►
From YouTube: Kubernetes SIG CLI 20190508
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
it's
May
8th-
and
this
is
another
installment
of
six
CLI,
my
name
is
notch
and
I'll
be
your
host.
Today,
before
we
jump
to
the
main
topics
or
two
of
them
I
pasted
the
link
to
the
agenda
in
the
chat
so
have
a
look
at
it.
For
some
reason,
zoom
is
not
working
nicely
for
me
today
when
I'm
trying
to
share
my
screen.
They'll
probably
want
do
that.
Anyhow
announcements.
A
The
next
meeting
in
two
weeks
will
be
cancelled.
You
do
cube
Connie
you,
but
we
are
having
face
to
face
during
cube
con
itself.
So
if
you
have
the
time,
I'm
gonna
paste
the
link
to
the
just
to
the
schedule
where
we
have
the
sick
face
to
face.
If
I
remember
correctly,
we
have
an
hour
and
a
half
or
two
hours
planned
for
ourselves,
so
we
wanna
have
some
PRS
merged
some
caps
reviewed.
That
would
be
the
perfect
place.
A
The
perfect
venue
to
discuss
those
or
eventually,
if
you
have
some
ideas
they
want
to
bring
up.
That
is
the
right
time
and
the
right
place
to
get
them
emerged.
It
applies
to
PRS
I'll,
be
there
for
the
entire
time,
if
I'm
not
seeing
anyone
I'll
just
probably
go
through
the
backlog
of
the
PRS.
That
I
owe
people
reviewing,
if
not
I'm,
more
than
happy
to
talk
with
anyone
else
about
the
direction
we're
going
with
the
60
IANA
QC
deal
project.
A
Also,
with
regards
to
keep
con
tu
a
reminder,
we
will
be
hosting
two
sessions
for
six
CLI,
one
will
be
intro
and
the
other
one
will
be
deep
dive.
Both
Phil
and
I
will
be
present
at
both
of
them
and
from
what
we've
been
discussing,
we'll
be
showcasing
customized
and
we'll
also
do
a
quick
kotor
for
four
cubes
CDL.
So
that's
it
and
then
a
quick
update
from
the
test
grid
Shaun
was
looking
at
the
tests
for
the
past
two
weeks.
How
good
or
bad
are
we
looking
Shaun
so.
B
B
A
The
work
on
that
one
so
I'm
guessing
I
can
send
your
name
for
the
next
two
weeks
when
it
comes
to
looking
at
the
test.
In
that
case
sure
cool
thanks
very
much
with
that
we
can
jump
to
the
main
topics.
Actually,
we
have
three
of
them
right
now:
customized
2.1
new
features
and
I'm
gonna
pass
the
baton
over
to
Jeff.
C
So
I
got
about
10
slides
here.
This
is
what
we're
gonna,
hopefully
release
this
month
with
customize
2.1
and
first
I
want
to
do
a
quick
review
couple
of
slides
to
remind
people
what
it
is.
It's
a
way
to
declare
your
changes
or
declare
the
settings
we
have
for
a
cluster
and
use
share
patches
and
share
changes
that
you'd
like
to
make
between
different
environments.
So
say:
you've
got
multiple
environments,
staging
production
and
development.
You
can
share
a
declaration
of
all
the
resources
you
want
to
use
on
those
environments
and
then
just
patch
things.
C
So
it's
template
free,
and
it's
give
me
another
example:
here's
how
it
works.
You
declare
what
you'd
like
to
do
in
a
customization
file.
This
file
points
to
other
yamo
files
that
contain
the
things
you
want
to
put
in
your
cluster.
Your
deployments,
your
config
maps
and
whatnot,
and
the
customization
file
has
descriptions
of
the
things
you'd
like
to
change.
So
in
this
simple
example:
you're,
adding
a
name
prefix
to
the
resources
that
you're
deploying
to
your
cluster.
