►
From YouTube: ROS 2 Security Working Group (14 Feb 2023)
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
Okay,
wait
we're
recording
now
all
right,
welcome
everyone
to
the
February
meeting
for
the
security
working
group.
I
will
start
just
to
request
your
approval
for
the
meeting
minutes
for
the
last
meeting
on
January
10th
2023.
You
can
check
those
in
there
are
pull
requests
in
the
GitHub
repo
for
the
group.
A
It
is
linked
in
the
agenda.
Let
me
share
it
again
for
those
of
you
who
just
joined
on
the
chat,
so
there
you
can
see
a
very
quick
summary
of
what
was
discussed
in
the
previous
meeting,
but
the
recording
is
also
linked
there.
So
all
of
the
recordings
are
there
available
in
case
for
those
of
you
who
are
new
to
the
group,
you
can
follow
any
discussions
that
are
interested
particularly
on
fascinating.
So
thank.
B
A
Yes,
there
are
people,
requests
from
past
meeting
minutes,
so
yeah
there's
a
little
bit
of
admin
cleanup
needed
there.
B
Okay,
all
right
all
right,
I'll
go
back
and
merge
those,
and
if
you
want
to
send
me,
your
GitHub
user,
ID
I
can
also
add
you
to
the
to
the
right
permission
to
the
repo.
A
Yeah
sure
So
currently
is
a
fork
that
I'm
making
PRS
with
to
the
groups
yeah.
We
need
somebody
with
approver,
I,
guess
rights
and
then
a
reviewer
too
much
I.
Suppose.
A
Okay,
thank
you
for
looking
into
that.
Nothing
right.
So,
let's
get
started
then,
with
the
first
item
in
the
meeting.
I
wanted
to
propose
a
discussion
about
about
security,
analysis
tools
for
Rose
code,
so
imagine
I
linked
posts
that
I
made
on
this
course.
Where
I
was
asking
for
some
Community
input
about
this
topic
and
invited
anyone
to
comment
or
to
join
the
meeting
and
contribute.
A
The
general
idea
is
I,
see,
there's
a
little
bit
of
a
gap
in
how
I
guess
the
rose,
Community
talks
about
or
address
the
stage
of
thought
Security
in
the
project,
just
because
I
don't
seem
very
much
being
talked
about.
Maybe
there
is
but
I'm
very
interested
in
that
aspect,
and
you
know
the
main
sort
of
community
questions
where
what's
standing
tools
we
use
in
your
projects,
such
as
a
static
analysis
tools,
some
Dynamic
analysis
like
fussing
and
how
are
you
finding
the
integration
with
loss
to
work?
A
So
there
are
a
number
of
projects
that
I
Community
projects
actually
that
I
Linked
In
the
this
discussion
document,
another
sort
of
wrappers
for
Ross
to
integrate
some
of
these
tools.
You
may
be
familiar
with
them,
like
the
call
called
Nixons
check
amidst
icos
Etc,
and
so
those
are
projects
that
are
interesting
precisely
because
they
may
they
make
the
situation
with
Ross
and
I
thought.
That
was
something
that
there
could
be
a
need
that
or
maybe
a
documentation
that
the
group
images.
A
Yeah
welcome
so
yeah
in
this
document.
I
had
these
discussion
questions
like
that
I
just
summarized
and
I'd
be
interested
in
in
hearing
your
perspectives
like
if
you
use
some
of
these
tools
and
especially
if
you
feel
like
and
I'm,
throwing
it
out
there.
Maybe
this
is
not
something
that
is
really
needed,
but
if
the
community
needs
more
documentation
too
know
how
to
get
started
like
integrating
these
tools
and
which
tools
to
use
I
know,
space
rules
have
been
making
a
big
effort.
A
In
this
sense
we
had
them
in
the
November
meeting
discussing
this
and
they
author,
the
I
meant
icos
integration,
for
example.
So
that's
the
main
push
that
I'm
seeing
in
that
direction,
but
maybe
there's
more
that
we
can
do
in
terms
of
documentation
or
even
supporting
these
Robbers
for
specific
tools.
A
B
Maybe
we
can
look
at
some
case
examples
of
of
other
open
source
Community
projects
spearheading
their
own
kind
of
like
I
I
feel
like
a
lot
of
this,
like
in
terms
of
quality
assurance,
was
sort
of
done
from
a
completeness
and
sounds
perspective.
B
I
haven't
seen
this
too
often
in
terms
of
from
just
a
purely
security
perspective.
