►
From YouTube: OpenActive W3C Community Group Meeting / 2019-11-25
Description
Entities and References. Agenda for the call and notes can be found at https://w3c.openactive.io/meetings/2019-09-25-references-vs-objects.
A
A
Okay
right
so
Chris
just
prior
to
who's,
starting
recording
a
call
you
mentioned,
maybe
for
fronting
the
events
and
roots
discussion
a
little
bit,
because
that's
mostly
your
interest,
which
ties
in
slightly
into
well
ties
in
directly
into
whether
or
not
or
to
what
extent
we
want
to
be
able
to
refer
to
entities
by
URL
rather
than
having
them
expanded
and
fully
represented
in
data
feeds.
Basically,
so
shall
we
start
from
there
Chris
yeah?
Let's
do
that
right
in
China.
So
just
have
you
got
the
URL
for
the
actual
issue.
A
A
A
It
seems
from
modeling
it
given
that
basic
fact
of
the
case
that
you'd
want
to
be
able
to
just
simply
cross
refer
to
have
a
separate
feed,
essentially
of
routes
and
then
cross
refer
from
events
to
routes.
So
you
could
say
you
know,
we've
got
a
5k
run
or
something
starting
in
such-and-such
a
time,
URL
whatever,
and
then
that
URL
points
you
to
to
the
route
itself
ons
from
some
other
source,
such
as
an
RPD
feed,
but
it
could
just
be
separately
published.
A
B
Keeping
feeds
fairly
object,
coherent,
I,
suppose
so
that
you
don't
have
nested
nested
different
objects,
no
beep,
so,
generally
speaking,
light
and
I'm,
not
on
a
unless
other
factors
mean
that
it's
much
better
idea
to
do
it
a
different
way.
I,
quite
like
the
idea
of
having
you
are
ice,
basically
pointing
out
merging
the
whole
thing
together.
Most
crawlers
are
perfectly
capable
their
API
clients
are
perfectly
capable
of
themselves
going
and
pulling
the
expediter,
and
you
lose
a
lot
of
concern
about
data
synchronization
know
if,
if
your
data
feed
that
contains
the
route
metadata.
B
A
A
B
A
The
other
aspect
where
urls
become
useful
with
route
metadata,
is,
if
you've
got
root
segments,
in
which
case
you
you're
saying
you
know
the
route
consists
of
you
know
these
five
segments,
and
so
one
is
a
parent
and
the
other.
Basically
there's
nothing
saying
that
you
can't
nurse
that
unless
you
have
alternative
route
segments,
so
if
you've
got
five
segments,
one
of
which
is
a
shortcut
between
two
of
the
other
points.
A
A
B
B
You
could
either
somehow
model
it's
a
segment.
1
then
choice
of
two
or
three
and
to
listas
four
and
then
both
forum
to
lead
to
five
or
you
could
potentially
unravel
it.
So
you've
basically
got
two
alternatives
for
the
entire
route:
1
2,
3,
5,
I'm,
sorry,
what
1
2
4
5,
whatever
I
said
right
now
you
get
the
idea
basic
racing.
B
A
B
Or
rather
yeah
ray
of
arrays,
it
is
an
array
of
arrays,
yeah
and
I.
Just
wonder
whether
we're
getting
to
you
in
the
weeds,
but
if
I
think
of
path
in
the
Peak
District,
for
instance,
you
can
get
from
one
station
stop
to
the
next
by
a
whole
set,
which
is
a
nice
way
to
do
a
walk,
because
you
know
one
way:
walk
and
catch
the
train
back
from
the
second
station.
B
A
Yeah
I
mean
I
think
in
the
use
case,
we
most
frequently
encounter
it's
quite
it's
quite
curated
as
an
int.
So
even
if
there
were,
you
know,
20
different
ways
of
completing
some
kind
of
circuit.
In
fact,
there
will
be
maybe
two
or
three
that
are
recommended,
and
then
the
existence
of
the
others
is
not
even
really
noted,
but
I
guess
for
race
fully.
You
do
want
a
routing
algorithm
there.
A
B
A
A
Yeah,
because
that's
a
much
more
complex
problem,
isn't
it
Nick
and
I
we're
talking
about
what
it
is
that
root,
mated
metadata,
actually
actually
models,
and
we
more
or
less
came
to
the
conclusion
that
what
it's
really
modeling
is
a
kind
of
guidebook.
You
know
of
the
kind
that
you
would
get
from
the
National
Trust
or
English
heritage,
or
something
like
that
Sanger.
You
know
stop
here.
