►
From YouTube: IETF-TAPS-20230515-1500
Description
TAPS meeting session at IETF
2023/05/15 1500
https://datatracker.ietf.org/meeting//proceedings/
A
A
C
A
D
D
A
Welcome
everybody
to
our
Taps
interim
meeting
and
I
sent
an
updated
agenda
in
which
we
are
going
to
be
discussing
potentially
some
of
our
outstanding
issues
on
our
documents.
First
and
then
we're
gonna
dive
into
our
implementation,
discoverability,
mappings
and
next
steps
agenda.
D
F
F
Like
thank
you
so
much
because
this
is
in
the
browser.
Strangely
yeah.
If
Safari
lets
me,
do
it
then
I'll
do
it.
F
A
You
mean
right
away:
yeah,
okay,
sorry
yeah.
Let's
see
Grant
English,
it's
Queens
here:
okay,
yeah
yeah,.
A
Yes,
I
can
see
it
great
sorry.
Did
you
Corey's
asking
for
screenshot
as
well.
F
Shall
we
go
through
this?
These
pull
requests.
B
F
Sure
all
right,
let's
see
I'll,
take
you
from
the
bottom
I
think
mostly
they
should
be
easy.
F
Yeah,
okay,
this
one
I
can
see
that
Tommy
here
tried
to
get
away
from
being
just
about
sctp
by
saying,
for
example,
but
for
example,
it
doesn't
really
fit,
because
that
indicates
that
there
would
be
a
general
explanation
about
this
in
the
sentences
before
and
then
comes
an
example,
but
it
doesn't
fit
in
that
way.
So
my
proposal
is
to
just
explain
in
general,
when
the
underlying
transport
protections
supports
multi-streaming.
F
Yeah
yeah
yeah
can
do
that.
Multi-Streaming.
H
E
E
I
mean
the
such
as
sctp
is
shouldn't
that
be
a.
E
H
F
C
F
C
I
F
F
This
was
really
just
something
that
our
reviewers
stumbled
over
that
it's
at
minimum
here
and
he
thought
that
maybe
it
should
say
maximum
or
should
I
say
both
or
should
it
say
which
so
we
fixed
that
these
are
the
easiest,
harmless
things
to
do:
system
policy.
Oh
yeah,
we
have
two
okays
already
anybody
any
comments
on
this.
This
was
what
is
about
even
yeah.
F
C
An
issue
sorry
yeah.
The
issue
suggested
a
number
of
things,
so
one
was
just
emitting
that
end
Allowed
by
the
system,
which
I
think
is
fine.
Then
there
was
a
broader
comment
from
the
reviewer
about.
Oh,
should
this
all
be
normative
and
I
I
think
we
should
avoid
that.
F
C
D
C
Right
and
then
the
other
part
was
if
there
is
an
essentially
not
an
error,
but
you
know
if
a
connection
can't
be
created
because
of
some
policy
reason:
where
does
it?
When
does
it
fail
and
I
think
that's
what
this
new
sentence
here
is
saying
is
like
it's
when
you
create
the
connection
like
through
initiator.
Listen,
it's
not
on
the
pre-connection
right,
which
is
good
yeah.
That's.
E
C
E
C
Yes,
maybe
we
should
clarify
instead
of
just
saying
such
failures.
You
say
like
if
such
failures
prevent
a
connection
from
being
established.
F
C
F
F
C
C
F
Yeah
exactly
right,
so
the
initiate
is
not
so.
F
F
Anyway,
I
pressed
the
wrong
button,
sorry
system
policy,
okay,
it's
updating,
and
then
there
was
this
from
Brian
with
three
your
case.
Do
we
even
need
to
discuss,
shall
we
save
time
and
just
merge.
F
H
F
F
H
F
F
C
B
What
I
think
it
is
we
haven't?
We
already
had
some
kind
of
consensus
that
we
wanted
to
avoid
uppercase
language
there.
Okay.
F
H
F
F
C
F
H
F
Is
Swift,
but
these
are
not
many
pull
requests
and
many
more
issues
left.
Oh
yeah,
that's
the
bigger
one
from
Brian
I
made
a
tiny
suggestion
which
I
think
he
committed
already.
F
F
This
one
this
was
suggesting
that
there
is
no
text
at
all
regarding
management
of
Opera.
J
I
think
this
is
sort
of
like
the
stock
up
steer,
yeah
comment,
where's
my
operations
and
manageability
section,
and
so
I
made
something
up.
Yeah.
H
J
F
F
Very
much
how
you
wrote
this.
It
does
also
this
co-existence
of
the
socket
API
question
here.
That
is
also
addressed
so.
H
G
D
G
D
G
F
Right
suggesting
that
we
are.
K
C
F
J
G
C
J
J
C
J
J
K
Yeah
I
mean
the
other
point.
Is
this
error,
logging
or
whatever,
which
is
not
wrong?
But
you
know
that's
a
very
big
topic
and
I
think
it's
really
out
of
scope.
Yeah.
J
J
C
F
Yeah
then
come
a
few
more
that
well
I
mean
I,
have
to
admit
I
didn't
even
just
get
too
carefully
read
and
deal
with
all
of
them.
It
was
a
lot
of
luck
to
even
just
copied
them
over
did.
Does
anybody
want
to
discuss
this
I
mean
I
can
just
make
a
stab
at
it
after
this
meeting
tomorrow
or
something,
and
we
can
just
I.
