►
From YouTube: Argo Contributor Experience Office Hour 8th Apr 2021
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
A
Okay,
so
we
had
several
topics
and
seriously
I
put
the
topic
to
review
feedback
before
upset
topic,
because
we
kind
of
we
wanted
to
talk
about
it
during
class
meeting
didn't
have
time,
so
I
think
it
yeah
just
don't
want
to
postpone
it
to
another
meeting.
Thank
you
admission.
A
So
basically,
like
a
couple
weeks
ago,
we
asked
people
to
give
us
feedback
in
general
about
argo
cd,
in
particular,
about
ui
and
originally
created
this
discussion.
So
and
I
guess
we
wanted
to
just
go
through
it
and
see
if
we
can
extract
useful
information,
so
originally
didn't
prepare
anything
or
I
can
just
go
through
it
and
you
know
looking
kind
of
free.
B
Some
of
them
you
did
respond
to
already,
and
a
lot
of
it
was
like
more
to
do
like
we'll,
add
this
as
an
issue,
and
so
I
think
the
main
point
of
this
is
to
discuss
each
each
piece
of
feedback
and
just
make
sure
that
we
have
it
like
an
action
taken
for
it.
If
it's
something
that
we
want
to
implement.
A
All
right,
okay,
so
in
this
case
I'm
just
going
to
start
from
from
the
top
one
and
then
go
for
all
of
them:
okay,
so
this
first
feedback,
I
think,
he's
mostly
talking
about
user
interface,
and
you
know
how
it
looks
like
on
smaller
devices.
So
he
is
complaining
that
show
orphan
button
is
not
visible
and
hard
to
click
and
in
general
you
know
some
some
buttons
won't
show
up
on
on
mobile
screen
and
I
think
that's
fair
and
even
like
yesterday,
someone
sent
us
pull
requests
that
improves
exactly
that
part.
A
But
I
think
we
have
a
lot
of
pages
that
don't
really
work
on
mobile
devices
and
I
can
give
some
context.
We
kind
of
knew
that
it's
not
working
on
mobile
devices,
but
so
far
everyone
was
thinking
that
people
usually
do
not
deploy
using
mobile
devices
and
not
really
using
you
know,
phone
to
access
august
gui,
but
I
might
be
wrong.
B
Right,
I
think
I
think
that's
been
brought
up
a
couple
times
if
we
should
account
for
mobile
devices,
and
I
think
that
there
should
be
some
kind
of
survey
sent
out
to
people
to
see
if
this
is
a
need
in
the
community
and
if
it
is
then
make
it
an
issue
for
it
and
if
it's
not,
then
just
state
somewhere
that,
like
we
don't
current,
we
aren't
currently
planning
on
implementing
this.
How.
A
About
if
we,
you
know,
get
kind
of
around
this
survey
by
creating
issue
and
then
just
ask
people
to
thumbs
up,
if
you
need
it,
that's
kind
of.
A
D
Yeah
so
yes,
it'd
be
good
to
capture,
I
guess
a
level
of
interest
and
like
mobile
support
for
it.
Of
course
you
know,
and
then
you
know
it
may
be
something
that
we
leave
for
the
community
pick
up.
Unless
you
know
someone
specifically
has
a
need
for
it,
but
at
least
we
would
have
identified
the
need
and
documented
it.
E
A
Yes-
and
I
think
there
are
a
few
ways
it
can
be
solved
like
maybe
maybe
the
better
way
is
to
create
mobile
application
because
we
technically
ui
doesn't
have.
We
do
not
expose
any
server
side
kind
of
capabilities
for
ui.
It's
just
the
expose
api
and
user
interface
is
just
a
single
page
web
application
that
uses
that
api.
So
mobile
application
can
use
the
same
api.
D
A
D
B
A
Yes,
I
mean
all
right,
so,
let's
move
to
next
one.
So
this
is
it's
kind
of
often
complain
and
I'm
sure
we
already
have
back
for
it
so
right
now,
user
interface
is
not
even
trying
to
see
what
kind
of
permissions
user
have
and
just
show
all
kind
of
buttons,
like
all
supported
buttons,
and
I
think
we
have
everything
we
need
to
improve
it.
We
just
need
to
work
on
it,
so
we.
A
That
can
tell
whether
you
have
access
to
let's
say
sync
application
or
you
don't
have
access
to
that
action,
and
so
ui
can
use
that
api
to
decide
if
button
should
be
even
shown
or
it
should
be
hidden-
and
I
kind
of
I
didn't
even
reply
anything
here
because
like
yes,
I
think
we
should
reply
and
we
should
say
yes
agree.
It
should
be
done
one
day.
C
What
is
the
future
of
this
area
of
work
and
then
do
it
in
future,
because
I'm
pretty
sure
whoever
has
asked
for
it
doesn't
have
an
urgent
need,
because
so
it's
fine
to
take,
take,
take
some
time
and
figure
this
out,
because
it's
a
bit
of
commitment
to
ensure
that
you
know
the
ui
works
across
multiple
things.