C
So
the
project
has
been
going
for
a
while
I'm
gonna
do
a
little
review
on
that.
In
a
second
but
here's,
some
numbers
we're
up
to
about
1200
commits
we've
got
a
lot
of
test
coverage
and
we
got
a
Bonner
and
tenant
outstanding
issues.
They're
growing
dramatically
and
health
is
wanted.
Thank
you
to
sunny
a
zero
eight
to
six
who
started
our
simplified
Chinese
documentation.
I
hear
some
history
the
going
back
to
August
of
2018.
That's
that's.
C
When
Bryan
grant
offered
authored
his
declarative
application
management
for
kubernetes
document,
that
was
an
inspiration
to
start
working
on
a
program
that
we
first
started
developing
in
the
coop
petal
repo.
It
was
called
Kay
expand
and
that
went
through
a
couple.
Different
name
changes
almost
a
year
ago
now
it
changed
to
from
conflate
to
customize,
and
we
settled
on
the
notion
of
a
customization
file
and
in
May,
which
is
about
a
year
ago.
I
think
it
was
May
7th.
C
We
continued
adding
features
of
the
course
of
the
next
year.
Those
are
listed
here
with
some
links
to
some
descriptions
of
how
they
work
back.
In
February,
we
started
to
work
on
our
January
and
February.
We
started
to
work
actually
December
on
moving
things
so
that
it
could
be
incorporated
into
coop
cuddle.
This
required
lots
of
refactoring
lots
of
isolation
of
kubernetes
dependencies
and
some
security
changes,
because
the
fact
that
we'd
be
customized
is
able
to
download
things
from
the
web
and
that's
new
behavior
for
coop
cuddle.
C
C
Apologies,
the
first
thing
is
we're
going
to
have
a
we're
gonna
remove
some
of
the
security
restrictions
or
at
least
allow
somebody
to
get
around
them
in
2.1.
This
is
a
hugely
a
popular
or
unpopular
feature
of
the
security
features
that
we
have
and
it's
kind
of
self-explanatory.
I'm
customization
files
are
unable
to
reach
out
of
their
directory
to
find
resources
by
default
and
by
adding
the
swag
you
can
get
outside.
You
can
share
patches
between
different
customizations,
sex
and
second
thing.
That's
kind
of
new
is
we've
switched
everything
to
go
modules
again.
C
This
is
something
that's
going
to
help
with
integration
into
coop,
cuddle
and
other
other
binaries,
and
this
is
also
going
to
be
the
last
release
where
we
have
kind
of
a
grab
bag.
Everything
in
the
kit,
everything
but
the
kitchen
sink
in
an
unsupported
pkg
directory.
We
don't
formally
release
the
packages.
We
only
formally
release
the
command-line
utility.
So
after
this
release,
we're
gonna
start
thinking
about
more
carefully
about
what
the
packages
should
be
and
how
they
should
be
released.
Third
thing:
that's
new
is
the
ability
to
extend
customize
via
plugins
and
plugins.
C
Customize
does
two
things:
really:
it
just
transforms
resources
and
then
generates
resource
resources.
Historically,
it's
been
able
to
generate
config
maps
and
secrets
and
transform
anything
change,
names
of
things,
patch
things
and
so
forth.
So
we're
adding
extension
points
to
allow
you
to
write
your
own
transformers
and
your
own
generators,
and
what
that
looks
like
is
in
your
customization
file.
You'll
have
two
new
keywords:
the
possibility
of
using
two
new
keywords:
one
is
generators,
the
others
transformers,
and
each
of
these
takes
a
list
of
the
ammo
files.
C
As
its
argument,
each
ammo
file
describes
either
a
generator
or
a
transformer
and
here's
an
example
of
the
secret
maker,
a
secret
generator.
That
looks
a
lot
like
how
customized
behaves
currently,
but
now
all
the
parameters
are
isolated
in
this
one
file,
that's
fed
into
the
into
a
generator,
so
there's
documentation
for
this
there's
a
tutorial
that
shows
you
how
to
do
last.
