►
From YouTube: Kubernetes SIG CLI 20201111 - bug scrub
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,
I
think
that
started
it.
I
will.
I
will
try
my
best
matcha
impersonation
good
morning
good
afternoon
good
evening,
depending
on
where
you
are
today
is
november,
11th
2020
and
thank
you
for
joining
us
for
the
60
li
bug
scrub
today,
we're
going
to
be
focusing
on
customize
and
jeff
is
actually
going
to
be
driving
for
the
first
part
here.
So
I
made
you
a
co-host,
jeff
and
I'll.
Let
you
take
it
away
if
you'd
like.
B
All
right,
let
me
try
to
share
my
screen
here.
Is
that
working
out?
Do
you
see
some
issues?
Yes,
fantastic,
all
right,
so
I
thought
I'd
lead
off
with
the
one
that
might
or
might
not
consume
the
most
time
and
discussion,
this
notion
of
enabling
globbing
in
the
resources
field
of
the
customization
file-
and
this
is
a
great,
well-written
issue,
talks
about
the
pluses
and
one
of
the
many
reasons
why
one
might
add
globbing
so
some
background
in
this.
B
For
the
longest
time
we've
customized
has
gone
with
this
notion
that
you
should
have
an
explicit
manifest
an
explicit
list
of
the
things
that
you
want
to
include
in
your
customized,
build,
as
part
of
you
know
the
whole
declaring
what
you're
doing
and
as
part
of
trying
to
be
as
transparent
as
possible
as
to
what
goes
into
the
build.
So
all
you
have
to
do
is
consult
the
customization
file
now
in
a
git
context.
B
That
may
not
be
as
important
in
a
git
context
if
your
customization
file
is
sitting
in
a
repo
and
it's
check
summed
and
so
on,
you
could
have
globbing
happen
or
you
could
even
have
just
say
you
drop
the
resources
field
and
that's
taken
as
a
signal
to
take
all
the
ammo
insight
into
accounting.
Your
build
now
it's
easy
to
understand.
B
Globbing
technical
problems
arise,
though,
because
when
you
run
the
build,
you
have
lots
of
files
that
you
may
look
at
that
aren't
full
resource
files
that
you
would
not
want
to
include
explicitly
in
your
build.
A
common
example
would
be
a
patch,
so
you've
got
a
patch
that
you
would
like
to
apply
to
some
of
the
resource
and
then
there's
things.
We
can't
anticipate
like
custom,
plugin
or
or
cm
function
that
gets
written.
B
That
has
some
sort
of
file
in
it
and
if
you
wild
card
that
into
the
resource
list,
it's
going
to
be
a
mistake.
So
the
question
is:
when
do
you
want
to
catch
that
mistake?
If
you're
going
to
catch
it
to
build
time,
then
that's
could
be
a
problem
for
you,
whereas
if
you
run
customize
edit,
add
resources
and
then
some
sort
of
glob
that
what
that
does
is
walk.
B
C
C
C
B
Let
me
give
you
another
alternative
related
to
this
is
another
formulation
of
sort
of
the
same
thing,
which
is
the
idea
that
you,
if
you
drop
the
resources
field,
that's
taken
as
a
signal
to
sort
of
take
everything
in
sight,
and
if
you
do
that,
then
you
also
have
to
do
is
introduce
some
kind
of
an
exclusions
field.
So
you
can
specify
the
things
you
don't
want
to
include
in
the
resources
list.
B
E
I
don't
know
how
would
you
I
would
do
this
with
scripting
like
because,
because
I
could
like
it
seems
like
it
would
be
reasonable
to
say
like
if
there
was
a
like?
Oh,
if
you
just
pipe
this
to
this,
then
like
don't
worry
about
it
right
or
something
like
that.
In
this
case,
I
guess
you'd
have
to
do
it
by
using
the
customized
commands
to
edit,
but
you
do
like.
Let's
say
you
do
like
find
with
dash.
Exec
right
is
kind
of
like
a
standard
way
of
doing
this
sort
of
stuff
is.
