►
From YouTube: WG Component Standard 20190507
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,
everyone
welcome
to
the
tuesday
may
7th
working
group
component
standard
meeting.
We
don't
have
much
on
our
agenda
today,
but
lubremir
is
going
to
talk
about
flag
overrides.
Take
it
away.
B
Yeah,
so
I
started
drafting
some
codes
basically
about
how
we
can
enable
basically
quotient
component,
configs
or
gvk's.
You
know
to
to
be
able
to
be
overwritten
from
the
command
line,
and
we,
like,
we
briefly
discussed
this
last
time.
I
think-
and
I
and
I
drafted
something
without
cobra.
B
Even
I
just
used
the
native
standard
library
fox,
and
I
was
able
to
get
something
working
basically
in
a
couple
of
hours,
the
whole
code
ended
up
being
something
like
200
lines,
but
there
are
some
questions
like
we
post
the
same
questions
last
time
like
how
do
how
do
we
override
a
specific
version
of
a
config
using
these
flags?
B
B
A
Yeah,
that's
true.
I
think
it's
probably
better
to
include
the
gbk.
B
C
Before
the
only
thing
is
like,
if
you
don't
have
a
right
access
to
a
file
system
or
you're
using
a
shell,
I
mean
I
can.
I
can
see
if
you've
got
big
management,
but
it
it
you've
got
to
like
teach
people
bash.
You
know
like
some
weird
features
like
hear
docs.
In
order
to
to
do
inline
config
like
do
we
have
the
ability
to
pass
a
config
as
a
string.
A
We
yeah,
so
that's
something
we
could
think
about.
It's
like
a
flag
that
just
takes
a
json
string
as
one
possible
option.
A
Yeah
I
mean
I,
I
guess
it
like.
It
would
be
worth
understanding.
You
know
what
workflows
different
people
have
that
they're
struggling
with.
Yes
in
my
workflows,
I
haven't
so
much
had
those
problems.
I
usually
edit
things
locally
and
when
I
you
know
my
own
machine,
I
do
have
access
to
the
file
system
and
then
push
it
to
get
an
automation
takes
it
from
there
or
you
know
you
just
edit
it
locally
and
push
it
right
to
an
api
server.
A
So
I
I
guess
I
don't
have
that
problem.
So
it's
worth
understanding
who
does
and
why.
C
C
B
They
yeah,
they
don't
know
what
the
config
file
even
does
and
they
haven't
read
the
docs
and
they
want
to
create
a
cluster
and
they
just
you
know
they
pass
the
bot
sub
network
as
a
flag
and
that's
it.
But
this
faces
the
problem
where
we
have
to
maintain
this
flag.
B
You
know
as
part
of
the
coi
the
cobra
interface,
and
this
is
not
only
a
problem
in
cuba-
dm
it's
everywhere.
It's
also
in
the
test
test
framework.
B
So,
basically,
I
was
thinking
of
a
way
for
a
generic,
basically
a
generic
way
to
expose
whatever
config
we
have.
But
I
guess
the
point
you
made
about
bash:
we
can
feed
the
configures
stream
this.
This
allows
us
to
basically
have
the
config
without
writing
it
to
this
the
file
system,
but
the
problem
is
that
this
is
not
portable
to
windows,
because
windows
does
not
support
this
type
of
pipes
and
we
either
have
to
write
the
file
to
disk,
or
we
have
to
basically
support
flags
directly.
B
A
C
Right,
I
mean
a
lot
of
people
just
like
to
work
with
a
single
binary
and
if
you're
trying
to
install
kubernetes,
you
have
to
figure
out
where
to
like,
put
the
binary
and
then,
if
they
can't
figure
out
how
to
override
where
the
config
file
is
read
from.
Then
that
can
be
frustrating
yeah.
But
that's.
A
A
C
D
B
Mean
the
other
reason
is,
people
are
like
we
discussed
with
people
are
lazy.
They
don't
want
to
understand
what
the
company
config
is
about.
They
just
want
to
use
a
single
flag
and
if
we
don't
want
to
maintain
the
separate
set
of
flags
for
because
of
these,
these
lazy
users,
we
basically
have
to
have
this
mechanic,
where
we,
you
know
copy
the
fields
that
basically
I'm
proposing
to
copy
the
fields
into
flags
automatically.
C
Yeah
I
mean
it.
