►
From YouTube: WG API Expression Bi-Weekly Meeting for 20200707
Description
WG API Expression Bi-Weekly Meeting for 20200707
A
A
B
C
So
this
is
a
slight
change
in
behavior
I.
Don't
think
a
lot
of
people
would
have
noticed
it,
because
it's
a
pretty
common
thing
to
do
so.
There's
some
pros
and
cons
listed
here.
We
went
through
like
seven
other
options
and
looked
at
a
lot
of
different
ways
of
dealing
with
it.
The
one
one
outcome
of
this
is
that
we're
probably
going
to
need
some
way
to
express
that
I
want
to
own
the
fact
that
a
field
is
unset
where
I
want
to
own.
C
The
fact
that
a
map-request
item
is
should
be
gone,
which
is
currently
not
expressible
very
well
and
to
express
that
we
can
and
we
can
introduce
the
idea
of
a
tombstone.
So
that
would
be
a
follow-up
to
this,
but
you
can
imagine
today
like
right
now.
If
there
are,
you
know
multiple
actors
that
are
working
on
an
associated
map.
There
is
right
now,
no
way
to
say
I
want
to
own
the
fact
that
this
map
item
shouldn't
exist.
There's
no
way
to
do
that
right
now.
So
that's
it.
C
That's
a
bit
of
a
gap
as
to
left
over
from
this,
but
what
this
does
do
is
it
makes
things
a
lot
more
consistent.
You
don't
have
to
remember
a
separate
rule
for
associative
map
with
then
four
fields
and
which
is
important,
because
not
sometimes
people
can't
even
tell
from
the
context
very
easily
if
they're
dealing,
what
or
the
other
I'm
just
have
any
single
rule
makes
things
less
important.
C
One
of
the
major
side
effects
of
this
is
that
if
you
are
doing-
and
you
are
a
field
over
to
a
controller
which
we
do
do
so
if
you
are
enabling
HPA
on
a
resource-
typically,
what
you
end
up
doing
is
happening
a
handover,
the
replicas
field.
So
it
used
to
be
that
a
user
or
supply
the
replicas
field,
and
it's
mainly
picking
the
value
now
they're.
Turning
on
the
HPA
they're,
yielding
control
that
field
to
a
controller
so
that
hand
off
is
really
important.
You
don't
want
to
accidentally.
C
Have
the
field
be
reset
to
its
default
of
one
there,
especially
if
you
have
like
a
large
set
of
replicas.
If
you
have
hundreds
of
replicas
or
something
you
said,
it's
a
one.
You're
gonna
break
your
service,
so
we're
we're
describing
how
to
do
that
transfer
very
carefully.
It
typically
works
out.
Well
because
what
will
happen?
Is
you
own
the
field?
The
controller
is
going
to
take
it
from
you
and
then
what
you're
going
to
do
is
you're
going
to
observe
the
controller.
C
Now
has
the
field
vendors
can
remove
it
from
your
pod
configuration,
so
the
strategy
actually
doesn't
work
really
nice
for
that,
but
it
was
different
than
the
behavior
before,
where
you
could
just
literally
abandon
it
and
when
the
controller
did
go
for
after
so
at
this
point,
well
transfer
transition
from
in
a
strategy
where
you
basically
can
abandon
something,
and
then
let
a
control
take
it
over
after
to
where
you
actually
have
to
do
a
field.
Ownership
transfer,
so
we're
gonna
have
some
steps
to
the
website.
C
So
it
it
can,
it
can
happen
on
accident,
but
you
do
I
mean
you
do
have
to
go
and
remove
the
field
right.
So
like
you,
you
can't
do
it
just
by
having
the
controller
take
over.
If
you
just
turn
on
HPA
a
pretty
same
things,
gonna
happen
and
you're
gonna
get
a
conflict
next
time
you
try
and
apply,
and
then,
if
you
resolve
that
conflict
by
removing
the
field,
do
we
get
the
right
thing?
A
A
D
D
D
C
A
A
B
Here
I
think
I,
don't
think
we've
seen
anyone.
So
today
we
are
tracking
everything,
including
the
future
gauge
fields
and
all
the
random
fields.
That
can
be
said
and
I
mean
we've
had
one
person.
That
is
the
status
thing,
because
Ferris
is
big
and
it's
very
common
to
be
raided.
I
think
we're
trying
to
fix
the
the
biggest
part
of
the
part.
E
D
F
B
A
B
A
A
B
A
A
A
B
A
A
A
E
So
if
you
look
at
the
you
know,
the
agenda
notes
for
the
API
machine
working
group.
There's
item
here
about
schema
verification
around
core
defaults.
I
think
the
idea
was
wanting
to
get
defaults
into
the
expression.
This
is
on
the
broader
topic
of
you
know.
We
have
this
kind
of
funky
semi
IDL
and
you
just
need
to
do
a
better
job
on
the
idea
and
the
idea
would
be
getting
defaults.
Expression
of
defaults
of
the
ideal.
A
A
B
A
E
C
G
E
Yeah
right
yeah
that
there
is
that
limitation
and
yeah
the
fact
the
matter
is.
We
have
behavior
that
has
more
complex
defaults
and
it
was
recognized
that
you
know
at
least
for
the
first
cut
we
just
have
no
hope
of
covering
at
all.
So
the
thought
was
just
you
know,
do
80%
job
and
get
constant
defaults
into
the
ideal.
Yeah.
C
It
gets
much
mileage
out
of
that.
It's
worth
tackling
the
stuff
that
we
can
like
things
that
are
static
or
both
for,
like
validation
and
for
defaulting.
It's
worth
trying
to
like
get
that
going
and
then
accept
that,
like
there's,
probably
always
going
to
have
to
be
escape
like
release
valves
for
things
that
are
only
expressible
in
code,
yeah.