►
From YouTube: DASH Workgroup Community Meeting June 29 2022
Description
PR 127 quick look
sonic-dash.yang quick look
Documentation updates
A
One
this
is
the
dash
community
call
for
june
29th.
Let's
get
started,
and
last
week
we
had
a
great
presentation
from
chris
with
the
continuous
integration
continuous
development
pipeline.
We
didn't
have
a
ton
of
time
for
q
a
and
so
this
week
I
was
wondering
number
one.
If
if
anyone
had
had
a
chance
to
try
the
repo
clone
the
branch
or
give
it
a
try,
provide
any
feedback.
B
D
B
Yeah,
actually
so
everything
the
realty
file
and
I
provided
some
report
to
chris
as
well.
Yeah.
D
A
Okay,
okay,
we
can
move
forward.
Then
I'm
wondering
if
we
have
group
on
the
call.
A
Okay,
did
you
happen
to
cover
your
item
where
you
were
working
with
andy
about
I'm
digging
back
in
my
memory
here.
The
p4
pna
remind
me.
E
Yeah
yeah
yeah
yeah,
I
so
yeah.
Thanks
for
your
reminder.
On
monday,
we
have
pretty
much
agreed
to
everything.
Now
I
just
sent
a
new
pr
to
andy-
I'm
not
seeing
him
here,
but
but
my
understanding
is
we're
we're
either
all
already
agreed
or
almost
there
so
and
I've
sent
you
an
email.
Just
you
know,
while
you
were
talking
a
minute
ago,.
A
E
Certainly
so
when
trying
to
so,
we
have
some
connection
tracking
use
cases
that
should
be
well
unders.
Most
of
us
here
would
know
about
it.
It
requires
us
to
have
tables
that
insert
entries
when,
when
a
packet
misses
the
table,
so
so
that's
typical
contract.
We
are,
we
have
input,
we
have
a
version
of
it
that
we've
implemented
and
that
required
some
changes
in
in
the
pna.
E
Spec
slash
the
p4
spec,
but
so
we
had
sent
a
proposal
to
pna.
E
A
D
Okay,
I'll
follow
up
with
andy.
If
there's
anything,
that's
remaining
as
long
as
all
the
you.
B
D
A
Okay,
so
different
topic.
Normally
what
I've
been
doing
during
the
weekly
notes,
which
is
kind
of
lengthy,
is
you
know
I'll,
go
ahead
and
do
these
upcoming
action
items
each
week
and
what
I've
decided
to
do
kind
of
instead?
Is
I've
gone
ahead
and
created
just
a
github
project
for
myself
to
kind
of
track
these
items
and
then
each
week
I'll
just
be
pulling
copying
and
pasting
from
here
so
that
I
can
keep
track
a
little
bit
better?
C
C
Oh,
oh,
I
see
you
want
to
just
use
this
as
your
as
your.
A
Yeah
because
we've
all
talked
in
the
community
about
all
the
kind
of
the
things
that
need
to
be
done
outside
of
behavioral
model
outside
of
h.a.
This
is
just
in
the
overall
community
and
you
know
like
we're
following
up
with
roop
here.
If
we
could
look
at
these
quickly,
we
we
need
to
write
more
test
cases
with
keysight.
We
have
scithrift
server
integration
with
bmv2
right
and
I
think
chris
is
working
with
marion.
A
C
Oh
yeah,
I'm
not
possibly
a
volunteer,
I
did
volunteer
okay,
but
but
you
know
open
to
all.
A
Yeah,
so
I'm
open
to
so
if
anyone
has
a
passion
for
testing
reach
out
to
chris
and
me
so
then
also
we
need
to
talk
about
this
in
the
behavioral
model
meeting,
but
the
white
box
versus
black
box
testing.
You
know
if
I
remember
this
was
we
were
looking
at
the
tcp
machine
state,
yeah
state
machine,
and
are
we
going
to
be
able
to
test
that?
And
so
that
needs
more
discussion
in
the
behavioral
model
meeting?
A
And
maybe
we
could
prep
to
talk
about
this
on
thursday,
because
that's
our
thursday
meeting
and
then
we
think
the
dash
container
will
be
ready
from
gohan
and
end
of
august
and
then,
of
course,
we'll
need
overlay
tests.