I'm
like
more
like
protecting
against
memory
management
bugs
is
more
like
a
byproduct.
Their
security
is
a
byproduct
of
that
and
they're.
Then
they,
the
initial
objective,
unless
you're
like
looking
at
a
crypto
Library.
B
But
I
I
can
I
can
see
the
the
the
Merit
and
having
sort
of
a
more
seamless
immigration
style,
we're
like
just
the
same
way
that
you
can
use
the
mixins
to
enable
lenters
or
code
cleanup
or
style
checking,
because
also
like
I
guess.
You
know,
space
Ross
is
what
they're
interested
is
that
having
the
the
memory
management
checks?
B
Is
there?
Are
there
purpose-built
tools
just
for
in
terms
of
security?
It
seems
like
that.
It's
very
domain
specific
like
like
in
our
case,
if
we,
if
we
want
to
make
some
kind
of
linth
or
like
maybe
we
would
check
that
your
permissions
aren't
excessive
in
terms
of
you're,
allocating
something
that
you're
not
using
in
your
launch
file
or
something
or
week
week,
and
maybe
focus
on
something
that's
domain
specific.
A
So
the
main
focus
while
I
was
thinking
about
this
was
tools
that
are
generally
are
generally
available,
they're
open
source
for
yeah,
linking
and
generally
reviewing,
C,
plus
plus
and
python
code.
A
You
mentioned
obviously
there's
some
more
raw
specific
aspects
like
a
launch
file,
but
this
is
more
like
two
seconds
that
exists
that
can
be
used
on
CPS,
Plus,
Code,
obviously,
for
managing
cases
like
memory
management
issues,
all
kinds
of
particular
ones,
but
I
guess
you
search
in
the
rust
Community
was
the
most
interesting
part
to
me
like
is
it
being
used?
A
Are
they
being
used
to
to
ensure
the
security
aspect
of
the
projects?
It's
a
big
question
for
me.
B
B
One
Direction
this
could
go
in.
Is
we
see
a
lot
of
other
like
specific
language
ecosystems
that
are
working
in
in
sort
of
Tandem
with
dependabot?
You
know,
like
the
you
know,
they
property
that
I
think
in
GitHub,
recently
acquired
or
Microsoft
acquired
in
terms
of
doing
the
automated
scanning
of
packages
and
dependencies,
and
so
you
know,
if
you're
reliant
on
a
down
on
a
on
a
Upstream
package
and
it
updates,
because
of
a
security
patch
or
something
like
that.
B
Based
on
on
your
inherent
dependency,
you
get
a
notification
to
either
bump
up
your
up
your
version
and
as
well
as
some
notice
on
that
there
you
need
compatibility
or
API
changes
so
like
that
you
see
that
in
Rust
you
see
that
in
Python
you
see
that
in
JavaScript
and
typescript
in
Ross
like
I,
feel,
like
you
know,
that's
something
that
would
be
kind
of
nice
to
have
in
terms
of
like
you're.
Just
we,
we
inherently
depend
on
packages
and
we
have
sort
of
this
abstraction
model
of
Federation.
B
B
You
know
like
if
a
package
updates,
but
it's
not
in
your
same
distribution,
that's
sort
of
orthogonal
to
what
kind
of
dependency
you
can
necessarily
accept
without
you
know,
overhauling
your
own
package
to
Target
the
next
latest
distribution
as
well,
so
it's
not
entirely
equivalent,
but
that
might
be
something
worth
exploring
and
see
if
how
viable
that
might
be
due.
That
might
be
able
to
to
do,
because
in
that,
because
they're
probably
going
a
long
way
in
terms
of
making
sure
all
these
Leaf
packages
in
the
ecosystem.
B
You
know
something
that
people,
don't
necessarily
Upstream
into
say
the
OS
or
F
repo
org,
that
they
are
still
inherently
using
the
the
latest
packages
with
all
the
security
updates
and
vulnerability
updates,
because,
like
big
big
projects
like
move
it
and
navigation,
you
know
they're
they're
pretty
huge
and
they
have
a
large
attack
surface.
That's
even
thoughthogonal
too,
just
necessarily
the
authentication
encryption
of
of
something
that
security
DLC
is.
B
You
could
like
do
denial
of
service
or
something,
and
that
would
still
be
kind
of
critical
in
terms
of
of
a
robotic
infrastructure.
So
if
they
patch
something
like
that,
it
would
be
worthwhile
to
get
sort
of
an
instant
feedback
on
all
the
other
Downstream
packages
that
that
use
that
so
that
yeah,
that's
just
my
kind
of
maybe
bring
up
to
on
what
what
might
be
viable.
