►
From YouTube: wasmCloud Community Meeting - 22 Feb 2023
Description
Welcome to the wasmCloud community! Tune in live where we discuss the latest developments in the wasmCloud ecosystem, WebAssembly standards, and break out sweet demos.
Agendas for wasmCloud community meetings can be found at: https://wasmcloud.com/community
A
I
think
I
think
it
worked
yeah.
It
worked
all
right,
hello.
Everyone
welcome
to
the
wasm
cloud
community
meeting
Wednesday
February
22nd
today
we
let
me
go
ahead
and
pull
up
the
agenda
really
quick.
We
actually
don't
have
a
demo
to
start,
which
is
a
little
uncharacteristic
of
us,
but
we
do
have
some
great
discussion.
Topics
I
think
it's
going
to
make
for
a
really
good
Community
call
around.
A
You
know
talking
about
application
deployment
manager
and
and
how
we're
going
to
advertise
our
Watson
Cloud
roadmap
and
then
kind
of
have
an
exciting
announcement
before
we
have
some
free
time
at
the
end
of
the
call.
So
this
agenda
will
be
published
right
after
this
meeting,
along
with
some
notes
from
this,
but
I
think
without
further
Ado.
We
can
get
right
into
discussing.
So
the
first
item
on
our
agenda
is
wadham
Titan
and
productionizing.
It's
a
fun
word
and
Taylor
you're
leading
this
session
right.
A
A
B
Oh
dang,
it
we
didn't
get
the
normal,
hey
Liam,
my
missing,
sharing
permissions
thing
that
normally
happens
so
anyway.
So
I
have
a
couple
things
I
wanted
to
talk
through
the
first
one
I
hope
will
be
relatively
quick,
quick,
but
I
wanted
to
solicit
some
ideas
so
I'm
just
talking
among
various
like
project
maintainers
and
stuff,
we
wanted
to.
Probably
we
wanted
to
really
start
using
GitHub
like
projects,
so
you
could
actually
like
track
where
things
are
at.
We
can
drag
it
through.
B
We
can
show
where
work
is
happening
as
far
as
I
know.
There's
two
ways
to
do
this:
we
can
try
to
have
like
a
top
level,
one
in
the
wasm
cloud,
Watson
Cloud
repo,
that
kind
of
shows
where
everything's
at
or
we
can
have
per
project
ones.
There's
good
Arguments
for
each
one
because,
like,
for
example,
wash
gets
a
lot
of
love
and
it's
always
being
worked
on,
so
it
could
be
better
to
have
that
as
a
separate
one.
B
But
we
also
could
say
that,
like
a
lot
of
these,
things
like
have
interdependencies
on
on
each
other,
so
I
just
wanted
to
put
it
out
there
for
the
community
like
what
would
be
best
for
for
everybody
to
follow
along
with
with
what's
happening.
Do
you
care
more
about
like
what's
happening
in
like
one
or
two
repos,
or
do
you
care
like
more
about
the
macro
level
about
what's
happening
everywhere,
because
I
want
to
make
sure
we're
addressing
the
like
needs?
The
community
here,
as
we
set
this
up.
B
The
answer
to
question
Stephen
so
for
Stephen
just
asked
the
question:
why
not
have
the
tracking
per
repo
and
pull
that
into
Mac
into
the
macro
level?
That
is
what
I
mean
by
macro
level
and
that's
just
a
lot
more
effort
like
if
I
set
up
a
project
there.
C
B
A
project
there,
that's
more,
that
I
have
to
go
manage
about
where
things
show
up
and
whatnot
I
personally
am
leaning
towards
the
per
repo
thing,
especially
as
the
project
grows
like
we're.
Gonna
need
that
anyway,
and
so
that
was
what
it
was
was
going
for
and
Stephen
how
about
you
just
come
off
mute
if
you
can
and
just
describe
what
you're,
what
you're
saying,
because
I'm
not
sure
what
you
mean
by
dashboard.
D
So
you
know
in
jira
how
you
can
a
lot
of
places.
I've
been
there's
been
a
lot
of
like
goals
and
things
underway
at
the
same
time,
and
when
it
came
to
release
tracking
what
they
would
end
up
doing
is
creating
a
dashboard
which
is
just
like
a
visual
representation
that
pulls
in
information
from
the
other
project
repositories
and
just
displays
it
in
some
kind
of
formatted
way.
I,
don't
know
if
that's
I've
never
used
GitHub
projects,
but
I
don't
know.
D
Maybe
there
would
be
a
way
to
use
a
markdown
readme
or
something
and
pull
in
information
or
use
GitHub
actions
to
update
some
kind
of
top
level,
readme
that
that
could
display
project
information
at
the
wassom
cloud
level.
So
if
someone
just
wanted
a
higher
level
scope
of
what's
going
on,
they
could
just
open
it
and
see.
But
then,
if
you're
looking
at
a
project,
the
Project's
still
set
up
in
the
repo
like
you're
saying
I,
don't
know
if
it
would
be
possible.
C
B
So
I
I
think
that's
probably
like
the
next
projects
plus
plus
I
I.
Think
it's
a
little
bit
too
much
effort
like
right
now,
like
I,
think
of
it.
This
way
like
this
is
a
very
selfish
way
to
think
about
it.
But
honestly
it's
the
way
I
think
about
it.
I
can
spend
five
hours
getting
that
working
or
I
can
spend
five
hours
implementing
something
in
Madame.
So
for
me,
I'm
gonna,
say:
I'm
gonna
spend
that
five
hours
working
on
the
damn
and
just
set
up
projects.
So
that's
my
personal
thinking
on
it.
B
I
don't
I
have
not
met
a
ton
of
people
who
want
to
track
the
whole
status
of
every
single
open
source
project
under
an
umbrella.
Before,
in
my
past
experience,
but
I
mean
I'm
willing
to
do
it.
If
there's
a
lot
of
people
who
really
want
to
see
that.
A
So
I
I
have
I,
have
a
thought.
I
I
feel
like
tracking
the
whole
dashboard
of
the
entire
wasm
Cloud
organization
is,
is
probably
most
important,
with
a
really
spread
out
set
of
projects
where
we
are.
You
know
individually
working
on
them
such
that
like
when
you
come,
and
you
want
to
experiment
with
wasmcloud.
You
actually
have
to
care
about
the
status
of
a
bunch
of
different
pieces.
A
You
know
from
a
maintainer
perspective,
it's
how
it
would
be
helpful
for
us
to
have
an
idea
of
what's
going
on
in
the
whole
organization,
but
from
a
a
user
of
wasmcloud.
They
probably
wouldn't
need
to
take
a
look
at
this
unless
it's
regarding
a
specific
feature
like
they.
