
and
then
importing
it
into
customize
will
be
sorry
importing
customizing.
The
coupe
pedal
will
be
no
more
different
than
importing
say
the
html
http
libraries
and
any
other
bill
package.
So
that's
the
plan
and
those
two
issues.
There
are
the
remaining
blockers.
A
J
Of
landed
on
this
place
to
use
custom
schemas
to
hook
into
the
open,
iepi
definition,
it's
sort
of,
like
probably
like
a
three-phase
rollout
to
get
where
we
actually
envision,
but
I
have
a
quick
demo,
so
I
have
cube
ctl
generate,
which
is
it's
right
now,
it's
just
a
plug-in,
and
so
I
can
generate
a
deployment,
and
so
this
spits
me
out
the
commented
out
deployment
very
straightforward,
very
basic,
but
how
this
is
actually
working.
Is
it's
hooking
into
open
api?
J
I
have
a
local
hard-coded
template
in
here,
which
we
imagine
we'll
have
to
do
at
first
with
some
of
our
like,
sanctioned.
You
know,
white
listed
templates,
and
so
this
is
just
a
hard-coded
template.
That's
open
api
spec
and
this
example
field
is
actually
built
into
the
open
api.
So
right
now
all
we're
doing
is
kind
of
merging
this
into
the
deployment
upstream
and
here's
our
example.
So
that's
cool,
but
that's
hard-coded
and
so
corey
actually
added
the
ability
to
use
a
local
file,
so
we
can
generate
with
a
local
deployment.
J
This
also
works
with
the
remote
url
too
right.
So
we
have
a
totally
different
deployment,
that's
defined
locally
right,
so
the
idea
behind
this
is
that
orgs
can,
if,
if
people
don't
like
our
are
suggested
or
recommended
templates
to
get
started,
they
can
provide
their
own
overripes
right.
So
maybe
you
have
a
an
org
specific
template
that
you
want
to
use
upstream,
then
everyone
in
your
org
can
have
the
same
template
when
generating
a
deployment.
J
J
So
that's
sort
of
the
phase.
One
approach
is
the
idea
is
we.
We
can
have
some
hard-coded
examples
that
we
merge
into
the
upstream
open
api
spec.
So
once
we
figure
out,
okay,
people
use
it.
This
makes
sense.
It's
a
good
idea
that
we
can
actually
get
those
definitions
added
upstream
to
the
open
api
specs.
We
don't
have
to
do
that
merging
anymore,
and
so
then
we
also
have
the
open
api
providing
those,
and
then
we
have
the
ability
to
use
remote
overrides
phase.
J
Two
is
more
of
an
idea
of
plugging
in
like
customized
setters
to
those
templates
where
we
can
pass
in
some
flags
and
define
setters
that
will
override
all
different
fields
in
our
templates.
So
maybe
you
can
provide
an
image
and
it
will
plug
in
your
image.
For
you
phase
three
is
more,
it's
definitely
longer
term
out
there.
J
The
idea
is
to
have
a
custom
open
api
extension
that
has
a
concept
of
a
template,
and
then
we
can
dynamically
parse
out
command
line
flags
that
are
provided
match
them
up
to
the
custom
api,
open
api,
spec
and
then
do
those
overrides
and
fill
in
the
templates.
We
can
also
sugarcoat
with
like
skylark
scripts,
which,
basically,
if
you
didn't
know,
skylark,
is
plugged
into
almost
every
part
of
kubernetes
right
now,
and
so
we
can
do
skylark
scripts
to
override
all
this.
J
So
so
that's
the
basic
idea,
we're
hoping
to
get
this
opr
opened
up
to
the
the
sig
experiments,
the
cli
experiments
and
then
we're
definitely
looking
for
some
people
to
play
around
to
get
feedback
and
and
pull
it
apart.
So.
C
Yeah,
that's
great
eddie.
Thank
you.
It's
awesome
to
see
a
problem
just
last
meeting
being
brought
up
and
already
having
working
code,
demonstrating
a
solution
and
I'll
just
add
to
what
eddie
was
saying.
Is
that,
like
one
of
the
motivations
for
this
was
the
have
more
support
for
the
create
commands?
And
you
can
see
that
the
architecture
shown
here
does
not
actually
support
some
of
the
the
the
requests
naively
right
so
like?
How
would
you
add
environment
variables
as
a
flag
right
or
these
sorts
of
things?
