►
From YouTube: Gateway API GAMMA Meeting for 20230606
Description
Gateway API GAMMA Meeting for 20230606
A
A
For
folks
who
don't
know
we
have
an
open
Agenda,
so
please
feel
free
to
add
any
topics
you'd
like
to
discuss
to
the
meeting
notes,
and
also
please
add
your
name
to
the
list
of
attendees.
So
we
can
track
off
and
go
into
these
meetings
as
we
try
to
make
sure
that
we're
working
at
time
zones
that
are
able
to
accommodate
also
a
range
of
hotels.
A
All
right
with
that,
we
can
go
into
a
brief
recap
of
last
week
to
start
out,
I
was
not
able
to
attend.
Last
week,
I
watched
the
last
eight
minutes
of
the
recording.
A
It
looks
like
the
first
half
of
last
week's
meeting,
didn't
quite
make
it
into
the
YouTube
playlist.
Yet
I
don't
know
if
there's
how
you
think
to
follow
up
on
there.
What
was
just
going
to
be
missing?
Oh
you
have
it
locally,
okay,
cool.
A
So
hopefully
we
can
get
that
up
at
some
point,
but
yeah
I
looked
like
kind
of
like
the
big
highlight
and
takeaway
and
celebration.
Is
that
our
gamma
ready
for
a
bunch
of
Milestone
closed
and
we
are
on
track
for
heavy
gamma,
officially
becoming
part
of
Gateway
API
in
the
upcoming
raid
over
lease,
which
I
am
super
excited
about?
A
B
A
A
There
is
one
issue
that
had
been
in
the
Milestone
that
has
been
removed.
So
it's:
oh,
that's
the
wrong
one!
That's
almost
done
there
we
go
so
this
issue
on
how
gateways
should
not
interact
with
gamma
running
configuration
is
still
work
in
progress.
I've
been
working
on
a
draft
with
Flynn,
but
we've
determined
that
this
should
not
block
gamma
being
part
of
the
spec,
because
there's
a
lot
of
complexity
to
this
and
it's
something
that
will
probably
require
some
implementation:
experimentation
to
figure
out.
A
What
kind
of
makes
sense
is
implementable
for
everybody,
so
this
is
Alice
Milestone,
but
we're
going
to
continue
to
work
on
it
and
yeah
with
that
Rob
I,
don't
know!
If
there's
anything
you
want
to
highlight
from
the
oil
Milestone
that's
relevant
to
gamma,
but
this
is
going
to
be
the
place
for
adding
anything
that
we
want
to
do
for
gamma
before
the
080
release,
part
of
it.
So
there's
one
issue
on
website
updates
that
I
think
would
be
great
to
get
in
there.
C
Yeah
yeah
thanks
I
I
think
that
080
is
really
the
the
release
that
we
want
to
leave
for
gamma
and
so
in
that,
in
that
case,
I
think
we'll
rely
heavily
on
when
y'all
say
it's,
it's
ready
to
go,
there's
a
couple.
Experimental
gaps
coming
down
the
pipeline
that
may
get
included,
but
there
aren't
many
release
blockers
here
that
are
not
gamma
related,
so
or
I.
Don't
think
I
don't
even
know
if
we
have
any
gamma
related
ones
other
than
documentation.
A
C
A
Yeah
and
I
just
caught
from
the
recording
last
week,
the
the
target
for
this
is
about
four
to
five
weeks
from
now
from
now,
maybe
three
to
five
weeks
from
now
for
entering
into
Sig
Network
API
review.
So
hopefully
everything
should
be
merged
and
basically
getting
to
a
point
where
we're
ready
for
folks
to
give
it
the
green
light
foreign
all
right,
what
else
yeah,
Keith
or
Rob?
Do
you
want
to
recap
any
of
this
conversation
from
last
week
about
like
experimental
versus
implementable.
D
Yeah
there
was
a
good
point
brought
up
by
by
Shane
last
week
around
you
know.
The
timing
of
gamma
was
was
experimental
a
little
while
ago
and
now
we're
moving
quickly
to
implementable
and
just
wanting
to
make
sure
that
that
timing
makes
the
most
sense
bit
back
and
forth
on
some
of
the
semantics
of
experimental,
implementable,
I.
Think
Rob
made
kind
of
the
seminal
point
that
really
gamma
was
implementable,
probably
a
while
ago
and
evidence
has
been
implemented.