If
you
use
an
open
source
project,
you
probably
don't
go,
take
a
look
at
all
the
open,
PRS
and
issues
and
things
unless
you
actually
have
an
issue
and
for
a
contributor
most
things
in
our
org
when
you
contribute
are
usually
in
one
project.
A
Sometimes
you
contribute.
You
know
something
to
was
on
cloud
OPP,
and
then
you
update
the
version
and
watch
or
something
like
that,
but
I
feel
like
I
have
to
sneeze,
but
I
I
think
that
what
would
be
really
high
impact
like
in
the
short
term
is
projects
per
repository,
and
then
people
can
get
a
good
idea
of
how
their
individual
thing
they
might
be
contributing
to
is
is
going
to
progress.
A
That's
kind
of
my
gut
feeling.
I
still
think
it
would
be
useful
to
have
it
in
some
Central
locations
say
like
here's
all
the
things
going
on,
especially
when
a
specific
feature
spans
more
than
one
repository.
So
you
can
get
an
idea
of
of
what
is
needed
to
wrap
that
up
rambled
a
little
bit
at
the
end
there,
but
does
that
kind
of
line
up
with
what
you're
thinking
Taylor
I
know?
You
said
that,
like
you
know,
time
spent
is
really
valuable
here.
B
B
When
you
become
a
control,
you
can
become
a
maintainer
of
a
specific
project
or
couple
specific
projects
or
you
can
become
an
org
maintainer
org
maintainers
are
like
kind
of
responsible
for
running
the
whole
org
and
everything
that's
going
on
at
that
level
and
project
maintainers
have
pretty
full
control
at
the
project
level,
and
so
the
project
maintainers
is
like
they're
going
to
want
to
be
able
to
manage
what
is
going
on
like
what
their
Milestones
are,
what
their
projects
are
so
I
think
having
that
is
critical.
B
What
I
think
I
can
do
is
start
with
that
and
see
if
I
can
create,
like
a
top
level
thing
that
can
at
least
query
or
pull
in
all
the
the
statuses
and
stuff
but
I
haven't.
There's
it's
been
about
like
I,
think
nine
months
since
I
last
heavily
interacted
with
GitHub
project,
so
there's
bound
to
be
new
features
that
I
that
weren't
there
before.
B
So
that's
the
that's
I
think
probably
the
easiest
way
going
forward.
If
that's
okay
with
everyone.
A
And
I
think
we
also
have
a
good
bit
of
you
know.
We
have
like
at
least
at
least
20
repositories
and
open
source.
Wasn't
cloud.org
I
think
there's
there's
some
of
the
effort
that
we've
talked
about
in
a
previous
Community
call
about
consolidating
some
of
them.
A
I
think
we
should
still
do
that,
but
that
just
makes
that
would
make
an
individual
project
more
and
more
useful,
the
less
repositories
that
we're
actively
contributing
to,
because
there
are
some
that
we
have
that
are
like
archives,
that
are
old
demos
and
things
that
that
don't
need
to
be
pulled
in,
but
the
simpler
we
can
make
our
GitHub
organization,
too,
makes
this
whole
process
easier.
I
think.
B
Okay,
sweet
we'll
go
ahead
and
just
do
that.
We're
gonna
I,
don't
want
to
take
too
much
time
on
this.
This
is
kind
of
Community
Management
stuff,
but
it's
good
that
people
know
because
everyone's
going
to
be
able
to
do
that
so
sweet.
That
was
my
first
thing
now
onto
something
slightly
more
exciting.
This
one
I
guess
I
can
share
my
screen,
but
I
will
also
share
the
link.
C
B
Tells
me
I'm
up
and
ready
to
go,
I'll
make
sure
I
have
the
chat
up
too
in
case.
Anyone
needs
to
give
me
that,
so
this
RFC
is
Slash
proposal
is
something
I've
put
up
for
wadham
or
Witham
as
I
like
to
call
it,
and
so
this
it's
a
it's
a
pretty
long
one,
and
it
is
mostly
in
the
technical
details.
The
big
thing
that
I
want
to
stress.
The
community
is
like
essentially
that
first
overview
of
the
sentence
overview
sentence.
B
That's
on
here,
like
we
kind
of
started
with
damn
and
then
kind
of,
let
it
sit
to
do
a
lot
of
stabilization
and
some
feature
stuff
inside
of
the
host,
and
so
it's
always
been
kind
of
like
a
rough
Alpha
level
of
support.
It's
been
there.
You
can
try
it
out,
but
there's
like
a
lot
of
things
that
have
even
changed
like,
for
example,
it
requires
two
databases
under
the
hood
because
Nats
didn't
even
have
KV
stuff
built
in
when
we
started.
B
So
that
is
all
that's
all
kind
of
like
accounted
for
in
here,
and
so
one
of
the
things
that
we
that
I
want
to
do
is
talk
about.
Okay,
what's
the
work
that
we
need
to
do
to
to
get
it
done
and
and
out,
and
so
that's
what
this
RFC
is.
It
goes
over
just
like
it's
typical,
like
proposal
stuff,
what
are
the
goals
or
the
non-goals?
What
are
the
different
key
technical
decisions?
B
These
are.
These
are
like
the
different
things
like
different
architecture
designs,
why
things
are
argued
for
how
we're
going
to
support
things
like
custom,
schedulers
and
then
at
the
very
bottom,
which
is
probably
more
important
to
people.
Is
the
like
basic
road
map
of
what
it's
going
to
be
I
didn't
want
this
to
be
prescriptive
and
say
like
this
is
how
everything
should
be
done.
B
C
B
Or
two
is
what
I'm
hoping
for
we'll
see
how
it
goes.
We
want
to
finish,
there's
like
some
stage,
one
stuff
and
then
stage.
Two
is
a
lot
of
spiking
things
to
see
how
how
this
is
going
to
like
actually
work,
because
there's
there's
some
it's
one
of
those
things,
that's
easier
to
tell
when
you're
implementing.
B
If
it's
like
what
what
kind
of
method
is
going
to
be
good
or
not
so
that's
more
spiky
and
then
the
whole
goal,
if
you
don't
notice
here,
is
like
hey,
I
want
to
be
able
to
have
e
to
e
things.
I
need
functional,
deploy
pipelines
that
deploy
it
as
like
Freight
plain
binary,
Docker
container,
the
full
shebang,
so
you
can
run
it
anywhere.
You
want
and
then
also
making
sure
that
it's
built
into
wash
all
documentation
is
there
and
then
deploy
guides
are
in
place
so.
C
B
This
is
a
like.
This
is
not
a
dictatorial.
This
is
a
hey.