C
And
the
notion
is
that,
like
the
tooling,
we're
building
long
term
and
really
investing
in
is
for
declarative
configuration
and
that
we've
we're
already
thinking
about
how
to
solve
that
sort
of
stuff
for
declared
configuration
so
like
eddie
brought
up?
For
instance,
we
have
a
notion
of
setters
that
works
on
their
declarative
configuration
and
allows
us
like
this
extensible
notion
of
doing
setters
and
substitutions
for
different
configurat
different
fields
in
the
configuration
file.
C
So,
rather
than
reinventing
a
new
mechanism
for
generating,
let's
just
reuse,
the
mechanisms
we're
using
for
configuration
files
to
expose
flags
and
commands
and
such
and
so
the
example
would
be
just
configuration,
it
would
have
the
appropriate.
You
know
setters
or
metadata
or
markup
or
whatever.
It
is
that
we
recommend
people
to
use
for
their
configuration,
and
then
we
would
parse
that
from
the
cli
into
flags
or
these
sorts
of
things.
C
Matcha
I
know
that
we've
talked
about
this
over
the
years.
I
think
we've
come
up
with
different
approaches.
I'm
curious
to
get
your
thoughts
on.
If
this
is
seems.
A
A
Kubecons
yeah,
when
when
eddie
was
talking
about
the
customized
style
setters
in
the
examples,
I
recall
that
you
are
literally
showing
me
something
like
that
using
the
kai
yam
library,
I
think
in
san
diego
a
couple
months
back
so
yeah.
It's
definitely
simple.
It
is,
most
importantly,
it
works
with
crds.
A
A
We
need
to
start
approaching
problems
from
this
side
rather
than
oh.
We
need
to
cover
the
buildings.
The
buildings
are
probably
the
least
of
the
problems
at
this
point
in
time,
so
yeah.
I
I
I'm
super
happy
that,
within
two
weeks
we
have
a
working
prototype
so
that
basically
means
that
plot
in
120
will
have
a
alpha
command
in
keep
cuddle.
A
So
if
you
have
any
questions
or
most
importantly,
ideas
about
the
the
generate
command
that
eddie
is
working
reach
out
to
eddie,
I'm
pretty
sure
that
we
will
definitely
spend
at
least
one
of
the
upcoming
meetings
talking
through
where
we
are
and
where
we
want
to
go
with
that
one:
okay,
nick
you're,
not
you're
you're
next.
D
Agree
I
mean
we're
working
on
some
internal
stuff,
one
of
the
things
that
service
from
that
is
some
imperatives
to
start
working
on
the
jobs
part
of
cuckoo,
which
is
all
pretty
darn
nifty.
I
I
have
a
big
serverless
background,
so
I
guess
everything
is
old,
is
new
again,
so
cooley
has
some
some
initial
support
for
enhancing
sort
of
working
with
jobs.
So,
for
example,
there's
some
sequence
diagram
views,
so
you
can
understand,
you
know
with
jobs
and
servers
generally.
D
The
problem
of
cold
starts
problems
of
of
you
know,
just
how
is
your
cluster
configured
in
terms
of
capacity?
So
you
understand
low
distribution
and
all
that.
So
we
have
some
support
for
that.
One
of
the
topics
that
came
up
that
actually
sort
of
plays
off
of
what
we
were
just
talking
about
in
terms
of
templates
and
and
customize
as
well.
D
It
occurred
to
me,
as
I
was
working
on
this
job
stuff,
that
there
might
be
some
interesting
crossovers
there
that
jobs
seemed
like
a
seems,
like
a
topic.
That's
perfectly
suited
for
this
kind
of
template,
driven
development
that
I
may
have
you
know,
I'm
defining
a
job
that
it's
a
template
and
every
time
I
run
a
job,
it's
some
sort
of
customization.
D
D
Right
now,
every
time
I
want
to
run
a
job,
I
have
to
craft
a
whole
new
yaml
and
apply
it
as
opposed
to
crafting
a
yam
or
a
template
up
front
akin
to
the
stuff
we
were
talking
about
and
for
my
job
definition.
If
you
will
and
then
every
time
I
run
a
job,
I'm
just
plugging
in
some
parameters-
some
different
environment
variables,
some
different
args
to
the
container
things
like
that.
D
A
Yeah
definitely
and
yeah.
I
know
the
pain
as
since
I
was
something
for
the
fact.
