►
From YouTube: ROS WebTools Working Group (2021-07-30)
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
I've
got
a
few
folks
who
dialed
in
anybody
who
hasn't
been
with
us
before
adrian,
I'm,
not
sure
if
you
actually
joined
us
in
previous
working
group
music.
A
B
A
Yeah,
do
you
have
anyone
else
in
here?
That's
going
to
the
the
web
tools
working
groups.
B
A
But
adrian's
from
fox
glove,
which
is
a
web-based
rus
tool
suite
and
he's
been
leading
a
lot
of
the
conversations
in
the
web
tools
working
group,
which
is
has
been,
I
think,
quite
valuable.
So
far,.
A
A
couple
weeks
ago,
maybe-
but
yes
I
believe
so
I
don't
know.
A
Yeah,
our
primary
focus
has
been
rosberg,
but
I'm
always
happy
to
talk
about
other
tools
in
the
ecosystem
as
well.
It
turns
out
that's
absolutely
big
project,
so
it
takes
up
a
lot
of
time.
C
Yeah
yeah,
what
about
colcon
and
like
the
whole
and
rostef
and
like
the
build
system?
Where
does
who?
Who.
A
B
A
You
know
there
isn't
a
working
group
where
that
stuff
gets
talked
about,
and
we
have
thought
about
that
talked
about
that
in
here,
a
bit
where
there's
almost
in
the
charter
for
this
working
group
of
tooling,
it
kind
of
falls
across
two
boundaries.
A
On
one
side,
you've
got
like
application
level,
tooling,
like
rosbag,
r,
cute
arviz,
and
then
you've
got
developer
tools,
which
is
you
know,
colcon
raw
step.
A
Maybe
I
meant
the
build
system,
kinda
accounts
for
it,
and
then
we,
you
know
we're
maintaining
github
actions
that
people
can
plug
into
their
repositories
to
do
builds,
and
so
the
majority
of
our
conversations
in
here
have
been
about
rosbag
and
about
those
github
actions.
That's
what
we're
actually
maintaining,
which
kind
of
fall
on
two
sides
of
this
boundary
and
I've
wondered
if
maybe
we
do
need
a
developer
tools
working
group,
but
I
don't
know
if
there's
anybody
looking
at
those
things
super
actively
right
now
at
all.
A
It's
hard
to
say,
and
I
don't
feel
plugged
in
enough
to
to
try
and
lead
such
an
effort
myself.
I
mean
I've
already
got
two
working
periods
going
on
picking
something
time.
A
A
Project
and
he
since
leaving
open
robotics,
has
not
been
super
active
on
the
ross
community.
He's
got
other
things
to
do
at
his
new
job,
so
we
are
getting
some
some
updates
in
here,
but
I
think
that
it's,
you
know
it's
a
quiet
project.
A
A
Live
but
we
could
also,
we
could
talk
about
those
things
here.
I
just
I
guess
my
point
was
I
don't
know
if
we
have
much
expertise
in
here,
but
it's
probably
an
okay
spot
for
the
conversation,
and
maybe,
if
you
notify
discourse
ahead
of
time
that
we're
going
to
talk
about
that,
then
we
could
get
some
people
who
are
more
familiar
with
this
code
base
in
here.
Talking
about
it
like
I,
don't
have
any
particular
thoughts.
I
was
just
sort
of.
A
Falls
but
I
mean
it
is
nice
to
have
these
public
forums
for
things
it
makes
it
feel
a
little
bit
like
a
proper
distributed
open
source
project
where
it's
easier
to
get
involved,
whereas
if
it's
like-
oh,
I
don't
know,
if
someone
open
robotics
knows
about
that,
then
it
kind
of
feels
like
you're
just
trying
to
edge
in
on
someone
else's
private
project,
yeah
yeah.
A
C
A
Know
if
we
had
any
action
items
from
last
time,
so
I
don't
think
there's
anything
to
review
there,
I'm
going
to
add
to
the
template,
because
I
did
this
in
the
other
working
group,
which
is
art
meeting
with
boarding.
It's
a
good
reminder
yourself.
I
don't
always
remember
to
do
it.
A
C
Yep
so
last
meeting
a
couple
of
weeks
ago,
we
were
talking
about
the
functionality
for
reading
multiple
bag
files
in
a
merged
way,
so
taking
multiple
bag
files
and
then
interleaving
the
data
based
on
timestamps
to
produce
a
single
output
stream
and
basically
the
one
of
the
foundational
aspects
of
conversion
bay
conversion
for
lag
merging.
