►
From YouTube: Network Service Mesh WG - 2018-07-20
Description
Join us for Kubernetes Forums Seoul, Sydney, Bengaluru and Delhi - learn more at kubecon.io
Don't miss KubeCon + CloudNativeCon 2020 events in Amsterdam March 30 - April 2, Shanghai July 28-30 and Boston November 17-20! Learn more at kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects
A
B
A
B
A
Okay,
so
as
always,
we
have
the
upcoming
event
coming
up
the
and
just
over
a
month
from
now
the
cloud
native
network
function
seminar
at
the
open
source
summit.
So
again,
if
you
are
signing
up
for
the
open
source
summit
and
if
you
are
able
to
make
it
make
sure
you
check
the
appropriate
box
while
you
are
registering-
and
that
will
be
in
thank
Coover.
A
A
A
All
right
so
for
for
the
action
items
so
I
see
that
someone's
filling
filling
out
items
so
I
will
go
to
the
bottom.
If
we
can
do
this
with
some
teamwork,
that
would
be
great,
so
Ducker
docker
hub
images,
so
I
recall
that
we
had
a
little
bit
of
work
towards
this.
If
I
recall,
do
you
remember
you
know
what
the
car
status
is?
It.
B
In
the
network
service
match
all
the
committers
for
for
the
network
service.
Mystery,
though,
should
have,
and
so
that
part
of
the
you
know
road
has
been
cleared
now.
We
just
need
to
publish
docker
images
to
that
right,
doctor
home
repo,
a
docker
hub
repo
upon
merge
and
I
think
we'll
be
good,
but
there's
still
a
little
bit
of
work
done
to
publish
those
images.
It's
not
entirely
clear
to
me
make
Travis
to
run
after
the
merge
in
order
to
do
so.
Maybe
someone
has
a
better
notion
of
how
that
works.
A
A
That
the
watches
master
master
branch,
which
is
not
as
good
as
having
Travis
fleshed
it
out,
but
that
is
an
option.
Another
one
is:
we
could
inject
one
of
our
credentials
into
into
Travis
if
Travis
cases
ability
to
do-
and
it
does
then
we'll
make
use
of
that
and
just
do
a
docker
push
directly
from
from
the
script.
So
I
think
those
may
be
actual
option.
A
A
B
D
B
C
D
D
B
D
D
A
Well
so
Tom
Herbert
put
together
a
PR
for
us
documenting
the
the
startup
process
and
so
he's
so
I'm
in
the
process
of
of
reviewing
that
we'll
go
ahead
and
work
to
get
to
to
get
that
in.
If
anyone
wants
to
help
review
or
run
through
it,
that
would
be
greatly
appreciated.
The
link
is,
and
the
isn't
in
the
network
service
mesh
github
in
a
in
a
pull
request
and
skip
the
little
request
number
right
now.
So
the
pull
request
is.
E
Yeah
so
let's
see
that's
ticket
165
right
can
y'all
hear
me
all
right.
Yes,
I'm
up,
he
doesn't
mean
it.
Okay,
so
ticket
165
is
kind
of
where
we're
drafting
that
and
working
on
something
similar
to
the
network
service,
mesh
use
case
Google
Docs
and
for
the
specific
use
case
that
we're
talking
about.
E
Oh
thanks,
listen
I
already,
did
it
so
the
ticket
165,
so
this
is
going
over.
The
CNCs
CNF
project
is
the
first
part
to
the
two
main
goals
that
we
have
our
performance
comparison
between
DNS
and
VN
s.
So
the
aims,
whether
that's
KVM
directly
or
using
something
like
open
sack
for
the
orchestration
and
and
then
docker
for
containers
and
then
eventually
one
is
the
orchestration
pieces
so
that
we
work
to
Brunetti
size,
N
and
we're
running.
C
E
So
the
V
CPUs
case
has
a
lot
of
needs
from
what
y'all
are
doing
in
the
NSM
that
seem
to
be
aligned
with
some
of
the
existing
goals,
like
the
in
the
NSM
use
case
document
the
LT
traffic
for
cap
lap
controller,
and
that
covers
some
of
the
items
that
we're
trying
to
do
like
passing
l2
frames
across
the
pod,
so
I've
added
some
of
that
in
as
far
as
what
do
we
need,
if
you're
looking
at
the
ticket,
what
do
we
need
there?
