►
From YouTube: CORE WG Interim Meeting, 2020-09-24
Description
CORE WG Interim Meeting, 2020-09-24
A
So
we
we
had
two
items
in
the
agenda
for
today
the
first
one
was
revisiting
latest
actions
on
the
this
is
directory
document.
I
understood
we
don't
have
actual
updates
since
last
time
to
to
check
today
correct
christian.
B
A
No
worries-
and
we
can
take
it
at
the
next
interim
just
as
well,
but
then
we
have
all
more
time
for
the
timelink
document,
and
alan
also
here
today
mentioned
his
pull
request
yesterday.
So
adding
more
material
for
the
discussion
and
I
think
bill
is
actually
going
to
drive
the
presentation.
C
Yeah
thanks,
so
I
I
have
to
confess
that
we
didn't
actually
progress
very
much
as
well,
but
but
since
there
was
an
interim
available
and
and
jaime
asked
me
to
to
come
and
present
something-
and
I
said
just
okay-
let's,
let's
just
do
that
so
thanks
to
thanks
to
ellen
also
for
giving
this
for
his
pull
requests.
Lately
I'll
share
my
screen
and
then
let's
see
if
this
works
again.
Sorry,
I
usually
thought.
C
All
right,
so
not
very
many
updates,
since
ietf
108,
but
I'll
just
give
you
a
summary
of
the
next
steps.
So
we
are,
we
are
right
now
in
dropped.
11
draft
11
has
was
was
the
last
last
one
that
was
submitted
just
before
just
before
the
108
meeting,
and
what
we
had
was
a
road
map
in
which
we
gradually
moved
towards
some
of
the
editorial
changes
that
changed
the
language
in
which
we
start
talking
about
state
transfers.
C
When
you
actually
have
the
the
conditional
observed
attributes
applied
on
resources
and
then,
while
while
we're
doing
that,
we'll
also
take
the
time
to
discuss
and
and
and
put
in
the
future
drops
that
will
address
some
of
the
issues
that
we
discovered
with
pmax,
particularly
when
you
have
the
the
co-op
cell
and
the
co-op
client
separated
by
a
proxy
that
might
or
might
not
behave
the
way
you
want
it
to,
and
then
proposals
from
ellen
and
and
live
with
m2m
people
for
ep
main
and
epmax.
C
So
I'll
just
have
a
couple
of
slides,
more
ellen
and
then
and
then
please
go
ahead
and
discuss
that.
So
the
the
idea
was
to
convene
small
research
groups
or
small
discussion
groups
before
ietf109.
C
So
please,
let
us
know
if
you
would
like
to
receive
an
invitation
to
join,
so
I
actually
extended
this
invitation
at
the
core
meeting
in
108,
but
then
I
didn't
really
get
any
anybody
else
who
wanted
to
join
in.
So
I
guess
we
are
still
in
the
same
same
group,
which
is
actually
kind
of
nice,
because
we
we
have
rather
rather
effective
discussions
and
also
very
informed
opinions.
C
But
but
if
you
wanna
join
in
any
of
you
are
here
right
now,
please
please
let
let
me
know
or
just
just
mail
the
office
of
dime
link
and
that
will
be
you'll
be
included
into
the
mails
okay.
So
the
ongoing
discussions
in
trough
11,
basically
just
discusses,
like
I
said
notifications
arising
from
setting
these
observed,
attributes
and
just
reporting
them
and
thinking
of
message
transfers.
C
So
there
are
substantial
editorial
changes
in
to
be
performed
in
order
to
alter
this
language
and,
and
the
idea
is
to
align
it
towards
using
language
that
that
okay,
so
that
was.
C
More
more
towards
how
how
it's
done
in,
for
example,
the
observe
rfc,
then
we
talk
about
restful,
state,
trans,
straight
state
transitions
and
and
transfers,
and
then
the
this
is
probably
the
last
letter
I
had
so
this
is
again
literally.
C
This
came
from
this
study
is
actually
taken
also
from
idea
108,
but
there's
a
proposal
basically
from
oma
to
to
support
two
new
attributes.
Epmax,
we
had
a
small
discussion
in
108.
It
wasn't
a
very
long
discussion
because
we
we
didn't
have
that
much
time,
but
allen
has
actually
augmented.
That
proposal
today
with
an
example
use
case,
which
I
hadn't
had
time
to
go
through.
I'm
sorry,
but
we
will
do
it
now,
so
epm
and
dpmax.
C
Basically,
they
differ
from
from
pmin
and
pmax
in
in
a
very
significant
manner,
and
then
the
entire
discussion
is
is
basically
found
in
this
github
as
an
issue
number
18..
So
that's
that's,
basically
tracking
the
opinions
of
of
everybody-
and
I
don't
know
if
I
can
actually
do
this.
But
let
me
try
to
see
if
I
can
oops
what
did
I
do
now?
Oops.
D
C
Just
a
minute
here
you
go
so
this
is.
This
was
what
I
wanted
to
show
you.
Can
you
still
see
my
screen
and
yes,
this
is
the
browser
yeah
okay,
so
this
is
the
this
is
the
current
issue
here
and
we
have
had
some
discussions
as
you
can
see,
it's
already
started.
C
Alan
allen
proposes
on
the
13th
of
april,
but
I
think
the
discussion
was
even
earlier
than
that,
because
the
original
information
came
on
the
core
meeting
list
and
then
we
proposed
that
ellen
basically
raised
an
issue
in
the
github.
So
so
this
was
even
earlier
than
that
and
and
then
there
have
been
already
some
opinions
on
that
from
others.
C
C
The
latest
the
latest
status
update
here
is
is
this:
this
is
the
text
to
consider
and
just
given
this
like
18
hours
ago
ellen,
I
think
I
think
it's
time
that
that,
basically
I
I
could
stop
sharing
my
screen
or
what
do
you
think.
C
E
E
C
Okay,
it's
yours.
A
E
Yes,
excellent,
okay,
so,
as
bill
said,
I
put
the
issue
in
and
after
we
got
through
this
and
we
we
did
have
some
very
good
discussions
because
it
needs
to
be
clear
to
the
reader
of
the
document
what
new
these
new
attributes
are
and
how
to
use
them.