D
Exactly
like
before
the
PA
merged,
it
was
implemented
by
istio,
of
course,
ran,
ran
ahead
and
did
it
really
early,
but
the
Linker
D
followed
relatively
soon
I
think
it
was
either
really
early
this
year
or
late
like
late
last
year,
with
a
with
a
gamma
compliance
content
implementation.
So
we've
really
been
incubating
for
a
while,
and
it's
been
implementable
for
a
good
bit.
D
We
just
kind
of
made
an
official
late
and
so
I
talked
to
Shane
after
the
meeting
last
week,
and
you
said
like
he
doesn't,
have
any
kind
of
remaining
hesitations
and
you
can
go
ahead
and
move
it
to
you
know
to
implementable,
but
he
just
wanted
to
make
sure
he
thought
about
it
and
I
agree.
It's
it's
good
to
have
the
conversation,
but
I.
Think
I'm
I
agree
with
Shane
that
this
that
this
that
this
movement
makes
sense
even
with
a
quick
turnaround.
A
Cool
yeah
I
think
additionally,
Kuma
is
also
making
good
progress
on
hunting
gamma
already.
So
that's
that's
also
been
encouraging
to
see
that
we
have
a
diversity
of
implementations
that
are
starting
to
explore
this,
and
it's
still
on
the
roadmap
for
console.
We
haven't
gotten
into
it
exactly
yet,
but
hopefully
we'll
be
able
to
start
that
in
the
near
future.
Now
that
we've
got
a
lot
about
the
work
out
of
the
way
for
us,
so.
A
All
right:
do
you
want
to
talk
about
off
the
policy
teeth?
I!
Guess
that
sorry,
that
that
covers
kind
of
the
recap
of
last
week
and
we
can
move
on
to
topics
for
this
week.
D
Yeah
so
looking
at
you
know
where
we've
got
for
where
we're
at
with
gamma,
because
we've
got
at
least
a
implementable
set
of
routing
semantics
to
find,
which
is
which
is
great.
For
me.
Looking
at
mesh
use
cases,
the
biggest
Gap
currently
is
something
standard
for
operation
policy.
D
We
have
talked
about
this
a
couple
of
times
and
we
talk
about
a
little
bit
about
how
they're
going
about
doing
it.
But
one
of
the
reasons
why
this
is
exciting
to
me
now,
not
just
because
we
don't
you
know
not
just
because
you
close
kind
of
closed
the
chapter
for
now
on
on
the
routing.
But
you
know
we
have
a
lot
of
work
happening
with
sort
of
the
Upstream
or
the
ingressing
API
side
with
policy
attachment
threat
policy,
and
things
like
that.
I
think.
D
We've
gotten
I
even
saw
some
I
couldn't
capture
a
pair
meeting
yesterday,
but
I
think
I
saw
some
information
on
there
around
essential,
tooling,
that's
being
created,
different
approaches
being
done
and
I
think
it
would
be
really
great.
D
You
know
for
us
who
want
to
see
kind
of
a
standard,
LG
policy
mechanism
to
to
kind
of
Stack
hands
on
that
and-
and
you
know,
many
hands
like
work
and
all
that
to
have
no,
you
know,
contributes
what's
happening
from
the
tooling
and
then
use
that
to
kind
of
try
to
brainstorm
a
a
standard,
Big
Air
quotes
there,
but
standard
policy
resource
all
right,
I'll
see
policy
resource
in
as
much
as
it's
possible
to
do
so
across
different
data,
plane
implementations.
D
So
that's
where
my
brain
is,
but
I
wanted
to
kind
of
see
what
other
people
thought
about
that
or,
if
they're
kind
of
content
with
using
imitation
Pacific
RC
policy
for
now
and
doing
something
else.
But
that's
what
came
up
to
my
mind?
What
does
everybody
else
think
about
like
our
do?
You
know
do
next.
E
E
So
it's
probably
better
to
start
with
this
and
and
I.
Definitely
it's
it's.
You
know
probably
the
most
important
thing
to
standardize,
because
it
would
be
really
bad
for
each
implementation
to
have
its
own
OD
and
but
let's
not
rush
because
we
did
set
before
and
didn't
turn
out
well.
