►
From YouTube: Kubernetes WG IoT Edge 20221012
Description
October 12, 2022 meeting of the CNCF IoT Edge Working Group where edge native white paper draft was discussed.
A
Hi
welcome
to
the
October
12th
meeting
of
the
cncf
iot
edge
working
group
on
the
agenda.
Today,
we've
got
a
discussion
of
the
Edge
native
white
paper
that
we've
been
working
on
for
a
few
months
now
in
preparation
for
a
talk
to
be
delivered
at
kubecon,
North
America,
there's,
probably
room
for
additional
items.
A
If
somebody
wants
to
add
something
to
the
agenda
notes
or
just
verbally
speak
up
right
now
to
drive
that
white
paper
discussion
unless
somebody's
got
a
another
item
to
nominate
I'll
take
Kate
to
do
that,
because
I
think
she's,
the
one
most
invested
in
this,
giving
the
talk
in
a
couple
of
weeks.
B
Sure
yeah
happy
to
kick
us
off
I
think
it
would
be
nice
to
build
scope
on
this
meeting.
Maybe
what
we
want
to
try
and
Achieve
by
the
end
of
it
for
one
I
think
we
could
try
to
we've
been
doing
this
of
late
go
through
and
resolve
comments,
get
things
in
a
good
place
to
start
and
then
I
think
there's
also
a
big
chunk
that
still
the
main
thing
that
is
not
ready
is
our
last
principle
of
external
device
connectivity.
B
So
we
could
also
start
with
that,
make
sure
that's
in
text
form
and
then
from
there
kind
of
see
if
we
can
bring
things
together.
Another
thing
that
I
was
thinking
is
the
list
of
Open
Source
projects
deciding
if
that
feels
complete
and
deciding
how
to
handle
that,
and
then
the
final
thing
would
be
references.
B
The
things
that
are
on
my
mind,
the
final
thing
is
making
sure
our
references
are
complete
and
also
footnoted
throughout
the
paper
we
get
to
sign
someone
to
that
after
and
the
people
who
listed
themselves
as
writers
or
reviewers,
making
sure
that
they
are
involved
in
the
way
that
they
would
like
to
be.
B
So
those
are
some
things
that
I
was
thinking
we
could
address.
Are
there
other
things
that
people
think
might
be
good
to
do
during
the
session.
C
Ask
if,
for
everyone's
thoughts,
if
we
want
to
put
this
in
GitHub
transitioning
from
Google
into
markdown,
I
would
assume
yeah
good
point
and
then,
where
would
that
go.
B
Yes,
so
I
think
that
is
a
goal
of
ours
and
ideally
to
do
that
before
kubecon.
If
we
feel
like
we're
ready
for
an
official
draft
to
be
published,
that
would
go
so.
The
cncf
iot
edge
working
group
is
under
the
runtime
tag,
so
it
would
go
under
run
time
and
I
was
just
gonna
PR
or
one
of
us
can
PR
into
runtime
and
say,
create
a
new
folder
or
directory
called
white
papers
and
Link
from
our
landing
page.
There
I
can
I'll
share
a
link
to
that,
but
yeah.
B
Ideally
we
can
do
that,
whether
that's
this
working
group
meeting
after
this
one
or
Thanksgiving
or
the
next,
that
would
be
kind
of
the
place
that
I
would
see
it
going
and
then
I
think
we've
talked
about
I.
Think
Brandon's
mentioned
this
after,
like
after
our
Draft,
when
we
feel
like
we're
in
a
good
state.
Are
there
other
places
that
we
want
to
put
this
and
kind
of
expanding
that
conversation.
D
A
B
Yeah
I
think
we
definitely
can
do
a
practice
run
next
meeting
I
think
we
I
wasn't
quite
sure
at
what
level
we
wanted
do.
We
want
to
do
full
practice
run,
or
was
it
also
just
to
look
at
the
slides
again.
A
Yeah
I
think
we've
looked
at
the
slides
at
least
a
rough
draft
with
the
agenda
last
time
and
looked
good
I'd
suggest.
Maybe
we've
got
a
lot
of
things
already
on
that
list
that
maybe
aren't
going
to
come
together
until
just
before
the
next
meeting.
So
if
you're
not
ready,
I'd
say
postpone
it,
and
you
know
you
kind
of
have
to
get
some
of
those
things
ready
to
have
a
because
they'll
feed
into
your
deck
I
suspect.
A
So
do
it
only
if
you're
ready
and
if
you're
not
my
experience,
is
sometimes
you
don't
get
that
practice.
We're
on.
You
know
until
you're,
even
physically,
at
the
conference
in
the
speaker,
lounge
or
something,
and
if
it
happens,
it
happens.