So
this
was
kind
of
my
list
that
I'd
be
pulling
from
instead
of
sending
the
other
one
that
you
just
saw
weekly.
If
that's
okay,
does
anyone
have
anything
else?
They'd
like
me
to
add.
C
I'll
just
say
that,
on
line
eight,
I
think
marion
agreed
last
thursday
to
do
an
initial
v-net
overlay
case.
C
They
kind
of
a
rough
draft,
and
that
I
volunteered
that
we
would
sort
of
formalize
it
to
automate.
It.
A
G
C
G
I
don't
know
christina
if
you
want
to
drag
that
one
here,
sorry
to.
G
No,
no
sorry
to
sound
like
a
broken
record
in
terms
of
you
know
we're
still
waiting
for
the
document
on
you
know
telemetry
and
you
know
metering
and
stats
and
so
forth
the
counters.
I
think
whether
whether
we
should
basically
track
it
here
and
then
probably
assign
the
ata
to
see
when
when
we
should
see
that.
C
A
Right
and
and
to
deliver
the
news
to
you
guys,
we
actually
have
been
going
through
and
vetting
the
list
of
counters
that
we
think
we
need
and
whether
it's
a
global
or
whether
it's
a
per
eni
counter
and
then
trying
james
grantham,
has
been
working
on
this
and
trying
to
transform
it
into
something
that
we
can
publish
here.
So
what
we
don't
want
to
do
is
you
know,
throw
out
250,
counters
and
say
here
we
go
figure
it
out.
You
know,
that's
not
what
we
want
to
do
so.
A
We've
been
going
through
trying
to
figure
out
each
one.
Do
we
need
it?
Do
we
know
what
it
is?
Do
we
need
it
because
we're
going
off
of
our
current
implementation
and
then
moving
the
bits
forward
that
we
believe
we
need,
so
we
have
been
working
on
it
in
the
background,
just
fyi
and
as
soon
as
we
can
can
get
it
out,
for
you.
C
G
One
other
thing
you
know
I
don't
know
again:
there
have
been
several
times
where
we
we
had
questions
about
scaling
or
any
other
functionality
right
and
then
we
went
ahead
and
you
know
deferred
it
to
saying
hey.
There
is
some
sort
of
a
poc
internal
poc
at
microsoft,
going
on
for
the
appliance
use
case
for
serious
and
then
and
then
we
will
have
the
upon
the
end
of
the
poc
and
then
once
we
have
the
results
of
it.
G
We'll
have
a
clarity
and
we'll
come
back
to
you,
and
then
you
know,
share
you
the
results
and
give
you
all
the
the
answers
to
certain
questions
that
we
had
as
far
as
scaling
or
or
any
other
questions
are
concerned.
So
I
guess
what
I
want
to
really
see
is
that
when
do
we
think
that
that
is
ending?
In
other
words,
when
should
we
expect
certain
results
of
those
plcs
to
be
shared
so
that
people
have
a
idea
that
you
know.
H
G
A
Yeah
yeah,
no,
I
agree,
and
I
know
that
right
now
james
has
been
creating
the
so
we've
been
reviewing
internally
the
the
sheets
and
and
what
we
think
we
can
deliver,
and
so
we
are
going
still
going
through
the
proof
of
concept
I'll
get
with
him
today
and
touch
base
back
on
this.
I
don't
see
him
on
the
call,
but
they've
been
so
focused
on
getting
this
proof
of
concept
out
the
door
that
it's
all
hands
on
deck.
C
Well,
it's
covered,
I'm
not
sure
if
it's
covered,
I'm
working
on
I'm
at
a
stopping
point,
I'll
get
some
feedback
from
marion
and
he
said
he
would.
You
know,
get
respond
to
me
at
some
point.
So
I
don't
know
what
that
will
be.
Okay,
you
can
change.
F
Yeah,
so
right,
so
if
I
may
add
to
that,
there
is
a
couple
of
standard
site
apis
that
need
to
be
implemented
pretty
much
by
hand
stuff,
like
capability,
query,
api
query
and
so
on,
which
is
missing
right
now.