A
A
A
F
E
A
Something
we
should
consider
next,
and
this
is
a
fair
complaint,
so
the
person
kind
of
pointing
out
that
we
have
a
lot
of
annotations,
that
kind
of
supposed
to
be
first
first-class
crd
fields,
and
so
you
know
if
it's
a
field
of
facility,
then
I'd
be
sure,
kubernetes
to
say
sugar
can
validate
it
and
here's
example
of
session
notations,
and
I
responded
kind
of
why
so
some
of
those
annotations
are
part
of
our
goal,
labs
project
and
I
feel
like
once.
A
We
maybe
decide
to
crucified
labs
project
into
and
merge
it
into
our
ocd.
Then
we
can,
you
know,
can
convert
annotations
to
two
fields
and
another
example
of
such
annotations
is
we
introduced
manifest
generate
paths.
So
these
two
annotations
are
part
of
argo
cd
already,
but
I
think
they
were
kind
of
experimental.
So
we
maybe
we're
not
sure
and
that's
why
we
introduced
them
first,
as
an
annotation,
so
pretty
much.
C
Yeah,
we
just
have
to
come
up
with
a
path
to
deprecate
the
annotation
usage
as
well,
because
that
is
an
api
contract
anyway,
people
who
are
using
that
we
may
continue
using
that
for
next
many
many
reasons
we
just
have
to
come
up
with
a
path
for
it,
but
I
agree
with
this.
A
And
okay
and
any
action
item
like
do
we
want
to.
Basically
we
don't
have
a
ticket
for
that.
We
don't
have
anything
to
you,
know
yeah,
maybe
should
we
add
a
note
to
create
ticket
and
then
later
we
can
create
it.
I
think
it's
good
to
track
that.
Eventually
we
should
do
it
right
and
next,
okay,
it's
thank
you
for
the
for
the
work.
A
Okay,
and
I
think
this
was
a
big
comment
and
several
unrelated
topics
mentioned
in
that
comment.
So
first
was
request
to
add,
make
argo
cd,
util
more
kind
of
easier
to
install
and
yeah.
This
is
it's
not
yet
done,
but
I'm
doing
it
today.
Basically,
we
have
so
maybe
unless
you
know
maybe
you
already
know
in
that
release,
we
changed
rbcd
binary.
A
So
basically,
all
the
binaries
that
argo
cd
has
are
now
kind
of
packaged
into
the
available
in
one
file
and
we
use
the
trick,
which
is
we
found
it
in
a
busy
tools
image.
So
basically,
we
have
a
binary
that
has
all
the
functionality
and
it
checks
its
own
name
and
if
it
looks
like
rbcd
util,
then
it
behaves
as
rfc
util,
and
so
what
that
gives
us.
It's
really
easy
to
publish
argo
cd
util
in
the
home
view.
A
It's
called
recipe.
I
guess
you
can
just
add
a
sim
link
to
existing
cargo
cd,
binary
and
then
next
to
the
next
person
who
installed
for
upgrades
argo
cd,
homebrew
to
2.0
will
get
two
binaries
instead
of
one,
but
without
you
know,
increasing
the
file
size
so
yeah.
Basically,
the
first
request
from
from
the
user
pretty
much
we're
going
to
address
it
today
and
next
request
was
about.
A
Okay,
that
was
about
a
new
command
in
rbc
util
that
he
had
implemented.
So
the
person
asked
about
ability
to
support
validating
policy
in
the
cluster
as
opposed
to
from
from
file,
and
I
think
he
said
that
he
thinks
it's
a
bug
and
I
kind
of
I
incorrectly
answer
it
that
I
think
it's
a
bug,
but
it's
not
about.
Basically,
the
cli
works,
as
expected,
he's
just
asking
for
a
new
way
to
you
know
additional
feature
for
that.
F
A
All
right
next,
I'm
kind
of
going
through
my
answers,
because
it
reminds
me
you
know
what
the
person
asks.
So
I
don't
have
to
read
it
from
beginning,
so
all
right,
so
he
the
person,
is
complaining
that
it's
hard
to
manage
declarative
configuration
and
he's
right.
It's
like
it's
not
easy.
We
have
a
lot
of
settings
and
I
try
to
point
him
that
we're
basically
going
to
we
try
to
improve
it
now
and
in
2.0
we
have
argo
cd
util,
which
can
generate
secrets
for
repositories
and
even
arrogant,
configmap
changes.
A
But
it's
not
yet
it's
not
enough.
It's
not
an
answer
and
I
think
we
pretty
much.
We
will
have
to
keep
working
on
and
the
direction
in
the
next
release,
because
obviously
util
can
have
more
commands
to
change
settings
and
we
can
simplify
the
settings
and
we
have
tickets.
I
kind
of
added
a
few
tickets
to
the
next
release
and
we
can
discuss
it
later
and
this
thing
is
related
to
kind
of
decomposing
keys
in
rbcd
config
map,
to
make
it
easier
to
manage
that
config
map
and
so
on.
C
Yeah
I
mean
on
on
that
note.
I
think
I
have
a
couple
of
enhancement
proposals,
not
an
a
couple
of
issues
with
proposals
on
starting
to
use
strongly
typed
apis
in
general.