But
since
you
have
two
speakers,
I
believe
you
you
kind
of
are
going
to
have
to
do
one
to
have.
It
come
off,
okay,
but
just
personal
opinion,
but
yeah.
B
Yeah
so
Omar,
just
hopped
on
the
question
from
Steve,
was
whether
to
have
a
practice
run
in
our
next
meeting
and
I'm
comfortable
with
that
we're
meeting
after
this
to
kind
of
look
through
our
slides
even
more
so
I'd
be
fine
with
that.
My
only
caveat
would
be
is
that
if
there's
another
topic
for
the
working
group,
we
can
make
it
a
one-off
with
a
smaller
group
I
think
sometimes
it's
hard
to
get
feedback
from
large
groups
but
yeah,
let's,
let's
tentatively
plan
on
that
for
the
next
meeting.
B
Unless
Amar
you
have
any
other
thoughts
on
that.
E
Is
this
keep
the
the
last
knitting?
If
the
sides
are
public
can
can
they
be
shared.
B
Yeah
I
think
that's
the
other
issues.
I'm
not
sure
you're
supposed
to
have
like
talk
like
on
a
recorded
like
I,
think
kubecon
wants
to
be
the
recorded
presentation,
and
so
that's
my
other
concern
with
doing
a
true
dry
run
in
a
recorded
meeting
so
yeah.
That
is
a
fair
thought.
A
Yeah,
we
could
always
just
pause
the
recording
too,
so
it's
not
there
I'm
expecting
you
know
if
you
literally
it's
the
first
time
you've
ever
done.
It
it'll,
be
so
rough
that
you
probably
don't
want
it
it's
circulation.
The
whole
point
of
it
is
to
you
know,
get
down
your
timing,
your
content
and
everything
and
very
few
people
get
it
right.
The
first
time.
A
Okay
well
anyway,
that
was
me.
I
I've
been
keeping
notes
in
this
of
what
the
agenda
is
for
this.
So
the
first
item
was
convert.
B
Yeah
I
guess
that
was
resolve
the
scope
for
today,
so
I
think
our
scope
is
maybe
let's
try
and
get
through
as
many
of
these
as
we
can,
and
we
did
resolve
scope
on
let's
present
next
time,
and
let's
do
that
after
the
end
of
the
meeting
and
stop
the
record
for
that
with
that
are
people
how
do
people
feel
about
moving
to
external
device
connectivity?
Looking
at
that.
B
E
I
I
just
want
to
say
to
say
that
this
bullet
point
is
just
my
outline
and
the
draft
of
the
thoughts
I
I
had
it
on
on
the
on
the
topic.
So
I
wanted
to
go
with
the
group
to
review
them
before
continuing
in
shaping
it,
but
that's
on
the
agenda
today.
So
that's
good.
B
Okay
and
I
think
part
of
this
is
we
originally
thought
we
were
going
to
have
a
paragraph
for
each
of
these,
and
now
it's
just
kind
of
larger
clumps,
so
inside
the
the
chart,
and
so
we
just
need
to
condense
all
this
into
three
sentences,
probably
yeah.
B
E
So
I
think
so
I
think
the
the
CNC
projects
in
the
space
are
are
all
already
covered
later
on
right.
So
so
we
can
already
removed
it.
B
So
if
you
were
to
kind
of
give
a
kind
of
elevator
pitch
on
what
this
is
about,
we're
trying
to
both
mention
the
issue
at
hand
of
there's
a
lot
of
sensors
and
then
recognize
that
you
need
to
deal
with
it
in
a
certain
way.
Is
that
kind
of.
E
So
then
examples
are
all
these
things
like
USB
cameras
or
or
local
radio,
sensors
or
devices,
or
all
of
that
require
us
to
to
have
a
special
consideration
for
workloads
running
on
the
edge
so
that
they
can,
for
example,
run
in
a
privileged
mode
or
with
raised
access
levels
to
to
be
able
to
to
access
those
resources
and
and
be
able
and
expose
those
resources
to
the
workload
so
containers
or
or
veggie
or
whatever.
We
we
use.
B
Okay,
and
so
my
follow-up
question
is
what
how
are
we
distinguishing
between
external
device,
connectivity
and
resource
aware,
since
we
have
both-
and
one
thing
Amar
noted-
is
that
we
have
some
groupings
going
on,
so
we
have
resource,
aware
external
device
connected
AV
and
wherever
and
aware
of
variable
availability
as
being
a
part
of
this
larger
idea
of
resource
and
device
awareness,
slash
monitoring,
so
maybe
I
for
starters-
can
I
move
this
resource
aware
grid
to
be
next
to
device
connectivity,
so
we
can
reconcile
those
next
to
each
other.
D
B
B
F
In
the
I
think
the
difference
is
stable.
