►
From YouTube: IETF-TOOLS-20211206-1800
Description
TOOLS meeting session at IETF
2021/12/06 1800
https://datatracker.ietf.org/meeting//proceedings/
B
C
I'm
not
sure
I
can
stay
on
for
the
entire
thing
robert,
because
it's
a
public
holiday
tonight
today,
okay
tonight
so.
B
C
Yeah
that
we
actually
cancelled,
but
it's
still
a
public
holiday,
and
we
were
we
just
came
back
today
from
another
like
trip
inside
of
finland,
and
my
my
week
is
totally
you
know,
filled
up
by
igf
stuff,
which
is
also
started
today.
So
I'm
trying
to
make
some
some
time
for
the
other
people
in
the
house.
B
So
we
do
have
a
pretty
tight
schedule.
I
want
to
get
started
pretty
soon,
but
I'd
like
to
see,
I
was
hoping,
we'd
have
jay
and
it
would
be
also
be
good
if
we
could
have
rust
before
we
get
started.
B
So
greg
maddie,
michael
rich
and
alexi,
please
do
go
ahead
and
share
video.
The
point
of
holding
this
meeting
this
way
is
so
that
we
are
simulating
being
around
a
table
at
an
ietf
meeting.
It
would
be
good
to
see
each
other's
faces
as
we're
talking
about
things.
B
Right,
I
was
also
hoping
to
have
more
community.
You
know
just
not
usual
tools,
team
participants
here,
michael
I'm
glad
that
you
were
able
to
join
us
advertising
this
just
to
the
tools
discuss
list
might
not
have
been
casting
the
net
wide
enough
to
get
people
to
spread
information,
but
we'll
go
with
who
we
have
here
thanks
everyone
for
spending
some
time.
B
B
Well,
their
ietfa
hosted
list
memberships.
That's
going
to
lead
us
into
a
discussion
about
minimum
3,
which
I
think
is
going
to
eat
a
large
bit
of
time.
B
B
Sorry
and
then
we
have
some
some
details
to
go
through
that
will
be
follow
on
from
mailman
three
that
you
can
see
in
the
agenda.
I'm
expecting
to
have
the
conversation
around
moderating
a
given
participant
to
take
a
little
bit
of
time
as
we
get
into
this.
Does
anybody
think
that
we
should
be
achieving
something
else
within
going
through
these
things?
In
this
conversation,
and
for
those
of
you
that
need
to
leave
early?
Is
there
any
reordering
that
you
want
to
inflict
before
we
get
started.
D
B
So
this
session
is
being
recorded.
It's
a
media
code
session.
We
will
make
the
link
available
to
people
later
so
keep
that
in
mind,
as
as
we're
having
the
conversation.
B
Let's
go
ahead
and
jump
in.
I
have
been
getting
some
interruptions
from
people
looking
for
the
links
that
have
how
how
to
get
here.
So
apologies
for
me
stutter,
starting
as
we're
going
through
this.
B
There
is
a
link
at
the
bottom
of
the
ietf
mail
processing
graphic
that
will
let
you
pull
up
a
full
resolution.
Version
of
it
feel
free
to
follow
that.
If
the
small
resolution
version
that's
in
the
main
link
hang
on,
I
need
to
paste
a
medical
link
to
some
people
that
can't
find
it.
B
So
the
way
that
male
gets
processed
is
we're
going
when
male
arrives
at
ietfa.
The
first
thing
that
it
does
is
it
goes
into
postfix
and
postfix
throws
it
through
some
checks
to
make
sure
that
it's
coming
through
places
that
it
should
come
through
then
runs
it
through
a
virus
scanner,
as
described
later
in
the
document,
runs
at
the
spam
assassin.
B
It's
very
permissive
about
what
it
does
with
the
results
of
spam
assassin.
At
that
point,
it
leaves
most
of
the
discard
for
spam.
Assassin
happen
later,
but
if
it's
very
clearly
spam,
it
will
get
killed
at
that
spot.
B
This
is
all
mail
going
to
ietf.org
or
any
at
our
irtf.org
anything
that
would
land
on
ietfa.
B
B
So
for
those
of
you
that
have
just
joined,
please
go
ahead
and
share
video.
If
you
can
we're
trying
to
replicate
the
experience
that
we
would
get
from
a
conversation
around
the
table,.
B
So
again,
this
is
all
mail
coming
in
to
ietfa
goes
through
a
virus
scanner
gets
marked
by
spam.
Assassin
then
goes
through
a
set
of
routing
based
on
aliases.
It
looks
at
the
these.
A
this
is
one
of
the
first
set
of
aliases
causes
it
to
go
through
post,
confirm,
post,
confirm
is
a
program
that
looks
to
see
if
this
sender
has
been
confirmed
as
a
sender
that
should
the
mail
should
be
allowed
to
go
through
the
iet
a
processing
chain.
B
If
it's,
they
don't
find
this
person
in
the
yes,
this
person's
been
confirmed
list.
They
get
a
message
sent
to
them
the
mail
they
sent
goes
into
quarantine
when
they
reply
to
that
message,
there's
a
key
in
the
message
that
causes
the
receiving
thing
to
say.
Okay,
I
can
release
this
cued
message
and
put
this
address
into
the
approved
senders
list
as
you'll
see
later
in
the
document.
There's
a
description
about
how
the
approved
senders
list
is
pre-populated
with
all
the
subscribers
to
the
ietf
list
and
several
other
feeds.
B
B
Once
it
gets
through,
this
mail
will
land
on
one
of
several
destinations,
the
the
as
you
go
across
the
top
of
the
page.
Here
things
might
get
sent
to
the
data
tracker.
We
have
several
data
tracker
ingestion
points.
B
These
are
python
management
commands
django
management
commands.
They
accept
the
mail
as
a
blob
and
and
cause
records
to
be
written
into
the
data
tracker.
The
big
feeds
here
are
nom-com
feedback,
some
ayanna
review
mail
and
there
is
a
an
adjuster
for
conversations
about
ipr
disclosures
when
the
secretariat
sends
a
message
to
someone
through
the
data
tracker.
B
There's
a
reply
to
that
has
a
key
in
the
reply
to
messages
coming
back
to
that
reply.
To
will
have
their
body
scraped
and
be
added
to
the
history
of
the
ipr
disclosure
that
particular
endpoint
gets
a
lot
of
spam.
The
spam
is
getting
dropped
on
the
floor.
Mostly
we've
got
some
work
that
we're
going
to
do
to
clean
up
the
the
way
that
particu.
What
happens
when
spam
arrives
at
that
spot?
B
B
There
are
places
where
we
have
explicitly
defined
external,
like
off
system
places
that
the
mail
should
go.
They
get
routed
to
that
spot,
the
if
they
get
routed
to
you
again.
I
guess
this
is
where
the
things
run.
Wrap
by
dmarc
would
go
to
that
destination
external
and
be
delivered.
B
On
that
list
we
go
through
the
adding
the
again.
The
post
confirm
challenge.
Adding
the
archive
dat
header
then
dropping
in
mailman
mailman
does
its
normal
checking.
Mailman
can
be
configured
to
remove
any
dkim
headers
that
came
in
on
the
message
and
our
instance
is.
Then
we
have
a
extra
code
that
we
have
added
that
on
outbound
processing,
we'll
put
in
a
new
dqm
signature.
B
B
B
E
Now
around
the
world
experience
go
ahead.
Nine
video,
apparently,
is
the
max.
Oh
you're
kidding
yeah.
Apparently
that's
an
alex
a
my
question
was
about
the
post
confirm
I
was
reading
the
mailman
upgrade
and
I
guess
that's
the
major
piece,
the
major
bit
that
is
different
between
two
and
three,
and
so
I'm
wondering
to
what
extent
is
our
post
confirm
wildly
different
than
the?
What
was
there?
B
Post
confirm
is
our
own
code,
even
in
what
we're
running
right
now
and
when
we
move
to
mailman
we're
going
to
have
to
re
rewrite
it
essentially
because,
right
now
it
works
by
mailman,
twos
extension
interfaces,
one
of
which
is
you
just
wrap.
The
the
command
line
called
mailman
the
command
line
command
called
mailman
to
get
into
mailman.
B
There
are
a
few
other
command
line
places
where
mailman
expects
to
be
able
to
get
list.
Manipulation
through
that
are
vastly
different
in
mailman
three,
so
we
will
have
to
rewrite
both
how
to
get
to
post
confirm
and
what
post
confirm
does
down
inside
to
work
with
mail
and
three.
Now
it's
one
of
the
smaller
differences,
but
it's
a
it's
a
big
one
and
it's
one
we're
going
to
have
to
do
it
is
one.