So
multiple
network
interface
support
different
types
of
interface.
E
We
want
to
be
able
to
do
PCI,
pass-through
and
other
things:
SRA,
V,
vs
user.
The
transporting
LT
frames
like
I
was
just
saying,
and
then
one
item
that
I'm
not
sure
if
it's
it,
maybe
somewhere
and
I'm,
not
saying
it
as
visible,
but
it
would
probably
tie
in
with
how
it's
how
we
implement
these
cases
more
fine-grained
control
over
resource
utilization.
So
we
when
we're
bringing
these
up,
and
we
need
a
network
function
to
be
isolated
with,
say
one
core,
and
we
are.
E
E
So
that's
not
as
clear
to
us
on
how
that
would
tie
in
with
bringing
it
up
and
I
think
so
the
other
part
would
be
ensuring
that
part
of
what
we're
doing
in
our
initial
testing
we're
having
to
do
the
work
outside
of
kubernetes.
You
have
to
come
in
and
and
run
things
to
get
stuff
connected.
So
this
is
all
the
networking
there's
also
the
stuff
about.
Looking
at
the
CPU
pinning
or
whatever
you
pass
so,
of
course,
using
kubernetes
principles
and
everything
else
and
tied
in
to
how
you're
orchestrating
all
of
this.
E
So
the
draft
is
there
we're
still
working
on
it
and
gonna,
keep
updating
on
that
165
and
add
in
where
the
blockers
so
that's
kind
of
after
the
overview,
which
is
pretty
close
put
in
some
diagrams
and
other
stuff.
Maybe
a
link
to
a
more
extensive,
detailed
talk
that
we're
working
on
and
then
we're
our
we
thing
were
blocked
as
far
as
kubernetes,
so
I
think.
E
A
B
Not
talk
to
them
a
quick
update
on
the
stuff,
that's
going
on
with
the
conversations
about
companies
works,
so
we
are
crying
we're
trying
to
figure
out.
That's
actually
the
right
vehicle
cig
network
think
so
there
are
some
other
folks
who
are
asking
interesting
questions
around
that
on
the
P
dark,
we're
going
to
talk
to
some
of
them
and
see
what
their
thoughts
are,
but
our
overall
position,
I,
think
and
I.
What
I
confer
that
this
is.
B
B
A
A
A
A
A
B
And
so
I
realized
I
was
converting
existing
plugins
to
use
this,
which
that
it's
very
easy,
when
you
add
a
new
dependency,
you
forget
to
either
initialize
and
close
your
dependencies
properly
and
it's
also
very
easy
forget
to
set
up
proper
defaults
for
them
within
the
mechanisms
that
create
a
new
plug
new
Verte
and
it's
this
is
that
plug-in
and
so
I,
a
deaf
stool,
Oh
burn.
B
That
being
said,
so
something
is
wrong
and
I
don't
know
what.
But
something
is
wrong.
You
know
the
goal
being
to
take
as
much
work
out
of
against
the
developer
as
we
reasonably
can,
so
they
don't
have
to
do
as
much
work
and
then
the
things
where
we
literally
can't
figure
it
out,
because
I
can't
figure
out
what
your
default
penance.
You
should
be
to
fail
as
early
and
clearly
with
as
clear
an
explanation
of
what
to
do
to
fix
the
problem,
as
is
humanly
possible.
A
B
B
B
There
we
go
so
you
know
this
is
a
quick
narrative,
introduction
and
effectively.
The
goal
here
was
I
had
two
things
that
people
giving
me
is
very
good
feedback.
What
is
this
is
everything
is
easier
in
a
story,
and
the
second
is
animation
makes
it
very
hard
to
read
the
slides,
even
though
it
can
be
cool
for
presenting
them,
and
so
you
know,
I
started
with
a
narrative.
Isis
Italy
meets
sera
sera.
B
Is
writing
kubernetes
app
to
be
deployed
in
the
public
cloud,
and
one
of
the
Politan
serves
that
needs
secure
access
to
a
corporate
intranet
right
so
from
service
point
of
view,
her
needs
look
like
there's
her
pod
there's
the
corporate
end.
On
the
other
side,
she
needs
an
l2,
l3
connection
between
them
and
there
needs
to
be
some
kind
of
security
in
there,
but
that's
not
really
something.
She
cares
about
right,
so
and
Sarah
still
once
or
normal
kubernetes
networking
that
that's
important
to
her.
