►
From YouTube: DASH Workgroup Community Meeting Aug 24 2022
Description
PR discussion
Welcome to 2 new participants
A
So
these
are
the
notes
we
have
from
last
week
and
the
things
we
covered
and
we
had
a
ton
of
attendance.
Last
week
we
talked
with
keysight
and
mercia,
particularly
about
the
different
pieces
here
with
the
the
config
generator
and
the
baby
hero
test,
and
then
we
moved
on
and
we
talked
with
intel
about
their
sciptf
update
and
we
reviewed
a
pr's.
A
We
reviewed
a
few
pr's
down
here
at
the
bottom,
so
I
was
curious
if
anyone
had
something
they
wanted
to
talk
about
on
this
particular
sheet
or
maybe
we
could
move
into
pr's.
A
Okay,
maybe
we
could,
like
I
said
gerald's
out
today
and
princess
prince
may
join
later,
so
maybe
we
could
move
into
prs.
Did
anyone
have
prs?
They
wanted
to
discuss.
A
B
So
I'd
like
to
bring
up
193,
it's
been
waiting
for
reviews
for
a
week
and
there'll
be
things
backing
up
behind
it.
If
we
could
get
some
closure
on
that,
that
would
be
nice.
That's
mukesh's
pr.
A
Would
it
be
okay
if
we
approve
this
pr.
C
B
So
you're
talking
about
201
marion,
the
one
you
just
submitted,
split
yeah.
I
saw
that
and
I
saw
the
pipeline
run.
Failure
and
I
made
some
comments
in
the
in
the
pr
about
that.
It
should
be
easy
to
fix
that
just
requires
some
additions
to
the
site,
api
query
or
some
changes
to
the
api
query.
B
You
know
which
has
to
provide
the
apis
to
libside.
I
don't
know
if
you've
looked
at
that
code
that
I
I
put
in
there
by
hand
several
weeks
ago.
C
Yeah,
I
know
what
you're
talking
about.
I
haven't
committed
the
changes
yet
but
yeah.
I
know
exactly
what
function
you're
talking
about.
B
Great,
it's
also
an
opportunity
to
consider
whether
we
ought
to
auto
generate
those
those
api
query
codes.
It's
all
pretty
much
boilerplate,
because
this
sort
of
an
example
of
who
changed
the
api,
then
the
build
breaks.
If
it
were
auto-generated,
it
would
take
care
of.
C
B
B
D
B
D
C
D
C
D
C
Yeah
we
just
yeah,
we
just
had
a
primary
goal
to
freeze
the
site
apis
and
I
I
think
we
achieved
that.
C
Yeah
like
today
or
tomorrow,
that's
what
I'm
doing
next.
I
I've
got
a
reply
from
microsoft,
sonic
team
who
said
that
they
believe
they
have
everything
needed
to
work
on
the
sonic
software
on
the
occasion.
So
we
kind
of
informally
declared
code
freeze
on
that
till
the
orc
agent
will
be
released.
C
So
now
we
can
back
to.
We
can
get
back
to
the
dash
repository
and
continue
with
that
yeah.
Sorry,
we
had
like
two
weeks
pause
on
that
now
we
can
get
back
to
it.
B
B
C
B
Right,
I
can
talk
about
a
couple
of
pr's
now's
a
good
time,
and
I
can
take
the
screen
over
okay.
B
So
first
one
I
want
to
talk
about
is
this
one
there's
been
an
outs,
a
long-standing
technical
debt
in
the
make
files
and
docker
files
and
the
way
they
were
constructed,
both
out
of
maybe
a
little
bit
of
sense
of
urgency
and
also
oversights
a
little
oversights
on
my
part.
Some
of
the
makefiles
had
a
lot
of
sudo
commands
in
them
in
order
to
get
around
some
permission,
issues
in
the
build,
because
the
containers
I'll
show
you
where
all
these
sudos
were.