So
the
first
thing
I
did
is
I
actually
created
a
pull
request
to
put
the
the
actual
changes
to
the
the
draft.
E
There
were
example
formats
and
because
of
other
confusions
on
the
use
of
pmax.
So
I
wasn't
sure
how
we
wanted
to
create
the
content.
So
I
just
put
it
here
in
the
in
the
issue
rather
than
adding
it
in
the
a
pull
request.
E
So
let
me
walk
you
through
some
of
the
logic
here
and
I
want
to
first
show
you
I
in
it
by
the
way
anybody
interrupt
me
at
any
time
or
if
you
think,
I'm
going
wrong
direction
and
you
want
to
pull
me
a
certain
direction.
Please
please
go
ahead
and
just
you
know
just
jump
right
in
so
in
the
lightweight
m
spec,
and
this
is
the
latest.
Well,
almost
the
release
version
of
the
spec.
That's
coming
out
quite
soon.
We
have
you
know
your
conditional
notification.
E
We
also
created
some
that
we
felt
were
important
for
lightweight
m,
and
so
we
created
these
ep
min
ep
max
edge
con
and
hq
max.
So
I
just
wanted
to
go
over
these
five
new
attributes
very
quickly
and
then
talk
about
how
I've
proposed
to
modify
dim
link.
So
far-
and
I
don't
know
some
I've
heard
some
people
call
it
didn't,
link
and
other
people
pronounce
it
dimelink
I
like
dynlink,
but
I'll
pronounce
it
any
any
way.
You
guys
want.
E
The
edge
is
a
new
attribute
that
actually
controls
the
the
evaluation
of
the
attribute
to
determine
whether
it
changed
did
it
change
on
the
leading
edge
or
the
falling
edge,
the
the
rising
edge
or
the
falling
edge,
the
leading
edge
the.
E
But
if
there's
an
important
notification
you'd
want
to
designate
on
this
observation,
this
notification
must
be
sent
confirmable
because
it's
critical,
so
you
know
a
an
alarm
of
some
sort
versus
just
an
informative
notification
and
then
we
also
have
a
historical
queue
depth.
So
if
you
are
maintaining
historical
notifications,
because
you
are
currently
unable
to
reach
the
the
client
with
the
notifications,
then
the
server
will
is
at
least
being
directed
to
say,
save
five
of
these
things.
So
when
we
come
back
up
at
least
I
have
a
history
of
five.
E
Those
are
things
we
felt
were
important
in
the
the
light
weight
and
specification.
E
So
when
I
relate
that
back
to
the
den
length
nine
link
specification,
the
only
ones
I
started
to
put
in
were
about
ep
men
and
ep
max,
but
I
think
these
other
concepts
are
also
important
that
we
need
to
consider
for
dynlink,
because
I
think
other
users
of
the
the
co-app
specification
may
want
to
take
advantage
of
these
concepts
all
right.
So
that
was
my
first
like
level
set
of
the
evolution
of
how
this
happened.
So
are
there
any
questions
so
far
and
by
the
way
I
do
have
the
chat
window
up.
F
E
So
I
I
don't
know
if
you
can
see
everything,
but
here
here
are
the
notes
on
this.
E
And
and
that's
interesting
because
we
just
added
this
capability
a
few
months
ago,
not
a
you
know
this.
This
hq
max
has
not
yet
been,
I
would
say
it
hasn't
been
implemented
in
many
distributions,
nor
has
it
been
widely
utilized
in
the
field,
so
I
I
do
suspect,
we'll
we'll
you
know.
Hopefully
we
understood
what
the
market
needed
and
that
this
will,
I
don't
know
evolve,
is
the
right
word,
but
there
you
may
see
some
changes
based
upon
deployments
all
right.
E
E
E
The
this
explains
the
logic
of
these
two
new
parameters
and
then
I'll
show
you
how
I
try
to
accommodate
them
in
the
dynlink
draft,
because
it
they're
not
notification
parameters,
they're,
really,
control
parameters
about
generating
the
notification
or
about
doing
the
measurements
in
order
to
generate
the
notification,
so
I
had
to
separate
out
conditional
notification
attributes
from
conditional
control
attributes,
which
is
what
I'm
calling
these
things
that
are
really
controlling,
how
the
notifications
are
generated
or
the
measurement
when
the
mo
when
the
measurements
occur.
E
Okay,
so,
as
my
example
I'm
going
to
say,
these
are
conditional
control
attributes
and
we're
going
to
do
an
observation,
we're
going
to
use
pmin,
pmax
and
greater
than
and
then
these
new
two
new
ones,
the
minimum
evaluation
and
the
maximum
evaluation,
okay
periods.
So
in
this
use
case
I
have
a
temperature
sensor
and
I'm
going
to
send
a
notification
whenever
it
goes
above
or
below
100
degrees.
So
I
said
gt
greater
than
100..
E
E
I
don't
think
there's
any
ambiguity
there
so
now
in
reality,
what
this
could
mean
is
I
have
my
device
and
I
generate
a
report
on
the
temperature
sensor
because
it
went
above
100.
E
So
if
I
wanted
to
based
upon
these
parameters,
I
could
just
go
to
sleep:
wake
up
once
every
60
seconds
and
set
my
temperature,
but
I
don't
think
that's
the
behavior
that
is
is
is
wanted
by
the
client
based
upon
these
parameters,
but
there's
nothing
that
guides
the
device
in
in
what
it
should
do.
Besides,
just
you
know,
waking
up
at
pmax
and
sending
the
the
report
so
in
order
to
control
how
frequently
the
the
server
measures
the
temperature
sensor,
the
client
can
configure
two
more
parameters
and
they're
called
ep
men
and
ep
max.
E
So
this
is
the
minimum
evaluation
period
and
the
maximum
evaluation
period.
So
in
order
to
control
how
frequently
the
server
measures
the
temperature
sensor
the
client's
going
to
configure
it
such
that
it
doesn't
do
it
more
frequently
than
five
seconds.
I
don't
want
to
waste
battery
life
every
you
know
waking
up
every
second,
I
want
to
wait
a
five
degree,
a
minimum
period
of
five
seconds,
and
so
that's
ep
min,
but
I
must
measure
it
at
least
every
10
seconds
and
that's
going
to
be
ep
max.
G
Alan,
why
did
it
send?
