►
From YouTube: GraphQL Working Group (Secondary, APAC) 2022-10-12
Description
GraphQL is a query language for APIs and a runtime for fulfilling those queries with your existing data. GraphQL provides a complete and understandable description of the data in your API, gives clients the power to ask for exactly what they need and nothing more, makes it easier to evolve APIs over time, and enables powerful developer tools. Get Started Here: https://graphql.org/
A
A
B
B
C
Hello,
sorry
I
realized
that
Benji
had
already
flagged
this
to
me
and
I
still
fell
in
the
Trap
of
the
calendar,
has
a
different
Zoom
invite
than
the
agenda
file.
So
it
goes,
but.
C
Yeah
only
only
actually
Michael
was
there,
so
I
showed
up
to
find
Michael
there
and
then
we're
like
well
we're
really
the
only
two
people
who
showed
up
nearly
on
time
and
were
like
wait
a
minute
wait
a
minute.
No
I
bet
sure
enough.
There's
a
separate
Zoom,
ID
anyhow,
we'll
Iron
Out
The
Kinks
here
well
welcome
everybody
to
this
is
the
first
secondary
one.
We've
done
ever
right.
It's
also
the
first
eight
pack,
time
zone
friendly,
one
that
we've
done
ever
and
we've
got
some.
C
Thanks
for
making
this
happen,
yes,
Andy,
it's
been
way
too
long,
since
we've
been
in
a
call
together
and
I.
Think
yeah
is
it
the
first
time
that
we've
met,
certainly
in
this
context
I
think
or
at
least
a
long
time.
E
E
C
B
E
C
Is
also
good
conference
is
all
scary,
I
think
we've
got
everybody
right.
D
Yeah,
it's
a
quick
quick
just
note.
Should
we
take
notes
in
the
same
document
as
the
primary
document?
If
so,
I'll
update
the
agenda.
D
E
C
Me
is
it's
okay.
Benji
mentioned
wanting
to
join
and
I
said
that
I
would
be
very
gracious
if
he
did,
but
he
did
not
have
to
thank
you
Matt
for
spotting
the
notes
by
the
way.
A
C
I
was
I
was
a
little
surprised
that
it's
just
the
six
of
us
I
saw
more
support
from
that
thread
of
folks
who
might
want
to
show
up.
So
that's,
okay,
I
think.
Maybe
I'll
go
back
to
the
discussion
thread
and
ping,
a
bunch
of
folks
and
rah-rah
about
how
we
did
the
first
one
of
these,
and
we
got
good
early,
healthy
attendance
and
we
should
do
more
yeah
Matt.
Do
you
have
our
our
node
stock
up
and
running.
D
C
Yeah,
that's
good
to
have
okay,
so
doing
things
a
little
out
of
order,
but
start
at
the
top
I
get
to
do
my
pithy
by
all
of
us
joining
here.
We
agree
too.
Membership
agreement,
participation,
contribution
guidelines,
code
of
conduct
links
all
those
in
the
agenda
files.
Some
of
those
I
wrote
those
are
really
well
written.
Some
of
those
somewhere
that
Linux
Foundation
wrote.
Those
are
boring.
C
C
Let's
try
to
go
in
the
order
that
it
shows
in
the
agenda
file
I'm
going
to
post
this
in
the
chat
just
in
case
anybody
doesn't
happen
to
have
it
open.
So
there's
me
at
the
top,
so
hello,
everybody,
Lee
director
of
the
graphql
foundation
and
coming
to
you,
live
from
San
Francisco.
D
Okay:
okay,
next,
hey
everyone,
I'm
Donna,
I'm
I,
help
out
with
graphql
Java
I,
also
work
with
Andy
at
atlassian,
coming
to
you
from
Sydney.
E
C
Thanks
for
joining
from
New
York,
hopefully
it's
too
late
there
right
end
of
the
day,
I.
D
No
and
I
hadn't
missed
the
last
meeting
because
of
flying
so
whatever.
B
I'm
Philip
I
work
for
edit
on
the
Grove
Hill
extension
that
we've
pulled
on
top
of
Grove,
kill
Java
and
that's
code
used
in
caucus
and
open
Liberty
for
IBM.
It's
my
first
meetings
I'm
mostly
just
here
to
listen.
A
Yeah,
that
leaves
me
hi
everybody,
I'm
Andy,
working
on
graphical
Java,
still
maintaining
it
today,
these
days
together
with
Donna,
actually
and
Pratt,
who
is
not
here
today
and
yeah,
still
working
at
that
last
year,
I'm
joining
from
Brisbane.
Actually,
this
time
from
Australia.
C
I
feel
like
we've
got
solid,
Australian
representation
here
today,
Sydney
Melbourne
Brisbane.
We
got.
B
C
A
A
C
Exactly
yeah
that's
right:
I
should
I
should
probably
write
that
in
as
like
agenda
item
zero
and
all
of
these
docs
as
to
like
a
reminder
for
everyone
that
we
do
record
and
publish
these
on
YouTube
and
so
or
a
reminder
to
myself
to
make
sure
that
I
don't
tell
Untold
jokes
about
the
quality
of
the
docs
that
people
have
written
or
something
else
yeah.
Next,
we
usually
look
for
a
note-taking
volunteer.
Matt
sounds
like
you
have
in
assembling
the
note
stock
there
become
the
initial
volunteer
yeah.
C
But
if
everyone
Matt
kindly
put
that
note
stock
in
the
agenda
file
at
the
top
of
the
file,
you'll
see
live,
notes,
there's
a
link
to
a
Google
doc
and
if
it
doesn't
work
reload
the
page,
because
he
just
added
that
in
there
in
the
last
few
minutes,
and
so
especially
if
Matt
chimes
in
a
discussion,
if
anyone
would
be
willing
to
jump
in
and
help
take
notes
along
the
way
that
would
be
super
useful
I
think
I
will
probably
skip
previous
meeting
action
items.