This
is
what
I'm
planning
on
doing
I've
done,
some
thinking
about
it.
Here's
the
ideas
and
so
I
invite
anyone
who's
interested
in
the
technical
details
to
go
comment
on
it
and
look
through
that.
I
have
some
example
work
of
like
what
I
was
trying
to
what
I
was
trying
to
do.
B
There's
a
link
there
I
think
this
will
be
done
as
a
feature
Branch,
that's
separate
from
the
main
branch,
so
we
don't
break
anything
that
anyone
that's
currently
using
it
and
then
we'll
we'll
push
towards
a
new
thing.
A
couple
key
points
that
I
wanted
to
point
out
here
with
with
some
of
this,
like
we
are
trying
this
these
last
two
sections
are
important
details.
B
We
are
building
in
not
implementing
right
now,
but
we
want
to
make
sure
we
have
support
for
custom
schedulers
that
you
can
just
extend
it
on
your
own,
but
we
also
want
to
the
thing
that
I'm
trying
to
push
for
here
is
that
we
create
something
that
could
be
used
as
an
example,
but
as
also
the
canonical
scheduler.
So
most
people
will
just
use
Witham
by
their
by
themselves,
just
plain
like
it'll
work
for
the
most
the
general
use
cases,
but
you
want
to
build
like
a
custom.
B
Scheduler
I
was
trying
to
learn
from
the
learn
from
the
experiences
I've
had
with
kubernetes
and
working
with
Nomad
and
stuff
that,
like
the
the
problem
you
have
with
with
tools
such
as
as
kubernetes
right
now
is
you're
kind
of
like
stuck
using
their
schedule.
You
have
there's
extension
points,
but
you
have
to
do
it
there.
You
can't
like
just
roll
your
own
scheduler
or
anything.
This
leaves
that
open.
B
If
you
have
an
advanced
use
case
and
leaves
on
the
building
blocks
around
one
of
the
questions,
Steven
asks
is
logging
tracing
one
on
stage,
that's
built
in
and
everything
in
my
opinion,
like
that's,
not
a
separate
thing.
B
I
already
actually
have
logging
and
tracing
in
there
and
the
stuff
that
I've
I've
started
to
build
like.
So.
If
I
click
on
that
and
show
you
we
have
like
if
I
go
through
to
like
so
there's
some
like
leader
election
stuff,
I
wrote,
we
can
see
specific
stuff.
That
happens.
There's
instrumentation
for
each
of
the
the
levels
there's
like
I'm
tracing
when
things
are
starting
like
all
that
is
just
part
of
the
like.
That's
that's.
B
For
me,
when
you're
writing
code
and
when
I'm
reviewing
this
code,
that's
if
you're
someone
who's
going
to
want
to
contribute
I
will
be
looking
for.
Like
hey,
are
you
logging?
Are
you
tracing
some
stuff?
We
want
to
basically
just
bring
this
to
production
level
quality.
It's
still
gonna
be
a
pretty
1.0
thing,
because
we're
iterating
we're
doing
that,
but
it's
going
to
be
production
level
quality.
It's
the
goal
at
the
end
here,
so
there's
a
whole
bunch
of
stuff.
B
Here,
I
tried
to
give
every
single,
like
reasoning
behind
what
I'm
I'm
trying
to
like
push
for
and
build
I.
Don't
want
to
be
the
only
one
doing
this,
but
I
was
trying
to
lay
down
a
lot
of
the
scaffolding
so
that
people
could
build
stuff
on
top
without
too
much
overlap
and
or
rebasing.
So
anyway,
this
is
new.
I.
Just
put
it
this
morning,
I
had
it
mostly
done
yesterday,
but
it
was
just
trying
to
do
one
more
grammatical
pass
and
just
checking
everything
else.
B
So
I
just
put
it
pushed
up
about
30
minutes
ago,
so
yeah,
that's
all
I
have
like
I
said:
there's
not
much,
there's
not
really
a
demo
here,
but
I
just
wanted
people
to
let
people
know
where
we've
been
thinking
about
it,
and
this
is
the
RFC
to
try
to
get
us
over
the
line.
B
I'm
glad
you're
excited
Steven.
It's
even
just
said
that
in
the
Chinese
I
said,
I
will
note
like
this.
The
behavior
should
stay
roughly
the
same.
It's
just.
This
is
more
like
underlying
technical
details,
rather
than
rather
than
like,
changing
the
whole
like.
Ideally,
so,
if
you
actually
look
at
how
like
what
am
works,
you
have
the
oam
stuff
like
there's
no
change
to
that.
B
Ideally,
this
should
work
pretty
much,
as
is
I
mean
we
I
I
I,
always
ascribe
to
the
philosophy
of
preservative
philosophy
of
Never
Say
Never,
but
like
ideally,
this
shouldn't
change.
You'll
use
the
exact
same
the
exact
same
thing
as
you
were,
using
before
just
some
of
the
deployment
mechanisms
and
how
we
get
it
running
and
including
it
with
wash
and
all
that
kind
of
stuff.
It's
going
to
be
like
part
of
the
main
thing:
okay,
CJ
had
a
question
about.
B
What
would
you
say
is
the
biggest
gap
between
like
current
and
Target
state.
A
lot
of
it
is
around
there's
not
a
lot
of
testing
like
as
it
currently
stands.
There's
not
a
lot
of
testing
there's
not
a
lot
of.
We
just
have
some
basic
examples.
We
have
it
like,
there's
no
way
to
easily
deploy
it
like.
B
How
do
I
add
this
in
you
also
have
to
like
have
redis
and
like
we're
trying
to
make
it
so
there's
not
anything
in
redis
at
this
point,
and
that's
just
that's
where
we're
going
to
store
that
that
data,
since
we
already
have
it
so
there's
a
lot
of
those
things
also
in
here
is
the
the
decision
link
we
needed
to
have
people.
B
This
is
a
critical
component
for
so
like
if
we
look
at
the
goals
here
like
this
is
something
that
is
mostly
there
at
the
top
level
we're
working
there
and
like
this.
Is
there
like?
That's
that's
the
idea
of
what
am
in
general,
but
we
don't
have
anything
that
shows
us
how
we
can
operate
it.
We
don't
we're
not
basically
acting
as
a
canonical
scheduling
implementation.
We
are
by
like,
like
de
facto,
but
we
aren't
actually
implementing
as
such.
B
This
one
is
like
we
need
to
be
a
project.
People
want
to
contribute
to
so,
for
this
is
why,
down
here,
one
of
the
things
the
frozen
the
pro
the
proposition
here
is
that
we
rewrite
with
Ammon,
rust
and
I,
go
into
extensive
detail,
because
that
is
a
change
from
what
we
have
about
the
justifications.