Why
did
it
set
p
max
to
60
if
it
really
wants
to
hear
from
the
server
every
from
the
sensor,
every
10
seconds.
G
Okay,
so
so
60
seconds
is
actually
the
refresh
time
and
not
the
the
the
edge
condition.
Time
that
you
must
do
sounds
like
pmax
is
actually
misnamed
is
what
I
it
sounds
to
me.
E
So
you
know
I
have
a
temperature
sensor
and
I
reported
because
it
went
to
101
degrees.
It
drops
down
to
100.
It
goes
up
to
102
degrees,
I'm
not
reporting
anything
because
it
hasn't.
You
know
my
greater
than
hasn't
been
violated,
it's
still
above
100,
even
though
it's
changed
so
after
my
pmax
is
60
seconds,
I'm
going
to
report
102
degrees.
G
G
Similarly,
the
name
is
similar,
but
actually
the
the
original
attribute
was
perhaps
poorly
named
and
for
the
purpose
of
what
you
were
doing
and
that's
why
the
the
the
that's
why
they
are
the
the
need.
The
need
for
this
other
thing
is
not
so
obvious,
because
it's
not
a
maximum.
It's
a
pmax
is
not
a
maximum.
It's
a
you
must
report
every
this
time
anyway
and
and
yeah,
so
I
can
see
why
it
was
named
that
way,
it
just
seems
like
maybe
some
renaming
here
would
be
helpful,
but
anyway.
E
Yeah
and
I'm
I'm
all
open
for
any
changes,
because
the
concepts
I
think
help
deployments
because
right
now
I
actually
do
know
of
a
device.
That
said,
I
just
wake
up
on
every
pmax
and
and
send
the
value,
because
that's
my
minimum
energy
usage.
G
E
G
E
That
kills
me
because,
honestly,
I
don't
want
to
wait
60
seconds
to
get
my
fire
alarm
just
because
you
allow
me
to
do
that,
doesn't
make
it
right.
So
that's
why
I'm
introducing
these
new
parameters
that,
if
you
don't
want
to
allow
me
to
to
go
to
sleep
every
pmax,
this
is
your
ability
to
tell
me
not
to
do
that.
E
So,
let's
go
through
the
timeline
because
I
actually
like
the
way
I
wrote
the
timeline
I
and
I'm
because
I'm
I'm
hoping
this
makes
sense
to
people.
So
I'm
going
to
start,
I
just
generated
a
notification
right.
My
temperature
went
above
100
degrees.
I
just
generated
a
notification,
so
I'm
considering
that
t
0.
E
now,
of
course,
because
of
p
min.
I
have
to
wait
five
seconds
before
I
do
anything
else,
because
I
can't
generate
a
report,
so
I'm
gonna
go
to
sleep
for
five
seconds
and
I'm
gonna
wake
up
at
the
end
of
pmen.
So
after
you
know,
paeman
expires,
which
is
t
equals
five.
Now
I'm
going
to
do
a
measurement
and
pretty
much,
I
think
that's
the
way.
E
Everybody
would
do
it
because
at
the
end
of
pmen
you
know
if
you
went
below
100
or
let's
say
you
were
below
100
went
above
100
you'd
want
to
generate
a
notification
as
soon
as
you
can
so
at
pminxprequel5,
I'm
going
to
do
a
measurement.
If
the.
If
the
temperature,
you
know,
met
the
gt
criteria.
So
if
it
was
a
100
above
100
went
below
or
if
it
went
below
and
went
above
of
course,
I'd
generate
a
notification,
but
let's
say
that
we
haven't
made
any
change
from
our
previous
state.
E
Of
course,
I'm
not
going
to
generate
a
notification,
so
I
payment
expiry.
I
wake
up.
Do
the
measurement
nothing
happen?
Go
back
to
sleep
so
from
my
new
parameter,
which
is
ep
men.
I
know
I
can't
do
another
evaluation
for
five
more
seconds,
so
I
must
wait
another
five
seconds
before
doing
another
measurement
and
now
I'm
at
t
equal
ten.
So
at
t
equal
ten,
I
say
you
know
what
I'm
still
in
power
saving
mode.
E
I
don't
want
to
do
a
measurement,
so
I'm
going
to
go
to
sleep
until
ep
max
expires
to
where
I
have
to
do
a
measurement
so
ep
min
says
you
can
do
a
measurement
at
the
end
of
this
ep
max
says
I
must
do
a
measurement
at
the
end
of
this.
So
now
it's
t
equal,
15
right
so
15
seconds
went
by
I'm
now
at
t
equal
15
because
of
ep
max.
I
must
do
a
measurement
now,
okay.
Well,
the
temperature
has
changed.
E
That's
my
use
case,
so
I'm
gonna
generate
a
notification
if
the
if
the
temperature
hasn't
changed
if
the
gt
hasn't
met
the
criteria,
I'm
going
to
go
back
to
sleep
at
least
for
another
five
seconds
right
or
I
could
just
go
back
to
sleep
for
another
10
seconds
because
of
ep
max,
but
it
the
device
it's
up
to
the
device
to
decide
its
cadence
in
terms
of
the
wake
up
based
upon
ep
min
and
ep
max.
E
Let's
say
60
seconds
goes
by
and
I
was
waking
up
every
10
seconds,
but
the
temperature
never
changed
doesn't
matter
at
t
equals
60.
I
have
to
generate
a
notification.
So
that's
my
example.
In
the
way
these
after
these
new
attributes
are
meant
to
be
used,
and
I
think
it's
important
that
the
logic
is
understood
before
we
shift
over
to
the
to
the
pull
request.
Does
that
make
sense.
C
C
E
Not
necessarily
because
there's
nothing
that
says
that
it
doesn't
say
that
the
so
when
it
when
the
way
it's
written
right
now
forget
my
new
proposal.
The
way
it's
written
right
now
is
you
know
I
can.
I
must
I
must
not
report
for
five
seconds
and
I
must
report
at
p
max,
but
it
doesn't
tell
me
about
anything
in
between.
So
what
should
the
device's
sleep
in
a
rule,
be,
it
doesn't
say,
there's
no
guidance.
E
So
if
we
wanted
to,
we
could
say
that
the
if
ep
men
in
epmax
are
not
defined,
that
p
min
may
be
used
to
guide
the
device.