C
F
C
G
H
F
L
C
C
G
G
C
F
I
H
F
Is
worthwhile
looking
for
the
implementation
job
to
see
if
there
is
anything
that
we
say
you
have
to
do
this
and
it
maps
to
behavior
in
the
interface
draft
to
make
sure
that
it
really
is
there
just
to
see
that
we're
not
missing
anything
I
can
do
that.
It's
an
annoying
job,
but
I
can
do
that.
I
think
it
makes
sense
to
be
thorough
here.
E
F
Okay,
so
I'll
write
this
here
all
right,
given
that
introduction.
H
H
F
F
Couple
of
the
others
to
other
people-
let's
see
yeah
the
next
one
is
three
nomenclature
and
language
needs.
So
that
was
quite
some
text
about.
I
E
C
I,
don't
yeah
I'm
looking
for
it.
None
of
this
seems
like
a
huge
problem.
I
mean
he's
not
pointing
out
that
we're
saying
that
some,
like
some
streams,
don't
require
a
handshake
and
so-
and
we
just
say
like
that,
so
it's
unlike
TCP
you're
saying
we
should
call
it
that
it
is
like
UDP.
Instead,
it
was
like
okay,
I,
don't.
H
F
Handling
connectionless
protocols
that
was
the
next
world-
oh
yeah,
this
was
I,
think
also
just
a
tutorial.
The
reason
I
didn't
do
it
yet
is
because
he's
proposing
to
shift
some
text
from
here
to
there.
So
it's
a
bit
more
work.
H
F
E
D
F
F
A
F
H
F
F
C
E
F
M
B
We
might
also
want
to
double
check
for
the
API
doc
whether
we
would
allow
to
magically
mapping
to
things
to
to
Multiplex
connections.
So,
let's
say
the
HTTP
case,
you're
opening
a
new
con
HTTP
Connection
in
quotes
to
send
in
request
the
questions,
whether
you
would
be
allowed
to
be
multiplexed
on
an
existing
connection.
B
F
F
C
H
F
E
E
E
C
C
E
F
E
F
E
H
M
M
It's
either
the
sort
of
protocols,
the
minimum
of
the
protocol,
specific
values,
yeah
or
something
smaller
based
on
a
previous
previous
connection,.
F
H
F
F
M
K
M
F
K
Yeah,
but
that's
more
a
system
parameter,
not
a
connection
parameter.
What
I'm
saying
is,
if
you
have
to
provide
the
data
already
when
you
establish
it
like.
When
you
create
the
connection
object,
then
it
doesn't
help
you
to
ask
the
maximum
length
after
that
anymore,
because
you
already
provided
the
data.
K
But
then
you
need
a
difference,
then
it's
not
a
connection
based
interface
or
you
have
to
ask
the
previous
connection.
It.
F
H
G
H
K
G
B
It's
that
way,
I
guess
for
you
today,
but
we
want
to
avoid
it
for
tips.
K
K
H
J
F
C
J
G
C
B
F
D
D
F
H
H
F
F
F
F
H
C
Right
right,
it
sounds
like
what
he's
saying
he's
like
our
original
text
is
essentially
just
saying
like
look,
your
five
Tuple
may
go
away,
and
then,
if
you
come
back
yeah
and
because
it
can
come
back,
you
know
don't
tear
down
your
connections
too
aggressively.
What
a
given
transport
protocol
does
with
this
is
entirely
up
to
it.
Like
sockets
already
can
handle
this
like
a
TCP
socket.
That
loses
a
route.
C
I,
don't
think
we
need
to
go
into
the
detail
of
what
every
transport
protocol
does
to
do
its
data
reliability
again,
it's
not
relevant
to
this
I
agree.
G
G
M
F
Or
profit
cost
that
Services
System
would
be
using.
We
do
not
think
that
such
details
at
here.
F
Yeah,
okay,
it's
okay,
oh
mine,
as
well
yeah.
F
It's
beautiful
specific
transfer
protocol.
I
B
E
H
H
E
F
E
F
C
Exactly
if
you
would
like
to
see
having
for
dtls
he's
writing.
D
F
B
F
F
J
F
F
F
F
H
M
This
this
is
just
the
this
is
what
happens
by
default.
If
you
don't
specify
a
source
address
right,
foreign.
M
C
F
F
F
F
Exactly
Let's
change
my
command
into
OK.
Somehow
I
don't
need
to
do
that.
I
can
just
merge.
It
find
the
buttons.
F
F
All
right,
so
we
do
these
things
soon
and
we're
done
with
it
can
I
stop
sharing.
A
Thank
you,
oh
okay,
so,
with
that
out
of
the
way,
let's
get
into
our
next
item,
implementation,
update
and
I
will
leave.
Tommy
has
a
slide
set.
C
Away
all
right,
thank
you,
yeah.
These
are
relatively
light,
slides,
but
just
a
bit
of
a
update
and
we
can
have
other
people.
You
know
chime
in
with
what
they're
aware
of
what
they
think.
C
And
you
know
we
have
a
couple
that
we
know
of
with
within
Apple
we
have
our
Network
framework,
which
is
Swift
and
C.
