►
From YouTube: IETF103-SACM-20181108-1610
Description
SACM meeting session at IETF103
2018/11/08 1610
https://datatracker.ietf.org/meeting/103/proceedings/
A
C
B
F
F
F
F
F
Now
so
Friday
is,
in
any
event
he's
getting
married
and
while
he
was
originally
optimistic
that
he
would
have
lots
of
spare
time,
we
were
skeptical
so
anyway
there's
a
small
update
that
he
wants
to
make
and
then
it
should
be
ready
for
working.
Your
blast
call
and
I
think
the
the
mimetype
issue
has
been
clarified
and
that
just
needs
to
be
rolled
back
into
the
document,
so
we
should
be
ready
to
go.
F
F
G
Pending
the
adoption
of
the
current
architecture
document,
we
are
postponing
major
refactoring
of
the
terminology,
because
it
would
not
make
sense
if
the
change
in
a
core
concept
which
I
hope
we
don't
so
I,
would
reinforce
Karen's
recommendation
to
read
please
the
current
version
of
the
architecture
and
maybe
even
remember
how
the
old
one
looked
like
I
dig
it
up
from
the
archive
because
it's
expired
and
then
assess.
If
this
is
addressing
what
you
need,
it
is
more
concrete.
It
has
very
specific
component
roles
in
it
and
terminology
wise.
G
We
are
still
on
the
modular
whole
concept
that
was
initially
proposed.
So
this
is
easy
to
specialize
a
generic
term.
It's
hard
to
generalize
a
specific
term,
so
I
think
that
there's
and
there's
not
a
big
challenge
to
adopt
the
terminology
document
through
the
architecture.
But
maybe
we
want
to
stay
a
little
bit
extensible
and
that
again,
I
have
not
really
read
about
in
the
architect
yet.
G
So
that
is
the
most
important
part
where
my
terminology
comes
in,
because
if
you
want
to
extend
something,
you
have
to
know
why
and
where
and
how
the
semantics
are
placed.
So
that
is
I
think
the
my
request,
added
to
the
I'll,
give
you
of
the
architecture.
If
you're
not
satisfied
with
the
semantics
they're
like
I,
don't
know
what's
assessed
here
and
I,
don't
know
what
the
outcome
is
and
I
don't
know
what
the
workflow
should
look
like.
G
G
G
Lights
just
in
time,
but
yeah,
so
we
created
an
update,
there's
no
about
concise,
off
identifiers,
a
basis
for
the
variation
of
target
end
points
because
they
are
at
least
semantically
one
half
of
it
is
software,
the
other
heart,
hath
speakers,
hardware,
components
and
we
had
want
to
use
the
concise
of
identifiers
to
identify
the
software
inventory
and
its
characteristics
and
semantic
interconnected,
as
we
are
concise
of
identifiers
next
slide,
and
actually
we
are
preparing
for
working
group
last
call
as
only
one
remaining
issue.
There
are
three
issues
open.
G
G
The
remaining
issue
is
a
very
specific
representation
of
media
in
that
is
originating
from
ISIL
spec
text,
which
is,
let's
say
open-ended,
and
nobody
really
could
explain
to
me
what
the
effect
of
purpose
is
so
I
included
an
example
of
how
this
is
a
phrase
at
the
moment.
So
there
are
expressions
and
they
interact
with
other
expressions
and
error,
a
bps
associated
with
that
so
I
guess
it's
a
conditional
language.
G
It
was
relatively
easy
to
convert
that
to
CDT
area
next
slide,
please
so
that's
the
CDD,
a
representation
of
the
media
attribute
for
concise,
often
divides
I,
guess
nobody
who's
really
ever
read
that
I
should
Raph
was
familiar
with
how
semantics
of
software
components
work
like
we
can
make
no
sense
of
this,
and
so
in
essence,
it
is
up
to
the
office
to
vet
this
validity.
That's
the
only
remaining
issue
here
and
yeah,
so
I.
This
is
just
to
show
how
a
structured
version
of
this
looks
like
next
slide.
Please
and
yeah
in
order
to.
G
Yeah,
it's
it's
as
though
the
media
is
hard
to
parse.
I
really
accept
that.
Even
then
the
originals
isida
document
it's
hard
to
get
what
what
this
is
about.
So
far
doing
last
call
there
are,
as
you
can
see,
some
tiny
details
you
have
to
do
so.