B
B
So
the
bigger
difference:
how
alexis
we
see
you
now,
the
max
of
nine
is
kind
of
surprising.
I
thought
we
worked
out
something
different
with
medico,
but
we'll
we'll
get
along
in
that
path.
Later,
the
bigger
thing
is
going
to
mailman
3,
they
refactored
the
way
their
application
works
completely.
B
B
B
Yeah
so
postorious,
we
will
probably
end
up
using
pieces
of
postorious
for
doing
things
like
allowing
chairs
to
moderate,
held
messages
unless
we
want
to
rewrite
those.
B
So
jay
is
there
anything
in
particular
that
you
would
like
to
call
out
as
something
that
is
perhaps
subtle
or
really
important
in
this
sketch
that
we
could
get
a
conversation
on
in
feedback.
F
Well,
no,
well,
maybe
I
just
had
it.
I
had
a
question
from
the
the
the
processing
bit
beforehand,
just
yeah
the
at
what
point
is
the
archive
url
decided
is
that
when
it's
delivered
to
the
archives
or
is
there
some
already
pre-calculated,
it
is
pre-computed.
F
So
it's
is
it
included
in
a
header
in
the
mail
message
or
anything
at
all
or
available
in
the.
B
It
is
added
it
is
added
to
the
mail
before
it
goes
into
mailman.
If
it
knows
that
it's
going
to
to
a
list.
The
archive
that
header
is
added
to
the
message
so
that
it's
present
as
it's
going
through
the
mailman
processing
and
it
ends.
B
In
the
archive,
so
any
any
message
it
has
been
sent
to
you
through
mailman.
If
you
do
a
view
source
and
hit
control.
U
there
is
an
archive.header
that
has
the
link
to
where
that
message
is
archived.
F
Yeah,
I
can
see
that
now.
Yes,
it
is
so
we
we
could
if
we
wanted
to
make
that
an
explicit
url
at
the
bottom
of
the
message,
as
well
as
part
of
the
the
footers
if
we
wanted
to
in
in
the
footer
yeah.
F
If
we
wanted
to
okay,
I
just
thought
so
that,
on
the
the
the
the
the
ux
stuff,
that
I've
done
here
on
improv
use
the
list
management
experience.
This
is
basically
bringing
together
everything
that
I
have
heard.
People
talk
about
about
the
list
management,
so
we
have
multiple
different
ways
that
people
subscribe
at
the
top.
F
I
know
this
from
doing
the
all
of
itf
survey
and
receiving
the
complaints
from
people
who
subscribe
using
their
own
domain
name
and
multiple
different
email
addresses
within
that
domain
name,
and
so
that's
why
we
have
the
private
domains
bit
there,
as
well
as
the
people
who
just
have
multiple
different
addresses,
subscribed
and
who
use
different
addresses
for
different
mechanisms.
So
that's
why
there,
the
pause,
mail
delivery
tool
addresses,
I
think,
murray
asked
for
that
or
someone
asked
that,
which
is
why
that's
there.
F
So
you
might
want
someone
else
to
explain
that
because
I
don't
fully
understand
it,
but
it's
there,
the
subscriptions,
as
you
can
see,
we've
got
two
at
the
very
top
which
are
like
special
ones,
the
important
news
one
and
the
direct
survey.
One
then
the
rest
all
follow
a
pretty
straightforward
mechanism
where
we've
got
them
in
categories
and
you
can
subscribe
for
everything
in
the
category
or
not,
and
you
can
change
your
address
and
you
know
a
variety
of
bits
and
pieces
like
that.
F
The
other
thing
that
I've
been
trying
to
do
here
is
help
people
understand
the
difference
between
a
highly
used
important
list
and
they
never
used
in
15
years,
but
everyone's
too
scared
to
close
it
list
and
which
to
me,
is
a
failure
of
nerve
but
never
mind,
and
so
that's
why
we
have
the
volume
listed
there.
So
that
which
is
the
best
indicator
I
could
think
of
for
giving
people
an
idea
of
what
it
is.
I
mean
it
may
be
too
complicated.
F
So
that's
basically
it.
I
think,
for
that.
B
G
E
G
G
No,
no,
not
me
sorry.
Let
me
think.
G
F
G
B
We
will
have
in
mailman
or
even
in
mailman
3
the
ability
to
cause
mailman
to
essentially
cue
delivery
to
say,
stop
and
hold
mail
until
I
release
the
floodgates
and
then
cause
many
months
of
mail
to
fall
out
later.
I
think
that
if
we
implement
this
at
all,
it
will
have
to
be
a
don't
deliver
directly
to
me.
H
It
was
eric
klein
that,
because
I
thought
I
was
going
crazy.
I
This
is
god
I
would
find
like
five
reasons.
One
is
sometimes
I
just
go
away
for
a
week
and
I
would
prefer
to
pick
stuff
up
from
the
archive
and
the
other
is
that
I
don't.
Actually.
I
pick
up
all
my
mail
from
the
imap
server,
so
I
you
know
I
saw
I
subscribe,
so
I
can
send
mail
to
list,
but
in
fact
I
have
other
ways
to
receive
it.
G
It
puts
a
burden
on
the
servers
and
I
don't
imagine
I
can't
think
of
a
package
that
a
list
package
that
has
ever
been
able
to
say
go
back
from
data
x
to
date,
y
and
resend
everything
I
should
have
gotten
to
me
now
yeah
for
one
thing
it
causes,
I
mean
unless
you
do
some
funky
things
it
makes
it
makes
threading
difficult,
there's
a
lot
of
reasons
why
that
would
be
a
difficult
thing
to
get
totally
right.
So
I
wouldn't
make
that
a
priority.
B
G
B
B
B
So
we're
running
kind
of
close
to
my
original
proposed
schedule.
A
little
bit
behind
we've
got
this
pending
desire
or
pressure
to
move
to
middleman.
Three.
I've
captured
some
of
the
reasons
why
we
would
want
to
do
it
in
the
list.
B
I'm
sure
that
that
there
are
many
more,
but
the
biggest
in
my
mind
is
is
that
mailman
2
is,
is
officially
no
longer
maintained
for
anything
except
critical
bug
fixes,
and
there
is
a
there
will
come
a
point
in
time
where
the
python
python
2
will
stop
even
getting
maintenance
and
the
packaging
for
that
on
most
on
a
lot
of
systems
is
already
falling
off.
B
B
So
there
are
a
lot
of
things
like
what's
suggested
in
the
graphic
that
jay
provided
the
mock-up
that
jay
provided.
That
will
be
a
lot
easier
to
do
with
the
python
libraries
that
mainland
3
exposes
compared
to
what
we
would
be
able
to
do
inside
mailman
2,
where
we
would
essentially
be
dancing
with
the
shell
command
line
interfaces
to
manipulate
lists,
and
perhaps
it
moves
us
closer
to
idn,
iea
eai
support.
But
we'll
talk
about
that
in
a
minute.
It
may
be
that
that's
still
a
little
bit
out
of
reach.
B
Because
I've
already
called
out
that
you
know
post
confirm
is
running
under
python
2.
Some
will
have
to
move
to
python
3
when
we
make
this
migration
python
3
is
fundamentally
a
slightly
different
beast,
as
I
called
up
before
it's
split
into
these
different
modules.
If
you
follow
the
links
to
the
official
documentations,
you
can
see
the
the
descriptions
of
the
core
module
postorious.
B
B
We
might
choose
to
extend
postorious
to
do
these
kinds
of
things
that
to
make
it
aware
of
what's
happening
in
the
data
tracker
or
write
a
parallel
interface
that
that
allows
the
integrated
manipulation
that
it's
talking
about,
they've
moved
archiving
off
to
sit
on
top
of
hyperkitty.
It's
something
that
I
don't
think
we'll
use.
We've
already
got
our
archiving
figured
out.
B
So
do
we
have
anybody
on
the
call
that
has
had
first-hand
experience
with
moving
a
large
number
of
lists
from
mailman
two
to
mailman
three
in
particular
a
large
subscriber
body
from
mailman
to
mailman
three.
B
So
I
probed
the
community
a
couple
of
years
ago
for
basically
experience
a
person
and
got
a
lot
of.
Please
don't
do
this.
It
hurts
so
much
the
pain,
the
pain
kind
of
response.
B
C
It
email
robert,
but
I
I
guess
we
have
to
do
this
migration,
given
the
state
of
mailman
two
and
python
two
right:
it's
really
not
an
option
to
like
not
do
anything
anymore.