B
You
know,
but
she
would
like
to
be
able
to
sit
and
receive
traffic
to
the
corporate
Internet,
so
Sarah's
definition
of
how
how
do
I
feel
you
were
to
do
this
in
a
way
we're
using?
How
do
I
find
out
that's
for
an
interface
that
connects
to
small
thing,
and
then
you
know
who
defines
that
supplement
and
what,
if
the
sudden
that
is
too
small
in
the
future,
and
how
do
I
define
you
have
to
define
a
second
interface
between
that
subnet
of
the
VPN
gateway,
pod
and
gee?
B
Do
I
have
to
go
insert
routes
for
my
corporate
IDs
into
my
pod?
What,
if
I
need
more
replicas
and
the
sudden
that
is
too
small?
What
if
the
sum
that
changes
do
I
have
to
change
all
the
other
pieces
of
the
puzzle?
You
know
what
happens
when
the
corporate
Enterprise
guys
complain
that
you
know
my
son
that
overlaps
with
some
sudden
that
inside
the
intranet
in
do
I,
have
to
go
redo
everything
you
know
what
am
I
network
guys
REI
P,
something
in
the
Internet
and
now
my
son,
that
is
incompatible
again.
B
B
What
happens
if
my
corporate
intranet
adds
Kidder's
or
prefixes
do
I
have
to
go
update
all
of
my
pods
with
new
routes,
and
at
this
point
you
know
our
little
sort
of
you
know.
Proto
mascot,
pops
up
and
says:
maybe
I
can
help
you
no
I'm,
who
are
you
I'm,
Adrian,
the
the
network
service,
mesh
spider,
what's
network
service
mesh?
Is
that,
like
sto
sort
of
you
know
all
the
cool
things
this
year
does
for
TCP
connections
in
HTTP?
You
might
do
that
for
IEP
Ethernet.
Another
l2
protocols
tell
me
about
your
problem.
B
I
have
lost.
My
told
me
about
your
problem.
Slight
apologies
I
need
you
to
fix
that.
So
you
know
this
is
how
you
do
NSM.
You
know
it
looks
like
services.
We
point
that
out.
You
know
so.
Okay,
just
do
I
need
this
deployment
from
an
APN
gateway
and
we
sort
of
talked
a
little
about
the
init
container.
B
No
interfaces,
no
routes.
How
does
all
this
work?
Well,
okay,
we've
got
three
concepts.
The
first
is
a
network
service,
for
example,
secure
internet
connectivity.
The
second
use
network
service
endpoints,
you
know
it's
a
pod.
The
protect
provides
the
service
you
want,
so
that's
sort
of
like
in
points
for
services.
B
So
is
that,
like
you
know
the
VPN
gateway
pod,
which
it
is
you
know,
the
third
concept
is
the
l2
l3
connection.
You
know
is
that
an
interface?
Well,
usually
it's
instantiated
as
a
kernel
interface,
but
there
are
high-end
use,
get
users
who
want
more
exotic
things,
but
you
probably
want
to
curl
on
her
face.
Okay.
Well,
what
about
subnets?
Well,
we
do
these
as
point-to-point
cross
connects.
So
you
don't
need
subnets.
If
you
want
a
bridge
domain,
that's
a
network
service
and
you
know,
okay,
what
about
routes?
B
Well,
generally
speaking
routes
for
that
are
gonna,
come
from
your
network
service
endpoint,
like
your
VPN
gateway,
you
can
do
other
more
sophisticated
things,
but
that's
the
normal
course
of
events.
Okay,
what
about
new
firewall
pop
the
new
firewall
pod?
They
want
to
add.
Well,
okay,
we
are
a
mesh
and
the
firewall
and
VPN
gateway
pod
work
together
to
give
you
the
network
service
you
want,
which
is
secure,
internet
connect.
B
You
know,
and
you
talk
about.
You
know
what
that
looks
like.
So
how
does
that
work?
So
you
know
virtual
hosts
and
rattles
in
sto
yeah,
so
we
have
something
similar
called
network
service
wirings.
So
a
network
service
wiring
has
a
target
secure
internet
connectivity.
That
means
it
applies
to
l2
and
l3
connections.
Trying
to
reach
that
network
service.
B
You
know
it
has
a
set
of
qualifiers
qualifiers,
say:
ok,
well,
the
source
is
not
a
secure
internet
connectivity
pod.