Mount
customization
in
the
home
chart
lots
of
examples
and
and
I'll
also
in
two
forms:
there's
built-in
plugins
that
you
can
look
at
and
there's
also
lots
of
tests.
C
That
kind
of
give
you
an
end
and
description
of
how
to
use
plugins
third
thing
and
I
think
Jing
phone
is
going
to
talk
about
this.
Some
more
is
the
notion
of
an
inventory
which
is
going
to
enable
safe
pruning.
So
in
your
customization
file,
you
can
specify
key
word
inventory
and
that's
going
to
just
point
to
an
object
like
a
config
map
or
any
other
any
object,
any
company's
object,
because
what
it
does
is
this
inventory?
Is
it
stores
when
you
do
a?
C
This
needs
customize
can't
do
this
by
itself.
Something
else
has
to
be
able
to
consume
the
inventory
and
that's
I
think
what
Jean
pong
might
be
talking
about
anyway,
there's
documentation
check
it
out
and
that's
it.
We're
gonna
fix.
We
fixed
a
lot
of
bugs
I'll.
Give
you
a
final
count
in
the
release,
notes
which
will
be
coming
soon.
That's
it.
A
A
D
D
We
want
to
apply
to
wait
a
block
for
the
status
for
the
aggregated
status
of
different
objects,
so
those
things
are
all
involved
in
Sierra
experimental.
So
today,
I
want
to
mention
that
I
I'd
like
to
request
a
new
repo
with
the
name
CRI
u2's.
So
there
are
mainly
two
reasons
for
this
request.
One
is
for
the
work
we
have
Dalian
Sierra
experimental,
so
we
developed
some
libraries
and
we
will
develop
more
in
that
vehicle.
D
Third
for
consumers,
for
the
users
who
are
going
to
use
that
so
the
name
with
experimental
will
easily
raised
concerns
about
the
maturity
of
what's
inside.
So
that's
why
we
want
the
Ricoh
was
a
different
name.
I
will
create
a
new
record.
Another
reason
of
requesting
base
is
by
this
new
ripple.
We
can
provide
more
modular
utilities.
D
For
example,
we
can
provide
app
like
man
up
library,
independently
from
other
cube
control
commands
like
currently.
If
you
want
to
render
cube
control
apply
library,
you
have
to
render
a
whole
of
it
right.
So
if
we
can
move,
maybe
apply
in
the
future,
we
can
move
up
the
library
into
this
new
utility
repo
and
that
user
can
just
use
the
ut3
who
they
don't
need
to
render
a
whole
of
true
control.
E
What
you
know
we
could
provide
is
just
test.
The
apply
code,
for
instance
and
say
well
actually
apply.
Code
is
good,
for
you
know
minus
eight
versions
or
something
like
that
right
and
you
can
always
run
the
latest
head
version
of
the
apply
container
and
then
now
we
actually
provide
stronger
guarantees
around
the
version.
Skew
of
that
particular
utility,
because
it's
published
independently
and
tested
with
this
semantics.
A
Back
of
utilities
and
helpers
for
both
the
lion
planet,
Earth
plug-in
implementers
and
whatnot
to
be
able
to
not
to
have
to
vent
our
entire
cube
CPL
in
the
near
future,
but
rather
be
able
to
reuse
the
same
primitives
that
we
are
using
already
in
cubes
ETL
and
build
either
plugins
or
their
own
full-blown
commands
if
they
wish
to.
So
that's
my
question.
Why,
rather
having
yet
another
repo
would
not
focus
on
expanding
this
year?
A
E
So,
potentially,
like
one
way
of
structuring
things
right
is
the
library
goes
in
CLI
runtime
and
then
the
utils
repo
is
just
a
place
to
publish
from
where
we
build.