C
G
B
G
E
Cool
robert,
I
I
think
the
the
critical
question
that
the
ietf
will
want
to
know
is
exactly
how
we're
going
to
make
the
transition.
I
think
that's
actually
probably
the
harder
question
than
the
actual
you
know,
objections
or
whatever,
and
to
that
extent,
that
the
question
might
be
more
along.
The
lines
like
one
way
is
to
move
some
lists
from
the
old
system
to
the
new
system
through
some
sense
of
a
forwarding
right,
and
so
the
question
might
actually
be.
Are
there
volunteers,
rather
than
are
there
objections
so.
A
I
wouldn't
write
on
the
same
machine,
sorry,
but
I'm
gonna,
I'm
going
to
jump
in
here.
I
put
my
hand
up
and
I
apologize
to
everyone
for
this,
but
at
jay's
direction
on
the
sandbox
server
we
set
up
mailman,
3
and
mailman.
3
is
now
running
on
the
sandbox
server
in
its
native
form,
and
we
are
also
running
mailman
2
on
the
sandbox
server
and
what
we
found,
at
least
with
the
current
releases,
is
that
it's
relatively
simple
to
move
a
list
from
mailman
2
to
mailman
3..
A
It
can
be
done
on
the
server
and-
and
so
you
can
have
in
essence,
lists
using
mailman,
two
and
different
lists
using
mailman
three
and
we
we
found
that
we
can
take
a
list
and
just
switch
it
over
it's
a
manual
process,
but
not
arduous
from
mailman
two
to
mailman,
3
and
and
have
everything
just
work
now
that
does
not
speak
to
post,
confirm
mail
archive
data
track
or
anything
else.
That's
just
a
test
setup
that
we
did
at
jay's
direction
and
and
just
completed
last
week.
A
So
apologies
to
all,
but
I
just
wanted
to
jump
in
and
say
that.
B
Well,
jumping
in
is
what
we
should
be
doing
on
on
the
call.
Don't
anyone
hesitate
to
jump
in
when
when
you've
got
something
to
say
so
that
is
news
that
I
didn't
expect
and
I'm
glad
to
hear
it
that
we
we
could
take
this
incrementally
as
we
go
through
the
lists
and
it
sounds
like.
Maybe
we
could
get
a
test
list
set
up
and
start
working
on
making
sure
that
we
have
a
functional
integration
from
mailman
three
with
post,
confirm
and
feeding
things
into
the
archive
before
we
start
moving
real
traffic
into
it.
B
A
Yeah,
I
can,
I
can
send
you
the
information
after
we
do
this
meeting.
If
you
like
it,
we
maddie
did
the
bulk
of
the
work
we
just
completed
it
a
few
days
ago.
I
haven't
had
much
time
to
play
with
it
myself,
but
I'll
se
and
again,
you
know
all
respect
to
michael
we've
only
done
testing.
We
don't
know
how
it's
going
to
work
in
production,
so
you
know
a
large
scale.
We
could
run
into
things
I
we're
just
at
the
beginning
stages,
but
I'll
send
you
that
info.
E
Robert
offline,
I
didn't
imagine,
it
was
possible
to
run
them
on
the
same
machine.
I
was
imagining
that
you're
gonna
wind
up
with
another
vm
and
that
that
there
would
be
essentially
a
gradual
but
effectively
a
forklift
upgrade
through.
You
know
use
of
smtp
between
the
machines,
but
but
what
you're
telling
me
is
it
sounds
much
simpler
than
I
had
imagined.
So
that's
probably
good
news.
A
Yeah
we
could
certainly
michael.
We
could
certainly
do
that,
and
you
know
I'm
no
opposition
for
me,
like
I
always
say,
I'm
just
the
operator.
I
do
what
they
tell
me,
but
but
we
we
did
want
to
see
if
it
could
be
done
on
the
same
machine,
to
make
the
lift
and
shift
easier,
and
we
did.
A
We
did
run
into
some
some
kinks
with
you
know,
making
it
work
with
the
data
tracker
because
of
the
django
interface
and
blah
blah
blah,
but
through
her
great
credit,
maddie
got
all
that
sorted
out,
and
so
yes,
robert,
is
running.
We've
got
the
hyper
kitty
which
we
know
we
will
want
to
replace
with
ryan's
much
much
superior
archive
software
and
we've
got
the
pistorius
and
we've
got
the
engine
all
running
and
we've
sent
messages
to
lists
and
it's
all
been
cool
and
groovy.
At
this
point,.
B
All
right
well,
when,
when
you
get
that
information
put
together
for
me,
I
will
send
some
a
summary
to
the
tools
discuss
list,
so
we
can
get
some
extra
eyes
poking
at
it.
B
Unless
somebody
else
has
something
that
they
know
of
to
bring
out
about
the
challenges
that
we're
going
to
run
into
moving
to
mailman
three,
I
think
I'd
like
to
move
ahead
into
some
of
the
rest
of
the
call
and
we'll
be
a
little
bit
ahead
of
our
schedule.
C
Plus
one
the
thing
that
michael
said
earlier,
I
think
when
we
go
to
the
itf
community,
they
will
want
to
hear
about
the
plan
at
some
level
of
detail.
So
I
think
we
should
probably
have
something
written
up.
C
B
So
I
got
a
a
quick
agenda
bash
for
murray
because
he's
going
to
have
to
drop
off
the
call,
let's
fast
forward
to
the
dmarc
rewriting
section
of
the
agenda.
B
So,
as
you
noted
in
the
the
flow
that
the
flow
diagram
at
the
very
top
of
of
the
graph
there's
a
spot
where,
when
email
comes
in
the
processing
chain,
looks
to
see
if
the
mail
was
to
at
dmarc.ietf.org
and
if
so,
if
the
username
part
was
formatted
in
a
way
that
was
expected
and
if
it
is,
it
is
unwritten
to
go
back
to
you
know
whatever
was
in
that
username
part.
So
as
in
the
example
here,
it
would
be
user,
dom
dot,
aim
and
mail
would
just
get
sent
directly
to
that
user.
B
There
is
some
cultural
memory
that
this
rewriting
would
be
dependent
upon
having
mail
go
through
in
the
other
direction.
First
and
the
database
was
kept
about,
addresses
that
were
safe
to
you
know
were
okay
to
unwrap
I've
looked
at
the
code.
There
is
no
such
thing.
It
will
unwrap
any
address.
That's
sent
to
you
at
dmarc.ietf.org
as
long
as
the
sender
that
is
sending
to
that
address,
passes
through
post
confirm.
B
So
it
does
provide
a
mechanism
where
somebody
might
choose
to
attempt
to
circumvent
a
block
by
having
their
mail
wrapped
through
dmarc
or
whatever,
but
it
would
be
someone
that
is
on
our
post
confirm
approved
list
already
and
if
changes
you
know,
if,
if
complaints
come
in,
if
we
start
to
see
abuse,
it's
easy
to
remove
the
abuser
from
the
post
confirm
allow
list.
B
Now,
as
we're
processing
mailing
we're
about
to
send
it
to
someone,
and
it
turns
out
that
the
dmarc
policies
for
sending
john
I'll
get
my
twos
and
froms
wrong
here,
yeah
the.
I
I
Yeah,
if
the
sender
has
a
restrictive,
dmarc
policy,
then
we
rewrite
it.
So
the
from
its
of
the
from
address
is
in
our
domain
and
it
passes
dmarc,
since
we
have
a
since
it'll
it'll
pass
our
dmarc
policy,
and
at
that
point
we
also
set
up
some
sort
of
temporary
forward.
So
if
it's
so,
if
somebody
even
replies
to
that
rewritten
address
it
will
work,
it's
not.
B
I
Oh,
that
turns
us
into
an
open
relay.
We
should
fix
that.
I
That's
basically
the
that's
basically
the
the
forward.
I
was
mentioning
just
implemented
in
sort
of
a
peculiar
way.
I
Yeah
I
mean
this
is
this
is
adapted
from
something
I
did
on
my
on
my
system,
where
I
actually
do
explicit
forwards,
but
the
overall
effect
is
the
same,
and
if
we're
going
to
deal
you
know
until
given
that
dmarc
is
not
going
away
and
the
arc
is
arc
which
is
supposed
to
fix,
this
problem
is
being
adopted
very
slowly.
I
don't
think
we
have
any
alternative
but
to
keep
doing
this.
I
G
I
mean
you,
you
can
try
being
part
of
the
experi.
It
was
published
as
an
experimental
draft
and
john's.
I
think,
john's
understanding
of
its
deployment
matches
mine
that
a
few
large
operators
have
started
doing
it,
but
not
enough
to
say
that
it's
to
be
able
to
conclude
whether
the
experiment
was
a
success
or
not.