A
Yeah,
that's
definitely
that's
a
good
point.
That's
not
what
I
have
mentioned
in
my
document,
but
absolutely
it's
more
like
a
dependency
analysis,
I
guess
like
ssca
kind
of
tool,
dependable
and
I,
guess
that
would
definitely
detect
when
your
package
is
depending
on
some
package.
That
has
some
vulnerability
that
has
been
recorded
like
a
public
cve
that
is
reported
and
you
need
to
update
to
another
version.
That's
not
vulnerable!.
B
Yeah
talking
about
CDE
is
we
had
the
CDs
filed
for
the
the
issue
that
we
discussed
one
or
two
security
meetings
ago
about
the
signature
delegation
or
or
being
able
to
sign
your
own
permissions.
So,
like
I,
think
two
of
the
three
DDS
vendors
have
already
patched
the
issue
and
their
main
branches
I'm,
not
sure
if
they've
cut
a
release
yet,
but
when
they
cut
a
release,
and
it's
then
packaged
into
the
the
Ross
package
repository
having
sort
of
depend
about
provide
those
notifications.
B
Everyone
hey
rebuild
your
your
own
packages
to
fix
this
particular
Target
issue,
to
make
to
kind
of
transitively
make
and
and
disseminate
the
issue
of
whatever
CDs
were
discovered.
D
B
So
it's
not
as
straightforward
yeah.
That's
the
issues
that
the
metadata,
the
versioning,
isn't
usually
fixed
in,
like
a
package
XML
file
or
their
cmake
or
whatever
it's
it's.
That
can
also
be
kind
of
frustrating
in
terms
of
when
you're
building.
You
know
Ross
packages
if
you're
like
you're
building
it
from
Source
you're,
like
what
version
of
opencv
did
they
use
to
get
this
deep,
Learning
Network
to
work,
but
that's
sort
of
the
Legacy
that
we've
kind
of
built
up
just
up
to
this.
D
A
So,
okay,
there's
a
comment
from
Mikhail
on
the
chat
two
years:
you're
referring
to
the
package,
XML
capabilities,
yeah,
there's,
definitely
some
issues
that
can
be
addressed
there.
D
Plus
I
mean
when
you
do
apt
update
it
already,
will
tell
you
there.
Is
these
updates
available
right?
If
people
don't
want
to
install
it,
I
mean
that's
kind
of
the
mentality,
a
lot
of
people
don't
ever
upgrade
their
packages
right,
I'm
sure
some
of
us
do
not
upgrade
all
the
packages
in
our
own
computer
right.
So
how
do
we
notify
them
that
you
know
there?
Is
this
Rose
to
core
update
available
available,
but
you
need
to
install
it.
It's
not
that
you're,
saying
it'll
be
good.
If
you
did
no,
you
need
to.
D
On
the
matter
of
static
analysis,
I,
don't
know
if
you
know
about
this,
these
are
guidelines
on
safety
in
C
plus
plus
it's
the
standard
used
to
write
the
C
plus
plus
code,
for
especially
for
automotive,
certifications,
it's
a
subset
of
the
language
and
there
are
static
analyzers,
yes
and
I,
like
whether
you're
combining,
although
it's
pretty
limiting
and
most
of
the
broad
shoe
code,
which
is
not
by
any
means
past
that.
But
there
are
things
like
that.
A
Yeah
I
guess
one
two
that's
been
used
in
the
Ross
project
for
certification
process
is
icos
that
space
process
pushing
for
in
this
sense,
which
is
also
very
very
powerful
tool
and
as
I,
was
mentioning
before.
There's
this
rubber
for
Ross
the
amen
psychos
picture
connected.
A
And
yeah
what
I
was
generally
going
for
is
what
what
we
can
do
to
sort
of
provide
better
documentation
or
promote
using
certain
tools
to
address
this.
For
these
issues
that
you're
bringing
up
the
dependency
issues
for
known
vulnerabilities,
it
could
be
just
detecting
your
own
issues
in
the
code
that
you
created,
you
can
the
most
common
languages
for
us,
XC,
plus,
plus
or
python,.
B
Estimate
I
think
your
microphone
is
people
whacked
up
I,
don't
know
what
it's
doing.
Do
you
have
a
secondary
microphone.
B
You
could
also
try
typing.
If
you
want
that's
a
little
bit
lengthy,
you
can
come
back
if
you
once
you
post
your
typo.
