►
From YouTube: wasmCloud Community Meeting - 28 Jun 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
All
right,
hello,
everyone
welcome
to
the
Watson
Cloud
community
meeting
for
Wednesday
June
28th.
Let
me
go
ahead
and
pull
up
the
agenda
here
so
that
we
can
get
started.
We've
got
one
demo
as
usual
and
then
we're
gonna
get
into
a
pretty
meaty
discussion
around
link,
names
and
I.
Think
this
will
be.
This
will
be
a
really
fun.
One.
We've
had
a
couple
of
discussions
in
the
community
slack
around
linkedin's
what
they're
useful
for
things
that
we
wish
they
could
do
so
I
think
this
is
going
to
be
a
great.
A
This
is
going
to
be
a
great
thing
for
this
discussion.
Thank
you
everybody
for
for
coming
on.
If
you
took
part
in
the
conversations
on
slack,
so
just
looking
forward
to
continuing
that,
but
without
further
Ado
Taylor,
why
don't
you
go
ahead
and
show
off
your
system?
Information
provider
should
have
permissions.
B
B
Okay,
let's
go
on
okay,
so
I
have
created
a
new
I've,
been
hacking
around
this
week
and
I
created
a
new
experimental
contract,
slash
provider
that
can
be
used
for
for
all
sorts
of
purposes.
Okay,
I
can
go
through
some
of
the
ideas
here,
but
this
is
basically
like
a
system
monitoring
thing
now.
I
know
that,
like
things
already
exist,
you
have
Prometheus,
you
have
different
agents.
B
You
can
run
all
that
stuff,
I
get
it
so
this
was
just
toying
around,
but
it
is
interesting
on
how
it
integrates
into
the
into
the
Watson
cloud
like
ecosystem.
So
this
new
contract
is
a
simple
thing
that
I'm
actually
going
to
pull
over
my
code,
so
I
can
show
some
of
this,
and
so
it's
a
really
simple
interface.
That
just
says
like
hey
I,
have
this
handle
metric
event
that
any
that
an
actor
can
can
Implement
and
it.
B
It
has
the
actual
metric
event
right
now.
It's
it's
super
simple
and
straightforward,
but
it
has
system
metrics
that
right
now
are
sent,
which
are
your
classic
ones
like
CPU
swap
memory,
just
those
super
basic
stuff.
C
B
Can
add
I'm
I'm
envisioning
in
the
future?
We
could
add
things
like
data
about
the
host
itself,
about,
like
whatever,
whatever
you
want
to
add
on
I
mean
it
already,
has
the
ability
just
to
add
an
arbitrary
data
if
you
want
to
inside
the
inside
of
a
contract.
B
So
basically
what
this
thing
does
is
on
you
can
you
can
set
it
up
so
that
it
can
grab
from
any
data
source
and
send
it
in
so
you
could
have
it
running
and
you
could
be
collecting
the
data
from
the
host,
which
is
the
example
of
the
data
from
the
thing
it's
running
on,
which
is
what
I'm
going
to
show,
but
you
could
also
theoretically
do
this
by
querying
Prometheus
or
whatever
you
want
to
do
and
then
sent
using
that
data
to
trigger
conditions
through
here.
B
So
you
could
have
something
that
sends
off
the
data
and
then,
when
the
actor
could
then
do
something
like
call
the
control
interface
and
start
something
for
you
or
or
update
something
in
the
dam
or
something
along
those
lines.
So
I'll
go
ahead
and
start
these
real
quick.
So
we
have
from
registry.
We
have
I
forgot
to
start
this.
One
first
I
was
going
to
start
this
one.
B
B
B
What
it
does
is
on
a
tick
I
think
it's
every
10
seconds.
It
grabs
the
system,
data
from
the
thing
it's
running
on
and
it
publishes
it
and
whatever
actor
gets
it
can
can
handle
it
and
right
now,
I
also
have
just
a
really
dumb,
dumb
actor
that
comes
in
and
I
can
take
it
and
it
will
just
receive
the
data
and
republish
it
on
NAS.
It's
doing
nothing,
fancy
it's
just
to
demonstrate
that
this
all
works.
B
So
I
will
start
that
up
now,
if
I
come
over
here,
I
can
Nat
sub
on
I
think
it
is
monitor.
B
B
On
there
it
has
a
specific
uuid
and
then
it
shows
me
like
my
current
memory
and
and
other
usage,
so
you
keep
showing
me
this
consistently
and
keeps
troubleshooting
to
the
topic.
But
the
thing
is
this:
isn't
limited
to
one
host
I
can
also
do
this.
This
is
a
second
host
I
connected
second
washboard,
so
I'm
going
to
start
from
file
over
here
come
back
and
start
the
provider.
B
Okay,
now
that
that's
going
we'll
give
it
a
second
and
then
we'll
start
to
see
a
new
one
pop
up
it'll
take
about
10
seconds
before
it
takes
the
first
time.
So,
let's
see.
B
Okay,
there
we
go
so
you're
starting
to
see
we
have
two
of
them
going
right
there,
the
two
different
uu
IDs.
So
obviously
this
is
reporting
from
me,
which
is
the
same
system,
but
the
the
whole
point
is.
You
could
technically
start
these
on
any
host
that'll
that
will
allow
you
to
start
it
on
it
that
can
measure
the
metrics
and
then
it'll
send
those
metrics
out
to
you,
and
then
you
can
use
those
metrics
to
react
inside
of
your
actor
to
do
something
in
this
case
I'm.
B
Just
publishing
the
information
out
there,
like
I,
said
you
could
come
in
and
and
say
like.
Oh,
my
CPU
percentage
is
over.
50
I
am
going
to
do
X,
whether
that's
you
spin
up
something
that
you
know
how
to
do
to
spin
up
a
new
awesome,
Cloud
host.
You
could
you
could
like
basically
put
that
into
here,
and
it
would
do
an
actor
to
start
a
awesome,
Cloud
host,
which
would
be
pretty
cool
if
you
like,
create
a
proper
contract
for
that.
