►
From YouTube: Kubernetes SIG CLI 20200812
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
I've
noticed
that
eddie
is
across
the
entire
agenda
of
ours
today,
but
before
we
get
to
him
two
announcements,
there's
a
cubecon
virtual
eu
happening
next
week.
There
are
day
zero
events
on
monday
and
the
rest
of
the
conference
or
the
main
conference
is
happening
tuesday.
Through
thursday.
I
link
the
agenda.
A
A
So
I'm
hoping
that
we
will
return
to
our
regular
meetings
in
two
weeks
and
additionally,
I'm
hoping
that
in
two
weeks
frame
we
will
do
120
planning.
A
It
should
be
around
ga
4
119,
because
if
I
remember
correctly,
the
dates
the
ga
is
planned
for
august
25th,
so
hopefully
by
then
we
will
know
the
time
frames
for
the
120
release,
but
either
way.
I
think
it's
reasonable
for
us
to
start
planning
for
the
next
release.
I'll
definitely
send
out
an
email
for
both
the
boxcraft
cancellation.
Well,
you
should,
if
you're
subscribed
to
our
calendar,
you'll
get
a
cancellation
email
but
I'll
send
another
reminder
about
the
planning
for
120
in
two
weeks.
A
So
I
guess
that's
pretty
much
all
that
I
wanted
to
start
with.
Eddie
mentioned
me
that
we
have
a
ton
of
newcomers,
and
so
we
would
like
to
give
you
the
floor
to
introduce
yourself
and
tell
about
a
little
bit
about
yourself
and
what
you
want
to
work
on
for
sexy
life.
B
Hi
everyone
good
morning,
my
name
is
cory
pratt.
I
am
a
devops
engineer
at
goodrx,
I'm
based
in
california,
and
I've
been
working
with
kubernetes,
for
I
want
to
say
two
or
three
years
now,
just
I
kind
of
started.
It
started
out
as
a
developer,
creating
home
charts
for
the
applications.
B
I
was
writing,
features
for
and
then
kind
of
jumped
into
devops
full-time,
because
you
know
helped
me
so
much,
and
I
found
it
really
interesting
in
my
contributions
as
a
part
of
six
cli,
I
want
to
be
able
to
work
on
just
anything
that
will
make
the
day-to-day
lives
for
people
use
kubernetes
much
better.
You
know
I
mean
anything
I
can
do
to
kind
of
improve
it.
It's
something
I
use
every
day,
so
yeah.
A
C
C
A
D
Hey
my
name
is
sheikh
and
I'm
from
india,
so
I
recently
completed
my
ccad
exam,
so
I'm
looking
forward
to
contribute
on
on
on
the
sixth
year.
This
is
one
of
my
favorite
communities
because
I
always
go
through
the
discussion
and
it's
really
interesting
previously.
My
my
first
contribution
to
the
open
source
was
on
the
easter
hto
side
and
the
conversations
with
the
six
eli
actually
helped
me
contribute
to
the
hd.
A
Do
we
have
anyone
else,
that's
nude!
I
want
to
introduce
themselves,
you
don't
have
to
be
entirely
new,
but
if
you
feel
that
you,
like
you,
wanna,
introduce
yourself
because
you're
only
listening
over
the
past
couple
of
times
also,
I
wasn't
here
two
weeks
ago,
so
I
still
need
to
mentally
catch
up
on
who's.
New
and
who's
not
feel
free
to
to.
You
know
to
speak
up.
A
E
A
Cool
welcome.
Thank
you.
We
have
an
amazingly
international
company
today,
I'm
super
happy
about
it
that
we
have
people
like
literally
from
almost
all
the
major
regions.
I
would
say
if
people
from
europe,
israel,
north
america,
india,
pakistan,
that's
freaking,
awesome
and
probably
a
couple
more
that
I
might
have
forgotten
to
mention.
Is
there
anyone
else
that
that
want
to
introduce
themselves.
A
Okay,
hearing
none
so
with
that
again.
Thank
you
very
much.
We
are
super
happy
that
you
joined
our
sick.
If
you
have
any
questions,
issues,
topics
that
you
want
to
bring
up
feel
free
to
add
them
to
the
agenda
and
we
can
cover
them
during
our
following
meetings.
A
You
can
ping
pingas
on
six
cli
slack
channel
or
the
mailing
list,
or
you
can
directly
approach
any
of
us
and
we
will
be
more
than
happy
to
to
help
you
with
with
issues
and
especially
for
pr
reviews,
and
I
and
I'll
be
speaking
here
for
myself.