C
Most
Checkers
are
not
different
source
and
even
the
standard
and
the
result
variation
you
have
to
pay
to
have
access
to
which,
like
kind
of
like
impacts,
usability
and
and
then
like
what
Russian
was
saying
like
when
things
will
Dependable
for
everybody
choose
massively
in
like
JavaScript
Community
as
well,
is
because
there
is
some
infrastructure
to
just
like
generate
like
Json
package
work
or
whatever.
C
So
it's
all
the
version
of
your
of
your
dependencies
that
can
get
updated,
and
today
it's
one
of
the
limits
is
that
like
yeah,
like
we
don't
have
this
infrastructure
in
Russ,
and
we
can't
expect
people
to
do
like
manual
changes
in
a
Federated
ecosystem
to
all,
like
actually
change
of
academics
and
I'll
change
our
dependencies
rebloom
rebuild.
It
would
be
like
we're
going
to
be
very
hard
to
get
critical
mass
on
this
and
now
I'm
gonna.
Let
us
read
more
test
events.
A
Yeah
is
there
one,
do
you
want
to
elaborate
more
on
this
question
you
put
on
the
chat.
B
I
think
the
Ross
tooling
working
group
is
the
they're
responsible
for
the
the
Ross
infrastructure
LCI
pipeline,
so
collaborating
with
them
to
to
bring
some
more
example.
Workflows
is
is
is
definitely
where
I
think
the
most
experience
would
be
if
we
find
if
we
settle
on
on,
if
there's
any
specific
security
domain
static,
analysis
tools,
honestly
yeah,
we
we
follow
up
with
the
the
tooling
working
group
to
maybe
incorporate
that
upstream
or
provide
some
more
Community
examples.
E
B
D
Static
analysis
they
they,
you
run
your
you
run
your
tests,
your
test,
suite
and
then
it
t-san
would
complain
about
the
potential
Deadlocks
and
data
races
another.
Sometimes
we
complain
about
memory
management
issues.
Basically,
regarding
you
know,
overflowing
array
or
writing
out
of
bounds.
Those
kind
of
things.
C
C
Packages
to
get
those
triggers
on
the
raspberry
phone
or
like
or
then
it
will
go
back
to
like
a
GitHub
actions
like
to
provide
a
simple,
simple
set
of
templates
that
people
can
use
to
be
able
to
like
run
those
like
Nike
or
whatever.
On
each
version
of
the
surface.
D
A
No
okay,
I,
don't
wanna
use
all
of
the
meeting
time
on
this,
but
yeah.
Thank
you
so
much
for
the
ideas.
It's
a
it's
a
good
idea
to
sync
with
a
tooling
working
group.
Actually,
if
we
go
with
the
integrating
into
CI,
which
obviously
these
tools
will
have
to
be
so
yeah,
that's
good
I
keep
thinking
about
this
and
we
can
keep
discussing
in
the
future
right.
So
nobody
else
has
comments
about
this.
Maybe
we
should
move
on
to
the
next
item.
D
D
You
know
put
comments,
improvements
whatever
you
feel
it's,
not
it's
not
very
long
right,
but
it's
about
to
make
making
sure
with
a
specific
example
or
use
case.
You
know
how
which
file
you
should
deploy
to
the
actual
robots
which
ones
you
should
keep
on
the
organization,
because
there
were
some
paragraphs
saying
not
all
these
files
you
need
to
deploy,
but
it
was
not
very
clear.
I
think
if
you
were
just
you
know,
proofreading
and
or,
and
copying
and
pasting
from
Snippets.
B
I
I
took
a
look
at
the
tutorial
I
I,
like
the
the
table
that
kind
of
clarifies
and
spells
out
what
goes
where
the
the
I
think.
What
you
could
we
might
I,
I
might
I
might
add,
maybe
a
PR
to
it
to
we
could
probably
codify
just
about
everything
so
like
I
guess
there
is
Merit
and
having
the
user
walk
through,
but
they
might
like
make
a
mistake
in
terms
of
of
doing
all
the
file,
manipulation
and
moving.
B
What
we
could
do
is
just
have
it.
The
user
interact
with
the
ros2
CLI
and
then
have
the
user
read
through
an
example
of
like
a
a
Docker
compose
in
which
that
is
like
mounting
the
certain
volumes
and
starting
to
containers.
That
way.
B
Maybe
you
don't
have
to
make
an
assumption
on
what
the
user
has
or
doesn't
have
on
their
host
and
they'd
be
left
with
a
very
replicable
static
setup
on
where
the
files
went
and
did
the
exact
same
thing,
but
that's
that's
fairly
minor,
as
it
is
right
now.
