►
From YouTube: Kubernetes SIG CLI 20200520
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
My
name
is
Martin
I'll,
be
your
host
today,
honestly,
I
don't
have
any
particular
announcements,
although
I
I
wanted
to
check
the
if
the
next
Wednesday
isn't
the
time
for
our
blocker
box
crap,
but
of
course,
I
run
out
of
time
and
just
looking
at
my
calendar.
Yes,
the
next
Wednesday
is
our
monthly
bucks
crop.
So
you
are
more
than
welcome
to
join
the
box
crop
next
week.
A
B
Okay,
cool:
can
everybody
see
this
yep,
okay
yeah?
So
in
last
week's
meeting
or
two
weeks
ago,
I
guess
on
that
mentioned
that
we're
pretty
close
to
being
done
with
the
work
for
supporting
multiple
indexes
and
crew,
and
so
I
just
wanted
to
show
a
quick
demo
of
that
and
so
I
just
set
up
this
like
dummy
kind
of
custom
index
in
my
own
github.
But
so
I
want
to
showcase
just
adding
that.
B
B
B
B
So
crew
doesn't
have
a
file
to
keep
track
of
indexes.
It
just
clones
the
index
into
this
index
repository.
So
we
have
this
default,
one
which
is
crew
index
like
the
central
index
and
then,
when
you
add
a
new
index
and
clones,
the
one
you
added
by
name
into
this
directory
here
in
the
in
like
the
dot
crew
folder,
where
it
stores
everything.
D
A
A
My
only
request
for
you
Alex
would
be
if
you
could
fit
in
because
I
remember
the
last
time
we
spoke
about
the
topic
for
the
majority
of
the
meeting.
I
would
like
to
time
box
that
and
within
the
next
15
minutes,
let's
say
so
that
we
have
to
top
it
at
the
time
necessary
to
cover
the
the
remaining
two,
especially
the
last
one,
so
I'll
keep
you
I'll.
Keep
you
in
touch
in
line
with
the
15
minute
deadline
for
the
topic,
I
like
to
call
yeah.
E
E
The
user
story
that
was
discussed
in
the
previous
six
year
by
meeting
is
has
been
act
by
Jeff
and
Phillip,
and
it's
also
merged
and
customized.
So
this
is
good
and
there
is
not
much
on
this
front
I
think
when
people
who
can
see
the
cap,
we
can
have
more
to
discuss
to
talk
about.
So
that's
about
it
from
my
side,
I.
F
A
F
A
G
Yeah
I
just
can't
hear
me
yeah
loud
and
clear,
yeah
yeah
so
on
the
the
the
III
actually
have
some
questions.
The
first
days,
as
we
all
know
the
cook
control
wrong
and
now
has
only
support
for
port
north
deployments
right.
It
could
control
wrong.
So
now
we
will
create
deployment
using
cook
consecrate
deployment,
but
this
command
doesn't
support
a
port
flag
and
the
replicas
fact
right
so
I
believe
there
are
some
peers
and
is
working
with
some
peers
to
add
the
port
flag
and
the
reflux
fare
to
cook
control,
crate
deployment
right.
A
At
the
same
time,
we
need
to
be
very
careful
not
to
add
each
and
every
single
possible
bit,
because
if
somebody
is
thinking
about
the
overall
goal
for
the
create
commands
is
to
help
you
get
started
with
writing
deployment
with
writing
stateful
set
or
any
other
resources.
This
is
for
newcomers.
If
you're
someone
that
is
a
long
term,
grantees
user
99%
of
cases,
you
will
be
writing
those
Jung's
or
Jason's,
whichever
you
prefer
by
hand,
because
you
will
be
tweaking
much
more
than
it
is
possible
with
the
create
commands.
A
I,
don't
want
to
end
up
in
a
situation
where
create
commands
are
super
bloated,
like
the
run,
was
because
the
problem
with
with
having
this
many
options
is,
it
will
be
confusing
to
users
and
honestly,
it
often
happens
that
when
I
need
sample
resources,
I'm
gonna
just
invoke
create
command
with
a
dry
run
and
pipe
it
into
a
file.
Just
so
that
I
have
the
template
of
a
resource,
and
then
I
can
add
the
additional
things.
A
A
lot
of
people
are
cleaning
and
I
got
a
feedback
from
dogs
that
it's
hard
for
them
to
replace
the
documentation
a
bit.
So
it's
I
think
it's
reasonable
to
ask
those
particular
cases
for
the
create
command.
So
yeah,
that's
that's.
Definitely
something
that
I
want
to
address,
but
I
like
I,
said
it's
always
case
by
case,
think
whether
it
makes
sense
in
the
overall
overall
picture.
It
should
not
blur
the
overall
perception
of
the
command
and
the
command
is
simple
jump
start
nothing
more.
