►
From YouTube: GraphQL Working Group (Secondary, APAC) - 2022-12-07
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
C
B
B
A
D
B
Lee
doesn't
seem
to
be
online
on
Discord
I'm,
not
sure
if
that
is
an
indication
that
he's
not
gonna,
join
us
or
going
to
be
a
little
late,
I'm,
not
sure,
but
in
the
meantime
I
guess
I'll
get
us
kicked
off
hello.
Everyone.
Thank
you
for
joining
the
meeting.
B
By
being
here,
you
will
have
already
agreed
to
the
membership
agreement,
the
participation
and
contribution
guidelines
and
the
code
of
conduct.
If
you
haven't
done
so,
please
leave.
They
are
all
Linked
In
the
agenda.
B
Let's
go
through
and
do
an
introduction
of
attendees
in
the
order
that
they
are
in
the
agenda.
With
the
exception
of
me,
hello,
everyone,
I'm
Benji,
don't
know.
C
A
E
Seattle
it
sucks
here,
no
I
represent
Microsoft.
F
G
B
Awesome,
thank
you.
Everyone,
okay,
I,
think
Lee's
running
a
little
behind,
so
I'll
continue
to
lead
for
now.
First
of
all,
if
anyone
can
help
taking
the
notes,
that
would
be
ideal.
I'm
gonna
try
and
take
some,
but
if
others
can
help,
that
would
be
great
and
let's
have
a
quick
look
at
the
agenda.
B
The
live
notes
link
is
in
fact
wrong,
as
you
quite
rightly
point
out.
If
you
go
to
the
primary
meeting
instead
and
open
the
document
from
there,
that
is
the
one
and
going
forward.
That
is
what
we
tend
to
do.
We
use
the
primary
meeting
notes
for
all
of
the
meetings
for
that
month,
so
I
have
already
kitted
those
out
in
there,
but
yes,
I'll
share
a
link
to
that
in
the
chat.
B
Thank
you
for
the
shout
out,
Matt
cool,
so
reviewing
the
previous
action
items
and
then
Donna
with
the
licensing
and
final
legal
sign
off
for
the
graphql
scalers
project,
and
then
Roman
would
like
us
to
have
a
look
at
and
potentially
review
a
number
of
pull
requests,
all
of
which,
against
the
spec
I
had
a
glance
through
them
just
now
and
they're
all
quite
short
ones.
B
Excellent.
So
let's
have
a
look
and
see
if
there's
any
action
items
ready
to
review
nope.
So
just
loading
up
the
remaining
action
items.
B
D
Yeah
I
wanted
to
follow
up
on
the
custom
scale
of
specification
project.
The
last
thing
that
we
need
to
oh.
Firstly,
thanks
to
whoever
connected
the
URL
I'm,
not
sure
if
that
was
Lee
or
somebody
else
so
great
to
see
that
we've
got
scalers
in
the
URL
and
then
the
last
thing
to
sort
out
is
the
legal
stuff.
So
the
license
and
just
make
sure
that
we're
doing
everything
by
the
book
and
I
wasn't
sure.
If
do
we
go
with
the
standard
graphql
foundation?
B
Yeah
good
questions,
questions
which
I'm
sure
Lee
would
be
brilliant
at
answering.
F
Should
we
table,
we
could
just
table
this
to
the
end
of
the
meeting,
to
give
the
lead
time
because
it'll
probably
be
a
lot
more
productive,
yeah.
B
Cool
well,
in
which
case
over
to
you,
Roman.
E
I'm,
not
even
sure
hi
guys
so
I'm,
not
even
sure,
should
I
actually
do
it
in
this.
One
bring
it
up
to
this
meeting
and
it
was
just
I
want
to
clean
up
the
number
of
my
PRS,
which
are
approved
which
I
discussed,
but
they
are
just
sitting.
I
have
no
merge
rights,
and
so
basically
I
thought
it
just
may
make
a
push
through
here.
E
E
You
know
things
like
this
or
not
so
tiny,
but
basically
I,
don't
think
there
would
be
any
opposition.
But
if,
if
you
want
guys
I,
we
can
go
through
them
one
by
one
or
you
can
just
click
on
the
links
and
see.
B
Can
I
suggest
that
we
start
by
going
on
the
ones
that
have
been
approved
already
by
two
two
people
with
murder
rights,
I
think
they're,
both
TSC
members
me
and
probably
Michael
in
most
cases
Matt,
maybe
you
could
have
a
look
as
well
stamp
your
approval
on
them,
hello,
Andy,
welcome.
B
E
Check,
I'm
not
sure,
did
I
yeah
I
think
I
share
the
screen,
so
basically,
this
is
the
first
one
974
and
what
it
I'm
not
sure
about
up
the
approved,
but
at
least
one
approval
that
is
there.
So
basically
what
says
that,
in
the
implementation,
the
sentence
says
that
the
implementing
field
should
exactly
match
names
and
types,
the
old
phrase
we
actually
allow
covariance
in
implementation,
and
basically
this
is
the
fix.
E
F
I,
don't
for
this
one
specifically
I
think
that
the
existing
spec
is
correct.
So
this
is,
if
I
understand
this
correctly
and
I
could
be
wrong.