There's
also
been
a
couple
python
attempts
that
we've
previously
presented
on
in
this
group
and
just
the
overall
comment
here
that
I
wanted
to
make
is
you
know
these
are
not
exact
matches
for
the
current
API
draft
they're
based
on
earlier
versions,
or
you
know
they
have
some
slight
inconsistencies
there
very
much
in
the
spirit
of
TAPS
and
they
follow
the
general
shape.
C
C
Overall
I
think
this
strategy
here,
then
that's
what
I've
been
thinking
of
for
our
implementations
is
that
you
know
we
probably
want
to
see
some
rev
of
the
API
and
implementations
and
see
more
implementations
of
that
API
once
we
have
kind
of
like
the
RFC
locked
up
and
finished,
and
maybe
we're
at
that
point
now
or
or
maybe
you
know,
we
just
say
you
know
in
a
couple
months
or
so-
we'll
be
done
enough
and
have
the
approvals
such
that
we're
not
going
to
see
a
major
change.
C
C
It
is
useful
to
have
that
API
be
stable
and
so
to
have
alignment
with
the
RFC
I
think
it
makes
sense
to
wait
until
that
RFC
is
done
just
as
an
example
like
I
I,
don't
we
have
not
been
updating
the
network
framework
apis
every
year
to
match
up
to
whatever
the
draft
has
currently
been,
because
that
would
cause
a
lot
of
application
churn
and
would
not
really
be
valuable
to
anyone,
but
I
think
once
we
have
the
rfcs.
C
What
I
would
hope
to
do
is
have
a
revision
of
our
public
apis
to
align
with
that
and
offer
more
kind
of
compatible
symbols
that
are
really
obviously
Taps
related,
not
as
a
python
Library,
author,
but
I
think
it
would
be
great
if
we
could
see
similar
updates
and
formalization
of
the
Python
bindings,
and
this
could
also
be
an
opportunity
for
seeing
more
formal
efforts
in
other
languages.
C
We'd
had
previous
discussions,
at
least
for
our
implementation
of
having
C
plus
bindings,
potentially
trying
to
pitch
those
as
standard
Library
bindings
as
well
it'd,
be
great
to
see
formal
things
can
go
and
rust.
Etc.
C
And
just
you
know,
Drew
this
a
little
flow
chart
here,
so
you
know
for
the
purposes
of
apis
I
think
the
first
thing
that
needs
to
come
to
really
cause
the
expansion
of
the
implementations
is
having
the
standard
reference
that
they
can
all
use
and
build
off
of.
Then
we
can
see
more
implementation
work.
We
can
see
that,
after
that
deployment
and
release
of
the
apis
is
things
that
are
built
into
different
platforms
or
into
different
libraries,
and
then
we'll
see
the
application
to
option.
C
You
know
a
a
bit
of
a
road
that
starts
with
the
RFC
kind
of
finalization
before
we
see
the
widespread
availability
and
then
adoption
of
something
that
is
more
formally
taps
I
think
it
would
also
be
interesting
once
we
have
this,
and
once
we
have
implementations
deployed,
that's
a
you
know.
We
update
our
apis.
We
get
some
other
apis
in
other
languages
available
to
have
a
wave
of
trying
to
increase
the
awareness
and
promote
those
implementations
and
talk
about
how
they
are
based
on
a
standard
right.
J
We
call
on
ipfix
right,
like
ipfix,
with
standardized
based
to
off.
You
know
largely
off
of
Cisco's
implementation,
and
then
a
bunch
of
people
tried
to
implement
it
and
that
led
to
significant
simplification
of
the
like
the
internet,
Standard,
Version
and
and
I
would
expect
that.
We're
probably
going
to
see
that
too.
If
we
get.
J
If
we
get
to
to
stage
four
I,
think
that
you
know
either
the
tax
working
group
or
some
area
successor
organization
should
probably
take
on
moving
the
rfc's
to
internet
standard,
because
I
suspect
we're
going
to
learn
a
lot.
C
That
was
actually
my
last
sentence
on
my
last
slide
of
hey.
You
will
probably
need
to
come
back.
Like
you
know,
one
of
the
questions
Martin,
you
know
in
general
the
chairs
and
ads
need
to
consider.
Is
you
know
what
do
we
do
with
the
group
and
we
can
do
work
on
mappings,
but
the
other
comment
here
based
on
this
chart
is
yeah
like
we'll,
probably
get
feedback.
We
may
want
to
do
some
businesses
at
some
point,
which
is
like
a
long
cycle.
C
C
Yeah
so
totally
agree
with
you
Brian
there
for
the
mappings
work.
I
think
that
is
the
only
other
thing
that
seems
clear
that
we
could
do
document
wise,
quick
and
HTTP
are
the
clear
candidates.
I
I
would
still
need
to
revise
the
work
I'd
done
for
the
quick
one.
C
I,
don't
think
it's
likely
that
those
would
exist
anywhere
outside
of
TAPS.
We
would
certainly
want
to
cross
pollination
in
review,
but
I
I
don't
see
a
world
where
the
quick
and
HTTP
protocol
groups
would
actually
be
defining
these.
So
if
we
want
to
see
them
happen,
we
should
do
them
here
and
that's
all
I
have.
So
what
do
people
think.
G
Yep
yeah
well
on,
would
it
be
adopted
in
quick
Martin
thinks
it
would
be
adopted
in
quick.
L
I
didn't
exactly
promise
that
I
mean
I.
Think
I
said
that
it
would.