We
have
that
grow
on
data
models,
etc,
etc.
So
my
point
was
which
I
mean
we
shouldn't
say
anything
different:
we
should
just
be
consistent,
I'm
actually
driving
so
I
can't
read
it,
but
that's
what
I
recall
yeah
I
think
I
put
that
in
okay.
B
I'm
gonna
take
this
out
and
make
it
a
note,
then
I
think
that's
a
good
Steve.
Maybe
you
can
add
that
to
the
to-do's
of
like
ensuring
having
editors,
ensure
that
we're
consistent
across
our
differences
and
similarities
tables
and
our
principles
tables.
B
I'm
gonna
just
put
it
in
here
Steve,
because
I
think
it's
easy.
If
we're
all
looking
at
it.
So
kind
of
the
summary
that
Steve
put
in
the
chat
is
external
device.
Connectivity
is
in
one
sentence
all
about
the
challenges
related
to
Edge
device
connectivity.
So
it's
getting
an
inventory
or
device
monitoring
health
of
device,
connect,
health
of
device
checks
or
health
checks
of
devices
and
connections
to
devices
update
management.
E
C
B
E
I
was
in
the
beginning
under
impression
that
resource
aware
is
more
related
to
the
actual
host,
the
the
the
resources
on
the
hosts
or
CPUs
memories,
storage
and
things
like
that
and
the
connectivity
is
then
then
the
the
thing
about
communicating
with
a
let's
say,
lead
devices
and
things
like
that:
yeah
I.
A
Think
it
is
I
think
in
general,
I
can't
see
how,
if
it's
an
if
this
is
specifically
external
devices
and
if
they're
external,
they
shouldn't
be
really
using
the
host
nodes,
memory
or
CPU.
A
If
the
connection
to
the
device
is
IP
based
with
some
of
your
orchestrators
and
container
runtimes,
you
have
a
semblance
of
establishing
limits
to
what
they
might
be
able
to
use.
But
if
these
devices
are
not
using,
you
know
traditional
IP
based
connectivity,
I,
don't
think
the
typical
container,
runtime
or
orchestrator
or
Os
will
do
anything
for
you.
So
there
isn't
necessarily
a
mechanism
with
existing
technology
to
monitor
or
impose
limits
on
resource
consumption
related
to
Communications.
A
We,
the
best
we
could
do,
is
indicate
it's
there.
I
don't
think
we
could
I
I,
don't
it's
a
falls
into
the
either
unsolved
problem,
space
or
out
of
scope,
so
whether
we
raise
Resource
Management
there
I
think
we
could
safely
leave
it
out
because
in
terms
of
best
practices
today,
I'm
not
sure
there
is
any
kind
of
recommendation.
We
could
do
other
than
observe
that
maybe
this
is
an
unsolved
problem.
B
You're
saying
the
unsolved
problem
is
rate,
limiting
the
number
of
connections
to
these
external
devices.
Yeah.
A
Well,
it
isn't
necessarily
even
the
number
of
connections,
but
one
connection
hogging
it
all.
If
they're
on
a
shared
bus,
you
know
like
a
4k
camera
at
60
frames,
a
second
if
it
goes
wild
Maybe
it.
You
know.
Maybe
if
these
are
smart
cameras
that
do
motion
detect
and
only
send
on
motion,
you
don't
want
one
going
wild
so
that
the
other
ones
don't
get
a
shot.
Whether
this
is
something
there's
existing
technology
on
I,
don't
know
if
there
is
I'm
not
aware
of
it,
but
yeah.
B
Thing
if
you're
talking
about
input,
then
I
guess
if
you
put
a
load
balancer
in
front
or
like
have
basically,
you
can't
establish
a
direct
connection
with
the
device
but
yeah
but
I
think
I.
Think
that's
fair
and
it's
sounding
like
you're
bringing
the
discussion
up
from
the
category
of
resource.
Aware,
like
the
the
principle
of
resource
aware
made,
you
bring
up
this
topic
and
the
reason
I
was
saying
that
is
because
I
see
that
topic
as
being
about
external
device
connectivity,
which
furthers
the
point
of
maybe
these
should
be
merged.
A
B
Well,
I
think
one
of
them
is
I,
think
resource
usage,
optimization
encompasses
some
of
that
and
I
think
the
word
external
can
be
removed.
So
we
could
say
resource
aware
and
that's
external
and
internal,
but
I
think
our
discussion
may
have
been.
We
wanted
something
explicitly
discussing
external
devices.
A
B
I
yeah
I
don't
see
a
need
to
address
management
of
devices
yeah
yeah
in
either
of
these
I,
don't
see
any
to
put
the
word
management,
so
I
think
what
we
could
do
is
I.