E
Yeah,
that's
also
a
thing
I'm,
not
sure
about
how
we
do
that,
because,
like
there's
the
main
meeting
where
we
have
a
lot
of
discussions
on
things
and
you.
B
E
C
That's
true:
maybe
what
we
should
do
is
to
the
degree
that
we
spot
actions
to
open
here
is
to
flag
which
meeting
they
came
from
as
like
a
GitHub
tag
on
them.
Yeah.
That
way,
when
we
open
up
action
items
for
this
meeting,
we
can
look
at
like
just
APAC
time
zone
ones
for
the
Europe
specific
one
secondary
one.
We
could
look
at
just
that
one
and
then
we
can
look
at
all
of
them
in
the
primary
meeting,
and
so
you
know
if
something
falls
through
to
the
primary
meeting.
C
Like
that's,
okay,
the
goal
there
is
to
just
make
sure
we
don't
miss
anything
that
we've
said
we
wanted
to
do,
but
probably
better
than
just
straight
up
skipping
action
items,
but
this
is
the
first.
So
obviously
we
have
no
previous
meeting
action
items
from
previous
times.
D
C
That's
a
good
call,
there's
usually
not
like
a
huge
amount
of
those
action
items,
so
maybe
that's
actually
the
simplest
thing
that
works
it's
just
to
like
we'll
open
it,
we'll
scan
it
and
anything.
That's
not
relevant
for
the
particular
group
in
the
meeting
with
just
Passover
I'm
happy
with
that
I
like
that.
C
My
goal
was
to
kind
of
block
on
calendar
enough
time
that
if
we
ever
have
like
a
lot
of
overflow
or
a
lot
of
stuff
to
discuss,
which
I
think
is
just
more
important
for
this
meeting
than
it
is
the
Europe
friendly
time
zone
secondary,
because
this
is
like
the
primary
way
in
which,
if
you're
in
an
APAC
time
zone,
that
you've
got
the
opportunity
to
get
like
live
time.
C
But
my
expectation
is
that,
like
we
probably
won't
fill
the
full
hour
and
a
half
all
the
time,
and
so
like
right
now,
I
see
we
have
like
one
major
topic
to
get
into
that.
We
maybe
spent
a
half
hour-ish
on
that
seems
like
a
sweet
spot
like
I
expect
that
most
of
these
will
have
like
one
or
two
major
things
to
dig
into
and
that'll
be
great.
So
the
one
and
a
half
hour
is
more
of
like
a
upper
threshold.
C
A
I
have
a
little
bit
meta
questions
about
topics
I'm,
not
sure
you
want
to
discuss
this
is
now
or
after,
but
but
my
main
question,
one
question
that
came
to
my
mind
was
a
little
bit
about.
Okay.
If
we
have
a
topic
that,
for
example,
we
bring
it
up
now
in
the
Apec
meeting.
A
This
is
like
should
also
be
covered
in
the
other
meetings,
or
are
other
interested
parties
kind
of
for
us
to
join
the
Apec
meeting.
If
they
want
to
contribute
something
about
it,
you
know
like
yeah,.
C
Yeah
so
I
think
one
thing
that
this
is
making
me
realize
we
definitely
need
is
a
report
back
step.
So
I
think
this
is
something
we
should
add
as
like
a
standing
agenda
item
to
the
primary
meeting
is
just
just
a
quick
like
basically
like
an
agenda
review
five
minutes.
What
did
we
talk
about
in
the
two
secondary
meetings?
C
C
Then
I
think
what
we
need
to
do
is
make
sure
that
there's
some
like
co-champion
for
proposal
or
they're,
just
like
someone
who's
gonna,
be
able
to
do
that,
and
so,
if
you
don't
have
that
right
off
the
bat
like
this
might
be
a
good
form
to
like
ask
for
a
volunteer
so
like
Michael,
Matt
and
myself
are
in
the
primary
meetings
as
well
and
so
like.
C
C
They're,
making
sure
that
you
fold,
one
of
us
in
to
carry
it
carry
the
torch
forward,
is
probably
the
right
way
to
do
that,
but
very
good
flag
on
the
like.
How
do
we?
How
do
we
make
sure
that
everyone's
aware
of,
what's
going
on
across
multiple
meetings,.
A
A
A
I
have
no
presentations,
I
have
no
real
outline,
but
I
will
talk
a
little
bit
and
please
feel
free
to
ask
questions
or
like
let's
discuss
stuff.
I,
give
a
little
bit
background.
Also,
maybe
somebody
will
watch
this
again.
Online
I'm
I'm,
not
sure,
but
let's
I
will
give
a
little
bit
context.
Even
for
people
who
are
not
here.
There
was
a
discussion
three
four
years
ago.
A
Basically,
it
came
out
of
the
real
life
need
in
atlassian,
but
also
in
other
places
that
okay,
there
are
non-trivial.
Custom
scalars,
for
example,
daytime
is
one
of
the
most
popular
ones.
B
A
Which
are
not
easy
to
to
get
right
across
different
implementations
in
every
detail
and
one
of
the
things
Lee
and
I
actually
looked
at
it
at
this
time
is:
is
it
possible
even
or
is
it
somewhat
even
desire
to
Define,
maybe
more
scalars
inside
the
spec,
like
a
daytime
scalar
implementation,
additionally
to
the
five
ones
we
have,
and
one
thing
that
that
we
realized
quite
quickly
was
that
why
possible
you
have
the
huge
burden
of
existing
conflicting
implementations
across
the
ecosystem
and
especially
with
more
complex
one
like
daytime
there's,
surely
no
no
agreement
what
it
really
means,
so
there
was
back
and
forth
and
we
ended
up
going
in
a
little
bit
different
direction
and
adding
the
specified
buy
URL
directive.
A
So
you
can
now
specify
a
URL
for
a
custom
scalar
which
should
point
to
a
stable
web
page
describing
the
custom
scalar,
so
that
was
kind
of
our
indirect
way
of
being
able
to
talk
about
custom
scale
as
implementation,
without
actually
forcing
a
specific
implementation
at
everybody
and
yeah.
This
was
all
fine.