What
that
helps
us
with
is.
We
are
coming
up
with
our
own
api
structure
inside
config
maps
and
secrets.
So
it's
not
that
we
don't
have
a
structure,
it's
just
that
there
is
no
validation
on
the
structure
and
we
have
to
build
tools
to
validate
those
structures
that
could
be
cli.
C
Instead,
if
we
had
a
crd
with
the
same
things,
we
get
client-side
validation
for
free,
which
means,
if
somebody
doesn't
put
it
it
just
refuses
to
you,
know,
get
accepted
or
get
get
created.
So
I
think
in
general
you
know
we
have
a
huge
config
map
at
this
point
and
the
best
way
to
know
it
is
to
go
and
read
documentation,
but
if
you
actually
have
them
as
a
crd,
you
get
documentation
for
free,
which
is
the
crt
itself.
There
is
no
reference,
that's
needed,
I
think.
C
In
general
we
have
a
ton
of
apis
that
we
we've
defined
in
loose
ways
in
config
maps
and
secrets.
We
should
probably
consider
moving
them
to
strongly
type
things
just
easier
to
manage
deprecate
over
versions
by
using
the
usual
crd
ways
of
deprecating
things
so
yeah,
I
think
that's
another
approach
we
should
consider.
I
think.
A
All
right,
there
was
one
more
comment,
and
I
didn't
answer
anything
so
pretty
much.
This
person
is
asking
to
support,
storing
applications
in
multiple
namespaces
and
he's
complaining
that
they
use
up
of
apps
pattern
and
they
need
some
way
to
ensure
uniqueness
of
application
names
and
they
want
to
pretty
much.
A
A
I
didn't
know
what
transfer,
because
technically
it's
not
easy
to
do
right
now,
because
rbcd
is
like
heavy
relying
on
storing
applications
in
one
namespace
and
I
and
then,
and
we're
going
to
discuss
the
same
topic.
You
know,
after
going
through
feedback,
wait
I
want
to
move
to
the
next
one,
and
that
was
again
big.
A
Kind
of
a
lot
of
feedback
in
one
comment,
so
one
was
about
sso,
reloading
functionality.
So,
right
now
argo
cd
tries
to
if
it
detects
that
you
sso
user
and
your
token
is
expired.
It
simply
starts
as
a
solo,
login
flow
and
usually
people
don't
even
notice
it.
It
looks
like
it
looks
like
a
page
refresh
and
I
think
he's
saying
it's
annoying
and
I
do
not
know
if
it's
possible
to
avoid
it
because,
like
you,
you
cannot
re
regenerate
the
system
talking
using
sso
without
showing
it
to
the
user.
A
But
yeah-
maybe
maybe
maybe
obviously
you
could
try
to
use
a
fresh
token
kind
of
an
ips,
but
currently
it
doesn't
do
it.
It
just
restarts
so
login
flow
and
that
requires
user
to
go
to
sso
and
typically
sso,
immediately
direct
user
back
to
ui
with
new
token
okay,
another
complaint
about
small
devices
so
basically
sometimes
filter
filtering
filters
are
not
in
the
perfect
state.
If
you
look
at
rcd
ui
from
mobile
or
any
other
small
device,
okay.
A
All
right
next,
this
is
it's
like
a
real.
I
think
it's
a
bug
basically
and
we
can
implement
it,
so
we
sometimes
unnecessarily
use
so
navigation
in
argo.
Cgui
is
done
through
history,
api
and
history.
Okay,
give
you
two
kind
of
methods.
One
is
you
push
something
to
the
history
and
so
user
can
click
back
button
and
go
back
to
the
you
know
previous
state
or
you
can
just
replace
current
like
url
and
it
won't
be
recorded
in
history,
and
so
we
use
push
method.
A
When
someone
clicks
on
the
panel
and
open,
let's
say
a
resource
date,
and
I
think
it's
not
correct-
it's
not
great
behavior
because
yeah
it's
unlike
so,
if
you
click
on
the
panel
and
then
you
click
back
button.
Instead
of
going
back
to
let's
say
list
applications
page,
you
will
just
hide
the
panel
and
I
promise
that
I
will
work
on
fixing
it
because
it
annoys
me
as
well.
So
I
don't
know
if
anyone,
if
you
think
it's
current
behavior
is
good.
A
Yeah
but
and
fix,
is
kind
of
I
mean
it's,
not
simple,
simple,
because
yeah,
that's
the
only
place.
We
have
a
lot
of
places
where
we
open
panel
and
we
just
need
to
go
through
all
of
them
and
change
one
parameter
from
false
to
true
kind
of
to
use,
replace
instead
of
push
right.
A
A
That
comment
kind
of
reminded
me
about
two
issues
that
we
have
already
in
argo,
cd
and
use
case
here-
is
that
if
you
have
a
user
who
has
access
to
the
cluster
admin,
access
to
cluster
and
user
has
credentials
to
call
git
repository
and
users
just
want
to
use
argo
cd
to
manage
cluster
and
deploy
manifest
from
the
tripod
that
user
in
this
case
user,
would
have
to
still
go
to
admin
and
kind
of
ask
him
to
add
a
register
cluster
and
report,
and
that
process
is
not
smooth
because
you
kind
of
have
to
share
your
credentials
with
the
admin
admin
have
to
edit
and
so
he's
asking
to
improve
it,
and
we
already
had
several
conversations
about
it.