B
Why
and
why
it
matters,
and
it
is
in
order
of
importance
in
these
sections
and
the
biggest
is
the
need
for
contributors
and
I
I
mean
I,
gave
the
tldr
right
there,
but
essentially
we
need
like
with
something
this
critical
in
an
ecosystem.
B
We
need
people
who
we
need
people
to
be
contributing
to
it,
and
we
need
people
who
can
go
and
understand.
What's
going
on,
contribute
with
a
language
that
they've
that
they're
a
little
bit
more
familiar
with
and
adds
a
couple
other
benefits
that
are
talked
about
underneath,
but
contributors
are
the
main
thing
there,
and
so,
like
that's
one
of
the
big
gaps
for
to
answer
your
question,
CJ
that
I
I
was
trying
to
address
with
that
decision.
B
Is
we
need
the
ability
for
people
to
more
easily
contribute
to
stuff,
and
we
can
I
can
go
over
that
some
other
time,
but
there's
there's
you
can
look
at
the
metrics
across
repos
and
you
can
see
where
contributions
are
happening
inside
of
like
the
different
laws
and
Cloud
repos
and
kind
of
get
a
feel
for
why
it's
important.
We
have
it
in
something
that
people
are
going
to
come
contribute
to.
B
So
that's
that's
one
of
the
big
things
is
we
have
to
have
that
because
if
it's
just
me
doing
it
like
yeah
I've,
written
schedulers
before
for
kubernetes
I've
worked
on
kubernetes,
and
that
makes
me
a
decent
expert,
but
that
doesn't
make
me
the
end-all
be-all
of
this
there's
people
who
literally
study
this
for
a
living
so
like
I,
don't
want
I
want
to
make
it
as
easy
as
possible
to
like
contribute
in
and
so
yeah.
We
talk
about
I
talk
about
all
the
different
things
inside.
C
B
Here
advantages:
disadvantages
for
Jordan
I,
specifically
explain
why
we
didn't
choose
go
even
though
that
was
a
friend
Frontline
competitor.
This.
C
B
Where
I
wish
Jordan
Jordan
could
camera
most
the
time,
because
I
really
want
to
see
his
laughter
at
this
point,
but
anyway,
so
anyway.
This
gives
us
a
whole
list
of
points
to
talk
about.
So
that's
the
those
are
the
the
gaps
we're
trying
to
to
solve.
There
is
just
really
the
productionizing
of
it
and
being
able
to
support
it
long
term
with
a
group
of
maintainers
that
can
grow
so.
B
Any
other
questions,
yes,
Stephen
has
a
question.
I'll,
just
read
it
back
for
those
on
live
stream.
Are
there
plans
or
discussion
around
bringing
terraform
or
similar
into
the
equation
after
what
AM
is
in
a
good
spot,
namely
for
the
addition
of
Hope
of
hosts
into
a
Nas
cluster
BSM
terraform
provider?
So.
B
Question
so
one
of
those
is
actually
about
spinning
up
hosts
and
one
is
about
spinning
up
with
damn
they
can
run
separate,
but
and
by
Design
we
actually
have
it
so
that
you
do
not
have
to
run
within.
We
figure
most
people
will
and
I'm
pretty
sure
and
we'll
when
we
get
to
implementing
and
we'll
debate
it
for
certain
but
I'm
pretty
sure
by
default.
When
you
run
wash
up
it's
going
to
start
with
am40.,
but
you
are
not
required
to
use
it.
B
You're
not
required
to
do
any
of
those
kind
of
things
with
it.
So
terraform
will
for
this
kind
of
set.
Those
are
the
kind
of
things
we
are
planning
on,
improving
on
the
documentation
side
of
Watson
Cloud
itself,
the
actual
host.
We
want
to
be
able
to
have
different
ways
you
can
deploy
like
multiple
hosts
across,
like
whatever
you
want
to.
B
Basically,
those
deployment
guides
are
kind
of
a
little
bit
of
a
weak
spot.
We
know
that
we're
going
to
be
working
on
on
adding
that
for
people
who
are
more
infrofocused
and
those
would
definitely
have
things
like
terraform
or
we're
also
completely
welcome
to
having
someone
contribute
terraform
that
they've
written
and
then
wadham
is
just
an
additional
like
add-on
process.
It's
essentially
a
sidecar
for
those
who
are
familiar
with
the
kubernetes
space
like
you
can
run
it
and
you
can
run
multiple
of
them.
You
can
run.
B
One
of
them
doesn't
really
matter
yeah.
Then
we
have
another
question
from
Oren
about
is
supervision
tree
a
major
Concept
in
the
scheduler
I'm
I
actually
kept
it
as
simple
as
possible
for
what
we're
doing
the
ideas
we
can
enter
it
into
I
went
back
and
forth.
I
I
wrote
State
machines
and
rest
for
kubernetes
in
the
crater
project.
I
specifically
call
that
out
for
those
who
are
rabbit
holy
like
many
of
us
engineers
and
want
to
go.
Take
a
look
at
that
that
detail,
but.
B
I
linked
it
right
here,
that's
greater
and
the
the
whole
idea
is
that
I'm
trying
to
this
is
supernet
I'm,
going
to
try
to
stay
too
much
out
of
the
weeds
because
we
could
get
real
into
the
weeds
here.
Basically,
I
want
to
make
them
commutative
so
that
it
doesn't
matter
which
order
you
run
through
so
basically
I
get.
B
You
can
give
it
an
event
or
like
a
manifest,
and
it
passes
through
any
arbitrarily
long
list
of
of
these
scalers,
the
things
that
can
make
decisions
and
then
no
matter
which
order
you
run
it
through.
It
ends
up
with
the
same
result
in
the
in
the
end.
So
it's
just
more
like
making
like
hey
what
decisions
need
to
be
made
down
the
tree
and
that's
why
I
said
like
I
need
to
skate.
I
need
to
actually
go
through
the
the
hackery
of
of
trying
to
implement
this.
B
To
just
see
some
of
the
like
nuances
of
it,
and
so
Jordan
asked
a
question.
This
is
one
we
talked
about
before,
so
why
aren't
we
talking
about
implementing
Witham
as
an
actor,
and
so
we
can
hold
State
and
talking
about
holding
State
providers,
so
the
main
thing
is
chicken
and
egg
I.
We
we
debated
this
extensively
beforehand
and
there
is
you
end
up
with
a
who
watches
the
Watchers
kind
of
situation.
B
We
could
get
around
this
by
adding
something
into
that's
like
the
equivalent
of
I'm
trying
to
remember
what
they're
called
I
can't
remember
if
they're
called
bootstrap
manifests
in
kubernetes,
but
basically
something
that
alt
like
is
loaded
from
disk
anytime
like
the
kublic
started,
and
it
would
start
those
pods
we
could.