B
You
can
also
use
this
to
maybe
like
scale
up
or
down
specific
actors,
or
things
like
you
could
modify
with.
Am
this
might
be
something
we
actually
use
as
a
data
source,
as
we
do
custom
Witham
scalars
in
the
future
like
when
you
can
just
bring
your
own
scalar
that
you
want
to
want
to
do.
B
That's
this
type
of
thing
that
could
be
used
here
as
well,
so
anyway,
pretty
straightforward,
mostly
terminal
demo,
but
I
just
wanted
to
show
that,
like
it
was
a
fun
thing
to
try
out
I
I,
it's
under
my
that
this
code
is
technically
all
available
under
my
GitHub
and.
B
It's
not
in
any
way
polished
or
anything
yet,
but
this
one
I
will
actually
send
in
chat.
So
people
can.
Where
did
my
can't
find
my
bar
okay
there?
It
is.
B
Okay,
so
I
I
just
put
the
link
in
there
in
case
anyone
wants
to
see
the
code
and
what
it
looks
like,
but
yeah
I,
I
defined
it
right
now,
as
an
experimental
I.
Put
that
so
I
have
like
the
dot
ex
or
the
the
EXP
thing
in
the
middle
here.
Just
because,
like
I'm
not
sure
if
this
is
a
Terra
bad
idea,
but
having
it
be
like
a
thing
that
publishes
to
an
actor
just
when
it
wants
to,
it
seems
like
a
pretty
fun
idea.
A
B
That's
exactly
what
it
does.
If
you
look
at
the
code,
it
it,
it's
actually
really
straightforward
and
dumb,
but
it
just
says
like
on
every
tick
grab
all
my
linked
actors
and
then
send
to
All
actors
so
that
that's
all
it
does,
and
then
I
mean
it's
just
Gathering.
It's
using
a
nice
crate
that
I've
I've
known
about
for
a
while
and
then
sends
the
data
out.
So
it's
not
it's
real
dumb.
This
provider
is
146
lines,
but
you,
like
I,
said
you
can
make
this
a
lot
more
complex.
You
could
be.
B
You
could
decide
when
to
do
it
out
based
on
your
own
monitoring,
so
this
provider,
like
I,
said,
could
latch
on
to
Prometheus
and
then,
as
it
queries
the
data,
it
could
say,
like
oh
I've
hit
a
threshold
now
I'm
going
to
send
a
metrics
event
and
that
metrics
event
will
get
sent
out
to
whichever
actors
are
linked.
However,
you
want
to
do
it
and
then
that
actor
can
can
take
action
based
on
that
that
information.
So
that's
how
it'll
work
like
I
said
super
dumb,
but
it
works.
A
A
Taylor,
what
was
the,
why
did
you
build
this
like?
What
was
the
did?
You
have
something
in
mind
that
you
want
like?
Was
it
custom
bottom
scalers?
Was
it
like
a
personal
project?
No.
B
I'm
I'm
working
on
a
couple
you'll
see
a
couple
more
that
I
might
demo
next
week
and
I've
been
just
having
this
itch
to
have
kind
of
what
I've
been
calling
the
util
contracts
and
I
know
like
having
a
util
folder
is
like
where
you
throw
the
kitchen
sink
and
I'm
not
trying
to
do
that.
But
there's
just
a
couple
like
little
things
that
I
I
wish
we
had
around
one
of
them.
B
I'm
working
on
is
I
kind
of
want
to
do
actor
config,
which
should
probably
eventually
be
a
built-in
to
what
we
do
inside
of
wasn't
Cloud,
but
I'm
going
to
do
like
an
actor
config
one
and
the
actor
config
one
can
like
I
was
planning
on
implementing
one
using
Nats
KB
and
one
using
hash,
Court,
Vault
and
then
I'm
hoping
to
actually
knock
that
out
today.
It
won't
be
fully
production,
ready
or
tested,
but
it
will
work
probably,
and
you
can
use
that
to
fetch
data
I'm,
also
working
on
Brooks,
already
I
think
demo.
B
A
single,
a
single
one
because
and
I
was
trying
to
see
if
I
could
come
up
with
a
distributed,
one
like
a
cron
provider
that
could
do
it
all
across
I'm,
just
trying
to
create
a
bunch
of
utility
contracts,
and
this
is
one
that
was
out
there
and
honestly
I
was
like
I
just
want
to
do
this
for
fun
to
see.
If
I
could
do
it,
but
then
I
ran
it
past.
B
People
like
Dan
who's,
our
infrastructure,
director
and
Lead
over
here
at
cosmonic,
and
he
was
like
I
want
this
and
I
was.
B
So
I
don't
know
like
if
it's
it
like
I,
said
I
know,
there's
other
tools
that
exist,
that
can
scrape
system,
metrics
and
whatnot,
but
I
think
this
one
could
be
interestingly
catered
to
was
in
Cloud
because
it
could
get
some
wasm
Cloud
specific
metrics
as
well
like
number
of
providers
running
on
the
host,
where
this
thing's
running
that
kind
of
thing.
C
A
Yeah
I
found
that
found
that
Chrome
provider
I
think
it's
still
in
a
it's
still
in
a
PR
right
now,
I
think
because
we
were
whatever
for
some
for
some
other
reasons,
but
it's
it.
It's
definitely
not
designed
to
run
more
than
one
unless
you
like
a
lot
of
ticks
well.
Thank
you.
Taylor.
Anybody
have
any
questions
for
Taylor
on
the
demo.
A
There
it's
fine,
but
we
will
go
on
to
our
next
agenda
item
which
the
rest
of
the
call
we
have
booked
just
kind
of
earmarked
for
talking
about
link
names,
and
so
you
know
whether
or
not
we
use
the
whole
time.
A
A
Names
are
actually
present
in
each
wasum
Cloud
application,
whether
you
configure
them,
whether
you
know
you're,
configuring
them
or
not
so,
and
you
may
have,
if
you're
in
the
wasm
cloud,
Community
you've
built
something
with
wasm
clouds,
and
you
do
know
what
a
link
definition
is,
which
is
a
runtime
configured
set
of
configuration
between
a
specific
actor
and
a
capability
provider
and
Link
definitions
do
two
things.