C
So
I've
got
some
very
extremely
rough
prototypes
in
these
two
branches.
Here
on
my
fork,
the
first
one
is
converting
the
reader
class
into
a
a
multiple
bag
reader
and
the
second
one
is
similar.
It's
it's
sort
of
taking
the
separate
aggregator
approach,
but
I
also
did
some
work
to
define
interfaces
separate
from
the
base
reader
interface
for
different
aspects
of
reading.
C
In
order
to
get
around
the
problems
of
the
current
reader
interface
not
really
being
suited
to
multiple
multiple
bag
files
at
once,
and
so
I
think
if
you
look
at
the
the
first
first,
if
you
go
down
to
where's
the
file
jump
on.
A
This
one's
the
one
where
you've
added
the
new
functionality
to
the
existing
reader
class,
so
that
we
can
read
multiple
files
right.
C
A
C
Yeah
yeah,
I
mean
I,
I
hadn't
had
another
solution
for
this.
That
was
the
hack
to
get
it
to
compile.
Basically,
there's
problems
like
that
in
it,
the
other
diff
is
much
more
complex.
It
doesn't
have
any
of
the
actual
functionality
in
it
yet
because
I
need
to
refactor
that
from
the
first
diff
into
it,
but
it
does
have.
If
you
look
into
reader
interfaces,
you'll
see
a
bunch
of
headers
there's
one
for
the
base
reader
interface.
C
So
I
said
I
decided
that
a
base
reader
is
basically
an
object
that
can
read
from
a
bag
file.
So
it
has,
has
next
read
next
and
get
all
topics
and
types.
So
those
are
the
three
fundamental
aspects
of
three
fundamental
methods
of
a
reader.
I
decided
every
reader
has
to
be
able
to
do
those
others,
it's
not
a
reader.
If
I
can't
do
them
from
that.
There's
an
interface
for
filtered
bags,
filtered
readers
where
you
can
have
a
couple
of
functions
for
setting
filter
and
resetting
filter,
there's
an
interface
for
metadata.
C
Yeah
yeah,
so
one
change
I
made
was
to
move
to
rti
rather
than
having
a
separate,
open
and
closed
methods.
When
you
construct
a
bag
reader,
it
opens
the
bag
for
when
you
destruct
it
or
closes
it
just
because
I
think
it
makes
the
airport
a
bit
cleaner,
but
I'm
not
tied
to
that.
Okay.
A
C
Is
gonna
be
basically
the
same
as
the
changes
to
the
reader
class
than
the
other
diff,
so
I'm
I'm
more
interested
in
the
interfaces
here.
What
this
is
what
the
objects
are
like,
what
the
inheritance
tree
is
going
to
look
like,
and
if
you
look
at
the
sequential
reader,
you
can
see
what
the
inheritance
is
going
to
look
like.
So
this
is
inheriting
the
base
reader
interface
because
it
needs
to
be
it
needs
to
be
a
base
reader.
C
It
also
inherits
the
single
bag,
open
interface,
the
metadata
reader
interface
and
the
filtered
reader
interface,
I'm
kind
of
not
married
to
the
single
bag.
Opener
interface
thing
it
doesn't
really
work
out
the
way
I
thought
it
would
because.
A
C
That
was
before
I
shifted,
rtai
and
now
now
that
now
that
we're
using
rti
and
the
stuff
it
kind
of
doesn't
make
any
sense
anymore.
So
I'm
quite
happy
to
get
rid
of
that
one.
C
C
Sorry,
it's
very
late.
I
understand
yeah
any
confusive
runtime
tight
identification,
so
the
merge
the
merged
reader
you
see
doesn't
take.
It
doesn't
have,
for
example,
the
the
metadata
reader
interface,
because
that
doesn't
make
much
sense
for
a
merged
reader,
because
it's
metadata
for
a
single
bag.
C
For
example,
it
doesn't
have
the
filtered
reader
interface,
but
it
does
have
the
base
re.
The
base
functions
of
a
reader
in
terms
of
has
next
redneck
schedule,
topics
and
types,
and
because
because
these
are
both
face
readers,
you
can
use
them
interchangeably
for
reading
from
bags,
but
the
the
sequential
reader
sequential
reader
is
capable
of
having
filters
set
and
metadata
read
from
it.