It
would
certainly
been
setting
aside
the
Practical
questions
like
I
think
it
would
be
nice
to
at
one
point
we
kind
of
had
this
and
then
the
different
protocol
groups
could
could
then
do
mappings,
rather
than
just
sort
of
be
in
the
matching
business
forever
and
Tommy's
the
chair.
L
They
see
this,
of
course,
so
I'm
going
to
refer
to
his
guidance
there
I,
if
we're
interested
in
showing
the
Downs
like
I'm
I,
think
it
would
be
sensible
to
to
ask
quick
what
they
think
and
if
they
want
to
take
it
I
mean
certainly
there's
a
lot
of
the
same
people.
So
I,
don't
think
so.
The
quick
working
group
would
have
plenty
of
caps
experience
that
wanted
to
take
this
on.
It's
not
like
it's
I,
don't
know
like
Co-op
or
something
on
the
on
the
dormant
question.
L
So
I,
don't
think
that
head
is
is
on
right,
he's
not
on
right
now
and
I
haven't
discussed
this
with
him,
so
I'm
not
so.
Take
this
with
a
grain
of
salt,
but
his
pattern
and
RM
cat
was
to
leave
it
open
until
the
until
the
documents
got
published,
so
that
gives
us
a
fair
amount
of
Runway
to
be
dormant.
Assuming
these
implementations
are
forthcoming
relatively
soon.
C
Yeah
but
Martin
I
mean
I
I
assume
you
just
mean
the
base
rfcs
being
published
and
like
given
my
phone
I,
don't
expect
I
mean
you
know
once
once
in
the
RFC
editor
Q,
then
I
think
it
makes
sense
to
do
a
revision,
but
you
know
for
ourselves
with
our
Network
framework,
like
we
do.
You
know
major
API
revisions
normally
come
out
in
June.
C
That's
not
going
to
be
this
June
because
we
don't
have
it
all
finalized
so
like
maybe
next
June,
it
would
be
like
the
earliest
that
it
first
goes
out
to
have
any
developers
try
and
then
we
want
to
talk
about
other
languages
like
I.
Think
it's
a
much
longer
time
frame
than
the
RFC
is
being
published.
C
L
Sure,
before
I
give
up
the
the
floor
here,
so
the
other
thing
I
would
like
to
sort
of
mention
as
a
possibility.
Here
is
my
my
transport
Discovery
draft,
which
you
know
I
my
previous
employer
I,
was
excited
about
I'm,
not
so
excited
that
my
current
employer
and
just
to
briefly
remind
people
of
the
idea
here,
is
that
if
you
like
user
space
protocol
implementation,
implementations,
there's
like
a
secure
way
just
for
register
those
with
the
Taps
that's
running
over
the
OS,
so
you
can
use
them
and
I
I.
A
A
I
I
But
it
just
seems
to
me
that
the
you
know
the
active
group
has
become
really
down
to
just
the
sort
of
the
Die
Hard
authors
and
I
feel
like
the
what
occurs
to
me
as
a
reasonable
way
forward
and
I'm.
Just
putting
this
out
here
as
a
a
proposal
for
discussion,
not
really
even
a
recommendation,
but
what
seems
intuitive
to
me
is
to
have
the
working
group
go
dormant
and
leave
the
mailing
list
open
as
a
point
for
collaboration
between
the
authors
and
any
implementers.
I
And
then,
when
you
get
a
critical
mass
of
topics,
you
know
when
there's
some,
like
you
know,
pressure
to
create
a
charter
and
restart
the
working
group
to
do
that
and
I.
Don't
think
it's
clear
that
the
mappings
work
I
mean
we've
tried
to
to
see
if
there's
a
a
critical
mass
folks
who
want
to
collaborate
on
that.
I
It
looks
like
the
answer
is
probably
no,
but
if
it
turns
out
that,
if
somebody
takes
it
up
and
they
run
into
some
problems-
and
it
generates
like
some
interesting
discussion-
then
you
know
that
might
be
sort
of
an
example
of
an
anchor
that
you
could
use
to
to
reset
the
charter.
But
I
feel
like
just
starting
to
add
more
topics
onto
this.
What
it's
going
to
do?
It's
just
going
to
put
more
stuff
onto
the
plates
of
the
folks
who
are
on
this
call
now
and
it.
I
I
M
Largely
agree
with
that,
what
we
seem
to
me
to
be
too
keen
on
closing
working
groups
and
which
raises
a
very
high
bar
to
restarting
the
work,
rather
than
putting
a
milestone
for
three
years
in
the
future
that
says
sort
of
review
status
and
decide
whether
to
continue
or
close
down
I
I
guess
you
know
the
mappings
are
yeah,
it's
not
clear.
We
have
the
energy
to
do
them
for
the
follow-on
implementations.
M
I
think
it
would
be
nice
to
see
implementations
and
standardized
apis
for,
like
rust,
for
example,
do
we
know
of
anyone
who's
doing
that
work
or
your
other
plans
to
encourage
someone
to
do
that?
Work.
C
M
B
J
I
B
So
the
other
option
would
be
leave
the
working
group
open
and
Charter
click
and
HTTP
gender
protocol
mappings
for
in
three
years.
M
I
mean
one
thing
we
did
in
RM
cat
was
to
add
a
milestone
for
like
a
couple
of
years
in
the
future.
