►
From YouTube: OpenActive W3C Community Call / 2021-01-13
Description
Data Set Site Specification
- Review of outstanding issues
- Matter arising
* Special opening hours specification
A
B
Yeah
sure
I'm
olly
sissman
digital
project
manager
at
london,
sport.
A
B
Hello,
nick
evans,
director
at
hyman.
C
Yep,
I'm
a
developer
at
prism,
developing
the
referral
uk.
A
Okay,
fantastic
and,
I
suspect,
dom
olly
and
I
will
I'll
have
more
to
talk
talk
more
to
say
about
social
prescribing
in
the
future.
However,
at
the
moment,
I'll
focus
on
today's
agenda
for
the
call,
let
me
just
start
sharing
my
screen
here.
A
So
most
of
the
agenda
for
the
call
is
focused
on
the
data
set
site
specification,
so
the
html
and
json
that
needs
to
be
present
on
the
basic
splash
page
that
you
encounter
when
first
spidering
for
open
active
data
and
there's
a
little
addendum
that
was
added
to
the
agenda
just
recently,
because
nick
flagged
up
some
conversations
that
were
happening
in
connection
with
the
mcr
active
project
regarding
the
complicated
matter
of
opening
hours,
particularly
in
coveted
realities,
we've
got
right
now,
so
that'll
be
at
the
end
of
the
discussion.
A
Now,
hopefully,
we
can
move
fairly
quickly
through
the
data
site
discussion.
It's
been
under
discussion
for
a
while
now
and
as
our
standards
go,
it's
one
of
the
more.
A
I
would
say
that
we
can
specify
things
it's
a
fairly
known
domain.
What
should
be
on
that
page.
So
it's
mostly
details
that
we're
discussing
rather
than
broad,
philosophical
differences.
I
thought
you
meant
less
dynamics
in
dull,
no
less
dynamic.
In
the
sense
it's
reassuringly
stable.
A
So
there's
three
issues
I
think
that
are
remaining
unresolved
in
connection
with
the
with
the
specification.
The
first
is
a
fairly
minor
one,
but
it's
the
guidance
concerning
human
readable
text,
because
the
data
set
site
obviously
has
human,
readable
content.
Most
of
the
business
work
is
done
by
the
json
objects
that
are
embedded
in
that
page,
but
there
is
something
there
for
people
to
read,
and
obviously
it
makes
sense
for
that
information
to
be
as
complete
as
possible.
A
There's
no
advantage
to
being
vague
in
the
human
readable
aspect,
but
the
question
is:
should
this
should
the
guidance
that's
given
concerning
what
should
be
in
that
content?
Should
that
be
required
as
a
status
or
recommended?
A
I
don't
think,
there's
personally,
I
don't
think
there's
any
problem
with
making
it
required
beyond
the
fact
that
it
was
enunciated
as
a
general
principle
that
required
fields
would
be
those
things
that
are
checked
by
validators,
that
if
you've
you've
got
something
that
you
can
definitely
put
a
green
tick
next
to
you
can
sensibly
make
that
required.
If
you
don't
have
that
capacity,
it
becomes
more
difficult
to
make
that
kind
of
strong
assertion.
A
Obviously,
for
human
readable
text:
that's
a
disproportionately
difficult
task
to
give
that
level
of
machine
intelligence
to
to
parse
that
out,
so
just
throw
it
open
to
the
floor.
Does
anybody
have
a
feeling
about
about
maybe
weakening
that
guidance
to
reflect
that,
or
is
human
readable
text
a
special
case,
and
we
don't
have
to
perhaps
be
quite
so
pedantic
about
that
about
that
difference?.
B
Yeah,
so
I
had
a
quick
read
of
the
the
open,
quick
read
of
the
open
id
spec
just
to
see
how
they
dealt
with
it,
because
there's
some
stuff
in
there
about
needing
to
have
certain
details
on
a
confirmation
screen
things
like
that.
It
seems
that
we
could
be.
B
We
could
reasonably
say
required
in
the
kind
of
sense
that
we
could
say
it's
required
that
you
have
human
human
readable,
like
a
human,
readable
form
of
something
that
includes
these
things,
not
that
those
things
will
ever
appear
in
a
validator,
as
you
say,
but
more
that
were
kind
of
really
describing
what
the
thing
is.
They're,
like
almost
non-functional
in
a
way,
I
suppose,
but
but
but
still
important
in
the
way
that
other
specs
have
you
know
the
consent
screen.
B
O
warf
obviously
needs
to
not
just
be
a
happy
face
for
the
with
the
green
button.
So
there
is
a
like
a
yeah
a
need
for
a
bit
more
detail.
There.
A
Yeah
I
mean,
I
wonder
if
the
difference,
though,
is
that
oauth,
presumably
that's
got
some
kind
of
legal
ramifications.
That's
responding
to
particular
legislative
demand
in
some
jurisdictions
anyway,
yeah
here
it's
about
it's
more
about
politeness
than
anything
else.
You
know
what
does
a
developer
make
of
it
when
they
come
to
your
site
and
what
is
what
is
the
sanction?
A
I
suppose,
if
somebody
fails
to
meet
that
requirement,
I
mean:
are
we
really
going
to
yank
them
off
the
status
page
if,
if
they
don't
have
some
information
present
in
the
human
readable
text,.
B
B
B
Do
we
end
up
with
a
usable
developer
experience
for
the
people
that
are
trying
to
find
the
information,
and
then
is
that
going
to
create
the
kind
of
supposed
open
ecosystem?