G
This
is
something
I've
asked
the
working
group
since
you
published
it
as
experimental
who's,
running
the
experiment,
what
who's
collecting
data
about
its
efficacy
and
whether
it's
should
be
folded
into
the
standards
track
version
of
dmarc
or
not
this.
I
saw
your
comment
in
the
document
about
the
software.
I
haven't
touched
it
in
a
few
years
because,
as
far
as
I
can
tell
it's
it's
in
a
usable
state,
there
are
some
pull
requests
that
I
have
to
go.
G
Look
at
if
there's
something
broken
in
it
or
something
some
feature,
you
need
that
you
don't
have
I'm
happy
to
go
work
on
it
for
the
ietf,
but
there's
I
mean
the
only
pers,
there's
four
or
five
filters,
and
there's
only
me
really
working
on
them.
So
I
my
time
is
spread
among
a
bunch
of
other
things,
but
if
something
urgent
needs
attention,
I'm
happy
to
work
on
it
with
you,
but
yeah.
I
I
don't.
I
don't
have
anything
dispositive
to
say
about
arc.
G
Yet
it's
still
early
in
its
life
and
I've
asked
the
working
group
for
status
on
it
and
I
haven't
really
received
any
likes.
A
few
people
have
said:
yes,
we're
trying
it.
I
haven't
heard.
Yes,
it's
effective.
Yes,
we
should
do
this.
You
know,
that's
that's
it's
still
too
early
for
those
conclusions.
Apparently.
B
So
I
think
my
recommendation
would
be
that
we
wait
for
the
community
to
say
in
in
large
numbers
hey.
Why
are
you
still
doing
this
dmart
thing?
Why
aren't
you
just
using
art
and
at
that
point,
start
to
look
at
whether.
E
I
G
That
would
be
great
this
because
if,
if
they
did,
I
can
ask
that
team.
Are
they
doing
it?
Because
if
we,
if
they
were
checking
arc
inbound
and
they
were
able
to
select
the
size
of
the
data
that
they
see,
they
could
definitely
say?
Yes,
this
is
effective.
No,
it's
not,
but
I
don't
like
john's
observations,
suggest.
That's
those
data
don't
exist
yet.
G
I
B
So,
let's
jump
back
to
the
idn
eai
support
and
john,
I
think
you
have
the
the
most
informed
opinion
about
what
we
can
do
at
the
moment.
So.
I
Yeah
eai
is
a
swamp,
and
the
problem
with
eai
is
that
it
is
only
one
way
if,
if
you
have
an
ascii
address,
you
can
write
to
an
eai
user,
but
eai
users
cannot
write
to
non-eai
users
where
an
eai
user
is
someone
on
a
system
that
supports
eai,
even
if
they
do,
even
if
they
don't
have
an
eai
address.
I
So
in
practice
this
means
that
among
male
systems
we
see
like
gmail
and
hotmail
have
full
eai
support.
Even
though
they're
all
ascii
addresses
anybody
running
x
and
repost
fix
it's
an
option
that
they
can
turn
on.
Although
most
people
don't
and-
and
then
there's
a
bunch
of
like
then
there's,
you
know
a
variety
of
systems
in
in
asia.
In
reality,
the
people
who
use
eai
are
some
non-english
speaking
communities,
there's
one
in
thailand
and
there's
one
in
in
india.
I
So
I
think
you
know
we
should
we
should
do
it
speak
to
you
to
eat
our
own
dog
food,
but
I
don't
think
we
should.
I
think
we
should
not
overestimate
how
useful
it
will
be
or
underestimate
how
hard
it
will
be,
because
because
of
this
of
this
one
way
issue
I
mean
so.
My
advice
is
since
since
repo
transfer
post
fix
underneath
we
can
turn
it
on
and
see
how
it
works
and
I
having
tested
it.
I
I
can
tell
you
the
postfix
support
works,
great
okay,
but
then,
once
you
once
it
comes
out
of
post
fix,
it
has
to
go
into
some
sort
of
imap
server
and
there
are,
as
far
as
I
can
tell,
there
are
no
imap
servers
that
actually
have
eai
support.
Although
it
turns
out
the
sort
of
legacy
post
fix
works,
pretty
much
okay,
usually
in
some
cases,
if
you're
running
dovecot,
you
know,
and
then
the
mail
program
you're
using
depending
on
what
it
is
like
thunderbird
supports,
support.
I
Cai
outlook
supports
eai,
pretty
much
anything
else
that
runs
on
your
desktop
other
than
oh,
so
apple
mail
supports
it
on
your
phone,
but
not
on
your
mac,
I
mean
so.
I
think
you
know
it's
worth
turning
it
on
on
our
for
our
own
mail
to
do
it
to
do
some
experiments
and
see
how
see
how
it
works
as
far
as
ending
adding
eai
to
mailman.
I
That
is
a
huge
project,
because
you
have
to
come
up
with
some
way
to
do
aliasing
to
send
to
send
mail
to
non-eai
recipients,
which
means
both
you
have
to
figure
out.
Who
is
a
90
ai
recipient?
Are
you
going
to
maintain
the
aliases?
Are
you
going
to
ask
people
if
they
have
aliases
at
home?
You
know
which
they
which
they
might
or
might
not
do,
and
I
don't
think
it's
something
we
should
do
on
our
own
and
just
fyi.
I
The
people
that
I
can
who
run
mailing
lists
also
think
they're
going
to
add
ea
to
mailman,
I'm
pretty
sure
they
have
no
idea
how
hard
it
is,
and
I
noticed
it's
also
somewhere
on
the
mailman
roadmap.
So
I
think
if
we
were
going
to
do
this,
I
think
we
would
my
advice
would
be
to
say
it
is
a
good
idea
and
then
talk
to
the
mailman
developers
and
see
if
we
can
come
up
with
some
people
who
understand
eai
and
actually
have
the
skills
to
do
it,
because
it's
a
big
project.
I
And
also
just
practically
speaking,
anybody
who's
writing
is
when
I
get
to
an
ietf.
Mailing
list
will
have
a
non-eai
address,
because
you
can't
get
your
work
done.
Otherwise
you
know
they
may
have
both.
I
have
both,
but
you
know,
and
people
in
china
and
india
have
both,
but
it's
it's
not
going
to
be
a
it's
not
going
to
be
a
blocker
for
anybody's
work
for
quite
a
while.
B
Anybody
else
have
anything
you
want
to
add.
I
think
the
takeaway
from
this
is
that
we
try
to
work
personal
relationships
and
or
just
the
submit
a
feature
request
interface
with
the
mailman
developers
and
wait
for
mailman
to
come,
say
that
we
have
the
mailman
part
of
this
figured
out.
Yeah.
I
But
I
think
it's
also
worth
doing
just
you
mean
for
people
who
actually
like
people
like
jay
who
actually
have
you
know.
I
don't
want
to
turn
j
into
my
only
guinea
pig,
but
for
people
who
have
actually
or
or
lars
anybody,
wasn't
anybody
who
has
an
ietf
mailing
address
if
we
just
turn
it
on
the
underlying
postfix
and
see
how
it
works.
I
think
that
would
be
interesting.
B
I
I
Post
fix
code
works
fine,
but
the
thing
is
nhs
post
fix,
there's
there's
you
know,
there's
imap
servers
and
pop
servers
and
muas
and
filters
and
spam.
Oh
yeah
and
oh,
it's
famous
acid.
It
works.
I
tested
spam
assassin,
no
problem
there,
but
I
think
you're
going
to
find
people's
people's
people's
mail,
clients
and
people
and
just
other
software
that
doesn't
hasn't
occurred
to
us.
Oh,
like
the
mail
archives,
you
know
like
they
don't
handle
ei
either
or
maybe
they
do.
I
don't
know.
B
All
right,
so
I
think
we
know
what
we're
going
to
do
with
that,
that
those
two
things
being
putting
pressure
on
the
mailman
three
development
community
and
we'll
talk
with
glenn
about
enabling
the
eai
and
our
basic
post
fix,
and
we've
got
some
folks
with
at
staff.ietf.org
addresses
that
can
nudge
it
and
see
what
really
breaks.
B
So
we've
got
a
next
section
talking
about
how
we're
dealing
with
spam,
how
we
are
keeping
our
lists
from
being.
B
As
rendered
as
useless
as
most
mailing
lists
in
the
world
have
been
just
by
the
the
rising
spam
tide,
you
can
read
through
the
mechanics
that
we
have
virus
scanning
and
the
use
of
spam
assassin
are
pretty
normal,
as
mentioned
earlier,
spanish,
the
where
we
have
spam
assassin
early
in
the
chain,
it's
fairly
permissive,
just
labeling
the
message
and
allowing
individual
list
managers
to
set
their
rules
for
dropping
spam
down
inside
mailman's
interface.