It
essentially
allows
an
actor
to
talk
to
a
capability
provider
it
allow.
A
A
Now,
whenever
you
create
a
link
definition,
you
need
a
few
pieces
of
information
which
you
may
know
of
the
actor
public
key
and
the
capability
provider
public
key,
so
they
can
talk
to
each
other,
the
contract
ID.
So
that
wasn't
Cloud
knows
what
types
of
operations
are
going
to
be
allowed
between
those
two
like
HTTP
server
or
key
value,
and
then
what's
actually
always
present,
is
the
link
name
of
the
capability
provider
and
actually
don't
know
if
I
have
a
awesome,
Cloud
thing
running
right
now,
but
I'll
pull
up
a
waslam
dashboard.
C
A
Name,
so
let
me
just
move
things
around
when
we
start
something
like
the
HTTP
server
capability
provider.
You
know
you
may
enter
the
oci
reference,
whether
you're
doing
it
here
or
through
the
control
interface
and
just
hit
submit,
but
you'll
notice.
This
linked
name
is
always
here,
and
it's
just
default
that
that
is
the
default
case
to
which
honestly
works
for
about
90
of
the
cases.
A
So
what
I
want
to
talk
about
is
just
Why
link
names
exist
and
when
you
may
use
different
link
names
and
some
of
the
things
that
I
want
to
talk
about
from,
there
are
the
examples
that
linked
names
aren't
really
robust
for
yet
and
some
of
the
discussions
that
have
popped
up
either
in
issues
or
in
GitHub
or
in
slack.
So
if
anybody
has
any
questions
as
I'm
going
along
or
has
comments
that
they
want
to
share,
please
feel
free
to
just
jump
in
like
raise
your
hand,
but
I'll
go
from
from
here.
A
D
Oh
sorry,
I
thought,
that's
what
you
were
asking
yeah
well,
yeah
I'll
tell
you
my
impression
of
this.
I
I
was
a
bit
sort
of
violated
the
principles
of
least
astonishment
for
me.
If
you
look
at
the
the
washboard,
the
the
link
table
is
is
suggests
that
it's
indexed
by
link
name,
it's
the
column
on
the
left
right,
and
so
it
looks
like
actors
and
providers
would
key
on
linkname
to
to
to
connect
right.
That's
how
this
looks.
D
That's
that's
the
impression
that
it
gives,
but
it
turns
out
that
that's
not
the
case,
and
you
can
just
ignore
the
link
name
and
leave
it
default
for
everything.
I
named
all
of
mine,
because
it
helps
to
understand
what's
going
on
right,
but
those
names
were
Superfluous.
So.
D
A
Okay,
yeah
I
think
that
that
makes
sense
and
I
think
that
there's
you
know
a
certain
amount
of
if
it
works
for
the
90
case,
then
maybe
link
name
shouldn't
be
exposed
as
a
as
an
early
option.
You
know
being
able
to
change
it
to
something
else
and
for
things
like
HTTP
server,
it
actually
doesn't
matter,
but
for
other
capability
providers.
It
could
matter
to
the
point
where
your
application
may
not
work
if
you
supply
your
own
custom
link,
name
and
so
I'll
talk
a
little
bit
about
that
as
well.
A
So
while
while
Vince
was
talking
just
to
get
everything
shown
up
on
the
dashboard,
I
set
up
the
the
classic
Echo
app
and
when
I
linked
the
actor
to
the
capability
provider,
it
was
on
the
link
name
default
now.
The
reason
why
the
the
base
reason
for
for
link
definitions
existing
is
so
that
or
for
link
names
existing
and
so
that
you
can
run
multiple
copies
of
the
same
capability
provider
and
provide
either
different
configuration.
You
know,
provide
isolation
or
just
have
a
separate
instance
of
that
capability
provider.
A
So,
for
example,
here
we
have
Echo
running
with
HTTP
server
default
if
I
hit
localhost
8080
I'll
get
the
regular
Echo
payload,
which
is
just
the
same
response
or
the
same
request
that
I
sent.
If
I
start
the
HTTP
server
provider
on
something
other
than
default,
we
can
call
it
something
like
inbound,
of
course,
they're
all
inbound.
Maybe
that's
a
bad
example,
we'll
just
call
it
demo
just
to
create
a
different
one.
A
A
So,
instead
of
instead
of
having
just
one
where
we're
listening
on
8080,
we
can
also
set
up
localhost
8081,
so
we
basically
have
two
endpoints
and
the
request
is
the
response
is
going
to
be
the
same.
A
So
if
you
have
one
HTTP
server,
that's
serving
there's
actually
a
new
feature
that
just
got
added
to
our
HTTP
server
around
making
it
so
that
you
can
only
do
read-only
requests
so
git
and
head
I
think
you
could
have
a
different
link
name
for
HTTP
server
like
read-only,
and
then
you
get
a
link
to
the
echoactor
and
then
essentially
have
it
wouldn't
be
able
to
respond
to
post
and
put
and
all
those
types
of
requests.
A
So
there's
the
flexibility
to
Define
different
configuration
and
then
there's
also
isolation,
which
is
actually
mentioned
in
the
link
definition
documentation,
which
I
can
go
ahead
and
send
in
the
chat.
This
is
the
example
that
we
have
in
terms
of
an
outbound
capability
provider,
so
I
call
HTTP
server
and
inbound
capability
provider,
because
it
calls
an
actor.
Outbound
would
be
something
like
key
value,
because
the
actor
is
calling.
A
The
key
value
provider
in
this
example
for
two
different
providers
is
that
we
have
a
redis
capability
provider
with
the
link
name
default
and
then
some
kind
of
low
latency
in-memory
cache
provider
running
with
the
link
name
cache,
and
when
you
make
a
call
to
a
capability
provider
in
in
the
actor
sense,
you
can
only
have
one
link
per
contract
per
link,
name,
and
that
sounds
a
little
complicated.