We
want
referencing
previous
conversations
we've
had
around
if
everyone
publishes
stuff
all
over
the
place
and
you
need
a
phd
in
open
active
to
figure
out
what
to
do
with
any
of
the
data,
because
it's
also
diverse
and
crazy,
then
yeah.
A
Yeah,
I
suppose
also
it's
hard
to
see
why
anybody
wouldn't
provide
that
information
in
a
sense
and
things
like
organization,
name,
licensing,
information
that
kind
of
stuff.
Actually,
maybe
licensing
information
is
where,
where
the
rubber
meets
the
road,
because
that
is
legally
actionable,
that's
that's
the
bit
actually,
where.
C
A
If
somebody
relabeled
the
data
as
their
own
or
violated
the
license
terms
like
that,
that's
actually
the
bit
that
there's
a
an
enforcement
mechanism
back
there,
that's
true,
okay
and
once
once
you're
saying
that
it
is,
it
is
licensed.
You
then
have
to
say
who
who
has
who
has
added
that
license
and
so
on
and
so
forth.
So
everything
enrolls
kind
of
out
of
that
in
a
way.
B
Yeah
right,
that's
true,
and
actually
that's
a
good
point,
because
the
only
example
we've
had
of
someone
customizing
their
page
so
far
was
good.
Jim
and
they've
still
got
it,
but
they
kind
of
made
their
own
page
before
they
realized.
There
was
a
template
because
they
had
a
ruby
template
and
they
just
stuck
some.
They
stuck
a
link
on
the
page
and
said
whatever
they
said.
I
think
we
had
a
conversation
with
the
times
in
the
bottom
line.
So
that's
really
exciting.
B
You're
doing
that,
that's
what
the
point
of
the
spec
is
going
to
be.
We
don't
have
a
spec
yet
so
I
can't
tell
you
if
that's
right
or
wrong,
but
it's
great
that
you're
enthusiastic
so,
but
but
they
did
miss
the
license
and
the
text
of
the
license,
specifically
that
kind
of
that
kind
of
thing
that
kind
of
odi
sanctions
blur
that's
at
the
bottom
of
the
data
site.
Template
was
not
was
not
present.
A
B
Yeah,
and
also
does
the
odi
have
because
this
is.
I
think
this
is
probably
originally
odi
guidance
that
became
a
dataset
site
early
on.
So
I
wonder
whether
the
odi
has
I
feel
like
there's,
probably
a
blog,
possibly
written
by
lee
from
like
a
long
time
ago.
B
D
Maybe
one
of
them
yeah,
I
think
again,
they're
a
bit
more
machine.
Actionable,
though
I
think
that's
memory.
B
Serves,
I
think,
I
think
that
the
certificate
is
but
getting
certificate.
I
think
you've
got
to
tick
a
lot
of
boxes
which
are
quite
fuzzy,
like
that's
true,
the
self-assessment
process,
yeah
yeah
right-
and
I
think
one
of
those
is
like.
I
have
a
data
set
site-
that's
well
described
and
blah
blah
blah.
A
Oh,
so
it's
a
bit
unfair,
citing
a
blog
post
by
lee,
simply
because
there's
there's
a
99
chance
that
that
blog
post
does
exist
like
the
potential
authority
on.
F
Everything
yeah,
that's
a
good
point.
It's
a
bit
unfair,
slashing
your
blog.
A
Yeah
well
also,
just
I
suppose,
the
I'm
not
accusing
you
of
this,
but
I
feel
like
it
would
be
possible
to
justify
anything
by
saying,
there's
a
blog
post
by
lee,
and
it
would
be.
B
A
B
Do
you
want
to
bring
up
what
the
data
site
is
actually
maybe
to
show
them
the
context?
I
just
realized
without
that
discussion,
as
in
an
example
of
why
on
the
stages,
because
that
might
be-
because
I
mean
actually
ollie
and
dom
are
perfect
examples
of
people
that
will
look
at
the
site
and
the
question
to
ask,
I
suppose,
is:
what's
the
minimum
on
this
webpage,
that
would
make
it
useful
for
you
classic
1610.
B
Okay,
so
is
everyone
seeing
the
set
site?
Now,
I'm
sorry,
if
you
jump
to,
if
you
go
to
a
glaston,
that's
actually
the
old
version.
I've
just
realized
it's
been
updated
in
since
then,
so
any
any
there's
a
few
that
are
new,
but
any
gaming
gladstone
for
me
led
would
be
actually
fine
or
maybe,
if
it's
got
there.
A
A
Okay,
so
yes,
ollie
and
dumb.
So
what
we're
talking
about
is
this
html
page
from
a
machine
point
of
view?
What's
important?
Is
these
links
session
series
scheduled
session
facility
facility
use
and
slot
those
all
point
to
our
pde
feeds
of
data,
but
obviously
there's
also
some
text
here
to
make
this
friendlier
for
developers
and
yeah?
This
is
this
is
what
the
discussion
pertains
to.
C
A
Right,
okay,
yeah,
so
that's
and
we
will
be
discussing
that
shortly,
two
points
down
on
the
agenda:
yeah,
so
right
now
that
takes
you
to
github.
Oh
sorry!
No
it
doesn't.
It
takes
us.
It
takes
you
to
the
developer,
open,
active
site
which
includes
guidance
on
how
to
harvest
rpd
yeah
on
the
bottom
left.
I
think.
A
That
documentation
right
now
doesn't
cover
the
booking
specification
which
we'll
be
talking
about
shortly.
C
Yeah
also
the
data,
the
fact
that
it's
open
source
is
saying
there
isn't
it
created
commons,
which
is
useful
to
know
yeah,
okay,
so
yeah,
the
licensing
I
mean
it's
only
worth
feedback.