You
can
see
this
and
then
you
can
take
one
of
these
two
paths
and
one
is
longer
and
nicer.
One
is
shorter
and
less
attractive.
B
A
B
A
B
A
B
So
if
you
have
a
URI
for
the
event,
then
the
segment's
within
us
are
basically
a
json
dot.
Notated
passed
down
the
structure
to
the
thing
that's
down
there,
so
you've
done
we'd
have
kind
of
a
uri,
plus,
probably
some
sort
of
hash
mark
type
construct
that
says
within
this
document,
go
to
this
place
just
just
again
thinking
in
terms
of
if
we've
got
code,
that's
expecting
things
to
have
united,
unique
ids.
B
A
A
B
A
It
sounds
a
much
simpler
that
way,
but
to
generalize
this
on
the
basis
that
somebody
somewhere
right
now,
the
open
data
Institute
is
going
to
be
running
validators
and
that
kind
of
thing
that
will
be
having
to
check
to
make
sure
that
the
data
is
actually
conforming
to
the
specification
and
so
on
and
so
forth.
So
if
it's
just
a
question
of
following
a
link
and
dereferencing
it
for
one
particular
user
agent,
that's
fairly
easy!
Most
of
the
time.
A
However,
if
we're
trying
to
do
this,
it's
scaled,
so
we're
trying
to
validate,
say
in
our
PDE
feed
or
our
PDE
feeds
from
thousands
of
publishers
all
at
the
same
time,
branching
out
in
a
linky
kind
of
way,
slows
that
whole
process
down
a
lot.
So
now
you're
adding
a
lot
of
latency
with
everything
that
you
have
to
check
and
every
time
you
have
to
follow
when
you
link
basically
as
opposed
to
just
sucking
in
the
the
objects
and
checking
them
locally.
B
B
B
So
III
don't
think
we
were
typically
mind
very
much
whether
we
were
because
we
are
not
indexing
or
validating
we're,
simply
consuming
a
fairly
limited
universe
of
instances
at
that
time
doesn't
really
matter
to
us
whether
we
need
to
make
multiple
calls
or
not.
I
can
see
that
there's
a
certain
year,
latency
improvements
without
making
many
calls
to
get
these
documents
all
pretty
readily
compressible
as
well,
which
means
that
you've
got
all
the
games
on
there
in
terms
of
throughput.
A
Okay,
yeah
I,
think
I.
Think
from
our
perspective,
it's
really
been
it's
almost
a
theoretical
purity
kind
of
thing
that
once
we
start
saying
well,
you
know
in
some
cases
it's
useful
to
have
a
URL
reference.
You
start
thinking
what
you
know
is
this
going
to
be
a
gotcha
for
data
publishers
where
some
things
are
you
or
also
the
ones
can
be
objects?
Bla
bla,
bla,
bla
bla,
you
start
thinking.
Oh,
should
we
just
generalize
this
you
know.
Can
we
make
everything.
B
The
there's
a
danger,
if
you
I,
think
I
matched
my
original
comment
about
sort
of
heterogeneous
documents.
If
you
like
is
typically,
publishers
will
be
pulling
data
from
an
events
table
and
from
allocations
table
and
from
a
user's
table
them
whatever.
You
know,
you
sort
of
see
how
these
these
these
documents
are
structured
and
it
does
take
for
them
to
construct.
The
flat
document
is
a.
B
B
Ultimately,
that's
still
probably
X
number
of
queries
for
constructing
the
one
document
that
you
then
cache
after
that
you're
saving
yourself,
because
they're
not
coming
back
again
to
find
those
documents
later
when
the
URL.
So
again,
it's
probably
actually
a
net
saving
to
these
documents.
A
bit
broader
right,
okay,
yeah.
B
A
A
A
So
I
attached
kind
of
an
issues
list
to
the
rest
of
the
agenda,
which
is
just
stuff
to
do
with
the
the
opportunity.
Spec
mostly,
was
there
anything
that
jumped
out
to
you
as
being
a
particular
interest
to
race
fully,
because
if
so,
obviously
we
can
talk
about
that
it,
but
if
not
I'd
be
happy
just
to
field
questions
concerns
observations,
anything
anything
you've
bought
from
race,
Felice
point
of
view.
There.
A
B
You
can
on
a
case-by-case
basis.
You
can,
it
can
be
fairly
obvious,
but
maybe
there
needs
to
be
some
sort
of
reserved
words
or
reserved
property
or
something
that
says.