B
Yeah,
so
the
the
place
where
we
talk
about
issuing
this,
we
it
that
that
doc
says
you
know,
try
to
do.
Try
to
use
out-of-band
use,
customize
edit,
add
resources
which
takes
a
glob
which
does
take
a
glob.
So
at
that
point
you
could
put
that
in
the
script
say
and
you'd
get
pretty
much
the
same
behavior.
B
So
if
you
did
have,
in
that
case,
a
patch
file
that
you
didn't
want
to
have
included
in
the
resources
list,
if
that
was
in
your
directory
and
your
your
script
rand,
this
ad
to
modify
the
customization.yaml
file
you'd,
see
then
you'd
see
the
wrong
output
you'd
get
the
wrong
output.
At
that
point,
I
you
know.
We
had
always
imagined
that
this
edit
thing
would
be
a
human-driven
interactive
thing.
That's
something
running
in
the
background,
because
then
you'd
be
making
the
explicit
choice
as
to
what
you
wanted
to
do
with
your
customization
file.
B
B
E
Yes,
and
like
I
I
mean
it,
I
don't
think
it
directly
answers
it,
but
I'm
also
not
sure,
like
there's
a
great
answer
for
the
question
I
think
like
the
way
like
in
it
and
not
because
I
think
maybe
we
don't
want
to
do
this
right.
So,
as
I
thought
about
the
way
I
think
you'd
do,
this
right
is
like
potentially
in
unix.
E
You
do
like
customize
takes
an
additional
list
of
files
as
arguments
right
in
like
in
an
imperative,
unix
system
right
and
then
she'd
do
like
find
whatever
thing
you
wanted
pipe
that
to
x,
args
and
then
customize
build
right,
and
then
that
would
be
your
like
one
line
of
okay.
You
want
to
cause
like
if
you
want
to
customize
whatever
it
happens,
to
be
in
a
particular
directory
someplace.
E
The
reason
like
I
don't
love,
that
is,
it
means
that,
like
it
moves
away
from
the
declarative
like
whatever
I've
checked
in
is
there,
and
that
would
like
that.
I
think
it's
a
worse
solution
for
the
user
than
having
it
just
take
a
directory
or
a
glob,
or
something
like
that,
because
it
makes
it
more
opaque
to
what's
going
on
right.
B
If
the,
if
the
goal
is
to
make
it
simple,
the
problem,
then,
is
what
about
what
do
we
do
about
the
files
that
are
just
meant
as
arguments
to
things
like
transformers?
You
know
that
aren't
actually
in
the
resources
list,
because
the
glob
will
catch.
Those
we'd
have
to
impose
new
rules,
saying,
like
such
files
have
to
be
in
a
sub
directory
and
then
the
the
globbing.
B
B
B
B
Reasonable
yeah
that
they
would
notice
at
that
point
that
the
that
patch
file
would
get.
D
B
E
C
Plus
one
for
expanding
customized
edit
to
accept
lists,
we
actually
use
customize
edit
for
a
lot
of
our
stuff.
I
mean
it's
kind
of
through
your
script,
but
we
are
using
annotations
on
the
incoming
resources
to
decide
what
we
do
kind
of
like.
Basically,
what
you
propose,
but
I
think
that
is
something
that
belongs
in
my
tooling
and
not
in
some
something
customized
proper.
B
I
think
it
probably
accepts
a
list
because
it
accepts
a
glob
but
we'll
make
sure.
E
B
B
Okay,
this
guy
here
is-
I
wrote
this,
but
it's
got
it
it's
from
input
from
other
folks.
It's
this
notion
of
having
right
now.
Customizations
have
patches
that
allow
inline
patches,
so
you
you
can
specify
a
patch
in
two
different
ways
as
a
file
path
or
as
an
inline
yaml.
That
is
your
patch
and
it
just
sits
in
a
customization
file.
So
this
issue
is
about
adding
the
ability
to
specify
configuration
for
cam
functions
and
plugins
inline
in
the
customization
file
instead
of
having
them
off
in
another
file.