A
Yeah
nothing's,
like
a
great
plan,
I
I,
definitely
contribute
on
behalf
of
like
what
console's
implementation
of
obviously
looks
like
with
our
intentions
currently
and
yeah,
making
sure
this
is
designed
the
way
that
it's
like
implementable
and
performant
across
all
specifications.
It
sounds
like
I,
think
of
like
you
could
share
Lessons
Learned
and
make
sure
that
we
have
support
like
an
intersection
set
of
features
like
I
know
some
messages,
but
not
others
might
be
able
to
do
things
like
L7
off-seat
policies
so
figuring
out
how
to
or
not
work.
C
Yeah
no
I
I
completely
agree,
I
mean
just
to
State
the
obvious.
This
feels
completely
in
scope
for
Gateway
API,
I'd
love
to
I'd
love
to
see
this.
You
know
moving
forward.
So
if
anyone
has
the
time
or
energy
to
start
a
gap
like
yes,
please
so
and
I
agree
with
what
costlin
said,
starting
with
the
survey
of
the
state
of
the
world
is,
you
know
probably
the
best
starting
point
for
this
skip,
especially
but
any
Gap.
D
Okay,
yeah
I
I'll,
go
out
on
a
limb
and
and
take
driving
that
I
started
this
conversation
back
in
October,
but
obviously
other
things
popped
up
between
then
and
now
so
I'm
happy
to
probably
pick
this
back
up
and
try
to
drive
some
consensus
across
different
implementations
here.
D
I
do
some
work
and
hopefully
maybe
have
something
next
week
next
by
next
meeting,
to
probably
start
some
conversation
on
no
promises,
though
I.
E
Have
a
question
there
is
we
had
some
discussions
before
about
some
some
sort
of
Middle
Ground
between
a
standard
across
all
implementation
and
a
standard
across
a
couple
of
implementations
like
only
based
implementations
could
use
something
like
external
mode
Z
or
which,
which
is
not
going
to
be
a
core
API,
probably
forever,
because
it's
a
voice
specific
and
it's
externally
making.
But
probably
there
are
quite
a
few
void-based
implementations
that
could
reuse
it.
Do
we
have
any
way
to
some
repository
some
common
place
where
we
can
store
it?
E
C
No
I've
been
wanting
to
see
this
happen
for
a
while
I'd
talk
with
some
people
in
Envoy
Gateway
team,
and
there
was
some
some
interest
at
a
previous
kubecon
in
you
know:
Envoy
branded
policies,
living
I,
think
outside
of
the
envoy
Gateway
branding,
but
in
an
Envoy,
repo
somewhere
and
I
think
that
that
really
would
fit
well
here
there
we
are
seeing
one
that
a
large
portion
of
Gateway
API
implementations
are
Envoy
based,
maybe
even
a
larger
portion
of
gamma
implementations.
C
I
know
it's
not
100,
but
a
large
portion
R.
So
certainly
if
we
can
standardize
around
that
and
provide
you
know
more
configurability
than
we
would
in
a
screwly,
fully
portable
option.
I
think
Envoy
branded,
something
would
be,
would
be
really
helpful.
So
yeah
I'd
love
to
I'd
love
to
follow
up
I,
think
the
people
that
are
working
on
Envoy
Gateway
probably
are
best
in
just
because
they
they
have
access
more
access
within
Envoy
organization
and
and
are
also
obviously
very
involved
in
Gateway.
E
E
Boy
only
I
mean
there
are
probably
other
implementations
that
are
shared
across
different
vendors
and
so
kind
of
Middle
Ground
between
each
event
or
on
its
own
versus
some
common
common
apis,
but
that
are
not
still
I,
don't
know.
I
I
I
don't
mind
either
way,
but
I
feel
much
better
if
it
was
under
when
I
decided
to
be
honest,.
C
Yeah
that
that
makes
sense,
I
I,
don't
have
a
strong
feeling.
I
I'd
want
someone
some
some
leadership
organization,
whatever
from
Envoy
side,
to
just
be
on
board
with
the
direction
we're
going.
If
we're
using
Envoy
in
our
naming,
but
otherwise
yeah
I
I
agree
with
it.
It
could
be
kubernetes
owned
as
well.
D
D
To
one
thing,
I
want
to
make
sure
that
that
we
discuss
is
that
I,
don't
know,
I
think
it's
probably
best
to
kind
of
take
the
a
page
out
of
gateway
api's
book
here
and
not
necessarily
assume
that
we're
gonna
have
to
break
across
implementations
it.