On
working
with
what
the
the
the
spam
score
looks
like
we've
mentioned
post
confirm
several
times.
B
This
section
provides
a
little
bit
more
of
a
concise
description
of
what
post
confirm.
Does
it's
turned
out
to
be
in
the
past
very
effective
at
deterring
people
that
just
had
simple
spam
bots
that
would
connect
to
smtp
dump
in
some
bits
and
run
away
because
it
required
a
round
trip
to
the
sender
address
before
any
mail
would
actually
go
through
the
list.
B
B
Down
inside
mailman,
the
mechanisms
that
we've
got
right
now
for
dealing
with
spam
are
mostly
working
with
that
spam
assassin
score
and
using
it
to
decide
what
to
do
with
the
message
before
it
lands
on
the
list.
B
We
do
have
the
global
white
list.
This
is
something
that's
recreated
every
day
and,
as
I
know,
we
should
do
the
work
when
it
comes
time
to
rewrite
it
in
python
3
to
call
it
an
allow
list
instead
of
a
white
list.
B
Lists
can
be
configured
to
allow
any
sender
that
is
on
that
list
to
pass
so
john
you'd
mention
that
you
subscribed
to
lists
so
that
you
can
send
to
them.
You
really
shouldn't
have
to
do
that
at
this
point
at
least
not
more
than
once.
B
You
subscribe
to
a
list
and
you
should
be
able
to
get
to
all
the
rest
of
them.
The
secretariat
has
periodically
gone
through
the
configuration
and
either
added
the
global
white
list
variable
to
the
except
these
non-members
or
poked
the
list
managers
to
suggest
that
they
do,
and
I
believe
that
almost
all
of
our
lists
are
configured
to
use
the
global
white
list.
At
this
point,.
B
This
use
pattern
of
reading
with
imap
is
going
to
make
some
of
the
statistics
that
jay
occasionally
asked
to
collect
a
little
bit
more
difficult.
Alexi's
done
a
little
bit
of
work.
We
haven't
really
tested
the
efficacy
of
it
yet
to
give
us
an
approximation
to
what
lists
people
are
subscribed
to
through
imap.
B
But
that
I
get,
I
guess
it's
worth
calling
out
at
this
point,
that
our
ability
to
see
how
many
people
have
subscribed
to
what,
given
that
we
have
a
lot
of
people
that
are
working
that
are
consuming
these
lists.
Using
imap
is
going
to
go
downwards
and
that
will
have
an
effect
on
the
utility
also
of
the
improved
list
management
interface
that
we
were
discussing
earlier
in
the
call.
B
So
I
can
perfect
I
can.
I
can
envision
for
myself
that,
outside
of
private
lists
that
I
I
I
could
see
a
future
where
I
am
not
using
mailman
to
subscribe
to
anything.
I
just
make
sure
that
I'm
in
the
global
white
list
somewhere
and
I
consume
everything
through
imap.
B
B
Now.
I've
noticed
just
by
inspecting,
what's
happening,
that
this
list
is
got
some
things
in
it
and
we
have
some
coming
mostly
from
data
tracker
accounts
that
have
active
email
addresses
that
are
starting
to
grow
rapidly
and
they're
growing,
with
addresses
that
don't
look
like
they're
really
legitimate
addresses.
B
You
know
it's
made
up
string
of
letters
at
gmail.com
kind
of
addresses
and
a
lot
of
them
and
a
bit
of
an
up
curve
recently,
as
I
note
right
now,
that
lists
nearly
got
a
hundred
thousand
addresses
in
it
and
the
the
active
participant
numbers
that
I
think
that
we've
got
in
the
actual
subscriptions.
B
I
don't
remember
the
exact
number,
but
I
think
that
it's
in
the
some
number
of
tens
of
thousands,
where
that
number
is
less
than
five.
So.
B
The
tools
that
we
have
right
now
are
setting
the
moderated
bit
on
the
list
so
that
their
posts
go
into
moderation
and
list
manager
deals
with
letting
those
posts
through
or
not,
and
the
secretariat
can
go
through
and
globally
set
that
bit
for
every
list
that
someone
subscribed
to.
If
the,
if
there's
a
need
for
moderation
across
the
the
entire
list.
B
There's
if
things
are
really
out
of
control,
somebody
can
put
the
entire
list
into
moderation
and
and
release
things
basically
from
a
manual
allow
list.
So
there
are
imagined
attacks.
We
have
not
had
this
happen,
but
there
are
imagined
attacks
where
somebody
that
is
persistent
could
just
generate
new
addresses
to
send
from
at
a
rate
faster
than
we
could
throw
into
moderation.
C
Robert,
do
we
have
data
on
how
much
these
existing
tools
have
even
been
used,
because
so
my
god
feeling
is
it's
not
a
problem
that
the
moderation
tools
that
we
have
aren't
working
or
aren't
enough?
I
think
it's
that
the
chairs
feel
uncomfortable
actually
using
them,
and
maybe
maybe
we
need
some
tools
that
sort
of
can
be
used
in
you
know
less.
You
know.
C
Pronounced
cases
of
harassment
or
what
are
you
going
to
call
her
abuse?
But
I
don't
know
right.
It
seems
like
that
the
tools
seem
pretty
comprehensive
to
me.
At
least
I
don't,
but
I
don't
think
I
don't
know
if
they're
actually
that
useful
to
us,
because
typically
it's
not
like
somebody's
spamming,
a
list
from
random
addresses
right.
It's
typically
a
participant
that
we
know
is
is
getting
out
of
line
again,
and
we
got
to
do
something
about
that
right
and
I
don't
know
if
the
tools
really
help
you.
A
If
I
may,
we
we,
at
the
secretariat
level,
take
action.
Almost
never
I
mean
there's.
I
can't
remember
the
last
time
I
actually
had
to
block
an
address
on
the
server
I
mean
I
block
addresses
for
egregious
spam
that
the
list
never
see
but
in
terms
of
being
asked
by
the
ietf.
A
I
can't
remember
who
was
president
then
so
it's
it's
almost
never
used
scenario.
A
We,
the
the
chairs,
do
as
lars
correctly
points
out,
have
the
ability
to
moderate
on
a
list,
and
I've
observed
that
he's
right
that
they
there
seems
to
be
a
lot
of
politics
rather
than
operational
issues
around
the
moderation
of
troubled
people.
B
There
have
been
discussions
in
the
past
about
things
that
are
not
necessarily
moderating
people
that
are
being
a
abusive
because
of
harassment
or
whatever,
but
we
have
conversations
that
are
going
a
little
bit
crazy
and
it
would
be
good
to
have
at
least
some
of
the
participants
in
those
conversations
slow
down.
B
C
C
This
is
something
that
mailman
3
would
already
allow
in
some
form
or
something
we
would
need
to
build.
We
don't,
I
don't
know.
B
It's
in
mailman
3.,
but
if,
instead
of
us
building
it,
it
would
suggest
that
it
would
be
something
that
we
would
ask
them
to
build.
C
Yeah,
I
think
it
could
be
useful
if
we
we
discussed
it
at
least
once
in
in
the
last
year
or
this
year,
whether
you
know
we
want
to
actually
sort
of
try
and
enforce
this.
But
it's
also
not
something
that
I
see
there's
only
a
few
lists.
I
think
where
this
might
be
useful
for
and
it's
yeah
we
can.
C
It's
never
really
yeah
and
sometimes
the
list
white
moderation
gives
you
a
bigger
hammer,
but
you
can
sort
of
if
you're,
if
the
chairs
are
sort
of
standing
by
right,
they
could
basically
do
something
similar
with
that
hammer
already
by
basically
just
pacing
out
how
they
approve
posts
onto
the
list.
C
I
don't
think
we
necessarily
urgently
need
new
tools
here.
I
think
it's
we're
sort
of
focusing
at
the
moment
on
sort
of
educating
the
chairs
a
bit
through
the
edger
team
and
emoji,
and
I
think
that
is
probably
where
we
should
spend
our
efforts.
I,
unless
somebody
has
like
a
really
great
idea
for
a
tool,
I
don't
have
one.
E
You're
right
lars
that
people
could
chairs
could
put
it
in
moderation
and
then
could
release
the
messages
slowly.
E
But
that's
that's
a
lot
of
work
for
the
chairs
and
it's
also,
and
particularly
there's
16
hours
when
they're,
not
working
with
the
messages
don't
go
through,
but
but
worse
is
that
the
allegation
then
becomes
that
they're.