B
B
They
they
write
into
your
file
system
and
I
needed
to
have
the
right
permissions
to
do
that
and
so
they're
sort
of
a
hacky
way
with
this
fixed
perms
build
target
that
was
called
frequently
that
would
change
the
access
to
various
directories
and
there's
nothing
wrong
with
it
technically,
but
from
a
logistics
point
of
view,
every
time
it
does
this,
you
might
have
it
might
interrupt
the
script
and
require
you
to
enter
your
sudo
password
and
hana
pointed
out
to
me
on
the
side
hey.
B
This
is
kind
of
inconvenient
for
automated
workflows,
so
I
finally
sat
down
and
worked
through
it
painstakingly
through
all
the
makefiles
and
all
the
docker
valves
and
figured
out
how
to
solve
this.
The
right
way,
and
so
one
way
one
thing
I
did
was
get
rid
of
this:
the
pseudo
commands
in
the
make
file
and
there's
a
number
of
places
where
it
was
called.
B
B
I
was
the
owner
of
the
file
systems
inside
the
docker
images,
which
is
fine,
but
it's
not
really
appropriate
for
long
term
and
the
reason
they
had
my
name
is
because
the
way
they
were
constructed,
they
used
a
common
technique
in
docker
files,
where
you
build
them
with
your
own
permissions
so
that
you
can
run
them
in
your
own
file
and
that's
that's
great
for
personal
use
but
for
let's
say,
docker
files
that
are
used
in
open
source
most
the
time
the
ownership
is
root.
But
then
you
have
all
these
other
problems
of
it's.
B
It's
got
root,
identity
built
into
it
and
that's
not
always
convenient.
So
I
created
something
called
dash
user
and
a
group
called
dash
users,
and
I
rebuilt
all
the
docker
files
to
have
this
identity
in
there.
It
doesn't
really
mean
anything,
it's
just
something
other
than
rooting
other
than
chris,
so
this
pr
will
basically
fix
all
the
sudos
in
the
make
files
and
also
fix
all
the
docker
files.
So
they
don't
have
my
identity,
sprinkled
throughout.
B
B
Yeah,
it
was
just
one
of
those
technical
bets
that
I
had
to
have
everything
else
off
my
plate
to
really
think
about
it.
It
only
took
one
day,
but
you
have
to
be
in
the
frame
of
mind
to
solve
these
kind
of
problems
where
you
can.
B
No
meetings,
no
big
projects-
okay,
so
that's
that'll,
be,
I
think,
that'll
be
welcome
by
people.
B
B
Automatically,
no
you
don't
what
happens.
Is
the
dash
user
in
in
the
make
files
now
that
when
you
run
them
at
the
right
time,
you
will
grant
ownership
to
allow
dash
users
to
write
into
your
file
system,
but
it
doesn't
require
sudo,
and
you
don't
have
to
this
owner.
The
stash
user
only
really
lives
inside
the
container.
B
Now,
when
you,
when
you
look
at
files,
you'll
see
if
it
created-
let's
say
a
psi
library
header
in
one
of
the
build
steps,
you'll
see
the
owner
dash
user
in
your
file
system,
but
it
doesn't
require
you
to
create
it
now.
B
B
Any
any
other
questions
I
did
a
bit
of
research,
you
know
latest
best
practices
and
this
problem
of
container
image
ownership,
and
it
is
a
problem
and
people
usually
do
things
like
this
or
they
do.
You
know,
shamath
or
or
chow
commands
liberally,
and
it's
kind
of
a
pain
that.
D
A
B
Yeah
and
then
the
other
pr-
let's
see
I
wanted
to
talk
about,
I
think,
is-
might
be
a
more
interest
to
people.
This
one
says
document
dash
as
a
sub
module.
I
have
this.
I
have
this
goal
that
you
can
take
this
entire
dash
repo
and
import
it
as
a
sub
module
into
your
workspace,
so
that
you
can
actually
incorporate
into
your
proprietary,
workflows
and
utilize
the
resources
of
the
dash
repo
in
your
own
environment,
for
building
your
own
proprietary
images
and
running
tests.