It's
that
the
types
available
the
like,
when
you
say,
interfaces
like
type
implements
x
y
z,
then
the
interfaces
This
Must
List
out
all
possible
interfaces.
I
could
be
wrong.
I
might
be.
F
The
implementation-
oh,
that
doesn't
change
so
just
change
it
yeah,
yeah,
okay,
I,
think
you're,
correct
I
have
to
I
have
to
read
through
more
context.
Yeah
I
I
will
review
this
further
I.
Think
you're,
correct.
A
B
I
also
think
that
you've
got
the
right
read
on
this,
but
I'm
not
100
certain,
which
is
why
it
doesn't
have
my
stamp
of
approval.
Yet
I
need
to
like
actually
yeah.
F
A
E
The
expansion
in
general,
the
implementation
rules,
which
surprised
me
early
all
kind
of
not
not
in
a
bad
way,
but
the
implementation
rules.
I
remember
from
the
early
that
of
the
interfaces
of
particular
Fields
is
that
type
matching
is
covariant,
so
basically
that
that's
the
general
rule
in
graphql,
more
relaxed,
much
more
relaxed
than
in
languages
like
dot
net,
so
basically
the
old
it
was
probably
left
over
from
some
earlier
times.
I,
don't
know
whether
other
times
yeah.
F
So
the
the
line,
the
sentence
going
to
covariant
types,
either
exactly
matching
or
of
covariant
types,
I'm,
pretty
sure,
you're
correct,
and
then
why
is
there
this
green.
B
F
E
Okay
about
uniqueness
of
directives
we
introduced
in
the
last
revision
we
introduced
repeatable
but
forgot
to
change
the
description,
and
basically,
we
are
still
about
that.
Repeating
directives
is
bad.
C
B
Proved
this
one
and
yeah
this
does
seem
to
be,
as
you
say,
a
leftover
have
you
scanned
through
the
spec
to
see
if
there's
any
other
locations
with
this
same
issue,
or
is
this
the
one
place.
E
What
the
same
I
just
went
through
the
entire
spec
and
tried
to
find
things
like
this
yeah
I
mean.
What
do
you
mean
the
same?
The.
E
E
For
it,
I
cannot
guarantee,
but
basically
I
was
just
going
line
by
line
and
trying
to
catch
all
these,
so
I
can
kind
of
see
Canada
well,
thank
you
about
directors,
but
kind
of
about
all
others.
Okay.
If
if
if
there
are
no
objections,
you
guys
have
a
look
at
it
again
and
then
please
approve
and
merge.
E
This
is
just
not
be
very
strange.
Capitalization
keys
purely
tutorial
looks
strange.
E
General
there
was
a
kind
of
capitalization
sentence
versus
first
sentence
versus
first
capital,
so
on
the
whole
section,
and
it
was
deep
then
in
the
entire
other
document.
So
the
entire
document
generally
follows
first
cap
on
every
word.
One
section
was
different,
so.
B
B
Just
merged
980,
because
yeah
that
seems
obvious.
E
E
A
E
Because
we're
going
from
execution
execute
yeah
okay.
So
if
no
objections,
let's
move
on.
B
Just
rolling
back
a
second
I
was
looking
at
981,
which
is
the
capitalization
of
subsections.
B
I,
see
that
you
fix
the
ones
in
section.
Seven
I
was
just
going
through
the
spec
itself
and
there
seems
to
there's
definitely
inconsistency
in
the
capitalization
of
the
headers.
It's
not
clear
what
the
capitalization
should
be.
B
B
So
I
think
we
should
make
an
editorial
decision
on
that
a
stylistic
decision
and
then
enforce
it
throughout
the
entire
spec
in
one
pull
request,
and
that
may
not
be
as
simple
as
it
sounds,
because
we
have
to
decide
which
style
guide
are
we
going
to
follow
for
capitalization
like?
Should
the
letter
a
in
and
be
capitalized
if
we're
going
for
sentence
capitalization,
maybe
we
want
to
use
like
the
Montreal
style
guide,
or
maybe
we
want
to
use
one
of
the
various
others.
B
E
Sure,
but
now
you
know,
I'm
actually
surprised
why
it's
only
this
I
believe
I
looked
through
the
entire
document.
You
say
there
are
still
different
case
of
capitalization
in
other
sections.
Maybe
it's
latest
some
kind
of
addition
edits,
but
at
the
time
I've
found
I
believe
I
found
only
this
or
maybe
there
is
something
lost
here,
but
I
believe
I
figured
out
that
throughout
the
document
most
cases
are
first
cap
and
I'm.
Not
an
expert
but
I
believe
that
when
you
have
first
cut
you
don't
capitalize
propositions
and
all
that
little
words
that.
F
There's
also
a
question
of
whether
errors,
like
errors
counts
as
something
like
in
the
kind
of
word
that
should
be
capitalized
or
not
given
the
rest
of
given
the
section
that
Benji
just
pointed
out.
So.
G
F
I,
basically,
I
think
the
action
item
is
that
we
we
should
exactly
as
Benji
said
we
we
should
have
an
action
item
just
like
we
need
to
decide
what
our
style
guide
is
and
then
we
can
apply
it,
but
this
is
like
this
is
good
that
the
pr
is
pointing
out
that
there
is
a
discrepancy
in
the
style.