So
I
have
a
list
from
chris.
Basically
from
the
compilation
results
of
scientrift,
I
will
do
that
and
then
we
will
have
the
full
support
for
that
side.
Rift
requires.
It
will
take
me
about
a
week.
F
It's
not
really
done
well,
I
can
do
that,
but
they
are
actually
used
like,
for
example,
psi
api
query.
It
is
used
because
it
is
querying
the
actual
callback
functions
from
the
implementation
that
will
be
used,
got.
C
It
got
it,
you
need
to
return
a
pointer,
how
about
the
type
conversion
errors.
I
F
In
yeah
I'll
need
to
follow
up
on
that
offline,
I
think
it's
not
really
either.
I
missed
something
or
it's
not
really
related
to
the
auto
generated
code,
because
it
seems
like
it's
not
from
the
files
that
are
able
to
generate.
C
A
Right
so
I
started
reading
through
this,
and
you
know
trying
to
figure
out
where
I
could
make
a
comment
or
something
like
that,
and
so
let's
see
this,
this
is
great
and
it's
very.
This
is
going
to
take
some
time.
I
guess
is
what
I'm
trying
to
say
to
everybody
is.
This
is
probably
going
to
take
a
good
15
minutes
to
read
through
I'm
sorry,
I'm
doing
this
wrong,
but.
A
C
G
C
To
find
them
and
fix
them,
I
did
a
but
we'll
see.
Oh
no.
C
C
A
Yeah
yeah
so
see
we
started
here.
This
is
the
right
one
right,
chris,
yes,
yeah
and
I
I
also
started
reading
through
this-
and
I
made
it
to
about
just
after
here,
so
everybody
this.
This
will
take
a
little
bit
of
maybe
15
minutes
or
so
to
read,
but
I
think
it's
pretty
valuable.
Did
you
want
to
talk
about
it
at
all?
Chris.
C
A
A
Then
the
clean
up
steps,
and
then
this
is
kind
of
where
I
ended
here,
and
so
I
need
to
get
through
the
rest
of
this
today.
C
A
C
That's
ongoing
obsession
of
mine
right
now,
but
but
I
got
pretty
quick
results
early
on
they're
already
in
the
pull
request,
the
the
initial
p4
code,
compile,
which
I
was
demonstrating
catching
errors.
It
now
only
takes
three
min
three
thirty
seconds
to
get
that
result
in
the
ci.
C
I
did
that
by
using
a
standard,
p4c
container,
that's
already
available
in
docker
hub,
instead
of
using
our
what's
all
the
hands,
instead
of
using
never
seen
that
that's
really
cute,
instead
of
using
the
larger
docker
image
that
has
like
everything
in
the
world
in
it,
I
just
pulled
the
standard.
One
makes
our
our
image
smaller
and
it
also
loads
into
the
ci
runner
much
more
quickly
like
23
seconds,
instead
of
two
or
three
minutes.
C
So
you
get
ci
results
on
the
p4
code
in
30
seconds
and
that's
the
thing
that's
usually
going
to
break
the
soonest.
So
I
think
that's
a
good
step
forward
and
I
also
shrunk
the
main
docker
image.
By
about,
I
don't
know
one
and
a
half
gig,
just
by
removing
the
grpc
source
code
from
it.
So
I'm
I'm
working
on
things
like
that
to
get
the
images
squeezed
down.
A
C
D
A
Thank
you
any
questions
for
chris.
While
he's
on
that
train
of
thought.
B
Yeah,
I
I
I
have
so
yeah
so
yeah.
I
tried
these
things
yeah.
I
can
see
that
the
docker
image
12.4
to
9.5
gb
drastically
reduced
and
things
are
yeah
compilation.
Time
is
also
it's
now
pretty
fast
thanks
chris,
it's
really.
We
are
in
the
right
direction
in
the
optimization
way.
A
That's
great
thanks,
you're
welcome
and
then
and
then
you
have
some
optional
things
down
here-
that
people
can
do.
C
Yeah,
I'm
just
breaking
down
the
make
file
you
don't
have
to.