B
B
That's
that's
that
that'll
be
in
this
pr,
and
I
tried
to
do
a
quick
start
where
it
just
gives
you
simple
instructions
on
how
to
incorporate
dash
as
a
sub
module
into
your
own,
and
also
I
did
a
test
project
right
here
in
my
own
repo
that
imports
dash
as
a
sub
module,
and
so
it
actually
explains
how
I
did
it.
I
just
imported.
B
It
followed
the
instructions
and
what
you
end
up
with
is
you
have
your
project
and
then
you
import
dash
as
a
sub
module,
and
then
you
can
run
make
selectively
and
reuse
a
lot
of
the
workflows
that
are
inside
dash
already
in
your
own
project,
and
you
can
use
them
selectively.
So
you
don't
have
to
reinvent
the
wheel
and
if
you
import
dash
into
your
project
as
a
sub
module
dash
also
imports
submodules
too,
as
part
of
its
build
structure.
B
So
that's
just
what
you
end
up
with
in
your
repo
it's
kind
of
a
complicated,
and
so
here's
the
steps
for
doing
that.
It's
really
not
that
much!
You
can
probably
do
it
in
5
or
10
minutes.
Just
do
a
scratch
project
and
import
it
and
try
it
out
and
what
I'd
like
to
know.
First,
let
me
show
you
a
couple
of
drawings:
here's
our
tried
and
true
workflow
for
for
the
dash
pipeline
with
all
the
software
workflows
and
the
automated
code,
generators
and
the
testing
we're
all
familiar
with
that.
B
Now,
if,
let's
say
you
wanted
to
have
your
own
lib
psi
generation,
your
own
tests,
etc,
you
need
to
customize
these
areas
here,
the
ones
with
the
exciting
vivid,
colors
yeah.
So
you
need
to
make
your
own
sci
library.
You
need
to
have
your
own
psy
back
end.
You
need
to
make
your
own
scithrift
server
in
order
to
talk
to
your
implementation,
then
you
have
some
kind
of
a
dash
data
plane.
It
might
be
an
soc
chip
might
be
software,
it
might
be
a
mixture.
These
might
all
be
in
multiple
processes.
B
There's
no
assumptions
about
how
it's
done
it's
third
party,
but
you
should
be
able
to
reuse
a
lot
of
this
other
stuff.
That's
already
been
built.
So
the
purpose
of
this
document
is
to
give
you
a
starting
point
to
do
this.
You
can
import
dash
in
and
then
override
and
customize
these
steps
as
needed.
B
So
that's
the
point
of
this
document
that
I'll
submit
as
a
pr
or
that
I've
submitted
as
a
pr,
and
I
encourage
people
to
to
read
it
through
and
undoubtedly
some
questions
will
come
up
like
well.
I
have
a
different
goal
than
what
you
specified
or
that
this
won't
work
for
me
how
about
trying
this.
B
So
I
encourage
that
kind
of
question
and
answer
and
I
can
work
with
you.
You
know
behind
the
scenes
to
try
to
make
this
work
for
your
particular
environment
and,
if
there's
learnings
that
come
out
of
that,
I
would
like
to
then
back
port
them
into
this
general
workflow,
so
that
everyone
can
benefit
from
the
most
generalized
approach
and
you
can
always
customize
behind
the
scenes.
B
A
I
have
a
comment.
I
I
like
how
you
change
the
color
of
the
pieces
that
were
specific
to
playing
the
what
you
need
to
change.
I
liked
that.
B
Good,
but
is
there?
Are
there
use
cases
that
people
have
that?
You
know
might
benefit
from
this,
like
if
people
have
let's
say
a
software
simulation
of
their
data
plane?
