►
From YouTube: IETF-TOOLS-20230516-1800
Description
TOOLS meeting session at IETF
2023/05/16 1800
https://datatracker.ietf.org/meeting//proceedings/
A
A
A
A
I
bet
we
spent
quite
a
bit
of
time
talking
about
how
we're
going
to
prepare
for
the
change
in
the
I.T
infrastructure,
how
we
were
handling
our
basic
deployments,
the
interaction
with
the
CDN,
how
we're
going
to
be
doing
releases
going
forward
automating
our
releases
into
a
you
know,
cloud,
an
infrastructure
that
assumes
that
we're
deploying
into
into
cloud
services,
and
then
you
can
read
the
rest
of
the
bullet
list.
If
you've
got
any
questions
about
the
things
that
we
went
into.
Let
me
know.
A
A
A
A
A
A
We
already
have
some
rfcs
that
have
linked
into
meeting
wikis
the
RSC
that
talks
about
running
hackathons,
in
particular,
has
many
links
into
these
older
meeting.
Wikis
we've
been
discussing
whether
or
not
to
continue
to
try
to
hold
on
to
these
things
as
not
Wiki
artifacts,
but
something
unrolled
that
could
be
attached
to
the
proceedings
or
if
we
should
just
move
to
not
preserving
them
ourselves,
but
using
archivets
Service
as
the
archive
and
I
think.
A
That
is
the
path
that
we
have
landed
on
as
an
operational
recommendation
so
that
we're
not
continuing
to
carry
separate
bits.
We
just
make
sure
that
the
wiki
themselves
got
wrapped
up
and
archive
it
at
some
point
and
we
just
drop
our
versions
of
the
wikis
at
the
point
that
the
meanings
are
finalized.
A
B
C
C
That
say
people's
project,
which
is
very
which
is
very,
very
secure
and
which
means
that
any
mail
that's
received
from
them
has
to
have
a
valid
dmarc
signature
or
come
from
an
authorized
IP
address,
and
it
turns
out
that
when
mail
goes
through
our
our
forwarder,
the
headers
get
reformatted
just
enough
to
break
the
dmarc
signatures,
and
since
it's
coming
from
our
from
our
forward
or
rather
than
from
the
original
Source,
it
doesn't
come
from
the
authorized
source.
So
the
dmarc
fails
and
the
mail
disappears.
C
So
it
seemed
kind
of
like
why
was
it
reorganizing,
the
the?
Why
was
it
reformatting?
The
header,
so
I
took
a
look
in
the
code
which
is
ancient
and
messy,
and
the
problem
is
essentially
that
it
uses
a
mind,
parsing
library,
to
parse
the
entire
message,
and
then
it
de-parses
it
on
the
way
out
and
the
mime
Library
thoughtfully,
reformats
the
headers
to
make
them
look.
C
Nice,
and
so
it
would
not
be
super
hard
to
kind
of
patch
around
that,
and
in
fact
it
does
that
in
a
couple
of
places,
there's
one
place
where
it
does
demarc
rewrites
and
it
actually
specifically
checks
to
say.
Oh
if
I
didn't
do
any
rewrite,
just
use
the
original
text
and
I
can
pretty
much
go
and
it
wouldn't
be
super
hard
to
go
back
and
Patch
around
the
places
where
it
does
it
in
other
places
would
be
not
totally
trivial,
because
sometimes
it
adds
some
headers.
C
You
need
to
part
you
need
to
splice
it
into
the
string
but
I'm
reluctant
to
do
it
just
because
the
module
is
such
a
piece
of
garbage.
It's
got.
Two
thirds
of
two-thirds
of
the
module
are
copies
of
random
random
libraries
that
duplicate
stuff.
That's
now
in
the
standard,
python
library
and
there's
places
where,
like
it
reads
its
own,
it
reads
its
own
source
code
to
look,
look
for
the
string
arguments
to
put
in
the
help
text
so
kind
of,
depending
on
how
long
it's
going
to
take
before
we
want
to
redo
it.
C
C
Oh
yeah,
the
end,
it's
the
final
thing
I
put
in
the
in
the
in
the
in
the
notes,
which
is
even
if
we
fix
the
dmarc
problem.