C
A
Yeah,
I
think
that
this
makes
sense.
Overall,
I
would
have
like
smaller
questions,
but
maybe
we
should
get
into
that
in
the
actual
review.
I
don't
want
it
like
on
a
high
level.
I
I
think
that
it
makes
sense
that
this
approach
is
a
little
bit
cleaner.
A
A
But
regardless
being
able
to
set
a
different
filter
for
every
sequential
reader
is
fine
because
you're
constructing
a
bunch
of
sequential
readers
which
could
have
their
own
filters
from
the
base
reader
interface
and
pass
that
to
the
aggregate
reader
yeah.
A
Yeah,
that's
fine.
I
mean
it's
a
simple
loop
where
you
construct
them:
yeah.
Okay.
I
think
that
I
agree
with
this
approach
and
I
mean
anybody
have
any
major
concerns
top
level
about
this
for
simultaneously
reading
multiple
separate
roster
bags.
C
So
my
only
concern
here
is
the
reader
facade.
It
still
has
methods
for
things
like
get
metadata
and
these
aren't
going
to
work
with
an
aggregate
reader.
So
I'm
my
feeling
is
that
we're
better
off
getting
rid
of
the
facade
and
just
working
directly
with
based
reader
pointers.
A
Yeah,
I'm
actually,
I
can't
remember
right
now
why
the
the
reader
facade
exists
as
opposed
to
the
implementation.
A
I
mean
we
do
have
clearly
multiple
implementations
yeah,
but
we
could
be
using
polymorphism
to
to
just
pass
around
the
base
interfaces
we
care
about,
rather
than
this
thing
I
wish
karsten
was
on
the
call,
because
I
wasn't
there
for
the
early
development,
so
I've
just
gotten
familiar
with
what
is
here
and
don't
always
have
an
answer
for
why
it
was
built
that
way
in
the
first
place,.
C
A
Yeah
I
mean
we
do
have
a
wrapper
for
this
reader
facade
for
python,
for
example,
it
probably
we
could
probably
still
just
pass.
C
A
Just
what
are
the
functions
that
are
a
problem
here,
open
and
close
kind
of
go
away
as
a
problem?
If
we're
considering
this
to
be
resource
allocation
is
initialization,
do
it
in
the
constructor
and
destructor
damn
has
next
to
them.
Read
next
are
all
good
right.
A
C
Gone
from
there,
the
aggregate
reader
won't
support,
get
metadata
or
the
set
filter,
reset
filter
and
get
implementation
handle
okay.
So
I
get
implementation
handler
will
be
fine
because
it
would
return
the
aggregate
reader,
which
is
okay,
because
there's
only
one
aggregate
reader
but
the
filter
to
filter
methods
and
get
metadata,
wouldn't
work.
A
Yeah
so.
A
Yeah,
I
I
mean
I
almost
think
we
could
implement
this
set
filter,
reset
filter
and
have
it
just
iterate
over
its
internal
readers
as
necessary.
C
C
A
A
A
writer
facade,
so
both
of
these
right
now
are
as
close
as
you
get
to
the
core
rosberg
2
interface,
so
I'm
hesitant
to
delete
them
entirely.
Yeah.
A
An
api
yeah,
it's
a
major
change
in
the
api,
that's
not
necessarily
something
we
have
to
avoid,
but
it's
something
I
just
want
to
feel
cautious
about
before
we
make
any
decision
like
that.
Yeah.
B
C
B
A
I
mean
you
could
just
create
a
mock
interface
that
inherits
from
the
interface
that
you're
testing
right.
C
A
It
it
might
very
well
be
yes,
which
is
another
part
of
why
it's
maybe
better,
to
find
a
an
alternative
to
deleting
this.
I
I
mean
I'm
wondering
if
it's
necessary,
like
I,
maybe,
instead
of
the
full
drastic
change
of
deleting
it.
Maybe
all
we
really
need,
let's
see
so
we're
in
the
reader
again
next
has
next.
A
A
If
we
modified
those
two
which
would
which
make
less
sense,
given
the
possibility
of
reading
multiple
bags,
we
don't
have
to
delete
this
interface.
Most
of
the
testing
infrastructure
doesn't
have
to
change
the
core
api
stays
the
same,
but
we've
just
removed
two
of
the
functions
within
it.