Think
resource,
aware
the
reason
we
I'm
trying
I
think
I'm,
remembering
the
reason
we
put
it
is
because,
in
the
similarities
and
differences
slide,
we
mentioned
how
the
The
Edge
is
more
heterogeneous,
so
applications
need
to
be
more
aware
or
of
what
resources
are
on
their
compute.
B
B
While
that's
a
different
discussion
than
external
device
connectivity
right.
Those
are
I
see
those
are
two
different
things.
Yeah.
A
I
up
to
now,
I
wasn't
thinking
of
a
GPU
as
being
in
the
category
of
an
external
device,
but
now
that
you
mentioned
it
I
suppose
some
people
could
think
that.
But
to
me
that's
belongs
more
under
the
category
of
compute,
even
though
it
is
a
device.
I
mean
the
CPU
is
a
device
too,
and
so
are
the
memory
chips
but
yeah.
F
Exactly
I
agree
with
that
as
well:
I
think
gpus
are
in
compute
mechanism
and
not
a
data
source.
A
B
I
I,
don't
I,
don't
think
the
word
external
is
confusing,
like
I
I
think
I
phrased.
What
I
was
saying
wrong:
I,
don't
think
a
GPU
is
an
external
device,
okay
and
I.
Maybe
if
someone
else
thinks
that
then
yeah
we
should
say
I
o
devices
but
I'm
aligned
with
that
I,
don't
think
a
GPU
is
an
external
device,
but
I
think
what
Omar
was
saying
about.
B
Maybe
we
talk
about
compute
resources
or
I.
Guess
we
should
scope?
What
are
we
trying
to
figure
out
right
now.
A
I
hate
to
throw
this
in
a
different
direction,
but
I'll
do
it
anyway,
that
maybe
this
isn't
external
device
connectivity
so
much
as
management,
which
is
broader
than
connectivity.
You
know
management
would
entail
to
me
taking
inventory
of
the
devices
discovering
them,
maybe
trying
to
keep
their
firmware,
updated,
and
that
goes
beyond
connectivity.
E
D
A
E
A
Could
choose
to
do
it
in
an
application,
but
it's
possible
that
maybe
someday
some
orchestrator
or
something
you
know
in
the
stack
would
take
it
on
too
so
I'm,
not
sure
an
application
is
a
way
to
address
the
challenge
one
way,
but
the
the
principle
is
still
there,
and
this
is
something
that
is
a
little
different
from
a
public
Cloud
data
center
environment.
You
know,
if
you're
hosted
in
AWS,
you
don't
worry
about
managing
devices.
A
The
cloud
provider
isn't
going
to,
let
you
get
to
any
devices
sitting
there
anyway,
even
if
you
wanted
to,
but
at
Edge.
This
is
one
of
the
fundamental
distinctions
that
you've
got
these
I
o
devices
and
we
need
to
call
them
out
in
some
row
of
this
table,
I
think
and
what
the
title
of
that
cell
is
on
the
left
side.
I
think
you
know
external
device,
something,
and
maybe
it's
management.
F
A
B
No
I
don't
think
we
should
get
rid
of
a
row
I
also
like.
B
C
Their
resource
aware,
the
way
you
just
reworded.
D
C
Really
resonated
when
you
were
describing
that
and
I
think
you
have
that
now.
B
I
was
I,
essentially
just
pasted
this
from
our
differences.
We
have
a
hardware,
awareness,
differences
and
so
I
repeated
the
wording
so
I
said,
rather
than
having
homogeneous
Hardware
platforms.
Developers
need
to
be
aware
of
a
wider
variety
of
hardware
and
interfaces,
and
that's
what
we're
defining
is
resource
aware,
because
that
removes
the
word
connections,
because
that's
what
was
really
throwing
me
off
and
local
resources
is
a
little
vague.
E
I
like
it
because
I
was
also
drawn
of,
especially
not
being
there
there
for
the
last
meeting,
that
supporting
connections
to
local
resources
is
something
that
threw
me
off,
that
it
might
be
the
same
thing
we
want
to
talk
about
in
the
in
the
external
device
connectivity,
so
I
think
right
now,
it's
it's
much
more
clear.
B
B
Changing
it
to
reflect
similarly
to
what
we
said
in
Hardware
awareness
slash,
should
we
rename
this
from
resource
aware
to
Hardware
awareness.
F
B
F
B
Okay,
so
for
external
device
connection
connectivity-
this
is
all
about
now
we're
talking
about
that
heterogeneous
environment
outside
of
the
node
and
how
applications
need
to
find
a
way
need
to
consider
all
the
connection
endpoints
around
them
in
their
environment.
And
what
is
that
process
like,
and
what
does
that
entail?.
E
And
I
think
we
can
talk
there
about
yeah,
as
is
estimoya,
suggested
to
to
it's
a
dynamic
and
changing
environment,
and
we
need
tools
to
discover
and
and
yeah
use
those
devices.