We
are
reading
the
extraversion
schema,
that
is
in
the
and
our
considerations.
We
proposed
a
entity
role.
Registry
I,
gave
examples
here,
so
texts
suffer
text,
I
have
a
creator
of
the
tech
itself
of
a
software
sales
or
semantic
to
software
components
that
are
entities.
G
We
want
to
have
a
registry
for
that.
We
also
want
to
have
a
registry
for
interconnectedness
of
components
again
yeah
some
software
requires
other
software
or
recommends
other
software
also
precedes
it,
patches,
etcetera,
etc.
So
this
is
another
type
of
semantic
interconnective
that
are
links.
There
is
an
existing
one
that
is
semantically
different,
so
I
think
the
Ayana
will
grant
us
a
separate
registry,
but
this
up
vienna.
I
think
it
would
be
an
a
discussion
and
then
there,
the
very
obvious
ones.
G
G
G
The
+
implies
absolute
semantic
equivalence
that
we
made
sure
to
make
this
provide
the
basis
for
this
by
cutting
out
every
extension.
That
is
not
mentioned
in
the
I
suspect.
For
example,
we
had
appendix
a
appendix
about
firmware
a
representation.
The
ISO
does
not
have
anything
like
that.
We
just
added
an
extension
point
for
other
resource
collections
and
will
do
from
were
in
another
draft.
We
covered
this
in
suit.
G
We
hope
that
the
manifest
of
the
suit
working
group
will
be
a
feasible
resource
collection
type
for
this,
and
if
that
happens,
we
have
high
synergy
with
this
huge
working
group.
So
that's
my
actually
hope.
In
any
case,
we
are
starting
with
C
war
in
order
to
test
this,
and
if
it's
not
possible,
we
will
go
with
Cosway.
It
same
goes
for
the
coop
content
format,
which
requires
Amir
type
registration.
G
It's
a
precondition,
prerequisite
so
to
speak,
and
it
is
again
then
for
co-op
applications
for
passive
or-
and
we
have
a
explicit
seaboard
data
item-
that
is
the
concise
of
identifiers,
and
you
need
a
magic
number
for
that
to
identify
this.
So
everybody
who
can
from
the
top
of
the
head,
translate
SKS
key
integer
representation
of
X
will
see
that
this
magic
number
is
means
Swit
and
Marisa,
pointing
to
a
map
construct,
and
it
will
define
the
semantics
of
a
concise
soft
identifier.
G
So
these
three
items
are
standard
registrations
and
the
three
first
only
for
registry.
So
this
is
actually
a
lot
of
I&R
considerations,
but
all
of
them
are
feasible
and
then
we
have
reusable
semantics
for
other
work.
In
this
domain,
because
some
software,
especially
film,
might
require
a
specific
habitat,
and
so
there's
reuse
value
in
this
next
slide.
G
Please
and
that's
basically
it
so
we
are
going
with
the
final
version
after
Oh
8,
it's
hopefully
oh
9
will
be
the
candidate
for
working
of
last
call,
and
my
request
to
the
chairs
is
that
we,
if
we
are
indicating
that
we
are
done,
that
we
will
request
on
the
list
start
of
last
call
and
proceed
from
there
and
yeah.
Then,
if
anybody
has
questions
or
comments
about
this
process
or
what
I
just
said,
please
speak
out.
F
G
G
There
was
no
real
tables
for
that
was
implicitly
captured
in
a
specification
texts,
so
English
text
always
leaves
room
for
interpretation,
so
that
was
the
only
place
where
we
had
to
be
really
careful,
I,
think
all
of
that
was
covered
by
Dave
in
a
final
polish
and
inquiry
by
him,
but
I
think
Dave
should
comment
on
this.
So
I
will
ask
him
to
give
a
comment
on
the
list
on
this
and
then
yes,
I
would
affirm
your
assumption,
but
it
can
only
be
really
sure
that
it
has
been
reviewed
by
the
I
suspect
office.
H
G
G
G
F
J
G
A
more
complex
network
device,
it
deploys
a
that's,
let's
call
it
a
firmware
blob,
but
if
you're
looking
at
it
closely,
it's
a
file
system
with
a
lot
of
files
in
it
and
recently
modern
network
equipment.
Basically
the
Linux
running
a
lot
of
software,
so
the
more
modern
items,
the
more
bigger
items
you
see,
the
more
applicability
is
them
smaller.