From
from
the
point
where
the
documents
happen
to
say,
so,
you
review
your
review
review,
whether
there's
been
any
implementations
and
whether
we
need
to
progress
with
the
group
or
close
down.
L
I'm
not
sure
what
the
difference
is
between
like
running
with
no
Charter,
no
deliverables
and
no
beatings.
But
closed
is
different
from
just
being
but
open,
as
you
think,
different
from
being
closed
because
ultimately,
you'll
just
have
to
I
mean
closing.
The
group
doesn't
mean
that
we
can't
reach
outer
it
someday
and
bring
it
back.
If
there's
like
a
lot
of
capital,
specific
work
but
I'm
not.
I
M
Creating
a
new
working
group
is
a
much
bigger
esque
than
you
know,
resurrecting
an
existing
group
and
just
saying
so,
do
we
need
to
do
any
work
here?
Let's
revise
the
milestones.
L
Well,
I,
I,
guess
I
guess
it
depends
on
the
Goodwill
of
the
transport
AIDS
at
the
time,
which
is
something
I
can't
I'm,
not
a
responsible
idea
anyway,
but
I
can't
really
oblige
the
the
obligate
the
future
ads
to
just
sort
of
say.
Yes,
this
is
an
obvious
thing
and
let's
rubber
stamp
it
to
start
again.
K
L
So
I
think
you
can
certainly
make
a
case
this
work
and
go
elsewhere.
But
if,
if
this
group
has
a
lot
of
energy
to
like
go,
do
a
lot
of
task
works
and
I?
Suppose
it's
fine
for
it
to
be
a
different
working
group,
but
I
I
do
think
that
we
get
more
eyeballs
in
these
other
working
groups
and
they
all
all
those
working
groups,
have
a
critical
mass
of
like
tax
people.
G
Martin
is
probably
serious
when
he
said:
tsbwg
could
kind
of
play
the
keeper
and
do
a
small
thing
or
do
a
presentation
to
try
and
bootstrap
restarting
this
work.
That
would
seem
well
within
the
charge.
What
tsbwg
does
so,
maybe
putting
the
thing
dormant
for
three
years,
with
the
only
item
being
a
milestone
to
see
if
we
ever
wake
up
seems
like
a
bit
of
an
odd
creation,
really.
A
G
No,
no,
no,
no
I
mean
once
the
ads
closed.
The
group.
If
there's
relevant,
work
on
a
topic
which
has
been
previously
published
in
another
working
group,
then
there
is
an
expectation
that
that
would
be
bootstrapped
somehow
by
tsvwg.
Whether
tsvwg
takes
it
tolerance
as
well
that
speed
to
work
we'll
work
on
or
whether
it's
talks
to
the
ad
and
after
it's
being
talked
about,
and
we
say
it
was
critical
mass
here.
How
do
you
want
to
proceed?
Maybe
they
want
to
quickly
spin
up
a
working
group
or
place
it
somewhere
else?
That's!
Okay!
C
And
I
you
know,
I
had
already
been
thinking
like
for
the
mappings,
depending
on
the
timing.
You
know.
Certainly
a
quick
map
I
mean
would
make
sense
to
present
to
quick.
But
let's
say
you
know,
quick
was
interested,
wanted
to
review
it
but
said
at
not
really
in
our
Charter
to
do
this
API
mapping
work.
Then
we
say:
okay,
you
know
the
quick
mapping
document
lives
in
tsvwg
as
far
as
how
it
gets
published.
C
That
is,
you
know
an
option.
We
could
try.
I
So
I
guess
the
question,
in
my
mind,
is
whether
Taps
is
sufficiently
defined.
I
mean
going
back
to
the
go
back
to
your
your
phase.
Chart
Tommy,
because
I
think
this
is.
This
is
I,
think
the
way
we
should
be
thinking
about
it.
I
You
know
what
we
want
is
application
adoption
and
is
tap
sufficiently
defined,
that
we
can
do
that.
You
know
modular
the
fact
that
we
might
learn
things
that
cause
us
to
want
to
make
changes.
I
You
know
is
there?
Is
there
enough
Taps
to
enable
that?
I
And
if
there
is
then
maybe
we
should
put
the
pens
down
and
let
that
work
happen
and
figure
out
what
the
issues
are
rather,
rather
than
invent
issues
that
we,
you
know,
we
don't
know
what
they
are
yet
or
you
know
be
a
a
zombie
working
group
if
there,
if,
if
the
gaps,
if
you
know
if
the
mappings
are
a
blocker
and
the
other
working
groups,
are
unwilling
or
unable
to
do
the
work,
then
then
that's
a
reason
to
do
it
right.
That
means
that
our
work
isn't
complete.
H
I
Don't
know
what's
going
to
come
up
in
the
editing
process,
that's
there
that
we
want
to
bring
in
more
than
just
the
authors.
I
I
mean
I,
don't
think,
there's
a
race
to
close
the
working
group,
but
I
think
you
know
what
we're
talking
about
here
is
whether
we
want
to
identify
new
Charter
items
to
work
on
right.
We
can
just
sort
of
we
can
leave
the
group
open
in
in
you
know,
kind
of
keep
it
flexible
for
a
while,
but
I
I,
don't
I.
I
Think
the
question
we're
talking
about
here
is
whether
we
want
to
identify
work
items
and
rally
the
12
folks
on
this
call,
because
there
don't
seem
to
be
many
others
to
to
do
that.