On
the
you
know,
the
the
reporting
evaluation
period.
We
could
say
that
I
have
no
problem
with
that,
but
I
don't
think
we
can
enforce
that
because
then
we'd
break
all
the
implementations
that
currently
use
pmen.
C
E
Let's
go
to
the
the
pull
requests
because
then
what
I
can
do
is
I
can
edit
this
online
on
the
pull
request
if
there
are
billets.
If
you
want
me
to
like
capture
some
changes
in
the
pull
request,
we
can
do
that.
E
So
let
me
go
ahead
and
go
to
the
pull
request
and
show
you.
You
know
what
I'm
proposing
as
the
changes
trying
to
get
this
checked
all
right.
I
can't
keep
the
chat
window
over,
or
else
I
wasn't
able
to
see.
So
let
me
go
to
the
the
rich
text,
the
the
the
first
concept.
E
I
said,
as
I
pointed
out,
I
said:
they're,
not
notification,
attributes
now
they're
control
attributes-
and
I
did
want
to
distinguish
between
the
two,
because
when
we
were
having
our
our
email
discussions
and
our
verbal
discussion
at
108,
it
was
but
these
these
ep
men
and
ep
max
have
nothing
to
do
about
generating
the
notification
and
that's
correct.
That
statement
was
100,
correct,
they're
about
control
of
the
device
when
evaluating
whether
the
notification
should
occur.
E
So
I
call
them
control
attributes
and
I
try
to
be
consistent,
making
sure
that
both
conditional
notification
and
conditional
control
attributes
are
covered.
So
then,
I
conditional
notification
attributes
define
the
conditions
that
trigger
a
notification.
That's
existing,
but
conditional
control
attributes
divide
the
cadence
of
measurement
of
the
conditions
that
trigger
a
notification.
E
And
I'll
just
keep
going
unless
I
hear
somebody
interrupt
me,
so
the
the
set
of
notification
attributes
defined
here.
This
is
the
pre-existing
text.
It
allows
the
trans
client
to
control.
How
often
a
client
is
interested
in
receiving
notifications
and
how
much
a
resource
file
you
could
change
when
to
make
it
interesting.
E
I
added
the
text.
The
set
of
control
attributes
defined
here
allow
a
client
to
control
how
often
the
server
performs
a
measurement
of
the
conditions,
and
so
that's
explicitly
the
ep
max
in
epmax,
which
controls
that
cadence,
I
added
them
into
the
table,
and
so
this
is.
This
is
just
a
list
of
the
attributes
and
I
change
the
name
to
control
notification
and
control,
conditional
notification
and
control
attributes
for
consistency
so
and
I'm
trying
to
distinguish
between
existing
text
and
new
text.
E
I
know
you
guys
can
look
at
this,
then
you
know
that
for
yourself,
but
just
to
review
control
notifications
should
be
evaluated,
but
control
and
conditional
control
attributes
are
used
to
configure
the
internals,
and
I
use
the
same
language
from
here.
I
like
the
way
I
did
this
sorry,
adding
myself
on
the
back.
So
conditional
notification
attribute
should
be
evaluated
on
all
potential
notifications
from
a
resource
and
whether
that's
the
internal
server
driven
sampling
process
or
if
it's
an
external
update
request
from
the
server.
E
But
here's
what's
interesting.
The
concern
conditional
control
attributes
are
used
to
configure
that
internal
server
driven
sampling
process.
So
that's
why
I
call
them
control
attributes,
because
that's
all
they're
doing
and
that's
of
course,
for
performing
the
measurement
bill.
Did
you
want
me
to
go
over
the
pieman
stuff,
while
I'm
here.
C
Not
necessarily,
I
was
just
wondering
whether
p
max
fits
into
the
conditional
control
attributes
or
not,
and
in
that
case,
if
max
is
one
of
those,
then
we
might
have
to
change
the
wording
for
output.
You
I'm
not
sure
if
they
scrolled
up
or
down,
but
it
was
near
the
table,
I
think
and
yeah.
So
they
said
just
just
sorry
if
you
go
down
yeah
so
so
there
is
the
the
last
line
there
that
says,
conditional
control
attributes
are
used
to
configure
the
internal
server
during
sampling.
C
E
E
Let
me
keep
going
through
the
new
text
and
then
we
can.
We
can
maybe
try
and
come
up
with
a
way.
I
could
modify
the
the
draft
to
reflect
all
that.
I'm
not
sure,
there's
really
much
change
there,
because
you
know
it.
I
didn't
have
to
well
we'll
go
through
the
full
thing
in
a
few
minutes.
Let
me
do
this
all
right
so
now
the
minimum
evaluation
period-
and
I
tried
to
craft
this
very
much
like
the
format
that
we
did
pimen.
E
So
you
know
when
present
indicates
minimum
time
in
seconds.
The
server
must
wait
between
two
consecutive
measurements
of
the
condition
of
a
resource
when
it
expires,
the
server
may
immediately
perform
a
new
measurement.
It
does
it's.
It's
saying
you
must
wait
this
time,
but
you
don't
have
to
do
it
at
the
end
of
this
column,
and
then
I
tried
to
capture
this
in
the
absence
of
parameter
and
is,
and
that's
not
used
by
the
server.
E
That's
where
we
would.
I
guess
inject
a
you
know:
if,
if
ep
min
is
not
defined,
keymin
can
be
used,
pimen
may
be
used
as
the
default
wait
period,
or
we
want
to
reflect
that
and
then
just
saying
it's
got
to
be
greater
than
zero
or
it's
a
bad
request.
E
Then.
Similarly,
the
in
pmax,
the
ep
max.
Excuse
me,
I
it's.
The
the
server
time
in
seconds
may
wait
between
two
consecutive
measurements
right,
because
you
know-
and
I
struggled
when
I
originally
wrote
this-
I
kind
of
struggled
with
the
wording,
because
you
must
wait
for
ep
men,
you
could
do
a
measurement,
but
you
must
do
a
measurement
after
ep
max,
so
you,
the
server,
may
wait
between
two
consecutive
measurements.
You
can
wait
up
to
this
value
when
the
maximum
period
expires.
E
You
must
perform
a
measurement
and
then,
in
the
absence
of
this
parameter,
it's
not
defined
used
by
the
server
and
then
it
must
be
greater
than
zero.