B
This
would
be,
you
know,
an
ideal
approach
and,
for
example,
this
might
be
the
approach
we
take
to
incorporate
p4dpdk
implementation,
which
would
be
an
open
source
one
ultimately,
but
there
might
be
vendors
might
have
their
own
software
implementation.
You
could
drop
this
in.
This
also
could
be
a
starting
point
for
your
hardware.
Implementation.
B
There's.
No
reason
why
this
workflow
couldn't
also
work,
for
you
know
the
real
cpu
cards,
so
once
people
have
had
a
chance
to
digest
this,
I
welcome
feedback,
and
you
can
you
know
we
can
do
it
in
the
community.
We
can
do
it
one-on-one,
because
obviously
this
gets
into
the
area
of
your
own
proprietary
backends
and
your
own
workflows
that
I
would
be
happy
to
try
to
make
this
work,
for
you
know
different
use
cases.
B
Yeah,
it's
it's
it's
right
here,
the
top
one
203.
B
So
the
pull
quest
is
there
you
can
you
can
download
my
fork
and
try
it
and
run
through
this
little
yeah
in
this
pull
request
you
can
just
download.
You
know.
I
want
the
best
way
to
view
it.
Okay,
let's
look
at
the
pull
request
itself.
B
Okay,
so
here's
the
pull
request
in
azure
dash.
All
you
have
to
do
is
is
go
to
this
project
and
you
know
get
clone
my
repo
for
dash
and
switch
to
this
branch,
and
you
can
try
it
and
you
know
I
have
a
sample
project
that
actually
goes
through
this
workflow,
so
give
it
a
shot
and
I'd
be
happy
to
if
someone
wants
to
set
up
a
one-on-one
teams
meeting
to
actually
go
through
this
together.
B
Yeah,
and
even
if
there's
a
hardware,
if
you
want
to
apply
to
your
hardware
workflow,
I
might
learn
more
about
your
use
case
and
try
to
generalize
it
so
that
everyone
can
benefit
from.
You
know
the
overall
learnings
and
we
had
a
few
different
people
trying
this
out
and
reacting
to
it.
We
might
come
up
with
something
that's
pretty
useful,
so
that
was
what
I
wanted
to
share
today.
E
A
pull
request,
109,
if
I
can
share
my
screen,
there
was
a
comment
here
about
making
it
pass.
It's
passing.
Oh.
E
The
other
one
is
a
baby
hero
test
so
based
on
all
these
packets
and
sending
the
keeper
live
I'll.
Keep
this
one
as
is,
and
I
will
add
another
one
that
does
not
basically
keeps
the
session
open
and
basically
close
it
as
fast
as
possible.
So
there
will
be
two
variants
and
yeah
this
one.
If
we
want
to
accept
it
and
I'll
add
one
more,
that
will
not
have
all
these
requirements
on
parallel
sessions
and
sending
the
extra
packet
and
it
will
be
as
fast
as
possible
and
we'll
have
both
versions.
A
Who
do
you
need
to
accept
it
like
me,
or
anyone
in
particular,.
E
E
A
Thank
you
was
there
anyone
else
on
the
call
that
had
something
they
wanted
to
talk
about
or
present
in
their
pr
or
issue
or
anything.
D
One
thing
is
that
chris
you
had
mentioned,
you
have
another
pr
on
top
of
the
readme
file
that
we
had
uploaded
for
the
ptf
user
guide.
A
B
Are
you
no,
the
readme
is
in
the
psi.
Ocpsi
pull
requests;
okay,
okay,
that's
the
ptf
user
guide,
so.
D
That's
okay
got
it
like.
We
can
sync
up
offline,
that's
a
different
one:
okay,
okay,
great.
D
Oh
yeah,
it's
I've
got
it.
It's
I'll.
B
I'll
paste
the
link
here,
restaurant
just
so
save
time.
It's
it's
side,
pull
request,
15.59.