E
Yeah,
I
think
that
that
makes
sense
the
like
yeah,
I
mean
it's
it's
one
of
like
you
said
it's
not
one,
it's
not
something
where,
like
a
user,
couldn't
use
the
system
without
that,
I
think
it
will
demo
more
nicely
like
some
of
these
things.
I
think
of
just
saying,
like
it's
easier,
to
show
examples
when
you
can
inline
like
several
pieces
of
related
content
into
a
single
file,
rather
than
like
showing
an
example
where
you
have
to
open
three
or
four
different
files
to
show.
What's
going
on.
B
So
the
the
open
api
schema
there's
a
person
at
google
who's
been
an
engineer
at
google,
has
been
doing
some
great
work
with
integrating
more
powerful
features
related
open
api
so
that
we
can
manage
custom
resources
better
and
also
manage
different
versions
of
the
not
that
they're
frequently
happening,
but
different
versions
of
the
kubernetes
resource
types.
B
So
this
particular
issue
is
about
right
now.
B
I
should
some
more
history
right
now,
there's
only
one
version
of
the
opa
open
api
types,
as
provided
by
kubernetes,
that's
compiled
into
the
kml
libraries
and
that's
like
117
or
something
I
think
phil
you
did
that
this
would
be
a
way
to
specify
in
your
customization
file,
which
version
you
wanted
to
use
and
allow
multiple
versions
to
be
available,
and
this
is
sort
of
a
segue
into
a
more
complicated
thing
where
you
could
specify
an
actual
file
path
to
a
json,
a
swagger.json
file
that
described
more
types
for
your
own.
You
know:
custom
use.
B
Yeah
it'd
be
part
of
the
customization
file,
so
in
your
customization
yeah
here
we
go
in
the
customization
file.
You'd
actually
specify
an
open
api
field,
which
would
be
a
new
thing
and
then
underneath
open
api
would
be
either
either
version,
which
would
specify
some
hard-coded
version
provided
by
kubernetes
to
kubernetes
type,
or
I
think
a
later
issue
allows
paths
to
be
specified
here
to
swagger.json
files.
Basically.
E
B
That
would
be,
I
think,
the
way
you'd
have
to
do
that
is
like
in
the
in
the
let's
say:
you
got
a
base,
it's
at
117,
you,
you
complete
the
processing
for
that
in
the
base,
and
then
that
comes
spewing
out
as
just
a
bunch
of
resources
and
then
in
the
overlay.
If
that's
asking
for
118
say
all
of
those
resources
are
now
interpreted
underneath
that
set
of
typing
rules,
so
it.
B
It
will
work,
it
will
do
the
right.
I
think
it'll
do
the
right
thing,
but
you
know
somebody's
doing.
That
would
be
that
might
introduce
some
weird
errors.
I
imagine
you'd
be
you'd,
be
downloading,
say
a
base
that
you
didn't
control
or
that
somebody
else
controlled
and
you
were
trying
to
use
a
more
modern
version.
I
don't
know
it'd
be
all
near
messages
for
that.
One.
D
Sean
I
interrupted
you.
I
just
thought
my
microphone
was
working.
Sorry,
just
a
quick
question:
we
don't
actually
contact
an
api
server,
we're
not
going
to
use
the
coop
config
to
contact
an
api
server.
B
There's
a
target
in
the
make
file
to
go
contact
a
server
and
and
grab
the
swagger.
The
open
api
information,
so
that
would
get
compiled
in
this
is
all
about
compiling
it.
So
there's
customizes
does
not
have
any
built-in
talk
to
the
server
stuff.
B
E
A
B
Bailey
by
the
way,
if
you're,
if
you
want
to
talk
about
anything,
go
ahead
and
chime
in
at
any
moment
like
if
you
have
a
something
in
the
back
of
your
mind,
that's
been
bothering
you
that
you'd
like
to
bring
up
the
meeting
that'd
be
terrific.
Otherwise,
I'm
just
gonna
keep
going
through
this
list
for
a
few
more
minutes.