It's
bad
request.
Then
this
was
the
conditional
between
the
two.
If
they're,
both
defined
epmax
must
be
greater
than
ep
min.
Otherwise
you
send
a
bad
request.
E
You
return
a
bad
request
and
then
there
was
just
a
spelling
error
found
there.
That
was
it.
Those
are
my
proposed
changes.
I
didn't
include
the
example
text.
I
think
example.
Text
may
be
valuable,
but
I
didn't
know
if
we
wanted
to
include
it
just
for
the
use
of
ep
min
ap
max
or
we
wanted
to
include
other
example
text.
E
So
I
didn't
include
any
example
text,
but
I
did
want
to
write
that
to
show
input
and
put
it
in
the
issue,
so
we
can
cut
and
paste
it
over
here
if
we
want
all
right.
So
that's
my
pull
request.
Where
do
we
want
to
go
from
here.
C
So,
from
my
perspective,
I
think
most
most
of
this
text
is
is
okay.
I
still
have
to
go
through
it
a
little
bit
more
carefully,
but
but
this
this
looks
sensible
to
me.
I
think
that
where
was
it
that
I
saw,
I
saw
something
that
that
yeah
there
was.
There
was
something
that
you
mentioned
regarding
the
if
ep
main
is
not
set,
then
then
the
server.
What
was
the
language
there?
C
I
need
to
check
once
more,
but
but
I
think
that
there's
that
there
should
be
something
written
in
the
implementation
considerations
about
about
what
to
do
in
the
absence
of
this
parameter.
Indeed.
So,
if
if
p
main
is
set
at
ep,
means
that
then
we
could
actually
reflect
some
text
in
the
implementation
considerations
that
that
that
the
the
server
can
use
pmin,
if
necessary,.
E
E
So
that
is
here
in
the
absence
of
this
parameter,
the
minimum
value
is
not
used
by
the
server.
The
server
may
use
p-min,
if
defined
as
a
guidance
of
the.
G
C
C
So
in
the
second,
in
the
second
paragraph,
when
we
talked
about
the
oh
hang
on
yeah,
so
so
there
was,
there
was
a
discussion
there
about
the
internal
sample
period
for
determining
the
resource
value
so
that
that
needs
to
be
considered
as
well.
So.
C
C
Yeah,
that's
that's.
Probably
the
the
internal
sample
period
is
probably
not
very
accurate.
Considering
now,
if
we're
introducing
ep
in
an
ep
max
there.
So
so
I
would
say
that
we
should
probably
rephrase
that
instead
of
saying
internal
sample
period,
that
we
could
actually
mention
internal
sample
period
and
and
the
ep
mean
ep
max
values,
so.
E
E
C
E
C
A
C
We
just
have
to
ensure
that
that
band
yeah
yeah
band
is
actually
a
modifier
for
for
the
behavior
of
greater
than
and
less
than
so
that's
as
long
as
that
doesn't
clash
too
much
with
ebb
and
eb
max
values.
Here
yeah.
I.
E
E
Somebody
put
something
in
the
chat:
oh,
oh,
okay,
sorry
christian!
Let
me
get
back
to
you.
Let
me
just
do
this
real,
quick,
I
I
did.
I
was
considering
putting
in
pull
requests
for
edge,
confirmable
and
maximum
historical
queue,
because
I
really
like
the
alignment
between
dynlink
and
lightweight
ms,
but
I
didn't
want
to
do
that
all
at
one
shot,
so
I
figure
if
we
can
get
an
agreement
on
ep9p
max
and
we're
comfortable
with
that,
then
I
would
put
in
a
separate
pull
request
for
each
of
these
other
ones.
E
E
B
Nope
so
sorry,
hello,.
B
You,
okay,
sorry,
so,
first
of
all,
I'm
I'm
rather
happy
with
those
attributes
that
are
more
related
to
the
sampling
and
not
to
the
resource
to
the
propagation
of
the
resource
state
being
separate.
B
One
particular
change
that
caught
my
eye
is
the
line
about
how
those
are
transferred
in
the
query
parameters.
So,
as
I
understand
in
lightweight
m2m,
they
are
always
transported
via
a
completely
different
channel
anyway,
and
in
the
div
line
91.
B
It
says
that
both
notification
and
control
attributes
can
be
inquiry
parameters,
and
this
would
be
the
very
point
where
we
could,
where
I
think
it
would
make
sense
to
say
this
applies
to
all,
and
this
applies
only
to
the
notification
parameters,
because
the
control
attributes
when
they
are
put
in
the
observe
requests
they
wind
up
in
a
as
part
of
a
safe
request.
So
they
shouldn't
really
change
anything
inside
the
server
and
the
sampling
interval
that
the
server
does,
which
would
probably
apply
to
all
the
observations
there
or
to
the
whole
sensor.
B
Those
should
not
be
changeable
by
a
safe
request,
especially
in
which
they,
which
the
observer
request
is.
So
would
it
make
sense
for
you
to
limit
this
to
the
notification
attributes
and
then
control
attributes
couldn't
be
used
wherever
there
is
an
actual,
unsafe,
oper,
an
operation
that
had
that
may
have
side
effects
involved.
E
B
B
That's
an
operation
that
can
that
affects
the
whole
device
and
it's
okay,
because
that's
visible
to
the
user
and
visible
to
the
protocol
as
long
as
they
are
only
applied
to
notification-
and
please
correct
me
when
I'm
wrong,
but
but
I
think
that
the
way
you're
using
them
in
lightweight
m2m
doesn't
even
use
those
those
query
parameters
in
in
an
observation
when
they're
used
there.
They
would
affect
all
the
other
observations
that
could
be
on
the
same
resource
in
a
get
request.
But.
E
B
But
at
least
one
of
the
attributes
does
not
make
sense
in
that
way
I
mean,
but
this
history
attribute
that's
something
that
affects
the
whole
resource,
because
it's
it's,
for
I
mean
it's
not
in
there
yet,
but
it's
in
the
other
table
that's
supposed
to
migrate
over
here.
If,
if
that
history
attribute
affects
how
a
later
observation
would
see
the
initial
state,
wouldn't
it.
E
You're
talking
control,
attributes
yeah.
