►
From YouTube: Keptn Community & Developer Meeting - Dec 16, 2021
Description
Meeting notes: https://docs.google.com/document/d/1y7a6uaN8fwFJ7IRnvtxSfgz-OGFq6u7bKN6F7NDxKPg/edit
Learn more: https://keptn.sh
Get started with tutorials: https://tutorials.keptn.sh
Join us in Slack: https://slack.keptn.sh
Star us on Github: https://github.com/keptn/keptn
Follow us on Twitter: https://twitter.com/keptnProject
Sign up to our newsletter: https://bit.ly/KeptnNews
A
Welcome
to
the
captain
community
meeting
today
is
december
16th
and
today
we
do
our
first
attempt
to
host
the
meeting
in
beverly,
so
it's
basically
community
cfio
portal
and
if
you
haven't,
joined
yet
consider
joining
because
we
intend
to
host
all
future
meetings
there.
B
C
Yeah,
I
can
give
you
a
few
words
about
that.
Since
christmas
is
around
the
corner
for
many
countries,
then
we
thought
about
giving
you
a
small
present
pre-release
of
captain
012,
which
the
highlights
are
mainly
too
for
the
core
side.
We
are
working
towards
have
a
better
support
for
zero
downtime
upgrades
and
higher
ability
and
for
the
bridge.
The
major
highlight
is
that
now
it's
possible
to
pick
data
from
an
event
via
the
ui
like
it
was
possible
for
the
secret.
C
A
C
Yes,
good
question:
you
shouldn't
do
the
normal
approach,
because
the
captain
install
or
upgrade
commands
is
not
supported.
You
should
install
captain
via
helm
at
the
bottom
of
the
quick
start
guide.
There
is
a
description
how
you
can
upgrade
or
install
captain
via
helm
directly,
and
you
should
follow
that
this.
A
C
A
Thanks
for
that,
I'll
probably
create
an
action
item
for
myself
to
document
upgrade
flow
somewhere
for
experimental
releases.
A
Yeah
and
thanks
for
that,
any
other
comments
regarding
this
holiday
present.
C
Yes,
you're
right,
we
released
it
last
week
the
major
improvements
there
are
some
fix,
especially
for
openshift,
where
we
were
using
too
many
permissions
since
openshift
most
restrict
is
more
restrictive.
We
change
a
bit
the
configuration
service.
A
D
D
Can
you
see
my
screen
now
yep?
Yes,
we
can
okay
yep!
You
see
that
there
are
cap
enhancement,
proposals
in
progress,
40.12
and
yeah.
Those
and
a
couple
of
ones
were
finished
for
o.11
and
the
bigger
items
there
are
robustness,
quality
and
scalability
improvements
and,
as
giovanni
mentioned
before,
it's
all
about
also
providing
capabilities
for
zero,
downtime
upgrades
and
high
availability
of
captain.
D
This
will
definitely
also
be
part
of
upcoming
releases
as
yeah
still,
some
work
is
needed
to
be
done
here,
but
what
I
would
like
to
also
mention
right
now
is
that
we
kicked
off
work
in
respect
to
our
role-based
access
control,
and
this
is
here
formulated
in
cap
60..
D
Please
take
a
look
at
this
cap
also
to
follow
the
discussions
there
or
the
ideas,
and
I
would
like
to
tell
you
yeah,
please
feel
free
to
read
through
the
cap
by
your
own,
but
I
would
like
to
show
you
that
the
first
chunk
of
tickets
and
issues
that
are
tackled
are
those
that
support
are
used
or
are
needed
for
supporting
oi
dc,
which
is
an
oauth.2
implementation
and
yeah.
D
Open
id
connect
is
well
an
additional
layer
on
top
of
oauth
2.0
and
it
allows
you
to
identify
the
end
user
based
on
the
authentication
server,
and
it
also
then
gives
back
the
information
which
permissions
this
user
has
and
at
the
end
we
will
need
these
permissions
in
order
to
get
access
to
the
api,
and
this
picture
should
also
give
you
a
high
level
overview
of
what
we
are
focusing
on
right.
Now.
D
It's
really
implementing
this
oidc
flow,
which
is
the
connection
between
an
oitc
provider
and
components
that
are
then
talking
to
the
captain.
Api
and
those
components,
are
the
cli
the
bridge,
as
well
as
the
distributor.