That's
all.
E
E
F
I
mean
probably,
as
a
has
like
a
group,
we
should
decide
on
the
right
styling
but
yeah,
basically
yeah
either.
G
A
E
Then
just
who
is
there
in
charge?
Please
may
make
an
action
item
and
we
will.
We
can
leave
it
alone
for
now
and.
B
Anyone
can
file
an
action
item,
you
can
file
the
action
item
against
GitHub
and
you
can
write
it
into
the
notes,
I'm
taking
some
notes
on
the
issue
itself
at
the
moment.
So
if
someone
else
could
do
that,
that
would
be
appreciated.
E
E
F
I
think
that
this
is
actually
intended
to
be
changed
and
the
reason
why
is
that,
if
you
are
creating
a
graphql
service
supporting
subscriptions,
requires
changing
the
some
underlying
architecture,
some
underlying
structure
for
how
the
service
works
so.
E
I
didn't
have
it
and
I'm
building
from
scratch
with
subscription
this
and
why
you
are
assuming
that
first
I
built
the
whole
thing
without
subscriptions,
and
then
there
is
significant
change.
E
You
know
that's
it's
not!
Nothing
is
changing
from
the
very
beginning,
I'm
building
good
subscription,
the
fact
that
subscriptions
represent
the
challenging
task
and
difficult
task.
Yes-
and
this
is
a
kind
of
logical
here,
but
here,
but
in
the
old
version
we
kind
of
the
reader
reads
and
asks
why
you
guys,
assuming
that
I'm
changing
like
I
I'm
done
with
the
first
part
and
now
chapter
three
subscriptions.
F
I
I'm
yeah,
the
I,
wouldn't
say
this
is
wrong.
Honestly,
the
correct
thing
might
just
be
like:
instead
of
supporting
subscriptions
is
a
significant
just
like
subscriptions,
have
a
different
operating
model
like,
instead
of
like
being
up
I,
think
that
the
the
feedback
here
is
like
the
the
whole
first
line
is
probably
not
great
in
a
spec,
because
it's
like,
oh,
you
need
to
be
like
this
is
going
to
be
hard
for
you.
Here's
like
a
big
mountain
you're
going
to
have
to
climb.
Instead,
we
should
just
be
like
more
fact-based.
F
A
E
Well,
okay,
but
at
least
you
know,
the
I
I
would
agree
that
probably
we
can
even
remove
you
know
something
like
this
or
revise
the
sentence
substantially,
but
I
wanted
to
do
the
minimal
change
which
doesn't
actually
cause
the
question
change.
What
change
you
know,
challenge
yeah.
It
will
be
difficult.
We
just
give
you
up
front.
E
Okay,
let's
then
think
about
this,
if
we
kind
of
can't
agree
and
but
but
let's
let's
finish
this,
it's
it's
one
of
these
small
things
should
be
just
fixed
one
way
or
another
and
move
on
right.
B
Yeah
I
I
agree
with
Matt's
I,
mean
I've
said
it
before
I,
don't
like
saying
that
this
is
challenging
because
it
isn't
necessarily
challenging
I
understand
your
concern
with
referring
to
change.
I
would
say
most
people
start
out,
craft
kill,
schemas
with
queries,
and
then
they
add
mutations
and
then
later
they
might
add
subscriptions.
That
does
seem
to
be
the
general
pattern
that
people
follow,
but
you're
right
they
could
Implement
them
all
at
once,
in
which
case
yes,
it's
not
technically
a
change
but
I.
Think
Matt's.
B
Take
on
this
with
his
suggested
wording
is,
is
better
I've
summarize
that
in
a
rough
quote
that
I've
put
into
the
issue,
but
some
some
wording
like
that
that
expresses
maths
what
Matt
coined
there
I
think
is
going
to
be
more
effective
than
using
emotive
words
like
challenging.
E
Guys,
let's,
let's
figure
out
the
proper
sentence,
it's
kind
of
one
of
the
sentences
just
introductory
sentence
to
the
subject.
When
you
don't
have,
you
know,
don't
even
know
what
exactly
to
say:
okay,
you
are
starting
a
new
chapter
right
and
kind
of
kind
of
that
kind
of
blah
blah
blah.
You
know
without
much
meaning
but
but
to
to
have
some
introductory
statement
all
right
next,
so
here's
the
sentence
and
I'm
removing
in
parallel.
E
Essentially
it
says
that
merging
Fields
emerged
when
executed
in
parallel,
and
basically
it's
not
quite
logical
and
raises
a
question,
so
it
means
that,
on
the
one
hand,
the
server
has
a
choice
to
execute
it
in
parallel.
Similarly,
and
in
the
sense
the
result
should
not
matter,
but
if
we
live
in
parallel
here,
it
says
merging
should
happen
only
if
you
execute
it
there
in
parallel,
it
means
that
it
will
be
different
results.
E
B
E
I
know
currently,
because
in
this
section
it's
user,
the
reader
is
not
in
the
streaming
kind
of
area,
and
in
this
section,
when
we're
talking
about
just
good
old
query
with
the
two
keys
matching,
then
here
it
means
that
the
response
should
be
merged.