I
feel
like
that's
how.
A
A
notch
sure
just
talking
through
it,
because
right
now,
I
don't
see
it
actually
used
anywhere,
except
in
tests.
A
Yeah,
I
mean,
I
think
it
does
make
sense
to
have
some
understanding
of
an
aggregated
metadata.
What
problem
did
we
run
into
with
that
jeff
of
how
it
would
be
difficult
to
aggregate
metadata
across
the
multiple
bags,
or
was
there
any
logical
inconsistency,
or
was
it
just
slightly
tricky.
B
A
Let's
see,
storage
identifier
was
maybe
a
problem
when
you
have
multiple
different
storage
inputs.
Relative
file
paths
is
kind
of
okay,
although
ordering
is
difficult.
Total
duration
is
computable
starting
time.
Total
message
count
topic
information
yeah.
So
there
are
a
few
of
these
that
are.
A
Yes
think
that,
like
these,
these
ones
at
the
end
here
are
the
ones
that
are
easy
to
aggregate
between
multiple.
If
effects
this
one,
you
could
just
list
all
of
the
relative
file
paths
from
all
of
the
bags,
but
the
ordering
is
maybe
weird
it
might
not
have
meaning
given
that,
but
maybe
it's
okay
in
that
case,
so
I'm
just
gonna,
say
medium
and
then
yeah.
We
got
easy
medium
and
these
ones
are
really
specific
to
a
single
bag.
A
But
we
could
set
them
to
some
null
value,
you
could
possibly
say
like
equals,
I
don't
know
an
a
or
unknown
and
then
the
output
bag
will
have
a
single
or,
if
you're,
writing
to
a
bag.
As
part
of
this
feature,
you
would
have
a
single
identifier
that
would
go
into
the
metadata
for
the
output,
but
the
input
would
have.
Oh,
you
could
even
just
say
I
don't
know
mixed
and
have
that
be
a
special
case.
I
guess
for
all
of
these.
B
A
I
don't
see
it
being
a
show
stopper,
but
it
is,
does
require
a
decision
to
be
made
about
the
logic.
There's
not
a
not
a
clear
answer.
A
I
was
wondering
yeah
beck:
where
did
you
lose
us?
Do
you
remember.
C
About
a
couple
of
minutes
ago,
I
was
just
saying
that
I
think
that
one
one
possible
approach
is
to
get
rid
of
the
assume
that
reader
implementations
should
be
configured
directly
rather
than
via
the
reader
facade.
C
A
A
So
I
yeah
jeff.
I
really
don't
think
we
have
to
get
rid
of
all
of
these
these
apis
here,
because
it
is
possible
to
aggregate
the
metadata,
at
least
in
some.
So
that
was
the
conversation
we
were
having
while
you're
offline.
C
A
Forget
metadata-
and
I
was
taking
a
look
at
the
metadata
here
where
some
of
this
is
really
easy
to
aggregate
this
relative
file
paths.
We
can
just
give
them
a
list
of
all
of
the
file
paths
between
things
yeah
and
then
we
could
possibly
for
these
ones
just
have
a
value.
That's
not
meaningful!
That
is
a
placeholder.
B
A
A
Yeah,
because
those
are
the
fields
that
that
don't
make
sense.
C
A
You've
got
multiple
bags,
but
it
could
yeah
just
be
a
placeholder
value
there.
The
the
one
yeah
even
then
set
filter
reset
filter.
These
could
be
global
overrides,
which
then
you
just
have
to
be
careful
to
you
know
not
even
be
careful.
Just
don't
call
that
if
you
want
to
have
separate
filters
on
the
different
writers
that
you've
configured.
A
A
You
would
have
to
deconstruct
and
reconstruct
the
object
to
replay
that
individual
bag
when,
when
we're
looping,
for
example,
yeah
and
right
now
that
wouldn't
be
possible,
be
within
this
or
within
the
player.
I
guess
because,
as
we
call
open
at
the
end
of
the
loop
or
close
at
the
end
of
the
loop
and
then
let's
see,
could.
A
You
I
don't
know
if
that
solves
the
problem,
that
we
we
need
to
open
and
close,
so
that
we
can
open
it
at
the
beginning
of
the
loop
and
then
reopen
it
to
loop
to
blue
play.
It
play
back
again.
Where
is
it
so.
B
A
Yeah,
michael,
I
think
that
you
don't
want
to
do
that
for
jump.