We
wrap
the
library
with
a
Cobra
Cobra
command
that
you
know
parses
flags
into
fields
of
the
of
the
runtime
library
organics.
That
would
be.
A
Yeah
that
would
be
reasonable
having
the
CLI
you,
those
being
more
of
a
set
of
actual
tooling,
that
you
can
use
either
as
plugins
or
standalone
commands
and
having
the
CLI
runtime
still
be
the
library
that
we,
then
we
allow
people
other
people
to
vendor.
I
think
that
distinction
between
the
two
would
be
reasonable.
A
F
A
And,
at
the
same
time,
we
are
actually
look
footing
ourselves,
meaning
we
are
the
primary
consumer
of
the
COI
runtime,
which
indicators
like
gives
higher
confidence
for
consumers,
so
that
we
don't
break
them
in
the
first
place
and
that
we
test
it
properly
before
publishing
yeah.
That's
all
that
okay
make
sense
and.
E
Any
plugins
for
COO
controller
plugins
for
customized
customized,
a
good
question.
Let's
see
so
the
plugins
themselves
could
be
commands
right
so
and
it
depends
on
I
think
the
plug-in
right,
like
you
could
have
a
plug-in
I
could
envision
having
a
plug-in
where
the
contents
of
that
plug-in
live
in
till
I
run
time,
because
it
is
within
the
scope
of
that
repo
and
then
vendored
in
and
then
just
published
as
a
command.
I.
H
C
C
A
A
E
A
E
Say
for
customized,
specifically
like
I,
think
it's
a
good
question:
how
much
a
like
what
what
is
their
responsibility
there
I
could
see
like
I
could
see
as
potentially
publishing
a
plug-in
which
is
like.
We
would
like
to
add
this
feature
to
core
customize.
However,
we
are
alpha
testing
it
by
publishing
it
as
a
plugin
first
and
then
gonna
get
feedback
like
I
could
see
that
sort
of
publishing
the
plugins
I,
don't
see
us
publishing
either
plugins
developed,
like
generally
by
the
community,
because
someone
wants
to
said
operation
right
like
that.
E
That
wouldn't
be
something
we
were
hosting,
nor
like
cloud
provider
specific
plugins,
the
other.
Maybe
the
sort
of
other
sort
of
plugin
we
could
provide
would
be
like
if
there
is
an
integration
with
a
common
ecosystem
component.
That
kind
of
everyone
wants
but
doesn't
really
fit
within
the
scope
of
core
customized,
where
the
project
potentially
aikka.
You
know
without
without
getting
into
specific
details.
I
won't
can't
say
like
whether
we'd
support
that
or
not
but
I
could
see,
like
generally
common
ecosystem
support,
plugins
that
don't
fall
in
scope
to
be
something
we
published
yeah.
E
Curation
is
probably
gonna
be
a
challenge
there.
Maybe
one
thing
we
could
do
is
I
think
get
repo
support.
Tagging,
maybe
I
can't
remember
this
is
true
or
not,
but
I
thought
geckos
support,
like
tagging
with
labels,
and
so
I
could
imagine
this
publishing
a
label
to
say
if
you're
publishing
a
plug-in,
you
can
tell
you
to
repo
with
this
label,
and
then
people
can
search
for
it
and
we're
not
guaranteeing
anything
about
that
plug-in
you're.
Just
aah
panting
into
a
search
index.
Yeah.
G
G
Think
they're
too
fabulous
concepts.
So
congratulations
for
coming
up
with
whoever
came
up
with,
but
from
the
dock.
It
wasn't
clear
if
you're
moving
in
a
particular
direction,
it
seems
with
plugins
you're
extending
the
new
keywords
and
the
customization
a
schema
and
in
the
data-driven
commands
you're
extending
the
CLI
using
new
commands,
yep.