They
get
involved
more,
it's
actually
firmware
without
multitasking
in
specific
from
our
parts.
We
require
the
resource
extension
that
will
be
crafted
in
suit.
G
J
G
F
G
F
K
The
purpose
of
our
draft
is
trying
to
define
a
minimal
set
of
configuration
and
the
status
parameters
of
the
security,
related
functions
and
the
service
or
network
infrastructure
device,
and
this
configuration
and
the
status
can
be
collected
by
the
second
collector
and
consumed
by
second
evaluator,
and
so
the
security
postures
of
this
security
of
the
network
devices
can
be
recorded
and
they
evaluated
in
general.
The
security
baseline
can
be
divided
into
three
layer.
K
Three
layers,
the
application
layer,
the
network
layer
and
the
infrastructure
layer
and
the
network
layer
can
be
also
divided
into
three
plans:
the
control
plan
that
they
play
and
the
management
plan
in
our
dropped.
Our
draft
a
is
about
as
a
paid
plan,
security
baseline
and
as
the
updates
of
our
from
version
2
to
version.
3
is
mainly
focused
on
the
data
modification
of
the
tape
model
structure.
In
the
new
version
we
reorganized
as
a
data
model
structure
of
the
security
functions.
K
K
The
the
Mac
limit
has
the
bulimic
limit
and
in
the
face,
manga
limiter
can
also
configure
the
maximum
Mac
entries,
the
right,
Oh,
Mac
learning
and
some
other
functions,
but
the
difference
is
a
key.
The
Mac
limit,
the
villa
Mac
limiter,
is
based
on
the
villa
ID
and
the
interface
Mac
limiter
is
based
on
the
interface
name,
and
the
traffic
surprise
is
similar.
K
K
G
Reading
would
be
overstating
it
but
I'm
following
them,
because
they're
young
modules
and
I
have
a
comment
on
this,
so
in
net
Kahn.
Finally,
maybe
a
little
bit
too
late,
but
finally,
the
event
notification.
This
is
a
subscribe
notification
and
the
yang
push
our
drafts
become
working
book
last
call,
and
then
this
information,
because
it
is
based
on
yang
data,
store
structures,
could
be
used
to
create
yang
telemetry.
So
that
would
eliminate
the
need
to
pull
these
some
of
these.
G
Standards
pushed
to
your
second
collector
and,
as
the
eliminates,
the
need
to
yeah
well,
overhead
of
periodic
falls,
which
is
a
nice
feature,
I
think,
and
if
you
have
something
really
important
to
communicate,
you
can
even
create
specific
notification
statements
and
the
yang
module
to
have
stuff
that
you
always
want
a
collector
to
subscribe
to
the
data
tree
module
can
be
big,
it
can
be
used
for
configuration
for
intend
and
all
other
flavors
that
nmda
provides.
But
the
notification
statements
are
very
specific
subsets
of
information.
G
You
can
hard
code
into
the
module
that
are
most
certainly
intended
to
be
telemetry
sent
to
the
second
collector
and
that
could
be
of
use
to
all
modules
and
yank.
This
is
not
limit.
This
commit
is
not
limited.
To
these
specific
set
of
young
modules
is
a
basically
a
good
thing
for
every
second
day
later
keg
module.
J
Mean
that
yeah
we
know
that
Netanyahu
has
defied
this
we're
a
new
sub
pop
and
push
in
Canada,
and
it
is
almost
published.
Yes,
we
we
also
want
to
reuse
them,
because
I
remember
in
the
very
beginning
of
this
draft
of
this
series
of
chat.
We,
our
a
very
initial
idea,
is
that
if
it's
only
about
the
security
stages
and
configuration
collection,
actually
the
yarn
push
Department
anything's,
it's
a
most
suitable
for
us.
So
so,
of
course
we
are
learning
of
the
Numa,
Canada
and
I
think
in
the
future
ocean
of
the
draft.
J
G
G
Okay,
that
does
not
in
most
cases
it
doesn't
make
much
sense
to
if
you
do
a
update
to
a
tree.
If
you've
write
something
to
into
a
node,
then
if
you
subscribe
to
that
node,
you
will
get
telemetry
that
you
just
wrote
that,
but
you
know
because
I
just
wrote
it
so
the
subscription
to
the
trees
are
typically
filtered
either
by
sub
sub
tree
sub
expressions,
which
would
not
be
enough
for
complex
writeable
notes
structures.