A
But
really
what
it
means
is
that,
if
you're
making
a
call
to
a
key
value
storm,
you
have
to
differentiate
which
one,
if
you
have
multiple,
so
you
can
have
default
in
Cache
or
the
one
that
we
bring
up
often
is
maybe
secret
with
the
hashicorp
vault
key
value
storm.
So
you
can
differentiate.
Those
requests.
One
example
that
we've
used
before
for
Nats
message.
Isolation
is
to
have
one
that's
inbound
and
one
that's
outbound,
so
you're
using
two
different
Nats
connections.
A
For
that,
and
that
is
kind
of
an
overview
around
kind
of
the
name
spacing
that
link
name
lets
you
do
so
you
can
create
that
kind
of
flexible
configuration
now.
This
is
good.
I
think
that
this
covers
maybe
95
of
the
use
cases
that
we
have.
But
we
have
two
discussion
points
that
came
up
recently
around
linked
names
which
one
this
one
is
in.
The
bottom
repository
actually
around
an
issue
that
we
have.
A
If
you
have
one
actor
and
a
provider
and
two
different
applications,
try
to
Define
overwriting,
link
names
or
overwriting
link
definitions.
Then
you
know
we
we
shouldn't
just
continually
fight
over
that
and
we've
gone
a
little
back
and
forth,
but
Lee
had
a
great
Point
here
on
their
use
case
around
holding
hosting
multiple
apps
on
a
single
lattice.
A
So
you
can't
have
different
configuration
for
the
same
actor,
the
same
provider
without
using
link
names
and
what
this
ended
up
coming
down
to
is
that
when
you
create
a
different
or
when
you're
interacting
with
that
provider,
you
have
to
use
a
special
link.
Name
they're,
not
some
something
that
wasn't
Cloud
discovers
at
runtime
dynamically,
so
discovering
at
runtime.
A
Dynamically
is
is
a
big
one,
because
we
generally
like
to
have
that
those
things
configuration
runtime
config,
be
something
that
you
can
do
at
runtime,
but
for
actors
it
is
actually
hard-coded
in
the
actor
which
link
names,
you're,
interacting
with
so
there's
the
issue
of
discoverability,
and
then
there's
also
the
issue.
If
we
take
a
look
at
one
of
the
other
awesome
Cloud
examples
of
reusing
the
same
actor
with
different
configuration
with
the
same
provider
and
the
one
that
always
comes
to
mind
for
this
one
for
me
is
Blobby.
A
Blobby
is
just
a
simple
fire,
a
reusable
simple
file,
server
actor.
It
interacts
with
a
blob
store,
so
you
can
basically
use
it
link
it
up
with
a
blob
store
Define
where
your
objects
are
going
to
live
and
then
Blobby
has
the
create,
read,
update,
delete
operations
automatically
included,
so
you
could
feasibly
have
the
Blobby
actor
and
a
blob
store
that's
running,
maybe
with
your
local
file
system.
A
You
know
you
may
want
to
host
two
folders
in
that
local
file
system
or
something
like
that
and
being
able
to
do
that
with
LinkedIn
starts
to
get
a
little
bit
complicated,
so
I
kind
of
wanted
to
lay
out
a
couple
of
different
examples
and
I.
Don't
know
if
everybody
who
took
part
in
the
conversation
is
here
on
slack
I
think
Lee
may
not
have
been
able
to
make
it,
which
is
okay,
we'll
be
able
to
send
them
a
summary
and
everything.
A
But
I
know
that
Vance
offered
a
really
good
a
really
good
perspective
in
there.
I
actually
think
you
sent
the
last
message
so
I'll
alternative
to
advance
to
share
some
some
of
his
thoughts.
D
Yeah
I
wanted
to
see
the
code.
You
you
provided
the
code,
you
you
showed
where
it's
supposed
to
be
statically
defined.
A
C
D
Yeah,
sorry
I
guess
that
was
it.
So
if
that
were
it,
the
argument
to
that
function
is
a
static
string.
But
if
it
were
a
variable,
then
it's
not
static
anymore.
Is
it
right
so
I'm
I'm
I'm
I'm
I'm
I'm,
calling
in
to
question
your
statement
so
I
I
it.
It
doesn't
seem
to
me
that
there's
anything
preventing
me
from
having
an
actor
that
is
dynamically
choosing
those
linked
names.
Maybe
it
just
picks
a
random
one
and
not
that
that
would
work.
But
but
that's
true,
is
it
not.
A
Yeah,
the
the
key
thing:
if
you
have
this
as
a
variable,
something
that
you
could
pass
in
at
runtime
is
how
you're
actually
passing
that
in
at
runtime
to
the
actor.
If
it
were
on
a-
and
there
was
one
example
that
we
talked
about
in
slack,
if
we
could
actually
put
that
on
the
link
definition,
but
there's
no
real
way
to
give
an
actor
configuration
with
wasm
Cloud
each
one
is
treated.
There's
no
startup
function,
there's
nothing
that
you
know.
A
C
A
You
definitely
could
do
that
right.
D
A
It's
not
static,
it's
not
100
static,
but
it
does
still
require
that
so
the
in
this
example,
sorry
here
where
you
create
a
key
value
sender
with
a
link
and
a
link
name
if
you
get
that
at
runtime,
so
you
just
let's
say
it's
an
HTTP,
it's
getting
an
HTTP
request
and
it's
interacting
with
a
key
value
store
and
as
as
a
part
of
that
HTTP
request,
you
give
it.
You
know,
hey
talk
to
this
link
or
talk
over
this
link
name,
you
can
create
a
dynamic
link
using
that.
A
So
you
use
the
link
name
that
you
get
in
on
the
request,
but
you
do
still
have
to
have.
Is
that
link
to
the
running
capability
provider?