There
are
a
significant
number
of
people
who
only
authorize
only
authenticate
with
SPF,
which
means
that,
even
even
if
we,
even
if
we
did
nothing
whatsoever
and
pass
through
the
literal
bits,
the
dmarc
would
still
fail.
So
we
still
need
to
do
some
sort
of
hack
similar
to
what
we
do
for
rewriting
the
addresses
on
mailing
lists,
which
is
ugly
but
anyway
around
it.
D
C
C
C
A
C
A
A
A
B
A
D
A
The
current
versions
of
the
connections
into
post
confirm
I,
mean
sorry
into
post
fix
and
the
current
connections
just
wrapping
the
current
connections
into
mailman.
We
might,
it
might
be
worth
pursuing.
But
let's,
let's
continue
that
conversation
offline
and
see
if
we
can
scope
how
big
that
effort
might
be.
C
A
A
A
A
Modifying
our
set
of
wrappers
our
extensions,
if
you
would
to
wagtail,
to
support
moving
the
IAB
website
onto
a
wagtail
instance,
that's
very
similar
to
our
www.itf.org
wagtail
instance,
because
I
already
want
to
talk
for
just
a
little
bit
about
the
activity.
That's
been
going
in
the
testing.
That's
been
going
on.
There.
A
E
Okay,
audio
in
place
great
so
as
explained
the
trust
have
explained
to
me
their
concerns
about
content,
licensing
for
wikis
and
con
contribution
attribution
for
Wiki
changes
proposed
on
GitHub
the
I'm
reading
this
out
loud
but
I'm.
Assuming
many
of
you
haven't,
read
it.
The
initial
view
of
the
truss
was
that
some
of
our
current
practices
could
no
longer
Contin
could
continue,
and
so
I've
proposed
some
possible
Technical
Solutions
to
them,
which
they
accept
and
which
I've
started
the
process
of
discussing
with
the
community.
E
E
Their
concern
is
that
contributions
made
via
GitHub
to
the
wikis
are
not
sufficiently
attributable
to
fulfill
the
requirements
of
the
note
well,
and
their
initial
view
was
that
we
should
not
allow
people
to
edit
the
wiki
via
GitHub
I
have
suggested.
Instead
that
we
have
a
mechanism
whereby
we
can,
as
the
maintainers
of
the
wikis,
see
whether
you
know
have
an
automated
process
that
tells
us
if
that
GitHub
user
is
somehow
registered
in
data
tracker
and
do
the
choose
whether
or
not
to
accept
it.
E
They
are
concerned
that
we
can't
have
a
blanket
ccby
4.0
license
across
the
wikis,
because
the
wikis
contain
things
that
our
contributions
and
need
to
be
licensed
under
the
TLP,
the
the
trust
legal
Provisions,
for
example,
excerpts
from
rscs
and
other
things.
Now.
The
initial
view
of
the
trust
is
any
such
content
should
not
be
permitted,
but
we've
actually
had
that
content
there.
E
For
many
many
years
on,
the
wikis
we've
just
never
been
particularly
clear
about
what
the
licensing
of
The
Wiggies
was,
and
so
I
propose
that
instead
we
build
a
macro
for
the
wiki
that
can
be
used
to
flag
content
under
a
different
licensing
scheme.
So
we
apply
one
blanket
licensing
scheme
to
the
wiki,
whether
that's
TLP
or
ccby,
is
to
be
decided,
and
then
anything
is
an
exception
to
be.
That
is
flagged
as
under
that
exception,
and
I
haven't
yet
raised
out
on
the
list
for
discussion
at
all.
E
E
A
All
right
we're
through
the
Section
that
I
anticipated.
We
would
have
discussion
time
on
a
really
quick
summary
of
what's
in
the
fyis,
if
we
had
a
spot
where
we
need
to
stop
and
discuss
just
speak
up,
the
rfps
for
the
change
to
the
iot
infrastructure
are
being
prepared.
Now
we
have
a
lot
of
mechanism
that
we
are
building
for
our
interaction
with
cloudflare.