A
A
And
I
really
like
that
proposal,
I
added
it
to
2.1,
milestone
and
kind
of
idea
that
we
could
have
two
type
of
repositories
and
projects
shared
and
one
shared,
and
if
it's
not
shared,
it
must
belong
to
one
project,
only
so
and
yeah.
If,
if
user
wants
to
manage
his
cluster
in
his
own
project,
users
should
be
able
to
edit,
and
if
you
want
to
share
that
project
across
multiple,
so
you
have
that
repo
or
cluster
across
multiple
projects,
then
you
need
help
of
admin,
because
there
is
no
other
way.
A
All
right
next,
it's
it's
a
good
kind
of
comment,
but
I
think
too
late.
So
he
he's
just
saying
that
we
have
syncoption
field
and
that
field
is
a
list
of
strings
and
it's
pretty
much
list.
Each
link
is
usually
key
equal
to
value
so
it
hit.
I
think
it
was
we
should
have
made.
You
know,
change
sync
options
and
made
it
just
map
of
convert
it
into
map.
C
A
We
have
that
was,
we
have
annotation,
but
we
also
have
it
was,
I
think
he
was
talking
about.
We
have
two
ways
to
specify
sync
option:
one
was
annotation
on
on
resource
right
and
I
think
it
was
kind
of
incorrectly
to
complain
and
so
to
complain
about
sync
option
in
in
this
first
case,
but
also
sync
option
is
a
field.
It's
a
first
class.
C
C
A
It's
a
hard
decision
anyway.
So
next
it's
a.
We
had
a
pull
request
and
the
pull
request
just
slipped,
and
I
think
the
request
is
to
please
review
and
finish.
You
know
working
on
the
pr
and
merchants-
and
this
is
it's
a
feature
that
led
argosy
users
to
kind
of
split
secrets
across
a
random
set
of
other
secrets.
A
So
you
could,
instead
of
requiring
user
to
just
use
argo
cd,
secret
secret
user,
can
maybe
create
n
secrets
and
just
apply
a
label
and
argo
cd
would
kind
of
try
to
get
all
secret
data
from
all
secrets
which
have
that
label
yeah
and
the
pr
is
kind
of
they're
already
99.
We
just
didn't
complete
it.
A
Right
so
we
going
next,
so
he
asked
for.
He
basically
said
that
it's
kind
of
takes
a
lot
of
cpu
and
memory
to
synchronize
a
lot
of
applications,
so
in
this
case
100
plus
applications
in
pretty
in
2043
repositories,
and
we
did
some
changes
in
2.0
to
make
it
better.
In
particular,
now
user
can
apply
sync
option.
A
I
forgot
the
exact
name,
but
the
thing
option
basically
tells
our
cd
to
sync
only
those
resources
that
have
differences,
and
I
don't
I
ask
him
that
and
I'm
not
sure
if
he
didn't
reply
anything
to
me,
but
it
should
help.
We
recently
saw
a
ticket
from
someone
and
basically
that
person
they
were
running
testing
for
us.
They
set
up.
I
think
50
applications,
each
application
had
400
resources
and
with
that
feature
they
basically
they
told
us
that
it.
It
helps
it
really.
A
Okay,
this
is
a
reminder
for
us
to
to
work
on
proposal
to
restructure
documentation
and
yeah.
So
the
complaint
here
is
that
so
it's
a
it
is
a
complaint
and
the
person
is
saying
that
it's
hard
to
find.
You
know
it's
hard
to
figure
out
how
to
manage.
Rbcd
declaratively-
and
I
guess
answer
is
argo
studio
that
can
help
to
do
it
plus
we
we
want
to
work
on.
You
know
refreshing
our
documentation
structure
and
I
like
fully
agree.
I
think
for
me
this
is,
you
know
priority
of
next.
A
A
All
right
moving
on
so
two
more
and
I
honestly
I
didn't
read
this
too,
and
I'm
going
to
read
it
right
now.
A
C
Yeah,
I
think
we
have
internally
at
red
hat.
We
have
a
lot
of
use
users
of
techtron
as
a
ci
and
argo
as
a
cd.
In
a
way
we
could
potentially
retrofit
techton
there
with
our
workflows
and
the
rest
can
remain
the
same.
C
A
A
A
All
right
so
and
okay
to
me,
I
feel
like
after
reading
this,
you
know
feedback.
C
A
C
A
A
Okay,
next
topic
is:
it
was
one
of
the
complaints
from
from
the
previous
list,
so
we
kind
of
started
discussing
how
we
can
try
to
improve,
self-service
or
pharmacy,
basically
how
we
can
help
end
users,
so
they
can
manage
our
business
and
yeah
and,
to
be
honest,
it
was
my
kind
of
I
started
as
a
pet
project
couple
weeks
ago,
and
I
was
going
to
present
something
during
next
meeting
and
and
I'll
show,
because
I
think
working
on
some
design
document
and
okay
and
just
to
give
a
little
bit
more
context
right
now.