B
We
could
Implement
something
similar
for
for
Witham,
but
the
thing
was
is
I
wanted
to
make
this
when
we
were
talking
about
as
well
and
I
kind
of
Revisited
that
as
I
thought,
I
didn't
address
it
in
here,
but
as
I
as
I
went
through
the
the
thinking
on
it,
I
want
to
create
something
robust
and
do
it
now
and
not
have
to
worry
about.
B
You
know
like
thing
to
thing
to
thing
inside
the
actor
and
then
worrying
about
which
thing
is
supervising
that
and
where
that
say
to
sword
like
I
I,
actually
really
want
to
do
that
personally,
but
I
wanted
to
start
from
somewhere
that
we
could
then
do
that
in
the
future.
Like
I'm,
not
saying
this
is
none
of
this
is
saying,
there's
very
I
should
say,
there's
very
few
things
in
this
RFC
that
are
like
Thou
shalt.
Not
like
we're
not
going
to
do
this.
B
We're
not
going
to
do
this,
it's
more
like
not
right
now,
and
so
to
keep
it
just
as
simple
as
possible
and
avoid
the
whole
chicken
and
egg
bootstrap
problem
like
we're
just
gonna.
Have
it
be
a
separate
process
that
doesn't
and
like
I
said,
be
down
in
this
section
that
I
kind
of
pointed
out.
The
idea
is
that,
like
the
key
functionality
is
reusable
and
I
mean
that'll
start
out
as
arrest
crate,
but
we.
C
B
Take
this
rest
stuff
and
build
use
as
building
blocks
into
the
actual
into
the
actual
actors
and
providers
in
the
future,
but
it
was
just
more
I
was
just
trying
to
follow
the
kiss
principle
here,
just
to
keep
it
there,
but
yeah.
We
went
back
and
forth
I'm
I'm
still
game
like
if
we
can
talk
about
it
but
yeah
the
bootstrap
problem
and
then
like
some
of
the
like
the
state
maintenance
stuff.
B
Was
it
and
yeah
ultimately
like
now
that
I'm
thinking
back
on
again,
ultimately,
what
it
comes
down
for
me
is
the
whole
like
who
watches
the
Watchers
like.
This
is
the
thing
that's
supposed
to
be.
B
Scheduling
this
is
supposed
to
be
scheduling
out
things
like
in
the
thing
that
it's
being
managed
by,
so
it's
just
trying
to
avoid
those.
B
Kevin
gets
you
in
just
second
James
I.
Had
a
question
in
chat:
what
about
the
supervisors
that
come
from
free
for
free
from
erling
or
Elixir?
If.
C
B
Isn't
supervision?
This
is.
This
is
totally
event
driven.
So
essentially
what
happens
is
when
it
gets
an
event
from
wasenbus.ebt.
It
says
well
like
what
was
it
was
an
actor
stop.
Was
it
a
hostock
that
can
trigger
different
kinds
of
actions?
It's
not
like
watching
a
process
to
be
alive.
It's
more
about
saying,
like
Okay,
did
by
state
change,
I'm
going
to
do
something.
I
did
my
state
change
I'm
going
to
do
something
now,
I
mean
you?
Could
you
can
code
it
up?
B
I'm,
not
saying
you
can't
code
it
up
on
that,
and
then
you
can
look
in
the
the
language
section,
for
other
other
reasons,
why
specific
things
were
chosen.
So
there's.
E
Yeah
so
along
the
lines
of
the
kiss
principle-
and
maybe
this
is
just
because
I'm
I
haven't
looked
into
it
in
that
much
detail.
Since
you
said
your
branch
isn't
that
old,
when
you
mentioned
the
the
branch
has
a
leader
election
part
of
my
my
distributed
system.
Brain
treats
that
as
a
warning
like
our.
Why
are
we
re-implementing
leader
election
because
implementing
leader
election
is
generally
something
that
most
people
do
wrong,
so
I'm
just
trying
to
figure
out
what
what
functionality
in
the
branch
requires
our
own
custom
leader
election.
B
So
that's
this
section
right
here:
we
need
the
ability
to
be
able
to
run
multiple
copies
just
in
a
high
availability
scenario.
This
the
simplest
thing
is
with
leader
election,
because
then
you
just
have
one
process
running:
there's
no
State
coordination
or
anything
like
that.
Yeah.
E
The
other
is
that
currently,
the
the
what
I'm
application
does
run
as
a
cluster,
with
only
one
leader
for
each
given
observed
lattice,
so
in
in
today's
code
base
we
essentially
get
leader
election
for
free
from
OTP,
and
you
know:
OTP
has
like
20
years
of
refining
the
leader
election.
So
what
I'm
worried
about
is.
E
B
E
It's
a
it's
a
thing
that
we
have
to
implement
new
in
this
rewrite
that
we
didn't
have
to
implement
before
so.
I
was
just
curious
where
the
need
came
from.
B
E
I
don't
want
a
rabbit
hole
on
this
I
was
just
trying
to
figure
out
where
what
what
part
of
wadham
made
a
society
in
our
RFC
that
we
needed
leader
election.
B
It
was
just
all
about
having
high
availability,
I
thought
about
Q
subscriptions,
but
maybe
I
misunderstood
something
about
Q
subscriptions
and
how
they
were
going
to
work
for
us
and
I'm
fine
just
going
to
key
subscriptions.
If
that's
easier,
the
the
main
thing
was
around
the
fact
that,
like
anything,
can
observe.
In
fact
that
was
my
plan.
Is
that
like
it'll
always
be
observing
and
having
like?
Basically
the
like?
What's
the
current
state
of
the
lattice
available.
C
B
But
like
what
what's
allowed
to
perform
actions,
what's
allowed
to
essentially
send
the
control
thing
like
start
actor
or
start
provider,
those
things
need
to
be
locked,
so
one
thing:
does
it
otherwise
you're
going
to
have
like
20
different
things,
try
to
start
and
for
start
provider
that
doesn't
matter
but
first
start
an
actor
that'll
bump
up
the
actor
count
for
every
single
one
of
the
Witham
processes.
That's
running
yeah.
E
That
makes
sense
you
know,
like
you,
said,
the
leader
election
is
one
way
to
do.
It
I
think,
there's
still
a
way
to
do
what
you're
talking
about
where
you
have
high
availability
and
only
one
thing
actually
issuing
control
commands
to
a
lattice
without
having
to
pick
a
leader.
B
Yeah
and
I'm
I'm
all
open
for
that.
So
sounds
like
a
good
one.