No
matter
what,
when
you
come
to,
you
can
put
a
note.
It
might
be
different.
You
know
in
the
other
section
when
there
is
a
streaming,
but
again
in
the
context
of
this
a
particular
section,
it
is
still
doesn't
depend
on
Parallel.
B
So
this
section
seems
to
be
about
value
completion,
which
is
what
happens
when
a
field
is
executed,
and
it's
talking
about
the
fields
with
the
same
response
name
that
are
executed
in
parallel.
There
may
be
more
that
are
executed
with
the
same
name
later.
That
I
mean
we're
talking
about
merging
selection
sets
here
and
value.
Completion
is
a
runtime
concern,
so
I
I'm.
E
C
E
F
Basically,
this
may
be,
we
shouldn't
be
using
parallel
in
this
whole
section,
but
within
this
section,
parallel
has
a
specific
and
the
purpose
and
meaning
of
parallel
here
is
discussing
when
you
have
within
the
tree
at
the
same
level
of
the
tree,
the
same
key,
which
parallel
is
probably
not
a
good
name
for
that,
because
it
invokes
parallel
processing
as
you're,
explaining
I
think
that
that's
correct,
but
we
keep
using
parallel
throughout
this
section
like
immediately
below,
we
have
an
example
operation
illustrating
parallel
Fields
with
the
same
name.
F
It
would
be
confusing
to
have
that
example,
operation
illustrating
parallel
Fields,
without
kind
of
the
context
that
we're
talking
about
executing
things
in
parallel,
I
agree
with
you
that
that
is
a
confusion
but
I'm.
Not
this.
That's
probably
specific
change.
I!
Think
that,
like
there's
some
word,
we
mean
for
like
same
level,
response
key,
like
response
keys
at
the
same
depth,
multiple.
E
What
does
me
parallel
again,
then?
Let's
be
careful
with
the
words,
and
maybe
there
are
somewhere
we
use
it
loosely,
but
in
certain
cases
it's
actually
become
confusing
because
parallel
in
many
contexts
you
know
it
actually
means
this
parallel
zero
execution.
E
So
and
in
this
particular
case
it
looks
like-
and
it's
actually
few
the
meaning
number
one
that
comes
in
mind.
What
we
put
into
this
phrase
is
that
a
parallel
execution,
because
it's
executed
in
parallel
it
immediately
causes
that
the
execution
versus
zero.
If,
when
it's
executed
zero,
then
it's
not
applied
I
would
say
so.
I
I
would
I'm
suggesting
to
improve
the
way.
At
least
we
use
these
terms
and
try
to
avoid
confusion.
E
F
E
But
but
but
there
is
one
there
is
one
thing
here:
it's
parallel
Fields,
it's
actually
the
context
is
slightly
loose.
You
know
the
first
phrase
that
I'm
fixing
is
executed
in
parallel.
This
is
the
direct
reference
to
execution
here.
It's
Loosely
used
term
parallel,
Fields
right
right.
F
A
E
B
B
If
we
change
the
wording
here
we
which
we
should
then
we
should
also
make
sure
that
a
similar
change
has
been
made
so
that
the
entire
section
is
consistent
and
I
think
using
response
names,
as
the
identifier
here
is
also
the
right
take
in
general,
because
that
is
what
governs
whether
or
not
it
should
be
merged.
I'm,
just
gonna
interject
for
a
moment
here
and
cut
back
to
Donna's
topic
for
a
moment
just
because
we
were
gonna
Circle
back
anyway.
B
So
Donna
Lee
says
that
he's
in
charge
of
all
of
the
things
that
need
to
be
done
and
I
believe
his
plan
is
to
basically
take
the
licensing
info
from
the
spec
itself
and
tweak
that,
for
whatever
the
needs
are
of
scalers
and
go
from
there.
So
if
you
don't
already
have
a
GitHub
issue
tasked
to
leave
for
that,
please
do
file
one
make
sure.
B
On
it,
if
you
want
ping
me
on
it
and
I
will
assign
him
to
it,
because
I'm
not
sure
if
you're
able
to
and
yeah
then
hopefully
we'll
be
able
to
get
that
sorted.
B
There
is
also
sorry
did
you
want
to
say
anything.
A
reply
to
me.
D
Yeah
and
also
I'm,
just
mindful
I,
don't
want
to
keep
everything
on
Lee.
If,
if
there
is
somebody
at
willing's,
Foundation
or
elsewhere,
where
I
can
help
out,
if
it
is
possible,
then
I'm
also
happy
to
pitch
him.
B
I
think
you
could
probably,
if
you
wanted
to
open
like
a
pull
request,
but
don't
actually
merge
it
and
then
have
Lee,
be
the
one
who
presses
the
big
old
merge
button.
C
B
Yeah,
okay,
yeah
awesome,
so
the
other
thing
that
Lee
has
reminded
me
to
tell
everyone
about
and
I'm
afraid
we
missed,
Russell
and
Andy.
For
this
is
that
we
have
the
TSC
elections.
They
are
open.
Now
they
were
delayed.