A
Basically,
we
use
multi
use
applications
from
multiple
namespaces
and
what
we
keep
hearing
from
end
user,
that
they
telling
us
that
if
argo
cd
could
read
applications
from
more
namespaces,
then
you
can
use
kind
of
following
trick.
You
can
just
give
a
namespace
access
to
end
user.
An
end
user
then
can
create
applications
that
he
wants
to.
He
or
she
wants
to
manage
in
that
namespace
and
that
kind
of
eliminates
the
need
to
talk
to
admin
and
ask
him
to
create
that
application
in
rbc.
Genuine
space
is.
C
It
yeah
and
and
alex
if
I
could
quickly
add
to
that
a
little.
I
thought
I
think
if
I
had
to
for
a
moment,
let's
say
forget
that
there
is
something
called
argo
cd.
Let's
I
mean
not
forget
that
it
exists,
but
rather
let's
say
I
as
a
user
somebody
gave
me
my
admin
gave
me
an
access
to
a
cluster
and
there
are
there's
just
one
namespace
called
showbig
in
it,
because
my
admin
just
created
a
just
assume
that
there's
only
one
show
big
out.
C
C
I
can
basically
do
a
lot
I
can.
I
can
provide
everything
needed
there
to
kind
of
tell
something
that
hey
here
is
where
my
path
to
my
here's,
the
path
to
my
git
repo
where
manifests
live-
and
I
assume
when
I
create
that
application
cr
it
would
automatically
pull
down
the
manifests
from
git
and
manage
them
now.
C
From
my
perspective,
as
a
user,
my
expectation
would
be,
if
I
and
and
sorry
alex,
I'm
just
not
trying
to
put
words
into
your
mouth,
but
then
I
think
in
general
my
expectation
would
be
that
if
I,
as
a
user,
was
given
a
certain
set
of
permissions
by
my
admin,
I
should
be
able
to
play
within
those
rules,
which
means
I'm
allowed
to
create
pods,
I'm
allowed
to
create
config
maps
and
how
to
create
secrets
in
showbix
namespace
and
those
should
all
work.
Now.
C
If
I
am
trying
to
be
funny
with
my
admin
by
putting
in
a
manifest
there
in
my
gate,
to
be
changed
as
something
in
the
cube
system
name
space,
then
the
application
controller
would
prevent
me
from
doing
that.
Why?
Because
well,
my
admin
never
allowed
me
to
do
that,
and
I
cannot
act
funny
by
trying
to
you
know,
jumping
over
it
by
using
an
application
cr.
So
my
expectation
is,
I
have
a
certain
set
of
permissions.
C
Just
let
me
create
an
application
cr
which
lets
me
play
within
those
rules.
I
think
just
from
a
and
and
this
application
cr
would
be
created
with
like
any
tool.
Like
cube
ctl,
for
example,
I
don't
know
like
I
said
I
do.
I
don't
even
know
argo
city
exists,
so
I
don't,
I
don't
know,
there's
an
august
3
cli.
I
just
know
there's
a
crd
that
I
could
create
as
a
cr
and
just
go
with
it.
So
yeah
alex,
I
hope,
you're
talking
the
same
thing
just
wanted
to
tell
you
yeah.
A
I
I
agree
with
everything
you
said
it's
I
have
the
same.
You
know
understanding
of
the
problem,
and
so
I
can
just
maybe
share
what
I
was
thinking
I
feel
like.
There
are
like
different
set
of
arguably
use
rbo
cd
users.
They
have
different
preferences,
some
actually
like
the
current
way.
They
really
want
to
just
manage
our
ocd,
like
they
won't
let
admin
admins
to
manage
argo.
A
Cd
admins
can
create
projects
set
up
permissions
and
then
users
can
create
applications
only
within
projects,
but
they
pretty
much
have
to
talk
to
admin
first
and
that's
right
and
I
think
it
works
for
some
users
and
later
we
started
hearing
more
requests
like
people
just
saying.
Oh,
I
want
to
use
git
I
like
to
use
git
to
create
applications
and,
and
we
created
upsets,
so
it's
one
opinionated
way
that
help
people
manage
argo
cd
and
I
feel
like
this
request
is
another
opinion.
A
Someone
saying
I
want
to
use
cluster
as
a
source
of
true
and
namespaces,
and
that's
when
I
kind
of
realized
that
I
decided
to
create
a
poc
and
I
have
not
completed
it
yet,
but
idea
was
that
I
was
thinking
to
create
one
more
crd.
It's
like
opinionated
crd,
called
application
source
and
it
could
live
in
target
in
the
managed
cluster
namespace
namespace
that
belongs
to
end
user
and
admin
would
have
to
install
that
crd
and
matching
and
corresponding
controller
into
the
cluster
and
that
controller
kind
of
implements
convention.