E
The
discoverable
or
data-driven
commands
I
think
facilitate
imperative,
workflows,
debugging
workflows
or
like
development,
workflows
or
bootstrapping
config,
workflows,
I,
imagine
and
do
not
I
do
not
facilitate
like
delivery,
workflows
and
management
application
workflows,
the
customized
ribbon,
transformer
generator,
declarative,
config,
workflows
are
or
commands
are
for
application,
management
and
delivery,
workflows
and
I
think
they
work
in
very
different
ways.
Like
I
can
imagine
the
customized
workflows
to
possibly
go
through
a
git
ops
workflow,
where
humans
never
run
any
of
those
commands
directly.
There's
just
a
repo.
E
That's
handed
to
humans,
I
get
repo
and
they
push
config
into
that
git
repo
and
then
some
team
is
set
up,
something
that
that
invokes
that
machinery
right
so
I
think
like
because
they're
both
part
of
the
CLI
right
in
terms
of
like
that
is
the
way
you
interact
with
them.
But
or
that's
that's
the
way
they
run,
but
in
terms
of
the
sort
of
workflows
they
support
or
the
the
roles
or
the
problems.
E
They're
trying
to
solve
I
think
they're
very
different,
and
that's
why
their
architecture
is
very
different
and
in
Boca
I'm,
not
sure
I
I
know
there
can
be
overlap.
They're
like
do
I,
could
play
the
deployment
declaratively
or
do
I,
create
it
imperative,
Lee
and
I'm,
not
really
sure
I,
don't
think
we're
gonna
have
an
opinion
in
the
future.
That
says
like
no,
you
can
never
to
create
anything
imperative.
E
G
E
E
G
You
know
we're
taking
parameters
so
to
do
negative
tests
to
add
parameters.
I
was
thinking
which
way
do
I
want
to
extend
in
order
to
sort
of
build
this
test,
including
negative
testing
so
and
then
I
thought.
Well,
maybe
the
creators
of
either
mechanism
have
a
thought
of
how
it
would
apply
and
I
could
see
doing
it
either
way,
but
I
didn't
get
too
many
hands
on.
This
would
be
a
good
way
to
do
it
under
these
scenarios.
But
what
you
just
described
actually
answer
some
of
my
questions.
So,
okay,.
A
D
D
A
Have
to
version
them
correctly
currently,
CLI
runtime
is
not
version
it
in
any
way,
I'm
guessing
that
at
some
point
of
time,
we
will
have
to
start
strength,
versioning,
similar
to
what
cubes
ETL
does
I'm
guessing
that
we
will
just
apply
similar
versioning
scheme
to
the
one
that
we
will
come
up
with
for
keeps
you
deal
when
it
gets
actually
to
its
own
repo
and
then
we'll
just
ensure
that
CLI
runtime
follows
the
same
versioning
as
they
keep
CTL,
since
they
keep.
Cto
is
the
first
consumer,
the
main
consumer
of
this.
A
E
E
F
E
G
C
One
customized
2.1
yeah,
that's
the
last
time.
In
fact,
the
PKG
directory
will
probably
be
deleted
and
new
new
packages
will
be
introduced
at
the
top
level.
That
will
have
you
know,
sort
of
formal
agreement
as
to
how
they'll
be
supported
and
they'll
be
version
right
now
there
is
no
formal
support.
It's
completely
using
your
own
risk.
Okay,.
E
A
E
Had
this
on
here,
I
I
have
slides
I
prepared
for
the
deep
dive.
I
mean
I
think
this
is
a
talk.
That's
really
about
the.
What
the
people
attending
want
to
hear.
Unlike
a
talk
on
a
specific
topic
right
that
you'd
give
like
about
a
particular
proposal
or
something
this
is
really
about
the
attendees
getting
access
to
developers
and
learning
about
stuff.
E
They
want
to
learn
about
so
I'm
planning
on
keeping
it
pretty
sparse
in
terms
of
slides
and
just
covering
very
high-level
stuff
and
then
being
able
to
dive
deep
into
whatever
topics
want
to
people
want
to
talk
about.