C
Sure
I
guess
I
did
have
one
topic
where
I've
been
working
on
a
plug-in
to
so
we
use
customized
spaces
everywhere
and
our
services
each
service
basically
consumes
a
centralized
base,
defining
like
what
a
go
microservices
and
one
of
the
things
that
is
getting
difficult.
C
Managing
that
is
upgrading
versions
of
that
base,
and
so
we've
been
pinning
it
with
semantic
versions
on
our
git
tags,
and
I
wasn't
sure
if
the
community
would
be
interested
in
a
function
for
running
gogetter
computing,
the
semantic
version
and,
if,
like
you,
have
customize
and
edit
update
and
that'll
update
the
wrap
that
you're
currently
consuming.
That's
right
now
my
approach
for
solving
it
internal
to
us.
I
haven't
open
source
that
or
anything.
C
If
there's
any
guidance
you
wanted
to
provide
on
that,
I
can
create
an
issue
with
basically
that
proposal.
B
Okay,
so
this
would
be
like
modifying
the
edit
command
to
be
more
aware
of
the
structure
of
the
base
target
right
in
the
resources
list.
Yes,
yeah
that'd
be
great.
I've
heard
this
from
other
people
too,
say
just
not
you
just
want
it
like,
like
an
image
tag.
You
just
want
to
modify
the
version.
Yeah
yeah
that'd
be
great.
Please
follow
it.
B
Okay,
mangled
manifest
this
came
in
a
couple
days
ago.
I
haven't
looked
at
this
yet
mangled
manifest
with
n
from
value
from
field
ref
field.
B
B
B
Should
probably
talk
about
plugins,
so
we
have
three
different
kinds
of
ways
to
extend
customize.
B
At
this
point
we
have
go
plugins
which
are
going
to
be
deprecated
in
the
before
we
reintegrate
with
coop
cuddle,
then
there's
just
exact
plugins,
which
are
which
are
what
they
sound,
like
customize
shells
out
to
a
program
that
reads
all
your
resources
performs
some
sort
of
modification
to
it,
and
the
output
of
that
is
absorbed
back
into
the
customized
processing
stream
and
then
there's
krn
functions
which
are
execs
but
then
containerized
and
and
following
a
slightly
different
way
of
expressing
accepting
input
and
providing
output.
B
B
B
B
So
it
looks
like
somebody's
trying
to
get
older
versions
using
the
bash
script
that
we
offer
install
customize.
That
sh,
which
I
think
takes,
takes
an
argument
and
by
default
gets
the
latest
one.
B
So
you
can
always
go
and
download
these
things
from
the
releases
page
by
hand,
so
yeah
I'll
click
on
that
later.
What
does
the
script?
Do
that
go?
Doesn't
at
this
point.
B
E
E
B
D
Can
I
ask
a
quick
question:
jeff
yeah,
so
so
I
know
that
separately
we're
we're
trying
to
to
get
the
latest
version
of
customizing
the
code
control
is.
Is
there
anything
in
our
list
here
where
we
need
to
prioritize
any
actions
or
prerequisites
in
order
to
to
get
that
done.
B
That
is
already
so
in
the
yeah,
so
brian
paisley,
I
just
merged
his
pr
this
morning-
did
I'm
sorry,
brian
prizzly.
He
did
a
bunch
of
implementations
of
that.
We
need
for
switching
away
from
api
machinery
to
kml
the
plan,
for
that
is
not,
as
you
got,
to
follow
issues
to
find
this
plan,
but
here's
the
plan
and
there's
two
issues
that
have
to
be
done.
B
That
thing
is
the
top
of
my
plate,
and
I
now
have
been
sort
of
freed
up
at
work
to
dedicate
myself
to
that,
so
those
two
things
get
done
and
then
what
we
do
is
we
flip
this
flag
to
true
right
now,
it's
false.
We
flip
it
to
true
we
ship
customize.
With
that
thing
on
assuming
this
doesn't
trigger
a
you
know
some
kind
of
bug
storm.
B
Okay,
so
this
straightforward
bug
upload
the
documentation
website,
that's
donnie
wants
to.