B
B
So
if
you
want
to
nominate
yourself
to
be
a
TSC
member,
you
can
absolutely
go
and
do
that
and
then
votes
will
be
taking
place
early
in
January
and
the
the
new
TSC
members
will
be
starting
at
the
very
beginning
of
February,
and
we
know
that
I
think
at
least
three
of
the
TSC
members
that
are
of
the
five
that
were
going
through
the
I
should
I
should
reverse
this
at
the
moment.
B
B
So
of
the
five
that
are
renewing
this
year,
three
of
them
have
decided
not
to
renominate
themselves,
so
there's
definitely
space
on
the
TSC.
If
you
feel
that
that
is
something
that
you'd
be
interested
in,
if
anyone
wants
to
ask
any
questions
about
the
TSC,
please
feel
free
now.
Otherwise,
we'll
go
back
to
Romans
topics.
B
I
think
just
look
for
the
words
parallel
and
find
a
consistent
wording
that
that
deals
with
all
of
them.
E
F
Of
honestly,
you
pointing
it
out
and
us
getting
agreement
that
this
is
not
correct
and
going
through
this
process
is
value
and
the
like
I
can
take
an
action
item
on
like
rereading
through
providing
providing
like
specific
or
even
creating
another.
E
Suggesting
specific
suggestion
specific
risk
right
exactly
that
will
be
great
yep,
okay,
so
action
item.
A
E
An
execution
in
mind
and
finally
I,
do
not
suggest
this.
This
is
not
in
the
list
on
the
agenda.
I
do
not
suggest
to
actually
push
it
to
something,
but
this
is
kind
of
had
been
hanging
there.
E
E
E
E
Excuse
me
so
I
I
want
to
just
then.
Finally,
in
this
can
I
ask
one
thing,
so
you
you
guys
say
that
this
is
different
from
what
you
think
can
I
just
see.
You
say
that
this
is
just
a
fragment
right.
That's
not
the
full
query.
C
E
No
I
mean
as
the
query
as
the
query
text.
It
is
full,
can
I
see
please
the
full
text.
E
E
B
Okay,
I've
written
into
the
zoom
chat
two
different
operations:
graphql
operations,
the
first
one
is
a
mutation
operation
with
the
selection
set
that
has
the
field
Foo
in
it.
The
second
one
is
a
mutation
operation
that
has
a
selection
set,
which
then
has
a
fragment
spread
in
it.
That
fragment
spread
is
a
fragment
that
is
on
the
mutation
type
and
it
has
a
selection
set
that
contains
the
field
Foo.
Both
of
those
selection
sets
that
contain
the
word
Foo
they
are
selection
sets
that
will
execute
serially.
B
But
at
the
moment
only
the
only
selection
sets
that
execute
in
serial
at
the
moment
are
those
found
on
the
root
level
mutation
type,
but
that
doesn't
necessarily
mean
that
that
will
always
be
the
case
in
graphql,
for
example,
there's
been
a
lot
of
discussion
about
allowing
nested
serial
execution
types
to
allow,
for
example,
name
spacing
of
mutations.
So
what
is
being
discussed
in
this
part
of
the
specification
is
execution
of
a
selection
set.
So
what
we
have
here
is
a
selection
set
with
a
field
that
is
going
to
be
executed
serially.
B
E
You
can
skip
the
operation
type
and
this
looks
like
absolutely
legal
request
with
the
just
missing
operation
type,
which
is
query
and
that's
what
is
confusing
I
understand
what
is
discussed.
It's
the
example
itself
again
this
whatever
might
happen
in
the
future,
next
invitation
and
so
on,
but
just
because,
in
the
context
of
this
particular
example,
I'm
following
on
I'm
reading,
the
explanations
I
see
the
example
I
expect
to
see
the
example
as
most
in
most
cases
that
this
is
entire
request.
E
Oh,
this
is
missing
operation.
Okay,
query
is
optional,
I
put
query
in
front
and
what
I
see
is
completely
illogical,
so
basically
to
make
to
land
it
in
the
context.
So
in
your
example,
you
show
that
it's
actually
might
be
mutation
on
more
a
little
more
advanced
example
when
it's
actually
a
fragment
so
leaving
this
more
advanced
example
with
a
fragment,
but
as
an
example.
Let's
just
put
mutation
there
and
then
all
the
questions
go
away,
because
the
confusion
is
that
I'm.
E
B
E
Course
you
can
no,
you
can,
because
you
are
executing
them
all
this
thing
on
the
result
of
the
result
of
the
result
of
something,
and
it
cannot
be
mutating
method.
As
far
as
understand,
mutating
method
can
be
only
on
the
top.
F
E
This
is
how
confusing
many
people
hit
this
I
don't
understand.
First
of
all,
if
we
put
mutation
there,
this
would
be
example
that
works
for
this
case
as
an
illustration
and
so
on.
It
won't
kind
of
change
in
Istanbul.
It
just
remove
all
this
mental
stop
points
that
not
only
me,
but
people
have
just
remove
this
mental
stop,
stop
Point,
even
if
you're
right
right,
it's
actually
not
exactly
mutation,
it's
selection
song,
but
it
it
will
be
absolutely
clear
and
not
non-stop
experience.
B
E
The
problem
is
that
it
is
Operation,
because
change
birthday
is
mutation
method.