And
if
it
didn't
exist,
then
you
would
time
out
or
you
would
fail,
the
request
at
runtime,
which
you
know
all
of
this
just
comes
down
to
yes,
it's
possible,
but
you
would
have
to
make
sure
that
everything
is
set
up
correctly
at
runtime
it
just
kind
of
creates.
B
Yeah
the
big
thing
here
when,
when
Brooks
was
saying
static,
I,
don't
think
I
think
we
want
to
be
careful
on
like
we're
not
meaning
static
in
the
programming
sense
of
the
word
like
the
in,
like
the
code
sense,
the
word
you
should
say
I'm,
saying
that,
like
anything
that
you
have
to
choose
from
in
the
act
or
any
of
the
linked
names
have
to
be
in
the
actor
when
it's
compiled,
there's
no
way
right
now
to
get
that
information
outside
of
doing
something
with
a
a
contract
and
a
provider.
B
For
that
reason,
because
we
want
to
be
able
to
say,
like
hey
here-
is
a
like
here's,
the
config
for
my
actor
I
want
you
using
this
link,
name
for
this,
for
this
contract
or
whoever
we
want
to
Define
it,
and
that's
like
that's
the
difference
here.
Is
that,
like
we're,
we're
not
talking
about
just
like
in
a
programming
sense
statically
compiled
we're
talking
about
that
like
it
has
to
be
in
the
actor
somewhere.
B
So,
even
if
it's
technically
a
variable
when
you
pass
it
in
that
variable
has
to
be
populated
with
information
that
is
statically
compiled
into
it
somewhere
like
whether
that's
a
list
of
options
you
provide
at
the
top
or
even
if
you
randomly
generate
a
list,
those
things
have
to
be
be
put
together,
like
in
the
actor
without
like
any
outside
input,
and
so
that's
the
that's.
D
B
D
I,
don't
see
that
limitation,
so
what
am
I,
not
understanding,
so
if,
if
the,
if,
if
I
receive
a
call
like
from
from
the
HTTP
server
and
I,
get
I
get
the
name
of
the
link
and
I
and
I
use
that
name
now,
I
have
dynamically
chosen
a
value
for
that.
For
that
link.
How
is
that?
D
What's?
What's
stopping
that
from
from
happening?
So
if
these,
if
these
link
definitions
are
made
after
the
actor
is
started
right,
and
that
would
mean
that
you
could,
you
could
feed
the
information
into
the
actor
and
the
actor
could
now
know
about
the
new
link
and
now
use
the
new
link.
Is
that
not
the
case.
B
No,
it
isn't
so
links
the
only
things
that
actually
receive
the
link.
Definition
are
providers
and
the
system
itself,
like
the
hosts
actors,
have
no
idea
what
they're
so
receiving
it
actually
doesn't
matter.
We
use
that
trick
all
the
time
and
receiving
it.
So
like
an
HTTP
server,
you
can
use
any
link,
name
and
it'll
still
work,
because
the
link
name
doesn't
actually
matter
there,
because
all
it's
doing
is
saying
like
it's
just
linking
it
to
a
specific
provider
that
started
with
that
linked
name.
B
That's
all
that
matters,
but
to
the
actor
it
receives
it,
no
matter
what
the
link
name
is.
So
we
do
that
to
sometimes
split
between
like
two
different
HTTP
servers,
but
the
when
it
tries
to
call
something
so
if
an
actor
tries
to
call
a
KV
store,
for
example,
that
link
name
actually
matters,
and
so
the
Brooks
it
might
be
easier
to
show
this.
Can
you
take
one
of
like
the
KV
counter
example
and
try
linking
it
to
like
Fubar
instead
and
show
what
happens?
That's.
A
Funny
that
this
is
actually
exactly
what
I
was
getting
set
up
and
so
yeah
exactly
what
what
Taylor
is
talking
about.
An
actor
doesn't
know
anything
about
the
link
names
that
it
has.
What
you
know
at
runtime,
it
doesn't
know,
what's
successfully
been
established,
what's
there
and
what's
not
now
on
the
and
as
Taylor
said
when
you
said,
receive
invocation
this,
like
an
incoming
request
like
the
HTTP
server,
the
link
name
is
not
provided
to
the
actor
here.
A
Whether
that's
you
know
something,
that's
good
by
Design
or
as
Kevin
put
it
on
his
to-do
list
for
about
two
years.
Is
it's
kind
of
we'll
kind
of
leave
that
one
up
to
you?
I
think
it
would
be
better
to
have
that
information
than
not?
Would
you
receive
it
on
whatever
linked
name
just.
D
So
just
so
we
don't
we're
not
we're
not
confused.
Here.
I
wasn't
suggesting
that
in
my
example,
I'm
saying
that
I
would
look
inside
like
the
Json
payload
yeah.
C
A
D
D
Okay,
got
it
right,
but
but
the
other
way,
if
I
link
to
to
to
a
KV
store,
for
instance,
and
the
actor's
making
that
call
and
and
it
could
then
dynamically
be
choosing
new
links,
a
link
that
was
only
created
a
few
moments
ago
after
the
actor
was
started
correct,
but.
A
D
A
The
the
problem
is
that
the
actor
would
have
to
be
explicitly
designed
to
do
that,
and
that's
that's.
It's
not
necessarily
a
bad
thing,
but
it
doesn't
feel
as
good
as
in
terms
of
just
interacting
with
business
logic,
to
use
that
Dynamic
link
there's
no
way
of
knowing.
If
that
request
would
succeed
at
runtime.
D
Well,
we
it's
important
to
understand
what
the
capabilities
are.
So
look.
There
is
a
question
and
I
guess:
I
raised
it
of
whether
or
not
a
discovery
mechanism
is
missing
or
or
could
be
a
beneficial
addition.
But
you
know
the
thing
about
Discovery,
registration
and
Discovery
is
that
it
happens
at
different
levels.
So
you
you
can't
stop
people
implementing
things
from
having
their
own
application,
Level,
Discovery
and
registration,
Etc
and
so
I
could
use
the
the
things
I
just
described
to
accomplish
that.