A
I
would
completely
advise
you
to
ping
me
directly
on
slack
if
you
want
to
have
your
prs
reviewed
sooner
than
later,
because
it
it
takes
me
a
while
to
go
through
my
github
notifications
and
sometimes
I'm
slightly
overwhelmed
with
the
amount,
I'm
pretty
sure
that
that's
not
only
to
me,
but
if
you
keep
on
pinging
me
on
slack,
I
will
respond
and
there's
and
I'm
pretty
sure
that
there's
a
couple
people
that
can
vote
for
that,
even
on
slack
it
it
can
take
a
little
bit
of
time
but
I'll
respond
sooner
than
on
through
github.
A
Okay,
with
that
out
of
the
way
eddie,
you
filled
up
the
entire
agenda.
There's
one
topic
for
me
for
plugins,
but
I
I
did
not
prepare
for
that
one.
I
need
to
do
it
for
the
next
time,
so
that
should
cover
that
one.
So
eddie
and
the
topic
of
additional
facts
in
cuba
create
coming
from.
F
Most
probably
yeah,
this
is
so
welcome
everyone
thanks
for
joining
us.
Of
course,
this
has
come
up
a
few
times
and
it's
come
up
in
different
pr's
and
issues
and
abdul
actually
raised
an
issue
where
he
addressed
a
lot
of
this
too.
The
the
tlbr
is,
we
kind
of
gimped
out
cube,
ctl
run
and
the
functionality
is
missing
for
some
folks
that
relied
on
it
and
there's
just
a
lot
of
oh.
We
should
have
this
flag
back
in.
We
should
have
this
sub
command
back
in.
F
I
think
corey
even
opened
a
pr
addressing
one
of
these
issues.
So
there's
a
couple
things
on
here
and
I
it's
kind
of
been
you
know,
stalled
for
a
while,
and
so
I
just
I
want
to
open
up
the
discussion
to
get
people's
thoughts.
I
know
phil
had
a
pretty
strong
opinion
on
growing
out
these
sub
commands
with
flags,
and
maybe
maybe
it's
best
to
start
with,
like
what
the
the
real
purpose
of
these
commands
are
like
what?
Why
are
we
providing
these
developers
and
what
what's
the
intended
behavior
for
that.
G
G
The
it
originally
did,
I
think,
pods
and
then,
when
we
created
deployments
it
switched
over
to
deployments
its
intent
was
never
to
like,
provide
the
full
capability
of
the
api
or
replace
like
specifying
your
workloads
in
a
declarative
manner.
G
The
create
commands
came
along
because
control
run
like
people
wanted
okay.
Well,
I
don't
want
a
pod.
I
want
a
job
or
I
don't
want
this.
I
want
that,
and
so
we
had
all
these
api
workload
apis
that
were
capable
of
creating
pods
and
so
folks
wanted
to
run.
They
wanted
an
imperative
command
to
do
run
their
particular
api,
then
with
a
service
right
and
then
so,
then,
all
of
a
sudden
run
became
this
like
single
imperative
command.
That
allows
you
to
do
anything
you
want
with
the
kubernetes
apis
right
through
one
command.
G
That
was
a
discoverability
nightmare.
So
then
the
create
commands
seem
to
make
more
sense.
Why
don't?
We
have
an
imperative
command
per
api?
I
think
the
create
command
still
didn't.
G
There's
still
some
question
about
like.
Why
do
we
have
these
things
and
I
I've
never
gotten
a
totally
satisfying
answer
about
why
we
have
these
imperative
commands
the
the
best
one.
I've
heard
the
the
two
ones.
I've
heard
that
kind
of
make
sense
are
one
is
in
a
teaching
environment.
G
It
gives
someone
that
first
day,
experience
which
I
think
doesn't
is
hard
to
justify.
Why
we
have
so
many
flags,
then,
because
we
don't
like
teaching
you
here's
how
you
run
a
container
that
makes
sense
then,
once
we're
adding
more
and
more
options,
it
seems
like
we're
teaching
you
to
do
the
wrong
thing.
G
So
that's
that's
kind
of
the
one
that
makes
sense
another
one.
I've
heard
is
debugging
like
I
just
want
to
like
run
a
container
and
attach
to
it,
and
it
needs
to
be
you
know
just
so.
G
I
have
access
inside
the
kubernetes
cluster
and
what
we
said
in
the
past
is
like,
for
both
of
those
kube
control
run,
isn't
probably
the
best
isn't
optimized
for
either
having
coop
control
run
with
like
a
million
flags
is
not
a
great
teaching
experience,
it
doesn't
lead
you
to
a
declarative
api
from
there
or
teach
you
how
to
do
stuff
outside
of
control
run
very
effectively
for
debugging
also
doesn't
make
a
lot.
G
The
other
third
reason
people
have
said
they
won't
run
is
because
they
want
to
do
because
they
like
the
imperative
interface
right,
because
it's
fun
and
is
easy
and
it
mirrors
gcloud
or
aws
cli
or
some
stuff
in
oc,
I'm
guessing,
and
so
I
think
the
answer
there
is
that
packages
like
a
lot
of
the
work
that
we're
doing
in
customized
does
a
better
job
of
that
right
rather
than
saying
okay,
why.