Oh
yeah.
Definitely
that
works.
I
can
definitely
speak
about
jobs
since
I
was
authoring
that
one
long
time
ago,
okay,
cool.
A
I
think
that's
pretty
much
all
that
we
had
in
the
agenda
for
today
is
there
anything
that
that
any
one
of
the
people
want
to
bring
up
to
the
rest
of
this
egg.
I
Yeah
and
I
was
going
through
the
contributing
in
that
sorry,
the
contributing
md
file
in
cube,
cutting,
and,
I
believe
some
portions
about
it-
are
outdated.
I
I
I
had
a
word
yesterday
with
philip
wittrock,
so
I
was
just
wondering
like
in
which,
how
do
I
get
information
like
how
much
of
it
is
outdated?
So
I
could
open
a
pr
for
all
removing
all
the
outdated
stuff.
C
Yeah
rohit
reached
out
to
me
yesterday
asking
about
the
mentoring
program
and
I
think
there's
some.
I
think
we
had
a
mentoring
program
at
some
point
back
and-
and
I
don't
really
remember
what
this
state
of
that
is.
C
So
I
was
suggesting
that
we
and
I
don't
even
have
access
to
the
google
group
that
it's
done
through
anymore.
So
I
was
suggesting
that,
if
that's
not
active,
we
should
remove
it
from
the
docks
of
the
contributing
section.
J
A
Correctly
they
did
recently.
George
castro
was
tweeting
about
a
set
of
updated
resources.
I
can't
remember
exact,
but
if
you
look
for
george
castro
on.
A
A
But
yeah
some
data
might
be
outdated.
I'm
not
surprised,
but
yeah
any
help
with
this
is
greatly
appreciated,
so
yeah.
That
might
be
your
very
first
contribution
to
the
kubernetes.
J
Project
I
had
one
quick
thing:
if,
if
you
all
are
creating
interfaces
or
implementations
of
interfaces,
make
sure
you
please
comment
them
and
don't
assume
everyone
can
tell
what
it
is
from
the
title
I've
been
struggling
going
through
building
this
generate
thing
like
I,
don't
know
what
the
hell
a
discovery
client
is
that
doesn't
make
sense
to
me.
I
don't
know
what
a
a
mapper
is
or
a
rest
mapper
resource
for.
J
C
Sean's
sean's,
I
think,
has
some
material
on
this.
In
fact,
I
think
he's
done
at
least
two
coupon
talks
on
like
what
the
on
the
those
sets
of
libraries
so
see.
If
you
can
find
them
in
a
google
search,
I
think
they
were
like
six
cli,
deep
diver
overviews
or
something
like
that,
and
and
if
not
reach
out
to
sean-
and
we
could
probably
just
take
the
videos
of
those
talks
and
put
them
on
the
new
doc
site
as
some
part
of
the
material.
A
C
But
yeah
you're,
not
the
only
one
eddie
that
has
been.
I
think
everyone
hits
that
that
pretty
steep
learning
curve
with
those
the
good
news
is
once
you
figure
those
out
everything
else
kind
of
makes
sense.
J
Well,
the
worst
part
is
like
I'm
still
not
a
big
fan
of
the
pooling,
like
it's
still
not
there.
So
I'll
do
like
give
me
the
definition
of
some
struck
that
I'm
supposed
to
provide,
and
it
takes
me
to
the
interface
and
I'm
like
okay,
and
then
I
do
get
me
all
the
inner.
You
know,
implementations
of
the
interface
and
there's
like
eight
different
things,
scattered
throughout
api
machinery
and
like
our
utils
package
and
like
open
ap.
It's
just
like
it's
very,
very
the
learning
curve,
especially
because
I'm
not
the
best
go
developer
out
there.
F
Yes,
just
to
chime
in
as
well
any
the
so
it
has
nothing
to
do
with
your
go
development
abilities.
This
is
this
is
a
very
opaque
area
has
been
for
a
long
time,
but
once
once
you
get
deeper
insight
into
it.
F
Then
you're
gonna
actually
have
really
good
insight
into
many
other
facets
of
kubernetes.
A
Okay
with
that
happy
thought,
I
think
we
can
close
today's
meeting.
Thank
you.
Thank
you
very
much
all
for
for
the
past
hour,
that
was
very
fruitful
discussion.
We
covered
a
lot
of
topic
and
see
you
in
two
weeks.