►
From YouTube: IETF104-EXTRA-20190326-0900
Description
EXTRA meeting session at IETF104
2019/03/26 0900
https://datatracker.ietf.org/meeting/104/proceedings/
A
A
A
B
A
A
B
A
C
Right,
I'm
not
going
to
already
right
through
goals,
I
think
we're
nearly
done
with
the
process.
That's
you
know,
although
it's
an
interesting
thing,
because
I
obviously
had
you
know
some
ideas
by
the
way
of
trying
to
optimize
for
clients
or
for
the
server
and
then
team
of
the
send
a
bunch
of
comments,
and
you
can
see
that
both
might
be.
C
A
C
Port
right
so
last
time,
I
had
two
slides
sort
of
you
know
things
to
add
and
set
things
to
remove
and
clarify
the
good
thing
that
one
of
them
is
empty.
Now,
so
I
think
most
of
clarifications
are
done.
So
we
agreed
last
time
to
add
the
move
command
to
the
base
back.
It's
quite
commonly
used,
clarified
and
solicited
fetches
that
they
might
continue
IDs.
C
Its
simplifies,
quite
significantly
else
up,
was
taken
out,
search
response,
recent
flag,
unseen
response
code
and
select
so
all
the
staff
that
people
either
there
are
better
ways
of
doing
this
now
or
they're,
just
pointless
things
that
service
need
to
do
and
Glines
don't
really
care
about
these
days,
I
also
found
looking
at
RFC
22
31
about
parameter.
Continuations
I
found
some
text
updating,
I
Maps
back,
and
this
was
actually
never
mentioned
an
IMAP.
So
this
was
a
like
RFC
who
are
updating.
Rc
2060,
which
is
imap4.
C
Things
remaining
I
think
more
examples
explaining
use
of
body
structure
with
various
complicated
messages,
especially
with
nested
messages,
and
for
that
would
be
very
useful.
The
still
a
significant
amount
of
bugs
and
various
implementations
in
the
yesterday's
again,
if
you
have
great
examples,
send
them
to
the
mailing
list.
That
would
be
very
helpful,
but
if
not
I
think
we'll
come
up
with
vari
and
I
will
come
up
with
something
yeah.
A
B
C
C
Would
start
with
examples,
because
the
document
has
one
example,
but
it
doesn't
demonstrate
lots
of
diverse
complexity.
Every
time
I
need
to
test
it.
I
test
it
against
my
own
server,
which
you
know
similar
to
Saracen
right
yeah,
so
I
would
say,
add
a
couple
more
examples,
and
then
we
can
see
if
we
can
improve
text.
C
C
So
when
you
can
upload
them,
it
keeps
it
a
little
bit
simpler
and
Chris
raised
the
question
which
I
had
on
to
do
list,
but
it's
not
reflected
in
the
message
in
their
in
their
latest
revision
is
we
need
to
be
clear
whether
binary
fetch
only
applies
to
live
body
parts
over
that
applies
to
intermediate
nodes.
So
what
does
it
mean?
If
you
fetch
binary
a
multi-part
mixed?
B
The
intent
of
binary
is
to
retrieve
a
binary
body
part
not
to
retrieve
in
binary
mode
a
segment
of
the
message.
So
I
think
we
should
just
stick
with
that
and
say
that
binary
is
a
single
body
part,
but
and
yet
a
single
leaf
body
part
does
anyone.
We
have
a
real
use
case
where
they
need
to
use
binary
to
retrieve
a
portion
of
the
tree
that
isn't
just
a
leaf
body.
Part.
C
A
B
C
B
A
B
C
E
D
Chris
Newman
sorry
yeah
implementing
this
for
a
leaf
node
a
leaf
part
is
pretty
straightforward
on
the
server,
but
implementing
this
to
support
yo,
a
subset
of
the
message
gets
more
complicated
and
if
you
actually
wanted
to
implement,
you
know
decoding
the
nested
encoding
case
message:
global,
which
I
never
would
then
it
gets
even
more
complicated,
so
I
I
would
say
we
you
know
I
would
say
the
trade-off.
Is
we
kind
of
want
I
met
four
rep
to
to
actually
deploy,
so
we
want
to
minimize
barriers
to
getting
deployed
within
reason.
C
B
C
C
D
D
A
C
B
C
C
C
B
C
D
A
B
D
Chris
Newman
I
agree
with,
should
not
the
two
cases
I've
encountered
where
it's
very
important
to
just
hang
up
is
like?
Is
you
know
in
the
middle
of?
If,
if
the
client
starts
delaying
in
the
middle
of
less
negotiation,
you
can't
send
it
by
you.
Cannot
the
only
option
is
closed
detection
and
then
you'll
also,
you
know
the
other
example
was
special
fetching
large
message
actually.
B
A
C
C
Well,
basically,
it
was
just
a
statement
of
fact
that
some
server
Street
mailbox
names
has
kids
a
sensitive
because
they
might
be
using
file
system
on
the
OS,
which
is
itself
case,
insensitive
and
ASCII
range.
So
this
is
client.