G
G
So
that
that's
the
background,
what
what
is
the
reason
that
we
need
does
this
fall
into
those
categories?
Why
not
packages?
Why
not
create
your
own
cli.
A
To
shed
some
more
light
to
what
phil
said
for
debugging
purposes,
lee
is
working
hard
on
having
the
keep
cuddle
debug
command,
which
is
designed
to
help
you.
A
A
A
We
are
currently
at
a
point
where
we
are
in
danger
sort
of
of
repeating
the
same
mistake
that
we
did
with
run
run
started
as
simple
command
app,
but
as
we
were
growing
the
the
spectrum
of
available
resources,
we
were
throwing
everything
into
keep
color
run
because
run
was
the
entry
point
for
people
coming
to
kubernetes,
and
that
was
a
mistake
obviously
and
create,
was
meant
and
is
still
meant
as
a
as
a
learning
tool.
A
A
As
a
the
first
step
as
something
that
you
can
quickly
showcase
how
to
use,
how
to
create
a
deployment
or
or
any
other
resources,.
A
For
blog
posts
for
for
tweets
for
for
documentation,
that's
perfect
there,
but
if
you're
interested
in
tweaking
your
object,
that
that
can
be
in
ingress,
replica,
replicas
of
replication
controller
or
a
deployment,
then
you
probably
already
know
the
initial
in
your
past
the
the
initial
learning
curve,
in
which
case
you
should
be
more
looking
into
tools
like
customize
or
some
kind
of
a
cicd
flow,
around
building
your
resources
from
scratch,
rather
than
building
up
the
create
commands,
because,
yes,
I
I
I'm
I'm.
A
Similarly
I'm
on
the
same
boat
as
phil
is
I
don't
want
to
grow?
Create
commands
to
be
monster.
Commands
like
run
is
these
days
where
we
have
a
full
page
of
possible
options.
That's
not
what
we
want
to
have.
We
don't
have
simplicity
in
the
first
place,
because
it
is
the
starting
place
and,
most
importantly,
if
a
newcomer
is
coming
to
cube
and
he
will
fire
he
will
or
she
will
fire
up
a
create
deployment
help
page
and
he
will
be
presented
with
millions
of
options.
A
G
H
G
Important
to
someone
enough
to
like
say
yes,
like
these
create
commands,
are
really
important
to,
for
you
know
some
reason
and
when
you
really
need
them
to
have
a
lot
of
flags,
we
said
okay,
why
don't
we
do
a
new
sub
project?
It
can
be
its
own
binary
and
then
it
can
evolve,
and
we
can
see
like
how
how
much
that
really
makes
sense.
G
I
think
that
that
would
make
more
sense
than
continuing
to
expand
the
surface
area.
That's
in
control
for
right
now,.
A
Yeah,
definitely,
I
think
we.
We
also
touched
this
on
in
one
of
the
previous
box
crops
and
the
general
answer
was
it's
a
case
by
case
and
if
you
provide
us
with
a
very
strong
case
that
it
does
make
sense,
it
will
be
super
useful
for
learning.
Yes,
okay,
we
can
consider
adding
that,
but
the
default
answer
will
be
no.
G
A
Issues
yeah
that
would
that
would
definitely
make
sense
to
have
the
most
frequently
asked
questions
being
answered
in
a
single
place.
Eddie
since
you're
in
charge
of
the
bug
scrub.
Do
you
think
you
could
put
together
the
most
frequently
popping
up
questions
together
and
then
we
can
collaborate
on
the
answers
and
then
we
will
put
it
together
in
in
contributing
and
doc
or
something
like
that.
F
Totally
cool
I
yeah,
plus
one
to
what
you
guys
said.
The
you
know
I
really
want
to
make
sure
we
nail
down
a
default
answer
is
no
other
than
special
circumstances,
because
you,
you
know
how
I
operate
with
the
policy
right
like
when
we
have
something
clear
and
easy
stated:
I'm
like
okay,
this
is
our
stance
on
this,
but
unless
we
figure
out
like
what
those
special
cases
are
like
education's,
obviously
a
good
candidate
but
yeah
this
is
going
to
keep
coming
up.
G
Annotations
before
or
something
we
did,
we
did
added
annotations
there
because
of
istio
or
some
special
case,
but
that's
really
the
exception,
and
that
was
because
it
just
didn't
make
sense
as
it
was
so.
It
was
kind
of
fixing
an
issue.
F
I
think
corey's
been
trying
to
chime
and
I
see
him
unmuted,
but
corey
actually
has
a
pr
open
to
add
environment
options
to
create.
B
Yeah,
actually
so
before
I
went
ahead
and
opened
up
here
for
this
or
tried
working
on
it.