Did
you
see
my
comment
on
the.
B
A
Yeah
but
realistically
we
don't
want
to
be,
you
know,
closing
and
then
reopening
the
bag
every
time
that
we
want
to
change
the
the
timing,
because
that'll,
I
assume,
be
very
slow
as
an
implementation.
A
A
I
think
that
makes
sense
not
for
sake
you,
okay,
there,
you
loosen
your
connection
again.
A
You
still
there
jeff,
no,
yes,
not
well
yeah
I
mean,
I
think
reopen,
would
still
do
the
thing
that
you
needed
to
do
for
the
the
current
naive
jump
in
the
implementation
right,
michael.
A
Yeah-
and
I
assume
this
would
probably
go
in
after
your
feature,
so
you
could
do
that
first
and
then
we
can
refactor
as
necessary
and
if
reopen
is,
does
the
same
thing
as
clothes
open.
Doesn't
it.
B
Just
looks
a
little
bit
odd
when
you
see
and
try
in
for
loop
like
here
and
do
and
the
first
time
you
see
it.
It's
like
you
open,
okay,
why
is
it
open?
We
haven't
opened
it
yet,
and
it's
already
opened
it's
a
little
bit
of
confusion,
but
it's
okay.
B
Yeah,
it's
more
generic
could
have
said.
We
already
probably
failed
what.
C
Yeah
I
dropped
off
again
yeah
I
switched
to
my
cell
phone
enough,
maybe
a
bit
more
stable.
That's
also
my
house.
His
wi-fi
is
broken.
I
was
saying
that
we
could
delete
open
and
close
and
replace
him
with
a
reopen
method.
C
A
And
I
think
that's
okay,
as
long
as
we
document
carefully
that
construction
is
opening
and
so
can
be
called
after
construction,
and
it
is
logically
consistent.
B
Yeah,
can
you
maybe
highlight
again
rationals
to
remove
clause
what
it
will
give
us
if
we
remove
clause,
what
advantages.
A
A
Well,
you
have
to
call
you
have
to
call
open
in
between
and
right
now
it's
you
know
it's
it's
a
configuration.
A
A
That
that's
the
main
concern
it's
just
the
split
between
the
yeah,
because
the
reader
implementation-
I
believe
you
go
to
this-
actually
is
a
little
bit
weird
that
you
have.
A
C
C
There
are
two
open:
if
you
look
at
the
reader
here,
there's
two
open
functions,
one
that
just
takes
the
url
and
produces
the
storage
options
itself.
C
A
But
we
don't
have
any,
I
don't
think
we
have
documentation
on
that,
but
we
added
converter
storage
options
later
and
this
one
was
the
original
api
and
we
probably
should
have
said,
like
you
know,
just
flash
deprecated.
However,
you
marked
that
here
yeah.
A
So
I
think
that's
fine
as
an
approach.
You
know
you
reopen,
in
which
case
you
know
this.
This
reader
facade
could
just
have
reopen.
No
clothes
has
next
read.
Next,
get
metadata,
get
topics
and
types,
I
think
even
the
set
filter's
fine
as
a
you
know,
as
a
first
step.
If
we
want
to
reconsider
it,
that's
okay
like
it
would.
C
A
Would
prefer
to
take
the
slightly
more
incremental
approach
here?
We
do
have
almost
a
year
until
humble,
so
we
have
time
to
use
the
rolling
version
and
and
see
whether
we
like
or
dislike
that
the
one
thing
is
get
implementation
handle,
I'm
not
entirely
sure
about
how
to
replace
that
right.
Now,.
A
I'd
see
so
that
yeah,
that
would
give
you
what
you
need.
Then.
C
Yeah
and
then,
if
you
want
to,
if
you
want
to
get
the
child
readers,
the
aggregate
reader,
the
reader
class,
would
have
its
own
method,
which
doesn't
come
from
any
other
interface.
So
obviously
we're
working
with
that
directly.
C
A
Great
that
sounds
awesome.
I'm
happy
we're
making
progress
on
this.
This
is
gonna,
be
really
useful
for
a
big
feature
set
yeah
michael.
Do
you
need
any?
This
is
just
me
now
thinking
about
major
features
on
rosberg
too
michael.
Do
you
need
any
reviews
right
now
that
I
haven't
been
doing.
