►
From YouTube: IETF110-ALTO-20210312-1430
Description
ALTO meeting session at IETF110
2021/03/12 1430
https://datatracker.ietf.org/meeting/110/proceedings/
A
B
C
A
B
D
C
C
A
G
G
D
A
I
think
for
me
the
discussion
is,
I
will
lead
the
discussion
to
introduce
a
charter,
new
goal
and
a
scope,
and
we
will
you
know,
list
the
proposed
work,
item
and
related
use
cases
and
associated
product
extension
and
then
and
for
the
charter
use
case
discussion.
Actually,
I
will
ask
the
problems
actually
to
champion
the
use
case
discussion
for
each
use
case
discussion.
I
will
give
you
five
minutes,
because
the
time
is
very
limited.
A
A
D
D
E
D
Next
slide,
please,
and
this
draft
is
pretty
mature,
so
this
is
also
you
know.
It's
pretty
much
done.
I
need
to
put
the
shepherd
right
about
and
it's
been
socialized
with
cdni,
so
we're
all
set
on
these
drafts.
The
first
two
drafts
path,
vector
and
unified,
props
and
cdna
requests
are
going
to
move
together
together
as
a
cluster.
D
A
A
Actually
they
really
deploy
auto
protocol
in
their
cdn
networker
and
a
key
component
is
flow
director
and
they
use
auto
protocol
to
optimize
the
cdn
content
delivery
and
also
they
actually
discuss
their
yeah
there's
a
key
work
actually
in
the
connex
conference
meeting,
also
presented
in
itf
and
itf
get
a
good
price,
so
the
second
auto
implementation.
Actually,
this
is
really
a
pork
actually
from
china,
mobile
and
tescent,
and
they
really
want
to
use
auto
protocol
to
expose
the
mobile
and
wireless
information
to
help
to
improve
the
user
experience
of
cloud
gaming
application.
A
So
there
are
qa
park
actually
test
has
been
conducted
and
for
implementation
from
spring
and
t-mobile.
They
claim
to
support
already
deployed
the
author
in
their
ctn
network.
I
haven't
get
a
detailed
information
for
some
reason
and
another
one
is
from
penny
fornica.
Actually,
he
is
a
taco
cdm
provider
and
right
now
they
actually
try
to
integrate
overlay
and
only
and
expose
capability
and
status
from
only
and
they
can
use
this
information
to
optimize
the
cdn
networker.
A
So
they
already
plan
to
move
to
the
production
in
the
second
house
this
year.
Also,
we
have
unicom
open
source.
Actually,
this
is
needed
by
year,
university,
richards
team,
and
it
has
all
been
open-sourced.
It
focuses
on
multi-domain
multi-controller
resource
orchestration
and
they
can
provide
the
service
discovery
and
resource
discovery
in
their
implementation,
also
the
implement
in
the
open
daylight
platform,
and
also
we
list
the
two
vendor
implementation,
which
maybe
is
outdated.
A
E
Just
one
quick
comment
regarding
the
last
slides:
the
last
slide
is:
we
did
have
many
presentations
on
these
implementations
over
the
years.
So,
if
people
are
interested,
they
can
maybe
also
review
the
presentations
we
had
at
previous
alto
meetings.
Yeah.
What
you
say
with
your
summary
is
absolute
correct,
but
just
for
people
who
are
who
have
a
deeper
interest.
We
have
presentations
from
some
of
these
operators
in
the
past.
H
Yes
thanks
one
question
about
this
slide,
so
the
first
one
says
deployment
num.
There
are
two
machines,
so
is
this.
So
what
is
a
deployment
versus
a
research
project?
I
I
I.
I
gather
that
the
sprint
thing
is
probably
a
deployment,
but
the
benox
thing
I
mean
I
that
looks
like
a
study
to
me
if
I'm
getting
with
the
or
proof
of
concept
of
some
kind.
If
I'm
understanding
the
bullets
there.
A
H
B
D
D
A
Can
you
go
back
yeah
for
this?
We
actually,
you
know,
boil
down
the
full
and
propose
the
work
island,
and
actually
we
reached
a
lot
of
gremlins
for
these
proposal
columns.
So
after
idea,
one
zero
nine.
Actually
we
began
to
document
recharge
scope
and
spend
time
to
discuss
the
operation
related
to
the
work
item
and
we
come
up
with
the
first
version
of
the
retired
text
and
at
the
end
of
the
last
year
and
sent
to
the
area
director
and
our
co-chair
for
review
and
also
to
facilitate
the
recharge
discussion.