I
want
to
take
a
look
at
you
know
what
the
thoughts
were,
and
I
was
looking
at
the
the
things
I
want
to
give
that
people
reported
and
how
the
stick
responded
to
stuff
this.
Pr
that
I
opened
adds
the
environment
variable
option
and
I
wanted
to
add
well
full
disclaimer.
B
My
reasons
that
I
mentioned
for
including
it
do
include
the
points
that
everyone
had
mentioned
previously
like
for
learning
and
debugging
and
stuff
so
and
I
haven't
worked
with
customize,
but
in
regards
to
some
of
the
options
that
were
recently
added
to
qgis,
upgrade
cue
cutter,
create
like
the
port
public
cassette
and
commands.
I
feel
like.
B
I
would
agree
that
there
are
some
options
in
cubes
and
q
pedal
run.
That
would
probably
be
kind
of
a
bit
much,
but
I
feel
like
the
option
for
environment
variables
was,
is
something
that
is
that
should
be
included
as
a
part
of
those
core
command
core
options
that
are
used
commonly
like.
With
with
what
I
mentioned,
the
portrait
set
and
all
that
stuff.
So.
G
G
G
There's,
actually
what
users
really
want
is
something
different
than
what
is
there
today
kind
of
like
we
were
saying,
so
I
guess
the
question
to
you
is
like
if
you're
using
this,
for
mostly
do
you
say
teaching
or
debugging,
which
one
is
the
main
both
thing
so
within
like
within
your
organization,
is
that,
like
kind
of
how
it
what
it's
about
yeah.
B
Yeah
for
myself
in
regards
to
like
debugging
and
stuff,
if
I'm
trying
to
get
poc,
I
mean
I
actually
still
use
to
create
deployment
or
something
I'll
quickly
do
like
qco
create.
You
know
a
deployment
with
all
the
options
stuff
and
then
doing
oemo
and
like
do
it
add
the
dry
run,
option
just
kind
of
get
something
basic
and
just
kind
of
adding
on
to
that.
G
So
how
like
there
is
a
high
cost
with
adding
stuff
to
kubernetes,
kubernetes
and
coop
control.
It's
just
it's
just
a
very
expensive
thing
to
do
right,
like
there's,
release,
notes
and
backwards
compatibility
guarantees,
and
once
these
flags
are
in
there,
we
can
never
take
them
out
right
and
those
are
stuff
like
seeing
the
deprecation
for
run
like
you
kind
of
get
an
idea
right
of
like
if
we
decided
that
this
was
a
bad
idea,
it
would
take
us
18
months
to
get
it
out
and
not
that
it
is
it's
just.
G
G
I
think
that
cost
is
worth
it
because
there's
so
many
other
tools
that
depend
on
it
and
it's
actually
very
difficult
to
get
like
to
to,
for
instance,
have
everyone
who
depends
on
apply
to
download
a
new
tool
right?
They
just
want
apply,
they
don't
want
to
say.
Oh
here's,
this
other
version
of
apply.
That
has
this
one
feature.
Therefore
I'll
get
that
right.
They
just
want
one
for
stuff
like
this,
where
it's
like,
hey
there's!
G
This
imperative
command
that
that
I
want
to
run
right
because
it's
because
it's
because
this
is
being
invoked
directly
by
an
end
user
and
not
being
invoked
by
spinnaker
or
tekton
or
flux
or
like
any
other
system.
G
It
means
that
the
end
user
is
actually
in
a
good
position
to
just
download
a
different
command
that
does
it
and
so
that,
for
that
combination
of
reasons,
it
being
very
high
cost
to
get
anything
in
or
out
of
control,
and
that
for
this
type,
this
set
of
commands
that
it's
pretty
low
cost
for
a
user
just
to
get
a
different
thing.
I
guess
my
question
to
you
is:
if
we
just
created
a
new
project
right
inside,
like
that,
you
could
go,
get
and
said.
G
B
I
mean
just
thinking
a
lot
I
mean
I
feel
like
the
first.
My
first
reaction
is
kind
of
that
having
that
other
tool
or
something
that
might
just
add
a
bear
of
entry
that
is
not
like
attractive
to
people
that
would
use
not
that
it's
not
a
the
right
way
to
do
so.
B
More
specifically,
like
another
point
that
I
was
thinking
about,
was
like,
I
saw
a
comment
somewhere
where
I
actually
received
the
my
ck
certification
earlier
this
year
and
was
kind
of
using
run
a
lot,
and
then
there
was
a
comment
in
one
of
the
issues
somewhere
that
someone
was
like.
Could
we
actually
make
a
revert,
because
I
have
to
take
the
cka
exam?
If
I
can't
like
I
need,
I
feel
like
using
two
control
run,
would
like
help
or
something
like
that
yeah.
B
So
I
feel
like
I
do
agree
with
some
points
just
going
back
to
it
all
that,
where
you're
saying
that
you
kind
of
the
purposes
for
control
create,
is
like
very
limited,
but
in
regards
to
all
the
uses
that
would
actually
use
cube.