So,
let's
let's
say
I
have
two
attributes.
You
know
I
have
an
observation
on
my
temperature
sensor
and
I
have
p
min
equals
five
p
max
equals
ten
forget
e,
p,
min
and
ep
knots
right
now.
I
have
another
observation
that
I
set
on
that
same
resource
and
I
said
another
observation
that
says
p
min
is
10
p
max
is
20..
B
Sorry
p
m
alpha,
p,
min
and
p
max.
Okay,
they
are
they're;
no,
they
don't.
No,
they
don't.
I
was
thinking
more
of
the
control
attributes
that
were
there
before
we
just
before
talked
about
pman
and
pmax,
going
over
to
control
attributes.
So
so.
B
Ep
mean
ep
max
and
and
his
and
history.
E
Okay,
I
I've
actually
been
doing
something
in
in
the
meantime
as
well.
Sorry
about
that,
so
I'm
trying
to
separate
out
here
the
the
conditional
notification
and
conditional
control
in
my
mind,
there's
they're,
completely
separate
in
terms
of
the
scope,
is
solely
within
the
the
specific
observation
we're
talking
about,
and
maybe
I'm
talking,
maybe
we're
talking
past
each
other.
E
I
don't
really
understand
your
question,
the
whether
their
controller
notification,
I
set
up
an
observation
and
that's
an
atomic
thing
and
the
device
what
the
device
must
do
is
the
device
must
take
all
the
notifications
and
all
the
ep
men
and
ep
max
values
of
all
notifications
to
decide
its
wake
up
cadence
because
it
asks
the
device
has
to
wake
up
to
at
least
perform
one
measurement,
but
that
doesn't
affect
the
other
cadences
of
the
other
measurements.
It
still
has
to
abide
by
the
ep
min
and
ep
max
of
its
individual
observation.
E
So
a
more
complex
example
is,
I
have
a
another
temperature
sensor
and
I
want
to
wake
up
every
one
second,
so
I
don't
set
ep
not
ep
min.
I
set
ep
max
at
every
one.
Second,
so
the
p
min
and
p
max's
are
the
same
at
five
and
sixty
on
the
time
equals
zero.
I
generate
notifications
for
both
time.
Equal
five
is
my
p
men
expertise.
E
So
I
know
I
can't
generate
a
report,
but
now
I'm
going
to
wake
up
every
second
to
evaluate
the
second
temperature
sensor,
but
that
doesn't
change
what's
happening
on
the
first
temperature
sensor,
even
though
I'm
waking
up
for
for
temperature
2
measuring
it
every
second.
I
can't
measure
temperature
one
for
five
seconds,
so
when
five
seconds
expires
on
temperature,
one,
I'm
already
waking
up
for
temperature.
Two
on
my
one
second
cadence,
so
I'm
gonna
measure
both.
B
Okay,
so
if,
if
two
observers
are
observing
the
same
temperature
would
would
the
device,
then
I
mean
what
will
if
this
is
actually
about
when
this,
when
the
sensor
is
being
evaluated,
what
if
they
conflict
there?
If
I,
if
one
request,
says
that
you
may
sets
an
two
requests,
two
observations
both
on
the
same
attribute,
one
says:
ep
min
ep
between
one
and
two
and
the
other
sends
ep
between
10
and
20..
B
E
It
can't
violate
the
parameters
so
temp
one.
It
cannot,
or
I
shouldn't
say
that
observation
one
okay,
that
is,
it
cannot
violate
what's
in
observation,
one
regardless
of
what's
in
observation,
two
just
because
the
observation
and
we're
talking
about
the
cadence
of
the
measurement,
not
the
the
reporting
criteria,
correct
so
as
long
as
you're,
not
violating
the
measurement
parameters
of
the
observation,
you're
solely
basing
it
on
the
reporting
criteria.
So
observation
one
says
I
can't
measure
I
can't
measure
more
frequently
than
every
five
seconds.
E
There's
nothing
here
that
says
you
can't
I
do
we
want
to
specify
that
I
mean
you
found
a
good
use
case,
which
is
same
resource.
I
have
a
single
temperature
sensor,
one
client
and
we're
in
a
multi-client
environment.
One
client
says
measure
it
every
one.
Second,
the
other
client
says
measure
it
every
10
seconds.
E
B
B
I
do
see
your
point
that
all
these
evaluation
product
periods
can
be
treated
independently
and
not
are
not
interpreted
as
this
is
when
you
must
take
a
measurement,
but
this
is
when
you
must
take
a
measurement
for
consideration
for
consideration
here,
so
I
may
take
some
time
to
to
process
all
the
the
implications
of
all
of
this.
If
ever,
if
everything
here
is
safe,
then
probably
yeah,
then
then
this
might.
This
may
all
work
out,
but.
E
E
Do
we
want
to
provide
guidance
to
say
that
you
know
this?
Is
these
values
are
recommendations
from
the
the
client
to
enable
the
server
to
maximize
its
its
power
savings
or
maximize
its
wake-up
cadence?
I
don't
know
how
we
want
to
cover
that,
but
I
I
I
think
it's
important
to
give
guidance
for
where
we
can.
E
All
right,
I
I
did
just
to
show
tell
you
guys
what
I
was
doing,
because
I
bill.
I
think
you
said
that
that
minimum
period
and
pman
and
pmax
are
actually
control
attributes.
So
I
thought
maybe
did
we
want
to
separate
out,
and
so
I
left
christian.
I
left
this
statement
alone
and
then
I
said:
okay,
here's
the
control
table
of
notification
attributes
and
then
I
have
that
same
statement
in
the
for
the
control
attributes
and
here's
the
table
for
the
control
attributes,
and
I
put
pminpmax
min
epmax
is:
is
that
does
that
work?
F
Missing
yeah,
so
I
I
was
trying
to
think
of
having
two
clients
and
one
sets
ep
min
and
et
max
both
to
seven
seconds
and
the
other
one
sends
it
to
13
seconds.
F
E
It's
a
it's
a
good
question,
because
if
the
device
is
doing
its
own
wake
up
cadence
because
of
you
know,
like
you
said
you
have
the
the
client
one
saying
wake
up
every
seven
seconds,
the
client
two
is
saying:
wake
up,
every
13
seconds
is
client.