It's
pretty
good.
D
Please
put
that
in
the
comments
and
we
will
take
care
of
it
for
sure.
We
were
actually
thinking
that
we
had
an
idea
of
maybe
having
or
at
least
worth
adding
kind
of
a
deploy,
deploy,
Enclave
verb
to
the
CLI
or
something
I.
Don't
know
if
that
could
STP
The
Enclave,
but
only
the
fines
that
need
to
be
put.
B
That's
something
that's
even
since,
like
sross
one
like
Morgan
Quigley
was
like
oh
it'd,
be
really
nice
to
have
some
kind
of
command
to
the
deployment
I
kind
of
felt
that
maybe
that
the
the
key
deployment
there's
a
little
bit
out
of
the
scope
in
terms
of
the
key
Generation
but
from
a
usability
standpoint.
I
think
it
if
there
is
Merit
and
sort
of
combining
those
utilities
under
a
common
command
line.
Interface,
but
I
think
it
sort
of
it
faces.
B
Sort
of
the
same
kind
of
big
issue
like
with
composition
or
multi-system
orchestration,
like
with
Ross
launch,
either
that
they're
trying
to
refactor
dealing
with
with
multiple
robots
with
one
Ross
launch
invocation
or,
if
you're,
trying
to
deploy
multiple
key
material.
B
So
maybe
maybe
we
should
look
and
see
if,
if
there's
already,
some
existing
tooling
that's
used
in
terms
of
key
deployment
that
we
just
kind
of
Leverage
on
rather
than
Reinventing
the
wheel.
But
if
not,
then
that's
also
something
we
could.
B
We
could
streamline
into
the
own
into
this
into
the
CLI
like
yeah,
just
using
SCP
under
the
hood
and
just
putting
the
target
checking
the
SSH
keys
are
matching
before,
like
deploying
the
the
nodes
by
the
keys
and
stuff
like
that,
but
then
to
take
it
to
your
other
example,
when
we're
maybe
using
PK
cs11,
that
kind
of
opens
does
that
complicate
anything
in
terms
of?
B
Was
it
really
mean
to
removing
these
Keys
when
you're
really
just
moving
strings,
but
we
can
get
that
discussion
a
little
bit
later.
C
Yeah,
that's
something
we
installed
in
writing
s
plus
two
and
the
CLI,
and
maybe
now
we
have
a
better
view
of
the
diversity
of
deployment
scenarios.
One
one
drawback,
or
at
least
thing
that
made
us
take
a
step
back,
was
that
if
we
are
targeting
like
people
using
like
TPMS
and
remote
connections
and
many
things,
it
would
be
like
pretty
hard
to
make
a
tool
that
actually
does
the
thing
people
want
and
because
it's
pretty
sensitive
material.
C
We
like
we
decided
to
take
a
step
back
and
like
not
Implement
that,
but
maybe
now
like
that,
the
tooling
has
evolved
a
bit
and
the
needs
are
evolved.
Maybe
we
have
a
better
like
set
of
use
cases
that
could
allow
us
to
say:
okay,
we
can
Target
some
of
them
and
provide
some
more
tooling
for
some
common
scenarios.
B
So
so
I'll
I'll
comment
on
the
on
the
tutorial,
with
the
maybe
using
a
Docker
composers
need
that
to
yeah
Independence.
B
B
B
So
dealing
with
like
if
the
user
has
different
file
versions
like
a
pen,
file
or
P11
file,
another
one
would
be
I
think
the
P11
file
extension
is
used
by
something
else,
but
is
that
sort
of
a
common
enough
extension
to
maybe
to
to
assume?
B
Token
and
string
kind
of
format
is
that
a
common
way
of
passing
ptsc11
configuration
to
other
libraries
or
or
applications
like
if,
if
something
else
is
using
a
TPM,
is
that
the
common
way
of
addressing
it
so
I'm
just
thinking
of
other
cases,
then
maybe
if,
if
the,
if
the
user
was
using
something
other
than
DDS
for
the
middleware,
but
was
using
Hardware
keys
to
facilitate
the
the
encryption,
would
they
it
would
be
fairly
trans
fairly
easy
for
them
to
use
this
URI,
this
PDA
711
URI
to
configure
the
whatever
underlying
library
and
application
for
the
encryption
anyway,
regardless
of
what
the
transport
is.
D
F
So
if
I,
if
I
understood
the
question
right,
the
biggest
is
11
Uris
are
pretty
much
coming
standard
for
addressing
these
requests
or
providing
these
requests,
but
how
to
actually
kill
the
URI.