We
should
probably
discuss
in
the
in
the
proposal
because
I,
if
there's
a
good
idea
here,
I'd
love
to
do
that.
Instead
of
doing
this,
I
mean
I
did
Implement
and
test
everything
here
just
to
be
sure,
but
that
doesn't
mean
I
didn't
do
it
Bud,
free,
because
I
always
write
bugs
so
yeah.
B
A
I
I
just
wanted
to
say
that
I
actually
had
a
really
similar
question
and
I
think
that
was
a
good
kind
of
bow
to
tie
on
the
leader
and
election
present
whatever,
but
I
I
had
a
similar
question
about
it.
So
I'd
love,
it
I
think
he
said
it,
but
just
to
reiterate:
let's
talk
about
it
in
the
issue,
because
I
want
to
be
able
to
read
some
of
the
rationale
behind.
A
B
B
A
All
right
well,
the
last
thing
that
I
wanted
to
last
thing
that
I
wanted
to
bring
up
in
today's
meeting
and
normally
I.
Do
you
know
I
I,
try
to
keep
you
know,
I
work
at
cosmonic
and
and
try
to
keep
this
meeting
like
kind
of
all
focused
on
on
the
community,
but
I
did
want
to
bring
up.
A
One
thing
that
we
launched
yesterday,
which
is
really
exciting:
cosmonic,
is
actually
sponsoring
a
hackathon
for
distributed
webassembly
applications
and
this
I'll
go
ahead
and
ping
the
go
ahead
and
ping
the
link
in
in
chat,
but
you
can
actually
find
it.
It's
like
one
of
the
first
hackathons
or
it's
in
like
the
first
page
of
hackathons
on
on
devposts.
A
But
one
of
the
reasons
why
we
wanted
to
do
this
is
to
really
outline
a
hackathon
experience
for
somebody
looking
to
build
something
using
wasm
cloud
and
then
essentially
just
take
that
and
host
it
on
the
cosmonic
platform.
So
the
reason
why
I
wanted
to
publish
that
and
talk
about
it
here
is
that
anybody
in
the
wasmcloud
community,
if
you
are
building
a
a
project
using
a
wasm
cloud,
if
you're
creating
an
application-
and
you
know
it's
it's
doing
something-
really
interesting-
something
that
is
or
even
just
really
fun.
We'd
love
it.
A
If
you
took
that
and
submitted
it
to
this
to
this
hackathon,
really,
the
only
requirement
is
that
you
have
a
wasmcloud
application
and
you
deploy
it
on
the
cosmotic
platform.
So
this
is
open
to
everyone
in
including
outside
of
the
us.
It's
going
to
be
running
for
the
next
I
think,
eight
weeks
or
so
so
you
have
a
good
amount
of
time
to
get
your
ideas
together
and
get
something
implemented,
but
nothing
gets
people
motivated
to
build
a
little
something
something
than
a
little
dollar
amount
attached
next
to
it.
A
So
we
have
a
few
prizes
for
the
people
who
you
know
have
the
most
interesting
project
to
you
know
the
honorable
mentions,
and
things
like
that
so
I
just
wanted
to
publish
this
here,
would
love
it
if,
if
you've
been
thinking
about
building
something
and
just
haven't
quite
gotten
to
it,
yet
this
would
be
a
great
thing
to
submit
to
and
yeah
so
just
wanted
to
just
wanted
to
announce
that
we've
been
working
on
putting
that
together
for
a
little
bit
super
excited
to
see
what
everybody
comes
up
with.
A
If
you
get
to
submitting
to
the
hackathon.
A
Unless
there
are
any
questions
on
that
piece,
I
think
this
is
the
end
of
our
kind
of
regularly
scheduled
agenda.
Let's
go
ahead
and
start
the
we
can
start
the
free
time,
part
of
the
community
meeting.
Anybody
have
any
high
level
questions
about
or
things
they
want
to
discuss
about.
Wasmcloud.
D
Is
there
a
way
to
see
like
any
logging
currently
in
the
terminal
right
now,
I
just
get
like
connecting
to
host
and
important
that
kind
of
stuff,
but
then,
when
I
investigate
like
the
wash
dashboard
I
see
like
links
have
been
added
and
actors
and
providers
haven't
been
in
if
I
try
to
list
the
app
I
get
like
erratics
connection.
Error
reason
closed.
So
I,
don't
really
know
what
that
kind
of
stuff
means
right
now.
A
Okay,
there
should
be
logs
coming
out
of
wadham.
It's.
A
A
Here,
I'll
send
you
because
Stephen
I
know
you're
in
the
slack
I'll
I'll.
Send
you
something
that
you
can
set
the
logs
for
wadham.
It
sounds
like
you
might
need
to
because
depend
on
redis
as
well.
So
might.
B
Yeah,
that's
what
I
was
mentioning
like.
That
was
one
of
those
things
that
we
we
like
were
like
when
we
created.
C
B
There
was
no
like
way
to
store
this
like
in
the
cave,
East
or
like
Nats.
Had
you
know
so
it's
like
well
red
as
it
is,
and
now
we're
like.
No,
we
don't
need
that
anymore.
So,
like
the
goal
was
to
just
be
like
no
we're
just
using
that
so
you'll
use
the
same
stuff
you're
using
for
your
like
lattice,
so
should
be
pretty
easy.
D
A
D
There
was
a
plan
to
remove
redis,
eventually
from
what
what
am
right.
A
Yeah
exactly
what
what
Taylor's
essentially
talking
about
there
at
the
time
there
wasn't
really
Nat's
key
value
when
we
first
wrote
it,
and
now
there
is,
and
that
would
be
great,
but
we
just
wanted
to
get
something
prototyping
and
and
working,
which
kind
of
the
easiest
way
to
do.
That
was
just
to
reuse,
a
redis
connection
that
you're
probably
already
running
for
for,
like
Watson
Cloud
stuff
Kevin.
Did
you
want
to
add
a
little
more
detail
because
you
were
there
when
the
the
book
was
written.
E
Yeah
I
was
there
back
before
the
war.
Just
the
the
context
I
wanted
to
add
is
that
ever
since
Nats
switched
from
the
the
product
they
called
Stand
to
the
new
products
called
Jetstream,
we've
always
been
kind
of
sitting
around
saying
yeah.
We
should
probably
modify
wadham,
so
it
uses
Nat
streams
instead
of
redis,
and
you
know
priorities
and
time
being
what
they
were.
E
We
just
never
got
around
to
it,
and
so
you
know
in
Taylor's
proposal
it's
just
kind
of
putting
in
in
concrete
terms
that
we've
always
wanted
to
switch
from
redis
demands.
We
just
haven't
in
the
current
product.