A
A
In
addition,
actually
we
work
on
some
auto
extension
like
multicast
and,
of
course,
canada,
which
allow
to
convey
multiple
cost
value
in
a
single
message
or
time
dependent
course
value
in
a
single
message
so
for
path,
vector
and
cause
constant
performance
metric
setting
high,
and
we
also
close
to
the
end.
So
what's
the
next
level
auto
working
over.
So
where
do
we
have
the
world?
A
Actually,
we
think
for
the
current
auto
product
actually,
especially
for
auto
deployment,
the
having
integrated
some
new
product
extension
we
work
on
in
in
the
old
working
group,
so
how
these
auto
deployment
integrate
the
many
new
product
extension
feature
and
in
an
in
a
deployment.
So
this
issue
we
need
to
investigate.
A
In
addition,
we
think
the
original
auto
protocol
is
very
limited
for
the
single
administrative
domain.
So
but
we
also
see
many
applications.
Actually
they
need
money
domain
support.
So
we
think
actually
money
domain
setting
is
the
the
the
issue
we
really
want
to
solve
and
for
also
we
actually
see
many
new
application,
new
scenarios,
like
sp1
scenario
and
edge
computing
scenario,
and
also
we
have
a
new
scenario
from
the
china,
mobile
and
passenger
the
cloud
gaming
use
cases.
So
they
really
want
to
introduce
some
new
mexicans
like
a
pop
star
mechanism.
A
A
So
discuss
with
our
auto
design
team,
we
come
up
with
five
proposed
chart
item
for
the
first
item.
Protocol
extension
for
policy
attribute.
We
have
listed
two
use
cases.
One
is
check
rate
after
adaptation
for
cloud
gaming,
which
is
from
china,
mobile
and
peasant.
We
have
another
cellular
nano
information,
exposure
from
nokia
and
for
proposed
product
extension.
A
A
We
think
we
actually
need
to
configure
autoclient
and
a
server,
especially
for
the
auto
server,
and
also
we
need
to
provide
a
performance
monitoring
for
the
auto
server
and
auto
client,
so
better
model
young
data
model
is
needed
for
this
and
item
for
money
domain
setting
support.
We
actually
list
three
use
cases
such
as
the
money
domain
resource
orchestration,
hierarchical
service
function
chain
in
the
distributed
data
center
network.
A
This
actually
has
been
proposed
by
charlie
telecom
on
the
list,
and
also
we
have
multi-domain
sd-wan
edge
recovery
and
the
required
market
command
support.
The
last
one
is
operation,
automation
and
main
use
cases,
hopefully,
and
under
the
integration
which
is
proposed
by
penny
fornica,
and
we
also,
as
we
mentioned
actually
auto
deployment
need
to
be
enhanced
and
we
need
to
think
how
to
upgrade
nano
information
combined
with
the
computer
information.
So
we
also
introduce
a
related
protocol
extension
here.
So
this
is
a
summary
and
any
comments.
A
C
F
F
F
When
the
the
af
received
the
afternoon
qs
profile,
then
the
af
will
change
his
encoding
scheme
that
adaptive
feeder
changes
to
prevent
the
congestion,
because
in
the
ue
side
or
the
client
side,
so
inside
the
case,
the
activity
will
change
the
bit
rate.
Along
with
the
grand
notification,
for
example,
it
remembers
that
I
only
can
separate
one
big
head
then,
where
he
changed
his
appetites
below
the
one
figurehead
if
they
said.
F
F
F
F
F
B
E
B
A
J
Yeah,
I'm
here
hi
everyone
I'll
quickly
present
the
auto
data
model
use
case.
This
is
pretty
straightforward.
If
the
device
where
we
are
implementing
the
alto
client
and
auto
servers,
they
already
support
yank-based
mechanisms
for
other
purposes.
Maybe
we
could
use
the
yank-based
mechanisms
for
configuration,
status,
monitoring
and
all
the
other
setups
for
our
all-pro
client
and
alto
server
as
well,
and
we
would
like
to
produce
a
standard
yang
model
so
that
all
the
advantage
of
yank-based
management
can
be
utilized.
So
the
scope
wise.
J
We
would
like
to
support
both
alto
client
and
auto
server,
especially
in
case.
We
are
also
now
exploring
multi
domain
case
so
server
to
server
communication
would
also
be
a
key
factor,
so
that
will
also
influence
how
our
young
data
model
would
look
like.