It
may
be
likely
like
an
external
C
case.
D
That
probably
is
a
fairly
likely
thing
that
we
are
going
to
have
to
you
might
have
to
we're,
definitely
not
to
discuss,
but
if
we
can
find
you
know
a
set
of
of
standards
that
at
least
you
know-
let's
say
Linker
de
proxy
and
onward
proxy
can
come
to
some
kind
of
consensus
on
bring
in
psyllium
in
their
model
and
see
if
they
can
have
some
attraction
consensus
there.
D
D
F
John
and
then
yeah
and
one
concern
we
should
keep
in
mind-
is
that
Envoy
is
not
a
human
API
and
it's
terrible
to
use
as
a
human.
F
So
you
know
it's
tough,
because
you
know
if
we
make
our
own
opinionated
API,
then
you
don't
have
the
same
opinions
than
everyone
else,
but
at
the
same
time,
I
really
really
don't
want
users
to
be
using
audible,
apis
directly
and
I.
Don't
think
users
want
that
either
so.
C
Yeah
I
completely
agree,
Envoy
API
was
not
does
not
feel
like
it
was
made
to
be
user
facing
and,
have
you
know
a
really
friendly,
ux
I
I?
Think
the
the
main
goal
of
anything
like
this,
if
we
were
to
do
if
we
were
to
move
forward,
would
just
be
to
expose
features
that
are
only
configurable
with
Envoy
and
don't
have
the
same
level
of
portability
across
other
products,
but
yeah
I,
I,
think
I'm
agreeing
with
what's
already
been
said
here.
E
Yeah
I
think
I
slightly
disagree
with
this
statement.
I
mean
I
mean
well,
I,
agree,
CPS
are
horrible,
but
taking
an
API
from
employees
that
is
bad
and
attempting
to
just
change
the
names
to
make
it
a
bit
more.
You
know,
lipstick
was
a
big
style
I,
don't
think
it
helps
too
much
either
so
I
would
rather
take
the
original
and
not
have
to
deal
with.
Oh,
this
name
is
translating
to
this
other
name,
because
it
doesn't
sound
good
or
in
practice.
You
know
if
you
look
at
our
externality
representative.
E
F
F
Superficial
changes
are
not
good
and
big
changes
are
hard,
so
it's
a
very
tricky
problem,
but
something
we
should
keep
in
mind.
E
A
Cool
yeah
I
just
wanted
to
like
plus
one
the
key's
point
of
like
I,
think
we
should
start
approaching
this
from.
Let's
see
if
we
can
do
this
in
an
implementation
and
proxy
abstract
way,
because,
as
we
all
know,
like
I,
think
every
mesh
implements
something
quite
similar
to
this.
So
hopefully
we
can
design
an
abstraction
that
is
going
to
make
sense
and
be
able
to
just
like
through
to
finding
something
else.
A
It's
extending
performance,
if
you
want
that
as
like
an
out
for
Xbox,
C
and
onboard
or
something
like
that,
might
be
a
way
to
approach
this,
but
yeah
or
something
yeah.
A
Think
like
yeah,
like,
what's
it
to
say,
like
let's
start
from
the
position
of
like
this,
is
something
we
should
try
to
design
for
and
then
the
like
proxy
based
thing
would
be
kind
of
like
if
there.
C
A
Cool
that
sounds
good
thanks,
Keith
for
volunteering
to
leave
on
that.
We'll
definitely
look
forward
to
reading
a
proposals
to
come
on
that
rob.
You
had
one
topic
here
on
issue.
You
want
to
race.
C
Yeah
I
think
this
has
just
kind
of
gotten
lost
and
it
seems
like
great
to
get
it
resolved
so
I,
don't
know
this
is
more
just
making
sure
we
don't
lose
track
of
it.
I
can't
remember
what
is
really
blocking
here.
I,
don't
know.
If
John
you
remember
any,
you
know
where
this
is
it
does
it
just
need
more
review.
F
A
Think
Port
required
is
probably
a
separate
scope,
separate
discussion,
because
this
feels,
since
our
favorite
was
two
weeks
ago
or
three,
but
it
did
come
up
on
and
again
video
and
we
discussed
it
and
it
makes
sense
to
me
I
haven't
given
an
explosive
rule.