E
E
That's
a
really
good
policy
that
I
would
like,
and
I
think
it
would
deal
with
most
of
the
runaway
tit
for
tat
threads,
except
that
people
are
seeing
themselves
each
other
right.
So
I
wish
that
I
wish
actually
that
didn't
happen,
but
it's
very
hard
for
us
to
do
anything
about
that.
D
I
also
agree,
I
think,
part
of
the
reason
why
it's
not
the
current
tools
are
successful
as
not
used.
Much
is
because
chairs
are
really
hesitant.
Speaking
is
one
to
do
it.
Oh
it's
going
to
stop
in
two
days.
Let's
just
let
it
go
that
down,
but
you
know
if
there
was
a
way
to
put
I'd,
say
a
list
wide
quota
in
terms
of
oh,
this
list
is
not
going
to
get
more
than
five
messages
an
hour,
no
matter
what
and
everybody
else
backs
up.
D
D
But
I
I
also
agree:
it's
not.
I
don't
think
it's
a
big
deal
yet,
but
I
think
a
large
reason
for
that
is
that
it's
a
big
scary
hammer
to
use
and
working
and
says
don't.
I
B
B
The
next
thing
that
we've
got
on
the
agenda
is
how
the
aliases
are
managed.
B
So
for
groups
we
go
for
groups
that
are
either
active
proposed
or
in
the
boss
state
and
for
of
type
working
group
research
group,
directorate
team
review
and
program
you'll
note
that
area
group
is
missing
from
that.
It's
something
that
we
should
add
and
we
should
review
and
see
if
other
group
types
make
sense
to
have
these
aliases
built.
B
B
Jay
recently
pointed
out
that
we
are
not
creating
aliases
for
rfcs
and
once
an
rfc
has
been
an
rsc
for
a
couple
of
years
and
there's
been
no
activity
on
the
draft.
The
draft
alias
for
that
roc
goes
away,
and
there
are
buttons
on
the
data
tracker
that
you
can
click
that
will
send
you
to
an
alias
that
doesn't
exist
and
mail
you
send
to
it
will
just
fail.
B
So
there
are
definitely
user
experience
issues
that
we
need
to
address
there
either
not
presenting
the
button
not
presenting
the
address
when
we
don't
have
it
or
perhaps
maintaining
a
list
of
well
that
for
sure
for
drafts
that
are
just
old,
very
stale
drafts
we
just
quit,
providing
that
here's
the
thing
that
would
send
to
the
authors
by
this
thing.
B
B
That
stays
around
indefinitely
and
we've
had
suggestions
for
different
kinds
of
document,
aliases
and
group
aliases.
The
ability
to
reach
the
authors
for
all
the
documents
for
given
groups
and
chairs
have
said
you
know.
I've
got
15
active
documents
and
I
want
to
send
a
mail
to
all
of
the
authors
of
those
active
documents
right
now,
because
it's
something
that's
pertinent
group
wide.
B
A
request
to
be
able
to
reach
all
of
the
chairs
for
a
a
given
area
director's
group,
so
barry
lieb
in
particular,
wanted
to
be
able
to
write
easily
to
the
chairs
of
the
group
that
he
was
directly
responsible
for,
without
bothering
the
chairs
of
the
groups
that
the
other
art
ad
at
the
time,
which
you
know
he
wanted
it.
We
could
build
it.
I
don't
know
how
much
use
it
would.
B
It
would
get
to
provide
that
kind
of
thing,
but
we
you
know
we
certainly
could
build
it,
and
the
secretariat
has
asked
us
for
an
alias
to
reach
liaison
managers
for
an
sdo
or
for
a
working
group.
C
B
F
For
sure,
so
it
sounds
as
if
we
need
perhaps
a
different
approach
to
this,
where
what
you're,
seeing
on
that
button
is
an
email
address
and
mailman
is
delivering
or
that
the
mail
processing
system
is
delivering
to
a
single
email
address.
But
we
have
a
separate
process
that
is
then
expanding
that
out
to
all
of
the
different
people.
So
we
perhaps,
for
example-
I
don't
know
if
this
is
possible-
could
have
a
separate
set
of
lists,
one-way
lists
or
something
like
that
that
are
automatically
created.
F
So
you
know
it's
a
just,
for
example
expander.iatf.org
or
something
like
that,
and
so
that
we
don't
have
to
worry
continually
about
the
number
of
aliases
and
that
are
being
created
in
the
table
there.
We
have
a
separate
mechanism
for
handling
that
expansion,
so.
B
Bringing
up
the
idea
of
using
a
separate
mechanism
when
maybe
this
is
something
that
you
could
help
with
what,
if,
instead
of
running
things
through
the
table
to
deliver
to
these
addresses,
we
have
a
melter.
That
knows
how
to
ask
the
data
tracker
for
what
the
addresses
are
and
the
melter
actually
builds.
The
you
know
replaces
the
recipients
in
real.
F
A
I
am
not
sure
that
I
should
still
answer
after
jay's
answer,
but
I
will
say
that
it
seems
to
me
that
a
militar
lookup
is
transactional
and
would
introduce
some
type
of
delay,
whether
detectable
or
not.
I
don't
know,
but
I
do
find
that
postfix
has
been
very
efficient
at
dealing
with
its
you
know,
with
its
tables
they've,
just
when
we,
when
we
do
our
next
upgrade,
there's
a
new
database
format
lmdb
that
they've
introduced,
and
I
I
don't.
A
I
would
be
comfortable
at
least
trying
to
let
the
data
tracker
build
out
a
much
larger,
alias
table
and
let
postfix
just
use
that
and
see
how
that
how
that
runs.
I
think
that
operationally
that
would
probably
be
smoother,
but
that's
just
my
opinion.
I
A
I
F
B
All
right:
well,
I
will
take
an
action
early
next
year,
depending
on,
unless
we
get
an
answer
back
from
the
john's
investigation
that
says
whoa
whoa
that
we
will.
We
will
just
try
this
and
see
see
what
kind
of
trouble
we
run
into
all
right.
So
one
of
the
big
problems
that
we
haven't
solved
yet-
and
it
would
be
very
good
to
solve-
is
none
of
these
aliases
get
the
dmarc
treatment
that
messages
going
through
lists
get.
B
So
if
this
list
expands
to
you
know,
someone
is
sending
to
this
list
and
the
list
gets
expanded.
The
and
they've
got
a
dmarc
policy
that
is
of
the
nature
that
john
described
their
address
doesn't
get
rewritten
on
the
way
out,
and
this
is
causing
failure.
You
know,
as
mail
is
not
being
delivered
to
the
intended
recipients
because
of
the
sender's
dmarc
policy.
A
Yep,
I
my
limited
knowledge
tells
me
that
the
aliases
themselves
are
being
serviced
by
post,
confirmed,
at
least
on
the
inbound
side,
and
the
dmarc
rewriting
is
being
served
by
post
confirm
on
the
outbound
side.
A
B
I
don't
know
if
it
would
require
the
insides
of
the
aliases
to
change.
We
don't
want
the
outside
of
the
alias
to
change
for
sure
right,
but
the
alias
that
somebody
would
actually
type
in
their
in
their
to
header
would
need
to
stay
the
same,
but
the
the
v
table,
aliases
that
we
generate
could
possibly
change
and
those
could
cause
things
to
get
fed
to
different
processing
chains
inside
postfix.
B
A
Sorry,
I'm
sorry,
I
was
just
agreeing
with
you.
I
we
could
probably
experiment
with
that
pretty
quickly
offline
robert
you
and
I
at
some
point,
okay
and
solve
that
I
feel
like
that
should
not
be
difficult,
given
that
we
already
have
post
confirm,
which
does
a
great
job
of
rewriting
and
doing
all
that
so.
B
Anybody
have
any
any
questions
about
how
we're
building
them
and
other
than
the
experiment
that
we,
the
plan
that
we
are,
are
working
towards
now
to
just
generate
them,
for
all,
dress
and
and
here
and
I'm
assuming
all
groups.
Then
at
this
point,
since
there
should
be
symmetry.
B
Anybody
has
anybody
experienced
any
problem
with
the
frequency
of
generation.
The
fact
that
you
know
somebody
changes.
These
chairs
changes
the
chairs
for
a
group.
These
aliases
don't
change
until
the
top
of
the
next
hour,
so
potentially
an
hour
at
most
an
hour
lag.
C
I
had
that
when
I
changed
gen
dispatch
chairs,
but
so
it's
fine
if
the
data
tracker
would
remind
one
that
you
know
when
you
read
when
you
change
chairs
that
would
have
put
an
alert
up.