We
have
work
to
refer.
One
of
them
is
the
http
client
server,
which
has
been
done
in
netcon
since
alto
uses
http.
This
will
be
a
very
good
reference
for
us
to
build
on.
J
It's
not
just
configuration,
I
think
the
operational
status
and
the
monitoring
part
would
be
very,
very
useful.
The
the
other
use
case
that
you
will
hear
today
also
talk
about
operational
stuff,
where
a
yang
based
mechanism,
which
I
think
itf
has
been
working
and
we
have
the
expertise
on
this-
can
also
be
utilized
for
all
protocol.
J
So
the
work
item
would
basically
require
us
to
explore
and
come
up
with
the
right
framework
for
what
should
be
the
data
model
design
and
the
final
output
would
be
the
young
data
model
for
both
client
and
server.
That's
it
from
my
side.
I
think
at
the
end,
if
there
are
any
questions,
I'm
here,
we
can
take
I'm
done.
Thank
you.
G
B
We
do
have
a
very
nice
work
done
by,
for
example,
separation
already
have
have
the
server
discovery,
and
now
I
think
the
number
one
thing
we
want
to
do
clearly
and
there
would
be
go
to
the
case
of
multiple
server,
auto
servers
in
a
multi-domain
and
here
trust
relationship,
privacy.
Accurate,
would
be
a
major
issue.
So
therefore,
that's
really
the
first
item.
We
think
we
should
get
done
and
the
second
one,
of
course,
is
the
we
want
to
really
do
this.
B
Auto
server,
communications
for
network
information
and
computing
information,
exchanges
and
basis
still,
of
course,
is
in
the
multi-domain
setting
and
number
three.
I
probably
that's
really
most
important.
What
I
think
would
be.
We
want
to
really
discover.
Of
course,
we
have
bgp,
but
the
bdp
is
not
really
fully
application
domain,
so
the
the
signaling
through
bdp
problem
is
very
difficult
and
so
on.
B
So
what's
really
happening
in
this
moment
is
we
want,
essentially,
is
extend
auto
and
to
allow
the
discovery
of
multi-domain
connectivity
and
with
all
these
kind
of
messagings
and
who
connects
to
whom
and
so
therefore,
then
we
can
build
this
essentially
whole
class
relationship.
That's
all,
and
of
course,
there
already
did
quite
a
lot
of
use
cases
and
quite
a
lot
of
documents
and
I'll
be
more
than
happy
to
go
for
details
in.
G
I
I
So
in
this
use
case,
the
the
user
equipment
receives
information
from
two
different
independent
channels.
One
is
directly
from
the
run,
so
it
receives
like
real-time
indications
that
are
reflecting
more
or
less
the
cell
load.
Okay,
and
this
is
to
help
out
deciding
whether
shall
I
connect
now
or
which
cells
shall
I
connect
to,
and,
on
the
other
hand,
from
the
alto
client
it
receives
like
longer
term
information
on
like
cell
kpis,
for
example.
I
I
I
Your
alter
server,
says
okay,
depending
on
this
value,
cost
is
v1
or
v2,
and
then
you
can
the
ue
when
it
needs
to
take
a
decision.
It
just
figures
out
just
fetches
that
indicator
and
says:
okay,
cost
is
this
value
and
I
take
this
decision,
but
this
is
one
example,
and
there
are
several
such
examples
of
indicators
that
you
fetch
directly
from
the
network,
and
that
gives
you
a
clue
of
what
of
whether
the
cell
is
loaded
or
not,
or
whether
the
kpi
that
reflects
the
some
kpis
on
the
network,
that
you
cannot.
H
H
And
they
like
have
it
under
control
or
or
like
each
operators,
their
own
stuff
and
they're
just
running
proprietary
software,
where
the
ues
know
what
to
do.
It's
not
clear
to
me
what
the
use
case
for
alto
is
there
but,
like
I
said
I
know
we're
short
on
time,
so
you
may
not
be
able
to
respond
fully
thanks.
I
Just
very
quickly,
maybe
I
was
not
clear
enough:
the
the
you
that
indicator
that
you
get
from
the
3gpp,
so
that
example
is
standardized,
and
I
am
not
saying
that
this
should
be
provided
by
the
alto
that
you
that's
that's
actually
the
trick.
The
ue
gets
it
this
time
of
indication
directly
from
the
network
on
its
own.