Control
create
for
the
option
to
add
environment,
environment
variables
is
something
that
is
should
like,
in
my
opinion,
something
that
is
as
common
as
specifying
the
fork
or
specifying
a
number
of
replicas
or
specifying
the
commands
or
something
like
that.
G
And-
and
I
think
all
of
those
were
mistakes,
none
of
those
should
be
placed
the
the
all
like.
I
I
agree
with
you
right
like
the
this
particular
flag,
is
probably
on
par
with
the
other
flags
that
exist
there
and
there's
probably
20
other
flags
that
are
also
on
par
with
the
other
flags
that
are
on
there
right,
like
image,
image
already
exists,
but
maybe
like
sidecar
container
right
or
I
I
know,
there's.
I
I
think,
there's
like
looking
at
coop
control.
G
You
can
see
from
run
just
the
number
of
flags
that
exist
like
the
restart
strategy,
for
instance.
Maybe
one
all
these
sorts
of
things
so.
A
I'll
go
ahead
and
quickly
defend
that
with
the
flags
that
we
approved
to
be
added
to
create.
For
now,
the
majority
of
them
were
problems
coming
from
docs
across
kubernetes
docs
we've
been
using
ron
way
too
frequently
than
we
should
have,
and
the
problem
with
removing
run
or
the
procating
run
for
creating
everything
else.
A
A
I'm
trying
to
think
how
this
actually
could
help
you
with
with
teaching,
but
every
single
time.
I'm
thinking
about
something
I
would
probably
prefer
to
have
three
four
five
images
with
that
particular
environment:
variable
hard
coded
for
teaching
purposes
rather
than
telling
people
to
do
it.
This
way,
also,
there's
the
other
option
that
you
create
a
deployment
and
then
you
for
teaching
use,
cube,
cuddle,
set
and
command,
which
allows
you
to
modify
an
environ,
an
existing
environment
variable.
A
So
the
combination
of
create
deployment
because,
for
example,
you
want
to
show
different
deployment
strategies
or
something
like
that.
You
created
the
the
basic
deployment
and
then
you
through
keep
cuddle
set,
and
you
are
changing
the
deployment
creating
newer
versions,
and
that
allows
you
to
to
show
the
part
of
deployments
the
reverting
and
the
entire
workflow
around
deployments
and
cube
cuddle
at
the
same
time,
without
the
cost
of
adding
the
n
f
to
create
command.
G
I
mean
I
I
don't
know,
I
think
the
the
flags
like
replicas
right
for
debugging
like
do
you
really
need
four
replicas
right
or
like
a
lot
of
these
things.
It's
like
the
debugging
scenario
I
see
is
like
you
want
a
pod
that
you
have
attached
and
here's
the
you
know
the
image
and
then
the
ones
there's
like
other
flags
that
make
more
sense
for
that
right,
like
mounting
in
your
tools
as
your
debug
tools.
G
As
a
you
know,
volume
like
through
a
container
image
right
or
like
there's
some
specialized
things
there
that
just
don't
make
sense
for
like
a
regular,
just
create
command
right
and
most.
A
For
debug
right,
that's
that's
where
the
cube
called
debug
command
goes
into
play
and
that's
that's
what
I'm
saying.
G
I
would,
and
then
yeah
for
teaching
for
teaching,
like,
I
think
like
having
like
the
better
teaching
command,
would
be
just
spitting
the
stuff
out
into
yaml
and
then
piping
that
to
control
apply
right.
That's
really
how
we
should
because
then
you
can
go
and
look
and
see
what's
being
generated
and
it
kind
of
forces
you
to
like
actually
experience
what's
happening
by
default,
and
then
you
pipe
it
to
apply
versus
it's
just
creating
the
thing
magically
and
then
you're
having
to
go.
H
I
used
to
use
the
cube
cuddle
run
command
with
oem
and
dry
run
just
to
spit
out
like
I'd
fill
it
all
out.
How
I
wanted
it
spit
out,
get
it
get
myself
a
yaml
file,
then
I
could
tweak
it
a
little
bit
and
then
I
would
use
apply
so
I
I
sort
of
missed
that
a
little
bit-
and
I
don't
know
I
feel
like-
and
I
know
speaking
as
someone
who
went
through
the
c-cad
exam,
a
lot
of
the
prep
tells
you
to
do
that.
H
G
G
I
can't
like
invest
the
energy
and
actually
doing
that
right,
like
brian
to
your
point
like
doing
like
outputting
that
stuff,
it
actually
fills
in
like
deletion
time
stamp,
which
you
probably
don't
want
right.
It
fills
in
all
this
random
stuff
that
you
don't
want
it
fills
in.
I
think
it
fills
in
replicas,
maybe
even.