A
So
there's
any
issues
yep,
so
there's
discussion,
that's
great!
That
takes
you
to
github
there.
That
gives
you
the
issues
attached
to
that.
C
Okay,
yeah,
it's
good,
I
mean
this.
Does
this
need
to
be
like
standardized
between
these
different
endpoints
at
all?
So
do
they
need
to
be
labeled
consistently
across
the
different
endpoints
or
it's
all
expected
to
be?
For
instance,
if
you
go
to
a
swagger
page
for
an
api,
and
it's
all
done
in
a
standard
format,
everything's
pretty
much
standardized.
A
C
Yeah,
that's
the
only
thing
I
was
thinking.
If,
if
I
had
to,
I
was
looking
at
multiple
ones
of
these
and
they're
all
made
me.
You
know
every
label
in
a
different
way
using
different
terminology
for
the
same
thing.
It
might
get
a
bit
confusing.
B
That's
a
super
interesting
point,
though,
because,
as
as,
as
is
it
as
are
having
people
like
good,
jim
and
others,
creating
their
own
kind
of
theme
on,
it
might
be
cool
for
them
from
a
data
user's
perspective,
you
actually
don't
really
want
to
have
to
look
at
every
single
page
and
go
where
on
this
page,
we
put
the
links
yeah,
that's
what
I
was
getting
at.
A
A
B
That's
really
interesting:
yeah
yeah,
you
almost
want
to
see
yeah,
you
almost
want
the
minimum
to
be,
I
mean
we
probably
can't
mandate
the
whole
template,
but
the
minimum
of,
like
the
the
links
to
the
feeds,
have
to
be
central
and
and
clear
and
the
documentation,
like,
we
probably
can't
say,
upper
white
right,
quadrant
upper
left
quadrant
for
the
buttons
but
yeah
I
see
I
I
do
see
what
dom
saying
like.
If,
if
it's,
if
it's
a
complete
maze
and
the
terminology
is
different,
the
colors
are
different,
everything's
different.
B
Then
this
is
going
to
become
a
minor
nightmare.
So
maybe
I
mean,
is
there
I
mean?
Could
we
could
we
provide
an
example
in
the
documentation
I
mean
in
in
the
in
this?
In
the
data
site
I
mean
that
there
are
legitimate
examples
and
other
yeah
and
other
specifications
I
mean
I
don't
know
if
it's
worth
screwing
almost
screenshotting,
one
of
them
or
making
one
up
and
or
making
a
like
a
like
a
what's.
B
It
called
that
ui
program
that
we
use
to
make
funky
like
balsamic
or
something
yeah
like
a
balsamic
sketch,
which
just
has
the
key
bits
that
this
has
on
it
and
just
like,
there's
a
so
there's
at
least
a
prototype
thing
that
that
can
be
pointed
at.
So
if
someone
makes
something
completely
crazy,
where
everything's
upside
down
in
the
inside
out,
we
can
just
point
them
to
the
image
and
say
great
great
first
try,
but
maybe
you
need
to
make
it
look
more
like
this.
Otherwise
you're
going
to
confuse
a
lot
of
people.
A
Yeah,
I
think
that's
easy
enough
to
put
into
the
guidance
as
well.
You
know.
Maybe
these
things
are
not
requirements,
but
presumably,
if
you're,
if
you're
releasing
a
data
set
site,
it's
because
you
want
people
to
use
it
and
you
want
them
to
be
effective
and
easy
for
them,
so
yeah
yeah,
I
can
imagine
some
explicit
indicators
of
what
that
would
look
like,
would
be
well
received.
A
B
Also,
sorry,
it
also
includes
versions
of
the
specs.
I
just
realized
on
that
on
that
page.
I'm
I'm
sure
this
is.
This
is
the
detail
now,
but
above
the
doc
docs
there's
the
version
which,
at
the
point
where
there's
multiple
versions
again
might
become.
F
Right:
okay,
getting
more
interesting.
F
A
And
do
we
need
to
is
that
in
the
json
anywhere.
B
B
Types
of
versions
and
the
booking
spec
versioning,
also,
interestingly,
we
haven't
thought
about
yet,
but
probably
do
need
to
think
about.
I
suspect,
where
the
booking
stuff
lives
on
this
page,
because
we
haven't,
we
haven't
added
booking
to
the
it's
in
the
json,
but
it
isn't.
I
did
have
a
quick
thought
on
this.
B
Actually
I
don't
know
what
you
guys
think
just
on
this
template,
because
yeah
the
kind
of
person
that
likes
to
make
like
fun,
wizzy
things
and
that
you
could
press
you,
could
press
a
link
alongside
documentation,
discussion
that
was
the
open
booking
api
and
the
card
could
flip
like
it
already
does.
When
you
open
the
page
around
and
on
the
other
side,
could
be
the
api
details
of
how
you
access
it,
the
swagger
document
and
the
stuff
so
as
not
to
make
the
page
too
much
more
complicated.
B
A
B
Yeah
you're
right,
that's
very
true,
so
css3
is
what
I
was
thinking
that
does
the
flip
but
yeah.
Absolutely
it
shouldn't
be
reliant
on
that,
so
it
will
need
to
be
yeah.
Maybe
that's
too
much
for
this.
Maybe
maybe
something
a
bit
more
like
scroll
down.
A
B
Yes,
although
they
shouldn't
have
to
parse
it
in
tennis,
because
it's
json
ld
structured
data,
that's
within,
but
but
but
but
still
yeah,
absolutely
having
having
a
standardized.
Most
basic
form
of
something
sounds
good,
so
is
that
is
that,
like
I
guess
in
terms
of
that
one?