This
has
merged
I.
Think
two
three
years
ago
now,
Looking
Back,
Now
I,
must
say:
I
have
not
seen
a
custom
scalar
specification
in
the
wild
outside
of
what
I
wrote
for
daytime
yeah.
It's
yeah
the.
E
Main
problem
here
is
and
I'm
pointing
for
daytime
I'm,
pointing
always
to
graphql
the
scalars,
but
the
main
problem
I
see
is
that
it's
not
something
from
the
graphql
specification
like
I
think
it
would
be
better
if
the
if
these
things
would
be
on
graphical
at
work
and
that
there
is
a
central
point.
We
we
guide
people
like
their
sponsors
from
the
gate,
all
these
scalars
for
long
and
stuff
like
that
or
money
and
things
a
ton
of
scalars
that
they
wrote.
E
We
actually
ported
them
to.net,
but
these
could
all
have
specifications,
but
it's
not
for
the
most
people.
I
talk
to
it's
not
and
there's
no
visible,
Central
Station.
To
put
these
things.
A
Oh
give
me
five
minutes:
I
will
come
back
to
this.
That's
definitely
something
I
want
to
discuss
so
yeah
I
will
come
back
to
this
detail
on
all
good
points,
but
we
see
two
challenges
with
custom
scalar
specification.
A
First
of
all,
it's
actually
quite
hard
to
write
a
good
custom
scale
specification,
because
the
mental
model,
what
a
custom
scalar
does
in
detail,
is
non-trivial,
especially
like
again
lean
and
I
worked
like
a
year
ago,
two
years
ago,
with
time
flies,
I'm,
not
sure
I'm
on,
for
example,
one
specific
extension
that
allows
translating,
from
a
literal,
to
a
no
sorry
from
a
external
value
or
raw
value
to
literal,
to
support
custom
scalars
in
all
aspects.
A
This
is
like
just
one
detail
and
it's
very
specific
knowledge
that
I
don't
see
widely
adopted
across
all
implementations
and
especially
people
outside
of
the
people
who
maintain
them.
This
knowledge
is
very
obscure
and
very
unclear,
so
we
set
together
at
Donna
and
I,
and
and
we
decided
we
tried
to
tackle
this
problem
in
two
fronts
and
I
will
come
back
at
the
second
part
to
what
you
said.
Michelle
is
like.
A
First
of
all,
is
we
want
to
simplify
or
want
to
provide
guidance
for
people
writing
custom
scale
as
implementation,
so
we
want
to
come
up
with
a
quiet,
easy,
understandable
template
that
references.
We
called
it
now
an
implementation
guidelines,
so
you,
you
should
have
a
clear
mental
model,
what
a
custom
scaler
does
and
what
is
required
for
the
implementation,
blah
blah.
But
this
is
a
static
document,
and
on
top
of
this
you
can
you
can
reference
a
specifically
and
you
only
have
to
do
minimal
work
to
Define
custom
scalars.
A
A
So
we
have
a
graphical
minor,
scalar
repo
actually
already
in
the
transfer
GitHub
orc,
which
was
always
intended
as
a
publication
Channel,
so
that
people
could
contribute
custom,
scalar
specification
and
they
will
show
up
on
craftgirl.org
to
quickly
jump
into
one
detail
that
will
come
back
very
lightly
and
come
people
will
surely
ask
I
mean
like
the
goal
is
not
to
be
an
authoritative
thing
about
every
scale.
It's
more
like
every
person
can
publish
their
own
specification
on
a
central
place.
A
So,
for
example,
I
can
publish
my
data
implementation,
but
it's
Donna
disagrees.
She
can
publish
her
datum
implementation,
so
it
would
be
like
a
namespaced
place,
but
it
will
be
hosted
under
craftwell.org,
scalers
or
spec
or
scalers.craftgrill.org.
Something
like
this
and
additionally,
it
should
have
like
a
common
format,
a
common
template
that
you
can
easily
understand
and
compare
custom
scalar
specs.
This
is
kind
of
the
vision
we
have
this
Vision
very
broadly
years
ago,
but
well.
A
We
are
all
doing
this
voluntarily
here
and,
and
so
we
never
got
around
to
really
make
this
happen,
and
but
now
we
found
some
time
and-
and
we
worked
on
the
first
step
specifically
and
just
what
we
want
to
present
a
little
bit
is
we
wrote
a
custom,
scalars
implementation
guideline,
it's
linked
in
the
agenda.
It's
a
markdown
format,
that's
hopefully
still
readable
and
it
talks
about
custom
scalers.
What
to
think
about
how
to
think
about
Implement
implemented
and
clarifies
a
few
things
around.
A
What
is
the
raw
value
and
literal
coerced
value
stuff
like
this?
It
it
yeah.
It
also
points
to
an
interesting
thing
that
the
official
like
the
spec
for
the
built-in
scale
is
not
the
clearest
part
of
the
spec.
To
be
honest,
makes
the
scalar's
description
in
the
in
the
spec
could
be
improved.
Also
as
a
as
a
follow-up,
we
believe
yeah,
but
this
is
the
first
part
we
want
to
get
this
implementation
guideline
out
and
we
also
have
examples
how
the
spec
would
look
like
for
the
built-in
scalers.
A
It
would
be
exactly
exactly
what
Michael
said
like
we
want
to
publish
it
on
craftgirl.org
yeah,
I
I
thought
I
I
talked
a
little
bit
longer.
Now
we
have
time
any
questions
or
discussions
before
I
show
maybe
a
little
bit
what
we
did
in
more
detail,
but
this
is
clear.
It's
like
any
major
concerns,
anything
that
goes
against
what
is
doing
yeah
happy
too,
to
talk
about
it.
C
I'm
still,
a
girl
like
I,
think
the
way
you
articulated
the
plan
aligns
directly
with
how
I've
been
thinking
about
it
too.