C
D
For
it,
so
what
I've
been
really
curious
about
is
the
The
Vault
and
wadham
integration,
because
right
now
something
I'm
doing
is
pulling
configuration
from
like
a
vault
provider
through
a
vault
provider.
But
wadham
has
KV
Vault
integration
so
like
would
it
be
reasonable
to
just
remove
the
KV
server
that
I'm
running,
like
does
wadham
startup,
like
some
kind
of
Vault
server
instance
that
you
can
access
and
then
use
that
for
all
secrets
in
your
lattice
or
cluster,
like
how
should
you
be
managing
vault?
In
that
scenario,.
A
Yeah,
that's
a
good
question.
So
there's
there's
a
few
pieces
there
there's
the
actual
Vault
server
like
hashicorp
vaults.
That's
actually
storing
your
your
secrets
and
everything,
and
then
we
have
Autumn
that
interacts
with
a
vault
server
to
pull
credentials
to
connect
to
a
lattice,
and
then
we
also
have
the
key
value.
Vault
implementation,
like
the
capability
provider
for
wasmo
cloud
So,
when
you
say
you're,
pulling
credentials
from
Vault
you're
talking
about
like
an
actor,
is
talking
to
a
vault
provider
that
pulls
from
a
vault
server.
Yeah.
D
Exactly
but
I'm
wondering
if
you
can
just
like
avoid
that
by
having
like
some
kind
of
wad
of
global
secrets,
that
you
can
didn't
pull
from
yeah,
that's
kind
of
like
there's
kubernetes,
where
it's
like.
You
have
these
Secrets
available
to
your
certain
name,
spaces
that
you
can
grab.
E
Go
ahead:
yeah
if
I'm,
if
I,
understand
the
core
scheme
right
I,
think
that
it's
a
that
providing
secrets
to
everything
within
a
lattice
is
a
separate
set
of
functionality
than
from
what
random
does
so.
Wadham
is
basically
an
observer
and
reconciler
robot.
It
Patterson
makes
changes
to
the
lattice
and
it
may
need
secrets
to
do
that
job,
but
its
job
isn't
to
provide
secrets
to
the
rest
of
the
lattice.
Okay.
E
If
you
want
to
give
an
actor
access
to
Secrets,
like
Brooke
said
you
can
give
it
a
you
can
hook
it
up
to
a
key
value
provider
that
is
backed
by
Vault,
there's,
also
a
configuration
service
that
we
have
where
you
could
write
an
actor
or
something
else
that
responds
to
configuration
service
requests
that
might
read
from
Abel,
but
yeah
I.
Think
the
the
key
idea
is
that
why
Adams
core
responsibility
doesn't
include
providing
secrets
to
what
that
is.
Okay,.
D
A
Kevin
said
it
a
lot
better
that
than
than
I
was
gonna
say,
but
I
was
just
going
to
talk
about
the
same
thing.
Like
the
separation
of
you
know
what
is
actually
a
vault
server
and
and
all
that,
but
that's
a
great
that'd
be
a
great
topic
to
bring
up
in
future
Community
call
if
we
can
get
some
RFC
or
thoughts
around
it
may
spin
off
on
a
slack
threat
after
this
call,
just
like
a
broader
I,
have
a
secret
I
want
it
available
to
the
whole
lattice?
A
A
Cool
well
Stephen,
if
you
have
any
other
Autumn
questions,
we're
happy
to
happy
to
go
through
that,
but
I
just
want
to
open
up
the
the
open
floor
to
an
even
bigger
open
floor.
If
there
are
any
questions
for
the
webassembly
community
as
a
whole
stuff
going
on
in
wasm
now
is
now
is
a
great
time
we
can.
We
can
chat
about
that,
so
this
is
essentially
totally
free
time.
If
anybody
has
some
on
their
mind.
D
D
Cool,
so
with
actor
to
act
or
communication
like
when
you're.
Is
there
anything
specific
you
have
to
do
in
bottom?
Like
there's,
no
link
definitions,
you
define
so
just
wondering
if
there's
anything
there,
that
needs
to
be
of
concern
like
are
you
just
using
an
interface?
You
don't
have
to
declare
it
anywhere
so,
except
in
the
in
the
actor
code.
D
Oh
that
reminds
me
like,
with
the
messaging
provider
I
noticed
that
there
was
message
subscribe,
subscriber
or
something
like
that,
which
it
was
kind
of
like
geared
towards
accurate,
active
calls,
but
there
was
only
a
handle
method.
So
I
was
curious
about
that
because,
like
when
you
implement
your
own
interface,
you
you
end
up
getting
like
the
sender.
D
So
what
was
there
like?
Was
there
a
goal
to
have
the
messaging
provider
be
able
to
do
actor
to
actor
communication
or.
E
Okay,
so
the
messaging
provider
is
definitely
not
for
actor
to
actor
communication.
Do
you
to
do
actor
to
actor
calls
you
don't
need
a
provider.
D
Yeah
I
set
provider
to
false
I,
just
I
was
curious
because
they
had
like
I.
When
you
look
in
the.
E
So
the
the
way
that
I'm
trying
to
figure
out
which
which
which
question
I'm
answering
in
general,
the
way
you
do
after
stuff
is
you,
create
a
contract
and
you
set
it
to
actor.
Handle
is
true,
and
so
then
you
define
your
actor
and
you
implement
the
handle
function
for
that
and
then
you
in
your
in
the
actor,
that's
making
the
call.
E
You
create
an
instance
of
the
sender
for
that
contract
and
then
the
target
is
specified.
I
forget
the
exact
function
name,
but
at
least
in
Rust
it's
something
like
with
actor
or
hue
actor,
something
like
that
and
so
you
get
to
and
from
from
I.
Don't
think
we
do.
But
when
you
create
the
sender
instead
of
doing
you
know
my
contract
sender
new,
you
would
do
my
contract
sender
like
to
actor
two
actor.
E
I
forget
the
exact
function
name,
but
that's
generally
how
that
works,
and
so
when,
when
you
tell
the
wasmcloud
host
that
you're
going
to
communicate
with
using
this
particular
contract
and
targeting
this
particular
actor,
it
doesn't
go
through
a
provider.
It
just
immediately
converts
that
into
an
RPC
call
and
the
target
is
pseudo.
Randomly
chosen,
among
all
of
the
instances
of
that
actor
that
are
running
in
the
last
cool.
D
E
E
So
in
that,
in
that
scenario,
the
public
key
changes
so
yeah
there.
What
you
would
end
up
doing
is
when
you
ask,
when
you
create
that
instance
of
the
sander
instead
of
supplying
the
instead
of
supplying
the
public
key,
you
would
Supply
the
Alias
and
yeah.