So,
if
there's
anything
you'd
like
to
see
discussed
again,
not
everyone
can
be
there,
but
if
you
send
either
to
the
mailing
list
or
or
hearsay,
I
really
would
like
to
see.
This
topic
discussed
I'll
make
sure
that
I
cover
it
and
I'm.
E
E
Give
an
update
on
the
docks,
I
still
haven't,
figured
out
how
to
get
this
stuff
out
of
Survey
Monkey
without
setting
up
a
bunch
of
accounts
and
whatnot
I'll.
Just
summarize
that
you
people
have
not
been
filled
out,
the
the
survey
I
think
out
of
a
hundred
visitors
or
out
of
like
a
thousand
visitors
a
week.
We
get.
E
This
is
a
this
is
an
improvement
but
like
there's
too
many
Docs
in
too
many
places
now,
can
you
go
clean
that
up
or
something
like
this
right,
or
can
you
add
a
little
more
detail
on
the
reference
doc
and
so
for
that
stuff,
I,
the
ones
that
I
I
can
address
like
I
added
the
reference
documentation
in
as
a
response
to
that
cleaning
up
all
the
documentation
everywhere?
It's
a
task!
That's
gonna
have
to
think
about
more
and
then
one
out
of
ten
people
is
like
I.
E
Don't
I,
don't
know
what
customize
is
and
why
are
you
telling
me
to
use
it
so
it's
it
follows
like
the
animal
like,
if
you
look
at
it,
probably
on
it's,
like
always
four
and
a
half
stars,
because
there's
always
like
you
know
a
couple
of
people
that
just
ate
any
any
product,
so
I'm
not
too
concerned
about
the
occasional
person
disagreeing
with
it.
As
long
as
the
general
trend
is,
people
are
excited
about.
It.
G
One
thing,
I
think
would
be
very
helpful
and
I
could
assign
myself
this
this
issue.
But
on
all
your
examples
you
know
how
you
sort
of
build
with
a
of's,
no
shell
scripts
and
at
the
end
you
do
the
customized
build,
and
you
say:
okay,
we're
done.
Sometimes
it
would
be
useful
to
have
the
actual
output
of
the
customized
build
embedded
and
the
example.
So
you
can
see
what
customized
actually
did,
rather
than
just
filtering
it
through
a
test.
Yes
or
no.
You
know.
C
Hi
yeah
I
know
exactly
what
you're
talking
about,
and
sometimes
the
output
can
be
long.
So
in
the
interest
of
keeping
short
examples,
we
usually
omit
those
things
that
particular
claim
an
example
is
meant
for
you
to
actually
go
ahead
and
run
on
your
own
on
your
own
workstation.
So
you
should
have
the
output
in
front
of
you.
C
E
E
A
Okay,
so
actually
Raul
Garcia
pasted
the
link
to
the
recording
of
the
last
week's
ask
our
candy
beers
I,
just
pasted
the
link
and
the
agenda,
but
I'm
gonna
paste
it
into
the
chat.
This
is
where
the
actual
code
tour
starts,
but
if
you
want
to
listen
about
other
stuff,
that
I'm
talking
feel
free
to
to
listen
to
the
remaining
part
of
the
meeting,
it
was
myself
and
David.
We
were
guests
of
that
meeting.
David
eat.
A
A
I'm
guessing
Shawn
had
to
leave
earlier,
and
so
we
don't
have
updated
on
the
under
split,
but
I
know
he
has
a
bunch
of
the
arts
out
there.
So
just
go
grab
for
his
name
as
the
author
and
you'll
be
able
to
see
the
progress,
there's
a
lot
of
moving
up
moving
parts.
Still
we
have
about
three
weeks
before
the
code
freeze,
which
is
on
if
I,
remember
correctly,
May
30th
but
I'm
good
quickly.
Look
at
my
notes
in.