C
G
G
Yeah,
exactly
like
the
there's,
there's
like
there's
a
lot
of
legit
reasons
why
people
would
want
to
imperatively,
generate
configuration
and-
and
I
think
mainly
mainly
it's-
that
there's
not
there-
there's
not
going
to
be
one
generator
that
sells
all
those
well.
A
G
Back
yeah,
I
wrote
a
cap
and
I
I
wrote
an
implementation
too,
and
we
could
revisit
that.
The
where
I
landed
on
that
is,
I
actually
found
like
package
it
like
can
just
like
getting
the
configuration
to
be
more
powerful
and
flexible
right
so
like
like
it
goes
back
to
like
there's.
Only
if
you
do
it
based
on
the
api,
there's
only
going
to
be
one
generator
right
and
maybe
brian
you
actually
have
more
knowledge
on
this
particular
thing
than
I
do
right,
but,
like
my
vision
would
be
like
which
you
actually
want.
G
Is
you
want
different?
Like
you,
you
want
a
generator
that
gives
you
a
deployment
with
a
config
map,
with
a
secret
all
wired
together
in
a
specific
way
right
and
then
you
give
it
the
name,
and
then
it
makes
sure
that
name
exists
in
all
the
places
right,
and
that
would
be
like
one
sort
of
archetypal
thing
you
get
with
some
some
different
things
that
you
fill
in,
like
in
sort
of
an
imperative
template
hybrid
or
you
may
want
like
something
else.
G
That's
a
different
set
of
resources
like
that
or
you
may
want
just
a
deployment
right,
but
but
with
an
init
container
filled
in
and
so
that
I
think,
like
what
is
really
going
to
be
best
for
folks
is
to
develop
like
a
canonical
set
of
those
sort
of
patterns
and
have
them.
You
know,
registered
dynamically
either
through
the
api
server
or
just
like
publishing,
container
images,
and
that
you
say
instead
of
instead
of
control,
create
deployment
with
the
image
you'd,
say,
control,
create
and
then
just
give
it
the
image.
G
And
then
it
expands
the
thing
and
then
apply
some
set
of
arguments
that
are
registered
dynamically
through
that.
G
A
Yeah,
I
would
like
to
cut
the
discussion
and
there's
one
action
item
that
I
could
take.
Is
I've
seen
the
cka
exam
being
mentioned
not
once
by
now,
and
maybe
we
could
reach
out
to,
I
think
that's
linux
foundation,
that's
behind
the
exam
and
work
with
them
so
that
they
they
accept.
A
The
fact
that
keep
cal
run
is
not
existent
at
this
point
in
time,
and
so
that
is
more
towards
the
proper
usage
of
of
of
cube
cuddle,
because
if,
if
the
exam
is
pushing
people
towards
run
that's
wrong
and
we
are
in
the
power
in
the
right
spot
to
actually
enforce
not.
A
B
F
I
think
if
you
guys
remember
when
I
actually
first
met
you
all,
and
you
asked
me
what
I
thought
qcpl
needed.
I
actually
brought
this
up
and
I
said
ctl
needs
generate
to
generate
a
scaffold
of
a
deployment
or
a
service
and
just
out
the
ammo
that
you
can
then
go
and
modify
all
right.
So
I
brought
this
up
like
a
year
ago
when
I
met
you
guys
and
I
think
it
still
rains
through
today,
especially
if
that's
what
the
the
the
the
ck
that
the
exam
is
like
they
every
develop.
F
So
I
work
with
beginners
right
like
that's
what
I
do
as
a
developer
advocate
is
I
usually
work
as
beginners
and
teach
them
kubernetes,
and
the
biggest
problem
that
I
see
people
get
started
struggling
with
is
getting
a
blank
amoled
test
to
start
filling
in
right.
If
we
don't
want
them
to
use,
create,
run
in
all
those
other
generator
commands,
we
need
to
give
them
another
tool
that
will
give
them
hey
here's
the
gamble
that
you
can
go
fill
out
with
all
the
fields
that
you
care
about.
F
So
maybe
this
calls
for
another
top
level
cube
cpl
command.
Maybe
we
should
investigate
like
opening
capture
that
right,
because
there
was
a
that's
exactly
what
corey
said
he
was
doing
was
he
was
doing,
keep
secure,
create
oh
yaml
dry
run
just
you
know
and
then
modifying
the
resulting
animal.
So
I
see
this
every
day
when
I
work
with
people,
and
it
sounds
like
it's
true
for
people
in
other
positions
as
well.
G
You
know,
but
really
four
months
when
you
consider
like
that,
you
have
to
get
the
code
in
and
get
it
into
a
release
and
all
that
sort
of
stuff.
So
I
agree
with
that.
I
think
I
think
it
requires
like
so
someone
has
to
like
pick
up
that
torch
of
like
carrying
that
vision.