Then
this
networks
wiring
applies
to
it,
and
then
it
has
a
route
that
connects
to
the
network
service
endpoints
that
how
label
firewall
equals
true.
So
that
ends
up
telling
things
that
are
not.
B
B
Making
the
point
that
there's
no
interfaces,
no
eyepiece
and
no
routes
to
be
configured
do
I
need
a
new
version
of
kubernetes.
Do
I
have
to
use
a
specific,
CNI
plugin?
Well,
no
network
service
measures
completely
orthogonal
the
normal
kubernetes
network.
It
doesn't
use
CNI
and
you
can
use
it
with
your
existing
kubernetes.
Ok,
how
does
it
work
and
we
talked
a
little
bit
about?
Ok.
Well,
you
remember
how
network
service
mesh
works?
Well,
we
have
something
called
the
network
service
manager
that
does
something
similar
that
runs
as
a
daemon
set.
D
A
So,
if
especially,
if,
especially
if
you're
wondering
what
network
service
mesh
is
about
and
simultaneously
I,
even
if
you're,
even
if
you're
consider
yourself
to
be
very
well-versed
in
what
network
service
Somesh
does
go
through
it
again
and
see
if
see,
if
you
can
find
any
any
problems
or
holes
or
anything
that
pops
up
unfulfilled
questions
that
they
come
to
mind
so
like
even
even
I,
haven't,
fulfilled
questions
and
will
often
have
conversations
to
try
to
work
out.
You
know
better
understanding
or
try
to
identify
holes
where
were
where
they
exist.
A
So,
okay
and
the
the
last
document
that
was
added
on
so
that
the
last
pull
request
listed
183.
If
someone
could
be
kind
enough
to
to
share
that
is
a
document
that's
going
in
the
readme
that
describe
what
network
service
meshes
and
I
took
this
particular
document
and
in
basically
two
parts.
So
one
of
them
is
trying
to
describe
the
problem
that
we
said
that
we're
saying
so
the
the
concept
that
we
need
more.
A
The
second
part
of
this
document
is
about
trying
to
explain
what
network
service
mesh
is
through
a
series
of
of
examples,
and
so
the
example
has
a
short
prose
and
it
shows
a
a
sequence
diagram
describing
each
of
the
steps
that
that
would
be
taken
in
order
to
instantiate
a
network
service
mesh
connection.
So
I
just
finished
this
up
yesterday.
There
there,
maybe
if
couple
grammatical
errors
or
typos,
though
definitely
point
those
out
to
me
as
well
as
you
as
you
see
them,
I
want
to
make
sure
that
this
is
as
readable
as
possible.
A
And
if
there's,
if
you
read
this
and
there's
something
you
don't
understand
at
this
particular
point,
it
definitely
feel
free
to
get
a
hold
of
me.
So
we
can
see
about
expanding
the
document.
One
of
the
things
that
I
did
was
I
stuck
with
the
easier
use
cases
at
this
particular
point,
because
I
I'm
worried
about
like
if
we
go
into
more
advanced
topics
as
to
what
network
service
special
do
I'm
worried
about
overwhelming
the
the
the
reader,
and
so
those
more
advanced
topics
would
still
follow
the
same
patterns.
A
But
you
know
you
want.
We
want
to
be
careful
as
to
how
much
we
throw
at
a
person
at
one
time
and
as
a
documentation
gets
written
up
and
then
we'll
we'll
add
links
to
this
as
well
saying
for
further
reading
on
the
architecture
for
further
reading
on
this
type
of
on
these
type
of
network
service
mesh
integrations,
and
so
people
will
be
able
to
find
that
information
as
it
becomes
available.
A
D
A
A
D
A
It's
whatever
the
branches,
so
in
my
case
is
github
f
Kouts
network
service
mesh.
I
and
then
you
open
up
the
the
branch
for
the
commit,
and
then
you
can
do
then
traverse
to
the
document
and
see
the
document
in
its
entirety
through
that
through
that
branch.
So
it's
is
not
as
nice
as
actually
integrating
lyrically
into
the
pull
request.
A
D
A
A
A
Then
it
shows
the
sequence
diagram
as
to
like
the
communication
that
is
done
between
each
system
and
in
terms
of
wanting
a
connection
down
to
actually
injecting
it
and
getting
the
connection
accepted
message,
and
so
so
I've
created
five
of
these
examples.