One
in
in
is
that
is
the
server
invalidated
from
providing
a
notification
at
seven
seconds
to
client
two,
and
I
I
don't
think
so.
E
I
think
it's
it's
the
device,
because
the
device
has
the
the
true
state
of
things.
Call
it
the
true
state
that
I
woke
up
and
I
know,
there's
something.
I
think
the
client
should
be
able
to
evaluate
the
notification
conditions
right
or
the
notification
attributes
to
say
I'm
up
anyway.
E
So
why
shouldn't
I
be
able
to
generate
a
report?
Generate
a
notification?
That's
I
guess,
that's
the
guidance,
I'm
thinking.
Maybe
we
should
capture
that
you
know
if,
if
independently
of
these
ep
men
and
ep
max
attributes
the
device
detects
a
or
is
a
is
able
to
do
a
measurement
because
of
its
internal
processing,
that
the
device
should
evaluate
the
notification
criteria
and
decide
whether
it
should
do
a
notification.
B
That,
from
considering
carson's
example,
do
we
really
need
to
phrase
this
as
a
minimum
as
a
maximum,
because
if
we,
if
we
phrase
this
more
in
terms
of
maximum,
if
if
we
phrase
this
one-sided,
then
then
the
problem
with
the
concrete
example
will
go
away
and
really
what
does
it
mean
to
have
a
minimum
period
between
samples
I
mean
the
this
is
assuming
that
the
device
is
actively
pulling,
but
it
could
just
as
well
receive
the
receive
the
values
from
from
some
external
interrupt
or
form.
Another
observation
for
all
that
matters.
E
E
So
I
had
no
guidance
on
a
timer
to
wake
out
of
this
mode.
All
I
knew
is
I
had
to
wake
up
at
pmax,
so,
no
matter
what
I
could
always
just
set
my
deep
sleep,
timer
to
be
60
seconds.
I'd,
wake
up
boom
send
out
my
report
that
doesn't
that's
not
a
good
implementation
right,
so
I
wanted
some
guidance
of
when
I
should
wake
up.
E
But
if
I
have
five
observations,
all
on
different
cadences,
with
different
ep
mins
and
ep
boxes
and
at
different
payments
and
pmaxes,
I
have
to
create
a
scheduler
and
which
is
what
I
did.
I
created
a
scheduler
taking
all
of
these
timers
into
effect
and
and
all
of
these
cadences
into
effect
and
my
own
internal
cadence
is.
I
know
I
need
to
wake
up
every
you
know,
30
you
know
or
like
100
milliseconds
to
to
be
polite
right.
E
Whatever
my
internal
cancers,
I
had
to
create
a
scheduler
and
then
based
upon
the
cadence
within
the
scheduler
for
all
of
these
accommodations.
That's
how
I
set
my
timer
and
that's
really
what
the
device
is
doing,
because
it's
not
a
single
observation
or
set
of
ep
man
or
pmax
ep
box
or
pn
or
pmin
or
pmax
for
a
single
observation
that
guides
the
device.
It's
all
of
them.
E
Whenever
ep
min
is
on,
I
use
ep
min
as
a
one
of
the
elements
of
an
optional
wake
up.
So
when
I'm
doing
my
scheduler,
I
show
all
the
the
mandatory
wake-ups
in
you
know,
so
I
create
a
list
of
the
mandatory
wake-ups
and
I
put
in
the
optional
wake-ups,
and
I
look
at
what's
my
first
mandatory.
So
that's
my
maximum
wait
time.
What
are
the
ones
before
that
and
do
I
want
to
wake
up
at
that
time?
E
Have
I
waited
long
enough
to
get
power
savings
and
if
I
have-
and
it
makes
sense
to
my
device
because
of
other
things
I
might
want
to
do-
then
I
wake
up
on
the
optional
and
set
my
timer
based
on
the
optional.
If
none
of
that
works,
I
wait.
I
set
my
timer
to
wake
up
on
the
the
first
mandatory
and
I
do
that.
Every
time
I
go
to
sleep
you
know,
because
I'm
basically
it's
a
dynamic.
E
E
E
B
E
E
E
E
F
E
Okay,
I
I
understood
all
the
words.
I
don't
understand
what
that
means
in
terms
of
trying
to
type
the
text.
Should
I
be
describing
the
example
and
then
say
that
the
you
know,
or
should
I
just
say
I
I
interdependency
of
control,
attributes
of
a
notific
of
an
observation.
E
Are
implementation
specific
or
are
you
know,
enabled
just
trying
to
figure
out
do?
Do
we
say
that
they,
the
I
don't
know,
I'm
struggling.
E
E
E
B
F
E
F
F
E
E
F
E
B
I'm
I'm
still
having
a
bit
trouble
coming
up
with
or
imagining
how
this
would
be
used
from
a
client
point
of
view.
So
if
I
were
to
implement
a
client-
and
I
don't
know
how
the
server
is
implemented-
which
I
generally
don't,
then
I
have
nothing
to
put
in
there-
I
could
put
in
a
zero
or
I
could
I
mean
so
from
from
the
power
re
from
the
the
power
considerations.
I
understand
you
would
be
using
this,
for
this
is
more
of
a
topic
of.
B
E
Okay,
so
not
it
is,
but
the
server
doesn't
know
the
preferences
of
the
client
so
that
that's
the
way
it
is
right
now
the
server
makes
this
decision
independently.
So
let's
take
the
case
of
and
believe
me,
clients
are
aware
of
the
servers.
I
know
my
server
is
actually
a
power
line
device
in
alaska
doing
a
measurement
of
a
pipeline.
E
I
know
that
because
it's
my
device
when
I'm
the
client
that
server
is
my
device,
it's
not
like
I'm
unaware
of
the
purpose
of
it
who
deployed
it.
You
know
how
much
it
costs
the
power
saving
I
need
out
of
it.
I
know
all
that,
because
it's
mine,
the
client,
is
only
the
one.
That's
saying
you
know,
I
want
to
be
able
to
configure
some
things.
E
So
in
this
case
I
have
that
device
out
on
the
the
oil
field
right
and
I
know
it's
power
line,
so
I
know
I
can't
wake
up
more
frequently
than
every
five
sec
five
minutes
to
do
a
measurement.