B
So
I
I
think
I
I
like
the
comments
made
by
martin,
so
basically
the
thinking,
but
we
I
understand
this
one
is
the
indicator-
should
be
total,
separate,
channel
and
part
of
auto
if
we
define
our
total
orthogonal
3gpp,
even
in
the
words,
even
in
some
extreme
case,
you
can
think
about
ip
layer
eca
as
some
kind
of
indicator
right
is
in
conjunction
bit
or
not
condition
bit,
and
then
you
read
all
the
information.
My
understanding
of
marketing,
a
suggestion
or
request
to
is
to
get
very
clear
use
cases
is
that
right.
B
H
H
To
know
if
the
people
who
are
running
these
real-time
indicators
actually
need
a
service
from
alto
and
if
it's
being
standard
in
some
other
sort
of
body,
it
would
be
great
to
get
a
liaison
statement
from
three
gpp
or
whoever's
saying.
Yes,
we
really
love
ulta
to
do
x,
y
and
z,
because
this
smells
to
me
like
something
a
little
speculative.
I
The
the
intent
is
not
to
help
out
3gpp
people
with
alto.
The
intent
is
that
if
you
have
a
device
in
a
cellular
network-
and
so
this
with
your
device,
you
can
get
like
direct
information
from
the
3gpp.
But
if
you
use
an
auto
client
that
is
connected
to
the
your
application,
if
you
have
an
application
client
running
into
your
ue
and
this
application,
client
uses
an
auto
client
to
get
larger
scope
information
on
the
alto
client.
I
You
cannot,
so
the
the
run
cannot
help
you
out
beyond
its
scope
and
the
alto
alto
cannot
give
you
real-time
de
information.
So
this
allows
to
make
to
compose
at
the
uee
level
at
the
at
alto
client
level.
You
compose
information
that
you
gather
on
one
side
from
the
3gpp
and
on
the
other
side
from
alto,
and
it's
not
meant
to
to
help
out
3gpp.
It's
meant
to
help
out
more
reactive,
auto
information
with
the
help
of
real
time
indicators.
K
Yes,
can
you
hear
me
yes,
okay,
perfect,
so
yeah?
I
will
go
quick
through
this,
so
essentially
here
the
goal
would
be
to
fall
to
enhance
or
or
enable,
even
in
some
cases,
the
the
automation
of
the
network
operation,
which
is
key
for
for
operators.
As
you
can
imagine
so
the
idea
the
objective
would
be
to
integrate
alto
either
with
with
other
network
entities
that
could
be
under
the
control
of
the
same
operator
or
even
outside
of
the
operator.
K
This
is
something
to
to
be
analyzed
according
to
the
different
use
cases,
and
the
objective
of
this
integration
would
be
to
allow
to
take
better
decisions,
triggering
form
actions
assisting
at
the
end,
the
decisions
that
could
be
taken.
There
are
four
cases,
let's
say
or
four
scenarios.
I
will
go
quick
through
them
taking
more
time
maybe
case
two
and
four,
because
probably
this
integration
with
other
parts
is
more
intuitive.
K
So
in
the
first
case,
essentially,
the
idea
would
be
to
provide
extensions
to
rfc
7971
leveraging
on
product
on
previous
protocol
extensions
that
had
been
performing
in
the
other
working
group
like
vendor
or
perfect
or
sorry
or
a
unified
product
property.
So,
essentially,
to
repeat
all
the
those
extensions
and
to
see
what
could
be
the
appropriate
protocol
appropriate
extensions
to
auto
protocol
for
doing
or
helping
on
this
direction
of
operation
automation
in
the
regard.
K
K
The
next
slide
is
okay,
so
the
third
case
would
be
essentially
to
consider
extensions
or
to
alto
for
acting
as
an
aggregator
of
information
coming
from
different
sources.
Different
sources
that
are
being
developed
in
I
mean
are
outcomes
of
itf,
and
this
could
be
the
traffic
engineering
database,
the
lsp
database,
and
so
so
collecting
all
those
information
and
making
alto
aggregating.
K
A
A
A
E
B
B
Why
so
and
pops
up
the
meaning
is
so
not
well
defined
in
some
sense
right
and
at
the
very
beginning,
when
I
I
say
pops
up
an
email,
I
would
think
about
like
a
message:
queue,
for
example,
kafka
and
rabbit
and
kill
your
name
and
all
kinds
of
very
different
mechanisms,
and
you
have
a
mechanism.
Essentially
we
introduce
this
thing
and
then
auto
servers
and
other
clients.