For
this
particular
document,
scroll
down
a
bit
and
you'll
see
the
second
example
is
connecting
to
services.
The
third
example
is
external
devices.
The
fourth
one
is
a
bridge
service
and
I'm
struggling
to
read
service,
but
it
could
easily
be
any
other
kind
of
bridge
service.
A
You
can
imagine
the
same
pattern
and
then
scroll
down
a
bit,
and
this
one
is
the
distributed
bridge
domain,
where
you
hook
your
pods
up
to
various
various
bridges
and
so
basically
building
on
each
each
particular
example
builds
builds
up
to
reinforce
the
idea
that
an
external
device
is
a
service.
They
a
bridge
is
a
service
hooking
up.
The
bridges
is
just
took
ya,
requesting
to
be
connected
to
yet
another
to
yet
another
service,
and
so
like
really
trying
to
emphasize
that
within
this
particular
this
particular
document
so
anyways.
A
A
D
I
added
an
action
item
for
me,
hopefully
to
actually
take
like
exact,
like
example:
five
in
your
doc
in
Fred's
document
and
actually
implement
it
using
VPP
is
that
they
deplane
and
putting
the
code
together
and
then
documenting
the
steps
forward.
It
might
take
more
than
a
week.
You
know,
but
I
thought
that
might
be
really
useful
since
we
started
we're
starting
to
talk
about
actually
gluing
stuff
to
data
plane.
D
A
D
And
it
probably
won't
be
the
only
one
doing
that
there's
also
another
thing:
I
think
it
was
Lucy
now
that
said
F's
is
there
an
agate
hub
issue
for
this,
so
that
led
me
to
ask
a
broader
question:
should
we
for
these
action
items
and
should
we
have
github
issues
written
for
these,
or
are
we
just
going
to
track
them
here
with
action
items
in
that
in
the
document?
Just
a
question.
A
A
A
One
thing
that
would
be
nice
is:
if
we
were
to
use
github,
it
would
be
nice
to
have
a
label
called
action,
item,
meeting
action
item
or
something
something
that
we
can
just
label
and
and
uses
the
filter
to
bring
up
in
our
in
our
meetings.
The
only
issue
that
I
can
see
with
it
is
as
well
with
that
kind
of
approach,
though,
is,
if
you
want
to
look
at
something
historic
and
say
what
what
do
we
talk
about
on
this
particular
meeting?
A
We
still
have
to
copy
them
over
to
the
document,
because,
once
someone
removes
that
label,
like
the
information
is
still
technically
there
because
github
says
added
label
at
this
time,
remove
label
with
this
time,
but
trying
to
extract
that
and
show
it
off
in
one
place.
There's
there's
no
easy
way
that
github
has
to
do
that.
A
A
Yeah,
so
back
to
the
to
the
example
that
you
were
talking
about,
you
know
even
showing
something
as
simple
as
let's
drop
a
new
interface
into
into
a
into
a
pod
like,
even
even
that
is
very,
very
useful,
like
we
could
drop
us
our
iove
with
that
example,
especially
if
it's
one
that
uses
the
device
plug-in
API,
you
manage
a
limited
number
of
interfaces,
then
that
would
enable
us
to
say
yeah.
This
works
with
with
SRS
our
iove
integrates
with
kubernetes
device
plug-in
API,
and
that
way
that
you
don't
over,
subscribe.
Your
pods
and.
A
Demonstrate
you
know
even
just
even
something
as
simple
as
you
know,
adding
adding
a
servi
or
add
a
radio
service
or
add
adding
and
adding
an
interface.
So
so,
even
something
like
that
is
useful,
but
yeah.
If
you
want
to
try
cycle
number
five,
you
know
by
all
means
and
make
sure
if
you
run
into
any
issues
or
if
there's
any
limitations
or
anything
you
run
into
you
know,
don't
definitely
difficut
hold
of
us
right.
C
C
All
right
so,
as
Frederick
mentioned
earlier,
I
was
working
on
a
very
simple
data,
plane
controller,
which
would
which
should
allow
to
stitch
together
an
SM
client
requesting
for
requesting
network
service
and
NS
e,
providing
that
that
service.
At
this
point
it's
done
by
using
with
interfaces,
and
so
basically
that's
what
I
would
like
to
show
you
right
now.
I,
don't
have
anything
running.
So
this
is
my
test,
bat
or
development
cluster
where
I
run
kubernetes.