If
you
go
to
the
very
you
can
go
back
later
to
the
topic.
I
just
throw
like
two
or
three
commands
to
try
this
out.
You
don't
need
to
read
all
this
to
get
started.
I'd
always
like
to
have
something
that
is
just
you
know,
type
a
few
commands
and
go.
A
C
It's
a
great
question,
and
maybe
I
didn't
clarify
this
entire
workflow
is
designed
to
you:
can
do
it
either
manually
step
by
step
or
with
the
summary
instructions
you
can
automate
it
with
your
own
scripts
locally
or
you
can
run
it
in
the
cloud.
C
So
the
goal
is
the
typical
real
life
week
to
week.
Kind
of
workflow
is
someone's
writing
p4
code
and
writing
tests
and
they're
doing
this
all
in
their
development
environment
for
long
periods
of
time
and
then
get
it
working
to
their
satisfaction,
so
it
can
be
done
totally
manually
or
with
your
own
script
and
then
eventually
you're
going
to
submit
a
pull
request
to
the
community.
Presumably
that's
the
end
goal
so,
while
you're
working,
hopefully
you've
got
your
test
working,
you
would,
you
know,
commit
it
and
push
it
to
your
own
fork.
C
C
So
once
you
pass
that
gauntlet,
then
you
can
do
a
pull
request
to
the
dash
repo
and
the
ci
will
run
again
on
the
pull
request
itself
before
it
ever
even
makes
it
into
the
dash
repo.
So
a
maintainer
will
look
at
that
and
go
oh.
It
passed
or
it
didn't
pass
right,
and
so
it's
kind
of
like
a
series
of
checkpoints,
but
primarily
people
do
this
in
their
own
workspace
or
in
your
own.
Let's
say
corporate
environment.
C
C
G
C
Basically,
yeah,
it's
like
a
cla
check,
but
sonic
sonic
is
pretty
picky
about
it.
It
immediately
yeah
immediately
flags
it
in
your
pull
request,
and
I
made
that
mistake
last
week
because
I
don't
commit
that
often
so
I
had
to
reach.
I
had
to
sign.
It
sounds
good.
Thank
you.
Does
that
answer
your
question.
It
was
kind
of
a
long-winded
answer.
No.
C
Yeah
and
if
that's
not
clear
from
the
beginning
I'll
go
back
and
add
that
I
do
have
a
goal
of
getting
this
merged
and
accepted
pretty
soon,
because
there's
several
things
that
that
are
depending
on
it,
one
is
marin,
wants
to
add
more
read
me
content
and
then
the
other
is
I
want
to.
I
want
to
start
filing
issues
against
the
model
itself,
the
bmv
2
model.
C
Can
we
set
a
goal
for
when
this?
You
know
people
can
say
it's
good
enough
to
me.
I
know
marion
already
said
it
looks
good
to
me.
He
approved
it.
I've
had
two
other
people.
Madhu
and
alex
bortock
have
both
run
through
everything
and
it
seems
to
work.
So
we
can,
you
know,
keep
this
open
for
review
or
just
take
a
chance,
and
you
know,
merge
it
and
then
go
forward
from
here.
There's
no
objections
to
any
of
the
content.
I
don't
think
I've
broken
anything
with
this.
A
Right
so
I
guess:
does
anyone
on
the
call
object
to
just
moving
forward
with
it,
or
would
you
like
more
time
with
it.
A
Sorry,
I'm
looking
around
on
the
screen
because
I'm
looking
at
multiple
monitors
trying
to
control
all
the
links
here,
but
so
I'm
not
really
hearing
a
no
yeah.
D
We
should
merge
it
and
yeah.
C
C
We
want
to
get
the
environment
ready
so
that
when
we
start
using
that
it's
using
a
consistent
environment
and
right
now
by
understanding
from
various
comments,
is
that
marion
has
a
fork
that
he's
been
working
on
for
that
and
others
are
going
to
be
sharing
it.
I
want
to
bring
everything
up
level
everything
up
so
that
we
don't
have
all
these
very.
C
A
All
right
and
then,
as
you
know,
prince
had
submitted
this
I
didn't
didn't,
know
other
than
hanif.
