►
From YouTube: TAPS WG Interim meeting 2019-12-20
Description
TAPS WG Interim meeting 2019-12-20
A
My
last
review
of
arch
that
made
me
think
okay.
Well,
maybe
we
need
to
rethink
that.
There
were
two
instead
of
knit
fix,
pr's
that
were
still
pending
that
I
just
merged
like
during
the
security
can
like
four
minutes
ago,
and
there
is
one
remaining
bug
on
arch.
If
you
on
arch
that
was
filed
by
Colin
18
hours
ago
from
the
Train
Colin,
do
you
want
to
July?
B
A
A
D
B
A
A
B
H
F
A
E
A
A
K
L
F
A
D
M
B
D
I
I
think
there's
probably
going
to
be
a
couple
different
cases.
Sometimes
your
transport
protocol,
the
littoral
transport
protocol,
like
it's
UDP
or
it's
quick
and
will
literally
have
messages,
and
then
you
do
anything.
Sometimes
you
have
in
your
stack,
something
like
HTTP,
which
effectively
gives
you
messages
and
can
be
viewed
as
a
transport,
although
is
not
always
viewed
as
classical
transport.
The
other
cases
are
you.
Can,
you
can
add
your
own
frame?
Yes,.
B
The
typical
case
is
probably
you're
using
an
existing
trance-like
protocol
like
HTTP,
for
example,
and
there
is
a
framer
implied
by
that
choice.
You
don't
implement
it
the
way
the
text
was
written.
It
seemed
to
imply
that
you
you,
as
an
application,
programmer
write
framers,
but
then
it's
option
to
write
framers
but
usually
they're
supplied
by
the
state.
Yes,
this
text
right
here.
A
C
K
H
H
K
D
B
D
M
K
C
P
Q
O
A
L
Q
O
A
A
K
H
H
A
B
D
B
N
A
D
D
D
A
K
S
B
P
I
A
I
would
I
would
actually
drop
this
as
well.
I
think
it's
this
overspent.
We
don't
need
to
specify
at
this
level
on
the
architecture
right,
it's
something
that
would
be
important.
For
instance,
right
here,
yeah
right,
the
implementation
can
try
it
at
the
implementation.
Decides
that
you
want
to
shoot
yourself
in
the
foot
and
lets
you
do
it.
Congratulations
you're
in
c++
time,
if
the
implementation
won't.
Actually
let
you
build
the
code.
If
you
try
to
do
that,
then
congratulations.
You're
in
ruslan
I
mean.
A
A
But
the
Apple,
so
the
question
is:
is
that
like
I
mean
we're
way
down
in
the
weeds
and
why
an
application
like
when
IP
that's
great
like
so
generally,
you
know
the
99.9%
cases.
You
set
the
properties
at
free
establishment
time
and
we're
done
this
so
like
this
is
already
you've
already
kind
of
broken
the
glass
and
we're
starting
to
mess
with
the
interns.
In
a
way,
unexpected
things
will
happen
to
you.
A
B
Q
B
B
A
A
A
D
Well,
I
mean
how
you
send
the
message.
I
think
is
broad
enough,
because
it
could
include
both
the
things
that
go
on
the
wire
in
how
you
send
it
and
the
things
that
do
not
go
on
the
wire
in
how
you
send
it
all
right.
You
give
like
oh
this
text,
my
timing
or
my
ordering,
whereas
this
other
thing
impacts
the
potentially
like
the
ECM
flags
that
get
marked
on
the
IP
packet.
B
A
That's
another
thing:
do
we
want
to
like?
Do
we
want
to
actually
have
one
er
that
just
rear
apps
all
the
text?
In
this
document,
we
want
to
keep
like
the
other
documents,
actually
keep
to
80
80
columns.
This
is
a
different
yak
to
shave,
but
it
also
means
that
all
outstanding
pr's
have
to
be
merged
and
nobody
has
to
have
anything
branched
because
it'll
break
emerge.
If
we,
if
we
reflow
that.
T
H
S
D
A
D
A
B
D
B
D
B
D
N
D
A
I
E
P
A
P
K
D
L
Talk
about
this
in
the
architecture
at
all,
because
I
think
there
is
a
way
in
the
interface
to
have
closer
connection
by
setting
setting
the
connection
to
unidirectional
at
the
certain
point
in
time.
So
there
are
too
many
ways
that
potentially
could
do
this
and
I
think
we
have
to
X
out
this
really
in
the
interface
and
don't
talk
about
half-closed
in
the
architecture.
It's
a
corner
case.
G
J
A
B
P
D
A
P
B
G
A
D
D
B
B
G
A
A
B
A
K
A
G
A
B
O
B
B
B
B
B
D
D
Well,
so
how
I
guess
Pat
selection
in
this
terminology?
What
well?
Okay
right,
if
you
zoom
out
to
the
overall
structure
of
here,
we
or
we
already
have
two
sections
here-
we
have
one
candidate
gathering,
one
for
candidate
racing
gather
just
like
the
gathering
is
merely
identifying
things.
That's
not
what
excess.
A
B
D
A
What
what
what
Tommy
is
saying
is
that,
at
least
in
his
concept,
those
are
the
same
action
because
in
his
implementation,
they're,
probably
the
same
action
in
a
protocol
like
is
they
are
not
necessarily.
There
are
different
phases.
They
are
necessarily
tempered
temporarily
separated
by
the
mechanics
of
the
well
and
what.
D
Know
that
I
think
we
are
miss.
We
are
disagreeing
on
what
the
word
selection
here
means,
which
I
think
is
understandable
right,
so
it
is
selecting
for
what
not
selecting
to
use
but
selecting
to
try
or
to
send
information
about
to
another
peer
selecting.
What
is
the
set
of
things
that
I'm
even
willing
to
operate
on
or
exchange
the
other
side
or
even
try
racing
on
that
is
what
I've
seen
a
Canada
pool.
So
that's.
Why
is
gathering
candidates.
B
B
D
D
G
D
J
A
F
I
D
A
A
M
A
I
M
D
N
B
A
A
F
F
B
F
A
B
N
B
F
K
F
A
A
F
A
Original
idea,
because
the
original
idea
from
Singapore
was
the
this
was
going
to
be
architecture,
call
and
I
think
it
was
many
think
it
and
then
the
next
call
it
was
going
to
be
and
I
think.
If
we
get
like
one
or
two
good
reviews
in
on
that,
and
we
can
have
sort
of
the
sync
kind
of
a
spectrum
on
your
face
today
and
we
can
turn
around
twenty.
A
A
F
So
so
my
thinking
is
exactly
along
the
lines
of
yours
Brian,
but
what
I
want
to
do
is
get
a
couple
people
to
commit
to
do
a
close,
read
and
sounds
like
clones.
Not
gonna
have
time
before
the
call
you
will
do
it
is
there
somebody
on
this
bridge
who's,
not
an
author
who
can
sign
up
to
do
for
you
to
the
document
before
the
19th
of
January
I
will.
R
F
That
addresses
my
anxiety
about
because
what
I
don't
want
to
do
is
have
a
whole
bunch
of
stuff.
Have
you
know
multiple
emails
that
Collin's
come
out
of
working
group
last
call
because
then
it's
going
to
take
a
lot
of
work
for
us
to
you
know
to
get
the
revision
done
when
we
want
to
be
turning
our
attention
to
the
implementation
document.
So
this
is
good.
This
works
wink
so
Kyle
max
you
don't
not
do
it
because
you're
not
there.