There
is
not
really
the
standard
way
so.
F
Hsn
manufacturers
are
using
those
wrapper
files
to
store
them,
and
this
is
pretty
much
because
of
the
Legacy
Legacy
Solutions,
of
giving
just
the
file
class
for
that.
Okay
configuration
files
where
you
define
the
key
so,
for
example,
Apache
web
servers,
I
think
still
not
understanding,
because
it's
11
URI,
so
they
use
files
instead,
that's
why
the
SSM
lenders
have
those
rubber
clouds.
F
Shortly,
there
is
no
standard
way.
Unfortunately,
of
course,
what
I
think
of
is
that
alternative
way
is
to
give
the
Pegasus
11
URI
directly
to
the
configuration
file.
B
E
B
Okay,
so
so
at
least
there's
some
precedent
in
this.
This
kind
of
approach,
the
the
in
terms
of
backwards
compatibility
or
priority
I,
don't
have
too
much
to
comment
that,
like
backwards
compatibility
is
not
my
my
not
my
focus
but
maybe
Mikhail.
B
Do
you
have
any
any
comments,
because
you,
you
went
pretty
deep
in
the
rabbit
hole
when
you
were
first
creating
that
infrastructure
in
terms
of
rmw
kind
of
figuring
out,
which
is
where
and
whether
it's
what
they
do,
and
what
to
sanitize.
C
Yes,
I
guess
I
think
that's
changed
a
lot
since
I'm,
not
sure
anything
I
would
say
would
be
very
relevant,
but
but
at
least
like
my
feeling
and
I
I
discovered,
like
the
P11,
like
reading
all
the
work
you
gave
me,
but
that
seems
to
be
more
common
like
looking
and
chatting
with
the
guy
at
like
developing
clinics.
C
They
were
supporting
also
like
URI,
not
into
TPM
and
and
when
we
explored
like
what
kind
of
like
support
and
Hardware
material
could
be
used.
C
C
Now,
if
we
just
have
a
certain
way
of
specifying
it
so
in
itself
it
seems
much
easier
than
it
could
have
been
before.
E
B
Sounds
like
most
of
us
are
pretty
comfortable
with
what's
being
proposed,
so
I'm
gonna
go
probably
comment
and
give
my
approval.
It's
good.
D
Yeah
then,
my
next
question
is
what's
left
or
where
should
we
go
for
for
this
to
be
merged
right?
If
we
are
comfortable
with
it
here
and
there,
it
has
approval
so
yeah
I
think
that
I
think
Michael
can
emerged.
I.
B
Don't
know
the
the
rmw
folks
have
any
any
comments
like
I
just
said.
The
only
difference
now
is
that
you
have
to
pass
in
a
Boolean
on
whether
you
support
picket
on
whether
the
rmw
supports
this
PK
11
specification
or
not,
or
the
the
it
can
parse
the
the
token
or
search
for
the
P11
file.
I
think
that
was
the
only
kind
of
from
the
DDS
common
API.
That
was
the
kind
of
the
only
extension
yeah.
D
D
I,
don't
think
so
we
opened
up
request
for
267
years.
It
will
be
ruled
out,
I,
don't
know
still
don't
know,
maybe
a
couple
of
weeks
because
it's
probably
will
bundle
some
other
things,
and
then
we
need
to
wait
for
the
next
humble
seat,
but
it
will.
It
will
be
on
on
Master
soon.
So
that
means
I'm,
broadly
rather
soon
abilities.
F
D
You
can
take
a
look
at
this.
The
thing
is
that,
if,
if
you
use
them
the
certificates
infrastructure
created
by
the
rose
to
CLI,
it
was
discovered
here
that
you
can
actually
use
your
identity
certificate
to
cite
permission
documents,
which
means
that
you,
that
a
malicious
note
can
change
its
own
permissions.
D
Some
of
the
implementations
were
allowing
intermediate
certificates,
and
that
means
that
if
you
use
the
same
certificate
for
both,
then
then
your
identity,
your
identity
certificate,
can
be
used
to
sign
permission
documents,
because
there
is
a
chain
of
trust,
so
this
PR,
what
it
does
is
that
it
force
it
forces
that
you
know
the
chain
of
trust
is
not
verified.
You
need
to
use
the
ca
that
the
permission
CA
to
sign
to
permission
documents.
You
cannot
sign
subsidiary
because
you
can
assign
them
with
that
subsidiary
certificate.