If
anyone
had
a
chance
to
go
through
this
here,
but
I
just
wanted
to
bring
it
to
everyone's
attention.
I
A
Awesome,
yes,
a
ton
ton
of
work,
so
it's
awesome
and
then
what.
C
Oh
prince,
while
you're
here
I
know
that
mercha
who's,
not
on
the
call
he
he
had
made
some
comments
about
the
json
schema
because
we're
we're
trying
to
create
test
cases.
C
I
I
thought
we
have
it
covered
in
the
hld
in
the
la
in
the
last,
commit.
I
C
With
an
example,
but
it
was
an
example
right:
it's
not
the
final.
I
Oh
yeah,
I
think
I
think
that
we'll
have
to
work
together
to
create
the
full
configuration.
I
So
basically
that's
exactly
the
the
schema
which
we
can
just
add
few
more
of
the
configuration
based
on
I
I
see
I
think
your
point
is
like
okay.
Axles
are
missing
and
those
things
right,
yeah
yeah.
C
Well,
yeah,
let
me
let
me
try
to
reframe
the
the
ask.
So
a
lot
of
us
are
working
sort
of
from
the
bottom
up
in
terms
of
the
overall
testing
and
we're
working
from
the
data
plane
on
up.
So
we
have
our
psi
apis,
but
at
some
point
we
want
to
write
test
cases
that
are
general
and
we
want
an
intermediate
representation,
which
json
seems
like
the
right
way,
similar
to
how
sonic
uses
json
for
config
import
and
export.
C
C
So
I
would
I
would
ask
for
some
some
more
priority
on
that
task.
I
know
it
doesn't
really
help.
Let's
say
the
sdn
controller
side
of
things,
but
it
really
is
a
kind
of
like
a
needed
item
for
people
testing
from
the
middleware
level.
I
Masonic,
I
think
we
have
the
basic
one.
I
I
need
to
go
and
look
at
what
needs
to
be
extracted.
C
I
I
just
to
emphasize
oh,
I
appreciate
that
prince
I
want
to
emphasize.
This
is
not
like
just
an
informal,
you
know:
here's
here's
an
example.
It
needs
to
be
like
attract
first
class
citizen
right.
It's
a
schema
that
will
all
will
build
a
significant
amount
of
work
on
top
of
so
from
my
point
of
view,
it's
more
important
than
gnmi.
C
A
Got
it
got
it
and
then
probably
the
last
thing
I
had
is
so
michael
miele
and
I
have
been
working
to
break
out
the
v-net
to
v-net
scenario,
away
from
it
used
to
be
called
the
sdn
transforms
document
and
it
was
in
the
general
folder
and
it
kind
of
described
the
scale,
the
apples,
the
routing
the
kind
of
everything
in
one
document
and
we're
trying
to
pull
it
apart,
so
that
the
the
items
that
are
common
to
all
scenarios
stay
in
the
sdn
transforms
document
and
then
we'll
break
out.
A
You
know
each
scenario
into
its
own
folder,
so
you
know
in
the
folder
structure
of
dash
we
had
done.
Hopefully
everyone
can
see
my
screen
documentation
and
inside
a
documentation.
We
had
placed
the
the
different
scenarios,
and
so
you
know
ideally
v-net
to
v-net
will
fall
here,
while
everything
that's
common
to
all,
the
scenarios
will
fall
into
general.
So
that's
what
we're
doing
right
now
and
of
course,
adding
some
content
to
v-net
to
v-net
scenario.
G
G
One
yeah
one
thing
just
I
have
for
my
clarity
sake.
I
think.
Whenever
we
talked
about
v-net,
you
know
we
just
said
v-net
and
then
then
we
started
saying
we
need
to
win
it.
So
whenever
we
said
v-net,
we
basically
also
have
another
use
case
which
has
been
appearing
right.
G
So
by
saying
v-net
I
always
understood
it
was
probably
intravenous.
In
other
words,
you
know
the
vms
belonging
to
the
same
unit,
they're
talking
to
each
other
and
so
forth,
and
then
and
perhaps
being
appearing
was
something
different.
So
I
think
it
will
be
probably
good
to
really
explain
the
difference
between.