D
Yeah
issues
and
and
epics
are
then
linked
here,
and
I
took
a
look
at
the
notes
and
saw
that
we
already
see
an
implementation
of
an
issue
today
and
yeah
in
the
next
upcoming
release.
Meetings.
More
will
then
be
talked
about
this
topic
as
we
make
progress
in
this
direction.
A
D
Right,
true
and
yeah,
for
example,
then
it's
possible
to
to
connect
a
key
clock.
Github
has.
E
D
A
A
So
while
we
are
talking
about
the
roadmap,
quick
updates,
which
changed
since
we
had
a
discussion
two
weeks
ago,
so
as
we
discussed,
I
started
putting
community
related
items
to
captain
roadmap
so
how
it
works.
Basically,
I
use,
I
always
use
community
label,
and
if
there
is
some
confusion,
please
let
me
know
so
we
can
think
how
to
represent
it
better,
but
for
now,
in
addition
to
technical
roadmap,
there
are
also
some
items
related
to
the
community
and
most
of
these
items
reside
in
the
old
captain
community
repository.
A
Okay,
so
back
to
the
more
technical
topic,
I
actually
have
a
suggestion
to
drop
a
dco
entirely
and
probably
to
replace
it
with
easy
cla.
It's
a
new
service
provided
by
the
linux
foundation
for
mambo
projects.
A
So
I
started
voting
in
a
stack
and
if
you
have
opinion,
please
comment
because
it
would
be
nice
and
for
me
the
main
use
case
for
that
is
to
allow
easy
editing
from
the
web
interface
so
from
elsewhere,
so
that
you
don't
have
to
care
about
dco.
You
don't
have
to
force
push.
If
you
somehow
make
a
mistake,
and
instead
of
that,
we
can
use
a
ecla,
which
is
also
quite
handy,
so
it's
basically
the
same
as
other
checks.
A
Just
that's
so
it's
one
of
my
examples
from
previous
experiments
and
here
how
it
happens.
A
You
just
have
a
github
up
which
adds
additional
branch
protection
and
also,
if
the
check
doesn't
pass,
it
provides
command,
pre
and
guidelines
how
to
register
and
once
it's
you're
registered,
everything
is
done
and
you
need
to
do
it
on
the
ones
for
multiple
github
organizations.
A
A
A
Yes,
that's
my
approach
too,
so
I
reached
out
to
the
cncf
to
get
confirmation
that
they're
finding
these
projects
using
easy
cla,
I'm
pretty
sure
they
are
fine,
because
there
are
projects
using
that
in
the
scenes
here.
But
once
we
have
official
sign
off,
if
no
objections,
I
will
set
up
a
proof
of
concept
in
a
few
repositories
and
then
we
can
expand
it.
A
And
yeah
another
things
related
to
the
contributing
flow,
so
you
may
have
seen
that
there
were
some
updates
in
tutorials,
but
also
they
will
update
some
contributor
guidelines.
Basically,
I
started
recommending
non-code
contributions,
including
documentation,
just
advocacy
and
outreach
to
for
extent.
The
channel
will
be
created
basically
the
same
for
internship
and
mentorships.
A
So
if
you
see
any
other
kinds
of
contributions
which
would
be
helpful
for
the
project
like,
for
example,
design,
etc,
again,
please
submit
a
ticket
for
me,
so
I
will
try
to
get
it
added
and
I
will
continue
working
so
that
we
can
aggregate
all
the
contributing
guidelines,
because
sometimes
it's
not
trivial
to
find
them
and
yeah.
Hopefully
we
could
extend
code
contributions
too.
A
We
have
a
lot
of
things,
for
example,
for
captain
core,
but
for
services,
etc.
Documentation
is
really
scattered,
so
maybe
we
could
aggregate
that
and
update
the
page
a
bit
more,
so
any
feedback
from
developers
would
be
appreciated
so
that
I
can
keep
working
on
this
documentation
before
I
try
developing
something
on
my
own.
B
Okay,
so
yeah
I've
been
working
on
a
lot
of
kept
concrete
and
captain
sandbox
stuff
and
the
first
one
I
want
to
show
you
is
an
extension
of
the
github
automation
repository
that
we
have
that
moritz
has
started
so
moritz
has
started
out
like
with
dco
workflow
in
here,
and
we've
now
extended
it
with
a
validate
semantic
pr,
workflow,
a
pre-release
integration,
a
release,
integration,
workflow
and
a
docker
build
action,
and
if
you're
wondering
what
all
of
this
means,
this
is
basically
github
action,
stuff
and
not
just
stuff
but
reusable
stuff.