Are
we
what's
the
best
way
of
moving
that
forward,
because
I
suppose
we
do
need
to
do
that.
A
A
B
Yeah,
I
guess
maybe
a
quick
question
for
for
dom
and
maybe
for
ollie
is,
is
if,
if
you
were
expecting
to
see
an
interactive
api
where
you
could
make
calls,
would
that
change?
What
you
were
looking
for
on
this
page
just
to
feed
into
that.
C
Well,
yeah,
that's
the
thing
you
see
being
a
you
know.
I
do
a
lot
of
work
with
apis
being
that
api
developer.
When
I
went
to
the
documentation,
I
would
expect
to
go
to
like
a
swagger
endpoint
and
then
be
able
to
make
the
calls
and
then
read
the
documentation
on
every
available
call
and
what
parameters
it
takes
and
so
on.
B
A
So
I
I
feel
like
okay,
just
to
be
clear
in
agenda
terms,
so
under
outstanding
issues.
That
was
the
third
point.
Api
description.
Are
we
content
to
offload
most
of
that
work
here
to
swagger
open
api?
Oh.
A
Well,
no,
I
wanted
to
clarify
that
this
was
sort
of
already.
There
has
been
some
discussion
about
this
already
and
it
sounds
like
the
steer
from
dom
is
basically
yes
and
the
conversation
is:
how
do
we
make
it
clear
that
that
is
what's
happening
and
that
actually
we're
talking
about
two
different
ways
of
interacting
with
resources
on
one
data
set
page.
C
Well
speaking,
someone's
written
ap
apis
with
swagger
and
without
swagger,
I
would
definitely
recommend
using
swagger
as
much
as
possible.
It
makes
the
whole
process
a
lot
easier.
We've
had
issues
where
we've
you
know,
made
new
changes
and
stuff
and
had
to
wait
quite
a
while
for
updates
to
go
through
so
actually
someone's
actually
updated
the
documentation.
So
this
makes
everything
a
lot
easier.
A
Yeah,
unfortunately,
we're
going
at
this
sort
of
from
the
other
angle,
so
the
swagger
is
really
doing
just
a
documentation
function
here.
C
Yeah,
and
also
if
he
has
the
full
the
full
support
you
can
start
generating
classes
and
stuff
off
it.
Can't
you
it's
quite
useful.
A
Yeah,
I
don't
think
I
don't
think.
B
Yeah,
the
slight
the
slight
problem
with
the
sweat
I
mean
yes,
I
suppose.
Actually
what
we've
been
talking
about
is
translating
the
models,
data
models
into
swagger
form
yeah,
which
is
fairly
non-trivial.
However,
that's
just
in
terms
of
the
data
model.
There
is
a
bit
of
the
booking
spec
that
isn't
the
data
activity
opportunity,
data
side,
and
that
is
much
more
uniform.
So
you
could.
C
So
it's
not
possible
just
to
decorate,
I'm
not
sure
what
what
you've
done.
But
it's
not
you
know
the
code,
I'm
not
sure
it's
not
possible,
just
to
decorate
the
classes
and
so
on
and
all
the
methods
and
then
so
that's
what
I
do
with
my
api.
I
can
just
decorate
all
the
methods
of
the
classes
and
then
he
just
runs
it
and
it
just
generates
it
all
automatically
for
you.
B
So
the
challenge
we
have
here
is
that
the
for
reasons
of
semantic
web
fun,
the
data
model
isn't
is
is
fit
is
fairly
flexible,
not
as
flexible
as
some
we've
actually
heavily
constrained
it
compared
to
the
schema.org,
which
is
the
kind
of
standard
for
for
doing
this
kind
of
data
sharing,
but
even
with
the
constraints
in
place,
it's
still
fairly
flexible,
which
means
that
you
can
describe
a
number
of
different
there's
the
there's
like
a
type
hierarchy,
and
things
like
that.
B
B
But
but
we
can
certainly
so
it
would
be
a
case
of
probably
manually,
creating
a
swagger
dock
or
creating
it
well,
yeah,
probably
manually
creating
it
from
the
end
points,
which
is
something
that
we
had
absolutely
intended
to
do.
It's
just
kind
of
to
do
to
do
list
item
yeah,
but
I
guess
tim.
Is
that
not
quite
the
point
you're
making
with
that
agenda
item
there?
I
just
realized.
A
Well,
yeah,
I
was
making
the
more
limited
point
that
are
we
happy
to
treat
documentation
of
the
api
and
the
api
endpoints
as
something
that
lives
more
or
less
in
swagger,
rather
than
something
that
lives
on
the
dataset
page?
That
was
that
was
that
much
more
restricted
view.
Less
architectural
focus.
B
Yes,
so,
and
so
from
what
I
am
I
mean,
if
I'm
understanding
that,
then
then
there
is
also
the
ability.
Well,
there
is
there,
isn't
there
isn't
there's
an
argument?
Potentially
there
isn't,
but
I
don't
know
the
ability
within
schema.org
to
describe
api
endpoints
in
detail
and
then
and
therefore
I
suppose
do
that
on
the
data
side.
A
B
So
there
was
a
conversation
about
this
in
the
w3c.
I'm
sorry
yeah
the
w3c
data
set
site
group
which
I
can
link
to.
I
still
remember
the
url
and
that
that
was
basically
saying
that
so
so,
having
done
an
analysis
of
all
the
various
documents
around
the
use
of
entry
points
and
actions,
it
does
appear.
That's
not
the
intended
purpose,
documenting
lots
of
different
endpoints
from
an
api
and
the
discussion
in
that
forum
with
the
chap
who's
who's
also
involved.