Oh
I've
got
one
of
these,
then
you
know
I
need
to
go
down
section
powers
route
rather
than
it's
just
here,
route,
easy
determination
to
make,
rather
than
looking
inside
the
object
and
going
Oh
have
I
got
one
of
these
in
there
does
that
mean
that's
missing,
or
does
that
mean
I
need
to
go
and
fetch
it
there's
something
else
to
look
all
right.
Yeah
yeah.
A
B
Kind
of
null
is
this:
where
there
is
a
an
either
or
where
you
can
have
a
an
event
that
has
an
embedded
root
or
an
event
that
has
a
referred
root
either
you
can
give
them
completely
different
names
as
property
names,
or
you
can
yeah
just
just
to
think
through
something.
That
means
it's
very
clear
and
hopefully
fairly
consistent
wherever
you
do
something
like
this,
that
when
it's
got
a
property
called,
this
kind
of
thing
is
going
to
be
referring
absolute
another
another
document,
the
you
see
fetching
this
place
right,
oh
yeah,.
A
That's
a
that's
a
good
point:
yeah
I
can
see
yeah
indeed,
because
most
of
the
yes
most
of
the
the
properties
that
you'd
use
you
know,
ID
or
URL
could
either
be
nice
information
to
have
about
about
the
the
object
or
just
simply
any
identifier,
yeah,
there's
nothing
there
to
twig
to
you
that
that
this
is
something
that
needs
to
be
followed
in
order
to
resolve
the
the
object.
Yeah
yeah,
some
kind
of
some
kind
of
flag,
all
right
yeah,
as
you
say,
a
naming
convention
like
right
now,
it's
geo
path.
A
A
B
A
B
A
It's
funny
I,
do
it's
a
frequently
articulated
principle
in
sort
of
the
linked
data
world
or
going
back
the
Semantic
Web
world
that
any
you
know
URL
tree
interchangeable
with
with
an
object
with
the
appropriate
time
but
yeah.
How
exactly
are
supposed
to
achieve
that
is
not
is
not
specified
so
yeah.
Indeed
it
must.
It
must
be
a
frequently
encountered
issue.
B
A
B
A
Data
set
I
think
I
think
it's
because
people
are
owned
faster
to
recognize
the
issue.
Actually,
this
has
happened
with
Nick
and
I
a
lot
talking
about
these
things.
You
know
sort
of
hold
on
we're
talking
about
the
object,
we're
talking
about
the
metadata
about
the
object,
kind
of
conversations,
but
because
everybody
who's
ever
had
anything
to
do
with
RDF
is
has
had
that
conversation
so
many
times
it
tends
to
resolve
itself
early.
A
A
A
A
Because
you
might
have
a
once
slightly
incommensurate
times,
basically
associated
you
know
all
together
as
in
us
in
collected
data
feeds
and
it
gets.
It
gets
fairly
intricate,
as
you
can
see
from
the
length
of
the
chain,
but
I
think
I
think
the
end
conclusion
was
you
do
sort
of
have
to
specify
time
zone
in
a
fairly
lightweight
way
at
some
point
in
the
data
publishing
chain,
because
otherwise
you
broad
term
anarchy,
basically.
B
A
B
B
How
we
calculate
their
weekly
totals
and
things
like
that,
but
we
have
had
a
few
support
requests
over
the
years
which
say:
oh,
we
didn't
take
into
account
this.
That
doesn't
help
also
gracefully
that
we
don't
really
know
what
your
prefer
preference
is.
Your
location
may
have
nothing
to
do
with
how
you
prefer
to
consume
your
dates
and
times.
B
Important
people
helps
with
their
calendars
and
that's
why
you,
who
kind
of
just
says
what
time
zones
changed?
Would
you
like
to
go
to
it
or
sticking
the
one
that
you're
in
before,
because
some
people
would
like
to
keep
what
we're
doing
in
time
that
they're
attached
to
and
other
people
like
it
remap
to
wherever
they
happen
to
be
that's
not
relevant
here,
but
it
is
just
give
there's
so
many
oddities
in
this
I
guess
I
can
see.
A
B
B
Yes,
my
concern
is
that
you're
potentially
offloading
interpreting
time
zones
onto
hundreds
of
people's
implementations
roads,
because
you
found
a
way
to
model.
Oh
well.
If
they
said
3
o'clock
in
the
afternoon
then-
and
it's
in
London-
then
what
means
that
well,
then
someone
else
has
to
go
and
work
out.
What
we
have
talk
in
the
afternoon
in
London
is
in
whatever
time
zone.
It
is
that
you're
converting
the
display
to,
and
that
potentially
is
a
hard
problem
to
give
them
all
the
complexities
of
time
zones.
Mm-Hmm
I
think
if
it
were
possible
to.