B
A
B
A
And
then
I
have
a
member
of
my
team
who
has
been
in
this
meeting
before
but
isn't
here
today.
Sonia
is
working
on
the
the
seek
api
for
storage
implementations
and
I
think
she
has
an
implementation
working,
but
it
still
needs
testing
so
we'll.
I
think
see
that
next
week,
okay
as
a
review
and
that
should
be
really
useful
as
a
new
api
for
storage.
B
A
Yeah,
I
think
we
can
go.
I
I
I
think
I'd
like
to
make
progress.
We
could
merge
with
what
you've
got
if
it
works,
and
then
we
can
replace
it
once
we
have
something
to
replace
that
that
portion
of
the
logic
with
and
then
I'm
working
on
right
now
the
sleep
for
so
I'm
working
on
sim
simulated
time
use
sim
time.
So
this
is.
A
A
Okay,
so
I'm
working
on
that
prerequisite
logic-
and
this
is
a
lot
of
this-
these
three
things
here
get
us
towards
that
that
advanced,
playback
control
story
that
we
defined
before
galactic,
and
there
were
just
a
few
last
stories
that
didn't
get
finished
and
it's
because
they
were
the
much
more
complex
ones.
A
So
I'm
glad
we're
making
that
progress.
I
think
we
will
have
that.
You
know
long
before
the
humble
release
and
we
can
have
some
testing
on
it
and
have
that
feature
set
fully
functional
and
then
another
thing
that's
going
on
that
I
wanted
to
bring
up,
is
in
the
middleware
working
group
meeting
on
wednesday
morning
or
on
wednesday.
Whatever
time
zone
you're
in,
we
talked
about
being
able
to
get
message-
definitions
on
the
wire
for
for
us
too,
and
here
we
want
to
use
those
to
store
them
in
the
bags.
A
That's
been
an
ongoing
conversation,
so
we
have
an
initial
conversation
in
in
the
roster
core
about
that,
but
we
don't
have
any
implementation
yet.
A
So
I'm
thinking
about
doing
some
prototypes
or
some
design
docs
around
that
don't
have
any
anything
concrete
to
show
yet,
but
just
wanted
to
share
that.
We
took
some
notes
from
the
middleware
group.
B
Yeah
but
as
far
as
understand,
it's
two
different
conceptual
approach,
one
of
them
to
rely
to
x
types
metadata
and
built-in
dds
information,
another
one
which
is
trying
to
add
one
more
service
which
will
provide
this
information
from
not.
A
A
I
think
we
need
to
maybe
prototype
both
approaches
and
see
which
one
works
better
yeah.
I
don't
really
know
because
right
now,
yeah,
the
x
types
will
give
you
the
idl
information,
which
is
a
it's
a
direct
mapping
from
the
dot
message
to
idl.
But
I
don't
know
if
you
can
reverse
that
transformation,
it
might
be
lossy.
So
so
that
might
not
be
good
enough,
but
maybe
it
is
I'm
not
sure.
C
A
Possibly
we
could
do
if
it's
easier.
We
could
do
a
an
implementation
that
only
supports
dds
based
implementations
and
that
would
at
least
be
helpful
for
a
lot
of
users,
but
I
agree:
it'd
be
nicer
to
get
the
dot
message.
Definition,
we
might
be
able
to
pass
that
as
user
data
on
the
discovery
traffic
yeah
it
just.
B
A
Investigation
anybody
who's
interested,
I'm
gonna,
start
a
thread
on
the
web
tools.
Working
group
mailing
list
about
it
and
adrian
had
asked
you
sent
me
an
email,
and
I
forgot
to
respond
to
it
yesterday
whether
there
was
a
mailing
list
for
this
working
group
and
I
haven't
set
open
up
properly
all
right,
we
hadn't
really
needed
it,
or
at
least
maybe
we
did
need
it
and
we
just
didn't.
B
Was
more
than
just
discussion,
I
was
mostly
looking
for
an
easy
way
to
subscribe
to
just
the
calendar
environment.
For
this
thing,
and
not
all
of
the
roster
working
group.
A
Yeah,
my
groups,
what.
C
A
C
Invite
work
the
google
group
for
getting
invites
to
the
the
working
group
account
events,
and
you
just
need
to
make
sure
that
you
put
include
that
invite
that
google
group
address
and
any
calendar
events.
B
C
A
Right
now,
okay,
right
now,
it's
only
it's
only
set
to
invite
ross
to
working
groups.