It
automatically
takes
care
of
converting
the
Alias
into
the
public
key
based
on
the
claims
that
are
cached
in
the
lattice.
D
At
least
for
the
caller:
yes,
that
was
that's
something
that
confused
me
about.
Wadham
is,
if
you
had
to
declare
an
alias
anywhere
there,
but
yeah.
E
Doesn't
do
aliases
either
it
really
doesn't
care
yeah.
That's
what
I
thought:
okay
cool
and
in
the
chat
there's
a
there
was
a
question
that
Brooks
already
answered
about
disabling
accurate
actor
calls,
and
so
by
default.
We
allow
them
all
the
time.
But
if
you
set
up
a
policy
server
that
policy
server
can
reject
specific
actor-to-actor
calls,
so
the
the
policy
server
will
be
told.
This
actor
would
like
to
talk
to
this
other
actor
and
you
can
choose
to
reject
her
or
allow
that
call.
A
Cool
yeah
we
we
had
some.
We
had
some
good
discussions
around
allowing
actor
to
act
or
calls
by
default.
Actually
I
think
that
it
might
even
be
an
ADR
that
we,
you
know
formalized
on
allowing
them
by
default.
We
could
certain
I'm
going
to
try
to
dig
that
up
while
I
while
I
talk.
We
could
certainly
think
about
like
a
flag
to
disable
them
entirely,
but
hey
I
found
it
here
we
go.
A
You
could
certainly
think
about
making
it
so
that
you
have
control
over
whether
it's
beat
you
by
millisecond.
You
know
Allowed
by
default
or
denied
by
default.
I.
Think
a
core
tenant
of
this
was
that
if
you
have
an
actor
trying
to
call
another
actor,
it's
really
hard
to
misuse
capabilities
in
this
way,
when
you
design
your
actors
to
be
a
little
more
single
responsibility,
because
you
can
never
really
tell
an
actor
to
do
something
that
it's
not
supposed
to
with
with
the
capabilities
model.
A
And
of
course
the
you
know,
speaking
in
absolutes,
there
is
is
probably
a
bad
way
to
approach
it,
but
the
the
flexibility
of
not
linking
to
actors
together
really
really
kind
of
outweighs
the
benefit
of
denying
it
by
default.
John.
Let
me
send
you
an
example
of
a
policy
service.
There's,
there's
an
actor
in
the
open
source
repository.
That's
just
a
a
single
actor
that
implements
a
policy
that
you
can
only
start.
A
You
can
only
start
webassembly
modules
or
or
providers
that
are
signed
by
wasmcloud.
So
hey
and
sorry,
if
we
didn't
didn't
read
that
question,
but
John
actually
asked
if
there
were
any
examples
for
this
type
of
policy
service,
and
we
have
one
in
the
open
source,
Watson,
Cloud
examples,
repo
under
actor
and
then
policy.
A
So
this
is
just
something
that
the
wasm
club
host
will
send
out
an
app's
message
to
and
say:
hey
am
I
allowed
to
do
this
thing,
and
this
just
lets
you
put
on
more
security,
you
could
prevent
actors
or
providers
from
being
started.
You
can
prevent
invocations
from
going
through
things
like
that,
and
this
is
It's
a
generic
thing,
so
you
could
implement
it
for
a
variety
of
different
back-ends.
F
B
Kevin
raised
his
hand
I'll,
let
him
do
it.
He
followed
the
rules
of
the
classroom,
no.
E
You
you
owned
in
first
so
go
ahead.
B
So
I
was
just
gonna
say,
like
the
the
reason
is
like
we
don't
everything
we've
we
tried
to
do
with
Blossom
cloud.
Is
we
don't
want
to
like
box
people
into
specific
Solutions
or
config
things?
So
we
don't
want
to
say
like
this
is
the
way
we
do
it
and,
like
you
have
to
know
our
like
our
syntax
and
our
way
to
approve
policies
and
our
way
to
do
that
with
config.
B
B
Opa
anything
and
so
yeah
and
like
Kevin,
basically
put
it
in
in
chat
like
it's
up
to
platform
Builders
and
available
at
runtime,
rather
than
having
it
be
like
something
we
make.
The
decision
on
as
the
maintainers
of
awesome
cloud.
G
Yeah,
it's
a
pretty
interesting
like
Dynamic,
that,
like
in
some
sense,
there's
no
space
between
a
static
thing
that
is
in
your
control,
versus
a
programming
language
that
has
to
be
called
over
the
network
to
make
choices.
A
Yeah
I
think
a
lot
of
our
goal
for
the
policy
Service
as
it
stands
today
is
to
prove
out
a
use
case.
You
know
the
the
policy
service
we
wanted
to
have
an
answer
to
people
who
are
saying
well,
I'm,
administering
a
wasm
cloud,
lattice
I'm
running
into
my
system.
How
do
I
make
sure
that
if
my
control
interface
connection
gets
compromised,
that
somebody
can't
take
arbitrary
actors
and
providers
and
start
them
on
my
wasm
cloud,
hosts
and
and
really
the
answer
to.
A
So
it's
it
we're
trying
to
actively
prove
out
that
use
case
in
a
generic
way
and
as
we
go
through
it,
it
may
end
up
that
95
of
the
time
when
people
are
interacting
with
this
policy
service
that
they're
using
some
implementation
that
uses
open
policy
agent
or
something
like
that,
and
in
fact
this
is
essentially
the
way
that
RPC
happened
with
wasmcloud.
We
went
through
a
couple
of
months,
maybe
years
proving
out.
You
know,
there's
there's
inner
process
calls
and
there's
Nats
or
PC.
A
We
just
found
out
that
we're
using
Nats
all
the
time
and
and
then
making
that
technology
Choice.
While
it
removes
a
little
bit
of
the
abstraction
it
just
makes
everything
better.
So
if
you
try
out
the
policy
service
and
you
you
take
a
look
at
it
and
you
find
the
abstraction
a
little
frustrating
or
or
doesn't
quite
feel
right,
it's
because
we're
we're
actively
trying
to
figure
out
the
right
right
way
to
commit
to
something
like
that
and
keeping
it.
Generic
allows
us
to
have
the
flexibility
to
actually
try
that
out.
Foreign.
A
Great
questions
everyone
I
really
enjoyed
this
kind
of
free
time
at
the
end
of
the
call
I
think
we
got
a
lot,
I
got
a
lot
done
and
a
lot
talked
about.
This
is
right
at
the
end,
though,
actually
Taylor
just
dropped
off
too
so
I'll.
Let
everybody
get
back
to
their
busy
days
and
thanks
for
thanks
for
coming
by
asking
great
questions.