Right,
like
I
can
imagine,
I
can
kind
of
give
direction,
I'm
kind
of
like
here's.
G
And
then
we
can
pull
it
in
like
if
it's.
If
it's
successful
and
people
like
it
exactly,
then
we
can
just
like
pull
the
thing
in
and
if
we
do
this
well,
we
can
vendor
it.
Instead
of
having
to
like
you,
there's
a
there's
enough
supporting
customized
libraries
now
that
manipulate
yaml,
rather
without
using
the
forbidden,
kubernetes
libraries
that
we
could
do
this
as
a
sub-project
and
then
vendor
it
in
at
a
point
where
it
hit
beyond
alpha
maturity.
G
G
A
So
the
fact
is
already
an
item
that
eddie
will
be
picking
and
then
eddie.
You
want
to
start
with
with
your
first
sick,
sponsor
project.
F
A
And
then
and
then
the
fact
will
be
just
a
pointer,
here's,
the
plugin
that
we're
working
on
that
does
this
for
you,
if
you,
if
you
feel
like
it,
help
with
it
and
and
try
to
push
this
one
forward,
and
we
will
see
where
we
are
in
in
a
couple
of
more
man
in
a
couple
more
months.
G
Help
eddie
with
fleshing
out
a
new
set
of
generator
commands
that
you
can
have
a
lot
more
input
into
the
division
and
really
apply
your
use
cases
and
target
them
at
that
with
the
intention
that
once
they
hit
a
you
know
appropriate
level
of
maturity,
they
could
then
be
pulled
into
the
control
as
a
new
sub
command
group.
B
I
think
it'd
be,
I
think,
that's
a
great
idea
and
that's
definitely
along
the
lines
of
something
I
would
be
interested
in
working
on
something
that
will
fit
my
use
cases.
I
guess
totally
okay,
shifting
conversation
to
focus
on
that,
but,
like
in
my
mind,
for
cube
cuddle,
create
all
the
all
the
options
that
exist
that
allow
you
to
modify
an
appointment
for
whatever
it's
used
and
having
environment
variables
is
something
that
it's
like
to
me
still
kind
of
core
and
doing
set
end
or
outputting
the
ammo
and
then
editing
it.
B
It's
still
something
that's
not
one
step,
and
you
know
I'm
totally
whatever.
What
do
we
decide
to
do
is
something
that
I
agree
with
I
I
would
be
okay
with,
given
that
the
conversation
of
the
given
the
conversation
of
the
what
he
had
just
mentioned,
but
yeah
still
to
me
kind
of
in
my
in,
like
I
guess
inside.
It's
just
something
that
I
feel
like
is
it's
core
to
modify
one
if
you
were
to
use
create,
but
yeah.
G
I
think
I
think,
if
you
and
eddie
I
I
don't
disagree
with
what
you're
saying
I
think,
if
you
and
eddie
are
successful,
all
the
crepe
commands
don't
make
sense
anymore,
like
what
you
will
have
done
is
will
just
be
better
and
so
we'll
deprecate
them.
And
if,
if
you
know,
if,
if
that's
the
case,
then
I
don't
think
I
feel
strongly
about
one
way
or
another.
If
we
add
the
environment
or
not,
because
our
intent
is
to
replace
the
entire
command
set.
G
But
yeah
I
hear
what
you're
saying
like
if
you
view
the
create
commands
as
the
generator
approach,
as
as
the
like.
Less
generator
approach
then
like
having
an
environment
as
part
of
the
I
want
to
generate.
Some
ammo
makes
sense
to
me.
G
If,
if
you
view
the
create
commands
as
effectively
like
a
you
know
a
way
to
not
generate
the
full
yaml,
but
maybe
like
just
get
a
deployment
up
and
running
with
with
a
couple.
G
B
G
I'm
less
convinced,
I
guess
I
think
the
original
intent
a
long
time
ago.
I
was
like
advocating
for
what
you
are,
and
I
wanted
the
great
commands
to
be
like
parody
with
run
and
and
the
discussion
that
came
out
of
that
with
other
folks
in
kubernetes.
Was
that,
like
these,
create
commands
are
supposed
to
basically
just
give
you
the
most
bare
minimum
thing
that
can
run
and
then
you
should
use
like
set
commands
or
other
commands
to
then
do
the
other
pieces.
G
So
that's
something
to
think
about
too.
I
guess.
G
A
I'm
definitely
in
favor
of
we've
had
this
conversation
for
a
year,
maybe
two
by
now,
since
we
I
think
I
seen
so
much
hate
around
my
name
when
I
announced
the
application
of
run
and
then
another
wave
when
I
actually
removed
the
functionality
that
I'm
definitely
in
favor
of
this
approach.
A
A
Alternatively,
I
have
a
directory
on
my
in
my
workspace,
where
I
keep
a
handful
of.