I
think
we're
we
need
to
get
a
better
understanding
of
how
the
to
ship
the
documentation-
and
I
know
phil
you've
been
doing
that
with.
What's
the
gentleman's
name,
that's
working
on
the
docs
can't
think
of
it
right
now,.
D
B
C
I
was
just
about
to
enter
the
one
for
the
not
mentioning
the
gcr.io
images.
I
was
pretty
excited
to
hear
about
that
I'll
definitely
be
using
them.
So
I
guess
we
can
skip
over
that
one,
since
it's
not
done
yet.
B
Okay,
that's
a
question.
Convert
all
built-in
customized
plug-ins
to
karen
functions.
It's
sort
of
a
packaging
thing
yeah,
so
part
of
the
api.
The
the
underlying
libraries
for
customize
are
in
the
v0
series
right
now
so
to
responsibly
integrate
them
into
cuddle.
B
We
want
to
release
1.0
and
take
that
opportunity
to
you
know
clean
up
the
api,
make
it
smaller,
maybe
less
packages
in
the
in
the
overall
module
and
not
today.
I
don't
want
to
discuss
this
today
but
decide
once
and
for
all
what
they
have
to
do
about
vars,
which
is
a
feature
that
some
people
really
hate
and
some
people
really
like
bailey.
Are
you
in
the
like?
Do
you
guys
use
vars.
C
We
use
vars,
but
not
I
I
want
to
change
to
setters
and
I
think
I
can
in
almost
all
cases
right.
B
B
So
I
think
that's
about
as
far
as
I
wanted
to
go
today,
unless
anybody
else
would
like
to
chime
in
with
something.
C
Are
you
is
customized
components?
Is
that
feature
going
to
be
kept?
Yes
going
forward
cool?
It
looks
like
we
are
about
to
make
a
pretty
big
investment
in
that,
for
example,
having
a
component
for
enabling
tls
and
that
would
apply
to
all
of
our
stuff.
B
Now
the
only
thing
to
be
dropped
for
sure
is:
go
plug-ins
because
go
plug-ins.
Never
the
go
team
just
really
didn't
decide
to
continue
working
on
that
thing.
B
B
It
does
make
the
code
more
complicated
and
we
do
get
issues
filed
about
it.
A
bunch
there's
an
issue
about
performance
in
bars
haven't
been
impacted
by
recent
changes.
I
don't
know
certainly
there's.
B
Oh
yeah
yeah,
so
one
of
the
things
because
we're
in
the
middle
of
this
transition
between
api
machinery,
parsing
and
kml
parsing
there's
a
lot
of
double
and
quadruple
parsing
going
on
because
of
these
various
layers
that
are
at
play,
that's
killing
performance
and
that
will
go
away
once
we
get
rid
of
the
api
machinery
and
delete
those
delete
that
code
path,
so
the
the
kind
of
performance
bugs
I'm
not
looking
at
them
right
now.
B
B
So
so
you
do
to
use
you
use
virus,
then,
okay,
all
over
into
setters
yeah
yeah,
the
other
thing
there
was
a
replacement
transformer
which
was
this
idea
of
specifying
you
know
what
vars
do.
Is
they
you're
obligated
to
find
you're
obligated
to
find
where
value
comes
from
and
then
use
vars
in
your
resources
to
specify
where
it
goes,
where
the
data
goes,
so
a
replacement
transformer
would
allow
you
to
do
both
of
those
things
in
the
context
of
a
transformer
configuration
you'd
specify
both
where
the
data
was
coming
from.
B
Like
you
say,
a
port
number
would
come
from
some
port
field
and
you
also
specify
where
it
should
go,
so
you'd
no
longer
have
to
sprinkle
var
dal
of
our
things
in
your
dollar
bar
tags.
If
you
will
in
your
resources,
you
would
just
specified
in
the
configuration
for
the
replacement
transformer
and
as
phil
is
brought
up
many
times,
we
can
also
you
know
this.
The
way
setters
work
with
you
know
annotations
directly
in
the
ammo.