It
can
be
only
under
operation
or
under
your
whatever
fragment,
which
is
a
bit
more
advanced
and
I
think
should
not
be
brought
up
here,
but
basically
change.
The
the
trigger
is
the
name
of
the
method.
If
you
put
food
there,
nobody
would
say
anything
but
change
method.
It
means
that
I'm
On
Top
I'm
right
under
mutation.
Why
mutation
is
missing.
F
I
I
think
this
is
you're
you're,
hitting
at
like
weaning
a
way
to
introduce,
because
we
do
this
actually
throughout
the
spec,
where
we
operate,
we
we
give
you
Snippets
that
are
selection
sets
rather
than
full
fragments,
rather
than
full
operations
rather
than
full
documents,
and
possibly
what
we
need
is
a
like,
quick,
intro,
somewhere
of
like
hey.
When
we
talk
about
a
selection
set,
it's
the
sub
tree
of
the
like
we're
talking
about
a
sub
tree
rather
than
the
full
context,
the
full
query:
the
full
mutation.
E
F
We
could
we
could
have
multiple
selections
just
because
you're
at
the
top
level,
you
can
have
multiple
top
level
selection
sets,
and
that's
right,
like
that's.
That's
actually
important
thing
about
graphql
is
that
you
can
have
multiple
selection
sets
that
will
need
to
be
merged
together
and,
and
you
can
have
them
anywhere
in
at
any
level
of
the
tree.
E
E
C
B
F
B
F
B
In
the
examples
we've
just
put
in
the
in
the
chat,
those
are
nested
selection
sets
but
they're
not
on
the
result
of
any
other
field.
They're
just
nested
selection
sets
they
could
have
like
directives
applied
to
them
or
things
like.
C
E
Spread
why
you
are
bringing
up
this
complicated
example,
because
yes,
many
other
things,
but
in
example,
you
should
put
essentially
the
simplest
possible
thing
which
illustrates
your
thing.
You
can
invent
things
and
so
on
and
say
this
is
just
but
essentially
in
the
straight
thinking
of
reader,
who
is
possibly
reading
this
for
the
first
time
it
really
confuses,
and
basically
even
if
what
you're
writing
is
possible,
maybe
but
I,
don't
think.
Then
it
is
a
good
example
to
keep
in
mind.
E
Oh
I
I
know
such
things
is
possible
in
graphql
and
you
you
are
trying
to
bring
it
up
in
the
background
on
the
user
on
the
reader
who
is
reading
it
the
first
time
and
the
mod
the
model
he
has
in
mind.
Is
that
oh,
it's
rotation?
That's
strange
one
Donna!
You
wanted
to
say
something.
E
F
We
need
the
spec
to
Encompass,
even
the
edge
cases
so
like
we
need
the
spec.
The
spec
kind
of
has
a
dual
role
right.
The
spec
has
a
role
in
educating
people
on
how
graphql
the
system
works,
but
also
in
being
so
descriptive
that
you
cannot.
If
you
follow
the
spec
a
word
for
word,
you
will
cover
all
of
the
possible
cases
that,
like
a
graphql
server,
you
will
not.
F
There
will
be
no
ambiguity
in
how
your
graphql
execution
operates
and
that
I
I
think
this
specific
example
probably
you're
getting
at
like
this
is
not
a
great
intro
example,
because
it's
confusion,
it's
confusing
that
this
is
not
like.
We
gave
nothing,
and
this
is
as
you're
pointing
out
this
is.
This
looks
like
a
valid
query
document
right,
because,
if
there's
no
because
earlier
in
the
spec,
if
there's
no
operation,
then
you
assume
that
it's
a
query
right
as
you're
saying,
if
you
just
paste
this
into
graphical
graphical,
will
treat
it
as
a
query
and.
A
F
We're
right
so
like
yes,
this
is
not
copy
pasteable
into
graphical,
as
is,
and
have
it
just
work,
but
we
do
this
a
lot
in
the
spec
where
we're
just
using
selection
sets.
We
use
selection
sets
that
aren't
even
Types
on
the
query,
root
type
right
or
like
where
we
have
fields
that
are
the
top
level
fields
in
the
selection
set
that
are
not
fields
on
query.
E
There's
you
see,
do
we
actually
need
the
mutation,
because
we
are
talking
about
in
general,
zero
execution
not
attached
into
the
streets?
What
what
triggers
is
this
combination
absence
of
operation
and
the
word?
It
means
that
you
know
you
guys
are
missing.
You
know
if
you
put
there
get
the
address,
read
some
read
method
in
the
cont
as
an
example
in
the
context
of
what
you're,
explaining
I
think
it
works
as
well.
It's
just
the
two
read
operations
working
as
in
serial
executed
in
serial
that's
yeah.
There
would
be
no
questions
right.
E
F
Read
it's
possible.
We
should
do
a
dot
dot
dot
on
mutation,
for
instance,
which
would
make
it
clear
that,
like
hey,
this
is
a
selection
set
specifically
on
mutation,
in
which
case
we
would
probably
want
to
update
all
of
the
selection
set
examples
that
we
have
to
do
dot
dot
dot
on
the
parent
type
throughout
the
document,
but
yeah
and
like.