Clients
cannot
rely
on
this
one
way
or
the
other,
and
you
know
its
server
implementations
have
a
choice
which
way
they
want
to
do
it
small
statement
of
fact
than
you
know
right.
C
B
B
A
B
A
C
C
Yes,
this
section
needs
your
work.
What
happened
is
Chris
contributed
text
to
move
modified,
you
gsm,
seven,
to
appendix
sort
of
as
an
optional
backward
compatibility
thing,
and
this
wasn't
cleanly
edited.
So
it
might
give
you
a
wrong
impression
that.
B
D
F
D
It's
enabled
it
then
the
the
behavior
of
a
bear
ampersand
is
different,
because
if
the
client
hasn't
enabled
the
bear
ampersands
an
error
because
it's
it
has
to
assume
the
MMU
tf7,
if
it's
enabled,
then
a
very
ampersand
yo
is
allowed,
and
if
your
internal
storage
is
mu,
t
f7,
you
turn
the
bear
amperes
and
into
a
frozen
so
yeah.
This
is.
This
is
implementable.
C
C
D
B
B
D
Chris
Newman
that
the
fact
the
matter
is
they're
stuck
with
plaintext
passwords
anyway,
so
the
as
much
as
I'd
like
to
promote
scram,
sha-256
I,
think
promoting
it
with
as
a
mandatory
to
implement,
isn't
the
right
way
to
do
it.
Yeah
it's
yo,
I
think
it
needs
to
be
encouraged
by
the
community,
but
I,
don't
think
you
know
I.
Think
people
will
just
ignore
that
and
and
we're
trying
to
impose
our
our
wishes
on.
You
know
on
implementers
that
can't
follow
that.
D
Basically,
the
issue
is,
if
you
tie
into
an
identity
system
and
any
medium
to
large
deployment,
has
to
tie
into
an
identity
system
yo,
you
can
only
implement
what
the
identity
system
supports.
So
yo.
If
we
put
in
a
mandatory
to
implement
we're,
saying
we're
saying
something
that
we
know,
people
will
have
control
and
can't
deploy,
but.
F
D
B
D
D
B
Just
for
the
record,
I
suspect
that
we
might
get
some
pushback
from
the
sec
ADEs
about
the
about
not
having
a
mandatory
damn.
It's
awesome
at
what
we'll
burn
that
we'll
burn
that
bridge.
If
we
get
to
it
well
sure
this
bridges,
throat
anyways,
but
that's
still
not
mandatory
to
implement
and
they
often
want
mentor
template
we'll
see
where
we
go.
That's.
G
B
C
B
B
Wise
you
just
stick
in
a
null
string.
We
can
certainly
deal
with
the
syntax,
wise
I
guess.
My
thought
is
that
if
well,
there's
two
things:
if
you
need
it:
syntax,
wise
and
you're
arbitrarily
sticking
in
a
null
string.
Why
not
stick
in
an
actual
string,
but
there
is
some
complexity
to
implementing
a
non
null
string
there,
as
I
found
out
when
I
first
implemented
it.
F
B
B
A
B
E
C
Elsa
Elsa
is
gone
yeah.
Well,
there
is
no
this.
The
comment
was
about
no
selects
when
you
do
select
and
string
right,
no
list
empty
string,
and
then
you
get
a
hierarchy
separator.
If
you're
writing
it
from
scratch.
I
would
have
fixed
it,
but
I'm
tempted
just
to
leave
it,
as
is
because
clients
still
need
to
handle
it.
The
way
it
is
so.
Yes,
it's
a
glitch,
but
I.
Just
yes
wouldn't
be
well.
B
C
B
C
C
A
B
C
A
C
C
C
B
E
C
C
C
B
B
B
A
A
B
B
B
B
D
A
C
B
A
C
C
B
B
C
C
B
B
C
C
C
Reflects
permanent
lights
be
sent
after
new
keyboard,
I
think
dance
is
know,
yeah
the
reason.
Well,
so
we
have
slash
star
special
marker,
saying
I,
support
88
right
new
keyboards
and
can
store
that
if
the
server
turned
this
marker,
you
store
a
new
keyboard.
There
is
no
need
to
announce
it
because
and
the
command
succeeded
it
stored
it.
There
are
some
implementations
that
have
a
limited
number
of
the
key
words
for
them.
At
some
point.
B
B
B
C
A
B
C
I
was
trying
to
remember
this
is
one
of
my
this
year
right,
so
I
put
the
fetch
preview
on
the
first
teller
jet
in
April,
so
I
nearly
put
it
before
but
transition,
but
it
we
had
quite
a
few
very
busy
torture.
So.
B
A
B
B
A
B
A
A
Planning
for
Montreal
will
definitely
I
expect
one
to
meet
there
again,
we'll
have
a
new
charter
and
a
new
work
to
look
at
wood.
We
don't
need
to
go
over
the
Charter
again
stay
unless
anyone
has
anything,
no
one
pose
for
that
dinner
at
tonight.
I
didn't
call
for
show
of
hands.
Anyone
who
didn't
already
say
they
were
coming
at
the
Jay
map
session
yesterday
or
a
mummy
got
up
and
walked
perfect
three
more
I
will
I
will
change
from
twelve
to
fourteen
excellent.