B
However,
a
whatever
it
is,
they
isolate
six
a1a
Thailand
timezones
representation
there.
This
wasn't
in
concrete
if
you've
got
that,
then
we'll
take
that
as
a
consumer
of
this
thing,
if
you
don't
have
that,
but
you
have
got
information
about
where
it
is,
then
we'll
do
a
best
guess,
based
on
Nick's
comment
coming
to
you
that
you
know
these
are
standard
standard
timezone
buckets
that
a
lot
of
systems
are
used
to
getting
people
to
choose
amongst.
B
B
A
A
A
B
A
A
B
Than
me,
retyping
yeah
I'll
put
that
in
and
it's
possible
that
Nick
or
someone
will
come
back.
I
can
see
a
counter
to
that.
That
says,
there's
a
danger
that
people
people's
software
will
accidentally
put
zeros
on
the
end
in
an
unspecified
time
zone
and
then
we'll
take
that
as
UTC.
Even
though,
actually
it's
not
specified
I,
don't
I'm,
not
sure
whether
I
was
actually
seeing
that,
but
some
I
can
see
how
he
might
accidentally.
A
B
So
I
put
it
up
there,
since
this
don't
see
a
specific
dimensions
but
and-
and
it
there's
no
doubt
that
that
representation
makes
it
easy
to
say.
But
we
really
do
mean
this
time,
rather
than
we
mean
what
you
interpret
to
be
this
time,
but
I
yeah.
It
is
an
interesting
question
and
whether
you
end
up
with
some
people's
publishing
systems,
putting
out
things
that
look
like
UTC
and
they
just
not
they're
local
and
they
just
haven't
really
bothered
with
the
conversion.
Yeah.
A
It
is
tricky
you
because
of
course,
Mumbai
most
of
our
publishers
will
never
have
to
deal
with
more
than
one
timezone
yeah
her
and
scale.
So
it's
not
surprising
that
they're
not
particularly
well
equipped
to
deal
with
it,
because
you
know
everything
everything
falls
under
GMT,
so
so
yeah,
the
the
risk
I
guess
that
you've
identified
is
is
reasonably
high,
but
you
know
that's
that
we
put
it
in
a
stack
of
me.
You
know,
but
appropriate
exclamation
points
in
the
warnings
around
it.
A
A
B
No
I
think
I
feel
like
we
had
a
pretty
good
shake
of
the
stick
at
the
session
and
then
in
the
in
the
calls
couple
of
calls
that
followed.
It
there's
nothing,
that's
sort
of
dangling
from
from
my
point
of
view
from
these
conversations,
I
thought
that
the
decision
to
distinguish
for
emotive
data
from
roots
was
a
good
idea.
B
That's
they
don't
tighten
up
a
lot
of
things
that
were
otherwise
feeling
very
repetitive
or
just
downright
odd
yeah,
no
I
think
it's
now,
I
think
my
my
observation,
general
is
I
really
want
to
see
some
real
data
in
this
spec
coming
out
of
some
people.
Who've
got
it
and
let's
just
see
what
it
feels
to
work
with
that
when
we
get
it
because
I
feel
like.
A
Okay,
sure
yeah
we've
done
a
little
bit
of
work,
which
was
discussed
on
the
previous
calls
and
where
I
just
I
just
mapped
some
national
trust
data
into
into
the
spec,
and
it
was
fine
I
guess
that
raises
two
questions
for
me,
which
is
if
I
were
to
take
that
exercise
from
a
couple
of
data
providers
and
mapped
out
some
example
sample
data.
Would
you
have
the
bandwidth
to
review
that
and
let
me
know
how
useful
that
actually
is
to
race
fully
yeah
I
can
take
a
look
at
that.
A
B
A
A
Great,
you
know
it's
very
much
and
it's
very
much
a
manual
process
right
now
that
I've
just
been
getting
little
bits
of
sample
data
and
sort
of
lining
it
up.
You
know
not
actually
pulling
live
out
of
the
out
of
the
database,
and
then
we
were
thinking
of
possibly
having
some
kind
of
hack
day
or
hackathon,
possibly
in
December
playing
around
with
roots
data.
A
A
Because
everyone's,
like
okay,
this
data,
we've
got
a
spec
for
it.
Let's
do
it
and
then
we
sort
of
hit
the
you
know
the
implementation
of
walls.
Okay,
we
need
to
find
a
developer,
so
it's
not
moving
as
quickly
as
we
hoped
but
they're.
You
know
there
is
movement,
so
hopefully
we'll
try
the
end
of
the
year.
We
can
start
actually
playing
with
it.