That
says:
don't
expect
this
alias
to
work
yet
it
completely
slipped
my
mind
that
there
was
a
little
light.
B
All
right:
well,
we
are
an
hour
and
a
half
into
our
three
hours.
This
gets
us
through
the
bulk
of
what
I
wanted
to
talk
about,
given
that
we've
got
extra
time.
If
there
is
any
topic
that
people
want
to
go
back
and
dive
into
a
little
bit
more
deeply,
this
would
be
a
a
good
we
could
do
so.
B
F
B
B
So
far
we
haven't
gotten
past
well,
there's
something
that
we
should
look
at
at
some
time
in
the
future,
so
it
would
be
possible
to
make
some
sort
of
assessment
over
the
viability
of
those
lists
to
just
run
through
again
and
see
if
the
address
has
any
real
activity
associated
with
it
other
than
just
the
creation
of
a
data
tracker
account,
you
know
if
it's,
if
it's
only
there,
because
the
data
tracker
accounts
been
created,
and
especially
if
that
account
is
over,
you
know
a
few
days
old
and
there's
been
no
drafts,
no
email,
traffic,
etc
that
we
could
pull
a
large
number
of
those
out
based
on
that
metric.
B
Is
it
worth
doing?
You
know
we're
we
we're
accruing
this
big,
this
big
thing
and
it
allows
you
know
I
could
envision
in
a
in
a
world
where
we
had
somebody
that
really
wanted
to
be
unfriendly
to
us
that
they
spend
quite
some
time
slowly.
Dribbling
dribbling
up
a
bunch
of
these
accounts
and
getting
them
on
the
white
list
and
then
launching
a
a
large
spam
attack
with
a
lot
of
these
addresses.
F
Well,
it
could
even
be
happening
accidentally,
so
are
we
making
it
too
easy
for
people
to
confirm
their
address
inside?
The
post-confirmed
response
is
that
I
mean
is
this?
Is
a
bot
just
auto
clicking
the
response
email
there?
For
example,
there
are
the
response,
url,
the
confirmation,
url,
sorry.
B
I
don't
think
they're
coming
in
through
explicit
post
confirm
actions
the
the
we're
getting.
B
B
There
are,
there
is
one
path
through
creating
a
data
tracker
account
where
you
don't
have
to
confirm
your
address,
but
I
don't
think
that
most
of
them
are
coming
through.
That
I
mean
I
guess
what
I'm
hearing
from
this
is
that
we
need
to
go,
get
some
metrics
on
what,
where,
where
growth
is
coming
from
and
make
sure
that
we're
looking
at
the
right
growth
path.
Yeah-
and
I
I
mean.
B
B
We
do
have
the
competing
pressures
to
balance
because
we're
using
data
tracker
accounts,
as
our
oidc
is
basically
our
ways
to
get
into
meetings.
There's
pressure
to
keep
the
bar
to
getting
a
data
tracker
account
low,
but
I
know
when
real
people
still
have
walked
up
to
get
data
tracker
accounts.
They
still
run
into
the
email
message.
B
J
Yes,
I
just
wanted
to
bring
up
bounce
processing
for
a
second-
that's
probably
a
non-issue
for
most
people
here,
because
they
are
not
trying
to
hurt
cats.
They
are
not
working
group
chairs.
J
When
you
have
a
large
working
group,
then
you,
you
have
a
large
number
of
people
whose
companies,
mail
admins,
do
weird
things
to
their
mail
addresses.
So
you
suddenly
have
these
these
occurrences
of
people
falling
off
their
mailing
list
and-
and
you
think
you
have
consensus
on
something,
but
the
whole
company
hasn't
gotten
the
consensus
call
and
things
like
that.
So
chairs
tend
to
switch
on
the
the
noisy
parts
of
the
bounce
processing.
That
mailman
provides,
which
is
great,
because
many
of
those
situations
are
then
becoming
visible
very
quickly.
J
So
unless
the
the
mail
admin
does
something
that
completely
swallows
those
mails
you
you
find
out
that
you
have
a
problem,
but
that
of
course
opens
the
next
problem
that
the
the
bounce
processing
is
a
magnet
for
spam,
and
that's
the
only
spam
I
ever
get
on
on
on
itf
on
through
idf
accounts,
and
fortunately
I'm
not
a
chair
anymore.
J
So
I
don't
have
that
problem
anymore,
but
I
can't
imagine
that
other
chairs
still
still
have
it
and
I'm
wondering
whether
anything
can
be
done
with
making
bonds
processing
a
bit
more
useful
for
for
chairs,
trying
to
hurt
cats.
B
No,
I
I
understand
the
problem
that
you're
describing,
but
I'm
not
immediately
imagining
what
improvements
to
the
bounce
processing
would
look
like.
J
Like
go
ahead,
carson,
probably
just
setting
up
a
higher
spam
threshold
or
a
lower
spin
threshold.
However,
you
counted
before
actually
forwarding
bounces
to
a
chair
that
would
probably
solve
eighty
percent
of
the
problem.
J
A
No,
I
was
just
going
to
say
that
the
problem
which
carson
reports
is
a
problem
that
comes
to
the
secretariat
periodically
two
three
times
a
year
and
unfortunately,
the
suggestion
that
carsten
has
offered,
which
is
a
good
one,
is
not
one
that
we
have
a
knob
to
turn
for.
A
I
can
adjust
the
spam
processing
server
wide
and
I
can
adjust
the
span
processing
per
list,
but
those
bounce
addresses
are
by
design
by
mailman
design
supposed
to
receive
everything
and-
and
they
do
anything
that
so
what
I
was
going
to
say
was
simply
that
what
it
would
look
like
fixing
this
problem
and
what
it
would
look
like
is
a
code
change
to
mailman.
A
Each
mailman
list
has
a
number
of
aliases
associated
with
it.
There's
the
list
name
at
ietf.org
to
which
people
send
mail,
but
there's
also
list
name
hyphen
bounces
at
ietf.org,
which
receives
it's
it's
the
envelope
from
for
outgoing
mail
and
it
receives
bounce
reports
and
mailman
failures,
delivery,
failures
and
so
forth
from
the
outside
world.
If
we
were
to
block
those
addresses,
then
we
wouldn't
know
about
bounces.
We
would
keep
sending
blindly
forever.
A
B
B
I
think
that
that
could
be
tuned
to
the
point
where
the
false
positives
would
not
be
harmful.
It's
just
something
we
should
look
out
for
the
ability
to
do
once
we
get
to
mount
three.
B
Because
I
think
that'd
be
the
easiest
thing
to
instrument
for
rate
of
bounce
messages
from
domains
and
set
alarms
when
we
start
to
see
large
number
of
balances
from
a
a
domain
to
to
recognize
that
something
is
happening
with
an
erickson
or
a
cisco
or
a
whatever.
A
If
I'm
understanding
this
correctly,
I
think
we
have
two
issues.
One
is
we
don't
want
companies
to
get
thrown
off
mailing
lists
for
whatever
reason,
and
that
issue
we
actually
do
have
knobs
the
mailman
list.
Owners
can
adjust
the
threshold
scores.
I
mean
we
could
do
a
global
adjust
too.
But
the
point
is
that
the
list
owners
can
adjust
that
behavior.
If
there's
a
list
that
has
a
problem,
you
know
with
erickson.
For
example,
that's
the
one
that
comes
to
mind
that
I've
dealt
with
recently.
A
That
is
getting
thrown
off
of
lists
because
of
bounces,
for
whatever
reason
that
the
list
owners
can
adjust
the
bounce
thresholds
of
accounts
and
even
disable
automatic
balance
removal
for
their
list
if
they
want
to
so
that
they
can
protect
their
participants.
That
is
a
choice
I
don't
know
off
the
top
of
my
head.
What
the
ramifications
of
that
would
be,
but
you
know
that
seems
to
be
the
best
choice.
A
The
other
issue
is
list
owners
getting
spammed
through
the
dash
bounces
vector,
and
for
that
one
we
I
think
we
do
have
to
go
to
the
mailman
three
people
once
we've
switched
over
to.
I
mean
it's
pointless
to
go
to
the
mailman
two
people
now
right,
so
I
think
we
should
switch
over
and
then
address
it
with
mailman
three,
but
I
think
that's
gonna
be
well
so.
B
I
think
you
messed
the
point
on
what
I
was
trying
to
suggest
when
you
dealt
with
when
you,
when
you
or
you
have
dealt
with
or
you've
helped
chairs
deal
with
an
organization
that
suddenly
organization-wide
is
having
problems
receiving
list
mail.