B
E
H
B
E
E
Needed
right,
but
if
it's
not,
if
it's
a,
if
it's
just
a
notification,
you
know
of
a
current
level
right,
then
so
a
liquid
level.
I
want
to
say
you
know
what
you
must
send
it
to
me
once
every
minute
or
you
must
measure
it
once
every
minute.
But
you
know
if
you
want
to
measure
it
more
frequently,
every
other
10
seconds
it's
up
to
you,
so
my
recommendation
is:
don't
do
it
more
frequently
than
10
seconds.
You
must
do
it
every
minute,
but
that's
your
schedule
here
do
what
you
want.
A
E
E
E
But
I'm
going
to
try
and
put
this
at
the
end,
the
clients.
E
E
I
love
wordsmithing,
you
know
online.
Sometimes
it
gets
into
stuff
all
right
reading
it
again.
One
present
the
minimum
evaluation
indicates
the
minimum
time
in
seconds.
The
client
recommends
for
the
server
to
wait
between
two
consecutive
measurements
of
the
conditions
of
the
resource,
since
the
client
has
no
interest
in
the
server
doing
more
frequent
measurements,
it's
a
bit
long,
but
it
does
count.
C
E
I
don't
want
to
lose
anything,
so
let
me
go
ahead
and
commit
these
changes.
Did
you
did
everybody
agree
on
the
separation
of
the
two
tables,
because
I
did
that
as
just
a
you
know,
a
proposal.
E
C
It's
good
yeah,
I'm
actually
weighing
up
whether
we
should
have
two
sections
to
describe
them,
but
but
let's
have
them
as
two
tables
for
now.
E
Okay,
so
let
me
just
keep
those
as
two
tables.
Let
me
go
ahead
and
commit
this.
I
just
don't
want
to
lose
these
changes.
C
C
Here
so
it
was,
could
you
scroll
a
little
bit
more
just
a
little
bit
more?
Yes,
if
both
ep
mean
and
ep
max
attributes
are
defined,
ep
max
must
be
greater
than
ep
min.
So
is
that
a
greater
than
or
greater
than
greater
or
equals
to
ep
min.
E
Is
if
it's
greater
than
or
equal
to,
I
saw
no
reason
why
you
would
ever
send
ep
men,
so
I
was
always
thinking
it
would
have
to
be
greater
than
okay,
it's
an
edge
condition.
But
to
me,
if
I
said,
ep
min
and
ep
max
to
seven
seconds,
it
makes
no
sense.
I
would
just
not
define
ep
net.
C
Sorry,
I
think
I
was
talking
about
gt
and
lt.
Sorry,
so
yeah
gt
and
lt
are
special
yeah.
They
were
different
yeah
exactly
so.
C
E
E
E
All
right,
good
stuff,
I
think
we're
good
the
the
question
I
guess
is:
did
we
want
to
put
this
in
different
sections,
because
I
could
just
do
that
offline
if
we
want
to
or
keep
it
in
the
same
section.
C
Well,
please
keep
it
in
the
same
section,
because
there's
still
editorial
changes
coming
in,
so
it's
better
not
to
yeah,
just
just
keep
it
and
see
this.
I
think
this
is
good.
E
B
So
as
a
if,
if
my
understanding
of
of
ep
min
and
epmax
is
correct,
this
should
so
the
let
me
first
differently
the
mental
exercise
I'd
like
to
do
or
invite
you
to
do
as
well
is
if,
if
two
observations
are
on
the
on
on
one
resource,
coming
in
through
a
proxy
that
is
aware
of
this
implementing
those
conditional
control
attributes,
can
the
proxy
come
up
with
a
set
of
parameters
based
on
which
it
would
start
just
a
single
observation
and
then
form
all
the
work.
F
E
G
F
Want
yeah
I
mean
this.
This
gives
leeway,
this
doesn't
give
guidance
so,
but
I
agree
with
christian.
This
requires
some
more
thinking.
We
are
not
going
to
do
this
in
my
mistakes.
Yeah
sounds.
E
Good
all
right,
so
the
the
the
pr
is
there
I
mean,
I
think
anybody
can.
You
know,
make
basically
a
pull
request
and
a
my
pull
request.
So
you,
I
don't
think
you
have
rights
on
my
branch,
but
if
you
wanted
to
make
suggested
changes
in
a
my
pull
request,
we
could
absolutely
do
that
and
then
we
could
review
them
in
in
the
next
interim
meeting.
E
All
right,
so
there
is
so
going
back
to
make
tables
readable
and
get
them
a
markdown.
So
I
should
merge
this.
C
E
Yeah
all
right,
we'll
figure
it
out
when
we
look
at
the
merge
conflicts.
Okay,
thanks
guys,
I'm
I
think
thank
you.
E
A
Alan
thanks,
indeed,
and
all
for
the
discussion
you
were
saying:
you're,
also
planning
to
file
more
prs
on
three
additional
parameters.
E
I'm
not
sure
do
you
guys,
I
mean
when
you
look
at
those
parameters.
Did
they
make
maybe
I'll
put
in
a
separate
issue
for
each
one
of
them,
so
we
could
start
a
thread
because
I
don't
to
me
I
would
like
to
I
don't
know
if
you
guys
would
agree
with
that.
So
I'll
start,
I
will
take
three,
those
three
other
attributes
and
I'll
put
each
one
in
a
separate
issue.
C
A
And
just
a
quick
thing
for
the
authors:
there's
still
an
outstanding
open
issue
on
github
number
17
on
proxy
and
caching
considerations.
C
C
A
We
have
four
open
issues
and
speaking
of
which
the
august
one
opened
by
carson
is
closed.
I
think
because
it
was
about
authorship-
and
I
remember
you
changing
the
outer
list
in
the
version
submitted
in
july,
so
this
is
pretty.
E
C
Yeah,
let's,
let's
skip
this
example
and
then
we
try
to
generate
let's
say
something:
that's
more
consistent
with
how
we
are
doing
this
in
the
in
the
in
the
draft.
Also.
I
Oh,
but
thanks
thanks
for
inviting,
of
course.
Of
course
we
can
come
back
to
this.
A
Okay.
Then.
I
think
we
can
adjourn
the
meeting.
Thank
you
all
for
today
have
a
good
day.
Okay,
thank
you.