D
A
Yeah
I
think
that
some
kind
of
configuration
discovery
that
goes
outside
of
the
normal
bounds
of
Link
names
and
Link
definitions
would
likely
over
complicate
this
scenario
right.
It's
like
some
kind
of
service
Discovery
is
is
out
of
scope
for
what
wasn't
like
that
kind
of
service,
Discovery
I,
think,
is
out
of
scope
for
what
wasmcloud
wants
to
do
here.
C
A
So
the
specific
scenario
that
that
Taylor
asked
me
to
show
here
was,
if
we
have
the
key
value.
Counteractor
that
interacts
with
it
receives
HTTP
requests,
interacts
with
the
key
Value
Store.
A
Then
we
know
the
capability
claims
of
the
KV
counter
is
HTTP
server
and
key
value,
and
so
we
set
up
two
different.
We
set
up
an
HTTP
server
on
demo
and
then
a
key
value
store
and
we
call
it
something
like
additional.
We
set
up
our
own
link
name
if
we
link
the
KB
counteractor
to
redis
on
key
value
and
additional.
That's
the
other
link
name
that
we
set
up
and
we
say,
go
talk
to
redis
at
the
normal
endpoint.
A
We
go
and
hit
that
after
it's
not
going
to
work
and
that's
because
it's
set
up
in
you
know
in
code
compiled
to
work
with
the
default
link
name,
which
is
just
what
we
do
as
a
default
with
a
key
value.
Sender
new
and
it
may
actually
pull
up
the
correct
thing
here.
I'm
not
sure
I
was
gonna.
A
Just
take
a
look
at
the
code
and
show
that,
under
the
hood
is
just
creating
a
new
one,
with
a
link
key
value
now,
I
think
what
essentially,
what
something
that
advances
is
getting
to.
A
Is
that
if
you
had
this
requirement
this
issue,
where
you
need
to
be
able
to
find
something
like
this
at
runtime,
then,
instead
of
creating
a
key
value
sender
with
new,
you
create
a
new
one,
with
a
link
that
you
receive
on
an
HTTP
request
and
then,
when
we
hit
this
endpoint
to
send
the
HTTP
request
to
increment
the
value
in
the
counter,
we
also
send
along
the
Json
payload.
You
know,
link
name
additional
fence.
Is
that
that's
right?
A
B
I
do
want
to
ask
Vance
I'm
just
trying
to
make
sure
I
understand
that,
because
I
think
you're
approaching
this
in
a
different
way
than
I
think
we've
heard
before
so
I
want
to
make
sure
that
we're
understanding
so
like
what
is
the
I
guess,
I'd
phrase
it
as
like:
what
are
the
shortcomings
you're
having
right
now,
if
any
with
how
the
link
definitions
work
and
like
how
is
the
like?
B
What's
the
use
case,
you're
you're
using
for
them,
and
maybe
like
we
talked
about
in
terms
of
use
cases
when
we
understand
a
bit
better
because
I
think,
like
I,
said
like
you're
you're
approaching
it
a
different
way
than
we've
seen
it
happen
before,
which
is
not
a
bad
thing.
I
think
that's
a
good
thing.
So
I
want
to
kind
of
hear
if
like
what?
What
are
the
shortcomings?
What
are
the
problems
you're
running
into
here
and
maybe
work
from
those.
D
Okay,
yeah,
the
the
that's
an
interesting
question,
an
interesting
perspective
and
I
guess
my.
What
I
need
you
to
understand
is
my
brain.
Doesn't
work
that
way
so
I,
don't
I'm,
not
use
case
driven,
I'm
capability
driven
and
so
the
I
I
I
understand
what
sort
of
use
cases
would
be
enabled
by
a
capability,
and
so
my
my
my
main
interest
is
usually
capabilities
and
use
cases
are
a
side
effect,
so
so
I
I
don't
have
a
a
particular
problem.
B
No,
no
I,
don't
I,
don't
think
you're
doing
things
I
wanted
to
make
sure
we
had.
We
had
answered
those
properly
because
I
wasn't
sure
if
we
had
and
that's
the
whole
point
of
this
discussion
to
make
sure
we
have
those
I.
B
Perfect,
okay:
Rex
did
you
have?
What
was
your
next
point
along
the
like
what
you
wanted
to
talk
through
with
this
yeah.
A
I
wanted
to
raise
one
more
thing.
This
was
kind
of
on.
Let
me
find
the
correct
the
correct
thing.
This
was
along
the
same
lines
of
if
you're
not
doing
that.
Dynamic
link,
name
creation
at
runtime,
so
you're
not
getting
that
information
from
a
request
and
you
are
compiling
in
different
link
names.
Let's
say
you
have
a
a
strict
set
of
capability
link
names
that
you
need
to
use
so
we'll
use
the
we'll
use
the
example
from
the
documentation
around
having
a
key
value
link
for
default
and
for
cash.
A
So
you
have
something
you're
doing
regular
key
value
operations
in,
and
then
you
have
a
different
capability
that
you're
storing
the
lightweight
cache.
There's
no
way
to
know
from
just
like
we
we
do
know
from
the
capability
standpoint.
There's
no
way
to
know
what
capability
links
an
actor
is
going
to
need
from
the
claims.
Something
that's
discoverable
and
here
I'll
make
this
a
little
more
smaller
something
that's
discoverable
at
like
before.
You
actually
run
the
actor.
So
when
you
run
KV
counter
and
you
it
says
the
capabilities,
HTTP
server
and
KB
store.
A
The
assumption
is
that
things
are
linked
on
default,
but
there's
actually
nothing.
That
is
really
saying
that
that
needs
to
happen
now
again
for
inbound
and
outbound
or
receiving,
and
then
sending
requests
on
your
own
HTTP
server
doesn't
have
this
restriction,
but
KB
store
does
so
if
I
take
a
module
and
I
like
KV
counter
and
I
distribute
it.
If
I
don't
use
the
link
name
default
that
could
be
really
potentially
confusing
to
the
person
running
the
actor.
It
would
just
need
to
be
documented,
but
I
think
Jordan
brought
up.