A
A
A
This
is
the
the
meat
of
this
and
the
implementation
part
of
that's
going
to
come
behind
the
Django
4
upgrade
process
after
we've
got
Django
4
deployed
because
the
what
we're
going
to
build
is
very
definitely
going
to
be
based
on
The
View
right
framework
and
that
we
really
want
to
have
django4.
Underneath
this
before
we
start
into
that.
A
The
road
map
has
been
up
for
several
weeks.
We've
got
feedback
that
it
is
useful.
We
added
quite
a
bit
of
content
to
it.
While
we
were
at
the
tools
Retreat,
please
review
it.
If
you
see
anything
at
any
time,
that
is
confusing
missing,
raise
the
point
on
ietf,
discuss
or
directly
with
us
and
we'll
we'll
make
sure
that
the
roadmap
is
giving
the
community
a
clear
view
of
what
we're
actually
working
towards.
A
We
had
a
hiccup
with
the
timestamps
that
required
pre-intricate
hoop
jump
to
fix,
but
we
got
through
that
we're
watching
performance
with
the
instrumentation
that
we've
got
we're
not
seeing
the
overall
performances
significantly
different,
but
performance
for
pages
that
people
were
particularly
sensitive
to
have
anecdotally,
improved
we're
suspecting
that
much
of
that
observed
people
observed
Improvement
is
in
parts
of
the
stack
that
are
current
tools,
aren't
instrumenting,
so
we're
looking
at
changing
the
our
instrumentation
to
see
further
up
the
stack
so
that
we
can
see
what's
happening
with
performance.
There.
A
We'll
be
talking
to
serious,
open
source
later
this
week
to
schedule
the
next
step
in
managing
our
postgres
instances.
The
next
task
is
turning
on
replication
between
the
postgres
instances
at
itfa
and
itfc
right
now.
Our
failover
strategy
is
that
we
would
take
one
of
the
daily
dumps
and
restore,
but
that
would
mean
that
if
we
had
to
failover,
we
would
lose
a
chunk
of
a
day.
We'll
have
replication
set
up
over
the
next
several
days
and
then
we'll
evaluate
whether
or
not
there
is
a
significant
performance
impact
for
having
that
replication
running.
A
A
A
We
had
a
period
of
time,
the
beginning
of
May,
where
our
asynchronous
container
processing
draft
submissions
lost
the
ability
to
send
email.
This
was
a
unexpected
change
in
the
behavior
of
Docker
on
our
primary
server.
Docker
compose
brought
the
network
for
the
containers
up
on
a
Network
that
was
not
what
had
previously
been
advertised
is
the
set
of
networks.
Docker.
Would
you
know
the
the
addresses
that
Docker
would
come
up
with
with
networks,
so
it
was
outside.
A
Of
the
firewall
allows
on
itfa
to
let
the
let
the
doctor
reach
out
to
the
out
even
to
the
in
to
the
mail
server.
Once
we
discovered
it,
the
fix
was
fairly
quick.
We
reached
out
to
every
draft
submitter
whose
submission
was
affected,
and
we
know
for
certain
that
their
submissions
are
complete.
So
we
don't
we've
not
left
anyone
hanging
and
we're
still
discussing
what
we
can
do.
While
we
are
still
on
ietfa
t
be
more
resilient
to
changes
like
that.
A
A
E
A
E
A
The
next
release
of
the
data
tracker
we'll
be
using
that
for
fonts
in
particular,
and
very
soon
after
that,
we
will
be
using
it
for
the
artifacts
that
we
have
that
are
always
and
changing.
So
the
internet
drafts
the
document
types
that
we
have
where
once
they're
out
there.
They
don't
change
the
they'll,
be
served
from
this
site.
A
Okay,
development
projects-
the
data
tracker
has
seen
a
bunch
of
releases.
We've
talked
about
some
of
the
things
that
went
by
already
we're
in
the
process
of
the
Django
4
transition.
We've
made
it
all
the
way
up
to
jennygo3.t,
there's
a
link
to
our
development
test
server
that
has
the
Django
Branch
running
on
it.
Please
spend
some
time
poking
around
there
and
help
us
find
things
that
are
broken.
The
diff
crawls
are
currently
not
finding
any
differences
between
the
sites,
but
those
don't
exercise
forms.