D
B
B
Foreign,
since
you
came
up
with
this
great
comment,
do
you
feel
like
this
is
a
way
we
could
word
the
whole
principle
itself.
B
Are
we
so
I
think
the
last
sentence
of
it
is
kind
of
a
reword
of
the
first
couple
sentences,
but
it's
also
very
good.
Is
there
a
way
we
can
incorporate
that
so?
The
word
capability,
for
example,
is
an
interesting
concept,
but
I
think
we'd
want
to
Define
it
before
using
it.
B
But
I
I
like
that.
Maybe
we
can
say
the
capabilities
of
applications
on
the
edge.
No,
maybe
that's
getting
too
too
much,
but
it
seems
like
what
you're
saying
here
is
a
capability
of
a
server
on
the
edge
is
also
that
of
all
of
the
resources
around
it.
It's
not
static
and
those
resources
aren't
static
and
therefore
the
capabilities
of
the
server
isn't
static.
E
E
What
would
be
that
referred
to
control
the
application
or.
B
B
So
is
there
anything
in
the
bullets
that
were
not,
including
here
we
don't
get
into
details
about
the
types
so
like.
E
Yeah,
yeah,
yeah
or
or
or
the
considerations
and
I
I,
don't
think
they.
They
belong
here
in
that
in
that
two
three
sentences
description,
so
you
can
describe
it
later
or
you
know
specialization
right.
B
Do
we
want
an
intro
sentence
along
the
lines
of
something
like
this
of
on
the
edge
applications
are
closer
to
the
source
of
compute?
They
must
know
how
to
connect
to
those
devices.
C
B
B
Okay
and
I
think
this
grid
of
the
paper
is
something
that
is
going
to
be
one
for
a
lot
of
offshoot
papers.
If
we
are
able
to
get
to
that
point,
because
there
is
a
lot
here
to
talk
about.
E
B
B
B
E
Anything
that
consumes
a
little
bit
is
that
we
talked
a
little
bit
about
it
in
the
portability
right,
seeing
different
Hardware
platform
and
local
resources
and.
B
Yeah
I
think
the
only
issue
here
is
is
an
I
o
device,
a
locally
attached,
Hardware
I
think
that's
where
we
get
grainy
with
external
device.
I
think
what
that
boundary
is
is
a
problem.
But
if
there's
a
slight
blending
of
these
I
think
that's
okay
and
we
can
even
have
a
footnote
that
discusses
that
that
boundary.
B
So
maybe
maybe
we
add
a
footnote
that
says
like
the
boundary
between
hardware
and
external
device-
is
vague,
but
regardless
of
that
boundary,
you
need
to
be
aware
that,
regardless
of
that
external
device
connected
activity
is
trying
to
discuss
how
that
can
change.
So
if
you
define
Hardware
as
something
that
can
be
plugged
in
removed,
then
take
into
account
like
external
device
connectivity
principle,
otherwise
just
take
into
account
that
your
Hardware
is
more
varied
in
general.
I
mean
that's
a
way
big
footnote,
but
we
could
have
a
footnote
along
those
lines.
B
Okay,
well
I'm,
going
to
resolve
these
unless
folks
feel
otherwise.
B
So,
okay,
at
this
point,
we
have
15
minutes
left.
What
do
we
want
to
do
next
on
our
agenda
on
the
list
of
things
that
we
said
we
have
to
do?
Do
we
want
to?
Maybe
next
on
our
list
was
list
of
Open
Source
projects?
Is
this
complete
so
I'll
go
to
that
next?
Unless
other
there's
other
ways
we
want
to
Swerve
next,
what
I
did
this
past
week
is
went
through?
B
Cncf
has
a
bunch
of
projects
I
downloaded
the
CSV
and
control
left
Edge
and
tried
to
put
everything
on
there
that
mentioned
Edge
and
then,
after
looking
at
it
did
seem
to
have
an
edge
use
case
so
that
added
Fab
Edge
Nats
Intel,
smart,
open
Edge.
B
B
Also
at
metal
lb
came
from
that
because
it's
focused
on
like
kind
of
bare
metal
Edge
scenarios
also
there's
some
from
Eclipse
that
I
added
just
from
looking
at,
like
they
have
a
list
that
I
think
I
lost
the
link
of,
but
I
did
link
it
at
one
point,
but
is
there
the
other
sections
a
little
concerning
because
obviously
we
didn't
get
them
all?
A
I
think
it
needs
some
work.
It
looks
a
little
small
mosquito
is
an
eclipse,
that's
under
other,
unless
that's
where
you're
putting
the
eclipse
things.