I
has
part
of
that
been
interacting
with
the
iet
department
at
that
company.
A
Found
a
way
to
interact
with
the
I.t
department
at
a
company.
I
get
complaints
from
participants
and
as
a
matter
of
procedure
for
me,
I
always
send
the
log
snippets
to
the
participant,
and
I
say
I
am
happy
to
work
with
your
I.t
people
and
your
male
administrators,
and
invariably,
I
I
can
think
of
no
exception
to
this.
The
answer
I
get
back
is
in
essence.
A
They
are
not
available
to
talk
to
you
or
to
me,
meaning
they
are
not
available
to
talk
to
glenn
or
to
the
reporting
participant,
but
I've
asked
them
to
fix
it
and
they'll
look
at
it
and
then
I
never
hear
anything
else
after
that.
I've
never
I'm
happy
to
work
with
it
people,
but
they
have
never
made
themselves
available.
C
A
And
so
yeah
and,
interestingly,
we've
never
we've
never
had
a
problem
with
exchange
people
like
I
know
a
lot
of
our
participants
use
microsoft.
I
for
my
personal
email
use,
microsoft,
e5
and
I
love
it,
and
we've
never
had
a
problem
with
exchange
or,
ironically
with
google.
It's
always
the
the
companies
that
run
in-house
anyways
rambling.
A
B
B
F
And
I'm
underwhelmed
by
the
feedback
on
the
the
improved
user
list
management
experience
interface,
that
I've
produced
there
that
bridge
I'm
going
to
presume
from
that.
That's
excellent
and
we
should
just
go
ahead
with
that,
which
is
highly
unlikely.
J
Yeah,
I
think
we
we
have
this
in
a
number
of
other
places
as
well,
where
people
only
actually
look
at
these
things
once
they
are
forced
to
use
them
so
yeah.
That
may
be
an
unpleasant
observation,
but
that
may
shape
the
way
we
handle
this.
B
So
our
future
workshops,
one
of
the
big
ones,
is
to
explore
more
deeply.
This
notion
that
we
have
been
referring
to
as
doc
dot
ietf.org.
B
B
How
much
work
do
we
need
to
put
on
top
of
these
things
for
improving
the
you
know,
seos
and
just
community
expectations
right,
so
that
one
would
probably
also
talk
about
when
we
are
what
our
long-lived
urls
should
look
like.
Building
things
that
communities
that
aren't
us
can
expect
to
use
and
expect
to.
You
know,
work
over
long
periods
of
time
as
we're
finishing
the
tools.itf.org
transition.
I'm
sure
many
of
you
saw
that
we
had
some
interaction
with
the
the.
B
Stack
exchange
right,
the
administrative
side
of
stack
exchange
for
the
people
that
were
running
stack,
exchange
actually
decided
to
go
through
and
fix
up
links
that
had
been
pointing
to
tools.iatf.org
and
all
of
their
messages
like
historically
to
point
to
somewhere.
That
would
still
work
going
forward
so
being
able
to
provide
organizations
like
that
lynx
is
part
of
the
discussion
we
would
have
at
that
workshop.
B
I'm
expecting
to
have
one
or
more
workshops
that
are
focusing
on
big
data
factor:
data,
tracker,
refactoring,.
B
Choices
right,
so
one
of
the
ones
that's
coming
up
is
whether
or
not
we
should
track
affiliation
differently
and
we're
tracking
affiliation,
where
we're
keeping
affiliation
in
the
data
tracker
and
and
how
we
try
to
keep
history
and
how
much
we
try
to
keep
it
fresh,
whether
or
not,
and
I
think
we
will-
but
maybe
it's.
B
I
will
phrase
this
more
as
what
are
the
consequences
of
modeling
rfcs
as
a
thing
different
than
a
draft
right
now
we
model
an
rfc
as
just
another
pseudo
version
of
a
draft
and
that
causes
all
kinds
of
of
pain
and
and
unexpected
behavior
for
users.
B
When
they're
trying
to
look
at,
for
instance,
the
last
version
of
a
draft
before
it
became
an
rfc
code
does
do
it's
easy
for
code
to
do
the
wrong
thing
in
that
place,
and
I
think
that
we
are
and
should
make
a
a
refactor
where
we
have
a
different
document
object
that
is
of
type
rsc
for
the
rscs
and
the
appropriate
related
document
objects
that
would
associate
an
rfc
with
the
draft
that
the
rfc
came
from.
B
So
we
should
have
won
one
of
these
workshops
discussing.
B
Mail
archive,
in
particular,
looking
at
archiving
in
general,
possibly
archiving
things
making
archive
things
like
chat,
logs,
easier
to
find
and
possibly
found
through
the
same
interface
as
you
would
find
male
threads.
B
This
will
become
pertinent
when
we
get
zula
running
are
and
if
we
get
adoption
across
the
ietf
of
our
zoolope
instance.
We'll
also
want
to
have
some
insight.
The
ability
to
search
across
that
integrated
with
searches
across
email
lists
that
workshop
or
one
close
to
it,
should
also
focus
on
statistics
like
out
of
our
archives.
What
tooling
do
we
need
to
get
better
statistical
view
into
the
archive?
What
are
the
number
of
posts
to
a
list
per
day?
B
B
I
will
note
that
we
have
on
the
beginnings
of
an
api
already
implemented
at
mail
archive
that
will
answer
that
number
of
posts
to
a
list
over
a
given
period
of
time.
The
I'm
expecting
will
finish
the
documentation
of
pretty
soon
and
can
get
it
out
into
people's
hands
and
john.
It
may
be
something
that
you
might
consider
using
for
generating
the
message.
B
The
message
that
you're
sending
out
you
might
be
able
to
provide
even
richer
historical
detail
than
what
you're
currently
doing.
B
B
So
that's
what
we
had
captured
jay
in
the
oh,
I
guess
talking
about
data
factor,
sorry,
data,
tracker
refactoring.
Some
of
the
other
things
we're
looking
at
are
changing
the
interfaces
for
building
an
interface
to
find
a
person
improving
the
interfaces
for
draft
submission,
making
things
like
the
agenda
interface,
more
mobile,
friendly
and
doing
things
like.
B
Starting
to
see
if
we
can
resolve
this
tension
around
having
multiple
forms
of
markdown
in
the
world,
we
have
these
notions
of
live
documents
or
semi-live
documents.
Authors
copies
of
documents.
Do
we
want
the
data
tracker
to
start
being
able
to
manage
these
things?
Do
we
want
data
tracker
to
manage,
live,
swarm,
editing
things
like
we're
doing
in
notes.itf.org
directly?
B
A
workshop
in
itself
just
talking
about
those
things,
and
we
might
have
a
workshop,
focusing
on
improving
how
we
have
how
we're
modeling
roles
so
that
we
have
better
data-driven
access
control
for
somebody
who
becomes
a
foo
for
a
bar
working
group
or
foo
for
a
bar
type
group
on
the
the
kinds
of
access
they
get
into.
The
rest
of
our
systems
happens
a
little
bit
more
automatically
than
it
currently
does.
B
B
F
So
if
anyone
has
any
other
suggestions,
please
let
us
know-
I
mean
the
the
doc
one.
Obviously
there
have
been
a
number
of
conversations
with
different
people
about
carlson,
for
example,
we've
spoken
about
that
be
useful
to
do
that.
The
mail
archive
one
should
probably
be
seen
as
more
as
you
said,
robert
more
about
archiving
other
things
and
how
we
bring
those
together.
F
So,
for
example,
could
you
could
we
imagine
going
to
the
mail
archive
tool
being
able
to
put
in
a
date
range
and
seeing
email
addresses
instant
messages
that
came
up
as
well
and
any
other
actions
that
took
place
across
a
you
know
within
say
a
particular
working
group,
or
something
as
a
as
a
timeline,
so
that
you
can
understand
things?
You
know
better
in
that
type
of
way.
F
That
sort
of
thing
might
be
useful,
and
then
we
have
the
the
complications
around
github
and
the
use
of
issue
trackers
and
those
sorts
of
things
and
how
that
then
appears
within
our
archiving
and
searching
tools
as
a
potentially
overall
thing,
so
yeah
any
things
or
suggestions
that
people
have
here
would
be
useful
for
us
to
understand
that
we
can
build
into
these
workshops.
B
Thanks
everyone
again
for
spending
the
time.
I
hope
you
found
this
useful
if
you
did
not
find
this
useful,
please
let
me
know,
don't
don't
don't
just
be
silent
about
it
and
give
some
of
your
day
back
and
hopefully
see
you
all
online
again
very
soon.