In
that
he's
very
much.
B
He
was
very
much
pushing
the
point
that
this
is
about,
describing
it
not
well,
there's
two
words:
isn't
there
there's
like
a
technical
description
and
there's
like
a
very
high
level,
fuzzy
description
and
I've
lost
all
the
terms
from
that
conversation.
So
I'm
gonna
go
quickly,
look
at
them
and
come
back
to
you
but
yeah.
It's
it
sounds
like.
B
Basically,
it's
not
the
the
preferred
approach
is
to
use
swagger
because
everyone
understands
swagger
and
that's
what
that's
designed
for
you
can
using
hydra
and
other
things,
try
and
do
that
within
the
schema.org
framework,
but
that's
not
generally
what
what
schema.org
intended.
A
And
from
from
what
dom
said-
and
I
think,
probably
speaking
to
developers
generally
swagger
is
kind
of
what
you'd
be
expecting-
that
that's
that's
sort
of
a
natural
flow,
I
think
for
most
developers.
A
A
So
it's
just
a
question
of
what
that
json
looks
like
specifying
both
what
software
is
responsible
for
supporting
booking
functionality
and
describing
what
kind
of
functionality
that
software
supports,
which
is
what
the
point
of
feature
list
is
here,
and
I
think
probably
this
is
not
a
forum
for
hashing
that
out
in
detail
because
there's
a
long
thread
to
go
through
there.
A
But
I
think
it's
more
like
a
call
to
action
of
please
go
through
the
thread
and
either
sign
off
on
it
or
dissent
from
it
as
as
desired,
because
it's
probably.
A
It's
I
feel
like
it's.
This
is
one
usable
representation
here,
and
it
would
be
possible
to
come
up
with
other
usable
representations.
However,
my
imagination
is
at
sort
of
an
end
with
this.
A
I
can't
think
what
the
alternative
representations
would
be
unless
nick
you've
got
something
off
the
top
of
your
head,
that
that
this
fails
to
match.
B
Yeah,
I
know
I
think
the
only
question
I
had
was
the
teacherless
being
an
array,
but
maybe
that
that
makes
sense
because
you've
got
well.
I
suppose,
if
it
is
an
array
which
is
true,
maybe
they're
all
true,
maybe
we
just
need
to
just
formalize.
B
B
Okay,
which
is
interesting
in
itself
because
you
might
have,
if
there's
if
it's
the
union
and
one
of
them
might
be
that
the
certification
site
and
then
another
one
might
be
just
documentation
about
other
add-ons
that
you've
added
to
the
to
the
thing
in
which
case,
how
does
a
machine
reading
a
machine,
that's
reading
the
certs
and
deciding
what
booking
endpoints
have
what
features
in
order
to
decide
what
they
can
connect
to?
How
does
it
know
which
one
of
those
is
machine
readable?
I
guess
it
will
have
to
jump
into
both
and
look.
B
Using
the
same
well,
I
guess
the
yeah
and
then
maybe
there's
a
structure
question,
because
the
machine,
readable
format
that
we're
currently
discussing
and
featureless
stuff
is
for
is
the
is
the
kind
of
test
certificate.
Is
the
test
suite
certificate,
specific
format
which
is
kind
of
kind
of
accustomed
to
that
on
the
basis
that
well,
it's
yeah.
So
it's
a
own
a
little
tool
at
the
minute.
B
B
B
Yes,
so
the
so
the
the
thread,
sorry,
the
the
the
the
certificates.
I
think
I
think
they
do
if
they
tend
to
do
a
few
things.
So
some
of
the
things
that
that's
jason
at
the
moment
states
is
that
it,
it
definitely
does
session
series
and
scheduled
session
and
whatever,
and
it
also
definitely
has
these
subset
of
features.
So
if
I'm
only
going
to
integrate
with
endpoints,
for
example,
that
accept
paid
bookings
to
filter
on
that
and
if
and
then
it
also
has
a.
B
So
I
guess
the
use
case
I
was
thinking
of
was
if,
if
one
was
two-
and
this
is
probably
a
kind
of
a
slightly
later
stage-
scaling
question
rather
than
immediate,
but
the
point
where
we've
got
100
endpoints
and
you
just
want
to
find
out
which
ones
are
x
feature
then
you
know
then
querying
that
information.
I
mean
it
wouldn't
be.
The
kind
of
spider
approach
to
gathering
the
data
for
the
data
site.
B
That
already
exists
is
only
one
step
further
to
then
gather
the
json
within
the
certificate
as
well,
and
keep
going
down
that.
So
you
could
just
extend
whatever
spidering
algorithm.
You
have
by
one
step
and
then
you've
got
a
database
full
of
all
the
features
of
all
the
apis
that
you've
got
access
to,
and
at
that
point
then
you
can,
for
example,
you
could
then
take
that
information
filter
it
to
find
the
ones
that
are
paid
and
then
talk
to
those
guys
to
get
connected
to.
B
You
know
in
that
in
that
way,
and
it
just
makes
that
spidering
slightly
more
onerous.
I
suppose,
if
there's,
if
there's
other
options
in
there,
that
are
of
different
formats.
A
A
We
would
be
residing
in
a
different
url,
basically-
and
I
think
again,
not
not
a
problem
right
now,
but
in
the
in
a
in
a
happily
envisioned
future,
you
would
have
more
than
one
system
capable
of
doing
that
and
you
would
want
to
advertise
the
fact
that
you
conformed
to
you
know
or
that
you
were
usable
within
you
know
different
sets
of
of
tooling.
That
way.