E
But
again,
why
not
change
just
the
name
of
the
method
where
it
should
be
mutational,
use
the
read
methods,
kind
of
not
non-mutating
methods
as
names
for
zero
execution?
Well,.
E
It's
just
the
mental
stop,
even
if
all
you
say
is
right
and
in
some
way
I
mean
educated
way
clear,
but
this
is
a
kind
of
huge
mental
stop.
Yeah.
F
E
F
It
may
be
the
case
of
what
we
really
want
is
like
runnable.
If
you
look
at
like
the
rust
documentation,
for
instance,
their
their
Rust
book
The
Code
Snippets
are
like
runnable
examples
and
if
we
could
get
closer
to
that,
that
might
that
might
help
like
unblock
but
I.
Don't
I,
don't
know
exactly
how
we
would
do
like
that.
That's
a
pretty
big
lift.
E
Again,
the
simplest
way
put
mutation
there
and
it
works
as
an
example.
Simpler
then
less
generous
but
simple
example.
Example
is
something
to
be
simple
and
it
will
be
absolutely
correct.
Request.
No
mental
stops,
no
questions
and
everybody's
happy.
That's
what
I
don't
understand.
Simple
fix
put
mutation,
just
simplify,
maybe
the
example.
B
F
E
E
I
think
right
now,
it's
it's
not
what
I
mean
allow
it's
allowed
anyway.
So
it's
up
to
the
server
it's
up
to
the
server,
how
to
execute
it,
but
at
least
down
there.
But
the
thing
is
to
be
honest:
I'm
in
the
future
world
I
would
actually
attack
the
whole
concept
of
order
of
execution.
E
It's
just
in
general
in
the
future.
You
know
nothing,
but
it's
it's
the
future
world
I
right
now,
I'm
concerned
only
with
this
okay
I
think
we
discussed
enough,
but
still
at
least
we
are
aware
and
agree
that
there
is
something
confusing
here:
mental
stop
right
and
sitting
there
and
thinking
what
the
hell
right.
E
F
Or
we
haven't
set
the
context
correctly
above
in
what
the
examples,
because
there's
actually
a
bunch
of
like
the
example
above
of
just
birthday
and
address.
That
example
is
not
meant
to
be
a
query
and
if
you're
reading
it
like
a
query,
then
we've
set
the
expect
like
we've
set
your
expectations
when
you're
reading
incorrectly
and
I
believe
we
have
other
selection
set
examples.
Wait.
E
E
E
F
B
So
to
to
expand
on
that,
this,
you
wouldn't
typically
have
like.
Maybe
a
birthday
field
right
at
the
root
level,
you
would
have
a
user
that
you
select
or
a
person
that
you
select
and
then
they
would
have
a
birthday
or
an
address.
So
this
is
again
a
selection
set.
It's
not
an
operation
or
a
document.
It
is
just
a
selection
set
that
we're
talking
about
and
Matt's
right.
B
We
need
a
better
way
of
indicating
that
I
mean
we
do
have
it
with
the
other
example,
where
we
explicitly
State
below
that
that
it
is
a
selection
set,
but
that
still
doesn't
seem
to
be
enough.
So
I
wonder
I
mean
I.
Think
runnable.
Examples
is
a
nice
goal
to
have
with
quite
a
lot
of
effort,
but
I
wonder
if
we
can
do
something.
That's
a
little
subtler
like,
for
example,
adding
a
title
or
a
caption
below
it
or
even
a
non-normative.
F
Selection
set
is
meant
to
imply
part
of
query
rather
than
whenever
we
say
selection
set
up
in
front
of
an
example.
That's
the
intention.
I
agree
like
so
right.
One
thing
we
could
do
is
around
the
blue
example
text
or
purple
example
or
Violet,
whatever
color,
that
is,
we
could,
for
instance,
expand
the
like
fragment
on
user
and
then
have
the
example
like
have
the
example
be
inside
of
that
fragment,
for
instance,
if
that
makes
sense.
So
it's
like
the
example
is
the
selection
set
right.
A
E
F
Yeah,
okay,
so
so
with
that
one,
for
instance,
it
would
be
clear
that
we're
talking
about
a
selection
set,
not
a
query.
If
we
had
like
a
little
wrapper
around
the
Violet
box,
that
was
like
fragment
on
fragment
user
on
user
birthday
on
user
and
then
the
example.
E
And
why
not
dance
show
that
the
entire
thing,
rather
than
this
fragment
in
the
first
case?
Yes,
that's
logical!
This
is
this
is
not
the
whole
thing
right.
It
can
be
down
the
road
somewhere
in
the
tree
in
the
request
tree
query
tree,
but
in
this
second
case
change,
but
the
immediate
indication
I'm
top
level
right
in
this
open
Embrace.
Something
is
missing
there.
Why
didn't?
They
just
put
it
to
show
the
whole
thing.
You
see,
you
see
the
logic?
E
Okay,
so,
let's
just
stuff
I,
don't
know
continue
discussion,
but
offline
come
up
with
something
because
it's
still
hanging
there
and
it's
still
mental
stop.
They
know
something
that
needs
to
be
fixed.
F
A
F
E
But
no,
no,
no,
no
I,
I!
Think
here
it's
a
kind
of
combination.