A
So
those
of
you
that
are
used
to
doing
Secretariat
things
or
iisg
things.
Please
take
some
time,
go
poke
around
on
that
site
and
make
sure
it
behaves
the
way
you
expect
it
behaves
so
far.
The
kinds
of
trouble
that
we've
been
running
into
as
we've
been
migrating.
You
know
I
have
been
of
the
Forum
that
dependencies.
A
A
Oh
yeah,
so
what
we're
planning
to
do
in
the
short
term
now
is
to
finish
this
migration
into
Django
4.
To
continue
to
move
these
artifacts
that
we
discussed
earlier
into
the
data
tracker.
We
have
a
couple
of
long
outstanding
problems
that
we're
going
to
dive
into
shortly.
After
that,
we
have
to
repair
our
reference
relationships,
because
the
submission
tool
was
building
them
on
submission
from
the
XML
without
expanding
the
includes.
A
So
anything
that
was
in
the
repository
in
XML
with
a
pronoun
and
X
I
include
to
a
reference
did
not
actually
get
captured
as
a
reference
in
the
related
document.
Since
in
the
data
tracker,
we've
got
a
workaround
for
that
until
we
get
to
the
point
in
the
data
tracker,
where
we
only
accept
fully
expanded
documents.
A
That's
on
the
roadmap,
but
not
as
soon
as
repairing
these
reference
relationships
and
we're
also
going
to
regenerate
all
of
bibexml3
using
modern
author
extraction
tools
dealing
a
little
bit
with
some
of
the
questions
about
initials
I
had
hoped
to
wait
to
do
this
until
we
had
agreement
in
the
community
and
the
change
at
the
RPC
to
where
we
just
used
author
names
as
blobs.
Instead
of
paying
attention
to
this
first
initial
last
name.
But
that's
still
looking
like
it's
going
to
take
significant
time.
So
we're
going
to
move
ahead
and.
E
A
We've
got
a
wave
of
changes.
We
need
to
make
to
support
nomcom
a
wave
that
we
need
to
support
the
meeting
scheduling
then,
of
course,
a
wave
to
support
the
meeting
itself.
Those
will
be
the
remaining
focus
and
in
the
very
short
term
and
as
I
mentioned,
getting
into
fully
expanded
drafts
and
then
the
work
to
support
the
RPC
tool
refresh.
F
F
So
rfcd
rxctive
can
be
used
to
compare
XML
files
or
background
files
without
converting
them,
and
then
another
change
was
the
document
comparison
on
Auditors
uses
rfcd
by
default.
There's
a
work
on
the
way
to
switch
between
RSV
depend
IDT,
but
I
guess
most
people
would
be
more
comparable
with
rxd
any
questions
about
other
tools.
F
F
F
As
for
upcoming
work,
I
am
compiling
a
a
GitHub
repository
all
the
fonts
that
Excel
taxi
uses
right
now.
The
instructions
are
to
use,
not
a
funds,
not
not
a
funds
from
Google
and
the
repository
that
then
the
link
to
the
not
funds
that
they
provide
on
accept
RFC
is
outdated.
It's
not
maintained
anymore,
so
in
the
future,
this
font
repository
would
be
the
front
repository
to
go
and
get
all
the
phones
required
for
excellent
RFC.
F
Apart
from
that,
a
couple
of
other
features
that
would
be
in
the
next
release
type
tables
which
allows
you
to
learn,
add
cable,
City,
majors
and
other
content,
and
we
will
have
a
widget
that
gets
the
notepad
from
the
ietf
GitHub
repository
and
show
it
whenever
that
widget.
Is
you
so
idea
behind?
That
is
the
not
available.
A
A
So
we've
had
a
long
running
project
that
was
cycle,
starved
because
we
were
focusing
on
off
PD
on
at
the
IMAP
server,
where
we
can
allow
mapping
from
one
username
to
another
to
support
people
who
have
changed
their
primary
login
at
the
data
tracker
and
move
to
the
point
where
somebody
can
use
any
of
the
email
addresses
that
the
data
tracker
knows
them
by
is
their
username
when
they
log
in
this
doesn't
mesh
well
with
the
core
implementation
of
user
at
the
IMAP
server.
So
Alexi
and.