So
I
think
it's
useful
to
keep
an
array
that
way,
but
is.
B
Well,
I
suppose
that
the
question
then
is:
is
it
worth
making
a
structured
object
if
that's
going
too
far
in
the
the
other
direction,
but
you
could
have
feature
list
when
each
item
in
the
feature
list
is
a
a
bit
a
bit
like
when,
when
the
endpoints
are
defined,
the
type
or
the
the
mime
type
whatever
is
in
there
and
then
the
the
endpoint,
I'm
not
sure.
B
A
B
Well
more
than
this
you've
got
so
this
is
a
debate
that
we
had
in
the
other,
in
that
other
forum,
around
the
difference
between
human,
readable
and
machine,
readable
documentation
and
whether
they
should
be
the
same
or
different
in
terms
of
properties
and
the
conclusion
there
seemed
to
be
to
have
them
different
so
that
you
could
reason
about
things.
You
don't
understand.
So
if
you
don't
understand
what
they
are,
you
can
just
say:
oh
human
stuff
csi
should
probably
present
it
to
the
browser
stuff.
B
That's
machine
readable,
if
I
don't
understand
it
I'll,
ignore
it
or
at
least
know
that
I
shouldn't
under
know
that
I
don't
understand
it,
and
I
should
so
that
was
that
was
kind
of
separating
from
that
perspective.
But
then,
within
the
machine
readable
content
there
there
is
the
actual
link.
There
is
a
a
way
of
describing
the
different
endpoints
with
the
content
type.
Oh
yeah.
This
is
a
posting.
B
And
in
fact,
if
you
go,
if
you
scroll
slightly
further
down
to
3.3.2
as
well,
to
give
you
another
example
yeah.
So
this
is
the
idea
that
there's
two
different
encoding
formats
and
they've
both
got
endpoints.
So
if
you're
looking
for
a
particular
format
of
description,
if
you're
looking
for
a
swagger
doc,
you
can
use
the
open
api
mime
type
there
to
get
your
swagger
dock
so
going
yeah.
So
you
see
you
could
this
is
this
is
in
a
different
part
of
the
data
set
size
as
well.
B
So
if
you
wanted
to
enumerate
all
the
swagger
docks
from
all
the
various
endpoints,
you
could
do
that
in
theory
by
just,
you
could
very
specifically
find
the
urls
for
those
and
retrieve
them.
So
wonder
whether
in
this,
this
is
a
similar
thing.
B
If
we're
gonna
have
a
situation
where
there'll
be
multiple
test,
suites,
which
makes
sense
or
different
validation
mechanisms,
then
maybe
having
something
like
this,
where
you
know
type
of
certification
provider
or
test
suite
or
something
in
coding
format,
maybe
we
could
just
inherit
from
creative
work
and
use
encoding
format
and
people
can
make
their
own
mime
types
up.
Okay,.
A
Yeah,
that
seems
that
seems
sensible.
Obviously,
it
adds
some
complexity,
but
at
least.
A
It's
a
pattern
applied
elsewhere,
as
you
say,
so
it's
it's
not
really
increasing
the
complexity
of
this
particular
specification.
Really,
if
you.
B
A
Fair
enough,
okay,
so
nick,
if
you
could
just
could
you
add
that
onto
that
github
discussion.
F
A
I
will
thumbs
up
that
yeah,
perfect,
okay
and
I'm
afraid
nick
I'm
gonna
make
you
continue
talking,
because
that's
the
data
site
data
set
site
outstanding
items,
however
emerging
earlier
discussion
is
this
morning-
was
an
inordinately
complicated
discussion
of
schedules
or
rather,
schedules
are
inordinately
complicated.
Therefore,
we
have
to
come
up
with
a
solution.
B
Yes,
yes,
okay,
so
shall
I
do
a
kind
of
summary
of
this,
then
there's
that
if
you
could.
A
B
Can
leave
you
to
that
yeah?
I'm
sure,
I'm
happy
to
so
this
yeah
this
this
this
is
yeah.
B
This
is
an
interesting
example
of
like
evolving
standards
and
pros
and
cons
of
that,
but
I
mean
there's
only
pros
of
evolving
standards,
I
suppose,
but
when
the
modeling
spec
was
put
together,
the
opening
hours
specification
within
the
modeling
spec
wasn't
fully
defined
and
then
it
was
left
to
the
implementer
to
figure
out
what
types
to
use
on
on
the
basis
of
a
very
rough
schema.org
definition,
and
so
assumptions
were
made
by
implementers
at
the
time
which
led
us
to
a
solution
which
is
the
one
that's
currently
in
use
in
a
couple
of
feeds.
B
Well,
I
I
think
gladstone
have
implemented
it
and
then
I
I
think
it's
in
like
two
other
systems,
maybe
so
that's
all
the
gasoline
fees
plus
two
others,
okay,
and
but
no
one
else,
and
so
and
so
that's
so
that's
that's
that's
interesting
because
then
we've
got
like
there's,
not
many
people
currently
doing
the
thing
and
then
the
other.
The
other
question
was
well,
so
so
going
back
a
step.
So
what
is
the
thing
we're
talking
about?
B
I
think
we're
talking
about
is
opening
hours
trying
to
get
to
this
kind
of
view
where
you
can
look
at
a
particular
place
and
say
when's
it
open
and
it
turns
out
that's
not
completely
as
straightforward
as
it
sounds
to
just
display
that
information.
So
that's
what
the
rest
of
this
this
little
issue
talks
about
so
features
from
a
bit
of
research
that
I've
done
there
with
some
of
the
the
the
images
further
up
so
you've
got.