G
All
we
are
XPath
filters
and
X
fast
with
us
would
be
in
this
case
required
if,
depending
on
how
your
structure,
the
tree,
if
you
have
a
one-half
of
the
three
is
about
configuration
settings
and
another
path
of
the
branch
of
the
tree
would
be
read-only
and
you
could
to
a
substrate
subtree
subscription
to
all
of
that.
If
you
just
want
specific
parts
of
that,
like
I,
don't
know
my
certificate
change
on
the
data
plane
I
got
a
report.
G
Is
a
bird
or
a
flapping
event,
or
something
like
that,
then
I
would
subscribe
to
those
notes.
Specifically,
we
are
separate
subscriptions
that
are
measured,
measured,
bundled
subscriptions
and
then
that
that
get
to
you
pushed
on
change,
for
example.
So
it
is
not
feasible
to
subscribe
to
a
complete
tree
of
the
writable
notes
in
there,
because,
typically,
you
use
these
nodes
to
set
settings,
and
you
know
that
you're
just
doing
that
would
create
arbitrary
telemetry.
That
is
not
useful
for
just
create
volume
and
I,
think
volume
and
scaling
the
down
is
a
big.
J
Yes,
thank
you.
I
I
got
your
point,
but
I
just
want
to
say
that,
although
our
initial
idea
is
that
we
want
just
want
to
specify
the
security
stages
and
the
changes
to
to
make
them
as
a
telemetry
and
collector
and
monitoring
the
device
network
devices
security
status,
but
in
factored
in
current
draft,
maybe
if
our
for
we
we
don't,
we
haven't
the
worry,
we
haven't
clearly
differentiate.
What
party
is
configuration
partner
and
what
party
is
elementary
or
the
state?
Your
part?
Actually
they
are,
they
are
together.
J
So
I
think
the
problem
for
us
to
be
solved
is
that
maybe
we
should
differentiate
it
and
we're
clearly
and
the
configuration
is
configuration
and
the
stadium's
monitoring
in
stages
monitoring
so
for
the
state
of
the
morning
during
part,
I
think
we
should
the
follow
the
net
conf
push
me
Canada
and
take
the
announced
that
Elementary
and
connection,
but
I
still
think
that
the
leftist
configured
in
part.
It's
also
useful,
but
I
am
not
sure
whether
it
should
be
here
or
should
it
be
net
mode
on
that
convoking
group.
L
D
M
M
M
F
D
C
F
F
So
I,
don't
I,
think
the
revelant
relevance
to
sockem
was
uncertain
and
but
in
any
event
they
requested
agenda
time.
We
gave
it
to
them
and
they're
not
here.
So
that's,
okay,
so
I
wanted
to
go
back
very
briefly.
The
ECP,
because
I
sort
of
I
did
actually
get
a
quick
update
or,
and
we
just
weren't-
we
both
lost
it
here
so
any.
F
In
any
event,
we
did
follow
up
with
Danny
about
what
the
status
of
ECP
was,
and
he
does
have
some
changes
that
he
wants
to
post
with
respect
to
terminology,
as
well
as
to
integrate
some
of
Adams
text
and
he's
looking
to
do
another
round
of
updates
in
another
round
of
review
and
then
see
where
we
go
next,
so
we
did
our
homework.
We
just
forgot
our
homework
so
so
at
this
point,
we're
at
the
any
other
business
part
of
the
agenda
and
Oh.
F
N
Hi,
yes,
so
I
created
a
sack
and
Friends
table
and
I
actually
did
some
work
on
Rolly
so
which
is
really.
N
F
F
F
But
you
don't
even
have
to
minute
that
Nancy
I
just
was
and
apologies
for
putting
you
on
the
spot.
We
just
we
knew
that
you
had
done
something
related
to
sockem
and
we
were
right,
yeah
and
and
deeply
appreciate
it.
Help
is
always
deeply
appreciated.
So
is
there
any
other
any
other,
any
other
business
all
right
so
way
ahead?.
D
Right
these
way
ahead.
These
are
the
items
that
are
listed
in
our
charter
still,
which
you
know
so
we
can
probably
do
some
updates
on
there.
But
two
we
have
working
group
last
call
on
coast
with
which
I
actually
feel
like
we're
we're
getting
close
I've
gone
through
the
last
draft.