H
Yeah
yeah,
just
just
sorry,
I,
just
let
it
emphasize
I
mean
I
think
we
are
all
on
the
same
page.
We
all
understand.
We
definitely
want
to
emphasize
the
apply
use
cases
and
not
the
create
use
cases.
This
is
there's
been
a
you
know,
obviously,
over
the
years
plenty
of
use
cases
for
for
create,
but
but
that's
that's
not
the
assemble
we
want
to
emphasize.
We
want
to
emphasize
the
declarative
apply,
which
is
you
know
the
the
direction
we're
moving.
It.
G
A
Written
for
the
run
command
and
when
we
eventually
stopped
the
ever-growing
run
command,
they
overlap
to
create
commands.
But
what
we
actually
want
to
see
within
the
run
commands
is
simple:
commands
create
I,
don't
know,
job
create
row
binding.
Several
commands
are
already
there
simple.
We
literally
have
a
a
resource
structure
inside
of
the
command,
how
it
produce,
and
it
just
throws
it,
there's
no
need
for
fancy
generators.
A
G
I
A
Well
aware
of
that,
one-
and
this
is
definitely
not
something
that
we'll
be
able
to
solve
within
a
couple
of
days-
it's
rather
a
long-term
goal
to
get
rid
of
the
generators.
I
already
started,
deprecating
the
generator
flag
for
create
commands,
because
it
does
not
make
any
sense
for
create
commands
if
I
say
explicitly,
create
deployment.
A
A
Okay,
cool,
so
that
we
can
move
on
to
topic
from
Sean,
which
is
Kip
Connell
edit
or
apply,
cannot
updates
that
as
resource
a
yeah.
Wasn't
we
yeah
go
ahead,
Sean.
H
Yeah
I
just
wanted
to
bring
this
to
the
attention
of
the
community.
There's
both
Tim
socket
man
and
Jordan
Leggett
said
also
think
that
this
is
important.
I
think
that
if
you
go
down
further
I
think
Jordan
mentioned
it's
not
just
editing
the
status,
but
he
I
think
he
mentions
multiple
like
the
patch
command,
as
well
as
even
apply
basically
being
able
to
modify
sub
briefs,
but
yeah
he
mentions
get
edit
apply
a
patch
seem
to
make
sense,
and
this
is
a
no
and
annoying
gap.
A
Try
to
write
it
down
a
proposal
how
this
could
look
like
it
will
be
definitely
something
bigger
and
especially
that
Jordan
is
mentioning
for
commands.
I,
wonder
I,
wonder
if
there
won't
be
more
than
just
these
four,
but
I
can't
think
of
top
of
my
head,
and
then
we
can
definitely
consider
this
as
a
feature
development
for
someone
that
is
interested
in
working
on
this
one.
It
would
be
nice
to
have
a
consistent
mechanism
across
all
four
of
these
commands.
H
Okay,
great
yeah,
so
I
just
I
just
wanted
to
point
out
that
up
until
now,
it's
the
the
spec
has
been
editable
and
the
status
is
not,
and
so
this
would
be
a.
It
would
be
a
significant
change
in
in
how
we
approach
resources
in
that
up
until
now,
the
server
has
only
been
the
the
actor
who
modifies
this
particular
sub
resource.
H
A
H
Great
I'll
take
the
I'll
take
on
the
action
item
to
to
try
to
escalate
this
to
Architecture.
Does
that
sound,
reasonable
yeah?
That's
something
and
and
I
think
that
it
will.
You
know,
I,
don't
I,
don't
imagine
a
lot
of
pushback
since
Tim
and
Jordan
are
behind
it
and
I'll
get
back
to
the
rest
of
the
community
and
if
anybody
else
who's
interested
in
it's,
you
know
a
pretty
significant,
visible
change
to
coupe
control.
Then
you
know
please
reach
out
to
me
or
to
my
chain
to
say
you
to.
G
C
F
A
It's
a
problem
I
mean
the
only
reason
that
we
did
the
only
reason
that
edit
is
using
patch
it's
because
at
the
time
when
added
was
has
been
added,
the
patch
was
the
only
option.
An
apply
was
not
there.
Yet
if
I
remember
correctly,
there
is
nothing
stopping
us
from
actually
changing
the
Edit
underlying
code
from
patch
to
apply,
especially
that
the
server-side
apply
is
becoming
more
and
more
important
for
us.
Then
it
will
eventually
go
to
GA.
C
K
A
A
F
Yeah,
that's
hi.
This
is
Jeff
I'm,
sorry
about
the
regressions.
We've
got
extensive
pre
submit
tests
and
release
tests
we
perform,
but
occasionally
something
slips
through.
The
release
process
is
described
in
the
releasing
directory.
It's
a
slow
directory
and
the
customized
repo
there's
a
contributor
who's
building
a
go
program
to
replace
the
bash
script.