B
B
B
The
ci
workflow
is
just
triggered
when
we
push
something
to
the
default
branch
or
when
we
create
a
pull
request
to
the
default
branch,
and
it
consists
of
a
prepare
step
that
has
always
been
there.
Don't
worry
too
much
about
it,
but
where
it
gets,
cool
now
is
unit
tests,
so
unit
tests
is
something
you
need
at
three
four
five
different
places,
so
you
need
it
in
your
ci.
You
need
it.
When
you
do
a
pre-release,
you
need,
we
need
to
release,
you
could
even
have
a
crown
job
running.
B
I
don't
know
every
now
and
then
to
do
it,
but
there
is
many
places
where
you
need
it,
and
here
is
the
first
cool
automation
or
the
first
cool
enhancement
that
we
have.
We
can
reuse
action
from
the
same
repository,
so
here
we
have
in
the
same
repository
the
unit
tests,
action
and
this
one
actually
contains
the
code.
B
How
do
we
execute
unit
tests,
and
this
is
the
one
that
each
repository
will
probably
have
to
customize
right,
but
this
is
one
one
line
or
maybe
two
steps
right
that
you
need
to
customize
depending
on
what
you're,
using
if
you're
using
go
or
something
else
and
in
the
ci
you
just
say,
unit
tests,
don't
care
what
it
is
just
run
them
and
the
second
cool
enhancement
here
is
than
the
docker
build.
So
you
probably
have
seen
github
actions
where
you
need
to
do
the
login
and
the
login
depends
on
certain
parameters
like?
B
B
Should
we
push
a
pre-release
version,
whatever
all
of
that
is
now
essentially
hidden
in
this
captain
github
automation,
action
which
we
released
today
actually,
and
the
only
thing
you
need
to
tell
this
action
is
how
should
my
docker
image
be
called
like
in
this
case
it's
in
a
docker
organization
that
has
a
name
and
what
version
does
it
have
or
what
tag
does
it
have
and
some
parameters
like
the
registry,
user
and
registry
password-
and
this
even
goes
as
far
as
what,
if
you
need
to
build
multiple
containers
like
job
executor,
has
two
containers.
B
Fine,
then
just
use
this
action
again
and
use
a
different
image
name
and
use
a
different
docker
file.
So
it
could
even
be
something
that
long
term
we
will
see
in
captain
core.
I
will
have
to
sync
with
moritz
with
that
regarding
that
matter,
because
there
are
some
limitations,
obviously
that
captain
might
have
to
well
bypass
somehow,
but
I
think
this
is
a
good
way
going
forward,
and
last
but
not
least,
we
have
the
pre-release
and
the
release
action.
Sorry
workflow!
B
So
here
we
do
a
docker
build
again,
but
then
we
also
do
a
helm
chart
build
and
by
the
way
helm
chart
means
you
get
a
file
at
the
end,
and
you
want
to
add
this
file
to
release.
So
you
can
use
the
github
cli
to
do
that,
like
github
release,
upload
the
release
tag
and
the
file,
so
essentially
we're
trying
to
make
really
really
small
building
blocks
that
are
super
easily
reusable,
and
I
think
so
far
this
is
looking
quite
promising
all
right.
Any
questions
regarding
this.
B
Cool,
I
have
the
captain
contrib
job
executor
service,
pull
request
linked
here.
So
if
you're
interested
in
the
details,
you
can
take
a
look
at
this
pull
request.
Then
I
have
made
several
bug
fixes
for
prometheus
service.
I've
linked
each
of
them
here,
I'm
not
going
to
go
through
all
of
them.
B
It's
it's
way
too
much
detail
in
them,
but
the
short
version
is
that
the
remediation
features
of
captain
with
prometheus
service
haven't
been
working
properly
for
quite
some
time
and
that
querying
certain
metrics
also
did
not
work
properly,
and
a
couple
of
smaller
bug
fixes
are
also
included.
There.
A
Should
we
do
a
retrospective
at
one
of
the
next
meetings,
just
to
see
what's
missing,
so
that
we
can
ensure
that
it's
being
tested,
at
least
for
major
releases,
because
my
guess
said
that
we
could
have
a
better
test
automation
for
that?