I
don't
know.
Maybe
I'm
the
only
one
who
is
a
little
bit.
You
know
confused
on
that
one.
So.
G
It
but
then
some
people
said
no.
The
v-net
use
case
also
covers
the
winner,
and
then
I
said
okay,
so
then
you
know
maybe
my
understanding
was
too
different
and
then,
if,
if
v-net
covers
the
same,
you
know
use
case
of
inner
peering
then
do
we
really
need
two
of
them
right
or
should
we
basically,
you
know,
combine
them
together.
So
just
you
know.
A
Yeah
yeah,
hey
jay
jay,
I'm
I'm
not
sure
how
you
pronounce
it.
Jay
you're
on
the
phone
jay
works
with
michael
over
in
sdn.
Is
that
something
you
can
take
up
and
take
back
to
michael
to
to
kind
of
work
to
get
an
answer.
A
Work
with
I'll
work
with
j
on
that,
so
j
is
new
to
our
group
here,
but
this
is.
This
is
kind
of
what
we're
doing
where
we're
showing
the
pipeline,
but
what's
interesting
down
here,
is
we're
trying
to
describe
the
steps
and
the
different
tables,
and
you
know
that
the
process
here
so.
H
I'm
correct
sorry,
if
I
interrupt
question
this
is
michael
really
here
we
use
the
the
the
example
that
the
json
schema
that
that
you
know
prince
and
chris
from
correct
were
talking
about
now.
H
Obviously,
this
is
an
example,
but
this
could
be
the
the
screamer
that
would
be
adopted
once
completed,
and
so
here's
just
the
the
basic
example
that
we
have
from
the
hld
document,
sonic
actually
document
that
we
are
using
here
in
the
in
the
scenario
and
show
them
that
from
clarifying
what
this
schema
means
in
terms
of
tables
graphically,
so
people
will
have
a
better
understanding.
H
Obviously
this
will
be
the
scenario
that
will
contain
also
more
actionable
steps
in
order
for
the
scenarios
to
be,
you
know
complete,
including
perhaps
the
the
effective
at
the
actual
schema,
and
then
there
are
other
scenarios
that
will
follow
the
same
format.
A
A
And
I
know
yeah
I
know
prince
and
others
have
brought
it
up
that
it
needs
to
stay
in
sync
for
sure,
but
we
were
trying
to
appeal
to
the
person
who
may
not
go
read
the
p4
right
away
and
just
kind
of
give
a
look
at
what
we
you
know,
a
visual
representation
of
what
would
be
happening
and
if
anyone
has
a
better
idea,
please,
like
call
me
I'll
give
you
my
cell
phone
so
and
that's
really
all
I
had
for
today
guys
does
anyone
else.
I
see
yeah.
A
A
Like
a
rock
concert,
we
just
put
our
hands
in
the
air.
Okay,
sorry,
cigarette.
A
So
that's
really
all
I
had
for
today
was
to
go
over
these
things
and.
D
This
vr
is
it
public
or
is
it
still
private?
This.
A
We're
gonna
merge
it
in
at
some
point
here.
A
Yeah
we
would
put
this
in
which
folder
the
data
plane.
C
A
Yeah
no
problem,
okay,
and
so
that
brings
me
to
my
last
point.
You
know
so
michael
who's
been
an
excellent
help
to
me,
is
taking
a
vacation
and
won't
be
with
us
for
a
while,
and
so
I
want
to
say
goodbye
to
michael
and
thank
you
for
his
help.
C
D
H
I
want
just
a
couple
of
words
christina
I
want
thank
you
all
of
you
guys.
Thank
you
christina
for
allowing
me
to
work
on
this
interesting
project.
I
learned
a
lot.
I
start
with
as
blank
slate.
I
learned
a
lot,
so
it's
really
an
interesting
project
and
just
some
point
in
time.
I
need
to
move
on
other
tasks,
but
I
hope
that
I
can
come
back
soon
and.
H
Perhaps
I
will
join
you
again
and
until
then
I
wish
you
good
luck
and
I
will
follow
you
because
it's
open
source,
so
I'm
all
forked
out.