It's
also
worth
considering
like
think
about
how
much
easier
a
docker
image
is
to
use
from
flags
versus
configs.
C
A
C
My
instance
is
that
it's
like
very
thoughtful
to
be
thinking
about
like
versioning
and
docs,
and
that
kind
of
thing,
but
honestly,
if,
like
the
flags,
were
best
effort
and
they
were
just
like
parse
the
dot,
notation
string
or
took
like
a
json
patch,
and
then
it
did
that
to
the
config
object
before
like
decoding
it.
Then
that
sounds
pretty
nice
to
me.
C
A
There
it's
it's
one
of
those
things
where,
as
soon
as
you
say,
it's
best
effort
going
forward,
then
there's
a
chance
that
something's
going
to
happen.
It's
not
clear
what
but
they're
you
know
they're
different.
I
can
already
see
different
things
right,
like
okay.
Well,
now,
config
is
coming
from
two
places.
Are
there
two
people
responsible
for
it?
You
know:
does
the
person
changing
the
flags
aware
of
what's
in
the
file
if
they're
not
even
bothering,
to
learn
about?
What's
in
the
file?
A
There
are
plenty
of
config
fields
that
are
interrelated,
so
if
you're
changing
one,
you
should
know
about
the
others
right.
So
that's
a
that
could
be
a
source
of
production
issues
if
you're
just
you
know,
blanket
overlaying
strings
in
this
thing
before
you
decode
it,
you
know
maybe
you're
gonna,
maybe
it's
easier
to
break
the
decoding.
A
I
don't
know
again
there's
the
version
resolving
what
version
of
the
api
retargeting.
So
I
I
think
it's
just.
I
think
as
if
it,
if
the
only
problem
is
that
people
are
unfamiliar
with
the
config
file
and
haven't
been
able
to
adapt
to
that.
Yet
I'm
I
think,
it's
too
early
to
say
that
that's
a
real
problem
because
most
things
the
config
files
have
only
been
like
alpha
or
maybe
beta,
without
full
coverage
of
the
config.
A
A
B
Yeah,
basically,
that's
what
I'm
suggesting
to
users
all
the
time
like
they.
We
don't
have
a
flag
in
kubernetes
for
something
I
say
this
is
already
available
in
the
config,
but
you
know
one
of
the
main
people
who
has
been
pushing
pushing
this.
Like
a
generic
question
in
the
kubernetes
space
is
tim
sinclair.
Who
is
basically
asking
where,
when
is
the
component
startup
working
group
is
going
to
start
working
on
the
flag
overrides,
and
I
realized
that-
probably,
oh,
probably
only
lucas
knows
about
this
yeah.
B
Yeah
yeah,
so
it's
kind
of
a
weird
space
like
do.
We
even
need
this,
like
you're,
saying
that
people
should
really
use
the
config
instead
of
these
flags,
and
so
basically,
what
I'm
going
to
do
is
I'm
going
to
draft
some
code,
probably
write
a
google
doc
about
this
and
I'll
show
you
guys,
maybe
next
time
or
the
week
after
what
I
have,
and
if
you
agree
with
this,
we
can
proceed
with
a
cap
or
hold
it
in
case.
The
idea
is
not
very
good.
B
Maybe
I
can
invite
team
as
well
to
join
the
to
basically
to
outline
the
whole
problem,
because,
according
to
him,
it's
a
generic
problem
in
kubernetes
and
it's
not
like
only
kubernetes.
A
A
B
I
have
one
example
with
the
couplet,
for
instance,
the
kublet
has,
I
don't
know.
Hobby
reflects
like
more
than
a
hundred
or
something
yeah.
It's
like
120,
or
something
like
that
yeah.
So
we
have
the
the
the
basically
the
the
version
api.
B
So
if
we
at
some
point
removes
all
the
flags,
we
can
expose
the
whole
api
via
flags,
but
again
it's
like.
Should
we
do
it?
Shouldn't
users
instead
just
use
the
config.
A
B
A
D
C
No,
I
just
finished
up
some
stuff
for
the
add-ons
sub
project,
so
I'm
going
to
be
looking
at
some
rebases
today
and
yeah.
That's
all.
C
Yeah
we
de-scoped
the
the
coupe
controller
manager
feature
for
115.
A
C
And
I
think
stuart
is
going
to
start
working
on
a
kep,
for
that
awesome
sounds.