B
A
Now
you
know
you
guys
saw
my
feeble
attempt
at
a
pr
here,
so
I
hadn't
had
an
update
about
the
multicast
counter.
Do
I,
since
I
open
it,
do
I
squash
and
merge
on
my
own,
or
do
I
wait
for
somebody
I
mean
I
did
put
j
there.
B
A
A
A
Yeah,
what
I
mentioned
was
that
michael
sigmund
mentioned
that
we
in
verbally
he
mentioned
to
me
that
we're
not
using
it
vfp
had
used
it
for
some
dhcp
advertisement.
I
believe,
but
I
just
wanted
to
capture
that,
for
when
we
talked
about
it
later,
because
we're
planning
to
yeah.
So
this
is
my
pm
attempt
at
a
pr
so.
A
Wait
for
that
one
spell
check:
okay,
good.
I
also
noticed,
if
there's
nothing
else,
to
talk
about
pr
wise,
a
couple
people
I'm
not
familiar
with-
maybe
brian
o'connor,
maybe
could
you
introduce
yourself
tell
us
who
you
are
or
or
we
can
tell
you
who
we
are
if
you'd
like
sure.
F
Yeah
hi,
I'm
I'm
brian
o'connor.
I
joined
intel
about
four
months
ago.
Prior
to
that
I
was
at
onf
for
about
a
decade.
F
I've
spent
the
last
five
years
or
so
as
a
member
of
the
p4
community,
working
on
primarily
the
software
stack
that
runs
on
hardware
targets
like
switching
asics,
so
I've
been
involved
in
the
stratum
project
if
you've
heard
of
that
I've
been
involved
in
pins,
which
is
another
sub
working
group
within
the
sonic
community,
and
I've
been
a
contributor
to
p4runtime
and
sort
of
involved
in
the
gnmi
open
config
community
for
a
while.
F
So
dash
is
a
little
new
to
me,
but
I've
been
kind
of
interested
in
following
along
and
seeing
what's
going
on.
A
It's
nice
to
meet
you
brian,
and
you
know,
usually
what
I
do
at
the
end
of
use
around
thursday
or
so
I'll
recap
everything
I
don't
know
if
you
can
see
my
screen,
but
I'll
give
a
recap
of
who
was
here
what
we
talked
about
blah
blah
blah,
but
at
the
bottom
I
do
have
the
sticky
links
for
what
you're
talking
about.
You
know,
p4
here,
for
example,
pins
pna,
ipk
things
like
that.
A
So,
as
people
join,
you
know
they
can
at
least
get
a
bead
on
what
we're
doing
here
so
nice
to
meet.
You.
B
A
A
So
nice
to
meet
you
brian
thanks
for
coming
and
then
also
I'm
not
familiar
with.
D
Yeah,
I
think,
like
I
introduced
myself,
I
work
for
napa
tech.
We
are
as
smart
and
so
I'm
seeing
the
prospects
you
know
to
integrate
dash
within
the
smart
mix.
So
at
this
time
I'm
more
of
an
observer,
so
yeah.
D
A
Yeah
appreciate
that,
so
that's
all
the
content
I
think
we
had
for
today.
I
can
give
a
half
hour
back.
If
you
guys
don't
have
anything
else
or
you
know,
I
I've
gone
through
what
I
have
so
okay.
Well,
thank
you
for
your
time
and
thank
you
for
everything
I
think
next
week
you
know
we
have
a
couple
things
on
deck
that
we
want
to
talk
about.
I
just
need
to
align
those
things
and
I'll
go
ahead
and
stop
the
recording
for
the
new
people.
A
I
tend
to
share
it
on
the
youtube
channel
and
through
the
meeting
notes.
I'll
probably
add
you
to
the
google
group
just
so
you
can
get
the
notes,
if
you
don't
mind,
and
if
you,
if
you
don't
want
just
send
me
an
email
and
let
me
know
and
I'll
put
my
email
in
the
chat
window
that
would
be
for
brian
and
yeah.