A
It's
saying
the
controller
would
look
for
all
application
sources
and
then,
if
it
found
application
source
in
the
namespace,
it
would
create
it
in
a
project
configured
by
admin
and
it
would
infer
destination
from
the
namespace.
Basically
it
I
mean
it
would
create
application
that
manage
resources
in
a
name
space
where
applications
source
crg
was
created,
hope
it
makes
sense.
But-
and
it's
not
it's
like
it's
theory-
cpuc,
it's
not
something
I
would
not
propose
to.
You
know
everyone
to
use
it.
I
just
wanted
to
share
it's
like
an
easier
way
to
express
my.
C
Idea,
yeah
yeah.
I
think
I
like
that
and-
and
I
think
for
a
moment.
What
I
would
just
want
to
do
is
to
let's
assume
that
the
target
cluster
is
the
same
cluster
you're
on
so
let's
say
I'm
on
cluster
a
there's,
an
argo
cd
in
dargo
cd
name
space
in
cluster
a
and
on
that
cluster,
a
I
have
namespace
alex
where
alex
who
lost
his
operator.
C
A
I
was
thinking
that
it's
not
necessary,
it
wouldn't
even
try
to
create
application
and
namespace.
It
would
talk
to
argosy
api
because
aerophytic
would
be
even
in
another
cluster.
So
imagine,
if
you
have
kind
of
add
bones
in
in
dozens
of
clusters
and
one
control
plane
cluster
with
ergocd
and
addons,
the
data
would
be
that
controller
and
basically
controller
knows
how
to
talk
to
argo
cd
and
it
knows
how
to
create
projects
so
applications,
and
then
it
keeps
watching
for
application
sources
in
in
managed
clusters
and
and
then
create
applications.
So
it's
yeah.
C
I
see
so
so
I
think
so.
I
think
what
you're
saying
is
that
there
is
a
way
for
the
controller
of
let's
say:
applications.
Let's
say
the
application
source
controller
knows
a
way
to
synchronously
talk
to
argo
cd
and
let
argo
cd
know
that
hey
somebody
requested
for
a
new
application
in
this
namespace.
Can
you
go
and
handle
it.
A
Yeah
yeah
that's
kind
of
ideal
and
another
flavor
of
the
same
is
you
know
some
people.
A
Much
create
applications
and
they
just
use
jenkins
so
or
like
any
other
ci
tool,
but
basically
that
ci
tool.
You
know
when
it
deploy
that
when
it's
ready
to
deploy
new
version
of
application,
it
just
runs
argo
cd,
cli
and
kind
of
ensure
that
application
exists,
so
it
kind
of
so
it
tries
to
create
application
and
uses
absurd
flag.
So
it
will
either
create
or
update
existing
application,
and
it
looks
to
me
as
the
same
pattern.
A
So
you
kind
of
have
some
automation
running
somewhere
and
user
end
user
drives
that
automation
and
kind
of
and
jenkins
approach
is
like
that
controller.
A
C
Yeah
I
mean,
I
think,
from
my
perspective,
I
would
say
if
you
are
letting
target
clusters
create
this
application.
C
The
current
model
of
you
know
a
target
cluster
wanting
to
create
you
know
an
application
source
or
application.
Cr
kind
of
would
also
imply
that
you
know
we
could
yeah
I
mean
we
are.
We
are
we
ain't
even
have
to
put
some
controllers
there
anyway
for
reconciling
that,
but
then
I
think
what
we
are
saying
is.
We
need
to
effectively
configure
an
agent
on
those
clusters
to
be
able
to
handle
these
new
crs.
I
mean
if
you
had
to
use
a
very
old-fashioned
computer
science
thing.
A
Kind
of
yes,
yes,
so
it
could
be.
Someone
would
have
to
someone
who
has
x,
who
has
admin
access
to
argo
cd
would
have
to
configure
that
controller
and
maybe
store
secret
store.
Argo
cd
token,
in
a
secret
of
that
controller
and
controller,
should
run
in
some
protected
name
space
like
cube
system,
or
you
know
some
some
namespace,
which
is
not
available
to
end
users,
but
that
controller
kind
of
configured
in
a
way.
So
it
would
not.
A
You
know
it
would
not
make
you
it
would
not
let
users
to
shoot
themselves
in
the
food
and
so
controller
would
maybe
it
would
allow
users
to
create
applications
only
in
some
projects
and
and
okay
and
now
it's
kind
of
I
didn't
really
think
carefully
about
security
aspect
of
it,
and
I
I
kind
of
I
feel
like
there
is
a
lot
of
other
things
can
be
added
to
that
controller
controller
can
create
projects
on
the
fly
as
well.
A
C
Yeah
yeah,
I
think
the
way
I
see
it
is
if
a
user
is
creating
an
application
source
cr
in
that
name
space.
Can
we
avoid
the
round
trip?
I
mean
again,
I'm
not
considering
where
we
have
the
whole.
You
know,
fleet
of
clusters
set
up
with
argo
city.
Let's
say
it's
not
I
mean
if
I'm
creating
a
cr
in
a
cluster,
that's
an
application
source
cr
and
that's
it.