So
let
me
start
I'm
starting
the
net
mesh
first,
which
is
the
NSM
code
all
right.
C
So
it's
coming
up
it's
running
now.
I
do
n,
SC
I
start
NS
C,
which
is
the
service
provider
for
that
specific
service
before
before
I
go
further,
I'd
like
to
show
you
that
this
in
a
seaport
has
just
two
interfaces:
loopback
and
the
normal
virtual
Ethernet
interface,
which
is
provided
by
the
kubernetes
api.
C
C
And
there
is
a
gel
PC
serve
your
PC
server
in
front
of
the
actual
data
plane,
which
is
accepting
your
PC
request
from
an
SM
to
accomplish
this
connection.
This
the
server
expects
from
the
NSM
just
to
good
name,
puts
the
name
and
the
namespaces,
and
then
it
goes
talks
to
the
API
and
then
basically
do
the
magic
with
the
dropping
they've
a
spare
and
then
plugging
in
the
right
namespaces
so
know
that
this
one
is
running
and
now
I
need
to
create
the
actual.
This
is
the
actual
and
SM
client.
C
So
it
starts.
As
you
can
see,
there
is
a
neat
container.
It's
running
now
they
need
of
portion
is
done
and
in
the
same
client
is
running
so
since
it's
not
crushing,
it
tells
me
that
basically,
this
it
was
successfully
acquire
network
service
and
we
should
have
it
now.
We
should
have
a
direct
connectivity
between
NSM,
client
and
NSC.
So
let's
do
that
command.
C
If
can
fig
minus
8,
as
you
can
see
now,
NSC
has
an
extra
interface
and
I
used
for
the
interface
name
portion
of
the
name
of
the
client.
So
it's
going
to
be
easier
to
correlate,
which
interface
used
for
which
client
and
I
statically
assigned
the
IP
address.
Just
I
mean
it's
working
progress
kind
of
so
I,
just
statically
assign
it
now
if
I
want
to
check
the
connectivity
being
1
1
1
1,
which
is
the
other
end
on
the
NSM
client.
C
C
So
that's
pretty
much
so
basically,
this
is
the
kind
of
way.
Now
it
seems
like
an
end-to-end
connectivity
and
I
was
planning
to
use
it
and
the
CI.
It
will
be
very
easy
to
test
cuz.
The
actual
image
is
like
a
forty
Meg.
When
I
looked
at
the
VPP,
it
was
three
gigs,
so
I
saw
that
I
mean
pulling
three
gig
from
the
docker
hub
would
be
kind
of
hard.
So
if
there
is
something
simple
as
this
in
the
CI
which
still
tests
the
end-to-end
connectivity,
it's
I
think
it's
a
good
tool.
D
C
C
Definitely
I
can
add
it.
I
actually
I
solicited
some
reviews
like
few
days
ago,
and
this
morning,
I
completed
all
the
code.
So
now
it's
stable
and
it's
running
in
in
the
NSM.
The
only
issue
I
still
have
is
there
is
a
dependency
and
Fredrik
I
was
I
was
hoping.
You
could
help
me
to
sort
that
out,
because
in
my
local
testbed
bill
goes
fine,
but
in
Travis
it's
failing.
So
if
you
could
help
me
sort
that
out.
That
would
be
great
sure.
C
And
the
second
thing,
while
working
on
this
example,
I,
actually
found
a
couple
of
bucks
in
the
net
link
net
net
net
library,
which
is
used
to
handle
Linux
namespaces,
so
I
pushed
two
fixes,
one
got
accept
it
and
it's
already
merged,
and
this
there
is
a
second
one
which
is
still
in
process
of
being
merged.
So
if
you
see
in
the,
if
you
see
in
the
review-
oh
sorry
in
my
PR,
there
is
a
change
in
the
vendor.
It's
not
missing.
Second
fix
which
I'm
still
waiting
for
to
be
merged.
B
A
Ok,
one
last
note
I'm
going
to
I'm
going
to
give
the
recommendation
from
Chris
a
try
and
there's
something
called
project
boards
in
github
and
so
I'm
going
to
go
through
and
open
up
github
issues
for
each
of
these
items
and
see
if
we
can
make
use
of
the
of
the
project
board
to
help
us
organize
that.
So
that
way,
that
way,
we
don't
risk
having
all
of
our
information
stuck
in
a
in
a
document
over
time.