I
think
it's
just
a
matter
of
someone
needs
to
just
do
it
and
something
especially
like
your
framing
is
right
here,
like
our
our
quality
threshold
doesn't
need
to
be
quite
as
high
as
what
we
would
usually
hold
for
most
of
our
other
work,
because,
right
now
what
we
have
is
I
think
you're
right
like
we
basically
have
nothing
like.
C
There
are
very
few
cases
where
people
are
pointing
at
a
common,
well-defined,
scalar,
spec
and
using
that
specified
by.
We
have
the
mechanisms
in
place.
We
just
don't
have
a
ton
of
like
examples
of
where
people
can
point
and
I
think
what
you're
framing
here
of
like
having
a
template
that
has
here's
the
questions
that
you
need
to
be
able
to
answer
to
be
able
to
have
like
a
decently
specified
scalar
with
like
a
moderate
threshold
like
I,
think
anyone
should
be
able
to
publish
these
anywhere.
C
They
should
be
able
to
copy
the
template.
Stick
it
on
any
website,
you're
good
to
go,
and
so
that
gives
us
some
license
to
have
a
little
bit
of
like
a
code
review
process
rather
than
just
rubber,
stamping
everything
that
shows
up.
But
we
don't
need
to
have
formal
approval
by
a
working
group
or
a
tsce
to
get
it
in
like
we
could
just
have
a
handful
of
trusted
delegates
to
say
this
looks
good
or
this
part
of
your
proposal
doesn't
make
sense
fix
that.
C
So
you
don't
have
something
confusing
and
make
sure
that
stuff
that
shows
up
there
is
like
decently,
quality
or
better.
A
A
But
distilled
wooden
boyfriend
like
a
manual
coat
of
manual
Revenue,
and
we
don't
expect
also
thousands
of
people
to
show
up.
This
will
be
still
a
limited
group
of
people
who
will
take
advantage
of
it.
So
having
a
the
revenue
process,
menu
will
is
still
scalable
well
enough
in
my
world,
I
would
be
happy
if
10
people
show
up
in
the
next
two
years.
I
would
put
this
under
success.
E
I
think
the
the
the
the
the
main
thing
is
visibility,
like
the
people
from
the
guilds
they
are
motivated.
So
they
do
a
lot
of
these
things,
but
I
think
it's
at
the
moment.
It's
not
visible
that
there
is
something
like
this
like
we
all
like.
There
is
no
guide
or
there's
no
hint
that
there
is
somewhere
where
I
could
publish
my
scale
up
right.
E
If,
if
I
didn't,
if
I
didn't
yeah,
if
I
didn't
know
that
graphql
their
status
exists
when
I
tell
people
about
that,
nobody
knows
and
also
the
specified
buy,
is
a
detail
in
the
stack
most
people
don't
even
know
that
the
scalers
have
added
that.
So
that
needs
a
place
where
we
think
where
we
explain
that
but
yeah
and
then
I
think
there
will
be
more
people
who
publish.
C
Yeah
I'm,
imagining
like
I,
mean
graphqlskillers.com
that
that
was
like
the
original
thesis
of
that
right,
which
is
like
super
high
level,
describe
what
it
is
point
to
the
spec,
where
you
need
to
point
to
the
spec
and
then
just
list
out
all
the
ones
you
got.
C
What
do
we
need
to
move
this?
One
forward?
Is
this
because
it
seems
like
we've,
got
pretty
broad
consensus
that,
basically
anything
we
do
here
is
going
to
be
useful
and
there's
not
a
lot
of
there's,
not
a
lot
of
opportunity
to
get
it
wrong
like
because
we're
in
a
state
where
no
one
is
really
paying
that
close
of
attention.
If
we
get
something
up
and
running
we're
like
yeah,
it
could
be
better.
C
It's
like
cool,
we'll
fix
it,
make
it
better
and
eventually
get
to
the
point
where
we
start
to
see
people
using
it,
and
then
that
kind
of
gets
dialed
in.
A
Let
me
show
you
a
little
bit
what
we
got
and
I
think
from
this
from
there.
We
can
talk
a
little
bit.
How
we
want
to
take
this
forward.
I
will
not
go
in
every
detail,
but
I
will
take
advantage
of
that
beyond
the
only
Topic
at
the
moment.
So
let
me
show
you
a
little
bit
if
that's:
okay,
yeah
Where
We
Are,
okay.
So
this
is
the
original
work
I
did
years
ago
about
on
crafts.com.
A
It
still
says
work
in
progress,
because
I
was
never
totally
satisfied,
satisfied
with
this
kind
of
spec.
This
is
a
daytime
scale
aspect
which
stands
alone
so
and
it
it
introduces
certain
naming
certain
mental
models
and,
at
the
same
time,
it
defines
what
a
daytime
scaler
should
look
like,
not
bad,
but
it
it
didn't
it.
It
would
not
allow
for
a
clear
template
to
emerge
from
it.
It's
it's
very
unstructured
overall
and
it's
usable
to
read
as
a
human,
but
I
was
not
happy
that
this
in
some
way
is
like
templatable.
A
A
Let
me
actually
try
to
it's
just
up
a
little
bit
yeah,
it's
like
it's
a
it
talks
about
serialization
and
destabilization
again.
Observability
corrosion
result,
corrosion
input,
corrosion,
literal
raw
input
value.
It
has
this
nice
picture
that
that
is
very
important
for
the
input
version.
How
to
think
about
these
things.
It's
it
sets
the
frame.
Let's
say
it
like
this,
and
and
what
we
found
is
when
you
have
this
document,
which
is
not
specific
to
any
custom
scalar
it
again
it
it.
A
It
clarifies
the
mental
model
and
the
constraints
you
have
when
implementing
a
custom.
Scalar,
then
what
we
found,
what
you
actually
only
need
per
custom
scalar
is-
and
that's
not
so
well
formatted
here,
but
what
you
actually
only
need
per
custom.
Scaler
is
three
things
on
the
core.
Of
course,
you
should
have
a
little
bit
explanation.
You
should
have
maybe
some
some
references,
But.