Essentially
can
pops
up
into
the
system
essentially
become
information,
a
broker,
and
we
design
this
way
that
one
potential
we
can
design
another
one.
B
But
if
you
look
carefully
like,
for
example,
trench
and
tencent
design
and
so
on,
they
are
not
really
pops
up.
In
that
sense,
this
pops
up
more
like,
for
example,
I
probably
would
see
a
little
bit
more
like
if
I
may,
and
and
given
I
probably
use
your
model,
but
if
I'm
using,
for
example,
grpc,
for
example,
a
google
model,
that's
essentially
like
a
streaming
request
and
streaming
of
reply,
and
then
of
course
you
have
connections.
So
therefore
this
continues
process.
H
Yeah
thanks
for
the
presentations
and
the
work
everyone
put
into
this,
I
think
you've
answered
a
lot
of
the
concerns
I've
raised
and
I
I
probably
need
to
since
spend
some
time
for
the
next
few
months
really
engaging
this
on
a
more
detailed
level
by
some
high
level
comments.
First
of
all,
like
my
sense,
is
that
that
I
mean
alto
is
a
request
response
based
protocol
and
this
pumps
up
thing
is
like
almost
the
whole,
or
I
mean
whether
or
not
it's
actually
pub
sub
per
se.
H
H
Second
issue
is,
I
think,
with
these
use
cases,
I
would,
I
would
kind
of
characterize
them
in
two
different
ways.
One
is
just
sort
of
protocol
maintenance
stuff,
you
know
making
it
easier
to
deploy
additional
information
fields
or
you
know,
upgrading
with
h2
or
h3,
and
that
that
seems
pretty
straightforward
when
we
continue.
H
But
there's
also,
I
mean
I'm
sensing
a
lot
of
desire
to
have
much
more
time.
Sensitive
data
get
channeled
through
this
and
I'm
skeptical-
and
you
know
this
like
network
feedback
thing-
is
something
that
keeps
coming
with
an
itf
and
things
like
plus
and
like
it
keeps
with
a
lot
of
people
being
thrown
at
these
problems,
who
are
very
experienced
in
this
area,
and
it
keeps
not
working
out
and
I'm
very
hesitant,
to
have
this
group,
which
doesn't
have
a
ton
of
people
in
it
frankly
to
try
to
tackle
this
problem.
H
H
My
final
point,
which
is
unrelated,
is
that
I
continue
to
be
a
little
hazy
on
the
deployment
status
of
this
stuff.
You
know
we
found
out
in
jabber
today
that
sprint
has
actually
not
deployed
this,
so
I
I
believe
that
my
understanding
of
this
right
now
is
the
deutsche
telekom
and
then
telefonica
in
the
future,
and
I
am
a
little
concerned
about
getting
ahead
of.
H
L
L
I
think
only
one
subscription
is
deleted
right,
but
if
we
use
a
request
response,
maybe
every
time
you
have
to
use
a
request
and-
and
the
addition
I
think
the
the
pops
up
is-
is
helpful
right.
You
reduce
the
signal
in
your
head
because
it
pops
up
is
even
bigger.
Right
only
say
some
conditions
is
satisfied.
L
D
Well,
we
have
a
limited
okay,
I'm
sorry,
I
can't
see
if
there's
a
cue,
but
we
have
a
limited
pub
sub
in
the
form
of
server
side
events,
so
you
know
I
I
don't
mean
to
focus
on
pubs
up
only,
but
I
think
we
need
to
have
a
value
proposition
that
what
are
the
gaps
that
server
side?
Events
sse
doesn't
fulfill
that
we
need
a
pub
sub.
A
A
K
Okay,
thank
you.
No
yeah.
Can
you
hear
me?
Yes,
okay,
okay,
let's
recommend
that
they
are
not
wrong.
There
are
supportive
documents
for
all
the
use
cases
that
the
point
during
the
last.
I
think
that
in
the
in
the
past
itf,
and
also
this,
we
concentrated
in
finishing
the
dependent
documents
from
the
previous
charter,
but
I
would
say
that
we
have
a
an
individual
draft
for
each
of
the
use
cases.
Maybe
in
some
use
case
there
is
just
one
single
document
for
more
than
one
use
case,
but
just
to
recommend
this
point.
A
Yeah
thanks
luis
for
clarification,
yeah
you're
right.
We
already
have
a
lot
of
existing
individual
drafts
so
for
this
discussion
we
just
summarize
the
key
point
and
product
extension
and
related
use
cases,
and
so
definitely
we
can.