A
Well,
he
wrote
Magic
Pony,
but
I
assume
it
was
something
else
when
he's
actually
coding
and
then
an
HTTP
client
link
on
default,
there's
just
not
a
way
to
discover
this
at
runtime
so
or
you
know
before
before
runtime.
So
the
next
point
of
discussion
around
link
names
is
really.
If
you
have
this
strict
set.
A
We
know
this
when
we
compiled
the
actor.
Unless
it's
something
that's
dynamically
defined,
what
links
or
what
link
names
it
requires.
So
should
that
be
present
in
the
claims.
C
Hey
quick
question
on
the
on
the
topic
of
the
same
discussion
in
advance:
is
it
not
possible
to
have
a
continuous
polling
of
those
names
and
just
have
them
available
when
not
any
you
know
any
need
arrives
to
you
know
to
realize
that
or
this
is
not
even
not
have
been
a
topic.
A
Yeah,
but
that
I
mean
that's
kind
of
I.
Think
at
the
core
of
the
this
kind
of
the
core
of
this
discussion
is
that
you
know
actors
shouldn't
have
to
deal
with
complex
logic
around
what
links
exist
or
don't
exist.
They
should
just
be
written
in
terms
of
business.
Logic
of
I
want
to
store
things
in
this
key
value,
store
on
a
link
that
I
kind
of
know
about
foreign
interesting.
A
So
for
for
the
other
question
around
discoverable
link
names,
we
did
get
Jordan
in
the
chat
who
said
yes,
which
of
course
but
I-
think
it's
Jordan
feel
free
to.
If
you
wanted
to
add
a
little
more
detail,
there
I'm
not
sure
if
you're,
in
a
point
where
you
can't
talk.
A
Okay,
so
around
the
different
link
names
and
making
it
so
that
you
could
see
those
in
the.
E
Clients
yeah,
so
my
my
use,
my
point
was
I
can
write
an
actor
right
and
I
can
take
maine.go
or
whatever.rust
and
I
can
take
a
hash
of
it
and
say
the
hash
is
like
three
I
can
compile
it
with
different
link
names
at
compile
time
where
technically
the
file
hasn't
changed.
Therefore,
I
would
argue
that
the
oci
reference
is
the
exact
same.
So
let's
say
I
decompilot
that
how
do
I
know
version
zero,
zero
one
and
zero
zero,
two
which
in
my
head
means
hey
it's
not
breaking.
E
But
it
is
breaking,
but
then
I
get
to
the
point
where
what
what,
if
I
need
multiple
link
names
in
the
actors
and
I
don't
want
to
put
like
30
of
them
in
there
all
at
once.
Right
so
am
I
doing
like
oci
reference
actor
one
and
then
zero
zero
one
oci
reference
actor,
two:
zero
zero!
Two
one
am
I.
Now
maintaining
oci
references
based
on
link
names.
E
It
just
gets
impossible
yeah.
So.
E
My
exact
use
case
was
go
to
link
hit
Echo
as
the
actor
hit
provider,
so
we
shouldn't
even
see
KB
redis
there
right
I
mean
in
the
washboard
it's
okay,
because
I
mean
this
is
a
Dev
board.
But
when
we
start
to
build
under
some
builds
like
next-gen
washboard,
we
we
shouldn't
even
see
KB
redis
there
and
then,
instead
of
like
a
link
name
default,
we
should
have
a
drop
down
there
because
we
know
that
that
actor,
it's
defined.
E
So
that's
kind
of
what
my
use
case
was
because
I
was
trying
to
do
exactly
that
filter
on
the
providers
it
can
and
the
links
it
has
the
capability
to,
and
then
we
take
out
the
entire
necessity
for
that
contract.
Id
block
right
because
we
know
the
provider,
we
know
the
link.
Therefore
we
know
the
we
know
the
contract
so
I
understand
that
not
being
a
function
of
the
washboard,
because
this
is
like
as
explicit
as
you
can
be,
but
it
kind
of
limits
any
like
future
feature.
E
Development
on,
say,
like
a
front-end
type
thing
and
also
I,
don't
want
my
my
oci
tags
to
be
like
zero
zero.
One
underscore
link
default,
underscore
link
Foo,
underscore
Link
Bar,
which
admittedly
I
do
have
out
there,
because
the
actors
are
literally
the
exact
same.
They
just
have
different
link
names
compiled
in.
D
A
All
so
Jordan
I'm
probably
going
to
ask
you
a
hard
question
here.
Would
you
need
to
have
those
different
actors
with
different
link
names
compiled
in?
If
you
discover,
if
you
got
those
on
an
incoming
request
or
something
like
you
did,
that
dynamically,
like
you,
just
had
a
variable
that
came
in
on
the
incoming
request
that
you
used.
E
I
mean
maybe,
but
then,
if
it's
like
an
HTTP
request
now,
I
have
to
write
front
ends
that
include
this
random
header.
That's
just
like
it.
That
now
becomes
a
wasm
cloudism
and
my
and
and
I
want
to
avoid
wasm
cloudisms
right,
because
I
want
to
be
able
to
take.
My
UI
have
been
an
actor
and
then
still
be
able
to
put
it
on
a
I.
Don't
know
an
ec2
somewhere
and
not
have
like
wasmcloud
isms
baked
in.
A
A
A
A
This
really
gets
down
to
what
we
want
to
isolate
within
a
single
lattice,
but
if
I
linked
key
value
counter
to
the
demo
link,
name,
HTTP
server
and
that's
all
fine,
then
maybe
I
could
pull
the
link
name
off
of
that
incoming
request,
basically
from
the
wasabus
context,
and
then
so
I
talked
to
the
key
value
store.
You
know,
demo,
link,
name,
I,
think
this
is
a
pretty
interesting
proposal.
A
I
think
it's
something
that
you
could
well
I'll
start
in
two
parts.
I
think
we
should
include
the
link
name
on
the
incoming
request,
because
it's
it
just
feels
like
more
information
that
that
doesn't
harm
to
have
it
there,
but
then
also
I.