It's
not
like!
Oh
others.
Examples
are
that
bad.
They
are
good.
Actually,
yes,
they
are.
They
assume
a
lot
but
they're
good
this
this
one.
Just
these
two
jump
up
this
and
the
next
one.
So
basically
I
I
wouldn't
spread
it
so
much
on
others,
yeah,
let's
just
decide,
maybe
okay,
we
can
leave
it
to
final,
but
at
some
point
we
will
have
to
kind
of
fix
it
right.
D
A
E
B
No,
that's
it
from
me.
Thank
you
for
the
the
discussion
and
anything
we
can
do
to
improve.
The
spec
is
good,
but
I
think
here
the
the
focus
needs
to
be
on
clarifying
that
what
we're
looking
at
is
a
selection
set
and
why
that
is
the
case
rather
than
turning
it
into
a
a
document.
B
I
think
that
will
give
us
the
best
stead
for
the
future
to
enable
those
future
discussions
about
extending
serial
selection
set
execution
to
other
areas
of
the
schema,
which
has
been
mooted
quite
a
number
of
times,
but
yeah.
It's
something
here
to
improve
Clarity
I
think
would
be
good
because,
as
you
point
out,
Roman
is
not
the
only
person
to
have
tripped
over
this.
There
have
been
three
or
four
people
that
have
tripped
over
this
I
think.
B
Cool,
thank
you.
Everyone
and
I
guess
we'll.
E
See
you
at
the
next
working
group
just
general,
ask
guys
those
of
you
who
have
merged
kind
of
Rights
permissions.
If
you
see
a
few
items
sitting
there
like
approved
PR,
then
after
sitting
there
for
a
month,
you
know
just
go,
look
look
over
and
a
few
if
they
are
approved
and
they're
ready
to
merge,
then
just
merge
them
like
half.
E
Discussed
today
they
were
just
primitive.
F
Yeah,
some
of
them,
some
of
them,
if
they're
merely
editorial
like
most
of
the
ones
we
discussed
today.
I
think
we
are
allowed
to
just
merge
them.
So
part
of
this
is
probably
like
we've
kind
of
historically
leaned
on
Lee
to
do
that
and
the
rest
of
the
TSC,
like
somebody
I,
believe
it
has
to
be
somebody
from
the
TSC
who
does
the
merge
and
the
rest
of
the
TSC
has
less
history
and
just
in
just
merging
editorial
things
so
I
believe.
F
Typically,
it
would
like
if
it's
not
just
a
merge,
it
requires
a
vote
and
like
a
little
bit
more
process
so
that.
F
E
C
B
The
other
thing
to
factor
in
is
that
one
approval
isn't
necessarily
enough
to
merge
something
into
the
spec
like
there
is
a
high
bar
to
merge
something
into
the
spec,
so
unless
it
gets
approved
at
an
actual
working
group
where
everyone
on
the
call
has
all
agreed
that
this
should
be
merged,
then
the
fact
that,
like
for
example,
I've
put
my
stamp
on
it,
does
not
necessarily
mean
that
I
haven't
missed
something
critical
there,
and
so
I
tend
to
wait
for
at
least
there
to
be
three
approvals
from
significant
working
group
members
and.
F
E
F
Yeah,
it's
Things
fall
through
the
cracks
and
we
probably
just
need
to
be
like
get
things
into
our
notifications
a
little
bit
better
and
there
might
be.
There
might
be
some
way
that
we
can
put
like
waiting
for
TSC
as
a
tag
or
something
like
like
there's,
probably
some
process
in
GitHub
that
we
could
do.
That
would
give
us
basically
things.
Tsc
needs
to
operate
on
as
action
items.
C
B
Like
but
it
might
actually
be
worth
scheduling
some
time
during
one
of
the
meetings
each
month
to
you
know
have
yeah.
B
C
B
Certain
certainly
is
quite
challenging
for
I
know
a
number
of
the
TSC
members
to
be
able
to
keep
up
with
the
quite
significant
amount
of
traffic
that
the
repositories
Garner
I
know
at
me.
At
the
moment,
I've
just
looked
at
my
GitHub
inbox
and
I've
got
189
unread
notifications
in
there
at
the
moment
and
I'm
one
of
the
ones
who's
actually
approving
some
of
these
things
right.
B
It's
like
there's
a
lot
of
stuff
going
on
and
it
is
hard
to
keep
up
with
no
I
understand
but
Matt's
right.
If
we
can
find
things
where
someone's
already
done,
you
know
like
if
one
of
the
TSC
members
has
already
approved
it,
there's
a
higher
chance
that
other
members
are
probably
going
to
go
yeah.
Actually,
this
does
look
fine,
so
surfacing
those
in
a
better
way
like
Matt,
says
I
think
would
be.
It
would
be
good
to
come
up
with
a
process
for
that.
E
Well,
let's
think
about
one
of
these
retails
Maybe
by
Lee
in
the
beginning,
look
at
the
queue
or
there
are
some.
We
will
spend
a
day
on
some
time.
Looking
at
PRS
or
I,
reminding
TSC
members
I,
don't
know
it's
just
yeah
some
routine.
There
are
already
PR's
waiting
for
the
second
third
approval
and
to
be
merged.