Work
and
and
I
just
based
on
and
I
I'm,
not
close
enough
to
the
implementation
efforts
and
and
Martin
did
say
that
there
was
somebody
wasn't
ready
to
identify
themselves.
I
But
if,
but
the
sense,
I've
gotten
from
Tommy's
comments
is
that
it's
going
to
take
a
while
before
we
start
to
get
really
useful
feedback
from
not
only
the
implementation
committee,
but
also
the
application
community
on
what
to
do,
and
it
seems
like
you
know,
what
are
we
going
to
do
between
now
and
then
other
than
like?
Try
to
to
justify
our
existence
as
a
working
group?
So
I?
I
G
I'm
right
on
that
last
sentence
you
made
as
in
I,
would
love
to
see
this
thing
fly.
I
think
it's
really
important.
I
would
happily
make
time
available
to
make
the
next
tax
document,
but
I
feel
what
the
responsibility
here
is
that
someone
has
to
enter
the
room
and
say
why
that
document
has
to
be
written.
So
if
it's
Tommy
coming
back
saying,
hey
we
implemented
it.
We
now
have
more
wisdom.
G
Nibs,
really
quick
is
what
we're
going
to
say,
but
it
falls
in
the
same
thing,
because
it's
not
us
that's
going
to
do
this.
It's
somebody
going
to
come
along
and
say:
Taps
is
wonderful,
it
doesn't
support
quick,
but
I
want
to
do
it
and
it's
the
I
want
to
do
it.
That
really
has
a
big
swear
with
me
right.
J
M
J
Have
a
we
don't
have
a
a
blog
post.
Somebody
should
write
it.
I
mean
I.
Think
I
wrote
one
for
apnic
like
a
long
time
ago,
when
this
was
you
know
before
we
were
mostly
done,
but
yeah
I
mean
I'm
I'm,
happy
to
I'm
happy
to
to
collaborate
on
that
with
somebody
Michael
Tommy
yeah.
C
And
you
know
we
should,
you
know,
talk
to
Craig,
Wood,
et
cetera,
exactly
but
I
think
it's
like
as
I'm
drawing
here.
Once
we
have
an
RFC
and
once
you
know
we
have
you
know
it's
like
our
API
has
updated
to
match
the
RFC
or
whatever,
like
once.
We
hit
some
critical
point
of
like
okay,
the
API
is
implemented
and
you
know
you
can
adopt
it
in
a
couple
different
languages.
C
That's
when
you
want
to
do
the
push
and
I
think
that'd
be
a
great
thing
to
have
the
working
group
list
open
for
and
all
of
us
coordinate
on
what
should
go
into
that.
But
we
don't-
and
you
know,
maybe
even
we
you
know
co-opt
some
tsv
area
meeting
sometime
and
you
get
some
time
there
to
talk
about
it.
But
we
don't
need
the
working
group
to
be
active
to
do
that.
F
I'd
be
very
happy
to
call
to
contribute
to
a
blog
post
or
something
like
that.
But
I
also
agree
that
it
would
be
great
if
Apple
at
some
point
further
down
in
time,
understand
that
it's
going
to
take
a
while.
But
if
at
some
point
there
will
be
a
more
clear
statement
about
really
being
aligned
with
Taps.
C
A
N
N
N
If
everything
goes
fine,
we
go
into
RFC
editor
artist.
Editor
r48
I
would
like
to
keep
this
working
group
up
and
running
until
the
output
is
done
so
that
anything
changed
in
the
document
that
we
have
a
chance
to
actually
call
on
a
working
group
and
say,
like
the
changes,
make
sense.
The
other
option
is
like
all
the
things
like.
We
can
close
the
working
group
right
now
and
I
do
ads
launcher
or
like
anybody,
any
area
register.
Does
an
area
sponsor.
That's
like
a
list
preferred
option
for
now.
N
What
are
the
other
options
we've
been
hearing
like
the
dormant
and
all
this
thing,
I
think
my
preference,
like
close.
If
we
are
not
doing
anything,
we're
waiting
for
information
to
happen,
then
we
close
the
working
group
and
come
back
again
later
on
with
all
our
questions
and
other
other
things,
and
then
you
have
like
tsbwg,
quick
or
other
working
group
or
just
somehow
keep
the
test
mailing
list
open.
So
we
can
have
discussions
there
rather
than
having
that
working
group
open.
N
So
that
would
that
those
are
like
my
preferred
Solutions.
So
keep
the
what
types
working
on
until
we
published
RFC
and
then
also
keep
the
mailing
list
up.
If
we
are
closing
the
working
group,
anything
else
other
than
that
would
need
some
sort
of
thinking
and
why
you
should
do
it
just
my
couple
of
cents
here,
thanks.
L
Regarding
just
this
Outreach
thing,
it
seems
to
me
that
our
actual
blocking
constraint-
this
point
is
os's.
I
mean
the
value
proposition
that
app
developers
is
look.
You
can
like
write
this
and
run
it
everywhere.
Right
now,
everywhere
is
iOS,
so
like
rather
than
just
I
mean
I
can't
block
those
are
good
and
it's
something
that
we
could
point
to,
but
to
the
extent
there's
like
a
Taps
cabal
I
think
we
should
try
to
try
to
actually
actively
Target
OS
communities,
whether
it
be
you
know,
the
Linux
Community.
L
Previously
someone
going
over
to
Microsoft
and
sweet
talking
them.