It
doesn't
seem
to
be
that
much
missing,
I
think
ECP
I
didn't
I,
haven't
kept
up
with
the
rate
of
change
for
for
Danny's
production,
so
I
need
to
do
more
reading
there,
but
hopefully
folks
can
do
some
more
reading
there
I'd
like
more
reviewers.
D
Obviously
Danny's
gonna
do
some
more
changes,
but
then
we
get
into
things
that
I
think
are
a
bit
more
fuzzy.
Well,
Roley
software,
descriptors,
I
think!
Yes,
we
were
according
to
Steven
and
and
I
and
I've
read
that
we've
just
been
waiting
for
that
I,
Anna,
media-type
registration,
I
think
that's
done
so
I
think
we
can
cross
that
off.
D
So
that
gives
us
architecture
and
then,
as
we
look
further
down
the
list,
things
that
I'm
not
so
sure
about
at
this
point,
where
we
stand
I
see
some
puzzled
looks
in
the
room,
but
we
chairs
we'd,
really
like
some
input
from
the
room
as
to
you
know
what
what
are
you
guys
thinking
about
this
list,
what
who's
willing
to
contribute?
What,
where
do
we
stand?.
G
Sorry,
it
is
expired,
so
without
that
we
don't
have
the
information
elements
to
construct
the
envelope
for
a
standard,
yank
push
notification
statement
that
would
then
be
able
to
incorporate
everything
we
define
for
yang
models
or
find
interesting
for
from
other
young
models
to
be
pushed.
We
aesthetic
notifications
for
second,
and
so,
if
we
want
to
go
ahead
with
that
either
we
do
a
arbitrary
data
model.
So
to
speak.
G
So
that's
that's
my
comment
on
the
last
item,
but
also
there
was
a
other.
Let's
call
it
blocker
and
there
was
a
heart
delay
on
working
group
last
call
init
conf.
So
this
is
finally
out
of
the
way
so
giving
that
I
have
time
again
at
some
point.
In
my
life,
I
could
start
a
data
model
without
information
model,
and
then
we
do
can
do
it.
The
other
way
around,
but
I
cannot
give
guarantees.
How
far
this
work
will
go
before
the
next
meeting
son.
F
G
Idea
is
that
yang
has
so
much
semantics
defined,
there's
so
many
things
out
there
that
are
defined
by
yang
models,
and
some
of
them
are
interesting
ii
to
be
circumcised,
I'm
a
tree,
and
if
we
have
the
second
statement,
concept
defined
and
the
information
had
the
information
model.
So
if
we
have
a
standard
notification
envelope
like-
and
there
comes
other
complimentary
work
like
the
bundled
messages
structure,
that
is,
that
I'm,
a
co-author
of
annette
conf
and
with
that
structure,
we
can
create
a
standard
second
envelope
with
metadata
in
it
like
this
changed
at
that
point.
G
This
is
relevant
to
this
device.
This
is
relevant
to
this
post
process,
etc,
etc.
If
the
device
knows
this
or
just
keep
this
MP
and
make
a
rather
raw
telemetry
output,
it's
it's
absolute,
have
a
very
scalable
envelope
for
notifications.
This
is
specifically
tied
to
second.
It
would
be
an
augment
to
the
message
spun
the
definition
that
will
be
produced
in
that
conf
and
with
VR
this
argument
we
would
be
able
to
use
literally
every
existing
yang
module
on
a
device
to
get
second
tailored
telemetry,
based
on
yang
modules.
G
That
is
the
core
concept
of
the
yang
post
transfer
mechanism
that
I
personally
as
a
potential
offer
that
it
hasn't
even
done
a
0-0
draft
yet
envisioned.
So
that
would
simplify
a
lot
stuff
for
I
think
it's
called
panic,
because
there's
our
network
equipment
devices
and
those
have
the
months
and
I've
developed
for
that
and
then
so
we
can
make
use
of
those.
We
are
that
generic
sacrum
glue
draft.
J
J
Four
I
think
I
did
it
for
our
information.
You
know
I'm
elope
and
we
can
work
on
this
direction.
Actually,
I
I
want
to
say
that
we
have
updated
our
our
security
airport,
a
network
to
a
security
post
q88
motor
to
after
for
several
motion.
If
you
look
at
those
draft
actually
the
content
industry
after
its
reality
with
stable
and
we
we
we
have
some
kind
of-
we
have
some
sort
of
confidence
that
those
include
some
useful
information
for
the
network
devices
security.