Maybe
some
checklist
for
tutorials.
I
don't
know,
but
it
would
be
nice
that
we
prevent
such
issues
in
the
future,
because
okay
parameters
is
one
of
important
use
cases
for
captain
and
great
to
ensure
that
it's
working.
B
I
don't
know
what
to
say
you're
right
yeah,
but
this
is
not
like
five
minute
task.
This
is
like.
B
A
B
B
Okay,
last
but
not
least,
with
all
this
automation
that
we've
shown
before
we
are
also
getting
a
cool
building
block,
which
is
called
automated
building
of
helm
charts,
and
this
now
means
that
the
prometheus
service
can
also
be
installed
using
a
helm
chart.
I've
laid
out
the
command
for
one
of
the
pre-releases.
Here,
it's
basically
the
same
as
some
people
already
have
seen
with
the
dyna
trust
service
and,
obviously
with
captain.
B
So
this
is
really
just
a
long
overdue.
Let's
say
feature
that
we
now
have
to
install
it
using
helm
and
it
will,
in
the
future,
help
us
to
maintain
the
installation
of
prometheus
service
on
a
cluster
much
easier
than
before,
because
no
more
cube
ctl
apply,
and
with
that
I
already
spoiled
something
we
have
a
pre-release
of
all
those
bug
fixes
for
prometheus
available,
but
I'm
still
working
on
verifying
if
the
release
is
working
as
intended.
So
please
allow
some
time
for
the
final
release
of
prometheus
service.
A
B
You're
welcome
and
with
that
I'll
hand
it
over
to
giovanni.
C
You're
still,
muted,
yes,
sorry
classical
mistake,
so
I
wanted
to
tell
the
community
that
we
are
working
towards
a
definition
of
what
is
a
breaking
change
for
captain
when
it
comes
to
rest
endpoints,
and
for
that
I
create
the
following
pr,
and
I
would
like
the
community
to
have
a
look,
have
a
review
and
provide
some
feedback
on
this.
But
the
general
idea
is
that
every
time
we
touch
an
endpoint
which
is
publicly
available,
so
it's
not
mark
as
internal
every
time.
C
We
change
that
end
point
in
terms
of
url,
adding
some
mandatory
fields
changing
the
data
model,
then
this
is
considered
a
breaking
change
and
also
every
time
we
change
our
cloud
event,
definition
for
event.
So
for
endpoints
that
support
cloud
events,
then
this
is
also
a
breaking
change,
so
bottom
line.
Please
have
a
look,
provide
some
feedback
and
finally,
we
have
an
agreement
of
what
is
a
breaking
change
in
terms
of
rest,
api.
F
G
So
where's
the
document
here
it
is
okay.
Today
I
want
to
discuss
three
items,
so
the
first
one
was
a
research
ticket
that
I've
worked
on
for
the
configuration
service.
G
Another
big
issue
is
that
this
service
is
not
really
suited
yet
for
having
multiple
replicas
running
at
the
same
time,
because
the
reason
for
that
is
that
it
basically
has
a
persistent
volume
mounted
to
it,
where
it
stores
the
the
repositories.
G
G
Actually,
the
one
I've
worked
on
was
first
how
to
enable
having
multiple
replicas
of
the
configuration
service
running
at
the
same
time.
So
to
do
that,
we
are
about
to
make
the
git
upstream
repository
mandatory.
G
I
think
that
was
mentioned
already
in
one
of
the
past
community
meetings
and
by
having
this
single
source
of
truth
in
the
upstream
repository.
We
can
now
have
multiple
replicas
of
the
configuration
service
running.
So
actually
we
don't
even
need
a
persistent
volume
anymore,
but
we
can
use
empty
deer
storage
within
each
of
the
replicas
of
the
configuration
service
and
basically
when
it
tries
to
fetch
data
for
a
project
it
will
try
to
clone
from
the
upstream
and
if
it's
there,
it
can
then
take
off
from
that
latest
state
and
yeah.
E
G
Implementation
detail
of
the
old
configuration
service
was
that
it
was
using
the
git
cli
within
the
container,
which
is
also
not
really
that
ideal
because
for
error
handling,
we
had
to
parse
the
cli
output
of
git
etc,
and
therefore
we
also
tried
to
look
for
alternatives.