A
B
When
would
it
be
ready
like,
what's
the
number
that
we
feel
like,
we
would
be
comfortable
with.
A
Well,
I
think
maybe
you
just
have
to
declare
a
date
and
those
who
want
to
I,
don't
think
there's
deletions
likely
from
this
list,
but
there
might
be
ads
and
it
would
kind
of
be
nice
to
not
just
list
them
but
go
to
the
work
of
making.
The
name
of
this
thing
be
a
link.
You
know.
If,
if
I
were
a
reader
of
this
white
paper,
it
would
be
a
value-add
if
these
things
were
actual
URL
links
to
go,
find
out
more
about
these.
B
For
sure
and
I
think
that
also
made
me
think,
should
this
whole
section
link
to
a
working
CSV
document,
what
if
it's
like,
a
just
literally
a
a
Excel
doc
or
a
Google
sheet
stock,
that
one
column
is
principal,
the
other
is
link
and
the
other's
name,
and
then
you
have
description
kind
of
like
what
cncf
has
and
this
whole
section
all
it
does
is
linked
to
that.
A
I
don't
know:
what's
the
value
versus
just
having
the
link
be
in
text,
you
know
if
it's
in
text
and
you
elect
to
have
a
slide
that
summarizes
this.
For
example,
you
could
just
cut
and
paste
it
but
think
if
it's
a
spreadsheet,
that's
a
little
Messier.
D
B
Being
able
to
say
this
is
a
working
document
is
representative
of
the
fact
that
it
is
a
working
list.
B
Pointing
to
a
list
and
saying
we
got
it
started,
I
think
like
in
the
presentation.
We
would
pull
out
several
of
them,
maybe
two
from
each
principle,
but
having
something
like
that.
E
A
B
I
think
it's
fine
to
keep
it
instead
of
diverting
us
to
a
different
thing.
Let's
focus
on
okay,
this
is
going
to
be
in
text.
That
seems
fine.
B
One
thing
when
we
were
creating
the
slides,
Amar
changed
this
to
be
a
grid
and
I
think
that
might
be
more
helpful
going
along
with
our
grid
format
of
have
this
via
principles
grid,
just
like
we
had
earlier
and
then
each
project
and
then
in
parentheses,
the
foundation.
So
it
could
be
cncf,
LF,
networking,
LF,
Edge,
LF
eclipse
and
then
just
add
it.
That
way,
maybe
and.
A
You
know
in
the
long
run,
if
we
wanted
this
to
be
a
living
thing,
taking
it
as
a
model.
The
cncf
itself
has
their
landscape
thing
right
now,
and
it's
driven
by
GitHub
check-ins,
mostly
by
people
who
want
to
appear
in
the
landscape
just
voluntarily
submitting
a
PR.
We
could
probably
have
a
thing
that
we'd
started.
A
We
are
not
going
to
get
it
done
in
the
next
two
weeks,
but
in
the
long
run,
if
we
had
an
edge
native
landscape
that
was
kind
of
maintained
by
people
who
want
to
appear
in
there,
meaning
we
don't
have
to
do
all
of
that
work
of
keeping
it
up.
It's
up
to
the
people
kind
of
self-nominating
to
be
relevant
to
the
landscape.
That
would
be
something
of
value
to
the
community
and.
A
I
think
that's
that's
where
we
should
go
as
an
end
game,
but
for
the
short
term
of
just
getting
a
useful
list
that
you
can
pop
into
your
presentation.
Just
saying
hey
here:
maybe
we,
since
this
is
a
graph,
we
don't
make
a
claim
that
this
is
exhaustive
right
now
it's
just
highlights
or
the
ones
we
came
across
and
leave
it
at
that,
because
we're
not
going
to
be
able
to
get
a
perfect
list
in
two
weeks.
B
Okay,
that
sounds
good
to
me.
B
H
This
is
Brandon
before
we
move
off
the
topic.
I
do
think
the
the
landscape
is
something
good
to
consider
in
the
longer
run.
It's
it's
built
on
a
free
tool
developed
by
the
Linux
foundation
and
I.
H
Think
they've
got
some
processes
whereby,
if
we
channel
the
request
through
the
cncf
and
make
the
case
that
we've
got,
you
know
a
critical
mass
and
a
interesting
and
useful
angle
on
edge
native
to
create
a
a
landscape
around
I
think
that
they
would
be
open
to
to
considering
that
and
I
think
that
would
have
value
as
well
to
give
this
initiative
more
visibility
and
elicit
more
participation.
H
One
of
the
things
having
built
out
a
landscape
for
LF
networking
I
know
that
we'll
have
to
come
up
with
categories
either
on
the
x-axis
or
y-axis,
or
both
that
talk
about
how
the
logical
groupings
of
these
projects
should
go.