J
F
E
E
E
F
Okay,
so
I
think
this
gives
us
a
little
bit
of
information
if
everybody
wants
to
take
a
quick
look
at
that
list
of
milestones.
That
I
think
part
of
this
is
you
know,
working
group
chairs
this
week
we're
talking
about
whether
or
not
we
should
update
our
milestones
and
oh
and
this
working
group
is
not
alone
in
this,
but
all
of
our
milestones
are
out
of
date.
At
this
point,
I
do
think
we
are
making
progress
and
we
have
some
things
that
are
coming
down
the
pike,
so
it's
great
anyway
way
forward.
F
F
Also
think
that
the
effort
in
the
hackathon
has
helped
focus
the
working
group,
and
so
we
are
again
a
number
of
the
people
that
are
regular
participants
in
that
are
not
here,
but
we
were
interested
in
those
of
you
who
are
regular
participants
in
the
hackathon.
Maybe
starting
some
conversations
about
what
kinds
of
things
you
might
be
willing
to
collaborate
on
in
the
context
of
IETF
104
I.
D
D
F
Thank
you,
I
think
I
mean
there's,
we've
done
it
a
couple
different
ways.
The
first
time
we
did
it
Adam,
you
know,
put
together
a
rough
plan
of
what
they
wanted
to
try
and
accomplish,
and
a
lot
of
work
went
into
coordinating
in
advance
and
then
I
think
after
that
people
showed
up
and
worked
on
what
they
were
working
on.
So
I
don't
have
an
opinion
on
which
way
we
want
to
do
that.
I.
Just
you
know.
F
N
Jordan,
just
a
procedural
element
for
the
working
group,
more
virtual
meetings
or
planning
sessions
and
doing
things
for
hackathons
are
doing.
For
it.
Works
is
a
forcing
function
if
we
know
that
we're
going
to
be
having
a
call
or
a
WebEx
or
something
and
someone's
going
to
call
us
on
the
carpet
to
see
how
far
along
we
are,
it
is
a
forcing
function,
so
it
it
does
help
focus
and
drive,
work
and
innovation,
because
it's
not
out
of
sight
out
of
mind
until
the
next
meeting
and
then
it's
two
weeks
before
and
you're.
N
Oh
my
gosh,
what
did
I
say?
I
was
gonna,
do
and
I
haven't
done
yet
and
so,
as
a
general
rule,
I
think
that
it
would
be
great.
You
know,
for
working
groups
across
the
IETF
to
try-
and
you
know
instigate
that
and
have
touchpoints
I'm,
not
saying
weekly,
but
every
three
months
is
way
too
long,
and
you
know
you
know,
for
us
vertically
challenged
people
I'd
like
you
know,
they
have
microphones
to
be
lowered
a
little
bit
so.
G
This
is
hang
from
the
floor.
There's
a
good
example
for,
like
yeah
like
a
role
model
yeah.
The
suit
working
group
has
an
interim
hackathon
that
was
really
successful.
It
saved
basically
a
IETF
meeting,
so
to
speak.
A
lot
of
insight
was
gained
a
lot
of
IOT
people.
There
are
new
vendors
that
are
really
interested
in
how
to
deploy
firmware
was
about
on-site
and
we
gave
new
requirements
and
it
was
really
fruitful.
So
maybe
this
is
this.
A
off-site
thing
might
might
have.
F
Yeah
I
think
the
I
just
a
couple
things
I
think
a
Brits
write
interim
meetings.
I
mean
it's
human
nature.
Unfortunately
there
they
are
very
few
people.
I
mean
myself
included
it's
hard
to
keep
things
moving.
If
you're
not
touching,
base
regularly.
I
agree
with
that.
I
think
the
one
advantage
suit
has
is
it's
the
new
hot
sexy
yeah
working
group?
That's
as
opposed
to
one
that's
been
around
for
a
while,
so
that
always
helps
I.
N
Would
guess
I
would
push
back
a
little
bit
on
that
is
that
it
they
don't
just
have
to
be
new
hot
and
sexy
in
order
to
get
people
to
work
they
just
have
to
be
active
and
a
lot
of
the
groups
you
not
just
soup
but
I
mean
obviously
some
of
the
big
ones.
You
know
host
these.
You
know
interim,
you
know
plugfest
hackathons
on
a
very
regular
basis,
sometimes
they're
in
person
and
sometimes
they're
virtual,
and
if
you
have
a
clear
you
know
list
of
things
you're
trying
to
accomplish.