So
let
me
add
this
new
guest
and
then
save
for
this
and
following
events,
okay
and
send,
I
think
that
should
solve
the
problem.
Okay,
if
that
was
the
thing
adrian,
then
I
think
you're
solved
now
you
should
have
gotten
a
new
invite
to
the
ross.
Tooling
working
group
invites
one.
C
I
think
we
mentioned
yesterday
but
john
on
our
team
is
probably
willing
to
help
out
there
too,
but
he's
just
out
this
week.
So
if
you
start
a
thread
somewhere.
A
It
out
to
the
the
web
tools
working
group
email
address-
I
didn't
get
around
to
it
yesterday,
but.
B
A
So
I'll
send
that
out
right
now
after
this
meeting
that
way
it's
out
and
we
can
coordinate
whoever
wants
to
help,
it
probably
doesn't
hurt
to
just
have
that
conversation
out
in
front
of
everybody.
They
can
always
filter
our
emails
if
they
want
right.
A
Yeah
great,
oh
fam,
we've
even
got
a
new
one
conversation
from
oh
from
that
link
last
week.
I
guess
I
missed
this
okay
yeah,
so
I'm
gonna
there
we
go
new
conversation.
I
have
it
up
in
front
of
me.
A
It
will
coordinate
on
there
and
anybody
who's
on
this
call
who
wants
to
get
involved?
Maybe
I
don't
know
probably
joining
that
mailing
list
is
overkill,
but
let
me
know,
and
then
we
can
start
a
side
thread
or
something
or
I
can
include
you
just
on
that
thread
other
than
that.
Is
there
anything
that
we
should
talk
about
before
we
get
going?
Are
there
any
major
feature?
Sets
that
we're
not
discussing
right
now,
we've
got
this
portion
is
talking
about
the
advanced
playback
api
we've
got.
A
A
I
think
that
there's
a
conversation
to
be
had
about
compatibility
like
forward
compatibility
of
ros
bags
figuring
out
a
way
to
ensure
that
bags
we
write
in
say
humble-
are
playable
in
foxy,
which
was
a
feature
of
ross
one
that
I've
been
informed
was
useful
for
people.
I
I
personally
didn't
have
that
experience,
because
I
was
mostly
using
internal
bags,
but
I'm
not
sure
how
to
ensure
that,
but
that's
something
maybe
we
can
just
have
on
our
minds
at
the
moment.
I
think
we
have
enough
other
things
going
on
to
pay
attention
to.
B
A
Basically,
it
covers
some
number
of
the
definitions
of
conversion,
but
not
necessarily
all
of
them.
So
I
think
we
need
to
be
more
more
precise
in
that
issue.
Maybe
we
want
to
close
it
and
reopen
multiple
slightly
more
directed
issues,
because
what
this
it's
part
of
work
that
will,
let
us
do
splitting
storage
format,
changing
merging
compression
format,
changing
yeah,
it's
basically
the.
A
Yes,
exactly
there
are
possibly
other
definitions
of
conversion
that
it
won't
cover,
but
I
think
we
just
need
to
be
more
specific
about
those.
I
can't
think
of
what
they
are
right
now.
A
All
right
yeah,
I
mean
that
was
super
useful
in
helping
us
start
the
conversation
and
clarify
a
lot
of
the
ideas
around
it.
A
That's
a
good
start,
let's
see
so
like
merge.
A
So
that's
this
is
directed
at
you
jeff,
if,
if
you,
if
and
when
you
have
a
pr
it
actually,
it
only
will
do
this
once.
A
Yeah,
so
both
of
these
issues
are
notable
for
this
functionality,
so
this
is
part
of
this
is
why
I
say
like
a
part
of
some
of
the
definitions
of
next.
A
So
we've
got
668
for
merging.
Splitting
is
5.99.
A
And
then
we've
got
seven,
do
you
say
727.
A
Things:
okay,
so
this
converter
then.
A
Yeah
it'll
cover
this.
This
portion
isn't
really
something
we
need
to
do
specifically
or
it
already
is
that
into
any
valid
respect
to
format
yeah.
So
this
doesn't
do
exactly
all
of
these
things,
but
it'll
do
a
lot
of
it.
So
maybe
we
need
to.
B
A
C
C
A
Yeah,
I
think
I'm
gonna
make
a
new
issue
for
the
generic
bag
rewriter.