A
A
This
is
for
the
built-in
scalers
here,
for
example,
the
ID
scaler
except
I,
just
literally
input
a
string
value
or
in
value
literal,
so
that
there's
no
restrictions
to
it.
So
this
is
all
what
you
need
and
then
what
you
need,
additionally,
is
the
Json
raw
input.
So
this
is
the
variables
passed
to
a
transfer
request.
A
In
which
way
can
you
represent
a
custom
scalar
as
variable
and
because
the
only
real
specified
sterilization
format
we
have
in
Crossville
is
Json,
so
this
is
expressed
as
a
Json
value,
and
so
here
an
ID
can
be
any
Json
string
and
then
the
result
again
is
is
when
a
scalar
is
used
as
output
type.
Again,
the
only
observable
serialization
decentralization
format
we
have
is
Json.
This
is
why
we
specify
it
also
in
Json-
and
we
say
an
ID
Scala
can
be
any
Json
string
as
a
result.
B
A
For
example,
the
float
one
has
actually
some
interesting
aspects
and
takes
the
float
value
or
as
little
as
literal
input
and
then
the
Json
raw
input.
Again,
these
are
the
variable
values
you
can
pass
into
request
and
there
are
some
interesting
details
about
that.
The
Json
spec
actually
does
not
specify
the
range
of
a
number
so
but
the
float
value
in
class
Square.
A
Does
it
clearly
so
there's
a
potential
conflicting
point
which
you
know
which
should
be
clarified
in
a
spec
and
the
results
of
a
float
again
is
the
Json
number
again
there's
also
things
around.
It
includes
excludes
Nan
and
positive
negative
Infinity,
while
the
Json
values
could
include
this
in
theory
as
I
understand
it.
So,
as
you
can
see,
this
is
a
kind
of
the
core
for
custom.
Scalar
spec
would
be
talking
about
these
three
different
things.
Literal
input,
Json
raw
import.
Json
result.
B
A
Would
be
the
the
template
we
talk
about
and,
as
you
can
imagine,
the
datum
version
would
be
much
smaller.
Now
like
this
again,
this
mixes
meant
to
model
in
the
introduction
of
topics
with
the
actual
spec,
and
we
would
have
much
simpler
points.
A
We
would
have
just
like
this
table-
and
maybe
this
one
and
and
a
little
bit
introduction
here
why
why
RFC
369
is
not
sufficient,
so
it
would
be
much
more
structured
and
much
much
more
convenient
to
read
and
would
follow
the
template
yeah
this,
and
so
our
suggestion
is
basically
using
this
kind
of
template
on
the
craft.
Bear
scales,
GitHub
repo,
and
maybe
we
need
to
spend
a
couple
of
hours
to
kind
of
build
a
simple
publishing,
Pipeline
with
Ali.
A
This
is
a
little
bit
where
you
come
into
play
like
maybe
you
want
to
prefer
or
maybe
want
to
be
consistent.
We
want
to
use
spec
MD
there.
You
know
make
like
a
second
D
pipeline,
that
that
is
consistent
with
the
other
ones.
That
might
be
good,
not
just
using
like
what
is
your
react
or
whatever.
So
these
are
some
details
to
talk
about,
but
this
is
this
is
what
we
have
MPS
to
take
criticism
or
ideas.
C
There's
definitely
like
two
very
separate
problems
here.
One
is
like
what
is
the
right
template
to
give
people?
What
information
do
we
need
to
have
a
good
scalar
proposal
and
then
the
separate
is
like:
how
do
we
build
the
website
where
these
things
come
together,
which
it
would
be
nice
just
for
you
know,
visual
consistency
to
use
spec,
MD
or
like
if
you
want
to
actually
the
thing,
that's
that's
quite
nice
is
that
it
allows
you
to
make
it
or
it
makes
it
easy
to
back
reference.
C
So
you
can
set
up
references
from
one
spec
file
to
another
spec
file,
and
so
it's
pretty.
A
C
But
you
know
appreciating
the
fact
that
there's
a
separate
novelty
piece
here,
which
is
like
aggregating
a
bunch
of
files
and
having
us,
build
step
for
each
and
like
having
a
main
page
that
pulls
them
all
together
and
like
spectrumsy,
doesn't
do
that
for
you.
So
that
would
be
a
new
thing
to
go.
Build
my
like
I,
think
your
frame
here
of
like
what
are
the
important
things
or
the
right
ones.
C
It's
like
you
need
to
Define
literal
inputs,
value
inputs
and
value
outputs,
which
does
not
necessarily
need
to
be
Json
for
what
it's
worth
it
just
needs
to
like
I
think
you
need
I
describe
the
fact
that
there
is
a
way
to
send
a
value
in,
and
then
you
should
also
probably
like
in
the
common
case
that
you
are
using
Json.
Here's
the
preferred
way
that
that
looks.
A
So
this
is
this
is
why
it's
interesting
like
when
we
work
through
it
I
believe
it
actually
has
to
be
Json,
because
we
don't
have
a
way
to
talk
about
other
things
in
a
implementation
neutral
wave.
So
this
is
the
observability
aspect.
I
know
what
you
mean,
and
this
doesn't
have
the
adjacent,
a
sealation
mechanism.
But
the
point
why
I.
A
Else,
yeah,
but
it's
it's
important
because
you,
you
need
to
have
a
Walmart
which
is
actually
understand
the
bill,
yeah
the
cross
implementation.
You
know
this
is
where
the
Json
aspect
comes
into
play.
This
is
a
little
bit
explained
more
here
on
on
the
implementation
guide.
There's.
C
There's
another
way
that
we
could
consider
kind
of
speaking
about
these,
which
is,
if
I
imagine
the
majority,
but
not
all,
but
the
majority
of
custom
scalers
are
actually
a
subtype
of
an
existing
scaler.
So
daytime
is
a
subtype
of
a
string
and
so
rather
than
saying,
like
here's,
how
you
serialize
this
as
a
literal
and
the
Json,
they
put
an
output
you'd,
be
like
it's
a
string
so
like.