I
don't
know,
deployments
jobs
whatever
I'm
using
for
testing
and
I'm
just
copying
and
tweaking
that
one
particular
piece
of
yum
and
then.
G
G
Let's
come
back
to
it,
because
this
meeting's
gone
on
a
bit
and
I
think
we
could
have
another
just
follow-up
discussion
after
after
we
kind
of
meditated
on
what's
happened
here,
but
why
don't
we
we're
definitely
want
to
do.
The
sub
project
sounds
like
eddie's
willing
to
leave
the
charge
on
that.
If
you
could,
if
you
want
to
co,
participate
in
that
with
him
and
then
help
contribute
your
vision
to
what
these
commands
should
look
like.
G
I
think
that's
something
we
can
move
on
now
and
then,
to
the
extent
like
do
we
want
to
add
environment
flags,
let's
like
revisit
that
in
a
week
or
two
after
we
have
the
sub
project
set
up.
Does
that
work?
For
you
definitely.
A
Okay
cool,
so
we
still
have
about
six
minutes
left
and
the
other
topic
is
the
infamous
get
all
which
pretty
much
falls
in
the
same
category
as
the
create
commands.
A
F
No,
we
can
punt
it.
We
can
put.
I
created
the
faq,
let's
start
filling
it
out
and
then
I'll
open
a
pr
and
get
emerged
as
a
markdown
document
right
in
the
root
of
the
cube
ctl
directory.
So
everyone
can
see
it
and
we
can
also
link
to
it
right
in
brian's
github
template
that
he
created
so
it'll
be
right
there,
but
yeah.
We
can
punt
this
for
next
time
or
or
the
faq
discussion.
A
I
mean
basically,
the
topic
itself
is,
is
pretty
simple:
people
want
to
have
discovery
service
right.
Is
that.
A
Get
all
and
that's
not
quite
get
all
that
each
and
every
single
person
wants
to
see
and
the
theoretically,
the
the
all
alias
is
controlled
on
the
server
and
not
in
the
client.
We
we
used
to
have
it
hard
coded
in
cube
cuddle,
but
once
we
move
the
majority
of
the
functionality
of
the
serve
over
to
the
server,
this
can
be
easily
tweaked
and
chris
nicely
pointed
out
that
there
is
a
plugin
that
is
available
in
crew
cat
all,
which
does
that
particular
thing.
A
So
the
oh.
It's
probably
it's
just
unfortunate
misuse
of
words
and
it
get
some
yeah
and
get
all
was
all
actually
at
the
beginning
of
kubernetes.
But
as
we
grew
three
other
resources,
it
just
wasn't
added
to
the
all
and
all
I
think,
jordan,
commanded
in
several
k
in
several
issues
that
we
are
not
going
to
fix
it,
and
he
put
the
reasons
why
behind
it.
A
So
yeah.
That's
that's.
Definitely
another
topic
for
for
our
fact
that
no
we're
we're
not
in
interested
in
doing
this,
because
that
all
will
behave
differently
on
a
different
server
and
that's
something
that
it's
hard
for
us
to.
Debug
them.
H
A
Definitely
something
that
we
could
consider
and
probably
we
we
should
start
doing
I'll
sync
with
the
sig
api
missionary
about
it,
because
that
particular
alias
exists
on
the
server.
So
we
would
just
need
to
ensure
that
they
start
deprecating
it
and
and
remove
it
eventually,
but
yeah.
That's
that's.
Definitely
a
good
idea.
G
Hey
eddie
and
corey,
just
a
quick
follow-up.
You
could
just
start
in
the
cli
utils
repo,
it's
already
there
for
kind
of
this
sort
of
stuff
anyway,
like
looking
at.
What's
in
there,
it's
like
mostly
next
generation
control
commands
like
there's,
apply
and
preview
and
diff
and
some
status,
so
I
I
think
it
fits
within
that
scope.
G
G
If
you
wanted
there's
some
vendoring
concerns
around
which
way
dependencies
can
go
and
which
arrows
work
too.
So
I'm
sure
we're
gonna
have
to
move
the
code
between
repos
at
some
point
anyway.
F
We'll
probably
just
start
as
a
plug-in,
so
we
don't
have
to
build
cube
ctl
to
test
it
out,
but
the
real
quick
to
close
the
loop
on
the
get
all
one.
That's
totally
fine.
I
just
I
want
to
make
sure
we
have
a
lot
of
issues
and
I
know
everyone's
really
busy,
but
we
have
a
lot
of
issues
that
like
have
a
ton
of
attention
like
this
one,
had
a
bunch
of
comments
and
thumbs
up
on
it
and
then
the
feta
bot
just
kind
of
closed
it
and
people
were
like.
F
A
G
A
If
you
come
up
with
something
like
that
that
we
missed
anytime
over
the
next
two
weeks
feel
free
to
add
it
to
the
future
topics
in
the
agenda.