That
would
actually
be-
and
you
know
of
course,
Android
as
well,
which
I
guess,
Brian
and
I
would
have
to
have
the
ball,
for
that
would
actually
be
I.
Think
the
most
effective
way
to
try
to
get
it.
Get
this
to
be
a
major,
a
major
part
of
the
ecosystem.
I
So
I
have
a
a
question
that
will,
as
usual,
expose
my
ignorance,
I
think
to
the
to
the
rest
of
the
group,
which
is
I'm
thinking
about
like
one
of
the
other
areas
that
has
activities
that
has
been
useful
in
encouraging
integration.
Implementation
has
been,
you
know
the
hackathon,
but
I.
You
know,
I
think
what
we're
talking
about
here
is
sort
of
implementer
support.
Answering
questions.
Helping
people
understand
things
you
know
make
trying
to
make
it
easy.
Is
the
hackathon
a
useful
form
for
doing
that?
I
We've
also
done
some
Outreach
I.
Think
Maria
went
to
like
you
know,
like
Linux
networking
stack
conference,
kind
of
thing,
I'm,
I'm,
I'm,
reacting
to
Martin's
comment,
which
I
think
is
really
dead
on,
which
is
you
know
to
enable
applications?
We
need
to
support
OS
implementation.
C
I
was
going
to
make
a
comment
to
Martin
that
well,
you
know,
while
I
agree
it's
critical
to
have
this
available
across
a
diversity
of
os's.
It
doesn't
necessarily
need
to
be
the
OS
implementers
doing
it.
You
know
it's
certainly
I
think
viable
to
have
it
be
libraries
that
are
open
source
and
cross-platform
or
implemented
in
languages.
C
I
Would
soliciting
maybe,
like
an
ia,
sorry
internet
Society,
grant
for
implementers
in
the
open
source?
World
be
a
useful
thing
here.
J
Maybe,
but
we'd
probably
want
to
identify
implementers
first
right,
I
mean
so
like.
We
can
certainly
help
with
that
with
that
framework,
but
we'd
want
to
we'd
want
to
identify
like
people
to
do
the
actual
implementation
work
before
we
go.
It's
like
let's
get
some
money
and
then
find
implementers
I
think
is
not
a
great
model
for
that
for
the
ice
hock
grants.
A
A
B
This
question
I
think
my
main
motivation
going
into
taps
or
moving
that
stuff
forward
was
originally
trying
to
do
fancy
stuff
with
http
and
additional
socket
options
and
finding
that
this
socket
API
was
completely
unfit
for
it
and
so
I
think,
as
we
see
that
most
traffic
today
is
http,
and
especially,
if
I
look
at
what
I'm
doing
my
day
work
at
the
moment
this
this
is
80
HTTP,
at
least
I
would
say
this
is
extremely
valuable,
especially
if
you
think
about
programmers
that
often
use
http
apis
and
would
love
to
see
it
more
on
the
operating
system
for
from
there,
I
would
say,
having
an
HTTP
mapping
and
probably
also
nice
HTTP
implementation
would
be
really
valuable.
N
I
N
N
So
you
also
have
to
convince
people
like
why
we
need
an
HTTP
tabs
mapping.
If
the
deployment
is
not
there.
B
So
I
think
web
trans
is
quite
of
orthogonal
to
Taps
mapping,
because
web
transport
is
just
piggybacking.
Different
transport
over
Taps,
sorry
over
HTTP.
So
we
I'm
rather
talking
to
application
developers
that
are
not
within
a
web
browser,
but
just
using
HTTP
to
access
apis
and
other
stuff
and.
H
B
B
Now
I
have
no
concrete
number
of,
but
I've
I've
seen
a
lot
of
API
traffic
in
the
in
the
background
where
we
just
have
a
rest
model
to
connect
some
sort
of
API
and
usually
the
libraries
used
there
if
you're
lucky,
it's
actually
lip
curl.
If
you're
unlucky,
it's
something
that
isn't
able
to
do
anything
close
to
happy
eyeballs
and
if
you're
unlucky,
it
does
not
even
Implement
a
connect,
call
based
on
Stevens
but
just
trust.
B
But,
on
the
other
hand,
most
of
I'm
I'm,
seeing
that
that's
not
eyeball
traffic,
is
a
lot
either
implemented
in
go
or
as
implemented
in
a
JavaScript
or
it's
implemented
in
Java
and
using
these
worlds,
apis,
which
are
not
perfect.
To
be
honest,.
N
Yes,
I
mean
yeah,
I
mean
I
mean
those
coming
into
those
weird
things
like
would
be
benefited
upon.
Http
mapping
over
tabs
I
mean
if
there
is
representation
from
them
asking
for
something
that
we
know,
and
we
can
judge
and
evolve
it
so
that
we
have
a
problem
to
solve
there.
Actually,
then,
it's
a
good
motivation.
A
Yeah,
so
I
wanted
to
not
just
towards
as
we're
talking
about
HTTP
mappings
and
I
saw
a
comment
about
quick
mappings
in
the
chat.
What's
the
status
of
a
or
quick
mapping
document
which
exists
but
I
believe
has
expired
and
be
in
a
potential
HTTP
mapping
document?
Do
we
have
any
volunteers
to
contributes.