D
D
In
many
cases,
I
don't
know
if
you
make
sense
for
the
I,
this
I
don't
know
I'm
not
I,
don't
have
a
strong
opinion
for
the
CLI
tool
to
generate
two
different
Cas
one
form
permissions
and
one
for
identity.
I,
don't
know
it's
not
necessary
according
to
the
specification
and
this
fixed
in
fastidious
and
I.
Think
in
other
implementations
it's
not
necessary,
but
I
think.
Maybe
it's
still
a
good
separate
of
concerns.
It's
a
variation
of
concerns.
I,
don't
know.
B
Yeah,
the
I
I
I
I've
asked
a
few
times
some
folks
in
the
OMG
group
and
like
what
were
the
original
intentions.
It
did
say
something,
but
it
it.
It
didn't,
stick
with
me.
So
it
was
a
little
unclear,
but
the
set
aside,
there's
probably
merits
to
to
make
this
more
configurable
and
the
first
iteration
of
sort
of
the
tool
I
made
with
like
teaching
and
sort
of
a
mix
of
event.
B
This
was
more
defined
in
terms
of
like
a
sort
of
a
hierarchical
kind
of
configuration
file,
and
you
can
delegate
on
what
signs,
what
and
and
what
kind
of
CA
you
want
to
use
and
how
the
ca
itself
might
be
configured
sort
of
like
a
replica
of
like
what
openssl
did
and
it's
sort
of
it's
weird
and
then
its
own
weird
configuration
language
but
more
like
a
more
structured,
yaml
or
Json
kind
of
set
up
that
we
didn't
end
up
porting
that
all
the
way
to
the
the
sros
CLI.
B
But
that's
something
that
I
think
it's
really
worth
revisiting
in
terms
of
making
the
sros2
CLI
a
little
more
configurable
and
how
you
want
to
initialize,
configure
or
use
your
Cas,
because
either
cases,
maybe
maybe
the
cas
or
elsewhere,
or
one
thing
that
sross
doesn't
do
that.
The
previous
iteration
did.
Is
we
sort
of
took
a
workspace
approach
on
sort
of
in
on
the
iterative
thing
where
you
might
have
certificate
requests.
B
So
that
way,
you
the
C
like
to
be
used
to
generate
all
the
certificate
signing
requests,
and
then
you
submit
that
elsewhere
to
get
them
signed
so
that
you
don't
have
to
co-locate
the
root
ca
key
material
on
the
same
machine
that
would
be
also
generating
the
key
material
for
the
nodes.
So
if
you've
had
some
very
strict,
that's.
B
Yeah
something
very
more
equivalent
to
like
the
webs
web
sphere
web
technology,
but
it
was
fairly
complicated
in
terms
of
substantiating
that
from
a
Ross
user
perspective.
But
we
should.
We
should
definitely
look
into
seeing
making
that
more
viable.
F
B
F
F
It
was
a
bit
confusing
also
in
the
beginning,
when
I
started
to
look
into
this
whole
resource
to
a
couple
of
years
ago,
but
and
as
an
example
in
our
our
system
of
VPN
building,
we
ended
up
making
our
own
sort
of
CA
logic
with
the
proper
cloud-packed
Cas,
and
then
we
are
using
the
srush
on
the
client
side
too,
generate
the
client-side
artifacts.
C
And
that
would
be
interesting
in
iterating
on
that,
because
it
seemed
to
us
that,
like
like
the
the
initial
approach
would
be
held
also
to
gain
adoption,
because
it
was
like
basically
Culkin
like
tool
on
elements
like
to
that.
I
did
a
lot
of
complexity
and
that's
why
we
went
like
a
merch
merch
single
rules,
but
but
we
see
all
the
limitations
of
trying
to
like
support
only
like
some
cases
and
rendering
everything
into
one
tool.
C
It
makes
it
like
hard
to
use
and
not
cover
all
the
cases
people
want
to
cover,
so
it
would
be.
It
would
be
interesting
to
like
gather
feedback
of
people
like
that
I
try
to
use
it
for
security
purposes
and
went
to
a
different
route
and
see
what
could
be
done.
I
mean
I,
don't
know
how
like
much
energy
and
time
people
would
have
to
like
spend
evolving
sros2,
but
it
would
be
interesting,
Gathering
feedback
and
see
like
what
are
the
different
approaches.
You
would
like
to
explore
and.