C
Yeah,
the
the
main
issue
for
me
is
that,
because
we
have
hundreds
of
services
that
are
consuming
those
bases
I
was
talking
about
before
it
means
that
that
far
sprawl
has
kind
of
happened
across
everywhere
and
so
that
tech,
debt
mountain
is
just
really
growing
a
little
too
quick,
so
moving
to
setters
is
definitely
something
we
want
to
do.
But
it's
kind
of
a
big
big
ask.
C
I
think
we
will
do
it
and
if
you,
if
you
really
hit
home
that
they're,
that
it's
vars
that
are
impacting
performance,
that
might
help
me
make
more
of
a
case
to
get
that
done
sooner
rather
than
later.
B
I
don't
know
that
it's
a
performance
issue
as
much
as
it's
a
you
know
it
sort
of
it
renders
the
yaml
unusable
so
yeah
kind
of
it's
the
one
thing
that
sort
of
breaks
the
spirit
of
like
keeping
the
ammo
usable
at
all
times
and
editing
it
in
the
in
the
course
of
processing.
I
guess.
C
B
I
might
be
wrong,
but
some,
I
think
somewhere
in
the
backlog.
There
is
a-
maybe
it's
only
in
my
mind,
but
there
is
an
issue
to
so
customize
uses,
gvkn
group
version
kind
name
as
a
primary
key
for
the
resources,
and
the
issue
is
to
just
replace
that
with
something
else.
You
know
a
unique
uuid
or
a
hash
of
the
content,
or
something
like
that
and
then
use
that
to
track
the
resources.
The
the
primary
key
is
used
when
you're,
when
you're
identifying,
where
you
want
a
patch
to
be
applied.
B
So
when
you
for
when
customers
first
reads
everything
in
it
could
convert
all
the
patch
targets
into
some
sort
of
unique
identifier
and
then
allow
the
name
field
to
be
modified
safely
and
right
now,
the
way
the
name
field
is
monitored.
The
name
field
is
modified
in
one
way
by
adding
prefixes
and
suffixes.
But
the
way
that's
implemented
is
pretty
pretty
gnarly.
B
C
Yeah,
my
first
prototype
used
prefixes
and
suffixes
and
people
kind
of
railed
against
it.
It
was,
I
guess
too
many
words
for
them.
I'm
not
sure
it
was
my
consumers
that
kind
of
pushed
us
back
into
vars.
There.
C
Exactly
that
that's
the
main
place,
so
maybe
I
can
pick
up
that
issue
and
see
if
I
can,
if,
if
I,
if
we
whittle
that
down,
then
that'll
make
it
easier
for
me
to
move
off
of
ours
in
the
future.
A
Yeah,
this
is
great,
I'm
glad
we
got
through
some
of
your
issues.
We
could
consider
doing
this
more
often
if
it
would
be
useful
right
now.
We
just
do
it
once
a
month.
A
With
okay
couple
pressing
ones
since
we
have
sean
here,
I
dug
into
this
one
yesterday.
A
Sean,
it
was
an
older
one.
A
Which
it
was
the
erd
one
this
one,
so
this
was
additional
printer
columns
for
crds,
don't
work.
This
is
number
517,
so
it
took
me
a
while
to
track
this
down
because
I
haven't
touched
this
part
of
the
codebase
before
and
so
I've
actually
nailed
it
down
to
the
api
server
where,
when
the
table
printer
was
implemented,
a
decision
was
made
to
only
support
one
item,
and
so
I
don't
know
if
this
was
an
oversight
or
we
don't
want
to
add
support
to
it.
But
this
is
exactly
where
that
issue
is
coming
from.
A
Is
we
only
grab
the
one
out
of
the
array.
D
Yeah,
I
think
this
is
this.
Is
this
is
an
issue?
Do
you
want
to
just
assign
it
to
me,
I'm
familiar
with
with
that
table,
converter
stuff.
D
There
yeah
thanks
for
looking
into
that.
It's
actually
you
know
this.
That
type
of
searching
I've
found
to
be
pretty
useful,
for
you
know,
being
able
to
figure
stuff
out
that
I
didn't
know
before
and
generally
get
more
information
or
get
get
a
better
context
on
the
on
the
code.