C
If
you
want
to
see
how
you
manage
strings
in
graphql,
go
look
at
the
strings
chapter,
but
then
the
contents
of
the
screen
string
have
to
map
this
particular
DSL,
which
is
defined
by
you,
know
the
the
RFC
3337
or
the
particular
chapter
within,
and
so
that
way
you're
like
not
worrying
too
much
about
serialization
and
deserialization.
It's
just
like
it's
a
string.
C
Let
me
tell
you
the
encoding
of
that
string,
but
I
appreciate
that
not
all
scalars
will
fit
that
model
like
if
you
wanted
a
you
know:
arbitrus
Cherry,
Json,
blob
type
that
was
like.
That
probably
has.
That
must
be
a
Json,
because
it's
the
definition
of
it
and
that
may
be
something
else.
D
B
C
But
like
there's
a
difference
between
a
64-bit
end
is
a
number
that
is
defined
by
64
bits
and
then,
when
you're
serializing
it,
for
example,
in
Json.
Like
you,
you
don't
necessarily
have
to
say
like
you
are
not
a
64-bit
if
you're,
not
in
Json,
because
like
that,
doesn't
really
make
sense.
But
like
acknowledging
the
fact
that
99
of
graphql
use
uses
Json
serialization
in
a
Json
serialization
out
yeah,
then
obviously
we
have
to
speak
to
it.
C
D
An
interesting
aspect
of
this,
though,
is
that,
as
long
as
we
get
the
scalars
all
defined
for
Json,
including
our
existing
graphql
scalers,
that's
almost
entirely
to
the
point
of.
If
you
wanted
an
alternative
mechanism,
all
you
need
to
do
is
take
all
the
scalars
Json
input
output,
whatever
and
explain.
D
This
is
going
to
be
the
format
in
this
other
like
if
we
had
flat
buffers,
for
instance
like
okay,
what
would
what
would
a
string
look
like
in
flat
buffers?
What
would
a
int
look
like
in
flat
buffers?
It
would
be
like
32-bit
if
you
had
a
custom
64-bit
in
flat
buffer.
It
would
be
this
and
like
having
that
one-to-one
comparison
would
allow
us,
in
the
future,
to
build
like
to
build
clients
and
servers
that
can
just
you
know,
toggle
switch.
C
Thousand
percent
agree
with
that
it
is
if
Json,
then
it
must
be
this.
Not.
It
must
be
Json
unless,
like
the
literal
type,
is
explicitly
tossed
with
talking
about
something
in
which
case
like
sure.
But
if
it's
like
64
minute,
it's
like
okay,
you
don't
have
to
be
doing
that
in
Json,
but
if
you
do
do
an
a
Json,
it
must
be
in
this
format
like
100.
That
makes
sense.
A
Yeah,
if
I
understood
you
correctly
met,
you
want
to
have
like
a
different
civilization
spec
for
different
formats.
So
we
have
in
the
official
Graphics
package,
Json
definition,
how
things
are
represented
in
Json
right
and
if
you
would
want
to
Define
your
scalars
in
in
protobuf
or
whatever
you
would
have
like
a
definition.
How
graphql
is
seriously
in.
D
D
Somebody
could
entirely
describe
another
format
and,
just
like
you
launch
another
graphql,
serialization
deserialization
format,
merely
by
walking
through
that
list
and
implementing
that
description
for
each
scalar.
In
addition,
you
would
need
then,
in
addition,
objects.
So
this
is
almost
a
like
Siri.
It's.
D
A
Hey
well
so
how
I
would
see
it
is?
You
would
have
to
duplicate
this
format
in
a
different
like
sorry,
this
this
chapter
is
a
section
of
suspect
in
a
different
format.
This
is
what
I
I
would
think
about
it
like.
If
you
want
to
duplicate
this
and
talk
about,
maybe
not
yeah.
Well,
both
of
them,
probably
and-
and
you
want
to
talk
about
how
is
Franco
and
mapped
into
civilization
format
a
then
you
can
definitely
write.
Custom
scale
aspect
referencing
this
differentialization
format,
but
I
think
there's
like
this
observability
aspect
to
it.
A
Without
a
sterilization
format,
you
can't
really
talk
about
custom
scalars
because
it's
it's
like
it's
inherently
some
part
of
the
implementation
and
you
need
to
have
an
observability
Channel
if
that
makes
sense
and
to
talk
about
it.
This
is
why
Jason
came
into
playing,
but
yeah
I
agree
with
it.
It's
not
nothing
exclusively,
there's
nothing!
Fundamentally!
That
says
it's
only
Jason!
Yes,.
C
C
I
really,
like
of
you,
need
to
be
thinking
about
literal
values
in
deserialized
values
in
that's
Json
and
like
here's.
What
the
Json
looks
like
and
then
serialize
output
values,
which
again
probably
Json,
in
which
case
here's
what
your
Json
value
looks
like
like
that's
a
good
one,
two
three,
but
my
thought
on
the
a
decent
portion
of
custom
scalers
will
be
subtypes
is
to
like
maybe
there's
a
templatized
section
above
or
section
below
of
like
what
is.
C
The
format
of
the
thing
like
daytime
is
actually
a
perfect
example,
because
the
thing
that
you
don't
want
to
encourage
people
to
do
is
write
out
like
all
this
details
that
had
a
parse
a
date
time
and
like
the
literal
value
and
then
repeat
themselves
in
the
Json
deserialization,
because
they
the
goal,
is
to
make
them
as
similar,
if
not
the
same
as
possible
and
for
daytime.
They
should
just
be
the
same
like
there's.
C
Nothing
special
about
that
string,
value
aside
from
the
fact
that
it
like
meets
that
particular
like
RC,
33,
37
DSL,
and
then
you
can
have
like
a
one
sentence
on
each,
which
is.
This
is
a
Json
string
which
follows
this
format
listed
in
the
chapter
below
this
is
the
Json
trick.
You
know
and
then
you're
like
okay
chapter,
here's,
the
like
specific
sub
chapter
of
RFC,
3337
that
we
are
filling
in.