B
B
So
that's
something
that
people
do
use
and
describe,
and
there
was
a
debate
that
happened
a
few
days
ago
with
the
ever
active
team
so
to
surface
that
here,
one
of
the
contentions
from
that
team
was
maybe
public
holidays
are
not
useful
because
it's
too
general
in
the
sector.
We
probably
have
easter
and
christmas
and
everything
and
like
maybe
all
these
are
all
different
times
so,
and
I
think
that
was
very
specific
potentially
to
everyone.
B
Active
way
of
working
evidence
shows
here
that
there
are
a
number
of
centers
that
do
actually
use
that
to
to
describe
their
opening
times.
Bank
holidays
is
generally
a
thing
that
is
done,
which
explains
also,
why
schema.org
included
that
in
their
spec,
so
so
yeah.
So
then
we
basically
taking
exactly
what
what
so
sk.
So
the
the
modeling
spec
doesn't
actually
give
us
the
detail
past
just
use
opening
hours
specification
at
the
moment.
B
So
what
this
is
proposing
is
that
we
actually
do
give
the
next
level
of
detail
really
so
that
we
can
get
a
bit
of
convergence
where
there
might
otherwise
not
be,
and
I've
linked
here
to
the
google
recommendations
which,
in
the
google
it's
own
structure,
their
own
structured
data
reading
and
they
do
have
a
kind
of
a
whole
set
of
scenarios
which
I've
effectively
just
lifted
and
shifted
into
the
proposal,
because
obviously
they've
thought
this
through
and
done
all
the
they've
had
all
the
different
angles
on
it,
and
so,
if
tim
on
the
right,
you
click
on
business
hours.
B
Table
of
contents
on
the
right,
yeah-
perfect.
Yes,
so
thank
you.
So
this
is
stunning
hours
late
night
hours
all
day
hours,
seasonal
hours,
you'll
recognize
these
headings
from
the
document,
so
basically
just
they're.
This
is
what
they
do.
It's
quite
obvious
how
this
works.
You've
got
the
day
of
week.
You've
got
a
numeration
there.
You
notice
they
don't
actually
have
schema.org
in
there.
If
you
scroll
up
a
little
bit
above
that
you'll
see,
it
says
we
accept
both
schema.org
notation
and
blah.
B
So
they've
recognized
that
some
people
are
using
schema
and
some
people,
don't.
Obviously
we
prefer
schema
as
a
policy
in
in
the
active
specs.
So
that's
why
our
stuff
has
got
that
in,
but
but
largely
apart
from
that
minor
difference.
B
Our
stuff
matches
this
as
as
they've
outlined,
and
it
matches
it
in
in
the
way
that
you've
got
the
array
of
days
of
week
and
that's
that's
important,
because
currently
the
models
that
we've
had
in
place
and
the
net
library
and
the
other
libraries
don't
have
an
array
for
day
of
week.
They
actually
only
allow
a
single
entry
there,
which
is
unfortunate
because
it
means
the
implementations,
as
they
have
already
cracked
on
have
influenced
that
as
a
single
entry.
B
So,
although
we
could
say,
let's
stick
with
the
single
option,
the
problem
is
that
that
then
makes
it
more
difficult
for
everybody,
because
if
what
you
really
want
to
be
able
to
say
monday
to
thursday
is
this
and
we
end
up
forcing
people
to
duplicate
data
and
then
on
the
data
consumer
side,
try
and
bring
it
back
together
again
to
produce
the
same
thing.
What
I
think
really
has
happened
here
is
it's
just
a
slight
oversight
in
terms
of
that
tooling,
which
I'm
probably
partly
responsible
for
about
two
years
ago.
B
So
it
was
one
of
those
kind
of
moments
of
like
oh
gosh,
yeah,
we've
just
assumed
it
and
cracked
on
and
didn't
think
to
read
other
implementations,
but
to
be
fair
to
us
at
the
time
there
was
a
big
surface
area
to
get
on
with
and
implement.
So
I
think.
A
Not
for
seeing
the
importance
of
different
opening
hours
specification
over
holidays
and
covered
lockdowns
2018
fair
enough
yeah.
B
Different
world
true,
so
yes,
so
so
yes,
so
tim,
if
you
would
mind
flicking
back
to
the
the
other
page
then
or
wherever
it
is
saying
great,
so
you
can
see
here
that
we've
got
day
of
week
as
an
array
we've
got,
closes
and
opens
which
is
exactly
the
same,
and
then
we've
got
some
optional
properties
further
down,
which
is
the
valid
valid
from
and
valid
through.
So
actually
they
have
in
their
example.
B
They
also
have
options
where
you've
got
seasonal
opening
hours,
so
they
can
be
open
for
christmas
or
you
know
the
ice
rink
is
open
for
christmas.
Only.
You
can
use
these
to
indicate
that
that's
the
case
you
can
also,
if
you
keep
going
further
down
in
in
schema.org,
you've
also
got
this
thing
called
special
opening
hours.
B
Special
opening
hours
means
that
you
can
specify
overrides
for
the
case
of
example,
lockdown,
and
what
that
ends
up
with
is
this
example,
as
tim's
scrolling
through
so
you've
got
special
opening
hours
there
at
the
bottom,
which
are
kind
of
saying
between
that
day
and
that
date
everything's
closed.
The
other
thing
that
google
defines
is
the
opens
and
closes
have
special
meaning
for
zero
zero,
zero
zero
if
they're
both
the
same.
B
It's
closed,
also,
twenty
zero,
zero,
zero,
zero
and
twenty
three
fifty
nine
means
always
open,
and
so
again
just
lifting
that
out.