The
code
base
yeah
for
sure.
So
hopefully
it
wasn't
too
much
of
a
waste
of
your
time.
Hopefully
it
was.
It
was
helpful.
A
No,
I
learned
a
lot.
It's
all
good,
okay,
this
one-
I
was
thinking
chris
about
this.
I
think
we
need
to
handle
this
soon.
So
there's
a
cube,
ctl
crew
plug-in
right
now
called
debug
and
we're
promoting
the
alpha
debug
command
to
top
level.
I
believe-
and
so
we
need
to
handle
this.
D
Before
it
gets
yeah
before
it
gets
promoted,
that's
when
the
collision
will
happen
right.
It's
yeah,
it's
not
not
right.
Now,
right,
I
think
I
think
verb
is
targeting
for
120,
though
yeah
so
verb
is
actually
the
guy
who
owns
the
debug
plug-in.
D
So
you
know,
hopefully
he
would.
He
would
be
able
to
to
manage
that,
or
at
least
he
would
have
the
he'd
be
in
the
best
position
to
to
manage
that
collision
since
he's
in
charge
of
both
of
them.
A
Okay,
I
slacked
it
to
him
so
we'll
see
what
he
says:
yeah
yeah
this
will
be
an
issue
for
other
plug-ins
too,
outside
of
crew
potentially
or
should
we
assign
him
yeah?
I
think
that's
fair
to.
D
Do
or
just
ask
him
what
his
what
his
plan
was?
Maybe
maybe
he
has
this
documented
somewhere.
What
his
plan
was
for
the
for
promoting
to
to
the
debug
command.
A
D
Scope
and
just
for
context,
we
are
testing
regimen
for
anything
other
than
linux
is
actually
woefully
inadequate
and
yeah.
I
mean
not
not
that
we
can
do
anything
here,
but
to
be
honest,
we're
not
we're
not
doing
nearly
enough
considering
how
many
people
use
control
in
windows.
A
A
Okay,
sally's,
not
on
the
call
sally's
been
looking
at
this
one.
A
A
Yeah,
I
also
ask
their
use
case
because
it's
it
seems
counterproductive
to
silence
the
errors.
You
should
be
using
a
you
know,
a
manifest
or
an
apply
because
imperative
versus
you
know
declarative,
okay,
so
I'll
reply
to
I'll
just
I'll
say
I
just
wanted
to
get
another
opinion
before
we
closed
it.
A
A
D
D
D
There
may
be
new,
our
back
stuff
that
I
haven't
heard
of,
but
if
you
look
at
the
so,
if
you
look
at
the
versioning
there,
the
server
version
is
17
in
the
client
version.
D
So,
first
of
all,
it's
out
of
there's
version
skew
there,
but
the
server
version
is
back
to
version
17,
and
I
let
if
you
assign
this
to
me
I'll,
look
up
what
this,
because
I
want
to
know
now
what
this
local
subject
access
review
thing
is
and
if
you
put
triage
accepted
my
guess
is
that
even
if
there
is
a
new
type,
it
just
got
added
in
a
newer.
F
Yeah
they're
sort
of
fake
types
or
dummy
types
that
are
used
to
do
validation
of
authorization.
I
wanted
to
debate
between
authentication
authorization,
but
authorization,
so
they
are.
They
are
used
internally,
they're
a
bit
like
the
cr
certificate
cs,
csrs
right,
which
are
like
sort
of
internal
type,
although
csrs
are
more
real.
This
is
a
this
is
a
very
fake
one.
F
Thank
you
but
yeah.
These
are.
These
are
even
more
like
ephemeral
types
as
my
understanding,
so
so
we
don't
expose
them
or
we
do
well,
you
wouldn't
want
to.
You
normally
wouldn't
apply
them,
for
example,
but
I
don't
know
why
you
would
hide
them
from
from
these
things,
but.
A
A
D
I
I
remember
running
into
this
a
while
back
at
the
very
least,
it's
poorly
documented.