You
know
revive
all
these
individual
jobs
to
have
documented
clear
scope
and.
B
Right
so,
okay
and
I
think
we're
running
out
of
time.
I
I
do
want
to
echo
number
one.
We
do
have
quite
a
lot
of
drops
and
then
somehow
people
got.
This
group
of
people
got
impression
that
we
shouldn't
work
on
the
draft.
We
should
first
focus
on
the
tax.
So
therefore
all
these
documents
just
expire,
so
I
think
definitely
the
clear
lesson
we
learned
is.
We
should
start
maintaining,
insert
objects
and
number
two
I
do
buy.
Actually
I
think
one
issue.
One
comment
I
want
to
make
to
address.
B
One
comment
made
by
marking
is
about
this
deployment,
and
I
do
agree
that
there
are
some
diploma
there's
some
deployments
and
one
issue
is
the
deployment
is
not
as
fast
as
we
did
and
even
for
this
one.
Actually,
we
had
a
very,
very
good
conversation
with
engema.
I
don't
know
if
incoming
online
are
not
here,
he
probably
is
not,
and
so
because
bannock
is
running
over
the
line,
one
of
the
largest
deployment
and
so
therefore
the
question
over
there.
B
Of
course,
we
can
all
like
issues,
people
using
using
p2p
as
a
way
and
then
global
one,
so
there's
actually
starting.
We
start
having
this
kind
of
major
effort
to
really
organize
all
these
cdn
operators
and
to
organize
network
operators
to
all
start
with
exchanging
information
and
talk
about
how
to
get
deployment
easier
and
so
on
and
get
more
wider
diploma.
So
that's
the
issue
which
we
start
to
really
address
and
so
therefore,
to
to
answer
to
answer
a
little
bit
of
the
comment
made
by
martin.
B
Of
course,
there's
still
a
long
way
to
go,
I
think
we're
trying
pushing
for
it.
I
know,
for
example,
one
common
made
by
ingama.
Was
it's
not
like,
like
ipv6,
you
need
a
white
deployment,
so
therefore,
essentially
there's
a
little
bit
chicken
at
the
prom
and
for
some
applications,
and
you
need
this
all
these
auto
servers.
B
For
example,
I
think
comment
made
by
animal
was,
and
you
need
more
network
to
really
provide
all
the
information
and
then,
of
course,
if
you
don't
have
a
network
provide
information,
the
application
may
not
be
interested
or
may
not
deploy
and
so
on.
It's
slightly
chicken
active.
So
I
think
we
are
trying
to
do
this
effort
and
we're
actually
even
to
try
to
seek
some
help
from
it.
For
example,
work
or
challenge
is
a
sitcom,
an
eye
workshop
and
so
on.
E
E
H
H
H
I
think
if,
if
what
richard
alluded
to,
where
they're,
just
like
gaps
in
the
current
alto
thing
that
are
preventing
wider
deployment
of
this
thing,
I
think
those
those
should
be
a
pretty
high
priority.
I'd
be
very
comfortable,
proving
those
you
know,
bob
I
and
mira
before
me
have,
I
think,
expressed
concern
with
some
of
the
real-time
trends
that
are
developing.
I
mean
I.
I
certainly
understand
the
desire
for
lots
of
network
feedback
to
the
end
points.
This
is
a
again
a
long-term
stating
theme
in
ietf.
H
I
just
think
it's
also
unlikely
to
be
successful.
Quite
frankly,
so
you
know
those
those
are
the
different
extremes
here
and
I
don't
know
if
we
ever
have
to
have
an
interim
or
we
just
need
to
have
an
informal
conversation
or
side
meeting
or
something
to
to
work
through
this,
and
I
I
should
also
confirm
zahid
about
this
and
see
what
he
thinks
offline.
So
we'll
do
that
and
then
and
yeah.
I
think
once
once
the
two
of
us
are
comfortable
with
it,
then
that
didn't
go
to
isg.
E
Okay,
all
right
I
mean
everybody.
We
always
have
donating
this
for
further
questions
or
discussions.
Right
I
mean
not.
Everything
has
to
happen
here
in
this
meeting.
I
just
wanted
to
reiterate
that,
but
I
think
we
had
a
good
presentation
on
the
current
proposals
and
on
the
current
charter
and
how
they
all
relate,
and
we
have
comments
from
our
director
who
is
basically
our
manager
for
rechartering
right
martin.
I
don't
want
to
dump
this
on
you,
but
you
are
the
the
the.