So
we'd
have
to
figure
out
those
those
categories,
whether
they're
aligned
under
our
Aid
or
so
principles
or
different
parts
of
the
Edge
from
where
they
are
on
the
edge.
We
would
have
to
figure
that
out,
but
I
do
agree.
H
H
If
we,
you
know,
go
forward
with
something
and
say
hey
this
is
it
and
it's
not
really
subject
to
change
or
we
make
change
difficult
that
might
wrinkle
some
feathers
from
folks
in
the
community
that
are
using
different
projects
that
think
that
those
projects
Merit
inclusion
on
the
list?
We
don't
want
to
create
the
friction
with
those
folks.
H
So
in
order
to
reduce
that,
if
we
have
it
in
a
spreadsheet
of
some
form
and
say
this
is
our
first
pass
at
creating
the
list
and
this
these
were
our
criteria
and
it
seems
like
we
are
sort
of
applying
criteria
around
the
principles
to
these
projects,
rather
than
trying
to
get
every
single
Edge
related
project
known
to
humankind
on
on
the
list.
But
if
we
just
have
a
statement
of
hey,
this
is
our
first
attempt
based
on
this
rough
criteria.
H
Here's
a
list
that
is
accessible
if
you
are
would
like
to
add
your
Edge
project
to
the
list.
Here's
what
you
do
and
sort
of
allow
people
to
do
that.
That's
why
that
way,
people
that
care
and
feel
that
they're
missing
can
get
it
added.
Otherwise,
you
know
other
folks
will
just
kind
of
Nod
and
say
hey.
The
list
looks
good
I,
don't
know
what
y'all
think
about
that.
A
I
agree
with
that
I
like
I,
like
that
plan
and
explanation.
B
So
would
have
making
sure
that
people
have
a
way
to
add
their
project.
Is
that?
Okay,
if
we
say
PR
into
this
draft,
to
add
your
project
or
are
we
saying
that
we
need
it
to
be
a
spreadsheet,
because
that's
lower
friction.
A
Well,
maybe
we
should
even
announce
that
we're
thinking
of
going
the
correction
of
the
landscape,
but
we're
not
ready
to
do
it,
yet
it
I'd
almost
seen
this
spreadsheet.
If
the
landscape
was
there,
that
should
be
the
one
source
of
Truth,
not
the
spreadsheet,
so
maybe
the
spreadsheet
should
even
be
viewed
as
a
draft,
but
people
can
get
themselves
on
the
interest
list
or
whatever
I'll
tell
you.
A
Another
thing
for
purposes
of
this
group
maybe
suggests
that
the
spreadsheet
would
also
serve
as
a
list
of
people
in
a
queue
that
we'd
ask
to
speak
at
meetings
of
our
group,
I
mean
at
some
time
to
keep
this
this
group
interesting
to
actual
Community.
We
can't
just
endlessly
have
it
be
white
paper
draft
discussions.
You
know,
I,
think
our
best
attendance
comes
when
we
get
speakers
on
interesting
projects.
A
So
having
a
self-nominated
spreadsheet
to
you
know
indicate
to
the
world
that
you've
got
something
related
to
Edge
would
be
useful
to
announce
at
kubecon,
and
if
you
announce
that
people
can
put
them
in
there,
one
thing
would
be
do
we
want
it
wide
open
or
is
it
restricted
to
open
source?
Only.
H
I
think
that's
a
great
idea,
Stephen,
and
if
someone
wants
to
nominate
their
project,
their
open
source
project
I
do
think
we
should
err
on
open
source
to
be
added
to
our
list
and
eventually
our
landscape.
Here's
the
process
for
doing
that
sign
up
here
and
then
we
also
invite
you
to
come
to
the
iotn
edge
working
group
to
talk
about
your
project
and
how
it
integrates
with
the
edge
native
ecosystem,
it's
great
to
link
those
two
together.
H
B
I,
wonder
is
that
so
is
the
process
just
that
anyone
can
access
a
spreadsheet,
anyone
can
add
anything
to
it
or
how
do
we
handle
access
rights
and
then
that
way
like
I,
think
that's
why
github's
kind
of
nice,
because
you
can
submit
a
PR
and
someone
can
go
in
and
look
up
the
project
and
be
like?
Oh,
this
is
related
to
the
edge
or
oh.
This
is
cool.
Also
come
speak
at
our
working
group.
B
If
you're
available,
for
example,
that's
how
the
cncf
landscape
works
is
you
have
to
submit
a
PR
essentially
into
a
GitHub
spreadsheet,
yeah.
E
But
maybe
we
can
do
it.
Similarly,
like
we
are
doing
with
the
white
paper,
so
first
period
be
a
a
sheet
that
any
any
group,
member
or
of
the
working
group
can
at
least
suggest
changes.