That's
in
there
right
now!
That's
that
we're
using
for
releasing
but
yeah.
The
goal
is
obviously
to
release
without
having
any
without
introducing
new
bugs,
but
sometimes
it
happens.
I
mean.
J
If
I
understand
Bailey's
question
correctly,
it's
that
her
organization
has
a
lot
of
tests,
probably
use
customized
in
some
pretty
bad
hit
some
edge
cases
in
pretty
extensive
ways
and
that
if
there
was
a
canary
channel
for
them
to
build
off
of,
they
could
run
through
their
own
test.
Integration
test
suite
and
since
they're
tested
had
some
of
their
aggressions.
Have
there
been
a
canary
channel
and
had
they
been
running
their
integration
tests
against
that
canary
channel
one?
L
Yeah
and
obviously,
if
I
ever
catch
anything
in
those
tests
step,
one
is
definitely
to
get
coverage
back
in
to
customize.
The
trouble
for
me
is
right,
where
not
all
of
our
manifests
have
released
yet
so
I
don't
have
those
out
on
github
right
now,
but
I'm
working
on
trying
to
get
like
a
canned
version
out
there
set
it
up
with
github
actions
so
that
anybody
can
get
feedback
off
of
even
on
their
PR.
F
So
it'd
be
fantastic
to
get
more
test
coverage
in
repo,
so
you
know
we
could
head.
We
can
take
advantage
of
those
tests.
There
is
no.
There
is
no
staging
release
channel
at
this
time.
So
if
you
have
any
proposals
about
how
that
might
work,
that'd
be
great,
there's
just
them
ahead
and
then,
of
course,
they
officially
released
versions.
J
F
F
Sure,
there's
also
built
into
our
current
process.
We
we
do
have
a
period
where
we
have
a
draft
release
up,
but
that
doesn't
sit
there
for
a
week.
Typically,
it
only
sits
there,
for
you
know
on
the
order
of
hours,
so
we
could
try
it
with
the
next
release.
In
fact,
it
was
going
to
cut
one
today,
moving
up
to
I,
think
3.6
and
we
could
put
a
draft
up
and
then
it's
available
for
download.
L
J
F
J
F
All
right,
I'm,
the
only
stand-up
so
as
they
just
mentioned,
we're
gonna,
do
a
release
today
and
we'll
leave
it
in
draft
status
for
and
available
for
download
for
the
course
of
the
week.
I
will
ping
ricochet
I
just
want
to.
Let
everybody
know
that
I'm
working
on
this
refactor
to
remove
all
of
the
kubernetes
core
dependencies
from
customize,
the
API
machinery,
etc.
F
H
There
was,
there
was
just
one
thing:
I
wanted
to
share
with
the
community
before
the
end
of
the
message.
End
of
the
meeting
I
just
wanted
to
give
a
big
shout
out
to
one
of
our
contributors
who
I've
seen
doing
lots
of
work
and
seen
on
lots
of
issues.
Doing
you
know
lots
of
resolution
and
that's
Brian
personally.
Is
you
know
I've
seen
you
know
him.
Tackling
I
know
that
there's
a
lot
of
people
out
there
doing
good
work.
E
Since
stand
up
and
talk
about
customized,
there
was
something
I
had
forgotten
and
had
asked
on
the
issue,
but
I'm
not
sure.
If
jeff
has
seen
this
Jeff
I
saw
that
you
have
merged
the
examples
for
for
the
composition
stuff,
but
the
call
is
not
there
yet.
So
how
do
you
want
to
tackle
this?
Do?
Do
you
want
us
to
write
an
update
to
dot
that
says
that
it's
not
it's
not
a
feature
that
that's
ready
yet.
F
Yeah
they'd
be
good,
just
throw
something
at
the
top
of
the
deluxe,
so
those
Ducks
a
lot
of
the
markdown
that
we
have
have
annotated
in
them
that
make
them
part
of
our
feature.
Tests
and
those
particular
examples.
Don't
have
that
yet
so
they're
not
really
they're,
not
dangerous
in
that
respect,
nothing's
failing
because
the
feature
is
missing,
but
yeah
I
agree,
sort
of
something
man
in
there
saying
that
this
is
part
of
the
cap,
pointed
to
the
cap
and
mention
that
it's
alpha.
That's
all.
Okay,.
E
D
E
A
D
It's
okay
for
context
for
everyone
else:
I'm
trying
to
abstract
everything
from
this
PR
of
adding
a
cubes
detail
tomb
and
into
documentation
so
that
we
can
point
to
like
here's
everything
that
we
recommend
for
best
practices
for
adding
a
command.
So
if
everyone
can
go
and
pile
on
a
review
and
just
add
your
input,
so
we
can
get
all
the
stuff
together.
So
I
can
document
it
up
and
I'll
drop
that
in
the
list
somewhere
here
to.