C
I
don't
I
mean:
does
the
user
now
want
the
whole
ui
thingy
as
well,
or
does
the
user
want
the
application
cr
to
just
associate
to
just
pull
things
from
gate
and
apply
them?
I
mean
in
that
case
the
way
I
see
it
is,
you
know,
do
we
need
the
round
trip
at
all,
or
do
we
actually
resolve
the
whole
problem
for
the
user?
On
the
same
cluster
itself,
you
could
just
visualize
from
a
different
cluster.
A
C
I
mean
yeah
kind
of
it.
I
mean
the.
I
mean
you,
it's
like
cluster
b.
You
basically
talk
back
to
cluster
hub,
let's
say
where
argosy
is
and
then
argo
cd
again
communicates
back
to
this
cluster
right.
So
you
have
basically
a
round
trip
that
happens.
Okay,
where
we
say
hey
hub
somebody
requested
for
this
and
then
argo
cd
responds
to
that
I
mean
I
and
I
know
you're
basically
trying
to
fit
into
the
current
argo
cd
model
here
and-
and
I
think
that
totally
makes
sense.
A
Yes,
let's
we
have
even
I
think
what
you're
describing
is.
I
would
call
it
github's
agent,
it's
like
a
simplified
version
of
argo
cd,
correct.
G
A
We're
kind
of
hoping
to
build
one
day
and
I
yeah
I'm
a
big
fan
of
that
idea.
We
just
never
get
time
to
to
work
with
it.
So,
let's
see
this
is
another
option.
It's
you
know.
C
Right
I
mean
I,
I
know
why
you
mention
projects,
because
we
need
projects
because
projects
help
us
figure
out
the
our
back.
For
you
know
that
thing
but
but,
like
I
said,
I
think,
like
I
said
in
in
the
in
the
in
the
initial
point
I
made
is
that
the
user,
who
created
that
whole
application
source
cr
already
had
a
set
of
permissions
to
his
name.
The
admin
had
given
her
right.
C
So
if
the
get
ops
agent,
the
fictitious
githubs
agent
could
effectively
make
use
of
the
same
permissions
to
determine
if
the
user
is
allowed
to
create
those
resources
or
not.
I
know
we
we're
not
doing
that
because
we're
trying
to
fit
it
in
the
current
model,
but
then,
like
I
said,
do
we
want
to
go
to
a
model
where
we're
saying
we?
C
If
a
user
in
a
cluster
like
if
a
cluster
user
and
of
course
not
everyone,
may
be
tested
users,
if
a
cluster
user
creates
an
application
or
application
source
cr,
we
just
use
the
permissions
of
the
cluster
user.
That's
it
if
the
user
is
not
a
cluster
user,
because
argo
siri
has
a
fantastic
concept
of
being
able
to
sit
above
clusters
and
users
could
only
live
in
argo
cd
through
a
through
a
model
where,
let's
say
from
active
directory,
etc.
In
that
case,
yes,
we
don't
have
a
cluster
user.
C
A
A
A
And
that's
why
they
don't
use
rbcd,
they
don't
really
like
it,
and
I,
and
that
was
that
was
inspiration
for
you
know
to
at
least
document
desire
to
build
github's
agent,
and
I
agree
we
have
kind
of
technology
to
do
it.
We
just
need
to
do
it
one
day
and
I
even
spoke
with
couple
such
users
like
someone
discovered
an
example
of
github's
agent.
It's
that
example
just
sitting
in
github's
engine
repository,
and
he
asked
me
like.
Okay,
we're
never
going
to
do
it.
C
Honestly,
like
I
mean
yeah,
I
mean
I'm,
I'm
ready
to
lead
an
effort
on
this.
To
be
honest,
because
I'm
really
interested
on
this
and
I
can
work
with
you
alex
and
come
up
with
a
proposal
so
that
it
addresses
three
things.
One
is
the
single
cluster
usage,
multi-cluster
usage
and
cluster
user
non-cluster
usage,
because
I
think
they
are
both
like
sorry,
cluster
user
and
non-cluster
user
concepts,
because
I
think
those
are
the
three
areas
that
we
need
to
address.
A
C
Kind
of
because
it's
yeah
the
ui
is
the
probably
the
simplest
thing.
I
think
I
think
I
think
to
be
fair.
Like
I
said
right
I
mean
the
way
I
see
it
is
the
ui
doesn't
even
have
to
catch
up
for
the
next
six
months.
If,
even
if
you
do
this,
I
mean
the
the
idea
should
be.
You
know,
if
I
I
mean,
especially
because
a
lot
of
users
in
many
organizations
never
have
access
to
any
ui
on
the
cluster.
So
I
think
regina.
C
If
you're
working
on
the
openshift
console,
there's
the
bad
news,
a
lot
of
companies
don't
even
give
people
any
access
to
any
console.
All
they
have
access
is
to
the
api
server
and
the
user.
Right,
let's
say
that's
all
I
have.
I
mean
if
we
can
at
least
address
that
use
case,
where
I
just
have
a
command
line
access
to
the
cluster.
In
a
couple
of
namespaces
and
like
I
said,
we
have
more
technology
than
we
need
today
in
argos
city
to
be
able
to
get
that
working.