So
I
haven't
done
any
creative
thinking
here.
I've
literally
just
elegantly
copied
what
google's
already
done
and
and
put
it
in
here.
So
just
that
then
means
that
we
we've
got
special
opening
hours.
B
So
you
can
then
have
a
lockdown
scenario,
so
you
could
put
in
all
the
hours
that
you're
usually
are
open,
and
then
you
can
say
actually,
but
between
these
two
dates:
we're
going
to
not
be
open
because
of
lockdown
and
add
that
in
or
but
that
same
information
that
same
special
opening
hours
could
be
used
for
lockdown.
It
could
be
used
for
christmas
opening
hours
that
are
specific
to
christmas
day.
B
It
could
be
used
for
any
type
of
very
day
specific
hours,
so
yeah,
and
so
this
is
and
then
to
help
unblock.
The
implementations
I
mentioned
are
kind
of
looking
to
crack
on
with
this.
At
the
moment,
I've
then
gone
ahead
and
made
the
update
to
the
tooling
to
reflect
that
array
change.
B
So,
even
though
that
is
a
breaking
change
to
the
to
the
validator
and
that
things
will
previously
did
validate
will
no
longer
in
that
respect,
there's
only
those
those
systems
I
mentioned
involved
and
so
talking
to
each
other
individually.
If
we
can
get
them
all
to
bump
and
add
an
array,
I've
already
emailed
two
of
them
this
morning,
just
to
change
it
to
an
array
and
if
we
can
catch
this,
I
suppose
before
it
gets
too
broad
widely
implemented.
B
And
then
we
don't
have
this
weird
because
there's
nowhere
else
in
the
specs,
apart
from
one
specific
exception
which
was
made
for
for
reasons
where
you
have
an
array
or
a
single
option,
because
if
you
have
this
kind
of
duality,
then
er
all
the
whole
model
gets
like
complexity
grows
in
all
parts.
So
that's
that's
a
good
summary.
A
Tip,
I
don't
know
yeah
yeah
I
mean
I
guess
my
only
reservation
is
this
does
seem
quite
complicated
to
parse
and
and
having
an
array
of
single
would
make
that
much
more
so
or
irritatingly
more
so
anyway,
I
mean
it
makes
perfect
sense,
and
you
know,
as
you
say,
it's
already
part.
You
know
well-established
part
of
schema.org
and
adopting
it
given
the
the
need
seems
pretty
sensible,
but
I
I'd
I'd,
ask
dom
and
ollie
if
they
had
any
opinions
on
this.
C
Give
me
any
ideas,
no
nothing
for
me,
don't
go
ahead.
I
think
we
originally
used
ical
in
open
referrals
that
was
quite
limiting
in
long
way,
so
this
won't
be
a
good
way
of
going
with
it.
B
It's
interesting,
so
we
do
use
ical
in
a
different
place,
which
is
for
schedules
for
recurring
events.
But
that's
that's
interesting,
but
I
would
suspect
that
you'd
have
challenges
with
that
right
because
that's.
C
F
A
Oh
sorry,
did
you
lose
me
or
I
I
lost
you
or
what
happened
there.
I
just
went
quiet
for
a
few
moments.
Okay,
yeah.
I
think
we
lost
dom
for
a
little
bit.
No,
I'm
sorry,
okay,
great
great
okay,
so
you
were
saying
you
were
saying
that
in
fact
your
recollection
is
an
open.
Referral
actually
uses
something
close
to
the
schema.org.
C
C
Yeah,
addressing
have
an
array
of
them.
I
don't,
I
think
there
are
like
notes
and
so
on
for
special
special
circumstances,
but
nothing
like
what
we
have
now.
I
imagine
you'd
have
to
wipe
your
current
opening
times
and
updated
all
those
or
something-
and
you
know
have
a
new
list.
I
mentioned:
that's,
not
you
how
you
do
with
the
current
spec.
A
C
The
referral
uk
yeah
I
mean
the
uk
one-
is
the
same
as
the
u.s
has
got
a
bit
chopped
out
for
the
most
part,
one
or
two
tables.
A
Okay,
I
will
take
a
look
at
that
then,
just
for
my
own
information,
I
think,
given
that,
given
the
sort
of
internal
consistency
of
this
I'd
be
happy
to
approve
this
for
integration
into
the
into
the
specification
2.1.
I
guess-
and
we've
all
got
a
couple
of
minutes,
but
did
anybody
else
have
any
other
business
they
wanted
to
raise
before
the
end
of
the
call.
A
Okay,
thank
you
all
very
much
for
joining.
I
feel
like
for
those
of
you
who
are
first-time
attenders.
This
was
a
fairly
deep
dive
introduction,
so
yeah,
it's
very
interesting.
A
Okay,
great
I'll,
be
circulating
the
agenda
for
the
next
call
in
advance,
we'll
be
looking
at
how
we
represent
courses,
or
you
know,
series
of
events.
B
In
the
next
call,
I'm
sorry
tim
on
the
courses
one
have
you
have
we
invited
lee
and
others
to
that.
Thinking.
Aren't
people
not
yet.
No,
though,
obviously
we
do
have
to
have
the
right
people
in
there.
B
A
B
A
It's
very
much
alive,
so
I
think
we
it's
definitely
definitely
next
on
the
agenda,
but
yeah
we
do
need
to
make
sure
anyone
immediately
affected.
Is
there
absolutely
okay
at
the
top
of
the
hour
or
at
the
bottom
of
the
hour?
Thank
you.
Thank
you
very
much
for
joining
and
hopefully
see
you
all
in
a
couple
of
weeks.
Thanks
very
much.