A
So
yeah
definitely
I
if
I
understood
you
correctly,
I
I
also
see
your
point
in
providing
a
simplified,
templating
or
straightforward
templating
for
for
narrow
down,
string,
values
and
daytime
would
fall
into
this
category.
Yes,
definitely
yeah.
This
could
be
even
simpler,
more
direct.
Yes,
it's
and
I
want
to
point
two
real
life
example.
We
experienced
why
this
all
this,
like
this
body
of
work
or
this,
this
thinking
about
and
getting
this
mental
model
right,
is
important.
A
First
of
all,
we
actually
had
incidents
involving
different
daytime
implementations
that
cause
problems
in
in
production,
so
that
that's
a
real
life
experience
and
then
a
second
example
is
probably
all
or
at
least
all
implementation
of
what
I
saw
of
Json,
for
example,
of
the
json's
custom
scaler.
This
actually
has
tricky
or
impossible
edge
cases.
For
example,
a
Json
scalar
allows
for
hyphens
inside
a
key.
Like
sorry,
a
Json
object,
the.
B
A
Can
be
hyphenated
and
a
Json
sorry,
a
craft
coil
object.
Literal
does
not
allow
for
hyphens
in
the
name
so
like
the
next,
so
the
the
possible
names
there
is
space
of
literals
does
not
map
directly
to
Json
name,
spacing
of
keys,
which
is
a
subtle
problem
but
which
we
also
encountered
in
real
life.
So
we
had
a
Json
value
coming
in
and
that
that
has
as
a
key,
for
example,
my
Miner's
house
is
key,
and
then
we
wanted
to
map
it
into
a
craft.
A
Bell
object
literally,
and
we
wanted
to
write
my
minus
object
as
a
name
or
as
a
key
of
this
object
and
the
this
was
not
a
valid
syntactic,
literal
and
graphical
anymore,
so
that
that's
that's
quite
a
subtle
thing
that
again,
yes,
HKC!
Yes,
not
a
lot
of
people
will
encounter
it,
but
I
think
it's
important
to
get
this
right.
Also,
if
you
truly
want
to
support
this
kind
of
things
across
ecosystems.
Just
to
give
you
two
to
things
where
we
are
coming
from,
why
we
invested
so
much
time
in.
B
C
C
And
I'm
I
think,
like
probably
like
a
good,
steady
state
like
how
we'll
know
we're
in
a
good
shape,
is
to
have
like
two
fairly
different
kinds
of
custom
scalers
as
like
clear
examples
to
pin
towards,
and
then
also
the
template,
because
the
template
just
says
fill
in
these
three
major
things
you're
like
great,
but
they
also
kind
of
want
to
know
like
mine,
is
a
little
bit
weird.
C
For
this
reason,
like
can
I
look
at
another
one,
and
the
two
primary
domains
of
weirdness
is
explicit,
some
type
of
string,
which
is
date
time,
which
is
what
motivated
all
this
in
the
first
place,
which
we
ought
to
just
put
up
and
then
also
like,
probably
the
Json
type
or
maybe
something
similar.
Maybe
it's
64-bit
is
another
good
one,
something
where
there's
like
some
subtlety.
That
needs
to
be
clearly
defined.
C
That's
like
very
clearly
a
different
kind
of
thing
from
a
daytime
that
would
give
like
a
great,
because
you
know
some
people
want
to
look
at
examples
to
pull
from
and
we
can
set
like
a
decently
High
bar
for
making
sure
that
those
look
pretty
good
because
whatever
you
put
up,
people
are
going
to
copy.
A
Yeah,
no
totally
true
that
makes
sense
again
to
have
simplified
versions
and
not
everybody
will
need
the
full
version.
Yes,
the
the
question
I
get.
The
question
would
be
okay,
we
have
this
implementation,
yet
we
have
a
idea
of
the
template.
What
would
be,
maybe
the
next
steps?
How
how
do
we
want
to
take
this?
For
what
like.
C
From
my
point
of
view,
you
are
unblocked
to
just
ship
something
Donna
I,
just
added
you
as
a
collaborator
alongside
Andy.
For
that
graphical
scalers
account.
Let
me
know
if
we
need
something
in
particular
to
set
up
domain
routing
or
anything
like
that
like,
but
if
we
have
a
a
website
clearly
working
but
at
the
wrong
domain,
like
that's
a
great
step
in
the
right
direction,
we'll
get
the
domain
stuff
figured
out,
but
I
think,
like
probably.
C
The
next
step
is
actually
like
getting
the
mechanisms
of
some
variation
of
spec
MD
in
a
index
page
generator
and
like
with
GitHub
actions
all
wired
up,
so
that
you
could
just
show
up
and
send
a
pull
request
with
markdown
file
and
once
it
lands,
it
becomes
a
URL
that
you
can
point
at
I.
Think
that's
probably
the
next
step
and
if
that's
something
you
all
feel
like
you're
ready
and
willing
to
be
able
to
go.
C
Do
that
then,
like
sweet,
make
it
happen
and
if
there's
something
there,
where
you
need
support
or
help,
then
feel
free
to
like
bring
that
back
to
the
group.
C
One
thought
I
had
is,
if
you
are
in
a
situation
where
you
need
help,
is
knowing
that
the
next
batch
of
the
graphql
community
Grant
is
going
to
open
up
sometime
soon.
C
One
of
the
things
we
didn't
do
last
time,
but
might
be
interesting
to
do
this
time,
is
to
write
like
rough
briefs
of
like
what
would
be
a
good
thing
to
apply
for
and
when
it
was
like.
Hey
I've
got
the
scale
this
project
you
want
to
come
build.
The
website
for
us
I
have
no
idea
if
anyone
will
bite
on
any
of
these,
but
if,
if
in
case
you
need
help,
then
that
might
be
one
Avenue
to
take
out
of
that
help.
E
A
Yeah,
like
I,
think
yeah
thanks
a
lot
for
listening,
I
think,
Donald
and
I
will
look
into
what
we
can
do
with
the
time
we
have
and
resources
and
and
effort
we
can
afford
to
make
something
separable
in
terms
of
producing
my
question
to,
for
the
people.