So
for
this
first
proof
of
concept,
we
were
using
the
goget
library,
which
was,
for
the
most
part,
working
quite
well,
except
one
limitation,
because
the
pull
methods
provided
by
this
library
cannot
actually
handle
fast
forward.
G
G
G
So
basically,
as
giovanni
mentioned
in
11.3,
we
had
to
do
some
changes
to
the
configuration
of
the
configuration
service,
which
also
involved
using
a
different
file
system
group,
because
the
one
we
previously
were
using
was
not
supported
by
default
in
certain
openshift
environments,
especially,
but
that,
in
turn
led
to
the
problem
that
the
projects
that
are
created
with
11.3
have
a
different
file
system
group
than
the
ones
with
with
with
11.2
or
all
versions
smaller
than
11.3,
and
that
led
to
some
incompatibilities
and.
E
G
G
Definitely
we
got
some
pain
from
it,
because
it
was
indeed
causing
problems
when
someone
tried
to
install
it
actually
on
openshift
and
I've
looked
at
actually
did
previously.
So
basically,
it
was
needed
for
supporting
the
blue
green
deployment
strategy
for
the
continuous
use
case.
G
G
H
H
Okay,
so
I
worked
on
some
smaller
bridge
issues
and
on
mockups
for
the
last
two
weeks,
so,
first
of
all,
I
built
in
a
validation
for
special
characters
for
payload
because
they
cannot
be
processed
with
the
weapons
service,
so
I
mean
they
are
now
yeah,
elevated
and
the
form
is
invalid.
If
you
enter
some
of
them,
so
you
can
try
it
out.
It's
like.
H
I
don't
know
for
something
again,
one
two,
three
four
and
this
another
jokes
that
this
is
forbidden
and
it
has
it's
not
allowed
and
we
have
to
remove
it
also.
If
we
try
to
fill
out
the
rest
of
the
form.
H
Now
we
can
see
it's
still
disabled
after
we
move
it,
and
everything
is
fine
again
so
yeah
just
to
prevent
some
weird
errors
on
on
the
website
good
then
I
also
increased.
So
this
is
more
of
like
a
quality
of
life.
Improvement
also
increased
the
field
size
or
I
actually,
I
made
it
the
text
area
for
the
ul
field.
H
So
if
we
put
in
a
really
long
url,
for
example,
that
one
you
can
see,
I
can
now
expand
it
to
have
the
full
full
picture
instead
of
what
what
it
was
again
before
that
it
was
really
cut
and
not
really
editable
in
with
this
issue
also
get
a
bit
of
rework
of
the
whole
of
the
whole
form.
So
it's
now
looks
a
bit
more
cleaned
up
than
the
line
and
stuff.
H
H
H
So
if
I
go
here
with
this
one.
H
Do
this
again,
it's
more
prominent,
so
it's
says
that
only
scope
captain
service
can
be
yeah
can
be
used
basically
and
how
to
configure
these
sort
of
secrets.
Basically,
so
the
bigger
issue
worked
on
was
mock-up
for
trading
in
seconds
from
the
bridge,
so
it
is
basically
that
the
outcome
for
this
bigger
mock-up,
so
I
did
this
in
figma
and
we'll
just
go
through
really
quick.
So
the
idea
is
to
trigger
sequence
directly
from
the
environment
screen.
H
Also,
if
you
go
to
the
sequences,
you
can
also
click
it
here
and
we'll
be
redirected
to
the
environment
screen.
H
So,
depending
on
what
you
choose,
you
get
the
proper
fields
that
are
necessary
to
configure
yeah
the
the
sequence
that
you
want
to
start,
for
example,
for
evaluation
you
can
use
timeframes,
start
end
time
and
so
on,
and
if
you
trigger
the
sequence,
you
will
land
here
and
we'll
see
the
sequence
that
has
been
triggered,
so
the
mock-up
can
be
reviewed
on
this
issue.
So
if
you
go
to
captain
captain
issues,
4507.
H
There
is,
there
are
two
variants
linked
here.
I
won't
spoil
anything,
so
I
would
ask
also
the
community
if
you
haven't,
voted
yet
take
a
look
at
the
mock-ups
and
give
you
a
vote
here,
and
I
would
really
appreciate
that
because
it's
a
short
ap
test
and
I'm
more
sure
about
what
what
the
right
choice
would
be
in
with
this
issue.