A
Yeah,
I
I
basically
agree,
and
I
think
if,
if
we
have,
you
know
desire
to
work
on
it
and
if
someone
wants
to
leave
it,
if
you
want
to
delete
it,
then
it
would
be
awesome.
It's
like
elsewhere
get
good
time
to
do
it.
I
know
I
won't
say
why.
C
Sure
I
think
that
sounds
good
to
me.
I
think
so.
I
think
as
a
next
step,
I
think,
do
you
want
me
to
start
a
document
or
at
least
list
the
use
cases
we
just
discussed
today
here?
Yes,
and
and
we
need
to
ensure
that
whatever
solution
we
come
up
with,
it
should
address
all
use
cases
all
personas.
At
least
it
should
not
lean
towards
one
person
or
more
than
the
other.
A
C
E
A
A
I
like
just
to
share
some
numbers,
so
we
kind
of
we
were
working
for
three
months
and
we
released
more
than
close
to
400
commits,
so
that
means
kind
of
like
three
four
commits
every
day,
including
weekends.
It's
amazing
a
lot
of
work
and
a
lot
of
good
changes,
and
it's
not
counting
three
more
projects
like
we
have
application,
sets
notifications
and
image
updater
so
that
release
is
kind
of.
I
think
it
deserves
to
be
2.0.
It's
really
mature.
C
Yep
and
thank
you
very
much
alex
and
you
know
all
the
other
maintainers
for
even
the
last
minute
changes
you
folks
had
to
do
to
get
things
to
work
and
please
feel
free
to.
Let
us
know
if
how
others
could
contribute
there,
because
that's
a
lot
of
work
that
went
in
in
the
I
think
last
few
days
to
one
get
the
packages
updated,
ci
in
good
shape,
etc.
C
A
You
want
to
talk
about
application
certificates.
G
Yep,
can
you
hear
me?
Okay,
yes,
awesome
yeah,
so
I
just
thought:
I'd
talk
a
bit
about
the
first
release
of
application
sets
that
came
out
yesterday,
so
it's
awesome
to
be
able
to
announce
that
application
set
controller
is
now
available
v010.
G
I
believe
I've
talked
about
it
on
this
call
before,
but
the
application
set
controller
makes
it
easy
to
target
multiple
clusters
to
support
monorepos
using
argo
city
or
to
allow
applications
to
be
created
by
unprivileged
users.
G
G
If
you
find
you're
spending
too
much
time
or
more
time
than
you'd
like
in
supporting
mono
repos,
if
you
find
that
you're
uncomfortable
with
the
security
exposure
of
allowing
tenants
of
a
multi-tenant
cluster
to
create
applications,
if
you're
a
cluster
operator
deploying
cluster
add-ons
or
if
you'd
just
like
to
to
explore
the
set
of
tools
that
application
sets
now
offers,
I
definitely
recommend
that
folks
check
out
the
application
set
controller
in
our
first
release,
and
let
us
know
what
you
think,
so
you
can
find
out
more
information.
G
You
know
what
better
place
than
the
blog
post
that
is
on
the
screen
here.
You
can
find
that
on
blog.goproj.io,
which
is
like
the
standard
argo
project
blog
space,
you
can
also
find
a
link
to
that
on
the
argo
project.
Twitter
account
at
argo,
proj
and
there's
also
a
link
in
the
meeting
notes.
G
G
Over
the
past
year,
the
work
began
on
application
sets
in
early
2020,
and
I
personally
didn't
come
on
until
late
2020,
I
was
able
to
fill
in
a
lot
of
the
project
structure
stuff,
like
ede
tests,
unit
tests,
release,
process,
bug
fixing
stuff
like
that,
but
the
actual
hard
work
of
that
initial
design
and
the
coding
of
all
the
great
features
that
went
into
into
the
release
things
like
the
list
generator
the
cluster
generator,
the
git
generators,
the
crd,
the
controller
scaffolding,
all
that
came
from
the
community,
all
that
came
from
the
fantastic
group
of
contributors
that
you
see
on
the
screen
here.
G
So
I
just
wanted
to
shout
out
ulmer
and
devin
and
michael
and
alex
jonathan,
not
sure
who
that
is:
mateo
jesty,
shan
liu,
william
rat,
in
the
deep
john
shubik,
alex
and
omer
for
all
the
contributions
we
would
not
have
been
able
to
be
sitting
here
today,
talking
about
the
application
set
controller
and
all
the
cool
things
that
that
it
does
without
the
folks
in
the
list
there.
So
thank
you
all
very
much.
In
the
blog
post,
I
included
a
link
to
their
github
page.
G
So
if
you
want
to
learn
more
about
them,
see
what
they're
all
about
definitely
check
them
out,
but
yeah
thanks
for
all
your
hard
work,
it
was
awesome
for
me
personally
to
be
able
to
join
you
all
on
this
journey
and
get
this
release
out
the
door
and,
finally,
a
special
thanks
to
the
argo
cd
community,
just
as
a
whole.
It's
it's
a
great
community
and
I've.