I
should
probably
go
do
that.
A
Basically,
it
looks
like
it
lowers
the
initial
implementation
burden
for
users
of
gamma,
but
in
actuality
as
soon
as
you
want
to
do
anything
with
it
other
than
just
like
have
a
sub
resource
there.
You
end
up
needing
to
add
additional
boilerplate
anyway,
so
it
really
doesn't
save
you
anything.
So
this
makes
sense
to
me
and
seems
like
it.
It
closes
some
gaps
where
we're
inconsistent
with
other
parts
of
the
stack
cells.
B
Yeah
I
mean
it
was
I
think
it
was
Muse.
Looking
for
poor
to
be
required,
so
I
mean
agreed.
We
can
just
make
a
separate
gap
for
that.
That's
not
I.
Don't
want
to
block
this
one.
F
C
A
A
All
right
are
there
any
other
topics.
I
have
one
channel
item
at
the
end,
but
Rick's
that
like
are
there
any
other
topics
that
we
want
to
discuss
during
the
meeting
today.
A
All
right
well
with
that
I'll
get
to
it
everyone
else
that
I'll
be
stepping
down
as
a
full
weight
of
gamma,
San
Diego,
but
definitely
kind
of
bittersweet.
It's
I
really
appreciate
working
with
everyone
over
the
past
year
to
kind
of
like
bring
this
project
from
an
idea
to
an
actual
instrumental
thing.
A
That's
coming
in
a
few
weeks,
so
really
excited
to
kind
of
like
see
everything
we
know
put
together
and
yeah
I'll
be
focused
internally
more
on
actually
implementing
gamma,
which
will
be
a
bit
of
a
nice
change.
So
yeah
again
was
very
much
still
something
that
console
intends
to
implement.
That
is
not
changing.
A
My
role
is
just
gonna
be
shifting
a
bit
and
we've
had
some
prerequisite
work,
and
hopefully
that
should
be
wrapping
up
soon,
so
we'll
actually
be
able
to
start
with
implementing
some
of
this,
and
hopefully
it
may
be
me,
and
maybe
other
folks,
still
being
able
to
like
chime
in
Upstream
as
we
kind
of
explore
and
hit
issues
or
successes.
A
A
Everyone
who's
on
this
call
and
also
a
book
for
watching
you
know
who
you
are
and
yeah
I,
think
the
Project's
in
good
hands
and
I'm
excited
to
see
continue
to
grow
and
continue
to
run
beyond
our
initial
scope
of
routing
tube
start
watching
things
like
off
the
egress
Etc,
so
I
have
a
bright
future
for
it
and
yeah
I'll
get
off
my
soapbox
now.
But
just
thank
you.
Everyone.
D
I
want
to
make
sure
everybody
here
understands.
You
know
just
how
how
pivotal
Mike
has
been
when
back
when
Mike
and
I
were
working
on
SMI.
He
suggested
looking
at
the
Gateway
API
and
seeing
what's
going
on
there
and
that's
a
integral
part
of
how
gamble
was
born,
and
we
wouldn't
be
here
if
it
weren't
for
Mike,
so
on
the
on
behalf
of
everybody,
using
gamma
and
working
with
it.
D
Thank
you
for
the
work
that
you've
contributed,
sorry
to
excuse
to
step
away
and
Hyper
World
change,
but
we
are
better
for
having
had
your
contributions
so
much
thanks.
C
Yeah
also
just
a
huge
thanks
to
to
Mike
yeah
you've
you've
been
awesome,
help
to
the
community
and
I'm
I'm,
so
glad
that
gamma
exists
and
I
know.
You
were
instrumental
in
that
and
you've
you've
pushed
it
forward
and
helped
us
get
to
where
we
are,
and
so
just
thanks
for
being
a
great
influence
on
the
community
and
even
though
you'll
be
less
involved
in
OSS,
I'm
sure
we'll
still
see
you
around
so
I.
A
All
right,
well,
thanks
everybody
that
will
wrap
it
up
for
me
each
day
a
I
will
probably
not
be
hosting
the
next
one
of
these,
but
yeah
stay
tuned
for
an
email
like
announcements
formally
and
yeah.
Thank
you
all
so
much
take
care
and
have
a
good
rest
of
your
day.