F
There's
also
another
flag
called
hierarchical
correctly,
also
logged
to
standard
error,
or
something
like
that,
and
I
think
that's
the
one
that
is
the
more
normal
one.
But
I
if
I
recall
correctly,
the
basically
the
default
k
log
configuration
logs
logs
only
to
standard
error,
and
so
there
is
a
log
file,
but
it
doesn't
get
activated.
F
So
if
I
were
to
write
to
a
log
file,
that
is
the
path
I
would
write
to,
but
by
default
I'm
writing
to
standard
error
only
and
you
can
say,
don't
elect
a
senator
or
also
log
to
sandwicher,
which
is
another
perhaps
misnamed
for
log
to
standard
file
as
well.
F
D
So
can
we
add
to
this
there's
also
the
also
logged
a
standard
error
flag.
Just
to
so
we
don't
lose
that
context.
The
that
knowledge
nugget
that
justin
just
dropped.
D
Go
ahead
and
assign
it
to
me
betty,
I
I
do
some
k-log
stuff.
F
A
D
D
Yeah,
I
would
imagine
that
this,
would
you
know
the
this
might
mess
up
so
so
this
the
resource
version
is,
is
what's
used
for
optimistic,
locking,
isn't
that
correct,
justin
and
so
yeah
the
the
ways
that
this
could
go
wrong
seem
really
high
and
that
it's
going
on
elsewhere
and
doing
it
on
the
client.
Might
maybe
I'm
wrong
that,
but
I
that's
that
just
seems
really
fraught
with
error.
D
So
so
this
is
yeah.
This
has
been
a
problem
forever,
but
it's
I
shouldn't
say
it's
a
problem.
It's
actually
the
way
that
things
are
and
always
will
be.
It's
you
there's
going
to
be
a
default
version
for
every
single
api
server
that
if
you
give
the
short
name
or
you,
you
ask
for
a
pod,
it's
going
to
try
to
figure
out
which
particular
group
version.
D
D
D
A
Okay,
I'll:
let's
keep
this
one
open
and
I'll
ping
mache
to
follow
up.
Does
this,
it
sounds
like
they're
confusing
something
with
this
chunk
size
which
there's
actually
another
issue
that
is
very
similar
to
this.
This
one.
D
A
D
A
A
A
I
feel
like
these
are
all
the
same
issue.
Should
we
put
a
reference
to
that
other
one
in
this,
then
yeah
I'll
I'll
put
that
together
after
how
much
time
we
got
three
minutes.
Let's
see.
D
D
The
minus
o,
the
you
can
specify
a
template
file
so
that
you
could
spit
it
out
into
the
format
you
want.
F
I
I
might
suggest
it's
only.
It
sounds
like
there
are
two
commands
that
they
actually
particularly
care
about.
We
could
presumably
have
a
a
different
k.
Log
back
end
that
does
output
json,
which
could
somehow
be
opted
into,
presumably
using
some
complicated
flag,
and
then
it
sounds
like
there's
a
handful
of
outputs.
I'm
guessing
like
you've
got
a
port
for
where
they
want.
F
A
A
D
Yeah,
can
we
can
we
add
that
to
the
yeah
to
the
agenda
for
next.
A
D
So
there's
some
underlying
philosophies
that
we've
had
around
for
a
long
time
that
describe
what
whether
or
not
we
allow
people
to
or
whether
or
not
we
honor,
whether
or
not
it's
an
api.
What
the!
What
what
things
look
like
in
what
format
that
they
come
back
and
and
mostly
we,
we
tell
them
this-
this
isn't
an
api
and
it
can
change
the
you
know
the
formatted
output,
and
so,
if
you
guys
are
relying
on
that,
that's
your
problem.
D
I
I
have
another
meeting
yeah
overtime,
anyways,
so
really
appreciate
again
the
the
setup
and
the
organization
and
moderating
eddie
yeah
thanks
guys,
thanks
for
coming
thanks
for
the
help
and
good
to
see
you
justin
thanks
for
for
showing
good
to
see
you.
This
is
fun
thanks.