A
C
I
mean
I
I
had
written
that
quick
mapping
document
I'm
happy
to
keep
working
on
it
as
time
permits
and
also
happen
to
help
work
on
HTTP.
It
would
be
useful
to
have
you
know
other
contributors
or
co-authors
I
I've
based
some
of
the
mappings
on
what
we
have
done
since
we
have
implemented
both
quick
and
HTTP
in
here,
but
I
I
think
to
be
viable
to
be
very
useful,
to
have
others
who
come
from
other
perspectives,
be
able
to
contribute
as
well.
C
H
N
Lucky
then
so
yeah
on
the
quick,
I
think
Martin
Martin.
You
made
a
comment
if
it
is
a
quick
mapping
and
then
quick
working
on
perhaps
we'll
have
more
audience
and
more
like
expertise
on
those
those
kind
of
things.
N
So
the
suggestion
was
like
that
work
could
go
to
a
quick
working
group
or
something
like
that.
Anybody
has
any
concrete
comment
on
on
Martin's
View
I
just
wanted
to
know.
I
A
Like
no,
we
can't
hear
you
yet
I
think
we
briefly
talked
about
whether
or
not
a
quick
document
would
be
a
quick
mapping
would
be
I,
guess
how
it
would
be
received
in
quick,
I
forget
where
we
landed
on
that.
D
C
Can
you
clarify
so
the
the
I.
N
Think
the
comment
I
I
saw
from
Martin
Duke
was
like.
If
we're
doing
a
quick
mapping
of
taps,
quick
working
group
would
will
have
like
more
audience
and
expertise
to
actually
because
they
will
be
the
mostly
the
user
of
that
API
right,
a
mapping.
So
it
might
be
good
to
actually
move
that
word
to
Quick
work.
We've
done
it
rather
than
keeping
it
in
Taps.
I
mean
I
just
wanted
to
know
like
what
all
other
people
here.
C
C
However
I
my
guess
is
that
it
would
still
not
be
something
that
the
working
group,
the
quicker
working
group,
would
actually
want
to
take
on,
given
that
that
is
where
the
protocol
happens.
Like
we
didn't
do
the
TCP
mappings
in
tcpm
and
I
I
do
disagree.
That
I,
don't
think
it's
the
people
within
the
quick
working
group
that
would
be
using
the
Taps
mappings
right.
C
So
I
think
it's
more
likely
that
we
would
review
that
work
in
quick
but
run
it
in.
You
know
either
some
form
of
this
spring
group
or
in
tsvwg.
N
I
So
this
reminds
me
of
you
know,
early
on
in
in
Taps
we
tried
very
hard
to
engage
with
the
apps
area
when
it
was
just
the
apps
area.
To
you
know,
to
try
to
identify
users
and
users
needs
from
taps
and
I'm.
Taking
my
Tommy's
point,
which
is
a
good
one,
you
know,
but
I
don't
know
if
we've
got
the
inside
of
this
group,
because
we
never
really
had
like
great
participation
from
the
apps
area.
I
But
but
it
sounds
like
the
you
know.
The
the
place
to
to
focus
on
user
support
are
people
who
who
would
get
benefit
out
of
TAPS,
but
don't
understand
it
enough
to
be
able
to
use
it
or
don't
or
not,
don't
understand
it,
but
but
would
use
it
if
it
existed.
I
You
know
so
early
adopter
types,
you
know
I,
guess
Tommy's
got
a
captive
audience
at
Apple
or
I,
don't
know
if
it's
captive
or
not,
but
but
at
least
in-house
users,
where
you
both
provide
an
operating
system
in
the
applications.
You
know,
maybe
we
should
try
to
find
a
you
know.
Some
other
examples
focus
on
me
supporting
them.
G
A
J
Yeah
I
I
think
that
I
think
that
well
I
mean
foreign
we're
talking
not
an
ietf
meeting
this
year,
possibly
toward
the
end
of
next
year.
Depending
we
probably
want
to
another
next
step
is:
is
the
ITF
blog
post?
We
probably
do
want
that
to
come
out
around
about
the
time
the
RFC
goes
and
I'm
happy
to
I'm
happy
to
hold
a
pen
there.
I
I
would
like
some
help
on
that,
but
I'm
happy
to
hold
the
pen
there.
G
Since
it
was
my
plan
for
my
plan
that
actually
probably
came
from
Tommy
in
creation,
I
mean
the
idea
that
we,
the
rfcs,
have
been
put
out
there
and
we
can
plan
a
hackathon
around
it.
That's
going
to
require
a
bit
of
planning
as
well.
So
we
do
that
and
we
figure
out
what
we're
trying
to
do,
and
maybe
at
that
point
we
bring
anything
else.
Bubbly
tough
and
say
well
tops
is
still
active,
but
this
is
not
happening
in
the
ITF
yet,
but
but
we
kind
of
discuss
it
working
out.
G
I
think
this
is
a
useful
side
meeting
and
I
think
we
could
even
make
it
a
little
bit
public
by
pointing
people
out
while
tapped,
isn't
meeting
as
a
working
group.
So
there's
no
active
documents
but
come
along
here
if
you're
interested
in
Taps
to
see
if
other
people
turn
up
or
not.
If
they
don't,
we
can
still
plan
it
if
they
do
all
the
better.
N
Okay,
cheers
I,
think
you
you
too,
and
I
need
to
talk
a
bit
more,
perhaps
on
these
things,
but
let's
catch.