I
also
did
something
a
bit
more.
H
I
basically
built
a
small
sigma
component
library
for
captain
for
reusability
issues.
So,
as
you
can
see
here,
I
added
a
lot
of
components
to
figma.
So
figma
is,
I
don't
know
if
you're
familiar
with
it,
and
this
is
a
prototyping
and
walk
up
to
basically
and
yeah.
H
H
H
I
Yep
so,
which
screen
this
one.
I
Okay,
so
for
bridge,
we
now
have
a
feature
for
open
id
connect,
so
you
can
now
enter
a
provider
like
google
or
yahoo,
or
something
like
that
and
use
that
for
identification.
The
bridge
there
are
some
additional
environment
variables,
for
example,
just
to
enable
the
the
sign
on
the
open
id
then
the
base
url.
I
That
is
then
used
for
the
redirect
the
discovery
like
the
one,
the
one
from
the
identity
provider,
the
client
id
that
is
also
provided
by
the
identity
provider
and
the
secret,
because
you
have
to
create
the
application
to
identify
to
the
to
the
provider
you're
using
and
then
also
the
algorithm,
because
it
doesn't
need
some
validation
which
algorithm
the
id
token
that
is
returned
should
be,
and
also
a
name
property.
I
Changed
or
we
in
the
bridge
we
just
used
the
name,
the
nickname
or
email.
So
if
the
one
is
not
available,
the
next
one
is
ticking
yeah.
I
can
show
a
quick
example
if
you
access
now
the
bridge
with
this
one,
okay,
yeah,
I'm
immediately
redirected
to
yahoo.
I
And
then
I'm
redirected
to
the
bridge
again
and
yeah,
I'm
logged
in
with
my
user
yeah
and
if
I
also
log
out
this
could
be
changed.
I
guess
because
the
session
is
cleared
but
then
because
you
are
at
the
main
screen,
you
are
then
already
redirected
to
your
provider,
the
sso
provider
and
because
you
are
already
looked
in,
there
is
immediate
redirect
to
the
bridge
and
you
are
logged
in
again.
E
Actually,
I
have
very
tiny
things
because,
as
florian
mentioned,
we
were
working
on
figuring
out
the
configuration
service.
One
thing
that
has
been
done
is
that
we
had
a
mistake
in
the
lighthouse
when
the
sli
provider
was
failing.
We
were
not
failing.
This
is
now
being
fixed.
So
if
we
receive
the
fail
result,
the
whole
things
will
be
filled
and
the
sequence
will
not
proceed,
and
then
a
little
feature
is
nicking.
E
Is
that
when
there
are
some
slo
in
the
file
which
are
actually
not
a
specific
slide
that
are
not
specified
in
slo,
we
show
them
as
a
message
in
the
evaluation
so
that
if
the
user
doesn't
realize
that
there
is
something
wrong,
he
will
see
it
here
now
then.
Another
tiny
thing
is
that
adding
the
problem
title
to
remediation
sequence
for
the
sequence
endpoint.
E
I
think
this
will
be
more
visible
and
understandable
why
it
is
useful
in
the
bridge
side
and
finally,
another
tiny
thing
which
I
don't
even
have
a
cluster
on
to
show
is
now
we
are
returning
the
value
used
for
the
comparison
and
in
the
lighthouse,
instead
of
recalculating
it
in
the
bridge,
and
this
will
be
used
in
the
bridge
for
more
accurate
explanation
of
what
happened
in
the
background
to
the
lighthouse
service.
A
Then
thanks
everyone
have
a
nice
christmas,
I'm
not
sure
I'll
be
doing
the
meeting
in
two
weeks
on
december
30th
or
in
one
week
on
december,
23rd.
D
That's
not
a
good
question.
Are
we
going
to
host
at
the
meeting
next
next
thursday.
D
D
This
means
next
week
we
we
having
the
meeting
and
then
the
next
one
will
be
13th
of
january.
A
And
speaking
of
meetings
on
21st
at
4
pm
utc,
we
will
have
capital
user
group,
they
will
have
kirotec
presenting
their
use
cases
and
the
experiences
with
captain.
I
guess
there
will
be
a
lot
of
discussions
about
parameters,
his
keys,
because
this
is
the
main
objective
at
the
moment.