And
then,
when
it
shapes
a
little
bit,
then
we
convert
it
to
the
markdown
or
whatever.
We
need
for
the
for
the
landscape
in
a
in
a
GitHub
and
and
continue
with
the
PRS.
H
I
think
so,
yeah
replace
it
with
a
line
and
a
link
to
a
Google
spreadsheet,
and
then
a
good
idea
for
control
might
just
be
hey.
If
you
want
to
add
your
project
to
this
list,
you
know
you
can
request
access
here
and
that
at
least
gives
us.
You
know
some
level
of
awareness
of
people
that
want
to
change
it
or
add
to
it,
and
then
it
gives
it's
a
conversation
starter
for
hey
I,
represent
at
this
project.
B
Great
and
I
think
also
that
could
be
a
good
opener
for
every
bi-monthly
meeting
is
looking
at
the
spreadsheet,
going
into
Google
and
seeing
the
delta
in
the
past
two
weeks
and
seeing
what
projects
were
added
and
then
saying.
Oh
cool
announcement.
There's
these
two
new
projects
quickly
go
to
the
link
and
just
spend
five
minutes
on
them
could
be
kind
of
fun.
H
Yeah
yeah
and
in
terms
of
how
many
projects
we
expect
I
mean
I,
think
this
lifts
is
a
good
first
swag,
and
this
is
maybe
half
the
Universe
we
may
get.
You
know
another
dozen
or
so
projects
wanting
to
sign
up
for
this
and
be
involved
over
the
coming
months
and
I.
Think
that
would
be
a
a
good
number.
You
know
as
long
as
it's
really
the
participation
and
the
contribution
part
that's
going
to
I
think
weed
out
others
that
just
want
their
project
added,
but
nothing
else
behind
it.
H
And
if,
if
that's
the
case,
then
we
feel
like
maybe
their
project
isn't
well
positioned
for
this
exercise.
You
know
we
we
have
to
have
some
level
of
Engagement
with
them,
but
if
they
can
request
that
through
requesting
access
to
the
sheet,
it's
it's
a
first
level
of
filter
to
make
sure
that
they're
adequately
I
think
representing
their
community
and
wanting
to
contribute.
B
That
sounds
good.
Is
anyone
opposed
to
the
spreadsheet
movement.
G
B
Yeah
I
think
one
flow
could
be
if
you
we
could
even
have
a
member
of
the
working
group.
Who's
like
this
is
a
working
group
responsibility,
but
when
you
see
a
new
project
added
put
up
a
GitHub
issue
on
their
GitHub
saying,
would
you
be
interested
in
talking
about
your
project?
B
That's
something
that
brought
me
into
this
working
group.
I
think
Steve
may
have,
or
maybe
it
was
the
I
think.
Maybe
runtime
tag
did
that
where
they
put
up
something
on
Awkward.
That
was
like.
Would
you
be
interested
in
presenting
in
the
runtime
working
meeting
and
that
brought
me
into
that
community,
so
I
think
having
someone
who
does
that
sees
the
project
goes?
The
GitHub
page
puts
up
an
issue
with
the
question
tag
and
invites
them
to
speak
could
be
good.
A
B
If
you
create
a
spreadsheet
and
just
put
it
here,
Steve
I
I'm
happy
to
help
Port.
We
can
just
delegate
who's
on
what
bullet.
B
And
then
I'll
leave
that
other
thing
is
references
we
I
guess
we
can.
We
need
to
start
like
adding
footnotes
for
when
the
references
are
relevant.
B
Anyone
interested
in
taking
on
linking
references
throughout
the
paper
so
there's
times
when
we
reference
these
things,
but
don't
have
necessarily
footnotes
and
maybe
having
something
where
these
are
numbered.
References.
C
D
C
And
I
can
help
with
that.
There
might
be
a
question
of
how
do
we
want
to
reference
that,
but
we
don't
have
to
discuss
how
now,
whether
it
be
numbered
or
what
have
you,
because
most
of
these
also
have
a
hyperlink
where
it's
stated
within
a
document
already.
B
Yeah
and
it
it
doesn't
need
to
be
pretty
because
we're
going
to
move
this
to
markdown,
soon,
anyways,
okay
and
then
I
think
folks,
who
don't
have
anything
explicit
to
do
but
want
to
be
involved
just
going
through
and
resolving
ads
and
replaces
where
they
seem
acceptable,
maybe
pairing
up
with
someone
else.
If
you
don't
feel
comfortable
doing
making
those
decisions
independently.
B
B
B
Okay
and
then
feel
free
to
ping
me
or
put
in
the
working
group.
If
you
want
help
with
that
sure.
B
B
D
B
B
Okay,
well
we're
certainly
over
so
I
think
that
can
conclude
our
meeting.