Kind
of
going
back
to
the
other
working
group
meetings
would
be
sharing,
maybe
a
little
bit
the
implementation
guide,
I'm
pretty
sure.
Some
of
our
most
speaker
theme.
B
A
The
terminology
is
not
completely
what
we
want
Even.
If
we
put
effort
in
to
make
it
nice
and
inconsistent,
it's
still
tricky
and
and
so
yeah
any
any
feedback
on
this
would
be
also
appreciated.
I,
don't
think
it
will
change
the
fundamental
layout,
what
we
need
to
Define
for
custom
Scala,
because
it's
still
these
three
things
but
yeah
having
feedback
on
the
implementation.
That
would
also
be
good.
B
A
A
I'm
just
excited
to
be
in
a
working
group
meeting
again
after
so
long
time.
It's
a
little
bit
of
the
first
one
which
we're
really
much
smaller
group
and
we
had
much
more.
There
was
no
agenda
to
read
and
items
and
it
was
much
more
unorganized
in
the
beginning.
So
yeah.
C
We've
really
dialed
it
in
since
then
awesome
yeah!
Well
so
I've
added
both
of
you
as
Andy
you've
been
on
there
that
Reaper
for
a
while,
but
Donna
you're,
now
a
collaborator
on
that
repo
too.
If
you
spot
anybody
else,
who
could
be
useful
that
you
want
to
add
just
let
me
know
and
I'll
get
them
added.
A
Sorry,
the
TCS
group
already
has
I
think
admin
rights
across
the
whole
world,
which
includes
this
repo
I
believe
so
so.
Matt
correct,
Michelle,
everybody
like
should
have
already
like
access.
If
you
want
to
do
stuff,
yeah.
C
That
is
very
true,
in
which
case
you
probably
have
plenty
of
access
controls.
In
the
case,
you
need
to
add
somebody
yourself,
yeah.
A
Which
I
think
so
I
think
a
down
the
road?
Probably
hopefully
we
get
there
ever,
but
there's
probably
something
to
figure
out
about
subdomains
and
stuff,
but
this
is
a
little
bit
further
ahead.
If
we
ever
get
there,
I
would
be
very
happy,
but
until
now
I
think
technically
we
are
very
unblocked
and
we
can
ship
things.
Yes,.
D
And
where
is
that
spec
MD
and
the
existing
documentation.
C
B
C
Does
yeah
if
there's
any
feature
ads
that
you
need
to
make
to
it,
feel
free
to
get
started
on
them
and
let
me
know
happy
to
edit
stuff,
but
right
now
it's
sort
of
like
a
Black
Box,
where
you
give
it
a
markdown
file
and
you
pipe
it
out
to
an
HTML
file
that
you're
going
to
put
it
in.
But
it's
just
like
a
command
line
tool
that
takes
markdown
in
and
HTML
out.
C
So
hopefully
easy
to
just
like
write
a
for
Loop
over
all
the
markdown
files
in
a
folder
and
then
there'd
be
there's
like
a
separate
net
new
thing,
which
is
building
sort
of
the
aggregate
like
making
an
index
page
that
lists
out
all
of
these
and
points
to
each,
but
it
probably
makes
sense
to
just
have
the
template
in
the
exact
same
folder.
That
way,
you
can
like
see
how
the
template
renders
on
the
website
simplifies
things
a
bit
and
then
you
can
have
an
index
page.
That's
just
I
would
imagine.
C
The
index
page
would
introduce
what
a
custom
scaler
is
point
to
the
relevant
parts
of
the
spec
that
you
need
to
while
doing
that
introduction
and
then
explaining
that
the
purpose
of
the
website
is
to
collect
scalars
and
give
them
canonical
URLs
here's
the
list,
if
you
want
to
add
your
own
here's,
a
link
to
the
GitHub
repo
done.
C
I'm
glad
we
talked
about
this
yeah.
This
has
been
sitting
a
little
stagnant
for
a
while,
but
if,
if
you
both
are
ready
to
inject
a
little
energy
back
into
custom,
scalers
I
think
that'll
be
very
good
yeah.
Once
it's
done.
A
C
I
think
this
is
the
other.
The
other
half
of
this
problem
is
like
getting.
The
documentary
in
one
place
is
great,
but
then
getting
them
used
in
the
primary
tools
that
people
use
is
I.
Think
the
other
half
the
battle.
It's
like.
We
go
there
next
and
I'm,
remembering
that,
like
full
cycle,
the
reason
why
we
get
on
this
was
because
we
tried
to
get
a
definite
date
time
type
and
then
big
surprise.
No
one
in
the
world
can
agree
on
dates
or
times
like
classic
programmer
problem
number
one.
C
But
if
we
had,
you
know
a
top
couple
of
them
and
had
them
integrated
into
Apollo
and
relay
and
graphical,
where
we're
like
I
see
you're
using
this
particular
variant
of
the
daytime.
Spec
I've
automatically
set
up
the
like
date,
encoding
for
you
or
like
I've,
set
a
validation
for
you
like
you
would
start
to
see
the
real
value
of
having
used
the
specified
by
tag
there,
and
it
would
also
like
pattern
match
the
right.
Okay.
C
This
is
what
it
looks
like
for
those
teams
who
maintain
those
tools
to
make
use
of
them
as
well.
So
that's
like
I
think
phase
two
we'll
get
there
when
we
get
there.
E
C
A
C
Sweet,
that's
our
agenda
for
today.
If
all
we
did
was
go
deep
on
this
one
thing
then
I
feel
good.
We
had
like
no
pressure
to
stop
early
to
move
on
to
the
next
thing,
and
we've
got
a
little
bit
of
time
left
over
so
I
said
we'd
be
here
for
roughly
an
hour,
and
here
we
are
right
at
about
an
hour
from
where
we
kicked
off.
I
think
we
did
it
all
right.
Folks,
thank
you
for
making
the
first
cities
a
success.