Think
that
this
would
you
know
you
may
have
more
than
one
link
name
that
you
use
in
response
to
getting
a
request,
and
so
it's
kind
of
a
slippery
slope
using
this
as
a
as
a
way
to
to
work.
E
Yeah,
but
after
doctor
calls
break
that
if
the
originating
thing
comes
from
a
different
actor,
for
example
right
so
right.
So
if
it's
coming
in
on
the
right
from
a
provider
right
that
makes
sense,
because
we
know
what
linky
came
over.
But
if
it
comes
over
a
provider
to
an
actor
that
actor
makes
a
decision
sends
it
to
another
actor
and
then
that
actor
sends
it
out
to
a
separate
provider
like.
E
Are
we
just
gonna
start
daisy
chaining
it
all,
because
I
I
do
have
that
example
too,
where
yeah,
where
I
don't
know
that
yeah
I
literally
have
no
idea
how
that
would
have
that?
How
how
how
I
would
dynamically
allocate
a
link
name
at
that
point?.
A
Yeah,
it's
a
good
point
because
you
would
have
to
just
basically
pass
that
around
or
pass
your
own.
Some
kind
of
value,
yeah,
so
I
know
that
we're
kind
of
approaching
the
end
of
the
hour.
So
I
want
to
try
to
to
wrap
this
discussion
up.
Thank
you
so
much
Jordan
and
Vance
and
Maxim
and
Taylor.
You
know
everybody
for
hopping
in
on
this
and
talking.
You
know
whether
it's
around
the
specific
use
case
or
what
you
could
do
from
a
capability
perspective.
A
It's
it's
really
important
around
link
names
and
how
we
may
want
to
improve
them.
I
think,
there's
still
some
things
to
iron
out
around
dynamically
finding
link
names
or
dynamically.
Getting
that
information
at
runtime
that
feels
very
specific
and-
and
we
there's
an
example.
A
I
know
that
we've
used
it
for
around
a
lattice
controller
being
able
to
get
the
lattice
ID
on
the
request,
and
then
that's
how
you
you
can
work
with
it,
but
the
the
only
I
think
there's
more
discussion
to
to
be
had
there
or
maybe
a
really
concrete
use
case
where
this
is
something
that
we
need
something
special
built
for
Linked
names.
What
I
do
think
we
should
do
and
I
don't
know
if
this
is
something
that
would
be
an
extra
tag
or
something
that's
core
to
the
capability
claims
themselves.
A
I
think
it
would
be
great
to
include
the
link
name,
information
in
the
claims
and
I
think
if
it's
something
that's
dynamically
configured
or
maybe
by
default.
We
have
like
a
little
like
link,
name
star
or
something
like
that.
Like
the
that's
the
status
quo,
or
if
you
don't
specify
it,
then
it's
just
default
or
if
you
don't
specify
it,
it
could
be
all
of
them,
not
sure
which
one
that
would
probably
come
out
in
an
RFC
but
I.
A
Think
if
you
are
using
an
actor,
you
should
be
able
to
know
what
links
it
needs
to
be
able
to
succeed,
run,
do
its
do
its
job
without
having
to
go
inspect
the
code,
you
should
be
able
to
know
that
from
the
module
or
from
the
claims
itself,
so
that
will
have
to
be
added
in
you
know,
there's
a
couple
different
places,
but
I
think
that
that
would
be
a
certainly
nice
quality
of
life
Improvement.
A
In
order
to
make
that
work,
Jordan
I
think
I
may
need
to
follow
up
with
you
to
see.
If
that
would
even
work
for
your
use
case,
because
you
might
need
to
have
you
know
the
exact
same
actor
with
different
link
names,
yeah
closed
Source
actors
where
you
distribute
just
the
webassembly
module
would
fall
under
this
specifically.
C
B
You
even
Blobby
falls
under
that
like
the
thing
is.
Blobby
is
a
perfect
example.
We
want
that
to
be
able
to
you,
can
plug
it
into
two
different
back
ends
and
that's
a
different
linked
name
and
so
to
be
able
to
talk
to
One
S3
like
this
at
the
S3
blob
store
versus.
You
know
like
something
that
encrypts
it
or
a
local
one.
You
have
to
have
two
different
link
names
and
so
like
being
able
to
know
like
what
the
link
name
should
be
for
your
configuration,
for
the
actor
is
pretty
important.
A
Okay,
well,
definitely
definitely
more
to
consider
there
we'll
kind
of
follow
up
with
the
notes
in
the
community.
A
It
will
follow
up
with
the
notes
in
the
community
call
around
how
we
can
proceed
in
adding
that
kind
of
things
and
the
claims
and
then
I
think
we
can
keep
using
Blobby
is
a
great
example.
What,
if
I
wanted
to
use
bloppy
with
two
different
blob
stores
and
then
go
from
there.
B
We
did
have
a
question:
that's
a
pretty
quick
hitter
Brooks
before
we
tie
up
I
want
to
yeah
sure
you
got
it.
Yeah,
someone
from
the
from
one
of
the
stream
sources
asked
about.
Do.
I
need
to
use
rest
to
do.
I
need
no
rest
use
little
Blobby
tables
I'm.
Glad
you're,
using
its
proper
name,
sounds
like
a
nice
utility
to
sync
a
DB
table
to
a
bucket.
It
is
meant
to
be
extremely
flexible
and
you
don't
need
to
need
to
know
rest
to
use
it.
B
So
that's
you
can
just
link
it
in
that's.
Actually.
The
whole
kind
of
selling
point
is
that
the
business
logic
can.
B
Whatever
and
we're
even
getting
closer
there
Bailey's
been
demoing,
some
of
that
and
I
think
it's
going
to
be
demoing
some
more
in
coming
weeks
about
how
any
language
can
do
this,
but
it
doesn't
matter
what
the
language
is:
it'll
pull
them.
You
just
know
that
it
has
an
API
that
it
exposes,
and
then
you
can
call
it
just
like
anything
else.