Unless
we
want
to.
I
think
it
might
make
sense
to
just
yeah
note
this
oops.
A
But
start
a
new
conversation
because
that's
a
little
overloaded
generic
bag
bags
to
bags
rewriter.
A
And
by
far,
I
and
also
believed.
A
Outputs,
possibly
different
prospects,
using
any
storage
options,
per
image,
etc,
options
for
all
outlets
and
outputs,
that's
something.
A
This
is
the
larger
project
I
think
so
related
issues,
I'm
going
to
say
this
is
related
to
717,
covers
much
of
the
functionality
that's
best.
When
it
doesn't
it's
done,
then
we've
got
joining
our
stitching.
A
At
least
we'll
have
this
here,
I
this
is
less
complete
than
it
could
be,
but
at
the
moment
I
don't
know
if
I
can
assign
you
in
the
org
I'm
assigning
hf
for
now.
Okay,.
A
A
Okay
and
if
yeah,
if
you
need,
I
think
we
will
definitely
want
to
just
leave
it
to
you
for
getting
this
first
portion
out
yeah.
But
then
we
could
pull
in
more
people
to
do
some
of
the
follow-up
issues
as
we
get
there.
C
C
A
Yeah,
I
think,
we're
in
a
good
spot
we're
making
progress.
One
note
I
wanted
to
make
on
that
is.
A
I
would
very
much
appreciate
it
if
you
could
add
doc
strings
to
all
apis
that
you
create
at
least
all
the
the
public
ones.
I'm
gonna
start
running
the
the
documentation
tool
on
rosbeg2
and
see
where
we're
missing
documentation.
Definitely
some
of
the
interfaces
are
are
very
underdocumented,
which
makes
it
unclear
why
they
exist
or
what
how
exactly
they're
meant
to
be
used.
A
B
I
mentioned
I
I
would
like
to
raise.
Maybe
a
question:
can
we
maybe
somehow
agree
to
what
style
to
use
when
documenting,
like
c
or
c
style,
using
the
oxygen
documents
documentation?
Because
I
see
a
proposal
with,
I
would
promote
proposal
to
uc
plus
for
new
descriptions
for
new
files.
If
it's
not
documented,
if
it's
already
a
lot
of
documentation
happened
in
c
style
to
preserve
the
c
style.
But
if
it's
entirely,
you.
B
B
Yeah,
it's
very
undesirable
to
have
it
really
because
it's
it's
it's
better
to
to
sustain
and
support.
Cpas
plaster
is
much
easier,
but
if
your
file
readies
have
this
most
of
them
just
keep
it
in
this
file.
But
if
it's
completely,
you
try
to
use
c
plus
class
really
what
is
the
c
plus
plus
style?
It's
just
three
backslashes.
B
A
I
thought
three
backslashes
was
just
the
the
bleeding.
B
B
A
I
don't
have
a
strong
opinion
about
this.
I
definitely
think
we
should
make
it
consistent
within
about
file,
but
if
we
want
to
do
that.
B
In
one
file,
but
if
it's
it's
new
well,
I
would
kind
of
encourage
people
to
use
this
new
series.
Lashes
style.
A
Okay,
I
yeah,
I
guess
it's,
why
is
it
better?
I
don't
actually
know,
I
don't
see
it.
B
B
By
the
way,
if
anybody
use
client,
if
you
will
just
type
three
slashes
and
enter
it,
will
actually
pop
up
snippet
for
this
comment
automatically,
you
don't
need
to
write
even
brief
details,
so
it
will
return.
It
will
automatically
insert
all
this
stuff
boilerplate
stuff
great,
it's
easy,
just
three
slashes
and
enter.
B
A
You,
okay,
we're
a
little
over
time
anything
else
we
need
to
talk
about
before
we.
A
Go
yes,
I
can.
Okay,
that's
great.
A
Okay,
thanks
everyone.
I
will
see
you
again
the
next
time.
We
don't
need
a
breakup
meeting
next
week.
Right,
I
don't
think
so,
and
then
the
next
one
yeah
the
following
meeting
would
be
on
the
the
13th.
A
You
know
what
I'm
not
going.
Oh,
no!
No!
That's!
Okay!
I'll,
be
here!
I'm
gone
the
following
friday,
so
we're
all
set
next
meeting
on
august,
13.
and
I'll
see
you
all
then.