E
C
Don't
know
if
it
answers
your
original
question
Eduardo,
but
we
could
also
look
at
like
a
very,
very
simple
solution,
because
for
Astros
too
I
I
don't
remember
properly
because
it
was
very
long
time
ago,
but
I
think
it
was
more
of
a
measure
of
like
Simplicity
that
we
just
use
the
same
CA
for
both
but
I.
Don't
know
how
hard
it
would
be.
You
just
have
like
two
different
Cas
and
I
mean.
Maybe
it
wouldn't
be
that
drastic
of
a
change
to.
E
C
Like
avoid
that
specific
vulnerability
in
doing
your
default
default,
enclave
and
key
material
generation,.
D
Yeah
I
guess
for
the
simple
case,
would
not
be
very
difficult
right
to
generate
one
instead
of
two
and
then
identity,
I
guess,
I
haven't
looked
at
the
code
really
but
and
I
think
it's
necessary
to
have
first
seen
a
simple
CLI
tool,
at
least
for
the
as
an
entry
point
as
it
as
it
is
now.
You
know
these
are
the
tutorial.
D
We
create
a
claim,
and
you
know
it's
three
commands
and
you're
good
to
go,
but
it
it
probably
makes
sense
to
that
also
has
to
come
more
like
a
more
advanced
version
where
you
can
really
configure
how
you
do
things,
so
it
might
be
that
it
supports
the
two
things
or
I
think
that
would
be.
B
If,
if
you,
if
you,
if
you
have
your
own
CA,
it's
it's
it's
about
as
simple
as
just
substituting
the
files.
The
other
thing
is.
The
sros
itself
is
also
using
Sim
links
to
point
to
the
permission,
CA
and
identity
CCA,
as
if
there
were
different
Cas.
It
just
happens
that
when
we
bootstrap
the
key
store,
it
just
generates
the
send
links
the
point
for
the
same
file.
B
So
the
end
users
could
replace
that
before
generating
enclaves
to
use
separate
Cas.
But
it's
it's
always
sort
of
been
a
balance
on
what
people
want
to
do
out
of
band
either
in
terms
of
CA
generation
or
key
distribution
to
targets
so
yeah
I
guess
we're
just
still
figuring
out
what
kind
of
scope
and
and
how
much
we
want
to
facilitate
within
the
same
bundle
tool.
D
B
If
it
would
be
nice
kind
of
have
a
survey
and
see
what
people
how
many
people
are
using
bras
with
secure
DDS
as
a
as
of
now,
we
just
submitted
a
a
paper
sort
of
recently
on
the
topic
and
one
of
the
criticisms
for
the
reviewers,
like
you
know
how
many
people
are
actually
using
this,
and
so
we,
as
the
authors
were
like.
B
Oh
man,
this
is
Ross
like
we
don't
really
know
like
who's
using
what
other
than
you
could
kind
of
estimate,
maybe
from
the
the
user
analytics
that's
been
published.
You
know
this
month,
but
that's
not
very
that's
not
specific
to
what
features
people
are
using
or
well
I
guess
you
know
we
could
figure
out
how
many
people
are
installing
s
Roth
too,
but
I
think
that's
that's
bundled
within
one
of
the
default
meta
packages.
So
that's
not
yeah.
It's.
D
B
It
may
be
worthwhile
to
maybe
asking
oh
sorry
for
the
if
they
want
to
start
at
a
survey
on
like.
Are
you
using
security?
What
features
do
you
see?
Security
are.
B
Or
are
you
using
a
secured
software-defined
Network
or
you
actually
encrypting
your
DDS
drafting.
B
Er
there
and
eventually
the
survey
we
can
probably
post
on
the
discourse.
C
I
think
we
need
to
check
so
it's
good
for
being
here
and
talk
to
her,
because
I
think
one
thing
she
wanted
to
avoid
and
that's
why
she
tried
to
be
the
focal
points
of
surveys.
C
Is
that
like
at
some
point,
a
lot
of
surveys,
emails
from
this
course
and
and
so
they
wanted
like
for
them
to
go
screw
up
in
robotics
so
that
they
could,
like
I,
don't
know
like
vendor
surveys
or
like
emit
surveys
at
specific
times
to
avoid
having
this
continuous
flow
of
survey
requests
to
the
community
so
that
we
can
start
at
least
like
iterating.
B
Yeah
yeah:
this
is
I,
wanted
to
get
cat
in
on
it,
because.
B
Only
get
people's
attention
once
because
otherwise
they're
like
just
gonna
spam,
so
you
want
to
make
sure
you
ask
your
questions.
Survey
right.
B
A
B
The
the
video
is
recording,
we
can.
We.